1 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
7 * tree.c (mem_ref_offset): Likewise.
8 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
9 * builtins.c (get_object_alignment_2): Likewise.
10 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
11 * gimple-fold.c (get_base_constructor): Likewise.
12 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
13 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
15 * ipa-polymorphic-call.c
16 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
17 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
18 (get_ancestor_addr_info): Likewise.
19 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
21 * tree-data-ref.c (dr_analyze_innermost): Likewise.
22 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
23 * tree-eh.c (tree_could_trap_p): Likewise.
24 * tree-object-size.c (addr_object_size): Likewise.
25 * tree-ssa-address.c (copy_ref_info): Likewise.
26 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
27 (indirect_refs_may_alias_p): Likewise.
28 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
29 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
30 (non_rewritable_mem_ref_base): Likewise.
31 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
32 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
33 * varasm.c (decode_addr_const): Likewise.
35 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
36 Alan Hayward <alan.hayward@arm.com>
37 David Sherwood <david.sherwood@arm.com>
39 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
42 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
43 Alan Hayward <alan.hayward@arm.com>
44 David Sherwood <david.sherwood@arm.com>
46 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
47 as a poly_uint64 rather than an unsigned int.
48 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
50 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
51 Alan Hayward <alan.hayward@arm.com>
52 David Sherwood <david.sherwood@arm.com>
54 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
56 (decode_addr_const): Update accordingly.
58 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
59 Alan Hayward <alan.hayward@arm.com>
60 David Sherwood <david.sherwood@arm.com>
62 * tree.h (bit_field_size, bit_field_offset): New functions.
63 * hsa-gen.c (gen_hsa_addr): Use them.
64 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
65 (simplify_vector_constructor): Likewise.
66 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
67 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
68 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
69 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
72 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
73 Alan Hayward <alan.hayward@arm.com>
74 David Sherwood <david.sherwood@arm.com>
76 * expr.h (emit_group_load, emit_group_load_into_temps)
77 (emit_group_store): Take the size as a poly_int64 rather than an int.
78 * expr.c (emit_group_load_1, emit_group_load): Likewise.
79 (emit_group_load_into_temp, emit_group_store): Likewise.
81 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
82 Alan Hayward <alan.hayward@arm.com>
83 David Sherwood <david.sherwood@arm.com>
85 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
86 an unsigned int to a poly_uint64.
87 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
88 sizes as poly_uint64s rather than unsigned ints.
89 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
92 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
93 Alan Hayward <alan.hayward@arm.com>
94 David Sherwood <david.sherwood@arm.com>
96 * lra-constraints.c (emit_inc): Change inc_amount from an int
99 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
100 Alan Hayward <alan.hayward@arm.com>
101 David Sherwood <david.sherwood@arm.com>
103 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
105 (add_stack_var, stack_var_cmp, partition_stack_vars)
106 (dump_stack_var_partition): Update accordingly.
107 (alloc_stack_frame_space): Take the size as a poly_int64 rather
108 than a HOST_WIDE_INT.
109 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
110 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
111 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
112 rather than a HOST_WIDE_INT.
114 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
115 Alan Hayward <alan.hayward@arm.com>
116 David Sherwood <david.sherwood@arm.com>
118 * target.def (return_pops_args): Treat both the input and output
119 sizes as poly_int64s rather than HOST_WIDE_INTS.
120 * targhooks.h (default_return_pops_args): Update accordingly.
121 * targhooks.c (default_return_pops_args): Likewise.
122 * doc/tm.texi: Regenerate.
123 * emit-rtl.h (incoming_args): Change pops_args, size and
124 outgoing_args_size from int to poly_int64_pod.
125 * function.h (expr_status): Change x_pending_stack_adjust and
126 x_stack_pointer_delta from int to poly_int64.
127 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
128 (ARGS_SIZE_RTX): Update accordingly.
129 * calls.c (highest_outgoing_arg_in_use): Change from int to
131 (stack_usage_watermark, stored_args_watermark): New variables.
132 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
133 (emit_call_1): Change the stack_size and rounded_stack_size
134 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
136 (save_fixed_argument_area): Check stack_usage_watermark.
137 (initialize_argument_information): Change old_pending_adj from
138 a HOST_WIDE_INT * to a poly_int64_pod *.
139 (compute_argument_block_size): Return the size as a poly_int64
141 (finalize_must_preallocate): Track polynomial argument sizes.
142 (compute_argument_addresses): Likewise.
143 (internal_arg_pointer_based_exp): Track polynomial offsets.
144 (mem_overlaps_already_clobbered_arg_p): Rename to...
145 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
146 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
147 Check stored_args_used_watermark.
148 (load_register_parameters): Update accordingly.
149 (check_sibcall_argument_overlap_1): Likewise.
150 (combine_pending_stack_adjustment_and_call): Take the unadjusted
151 args size as a poly_int64 rather than an int. Return a bool
152 indicating whether the optimization was possible and return
153 the new adjustment by reference.
154 (check_sibcall_argument_overlap): Track polynomail argument sizes.
155 Update stored_args_watermark.
156 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
157 (expand_call): Likewise. Maintain stack_usage_watermark and
158 stored_args_watermark. Update calls to
159 combine_pending_stack_adjustment_and_call.
160 (emit_library_call_value_1): Handle polynomial argument sizes.
161 Call stack_region_maybe_used_p and mark_stack_region_used.
162 Maintain stack_usage_watermark.
163 (store_one_arg): Likewise. Update call to
164 mem_overlaps_already_clobbered_arg_p.
165 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
167 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
168 * config/microblaze/microblaze.c (microblaze_function_prologue):
170 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
171 TARGET_RETURN_POPS_ARGS interface.
172 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
174 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
175 * config/i386/i386.c (ix86_return_pops_args): Update for new
176 TARGET_RETURN_POPS_ARGS interface.
177 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
178 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
179 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
180 TARGET_RETURN_POPS_ARGS interface.
181 * config/vax/vax.c (vax_return_pops_args): Likewise.
182 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
183 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
184 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
185 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
186 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
187 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
188 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
189 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
190 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
191 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
192 and x_stack_pointer_delta from int to poly_int64.
193 * dojump.c (do_pending_stack_adjust): Update accordingly.
194 * explow.c (allocate_dynamic_stack_space): Handle polynomial
195 stack_pointer_deltas.
196 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
197 (pad_to_arg_alignment): Track polynomial offsets.
198 (assign_parm_find_stack_rtl): Likewise.
199 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
200 * toplev.c (output_stack_usage): Update reference to
201 current_function_pushed_stack_size.
203 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
204 Alan Hayward <alan.hayward@arm.com>
205 David Sherwood <david.sherwood@arm.com>
207 * function.c (in_arg_offset, var_offset, dynamic_offset)
208 (out_arg_offset, cfa_offset): Change from int to poly_int64.
209 (instantiate_new_reg): Return the new offset as a poly_int64_pod
210 rather than a HOST_WIDE_INT.
211 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
212 (instantiate_virtual_regs_in_insn): Likewise.
214 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
215 Alan Hayward <alan.hayward@arm.com>
216 David Sherwood <david.sherwood@arm.com>
218 * rtl.h (get_args_size, add_args_size_note): New functions.
219 (find_args_size_adjust): Return a poly_int64 rather than a
221 (fixup_args_size_notes): Likewise. Make the same change to the
222 end_args_size parameter.
223 * rtlanal.c (get_args_size, add_args_size_note): New functions.
224 * builtins.c (expand_builtin_trap): Use add_args_size_note.
225 * calls.c (emit_call_1): Likewise.
226 * explow.c (adjust_stack_1): Likewise.
227 * cfgcleanup.c (old_insns_match_p): Update use of
228 find_args_size_adjust.
229 * combine.c (distribute_notes): Track polynomial arg sizes.
230 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
231 end_true_args_size, beg_delay_args_size and end_delay_args_size
232 from HOST_WIDE_INT to poly_int64.
233 (add_cfi_args_size): Take the args_size as a poly_int64 rather
234 than a HOST_WIDE_INT.
235 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
236 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
237 polynomial arg sizes.
238 * emit-rtl.c (try_split): Use get_args_size.
239 * recog.c (peep2_attempt): Likewise.
240 * reload1.c (reload_as_needed): Likewise.
241 * expr.c (find_args_size_adjust): Return the adjustment as a
242 poly_int64 rather than a HOST_WIDE_INT.
243 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
244 to a poly_int64 and change the return type in the same way.
245 (emit_single_push_insn): Track polynomial arg sizes.
247 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
248 Alan Hayward <alan.hayward@arm.com>
249 David Sherwood <david.sherwood@arm.com>
251 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
252 from HOST_WIDE_INT to poly_int64.
253 * expr.c (push_block, emit_push_insn): Likewise.
255 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
256 Alan Hayward <alan.hayward@arm.com>
257 David Sherwood <david.sherwood@arm.com>
259 * function.h (frame_space): Change start and length from HOST_WIDE_INT
261 (get_frame_size): Return the size as a poly_int64 rather than a
263 (frame_offset_overflow): Take the offset as a poly_int64 rather
264 than a HOST_WIDE_INT.
265 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
266 (assign_stack_temp): Likewise for the size.
267 * function.c (get_frame_size): Return a poly_int64 rather than
269 (frame_offset_overflow): Take the offset as a poly_int64 rather
270 than a HOST_WIDE_INT.
271 (try_fit_stack_local): Take the start, length and size as poly_int64s
272 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
273 rather than a HOST_WIDE_INT.
274 (add_frame_space): Take the start and end as poly_int64s rather than
276 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
277 (assign_stack_temp): Likewise for the size.
278 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
280 (find_temp_slot_from_address): Handle polynomial offsets.
281 (combine_temp_slots): Likewise.
282 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
284 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
285 poly_int64 rather than a HOST_WIDE_INT.
286 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
287 than a HOST_WIDE_INT.
288 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
289 polynomial frame offsets.
290 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
291 as a poly_int64 rather than an int.
292 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
293 * config/v850/v850-protos.h (compute_frame_size): Likewise.
294 * config/v850/v850.c (compute_frame_size): Likewise.
295 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
296 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
297 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
298 * config/pa/pa.c (pa_compute_frame_size): Likewise.
299 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
300 rather than a HOST_WIDE_INT.
301 * explow.c (get_dynamic_stack_base): Likewise.
302 * final.c (final_start_function): Use the constant lower bound
303 of the frame size for -Wframe-larger-than.
304 * ira.c (do_reload): Adjust for new get_frame_size return type.
305 * lra.c (lra): Likewise.
306 * reload1.c (reload): Likewise.
307 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
308 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
309 * rtlanal.c (get_initial_register_offset): Return the offset as
310 a poly_int64 rather than a HOST_WIDE_INT.
312 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
313 Alan Hayward <alan.hayward@arm.com>
314 David Sherwood <david.sherwood@arm.com>
316 * reload1.c (elim_table): Change initial_offset, offset and
317 previous_offset from HOST_WIDE_INT to poly_int64_pod.
318 (offsets_at): Change the target array's element type from
319 HOST_WIDE_INT to poly_int64_pod.
320 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
321 (elimination_costs_in_insn, update_eliminable_offsets)
322 (verify_initial_elim_offsets, set_offsets_for_label)
323 (init_eliminable_invariants): Update after above changes.
325 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
326 Alan Hayward <alan.hayward@arm.com>
327 David Sherwood <david.sherwood@arm.com>
329 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
330 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
331 (decomposition): Change start and end from HOST_WIDE_INT
333 (decompose, immune_p): Update accordingly.
334 (find_inc_amount): Return a poly_int64 rather than an int.
335 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
338 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
339 Alan Hayward <alan.hayward@arm.com>
340 David Sherwood <david.sherwood@arm.com>
342 * tree.h (get_inner_reference): Return the bitsize and bitpos
343 as poly_int64_pods rather than HOST_WIDE_INT.
344 * fold-const.h (ptr_difference_const): Return the pointer difference
345 as a poly_int64_pod rather than a HOST_WIDE_INT.
346 * expr.c (get_inner_reference): Return the bitsize and bitpos
347 as poly_int64_pods rather than HOST_WIDE_INT.
348 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
350 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
351 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
352 (optimize_bit_field_compare): Update call to get_inner_reference.
353 (decode_field_reference): Likewise.
354 (fold_unary_loc): Track polynomial offsets and sizes.
355 (split_address_to_core_and_offset): Return the bitpos as a
356 poly_int64_pod rather than a HOST_WIDE_INT.
357 (ptr_difference_const): Likewise for the pointer difference.
358 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
359 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
360 * dbxout.c (dbxout_expand_expr): Likewise.
361 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
362 (loc_list_from_tree_1, fortran_common): Likewise.
363 * gimple-laddress.c (pass_laddress::execute): Likewise.
364 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
365 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
366 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
367 * tree-affine.c (tree_to_aff_combination): Likewise.
368 (get_inner_reference_aff): Likewise.
369 * tree-data-ref.c (split_constant_offset_1): Likewise.
370 (dr_analyze_innermost): Likewise.
371 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
372 * tree-sra.c (ipa_sra_check_caller): Likewise.
373 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
374 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
375 (instrument_bool_enum_load, instrument_object_size): Likewise.
376 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
377 to get_inner_reference.
378 * hsa-gen.c (gen_hsa_addr): Likewise.
379 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
380 * tsan.c (instrument_expr): Likewise.
381 * match.pd: Update call to ptr_difference_const.
383 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
384 Alan Hayward <alan.hayward@arm.com>
385 David Sherwood <david.sherwood@arm.com>
387 * fold-const.c (fold_comparison): Track sizes and offsets as
388 poly_int64s rather than HOST_WIDE_INTs when folding address
391 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
392 Alan Hayward <alan.hayward@arm.com>
393 David Sherwood <david.sherwood@arm.com>
395 * expr.h (get_bit_range): Return the bitstart and bitend as
396 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
397 as a poly_int64 rather than a HOST_WIDE_INT.
398 * expr.c (get_bit_range): Likewise.
399 (expand_assignment): Update call accordingly.
400 * fold-const.c (optimize_bit_field_compare): Likewise.
402 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
403 Alan Hayward <alan.hayward@arm.com>
404 David Sherwood <david.sherwood@arm.com>
406 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
407 (aarch64_legitimate_address_p): Use it instead of an rtx code,
408 as an optional final parameter.
409 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
410 (aarch64_legitimate_address_p): Likewise.
411 (aarch64_print_address_internal): Take an aarch64_addr_query_type
412 instead of an rtx code.
413 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
414 (aarch64_legitimate_address_hook_p): Likewise.
415 (aarch64_print_ldpstp_address): Likewise.
416 (aarch64_print_operand_address): Likewise.
417 (aarch64_address_cost): Likewise.
418 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
419 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
420 (aarch64_mem_pair_lanes_operand): Likewise.
422 2017-12-20 Richard Biener <rguenther@suse.de>
424 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
425 update_stmt_if_modified.
427 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
429 PR tree-optimization/83491
430 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
431 before walking uses. Improve coding style and comments.
433 2017-12-20 Tom de Vries <tom@codesourcery.com>
435 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
437 2017-12-20 Jakub Jelinek <jakub@redhat.com>
440 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
442 (pass_ipa_free_fn_summary): Add small_p private data member,
443 initialize to false in the ctor.
444 (pass_ipa_free_fn_summary::clone,
445 pass_ipa_free_fn_summary::set_pass_param,
446 pass_ipa_free_fn_summary::gate): New methods.
447 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
448 | TODO_dump_symtab if small_p.
449 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
450 entry and add another instance of the pass with false parm after
452 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
453 ipa_free_fn_summary here.
455 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
457 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
459 2017-12-20 Martin Sebor <msebor@redhat.com>
462 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
463 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
464 (expand_builtin_strncmp): Same.
466 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
469 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
470 allowed debug stmts before labels.
471 (expand_gimple_basic_block): Likewise.
472 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
473 * gimple-iterator.h (gsi_after_labels): Likewise.
474 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
475 rename reused variable, and simplify using gsi_move_before.
476 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
477 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
478 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
479 (gimple_verify_flow_info, gimple_block_label): Likewise.
480 (make_blocks): Move debug markers after adjacent labels.
481 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
482 allowed debug insns outside blocks.
483 * df-scan.c (df_insn_delete): Likewise.
484 * lra-constraints.c (update_ebb_live_info): Likewise.
485 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
486 (vt_initialize, delete_vta_debug_insns): Likewise.
487 (reemit_marker_as_note): Drop BB parm. Adjust callers.
489 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
490 Alan Hayward <alan.hayward@arm.com>
491 David Sherwood <david.sherwood@arm.com>
493 * poly-int-types.h (round_down_to_byte_boundary): New macro.
494 (round_up_to_byte_boundary): Likewise.
495 * expr.h (get_bit_range): Add temporary shim.
496 * gimple-ssa-store-merging.c (store_operand_info): Change the
497 bitsize, bitpos, bitregion_start and bitregion_end fields from
498 unsigned HOST_WIDE_INT to poly_uint64.
499 (merged_store_group): Likewise load_align_base.
500 (compatible_load_p, compatible_load_p): Update accordingly.
501 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
502 (split_group, imm_store_chain_info::output_merged_store): Likewise.
503 (mem_valid_for_store_merging): Return the bitsize, bitpos,
504 bitregion_start and bitregion_end as poly_uint64s rather than
505 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
506 (handled_load): Take the bitsize, bitpos,
507 bitregion_start and bitregion_end as poly_uint64s rather than
508 unsigned HOST_WIDE_INTs.
509 (pass_store_merging::process_store): Update call to
510 mem_valid_for_store_merging.
512 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
513 Alan Hayward <alan.hayward@arm.com>
514 David Sherwood <david.sherwood@arm.com>
516 * builtins.c (get_object_alignment_2): Track polynomial offsets
517 and sizes. Update the alignment handling.
519 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
520 Alan Hayward <alan.hayward@arm.com>
521 David Sherwood <david.sherwood@arm.com>
523 * tree.h (get_inner_reference): Add a version that returns the
524 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
525 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
526 the case in which bitpos is not associated with the first byte.
528 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
529 Alan Hayward <alan.hayward@arm.com>
530 David Sherwood <david.sherwood@arm.com>
532 * tree-affine.h (get_inner_reference_aff): Return the size as a
534 * tree-affine.c (get_inner_reference_aff): Likewise.
535 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
536 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
538 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
539 Alan Hayward <alan.hayward@arm.com>
540 David Sherwood <david.sherwood@arm.com>
542 * fold-const.c (pointer_may_wrap_p): Take the offset as a
543 HOST_WIDE_INT rather than a poly_int64.
545 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
546 Alan Hayward <alan.hayward@arm.com>
547 David Sherwood <david.sherwood@arm.com>
549 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
550 HOST_WIDE_INT to poly_int64_pod.
551 (perform_symbolic_merge): Update accordingly.
552 (bswap_replace): Likewise.
554 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
555 Alan Hayward <alan.hayward@arm.com>
556 David Sherwood <david.sherwood@arm.com>
558 * tree-affine.h (aff_tree::offset): Change from widest_int
560 (wide_int_ext_for_comb): Delete.
561 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
562 constants as poly_widest_int rather than widest_int.
563 (aff_combination_constant_multiple_p): Return the multiplier
564 as a poly_widest_int.
565 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
567 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
568 version static and add an overload for poly_widest_int.
569 (aff_combination_const, aff_combination_add_cst)
570 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
571 the constants as poly_widest_int rather than widest_int.
572 (tree_to_aff_combination): Generalize INTEGER_CST case to
574 (aff_combination_to_tree): Track offsets as poly_widest_ints.
575 (aff_combination_add_product, aff_combination_mult): Handle
577 (aff_combination_constant_multiple_p): Return the multiplier
578 as a poly_widest_int.
579 * tree-predcom.c (determine_offset): Return the offset as a
581 (split_data_refs_to_components, suitable_component_p): Update
583 (valid_initializer_p): Update call to
584 aff_combination_constant_multiple_p.
585 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
586 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
587 as a poly_int64 rather than a HOST_WIDE_INT.
588 (get_address_cost): Handle polynomial offsets.
589 (iv_elimination_compare_lt): Likewise.
590 (rewrite_use_nonlinear_expr): Likewise.
592 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
593 Alan Hayward <alan.hayward@arm.com>
594 David Sherwood <david.sherwood@arm.com>
596 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
597 as a poly_int64_pod rather than a HOST_WIDE_INT.
598 (get_addr_base_and_unit_offset): Likewise.
599 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
600 (get_addr_base_and_unit_offset): Likewise.
601 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
602 to poly_int64 in example.
603 * fold-const.c (fold_binary_loc): Update call to
604 get_addr_base_and_unit_offset.
605 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
606 (maybe_canonicalize_mem_ref_addr): Likewise.
607 (gimple_fold_stmt_to_constant_1): Likewise.
608 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
610 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
611 * match.pd: Likewise.
612 * omp-low.c (lower_omp_target): Likewise.
613 * tree-sra.c (build_ref_for_offset): Likewise.
614 (build_debug_ref_for_model): Likewise.
615 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
616 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
617 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
618 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
619 (constant_pointer_difference): Likewise.
620 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
621 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
622 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
623 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
624 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
625 (set_ssa_val_to): Likewise.
626 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
627 (maybe_diag_stxncpy_trunc): Likewise.
628 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
629 * tree.c (build_simple_mem_ref_loc): Likewise.
630 (array_at_struct_end_p): Likewise.
632 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
633 Alan Hayward <alan.hayward@arm.com>
634 David Sherwood <david.sherwood@arm.com>
636 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
637 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
638 (get_ref_base_and_extent_hwi): Declare.
639 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
640 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
641 (get_ref_base_and_extent_hwi): New function.
642 * cfgexpand.c (expand_debug_expr): Update call to
643 get_ref_base_and_extent.
644 * dwarf2out.c (add_var_loc_to_decl): Likewise.
645 * gimple-fold.c (get_base_constructor): Return the offset as a
646 poly_int64_pod rather than a HOST_WIDE_INT.
647 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
648 * ipa-polymorphic-call.c
649 (ipa_polymorphic_call_context::set_by_invariant)
650 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
651 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
652 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
653 rather than get_ref_base_and_extent.
654 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
655 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
656 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
658 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
659 call to get_ref_base_and_extent.
660 * tree-sra.c (create_access, get_access_for_expr): Likewise.
661 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
662 (stmt_kills_ref_p): Likewise.
663 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
664 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
666 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
667 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
668 when calling native_encode_expr.
669 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
670 call to get_ref_base_and_extent.
671 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
672 get_ref_base_and_extent.
673 * var-tracking.c (track_expr_p): Likewise.
675 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
676 Alan Hayward <alan.hayward@arm.com>
677 David Sherwood <david.sherwood@arm.com>
679 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
680 HOST_WIDE_INT to poly_int64_pod.
681 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
682 polynomail parameter offsets.
684 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
685 Alan Hayward <alan.hayward@arm.com>
686 David Sherwood <david.sherwood@arm.com>
688 * gengtype.c (main): Handle poly_int64_pod.
689 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
690 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
691 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
692 from HOST_WIDE_INT to poly_int64_pod.
693 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
694 (copy_cfa): New function.
695 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
696 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
697 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
698 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
699 (initial_return_save): Treat offsets as poly_ints.
700 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
701 offset is nonconstant.
702 (reg_save): Take the offset as a poly_int64. Fall back to
703 DW_CFA_expression for nonconstant offsets.
704 (queue_reg_save): Take the offset as a poly_int64.
705 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
707 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
708 Alan Hayward <alan.hayward@arm.com>
709 David Sherwood <david.sherwood@arm.com>
711 * rtl.h (operand_subword, operand_subword_force): Take the offset
712 as a poly_uint64 an unsigned int.
713 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
715 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
716 Alan Hayward <alan.hayward@arm.com>
717 David Sherwood <david.sherwood@arm.com>
719 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
720 'p' format code. Use INT_LIST rather than SUBREG as the example of
721 a code with an XINT and an XEXP. Remove the implication that
722 accessing an rtx field using XINT is expected to work.
723 * rtl.def (SUBREG): Change format from "ei" to "ep".
724 * rtl.h (rtunion::rt_subreg): New field.
725 (XCSUBREG): New macro.
726 (SUBREG_BYTE): Use it.
727 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
728 Update constructor accordingly.
729 (subreg_shape::operator ==): Update accordingly.
730 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
731 than an unsigned int.
732 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
733 a poly_uint64 rather than an unsigned int.
734 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
735 than an unsigned int.
736 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
737 (subreg_size_highpart_offset): Return a poly_uint64 rather than
738 an unsigned int. Take the sizes as poly_uint64s.
739 (subreg_offset_from_lsb): Return a poly_uint64 rather than
740 an unsigned int. Take the shift as a poly_uint64 rather than
742 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
743 as a poly_uint64 rather than an unsigned int.
744 (simplify_subreg_regno): Likewise.
745 (byte_lowpart_offset): Return the memory offset as a poly_int64
747 (subreg_memory_offset): Likewise. Take the subreg offset as a
748 poly_uint64 rather than an unsigned int.
749 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
750 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
751 poly_uint64 rather than an unsigned int.
752 * rtl.c (rtx_format): Describe 'p' in comment.
753 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
754 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
755 offset as a poly_uint64 rather than an unsigned int.
756 (byte_lowpart_offset): Return the memory offset as a poly_int64
758 (subreg_memory_offset): Likewise. Take the subreg offset as a
759 poly_uint64 rather than an unsigned int.
760 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
761 mode sizes as poly_uint64s rather than unsigned ints. Return a
762 poly_uint64 rather than an unsigned int.
763 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
764 (copy_insn_1): Handle 'p'.
765 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
766 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
767 an unsigned int. Return the shift in the same way.
768 (subreg_lsb): Return the shift as a poly_uint64 rather than an
770 (subreg_size_offset_from_lsb): Take the sizes and shift as
771 poly_uint64s rather than unsigned ints. Return the offset as
773 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
774 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
776 * rtlhash.c (add_rtx): Handle 'p'.
777 * genemit.c (gen_exp): Likewise.
778 * gengenrtl.c (type_from_format, gendef): Likewise.
779 * gensupport.c (subst_pattern_match, get_alternatives_number)
780 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
781 (subst_dup): Likewise.
782 * gengtype.c (adjust_field_rtx_def): Likewise.
783 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
784 (match_pattern_2): Likewise.
785 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
786 (rtx_test::subreg_field): New function.
787 (operator ==, safe_to_hoist_p, transition_parameter_type)
788 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
789 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
790 * genpeep.c (match_rtx): Likewise.
791 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
792 (rtx_writer::print_rtx_operand): Handle 'p'.
793 (print_value): Handle SUBREG.
794 * read-rtl.c (apply_int_iterator): Likewise.
795 (rtx_reader::read_rtx_operand): Handle 'p'.
796 * alias.c (rtx_equal_for_memref_p): Likewise.
797 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
798 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
800 * calls.c (expand_call): Likewise.
801 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
802 (make_extraction, gen_lowpart_for_combine): Likewise.
803 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
805 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
806 rather than an unsigned int. Treat subreg offsets as poly_ints.
807 (exp_equiv_p): Handle 'p'.
808 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
809 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
810 * dse.c (find_shift_sequence): Likewise.
811 * dwarf2out.c (rtl_for_decl_location): Likewise.
812 * expmed.c (extract_low_bits): Likewise.
813 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
814 (expand_expr_real_2): Likewise.
815 * final.c (alter_subreg): Likewise.
816 (leaf_renumber_regs_insn): Handle 'p'.
817 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
818 Treat subreg offsets as poly_ints.
819 * fwprop.c (forward_propagate_and_simplify): Likewise.
820 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
821 * ira.c (get_subreg_tracking_sizes): Likewise.
822 * ira-conflicts.c (go_through_subreg): Likewise.
823 * ira-lives.c (process_single_reg_class_operands): Likewise.
824 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
825 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
826 as a poly_uint64 rather than an unsigned int.
827 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
828 subreg offsets as poly_ints.
829 * lra-constraints.c (operands_match_p): Handle 'p'.
830 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
831 * lra-spills.c (assign_mem_slot): Likewise.
832 * postreload.c (move2add_valid_value_p): Likewise.
833 * recog.c (general_operand, indirect_operand): Likewise.
834 * regcprop.c (copy_value, maybe_mode_change): Likewise.
835 (copyprop_hardreg_forward_1): Likewise.
836 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
837 (record_subregs_of_mode): Likewise.
838 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
839 * reload.c (operands_match_p): Handle 'p'.
840 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
841 * reload1.c (alter_reg, choose_reload_regs): Likewise.
842 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
843 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
844 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
845 (simplify_const_poly_int_tests<N>::run): Likewise.
846 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
847 a poly_uint64 rather than an unsigned int.
848 * valtrack.c (debug_lowpart_subreg): Likewise.
849 * var-tracking.c (var_lowpart): Likewise.
850 (loc_cmp): Handle 'p'.
852 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
853 Alan Hayward <alan.hayward@arm.com>
854 David Sherwood <david.sherwood@arm.com>
856 * ira.c (get_subreg_tracking_sizes): New function.
857 (init_live_subregs): Take an integer size rather than a register.
858 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
859 to init_live_subregs.
861 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
862 Alan Hayward <alan.hayward@arm.com>
863 David Sherwood <david.sherwood@arm.com>
865 * expr.c (store_constructor_field): Change bitsize from a
866 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
867 HOST_WIDE_INT to a poly_int64.
868 (store_constructor): Change size from a HOST_WIDE_INT to
870 (store_field): Likewise bitsize and bitpos.
872 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
873 Alan Hayward <alan.hayward@arm.com>
874 David Sherwood <david.sherwood@arm.com>
876 * expmed.h (store_bit_field): Change bitregion_start and
877 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
878 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
879 (store_bit_field_1, store_integral_bit_field, store_bit_field)
880 (store_fixed_bit_field, store_split_bit_field): Likewise.
881 * expr.c (store_constructor_field, store_field): Likewise.
882 (optimize_bitfield_assignment_op): Likewise. Make the same change
883 to bitsize and bitpos.
884 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
885 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
886 same change in the constructor arguments.
887 (get_best_mode): Change bitregion_start and bitregion_end from
888 unsigned HOST_WIDE_INT to poly_uint64.
889 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
890 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
892 (bit_field_mode_iterator::next_mode): Update for new types
893 of m_bitregion_start and m_bitregion_end.
894 (get_best_mode): Change bitregion_start and bitregion_end from
895 unsigned HOST_WIDE_INT to poly_uint64.
897 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
898 Alan Hayward <alan.hayward@arm.com>
899 David Sherwood <david.sherwood@arm.com>
901 * rtl.h (simplify_gen_subreg): Add a temporary overload that
902 accepts poly_uint64 offsets.
903 * expmed.h (extract_bit_field): Take bitsize and bitnum as
904 poly_uint64s rather than unsigned HOST_WIDE_INTs.
905 * expmed.c (lowpart_bit_field_p): Likewise.
906 (extract_bit_field_as_subreg): New function, split out from...
907 (extract_bit_field_1): ...here. Take bitsize and bitnum as
908 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
909 extractions, check that BITSIZE matches the size of the extracted
910 value and that BITNUM is an exact multiple of that size.
911 If all else fails, try forcing the value into memory if
912 BITNUM is variable, and adjusting the address so that the
913 offset is constant. Split the part that can only handle constant
914 bitsize and bitnum out into...
915 (extract_integral_bit_field): ...this new function.
916 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
917 rather than unsigned HOST_WIDE_INTs.
919 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
920 Alan Hayward <alan.hayward@arm.com>
921 David Sherwood <david.sherwood@arm.com>
923 * expmed.h (store_bit_field): Take bitsize and bitnum as
924 poly_uint64s rather than unsigned HOST_WIDE_INTs.
925 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
926 that returns the byte size.
927 (store_bit_field_1): Take bitsize and bitnum as
928 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
929 to simple_mem_bitfield_p. Split the part that can only handle
930 constant bitsize and bitnum out into...
931 (store_integral_bit_field): ...this new function.
932 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
933 than unsigned HOST_WIDE_INTs.
934 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
936 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
937 Alan Hayward <alan.hayward@arm.com>
938 David Sherwood <david.sherwood@arm.com>
940 * lra-int.h (lra_reg): Change offset from int to poly_int64.
941 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
943 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
944 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
945 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
946 offset as a poly_int64 rather than an int.
947 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
948 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
949 * lra-constraints.c (equiv_address_substitution): Track offsets
951 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
952 (curr_insn_transform): Handle the new form of sp_offset.
953 * lra-eliminations.c (lra_elim_table): Change previous_offset
954 and offset from HOST_WIDE_INT to poly_int64.
955 (print_elim_table, update_reg_eliminate): Update accordingly.
956 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
957 (get_elimination): Update accordingly.
958 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
959 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
960 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
961 poly_int64 offsets generally.
962 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
963 (mark_not_eliminable, init_elimination): Update accordingly.
964 (remove_reg_equal_offset_note): Return a bool and pass the new
965 offset back by pointer as a poly_int64.
966 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
967 rather than a HOST_WIDE_INT.
968 (do_remat): Track offsets poly_int64s.
969 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
971 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
972 Alan Hayward <alan.hayward@arm.com>
973 David Sherwood <david.sherwood@arm.com>
975 * rtl.h (mem_attrs): Add a default constructor. Change size and
976 offset from HOST_WIDE_INT to poly_int64.
977 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
978 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
979 (widen_memory_access): Take the sizes and offsets as poly_int64s
980 rather than HOST_WIDE_INTs.
981 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
982 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
984 (adjust_offset_for_component_ref): Change the offset from a
985 HOST_WIDE_INT to a poly_int64.
986 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
987 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
988 * dce.c (find_call_stack_args): Likewise.
989 * dse.c (record_store): Likewise.
990 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
991 * print-rtl.c (rtx_writer::print_rtx): Likewise.
992 * read-rtl-function.c (test_loading_mem): Likewise.
993 * rtlanal.c (may_trap_p_1): Likewise.
994 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
995 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
996 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
997 (mem_attrs::mem_attrs): New function.
998 (set_mem_attributes_minus_bitpos): Change bitpos from a
999 HOST_WIDE_INT to poly_int64.
1000 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1001 (clear_mem_offset, clear_mem_size, change_address)
1002 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1003 initialize mem_attrs.
1004 (set_mem_offset, set_mem_size, adjust_address_1)
1005 (adjust_automodify_address_1, offset_address, widen_memory_access):
1006 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1008 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1009 Alan Hayward <alan.hayward@arm.com>
1010 David Sherwood <david.sherwood@arm.com>
1012 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1013 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1014 rather than 0 to represent an unknown size. Assert that the size
1015 is known when the mode isn't BLKmode.
1016 (may_trap_p_1): Use -1 for unknown sizes.
1017 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1019 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1020 Alan Hayward <alan.hayward@arm.com>
1021 David Sherwood <david.sherwood@arm.com>
1023 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1024 to poly_int64. Update commentary for positions_needed.large.
1025 (read_info_type): Change offset and width from HOST_WIDE_INT
1027 (set_usage_bits): Likewise.
1028 (canon_address): Return the offset as a poly_int64 rather than
1029 a HOST_WIDE_INT. Use strip_offset_and_add.
1030 (set_all_positions_unneeded, any_positions_needed_p): Use
1031 positions_needed.large to track stores with non-constant widths.
1032 (all_positions_needed_p): Likewise. Take the offset and width
1033 as poly_int64s rather than ints. Assert that rhs is nonnull.
1034 (record_store): Cope with non-constant offsets and widths.
1035 Nullify the rhs of an earlier store if we can't tell which bytes
1037 (find_shift_sequence): Take the access_size and shift as poly_int64s
1039 (get_stored_val): Take the read_offset and read_width as poly_int64s
1040 rather than HOST_WIDE_INTs.
1041 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1042 non-constant offsets and widths.
1044 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1045 Alan Hayward <alan.hayward@arm.com>
1046 David Sherwood <david.sherwood@arm.com>
1048 * inchash.h (inchash::hash::add_poly_int): New function.
1049 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1050 Use poly_int64 rather than HOST_WIDE_INT.
1051 (ao_ref::max_size_known_p): New function.
1052 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1053 rather than HOST_WIDE_INT.
1054 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1055 to temporaries until its interface is adjusted to match.
1056 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1057 (aliasing_component_refs_p, decl_refs_may_alias_p)
1058 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1059 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1060 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1062 * alias.c (ao_ref_from_mem): Likewise.
1063 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1064 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1065 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1066 (maybe_trim_complex_store, maybe_trim_constructor_store)
1067 (live_bytes_read, dse_classify_store): Likewise.
1068 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1069 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1070 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1071 (vn_reference_lookup_3): Likewise.
1072 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1074 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1075 Alan Hayward <alan.hayward@arm.com>
1076 David Sherwood <david.sherwood@arm.com>
1078 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1079 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1080 instead of ranges_overlap_p.
1082 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1083 Alan Hayward <alan.hayward@arm.com>
1084 David Sherwood <david.sherwood@arm.com>
1086 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1087 sizes as poly_int64s rather than HOST_WIDE_INTs.
1089 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1090 Alan Hayward <alan.hayward@arm.com>
1091 David Sherwood <david.sherwood@arm.com>
1093 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1094 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1095 * gimple-fold.c (fold_ctor_reference): Likewise.
1097 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1098 Alan Hayward <alan.hayward@arm.com>
1099 David Sherwood <david.sherwood@arm.com>
1101 * target.def (dwarf_poly_indeterminate_value): New hook.
1102 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1103 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1104 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1105 * doc/tm.texi: Regenerate.
1106 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1107 offset as a poly_int64.
1108 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1109 offset as a poly_int64.
1110 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1111 Take the offset as a poly_int64.
1112 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1113 (frame_pointer_fb_offset): Change to a poly_int64.
1114 (int_loc_descriptor): Take the offset as a poly_int64. Use
1115 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1116 (based_loc_descr): Take the offset as a poly_int64.
1117 Use strip_offset_and_add to handle (plus X (const)).
1118 Use new_reg_loc_descr instead of an open-coded version of the
1119 previous implementation.
1120 (mem_loc_descriptor): Handle CONST_POLY_INT.
1121 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1122 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1124 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1125 Alan Hayward <alan.hayward@arm.com>
1126 David Sherwood <david.sherwood@arm.com>
1128 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1129 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1130 * inchash.h (inchash::hash::add_poly_hwi): New function.
1131 * gengtype.c (main): Register poly_int64.
1132 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1133 offset as a poly_int.
1134 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1135 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1136 offset as a poly_int64.
1137 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1138 * print-rtl.c (print_poly_int): New function.
1139 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1141 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1143 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1144 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1145 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1146 Enforce track_offset_p for parts of a PARALLEL.
1147 (vt_add_function_parameter): Use const_offset for the final
1148 offset to track. Use get_tracked_reg_offset for the parts
1151 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1152 Alan Hayward <alan.hayward@arm.com>
1153 David Sherwood <david.sherwood@arm.com>
1155 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1156 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1157 * doc/tm.texi: Regenerate.
1158 * hooks.h (hook_bool_uint_uint_true): Delete.
1159 (hook_bool_puint64_puint64_true): Declare.
1160 * hooks.c (hook_bool_uint_uint_true): Delete.
1161 (hook_bool_puint64_puint64_true): New function.
1162 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1163 instead of unsigned ints.
1164 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1165 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1167 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1168 Alan Hayward <alan.hayward@arm.com>
1169 David Sherwood <david.sherwood@arm.com>
1171 * optabs.h (expand_operand): Add an int_value field.
1172 (create_expand_operand): Add an int_value parameter and use it
1173 to initialize the new expand_operand field.
1174 (create_integer_operand): Replace with a declaration of a function
1175 that accepts poly_int64s. Move the implementation to...
1176 * optabs.c (create_integer_operand): ...here.
1177 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1178 the mode preserves the value of int_value, instead of calling
1179 const_int_operand on the rtx. Use gen_int_mode to generate
1182 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1183 Alan Hayward <alan.hayward@arm.com>
1184 David Sherwood <david.sherwood@arm.com>
1186 * dumpfile.h (dump_dec): Declare.
1187 * dumpfile.c (dump_dec): New function.
1188 * pretty-print.h (pp_wide_integer): Turn into a function and
1189 declare a poly_int version.
1190 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1192 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1193 Alan Hayward <alan.hayward@arm.com>
1194 David Sherwood <david.sherwood@arm.com>
1196 * doc/generic.texi (POLY_INT_CST): Document.
1197 * tree.def (POLY_INT_CST): New tree code.
1198 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1199 * tree-core.h (tree_poly_int_cst): New struct.
1200 (tree_node): Add a poly_int_cst field.
1201 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1202 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1203 instead of a wide_int_ref.
1204 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1206 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1207 instead of an unsigned HOST_WIDE_INT.
1208 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1209 (ptrdiff_tree_p): Declare.
1210 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1211 extern inline implementations if the target doesn't use POLY_INT_CST.
1212 (poly_int_tree_p): New function.
1213 (wi::unextended_tree): New class.
1214 (wi::int_traits <unextended_tree>): New override.
1215 (wi::extended_tree): Add a default constructor.
1216 (wi::extended_tree::get_tree): New function.
1217 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1218 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1219 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1220 (wi::tree_to_poly_wide_ref): New typedefs.
1221 (wi::ints_for): Provide overloads for extended_tree and
1223 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1224 (wi::to_wide): New functions.
1225 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1226 * tree.c (poly_int_cst_hasher): New struct.
1227 (poly_int_cst_hash_table): New variable.
1228 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1229 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1231 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1232 (init_ttree): Initialize poly_int_cst_hash_table.
1233 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1234 a poly_int64 instead of a HOST_WIDE_INT.
1235 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1236 instead of an unsigned HOST_WIDE_INT.
1237 (wide_int_to_tree): Rename to...
1238 (wide_int_to_tree_1): ...this.
1239 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1240 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1241 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1242 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1243 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1244 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1246 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1247 (streamer_read_tree_body): Likewise.
1248 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1249 (streamer_write_tree_body): Likewise.
1250 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1251 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1252 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1253 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1254 * gimple-expr.h (is_gimple_constant): Likewise.
1255 * gimplify.c (maybe_with_size_expr): Likewise.
1256 * print-tree.c (print_node): Likewise.
1257 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1258 * tree-pretty-print.c (dump_generic_node): Likewise.
1259 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1260 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1261 * tree-vrp.c (compare_values_warnv): Likewise.
1262 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1263 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1264 (force_expr_to_var_cost): Likewise.
1265 * tree-ssa-loop.c (for_each_index): Likewise.
1266 * fold-const.h (build_invariant_address, size_int_kind): Take a
1267 poly_int64 instead of a HOST_WIDE_INT.
1268 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1269 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1271 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1272 (int_const_binop_2): New function, split out from...
1273 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1274 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1275 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1276 * gimple-ssa-strength-reduction.c (slsr_process_add)
1277 (slsr_process_mul): Check for INTEGER_CSTs before using them
1279 * stor-layout.c (bits_from_bytes): New function.
1280 (bit_from_pos): Use it.
1281 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1282 by BITS_PER_UNIT to get the TYPE_SIZE.
1283 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1284 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1286 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1287 Alan Hayward <alan.hayward@arm.com>
1288 David Sherwood <david.sherwood@arm.com>
1290 * doc/rtl.texi (const_poly_int): Document. Also document the
1291 rtl sharing behavior.
1292 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1293 * rtl.h (const_poly_int_def): New struct.
1294 (rtx_def::u): Add a cpi field.
1295 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1296 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1297 (wi::rtx_to_poly_wide_ref): New typedef
1298 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1299 (poly_int_rtx_p): New functions.
1300 (trunc_int_for_mode): Declare a poly_int64 version.
1301 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1302 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1304 (strip_offset): Declare.
1305 (strip_offset_and_add): New function.
1306 * rtl.def (CONST_POLY_INT): New rtx code.
1307 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1308 (shared_const_p): Use poly_int_rtx_p.
1309 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1311 (gen_int_shift_amount): Likewise.
1312 * emit-rtl.c (const_poly_int_hasher): New class.
1313 (const_poly_int_htab): New variable.
1314 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1315 (const_poly_int_hasher::hash): New function.
1316 (const_poly_int_hasher::equal): Likewise.
1317 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1318 (immed_wide_int_const): Rename to...
1319 (immed_wide_int_const_1): ...this and make static.
1320 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1321 instead of a wide_int_ref.
1322 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1323 (gen_lowpart_common): Handle CONST_POLY_INT.
1324 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1325 * cselib.c (cselib_hash_rtx): Likewise.
1326 * dwarf2out.c (const_ok_for_output_1): Likewise.
1327 * expr.c (convert_modes): Likewise.
1328 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1329 * rtlhash.c (add_rtx): Likewise.
1330 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1331 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1332 Handle existing CONST_POLY_INT rtxes.
1333 * expmed.h (expand_shift): Take a poly_int64 instead of a
1335 * expmed.c (expand_shift): Likewise.
1336 * rtlanal.c (strip_offset): New function.
1337 (commutative_operand_precedence): Give CONST_POLY_INT the same
1338 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1340 * rtl-tests.c (const_poly_int_tests): New struct.
1341 (rtl_tests_c_tests): Use it.
1342 * simplify-rtx.c (simplify_const_unary_operation): Handle
1344 (simplify_const_binary_operation): Likewise.
1345 (simplify_binary_operation_1): Fold additions of symbolic constants
1346 and CONST_POLY_INTs.
1347 (simplify_subreg): Handle extensions and truncations of
1349 (simplify_const_poly_int_tests): New struct.
1350 (simplify_rtx_c_tests): Use it.
1351 * wide-int.h (storage_ref): Add default constructor.
1352 (wide_int_ref_storage): Likewise.
1353 (trailing_wide_ints): Use GTY((user)).
1354 (trailing_wide_ints::operator[]): Add a const version.
1355 (trailing_wide_ints::get_precision): New function.
1356 (trailing_wide_ints::extra_size): Likewise.
1358 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1359 Alan Hayward <alan.hayward@arm.com>
1360 David Sherwood <david.sherwood@arm.com>
1362 * emit-rtl.h (gen_int_shift_amount): Declare.
1363 * emit-rtl.c (gen_int_shift_amount): New function.
1364 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1366 * calls.c (shift_return_value): Likewise.
1367 * cse.c (fold_rtx): Likewise.
1368 * dse.c (find_shift_sequence): Likewise.
1369 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1370 (expand_shift, expand_smod_pow2): Likewise.
1371 * lower-subreg.c (shift_cost): Likewise.
1372 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1373 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1374 (expand_vec_perm_var): Likewise.
1375 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1376 (simplify_binary_operation_1): Likewise.
1377 * combine.c (try_combine, find_split_point, force_int_to_mode)
1378 (simplify_shift_const_1, simplify_shift_const): Likewise.
1379 (change_zero_ext): Likewise. Use simplify_gen_binary.
1381 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1383 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1385 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1387 * doc/invoke.texi (ARM Options): Document accepted extension options
1388 for -march=armv8.3-a.
1390 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1393 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1394 or --with-float=softfp, set the default CPU to arm10e.
1396 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1398 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1399 * config/visium/predicates.md (const_shift_operand): Likewise.
1400 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1401 (visium_legitimize_reload_address): Likewise.
1403 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1406 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1408 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1411 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1413 2017-12-20 Julia Koval <julia.koval@intel.com>
1415 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1416 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1417 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1418 * config.gcc: Include vpclmulqdqintrin.h.
1419 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1420 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1422 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1423 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1424 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1425 * config/i386/i386.c (isa2_opts): Add -mcx16.
1426 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1427 (ix86_option_override_internal): Move mcx16 to flags2.
1428 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1429 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1430 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1431 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1432 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1433 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1434 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1435 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1436 * doc/invoke.texi: Add -mvpclmulqdq.
1438 2017-12-20 Tom de Vries <tom@codesourcery.com>
1441 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1442 * calls.c (rtx_for_static_chain): ... here. New function.
1443 * calls.h (rtx_for_static_chain): Declare.
1444 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1445 instead of targetm.calls.static_chain.
1446 * df-scan.c (df_get_entry_block_def_set): Same.
1448 2017-12-19 Tom de Vries <tom@codesourcery.com>
1450 PR tree-optimization/83493
1451 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1454 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1456 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1457 inform with hardcoded english plural handling.
1459 2017-12-18 Jeff Law <law@redhat.com>
1461 PR tree-optimization/83477
1462 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1463 a non-virtual PHI, always push a new range.
1465 2017-12-19 Martin Sebor <msebor@redhat.com>
1468 * builtins.c (compute_objsize): Handle non-constant offsets.
1470 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1472 PR tree-optimization/83444
1473 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1474 character load case, if get_stridx on MEM_REF's operand doesn't
1475 look usable, retry with get_addr_stridx.
1477 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1480 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1481 (variable_tracking_main_1): Keep markers even when VTA fails.
1484 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1485 even if there are markers before it.
1486 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1488 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1490 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1491 typical order conditions.
1492 * sel-sched.c (extract_new_fences_from): Likewise.
1493 * config/visium/constraints.md (J, K, L): Likewise.
1494 * config/visium/predicates.md (const_shift_operand): Likewise.
1495 * config/visium/visium.c (visium_legitimize_address,
1496 visium_legitimize_reload_address): Likewise.
1497 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1498 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1499 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1500 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1501 SET_DUMP_DETAIL): Likewise.
1502 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1503 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1504 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1505 avr_set_core_architecture, avr_set_current_function,
1506 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1507 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1508 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1509 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1510 avr_map_decompose, avr_fold_builtin): Likewise.
1511 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1512 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1513 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1514 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1515 * config/m32c/m32c.c (m32c_conditional_register_usage,
1516 m32c_address_cost): Likewise.
1517 * config/m32c/predicates.md (shiftcount_operand,
1518 longshiftcount_operand): Likewise.
1519 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1520 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1521 can_use_cdx_ldstw): Likewise.
1522 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1523 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1525 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1526 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1527 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1528 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1529 * config/vax/vax.c (adjacent_operands_p): Likewise.
1530 * config/ft32/constraints.md (L, b, KA): Likewise.
1531 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1533 * cfgexpand.c (expand_stack_alignment): Likewise.
1534 * gcse.c (insert_expr_in_table): Likewise.
1535 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1536 * cgraphunit.c (cgraph_node::expand): Likewise.
1537 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1538 * emit-rtl.c (add_insn): Likewise.
1539 * input.c (dump_location_info): Likewise.
1540 * passes.c (NEXT_PASS): Likewise.
1541 * read-rtl-function.c (parse_note_insn_name,
1542 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1544 * sched-rgn.c (sched_rgn_init): Likewise.
1545 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1546 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1547 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1548 * explow.c (eliminate_constant_term): Likewise.
1549 * final.c (leaf_renumber_regs_insn): Likewise.
1550 * cfgrtl.c (print_rtl_with_bb): Likewise.
1551 * genhooks.c (emit_init_macros): Likewise.
1552 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1553 * tree-data-ref.c (conflict_fn): Likewise.
1554 * selftest.c (assert_streq): Likewise.
1555 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1556 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1557 fold_binary_loc, multiple_of_p): Likewise.
1558 * reload.c (push_reload, find_equiv_reg): Likewise.
1559 * et-forest.c (et_nca, et_below): Likewise.
1560 * dbxout.c (dbxout_symbol_location): Likewise.
1561 * reorg.c (relax_delay_slots): Likewise.
1562 * dojump.c (do_compare_rtx_and_jump): Likewise.
1563 * gengtype-parse.c (type): Likewise.
1564 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1565 simplify_const_relational_operation): Likewise.
1566 * reload1.c (do_output_reload): Likewise.
1567 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1568 * gengtype.c (type_for_name): Likewise.
1569 * gimple-ssa-sprintf.c (format_directive): Likewise.
1571 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1574 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1575 accessing it. Adjust comment.
1577 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1580 * predict.c (zero_one_minusone): New function.
1581 (apply_return_prediction): Avoid return prediction for functions
1582 returning only -1, 0 and 1 values, unless they only return -1 and 0
1585 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1587 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1588 fall-through warning.
1589 (arc_compute_frame_size): Remove unused variables.
1590 (arc_print_operand): Fix fprintif format.
1591 (arc_can_follow_jump): Clean fall-through warning.
1593 2017-12-19 Marek Polacek <polacek@redhat.com>
1596 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1599 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1601 * config/arc/arc.c (overriderregs): New variable.
1602 (arc_override_options): Track fixed/call saved/call options.
1603 (arc_conditional_register_usage): Check against overrideregs
1604 variable whenever we change register properties.
1606 2017-12-19 Nathan Sidwell <nathan@acm.org>
1608 * opts.c (finish_options): Don't prefix dump_base_name if it
1609 already contains directories.
1611 2017-12-19 Martin Liska <mliska@suse.cz>
1613 PR rtl-optimization/82675
1614 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1615 more element in sbitmap.
1617 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1619 * gimplify.c (gimplify_expr): Use error_operand_p.
1621 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1624 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1627 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1629 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1630 the jump is to a label.
1632 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1634 PR tree-optimization/83444
1635 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1640 * ipa-inline.c (flatten_remove_node_hook): New function.
1641 (ipa_inline): Keep only nodes with flatten attribute at the end of
1642 the array in the order from ipa_reverse_postorder, only walk that
1643 portion of array for flattening, if there is more than one such
1644 node, temporarily register a removal hook and ignore removed nodes.
1646 PR tree-optimization/80631
1647 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1648 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1649 IFN_REDUC_MAX or IFN_REDUC_MIN.
1651 2017-12-18 Jeff Law <law@redhat.com>
1653 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1654 record symbolic equivalences from backedges in the CFG.
1657 2017-11-19 Jeff Law <law@redhat.com>
1659 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1660 of degenerates resulting from ignoring an edge.
1662 2017-12-18 Martin Sebor <msebor@redhat.com>
1665 PR tree-optimization/78450
1666 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1667 (handle_builtin_strlen): Call it.
1669 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1671 PR rtl-optimization/83424
1672 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1674 2017-12-18 Marek Polacek <polacek@redhat.com>
1677 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1678 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1681 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1683 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1684 (umaddsidi4, umaddsidi_split): Likewise.
1686 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1688 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1691 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1694 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1696 2017-12-18 Richard Biener <rguenther@suse.de>
1698 PR tree-optimization/81877
1699 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1700 (outermost_indep_loop): Adjust.
1701 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1702 (can_sm_ref_p): Adjust.
1704 2017-12-18 Richard Biener <rguenther@suse.de>
1707 * tree.c (array_at_struct_end_p): Return true if the underlying
1708 object has space for at least one element in excess of what
1709 the array domain specifies.
1711 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1713 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1714 example to avoid overfull hbox.
1715 * doc/invoke.texi (Option Summary): Add missing @gol.
1716 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1719 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1720 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1722 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1723 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1724 (AArch64 Options): Document -mverbose-cost-dump.
1725 (ARM Options): Likewise, plus -mflip-thumb.
1727 2017-12-17 Martin Sebor <msebor@redhat.com>
1730 * gimple-ssa-warn-restrict.c
1731 (builtin_memref::offset_out_of_bounds): Correct the handling of
1734 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1736 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1738 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1740 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1742 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1744 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1747 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1749 * doc/invoke.texi: Fix some typos.
1751 2017-12-16 Martin Sebor <msebor@redhat.com>
1753 PR tree-optimization/78918
1754 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1755 * builtins.c (check_sizes): Rename...
1756 (check_access): ...to this. Rename function arguments for clarity.
1757 (check_memop_sizes): Adjust names.
1758 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1759 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1760 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1761 (check_strncat_sizes, expand_builtin_strncat): Same.
1762 (expand_builtin_strncpy, expand_builtin_memset): Same.
1763 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1764 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1765 (maybe_emit_sprintf_chk_warning): Same.
1766 (expand_builtin_strcpy): Adjust.
1767 (expand_builtin_stpcpy): Same.
1768 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1769 in pointer-checking forms of memcpy, memmove, and mempcpy.
1770 (gcall_to_tree_minimal, max_object_size): Define new functions.
1771 * builtins.h (max_object_size): Declare.
1772 * calls.c (alloc_max_size): Call max_object_size instead of
1773 hardcoding ssizetype limit.
1774 (get_size_range): Handle new argument.
1775 * calls.h (get_size_range): Add a new argument.
1776 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1777 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1778 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1780 (gimple_fold_builtin_memory_chk): Same.
1781 (gimple_fold_builtin_stxcpy_chk): New function.
1782 * gimple-ssa-warn-restrict.c: New source.
1783 * gimple-ssa-warn-restrict.h: New header.
1784 * gimple.c (gimple_build_call_from_tree): Propagate location.
1785 * passes.def (pass_warn_restrict): Add new pass.
1786 * tree-pass.h (make_pass_warn_restrict): Declare.
1787 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1789 (handle_builtin_strcat): Same.
1790 (strlen_optimize_stmt): Rename...
1791 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1792 stpncpy, strncpy, and their checking forms.
1794 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1796 PR rtl-optimization/82849
1797 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1798 and get_max_loop_iterations_int.
1800 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1801 Alan Hayward <alan.hayward@arm.com>
1802 David Sherwood <david.sherwood@arm.com>
1804 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1805 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1806 size as a poly_uint64.
1807 (mode_for_vector, mode_for_int_vector): Take the number of vector
1808 elements as a poly_uint64.
1809 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1810 size as a poly_uint64.
1811 (mode_for_vector, mode_for_int_vector): Take the number of vector
1812 elements as a poly_uint64.
1814 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1815 Alan Hayward <alan.hayward@arm.com>
1816 David Sherwood <david.sherwood@arm.com>
1818 * machmode.h (MACRO_MODE): New macro.
1819 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1820 * targhooks.c (default_libcall_value, default_secondary_reload)
1821 (default_memory_move_cost, default_register_move_cost)
1822 (default_class_max_nregs): Likewise.
1824 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1825 Alan Hayward <alan.hayward@arm.com>
1826 David Sherwood <david.sherwood@arm.com>
1828 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1829 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1831 * genautomata.c (main): Likewise.
1832 * genconditions.c (write_header): Likewise.
1833 * genemit.c (main): Likewise.
1834 * genextract.c (print_header): Likewise.
1835 * genopinit.c (main): Likewise.
1836 * genoutput.c (output_prologue): Likewise.
1837 * genpeep.c (main): Likewise.
1838 * genpreds.c (write_insn_preds_c): Likewise.
1839 * genrecog.c (writer_header): Likewise.
1840 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1841 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1842 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1843 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1844 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1845 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1846 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1847 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1848 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1849 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1850 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1851 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1852 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1853 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1854 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1855 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1856 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1857 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1858 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1859 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1860 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1861 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1862 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1863 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1864 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1865 * config/darwin.c (IN_TARGET_CODE): Likewise.
1866 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1867 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1868 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1869 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1870 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1871 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1872 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1873 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1874 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1875 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1876 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1877 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1878 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1879 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1880 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1881 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1882 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1883 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1884 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1885 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1886 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1887 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1888 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1889 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1890 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1891 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1892 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1893 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1894 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1895 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1896 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1897 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1898 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1899 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1900 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1901 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1902 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1903 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1904 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1905 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1906 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1907 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1908 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1909 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1910 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1911 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1912 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1913 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1914 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1915 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1916 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1917 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1918 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1919 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1920 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1921 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1922 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1923 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1924 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1925 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1926 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1927 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1928 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1929 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1930 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1931 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1932 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1933 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1934 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1935 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1936 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1937 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1938 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1939 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1940 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1941 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1942 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1943 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1944 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1945 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1946 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1947 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1948 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1949 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1950 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1951 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1952 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1953 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1954 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1955 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1956 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1957 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1958 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1959 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1960 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1961 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1962 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1963 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1964 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1965 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1966 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1967 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1968 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1969 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1970 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1971 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1973 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1974 Alan Hayward <alan.hayward@arm.com>
1975 David Sherwood <david.sherwood@arm.com>
1977 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1980 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1981 Alan Hayward <alan.hayward@arm.com>
1982 David Sherwood <david.sherwood@arm.com>
1984 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1985 * doc/md.texi (vec_series@var{m}): Document.
1986 * tree.def (VEC_SERIES_EXPR): New tree code.
1987 * tree.h (build_vec_series): Declare.
1988 * tree.c (build_vec_series): New function.
1989 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1990 * tree-pretty-print.c (dump_generic_node): Likewise.
1991 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1992 * tree-inline.c (estimate_operator_cost): Likewise.
1993 * expr.c (expand_expr_real_2): Likewise.
1994 * optabs-tree.c (optab_for_tree_code): Likewise.
1995 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1996 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1997 * expmed.c (make_tree): Handle VEC_SERIES.
1998 * optabs.def (vec_series_optab): New optab.
1999 * optabs.h (expand_vec_series_expr): Declare.
2000 * optabs.c (expand_vec_series_expr): New function.
2001 * tree-vect-generic.c (expand_vector_operations_1): Check that
2002 the operands also have vector type.
2004 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2005 Alan Hayward <alan.hawyard@arm.com>
2006 David Sherwood <david.sherwood@arm.com>
2008 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2009 (VEC_COND_EXPR): Add missing @tindex.
2010 * doc/md.texi (vec_duplicate@var{m}): Document.
2011 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2012 * tree.c (build_vector_from_val): Add stubbed-out handling of
2013 variable-length vectors, using VEC_DUPLICATE_EXPR.
2014 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2015 * cfgexpand.c (expand_debug_expr): Likewise.
2016 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2017 * tree-inline.c (estimate_operator_cost): Likewise.
2018 * tree-pretty-print.c (dump_generic_node): Likewise.
2019 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2020 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2021 (test_vec_duplicate_folding): New function.
2022 (fold_const_c_tests): Call it.
2023 * optabs.def (vec_duplicate_optab): New optab.
2024 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2025 * optabs.h (expand_vector_broadcast): Declare.
2026 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2027 vec_duplicate_optab.
2028 * expr.c (store_constructor): Try using vec_duplicate_optab for
2030 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2032 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2035 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2036 div/mod latencies a bit.
2038 2017-12-15 Jeff Law <law@redhat.com>
2040 PR tree-optimization/36550
2041 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2042 (mark_threaded_blocks): Rewrite code to avoid block copying when
2043 optimizing for size. Don't pessimize blocks which will be
2044 copied, but all the statements will be dead.
2046 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2048 PR tree-optimization/81165
2049 * tree-ssa-threadupdate.c (uses_in_bb): New.
2050 (estimate_threading_killed_stmts): New.
2051 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2052 * tree-ssa-threadedge.c
2053 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2056 2017-12-15 Jeff Law <law@redhat.com>
2058 PR tree-optimization/83410
2059 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2060 threads when parallelizing loops.
2062 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2064 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2066 * config/alpha/alpha.c (vms_attribute_table): Swap
2067 affects_type_identity and handler fields, adjust comments.
2068 * config/mips/mips.c (mips_attribute_table): Likewise.
2069 * config/visium/visium.c (visium_attribute_table): Likewise.
2070 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2071 * config/microblaze/microblaze.c (microblaze_attribute_table):
2073 * config/spu/spu.c (spu_attribute_table): Likewise.
2074 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2075 * config/arc/arc.c (arc_attribute_table): Likewise.
2076 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2077 * config/v850/v850.c (v850_handle_interrupt_attribute,
2078 v850_handle_data_area_attribute): Formatting fixes.
2079 (v850_attribute_table): Swap affects_type_identity and handler
2080 fields, adjust comments.
2081 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2082 * config/arm/arm.c (arm_attribute_table): Likewise.
2083 * config/avr/avr.c (avr_attribute_table): Likewise.
2084 * config/s390/s390.c (s390_attribute_table): Likewise.
2085 * config/sh/sh.c (sh_attribute_table): Likewise.
2086 * config/i386/i386.c (ix86_handle_cconv_attribute,
2087 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2088 (ix86_attribute_table): Swap affects_type_identity and handler
2089 fields, adjust comments.
2090 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2091 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2092 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2093 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2094 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2095 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2096 * config/rx/rx.c (rx_attribute_table): Likewise.
2097 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2098 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2099 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2100 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2101 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2102 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2103 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2104 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2105 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2106 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2107 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2109 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2111 (register_scoped_attributes, decl_attributes): Formatting fixes.
2113 PR tree-optimization/83269
2114 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2115 subtraction in arg0's type if type is signed and arg0 is unsigned.
2119 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2120 convert? on @0 instead of convert. Check type of @1, not @0.
2121 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2122 convert? on @0 instead of convert. Check type of @1, not @0.
2123 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2124 only optimize if either both @1 and @2 types are narrower
2125 precision, or both are wider or equal precision, and in the former
2126 case only if both have undefined overflow.
2128 2017-12-15 Richard Biener <rguenther@suse.de>
2131 * internal-fn.def (IFN_NOP): Add.
2132 * internal-fn.c (expand_NOP): Do nothing.
2133 * lto-streamer-in.c (input_function): Instead of removing
2134 sanitizer calls replace them with IFN_NOP calls.
2136 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2137 Alan Hayward <alan.hayward@arm.com>
2138 David Sherwood <david.sherwood@arm.com>
2140 * dse.c (store_info, read_info_type): Replace begin and end with
2142 (print_range): New function.
2143 (set_all_positions_unneeded, any_positions_needed_p)
2144 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2146 (record_store): Likewise. Optimize the case in which all positions
2148 (get_stored_val): Replace read_begin and read_end with read_offset
2150 (replace_read): Update call accordingly.
2152 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2154 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2155 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2156 (loop_cand::loop_cand): Initialize above members.
2157 (loop_cand::supported_operations): Delete.
2158 (loop_cand::can_interchange_p): Inline above function.
2159 (loop_cand::classify_simple_reduction): Record number of constant
2160 initialized simple reductions.
2161 (should_interchange_loops): New parameters. Check stmt cost of loops
2163 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2164 Update call to should_interchange_loops.
2165 (should_interchange_loop_nest): Update call to
2166 should_interchange_loops.
2168 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2171 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2172 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2173 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2175 2017-12-15 Julia Koval <julia.koval@intel.com>
2177 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2178 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2179 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2180 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2181 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2183 2017-12-15 Julia Koval <julia.koval@intel.com>
2185 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2186 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2187 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2188 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2189 _mm_aesenc_epi128): New intrinsics.
2191 2017-12-15 Julia Koval <julia.koval@intel.com>
2193 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2194 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2195 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2196 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2197 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2199 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2201 * gimple-ssa-strength-reduction.c (analyze_increments):
2202 Distinguish replacement costs for constant strides from those for
2205 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2207 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2209 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2211 * doc/invoke.texi: Document -Wcast-function-type.
2212 * recog.h (stored_funcptr): Change signature.
2213 * tree-dump.c (dump_node): Avoid warning.
2214 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2216 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2219 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2222 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2224 PR tree-optimization/83312
2225 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2226 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2227 "val" param, and to cope with arbitrary basic blocks.
2228 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2229 handle NULL_TREE for "val", dropping "bb" param.
2230 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2231 drop "bb" param. Handle NULL_TREE for "val".
2232 (find_case_label_for_value): Make "switch_stmt" param const.
2233 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2235 (vrp_prop::check_all_array_refs): Reimplement as...
2236 (check_array_bounds_dom_walker::before_dom_children): ...this new
2237 vfunc. Replace linear search through BB block list, excluding
2238 those with non-executable in-edges via dominator walk.
2240 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2242 * config/arm/arm.opt (mverbose-cost-dump): New option.
2243 * config/arm/arm.c (arm_rtx_costs): Use it.
2245 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2248 * reload1.c (emit_input_reload_insns): Skip debug markers.
2250 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2252 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2256 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2257 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2258 * config/sh/sh.c: Skip debug insns besides notes.
2259 * config/sh/sh.md: Likewise.
2260 * config/sh/sh_treg_combine.cc: Likewise.
2261 * config/sh/sync.md: Likewise.
2263 2017-12-14 Tom de Vries <tom@codesourcery.com>
2265 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2266 Add item for weak_undefined.
2268 2017-12-14 Richard Biener <rguenther@suse.de>
2270 PR tree-optimization/67842
2271 * sese.h (bb_in_region): Remove #if 0'ed code.
2273 2017-12-14 Richard Biener <rguenther@suse.de>
2275 PR tree-optimization/83326
2276 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2277 may_be_zero parameter and handle it by not marking the first
2278 peeled copy as not exiting the loop.
2279 (try_peel_loop): Likewise.
2280 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2281 to handle the case of constant or zero iterations and perform
2282 loop header copying on-the-fly.
2284 2017-12-14 Richard Biener <rguenther@suse.de>
2286 PR tree-optimization/83418
2287 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2288 Instead of asserting we don't get unfolded comparisons deal with
2291 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2294 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2295 basic blocks. Assert debug bind insns don't appear outside of bbs,
2296 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2299 PR tree-optimization/83198
2300 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2301 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2302 value if arg is a REAL_CST with incompatible type.
2304 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2305 Bin Cheng <bin.cheng@arm.com>
2308 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2310 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2313 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2315 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2316 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2318 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2319 Alan Hayward <alan.hayward@arm.com>
2320 David Sherwood <david.sherwood@arm.com>
2322 * poly-int.h: New file.
2323 * poly-int-types.h: Likewise.
2324 * coretypes.h: Include them.
2325 (POLY_INT_CONVERSION): Define.
2326 * target.def (estimated_poly_value): New hook.
2327 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2328 * doc/tm.texi: Regenerate.
2329 * doc/poly-int.texi: New file.
2330 * doc/gccint.texi: Include it.
2331 * doc/rtl.texi: Describe restrictions on subreg modes.
2332 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2333 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2334 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2335 * targhooks.h (default_estimated_poly_value): Declare.
2336 * targhooks.c (default_estimated_poly_value): New function.
2337 * target.h (estimated_poly_value): Likewise.
2338 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2339 (wi::unary_traits): Delete.
2340 (wi::binary_traits::signed_shift_result_type): Define for
2341 offset_int << HOST_WIDE_INT, etc.
2342 (generic_wide_int::operator <<=): Define for all types and use
2343 wi::lshift instead of <<.
2344 (wi::hwi_with_prec): Add a default constructor.
2345 (wi::ints_for): New class.
2346 (operator <<): Define for all wide-int types.
2347 (operator /): New function.
2348 (operator %): Likewise.
2349 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2350 (ASSERT_MAYBE_NE_AT): New macros.
2352 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2353 Dominik Vogt <vogt@linux.vnet.ibm.com>
2356 * emit-rtl.c (init_emit): Remove ??? comment.
2357 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2358 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2359 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2360 alignment of 3 virtual registers to BITS_PER_WORD.
2362 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2364 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2366 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2367 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2369 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2372 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2374 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2376 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2377 with the exception of debug begin stmt markers appear before
2381 * final.c (rest_of_handle_final): Call variable_tracking_main only
2382 if !flag_var_tracking.
2384 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2385 Jakub Jelinek <jakub@redhat.com>
2389 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2390 labels debug stmts that can only appear after labels.
2392 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2394 PR rtl-optimization/82398
2395 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2396 EXPR_USEFULNESS in priority comparison.
2398 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2400 PR rtl-optimization/83393
2401 * combine.c (move_deaths): If reg_stat points to a too new insn in
2402 last_death, do not use it: find the proper insn instead.
2404 2017-12-12 Jeff Law <law@redhat.com>
2406 PR tree-optimization/83298
2407 PR tree-optimization/83362
2408 PR tree-optimization/83383
2409 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2410 push_value_range a public interface. Add new argument to
2411 record_ranges_from_stmt.
2412 * gimple-ssa-evrp-analyze.c
2413 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2414 Update comments. Handle recording temporary equivalences.
2415 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2416 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2417 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2418 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2419 gimple-ssa-evrp-analyze.h.
2420 (record_temporary_equivalences_from_phis): Add new argument. When
2421 the PHI arg is an SSA_NAME, set the result's range to the range
2423 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2424 from statements too.
2425 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2426 Pass it down to children as needed.
2427 (thread_outgoing_edges): Likewise.
2428 (thread_across_edge): Likewise. Push/pop range state as needed.
2429 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2431 2017-12-12 Julia Koval <julia.koval@intel.com>
2433 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2434 (PTA_CANNONLAKE): Remove PTA_CLWB.
2436 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2439 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2440 (vcondv2div2df): Likewise.
2441 (vconduv2dfv2di): Likewise.
2443 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2445 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2446 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2447 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2448 ATTR_MATHFN_FPROUNDING_ERRNO.
2450 2017-12-12 Richard Biener <rguenther@suse.de>
2452 PR tree-optimization/83385
2453 * graphite-scop-detection.c (get_order, order): Remove.
2454 (bb_to_rpo): New global.
2456 (build_scops): Sort pbbs in RPO order.
2458 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2460 * combine.c (simplify_set): Do not transform subregs to zero_extends
2461 if the destination mode is a vector mode.
2463 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2465 PR tree-optimization/83359
2466 * tree-cfg.h (fold_loop_internal_call): Declare.
2467 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2468 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2469 (find_loop_dist_alias): New function.
2470 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2471 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2474 PR tree-optimization/80631
2475 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2476 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2477 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2478 hardcoding zero as the value if COND_EXPR is never true. For
2479 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2480 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2481 hardcoding MAX_EXPR as the reduction operation.
2482 (is_nonwrapping_integer_induction): Allow negative step.
2483 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2484 vect_create_epilog_for_reduction, if no value is suitable, don't
2485 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2487 2017-12-12 Richard Biener <rguenther@suse.de>
2489 PR tree-optimization/81889
2490 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2491 range info from the non-wrapping IV instead of just the range
2494 2017-12-12 Julia Koval <julia.koval@intel.com>
2496 * config.gcc: Add vaesintrin.h.
2497 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2499 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2500 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2502 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2503 * config/i386/immintrin.h: Include vaesintrin.h.
2504 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2505 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2506 _mm_aesdec_epi128): New intrinsics.
2508 2017-12-12 Julia Koval <julia.koval@intel.com>
2510 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2511 OPTION_MASK_ISA_VAES_UNSET): New.
2512 (ix86_handle_option): Handle -mvaes.
2513 * config/i386/cpuid.h: Define bit_VAES.
2514 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2515 * config/i386/i386-c.c (__VAES__): New.
2516 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2517 (ix86_valid_target_attribute_inner_p): Ditto.
2518 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2519 * config/i386/i386.opt: Add -mvaes.
2520 * doc/invoke.texi: Ditto.
2522 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2524 * debug.h (gcc_debug_hooks): Add inline_entry.
2525 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2526 * debug.c (do_nothing_debug_hooks): Likewise.
2527 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2528 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2529 (dwarf2_lineno_debug_hooks): Likewise.
2531 * common.opt (gstatement-frontiers): New, setting
2532 debug_nonbind_markers_p.
2533 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2534 * toplev.c (process_options): Autodetect value for debug statement
2536 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2537 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2539 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2540 markers. Integrate source bind into debug stmt expand loop.
2541 (pass_expand::execute): Check debug marker limit. Avoid deep
2542 TER and expand debug locations for debug bind insns only.
2543 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2545 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2546 * final.c (reemit_insn_block_notes): Take current block from
2547 nonbind markers. Declare note where it's first set.
2548 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2549 begin stmt markers if enabled.
2550 (notice_source_line): Handle nonbind markers. Fail if their
2551 location is unknown or that of builtins.
2552 (rest_of_handle_final): Convert begin stmt markers to notes if
2553 var-tracking didn't run.
2554 (rest_of_clean_state): Skip begin stmt markers.
2555 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2557 * function.c (allocate_struct_function): Set begin_stmt_markers.
2558 * function.h (struct function): Add debug_marker_count counter
2559 and debug_nonbind_markers flag.
2560 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2561 * gimple-low.c (lower_function_body): Adjust
2562 debug_nonbind_markers.
2563 (lower_stmt): Drop or skip gimple debug stmts.
2564 (lower_try_catch): Skip debug stmts.
2565 * gimple.c (gimple_build_debug_begin_stmt): New.
2566 (gimple_copy): Increment debug_marker_count if copying one.
2567 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2568 * gimplify.c (rexpr_location): New.
2569 (rexpr_has_location): New.
2570 (warn_switch_unreachable_r): Handle gimple debug stmts.
2571 (shortcut_cond_r): Call expr_location.
2573 (find_goto_label): New.
2574 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2576 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2578 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2579 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2580 (LANG_HOOKS_INITIALIZER): ... this.
2581 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2582 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2584 (update_ebb_live_info): Skip debug insn markers.
2585 * lra.c (debug_insn_static_data): Rename to...
2586 (debug_bind_static_data): ... this.
2587 (debug_marker_static_data): New.
2588 (lra_set_insn_recog_data): Select one of the above depending
2590 (lra_update_isn_regno_info): Don't assume debug insns have
2592 (push_insns): Skip debug insns.
2593 * lto-streamer-in.c (input_function): Drop debug stmts
2594 depending on active options. Adjust debug_nonbind_markers.
2595 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2596 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2597 begin stmt marker notes.
2598 (print_insn): Likewise.
2599 * recog.c (extract_insn): Recognize rtl for debug markers.
2600 * rtl.def (DEBUG_MARKER): New.
2601 * tree-inline.c: Include params.h.
2602 (remap_gimple_stmt): Handle nonbind markers.
2603 (maybe_move_debug_stmts_to_successors): Likewise.
2604 (copy_debug_stmt): Likewise.
2605 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2606 markers regardless of no side effects.
2607 (tsi_link_before): Don't update container's side effects when adding
2608 a begin stmt marker.
2609 (tsi_link_after): Likewise.
2610 (expr_first): Skip begin stmt markers.
2611 (expr_last): Likewise.
2612 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2613 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2614 Disregard nonbind markers.
2615 * tree.c (make_node_stat): Don't set side effects for begin stmt
2617 (build1_stat): Likewise.
2618 * tree.def (DEBUG_BEGIN_STMT): New.
2619 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2620 * var-tracking.c (delete_debug_insns): Renamed to...
2621 (delete_vta_debug_insns): ... this.
2622 (reemit_marker_as_note): New.
2623 (vt_initialize): Reemit markers.
2624 (delete_vta_debug_insns): Likewise.
2625 (vt_debug_insns_local): Reemit or delete markers.
2626 (variable_tracking_main_1): Likewise.
2627 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2628 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2629 (gimple_debug_nonbind_marker_p): New.
2630 (gimple_build_debug_bind): Adjust.
2631 (gimple_build_debug_begin_stmt): New.
2632 * doc/invoke.texi (max-debug-marker-count): New param.
2633 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2634 (debug_parameter_ref, debug_marker): New.
2635 (NOTE_INSN_BEGIN_STMT): New.
2636 (DEBUG_INSN): Describe begin stmt markers.
2638 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2639 without debug insns, we wouldn't, but clean up debug insns
2640 after a control flow insn nevertheless.
2642 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2643 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2644 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2645 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2646 (rtl_tidy_fallthru_edge): Likewise.
2647 (rtl_verify_fallthru): Likewise.
2648 (rtl_verify_bb_layout): Likewise.
2649 (skip_insns_after_block): Likewise.
2650 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2651 * dwarf2out.c: Include print-rtl.h.
2652 (dwarf2out_next_real_insn): New.
2653 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2654 Dump debug binds in asm comments.
2655 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2656 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2657 callers to use gsi_start_nondebug_bb instead.
2658 (gsi_after_labels): Skip gimple debug stmts.
2659 (gsi_start_nondebug): New.
2660 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2661 (proper_loop_form_for_interchange): Adjust.
2662 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2663 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2664 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2665 (collect_fallthrough_labels): Likewise.
2666 (should_warn_for_implicit_fallthrough): Likewise.
2667 (warn_implicit_fallthrough_r): Likewise.
2668 (expand_FALLTHROUGH_r): Likewise.
2669 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2670 (graphite_copy_stmts_from_block): Skip nonbind markers.
2671 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2672 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2673 * jump.c (clean_barriers): Skip debug insns.
2674 * omp-expand.c (expand_parallel_call): Skip debug insns.
2675 (expand_task_call): Likewise.
2676 (remove_exit_barrier): Likewise.
2677 (expand_omp_taskreg): Likewise.
2678 (expand_omp_for_init_counts): Likewise.
2679 (expand_omp_for_generic): Likewise.
2680 (expand_omp_for_static_nochunk): Likewise.
2681 (expand_omp_for_static_chunk): Likewise.
2682 (expand_omp_simd): Likewise.
2683 (expand_omp_taskloop_for_outer): Likewise.
2684 (expand_omp_taskloop_for_inner): Likewise.
2685 (expand_oacc_for): Likewise.
2686 (expand_omp_sections): Likewise.
2687 (expand_omp_single): Likewise.
2688 (expand_omp_synch): Likewise.
2689 (expand_omp_atomic_load): Likewise.
2690 (expand_omp_atomic_store): Likewise.
2691 (expand_omp_atomic_fetch_op): Likewise.
2692 (expand_omp_atomic_pipeline): Likewise.
2693 (expand_omp_atomic_mutex): Likewise.
2694 (expand_omp_target): Likewise.
2695 (grid_expand_omp_for_loop): Likewise.
2696 (grid_expand_target_grid_body): Likewise.
2697 (build_omp_regions_1): Likewise.
2698 * omp-low.c (check_combined_parallel): Skip debug stmts.
2699 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2700 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2701 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2703 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2704 (make_edges): Likewise.
2705 (cleanup_dead_labels): Likewise.
2706 (gimple_can_merge_blocks_p): Likewise.
2707 (stmt_starts_bb_p): Likewise.
2708 (gimple_block_label): Likewise.
2709 (gimple_redirect_edge_and_branch): Likewise.
2710 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2712 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2714 * tree-pretty-print (print_declaration): Omit initializer in slim
2716 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2718 (eliminate_unnecessary_stmts): Stabilize block removal order.
2719 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2720 * var-tracking.c (get_first_insn): New.
2721 (vt_emit_notes): Call it.
2722 (vt_initialize): Walk any insns before the first BB.
2723 (delete_debug_insns): Likewise.
2725 * gimple.h (enum gimple_debug_subcode): Add
2726 GIMPLE_DEBUG_BEGIN_STMT.
2727 (gimple_debug_begin_stmt_p): New.
2728 (gimple_debug_nonbind_marker_p): New.
2729 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2730 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2731 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2732 * insn-notes.def (BEGIN_STMT): New.
2733 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2734 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2735 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2736 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2737 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2738 (INSN_DEBUG_MARKER_KIND): New.
2739 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2740 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2741 (INSN_VAR_LOCATION_PTR): New.
2742 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2744 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2745 deps only if debug bind insns are enabled.
2746 (pass_expand::execute): Avoid deep TER and expand
2747 debug locations for debug bind insns only.
2748 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2749 debug stmts special handling down to debug bind stmts.
2750 * combine.c (try_combine): Narrow debug insns special handling
2751 down to debug bind insns.
2752 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2753 Narrow debug insns preexisting special handling down to debug
2755 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2756 handling down to debug bind insns.
2757 * function.c (instantiate_virtual_regs): Skip debug markers,
2758 adjust handling of debug binds.
2759 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2760 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2761 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2762 insns to debug bind insns.
2763 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2764 special handling of debug stmts to debug bind stmts.
2765 * ipa-split.c (split_function): Likewise.
2766 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2767 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2769 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2770 * regrename.c (build_def_use): Likewise.
2771 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2772 (pass_cprop_hardreg): Narrow special casing of debug insns to
2774 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2775 * reload1.c (reload): Likewise.
2776 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2777 casing of debug stmts to debug bind stmts.
2778 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2779 * ssa-iterators.h (num_imm_uses): Likewise.
2780 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2781 debug stmts to debug bind stmts.
2782 * tree-inline.c (tree_function_versioning): Narrow special casing
2783 of debug stmts to debug bind stmts.
2784 * tree-loop-distribution.c (generate_loops_for_partition):
2785 Narrow special casing of debug stmts to debug bind stmts.
2786 * tree-sra.c (analyze_access_subtree): Narrow special casing
2787 of debug stmts to debug bind stmts.
2788 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2789 stmts to debug bind stmts.
2790 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2791 casing of debug stmts to debug bind stmts.
2792 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2793 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2794 casing of debug stmts to debug bind stmts.
2795 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2797 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2798 debug stmts to debug bind stmts.
2799 (gimple_replace_ssa_lhs): Likewise.
2800 (insert_debug_temp_for_var_def): Likewise.
2801 (insert_debug_temps_for_defs): Likewise.
2802 (reset_debug_uses): Likewise.
2803 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2804 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2805 (adjust_debug_stmts): Likewise.
2806 (adjust_phi_and_debug_stmts): Likewise.
2807 (vect_do_peeling): Likewise.
2808 * tree-vect-loop.c (vect_transform_loop): Likewise.
2809 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2810 * var-tracking.c (adjust_mems): Narrow special casing of debug
2811 insns to debug bind insns.
2812 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2813 (compute_bb_dataflow, vt_find_locations): Likewise.
2814 (vt_expand_loc, emit_notes_for_changes): Likewise.
2815 (vt_init_cfa_base): Likewise.
2816 (vt_emit_notes): Likewise.
2817 (vt_initialize): Likewise.
2818 (vt_finalize): Likewise.
2820 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2821 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2822 (next_nonnote_nondebug_insn_bb): New.
2823 (prev_nonnote_nondebug_insn_bb): New.
2824 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2825 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2826 (prev_nonnote_nondebug_insn_bb): Declare.
2827 (next_nonnote_nondebug_insn_bb): Declare.
2828 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2829 * cfgrtl.c (get_last_bb_insn): Likewise.
2830 * lra.c (push_insns): Likewise.
2832 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2835 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2837 * selftest.c (selftest::test_runner::test_runner): New ctor.
2838 (selftest::test_runner::~test_runner): New dtor.
2839 * selftest.h (class selftest::test_runner): New class.
2841 2017-12-11 Carl Love <cel@us.ibm.com>
2843 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2844 vec_extract_fp32_from_shortl]): Add #defines.
2845 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2846 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2847 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2848 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2849 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2850 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2851 * doc/extend.texi: Add documentation for the added builtins.
2853 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2855 PR rtl-optimization/80693
2856 PR rtl-optimization/81019
2857 PR rtl-optimization/81020
2858 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2859 are not mentioned in i3. Place the REG_UNUSED note on i2,
2860 possibly modified to REG_DEAD, if it did not originate in i3.
2862 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2864 * recog.c (store_data_bypass_p_1): New function.
2865 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2866 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2868 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2870 PR rtl-optimization/83361
2871 * ifcvt.c (if_convert): Call fixup_partitions.
2873 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2875 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2876 early folding of splat_u{8,16,32}.
2878 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2880 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2881 output_operand_lossage first argument with capital letter.
2882 (aarch64_override_options): Don't start error and sorry first argument
2883 with capital letter.
2885 2017-12-11 Andi Kleen <ak@linux.intel.com>
2887 PR gcov-profile/83355
2888 * auto-profile.c (string_table::get_index_by_decl): Don't
2889 recurse when abstract origin points to itself.
2891 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2893 PR tree-optimization/83320
2894 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2895 (prune_datarefs_not_in_loop): Ditto.
2897 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2899 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2901 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2903 PR tree-optimization/83337
2904 * gimple-loop-interchange.cc (compute_access_stride): Handle
2905 bitfield DRs properly.
2907 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2909 PR tree-optimization/83338
2910 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2911 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2912 vector of unsigned integers to vector of signed integers.
2914 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2916 PR rtl-optimization/83317
2917 * lra-constraints.c (process_address_1): Add insn code check.
2919 2017-12-08 Michael Matz <matz@suse.de>
2921 Fix PR tree-optimization/83323
2922 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2923 head-controlled loops and loop BBs.
2924 * common.opt (funroll-and-jam): Remove, instead ...
2925 (floop-unroll-and-jam): ... reuse this option.
2926 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2927 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2928 (-floop-unroll-and-jam): ... this option.
2930 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2932 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2933 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2934 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2936 2017-12-08 Richard Biener <rguenther@suse.de>
2939 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2940 handle accesses outside of zero-sized vars.
2942 2017-12-08 Martin Jambor <mjambor@suse.cz>
2944 PR tree-optimization/83141
2945 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2946 test for MEM_REFs implicitely changing types with padding. Remove
2948 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2950 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2952 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2954 * config/avr/avr.c (avr_attribute_table): Likewise.
2955 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2956 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2957 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2958 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2959 of struct attribute_spec.
2960 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2962 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2965 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2967 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2970 * profile-count.c (profile_count::from_gcov_type): Move from
2971 profile-count.h; handle overflow.
2972 * profile-count.h (profile_count::from_gcov_type): Move offline.
2974 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2976 PR rtl-optimization/83304
2977 * combine.c (move_deaths): If we do not know where a register died,
2980 2017-12-08 Richard Biener <rguenther@suse.de>
2982 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2983 Provide -fopt-info-loop feedback when we interchange in a nest.
2985 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2987 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2988 for armv6 ARM CPU IDs.
2990 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2992 * common/config/arm/arm-common.c: Include <algorithm>.
2993 (INCLUDE_VECTOR): Define.
2994 (compare_opt_names): New function.
2995 (arm_rewrite_selected_arch): Only strip out extensions that can be
2996 expressed through -mfpu. Sort the remaining extensions
2999 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3001 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3002 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3003 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3004 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3006 2017-06-08 Tristan Gingold <gindold@adacore.com>
3009 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3010 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3012 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3014 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3015 of struct attribute_spec.
3017 2017-12-08 Julia Koval <julia.koval@intel.com>
3019 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3020 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3021 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3022 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3023 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3024 _mm_maskz_dpwssds_epi32): Ditto.
3026 2017-12-08 Richard Biener <rguenther@suse.de>
3028 PR tree-optimization/81303
3029 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3030 conditions try to create a comparison vector type matching
3031 the data vector type.
3032 (vectorizable_condition): Adjust.
3033 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3034 Leave invariant conditions alone in case we can vectorize those.
3036 2017-12-08 Julia Koval <julia.koval@intel.com>
3038 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3039 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3040 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3041 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3042 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3044 2017-12-08 Julia Koval <julia.koval@intel.com>
3046 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3047 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3048 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3049 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3050 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3051 _mm_maskz_dpbusds_epi32): New intrinsics.
3053 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3055 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3057 (ld<bh>io_signed): Likewise.
3058 (st<bhw>io): Likewise.
3059 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3060 SMALL_INT12 constant integer operand.
3062 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3064 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3066 (nios2_large_constant_allowed): ...this. Adjust uses.
3067 (nios2_plus_symbolic_constant_p): Rename to...
3068 (nios2_plus_large_constant_p): ...this. Adjust uses.
3069 (nios2_legitimate_address_p): Correct CONST_INT handling.
3070 (nios2_symbolic_memory_operand_p): Rename to...
3071 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3072 (nios2_large_constant_p): Check for large constant integers too.
3073 (nios2_split_large_constant): Handle constant integers.
3074 (nios2_split_symbolic_memory_operand): Rename to...
3075 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3076 (nios2_legitimize_constant_address): Handle constant integers.
3077 (r0rel_constant_p): Handle small constant integers.
3078 (nios2_print_operand_address): Handle r0-relative integer addresses.
3079 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3080 * config/nios2/nios2.md: Adjust for renamed functions.
3082 2017-12-07 Andrew Waterman <andrew@sifive.com>
3084 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3085 (TARGET_HAVE_SRODATA_SECTION): New define.
3086 (riscv_select_section): New function.
3088 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3089 Alexander Monakov <amonakov@ispras.ru>
3090 Jakub Jelinek <jakub@redhat.com>
3093 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3095 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3097 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3098 a bool success value. Don't call output_operand_lossage here.
3099 (aarch64_print_ldpstp_address): Return a bool success value.
3100 (aarch64_print_operand_address): Call output_addr_const if
3101 aarch64_print_address_internal fails.
3102 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3103 'y'; call output_operand_lossage instead. Call output_operand_lossage
3104 if aarch64_print_ldpstp_address fails.
3106 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3108 * tree-vector-builder.h
3109 (tree_vector_builder::binary_encoded_nelts): Declare.
3110 * tree-vector-builder.c
3111 (tree_vector_builder::binary_encoded_nelts): New function.
3112 * fold-const.c (negate_expr_p): Likewise.
3113 (operand_equal_p, fold_checksum_tree): Likewise.
3114 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3115 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3116 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3117 (uniform_vector_p): Likewise.
3118 * varasm.c (const_hash_1, compare_constant): Likewise.
3119 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3120 (valid_lattice_transition): Operate directly on the VECTOR_CST
3122 * ipa-icf.c: Include tree-vector-builder.h.
3123 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3124 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3126 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3128 * tree.c (build_vector): Delete.
3129 * tree.h (build_vector): Make static and move into the self-testing
3132 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3134 * vector-builder.h (vector_builder::derived): New const overload.
3135 (vector_builder::elt): New function.
3136 * tree-vector-builder.h (tree_vector_builder::type): New function.
3137 (tree_vector_builder::apply_step): Declare.
3138 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3140 * gimple-fold.h (tree_vector_builder): Declare.
3141 (gimple_build_vector): Take a tree_vector_builder instead of a
3142 type and vector of elements.
3143 * gimple-fold.c (gimple_build_vector): Likewise.
3144 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3146 (get_initial_defs_for_reduction): Likewise.
3147 (vectorizable_induction): Likewise.
3149 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3151 * tree-vector-builder.h
3152 (tree_vector_builder::new_binary_operation): Declare.
3153 * tree-vector-builder.c
3154 (tree_vector_builder::new_binary_operation): New function.
3155 * fold-const.c (fold_relational_const): Use it.
3156 (const_binop): Likewise. Check that both input vectors have
3157 the same number of elements, thus excluding things like WIDEN_SUM.
3158 Check whether it is possible to operate directly on the encodings
3161 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3163 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3164 new_unary_operation, operating only on the encoded elements.
3165 (const_unop): Likewise.
3166 (exact_inverse): Likewise.
3167 (distributes_over_addition_p): New function.
3168 (const_binop): Use tree_vector_builder and new_unary_operation
3169 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3170 on the encoded elements unless the encoding is strided and the
3171 operation does not distribute over addition.
3172 (fold_convert_const): Use tree_vector_builder and
3173 new_unary_operation. Operate only on the encoded elements
3174 for truncating integer conversions, or for non-stepped encodings.
3176 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3178 * config/sparc/sparc.c: Include tree-vector-builder.h.
3179 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3180 * expmed.c: Include tree-vector-builder.h.
3181 (make_tree): Use tree_vector_builder instead of build_vector.
3182 * fold-const.c: Include tree-vector-builder.h.
3183 (const_binop): Use tree_vector_builder instead of build_vector.
3184 (const_unop): Likewise.
3185 (native_interpret_vector): Likewise.
3186 (fold_vec_perm): Likewise.
3187 (fold_ternary_loc): Likewise.
3188 * gimple-fold.c: Include tree-vector-builder.h.
3189 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3191 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3192 (simplify_vector_constructor): Use tree_vector_builder instead
3194 * tree-vect-generic.c: Include tree-vector-builder.h.
3195 (add_rshift): Use tree_vector_builder instead of build_vector.
3196 (expand_vector_divmod): Likewise.
3197 (optimize_vector_constructor): Likewise.
3198 * tree-vect-loop.c: Include tree-vector-builder.h.
3199 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3200 of build_vector. Explicitly use a stepped encoding for
3202 * tree-vect-slp.c: Include tree-vector-builder.h.
3203 (vect_get_constant_vectors): Use tree_vector_builder instead
3205 (vect_transform_slp_perm_load): Likewise.
3206 (vect_schedule_slp_instance): Likewise.
3207 * tree-vect-stmts.c: Include tree-vector-builder.h.
3208 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3209 (vect_gen_perm_mask_any): Likewise.
3210 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3211 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3213 (build_vector_from_val): Likewise. Explicitly use a duplicate
3216 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3218 * doc/generic.texi (VECTOR_CST): Describe new representation of
3220 * vector-builder.h: New file.
3221 * tree-vector-builder.h: Likewise.
3222 * tree-vector-builder.c: Likewise.
3223 * Makefile.in (OBJS): Add tree-vector-builder.o.
3224 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3225 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3226 (tree_vector): Change the number of elements to
3227 vector_cst_encoded_nelts.
3228 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3229 (VECTOR_CST_ELTS): Delete.
3230 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3231 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3232 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3233 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3234 (VECTOR_CST_ENCODED_ELT): Likewise.
3235 (vector_cst_encoded_nelts): New function.
3236 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3237 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3238 (vector_cst_int_elt, vector_cst_elt): Declare.
3239 * tree.c: Include tree-vector-builder.h.
3240 (tree_code_size): Abort if passed VECTOR_CST.
3241 (tree_size): Update for new VECTOR_CST layout.
3242 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3243 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3244 (build_vector): Use tree_vector_builder.
3245 (vector_cst_int_elt, vector_cst_elt): New functions.
3246 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3247 encoded elements and then create the vector in the canonical form.
3248 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3249 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3250 (tree_c_tests): Call test_vector_cst_patterns.
3251 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3253 (hash_tree): Likewise.
3254 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3255 (streamer_write_tree_header): Likewise.
3256 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3257 (streamer_alloc_tree): Likewise. Update call to make_vector.
3258 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3260 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3262 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3263 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3264 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3266 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3267 Richard Biener <rguenther@suse.de>
3269 PR tree-optimization/81303
3270 * Makefile.in (gimple-loop-interchange.o): New object file.
3271 * common.opt (floop-interchange): Reuse the option from graphite.
3272 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3273 -floop-interchange and mention it for -O3.
3274 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3275 * gimple-loop-interchange.cc: New file.
3276 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3277 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3278 * passes.def (pass_linterchange): New pass.
3279 * timevar.def (TV_LINTERCHANGE): New time var.
3280 * tree-pass.h (make_pass_linterchange): New declaration.
3281 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3282 interchange. Record IV before/after increment in new parameters.
3283 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3284 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3286 (check_reduction_path): ...New function here.
3287 * tree-vectorizer.h (check_reduction_path): New declaration.
3289 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3292 PR rtl-optimization/80818
3293 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3294 always early clobbered.
3295 * lra-lives.c (process_bb_lives): Check input hard regs for early
3296 clobbered non-operand hard reg.
3298 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3301 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3302 types_compatible_p, just that TYPE_MODE is the same.
3304 2017-12-07 Martin Sebor <msebor@redhat.com>
3307 * attribs.c (empty_attribute_table): Initialize new member of
3308 struct attribute_spec.
3309 (decl_attributes): Add argument. Handle mutually exclusive
3310 combinations of attributes.
3311 (selftests::test_attribute_exclusions): New function.
3312 (selftests::attribute_c_tests): Ditto.
3313 * attribs.h (decl_attributes): Add default argument.
3314 * selftest.h (attribute_c_tests): Declare.
3315 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3316 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3318 * doc/extend.texi (Common Function Attributes): Update const and pure.
3319 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3320 of struct attribute_spec.
3321 * config/arc/arc.c (arc_attribute_table): Same.
3322 * config/arm/arm.c (arm_attribute_table): Same.
3323 * config/avr/avr.c ( avr_attribute_table): Same.
3324 * config/bfin/bfin.c (bfin_attribute_table): Same.
3325 * config/cr16/cr16.c (cr16_attribute_table): Same.
3326 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3327 * config/h8300/h8300.c (h8300_attribute_table): Same.
3328 * config/i386/i386.c (ix86_attribute_table): Same.
3329 * config/ia64/ia64.c (ia64_attribute_table): Same.
3330 * config/m32c/m32c.c (m32c_attribute_table): Same.
3331 * config/m32r/m32r.c (m32r_attribute_table): Same.
3332 * config/m68k/m68k.c (m68k_attribute_table): Same.
3333 * config/mcore/mcore.c (mcore_attribute_table): Same.
3334 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3335 * config/mips/mips.c (mips_attribute_table): Same.
3336 * config/msp430/msp430.c (msp430_attribute_table): Same.
3337 * config/nds32/nds32.c (nds32_attribute_table): Same.
3338 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3339 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3340 * config/rl78/rl78.c (rl78__attribute_table): Same.
3341 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3342 * onfig/rx/rx.c (rx_attribute_table): Same.
3343 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3344 * config/sh/sh.c (sh_attribute_table): Same.
3345 * config/sparc/sparc.c (sparc_attribute_table): Same.
3346 * config/spu/spu.c (spu_attribute_table): Same.
3347 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3348 * config/v850/v850.c (v850_attribute_table): Same.
3349 * config/visium/visium.c (visium_attribute_table): Same.
3351 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3354 * config/arm/arm.c (INCLUDE_STRING): Define.
3355 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3356 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3359 2017-12-07 Michael Matz <matz@suse.de>
3361 Add unroll and jam pass
3363 * gimple-loop-jam.c: New file.
3364 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3365 * common.opt (funroll-and-jam): New option.
3366 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3367 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3368 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3369 * passes.def: Add pass_loop_jam.
3370 * timevar.def (TV_LOOP_JAM): Add.
3371 * tree-pass.h (make_pass_loop_jam): Declare.
3372 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3373 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3374 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3375 to flow_loop_tree_node_add.
3376 (duplicate_subloops, copy_loops_to): Append to sibling list.
3377 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3378 * doc/invoke.texi (-funroll-and-jam): Document new option.
3379 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3381 2017-12-07 Richard Biener <rguenther@suse.de>
3383 PR tree-optimization/83296
3384 PR tree-optimization/67769
3385 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3386 flow sensitive info in an unrelated BB.
3387 (value_replacement): Use reset_flow_sensitive_info.
3388 (minmax_replacement): Reset flow sensitive info on the def
3389 we move. Do not reset flow sensitive info in the whole BB
3390 we move the stmt to.
3391 (abs_replacement): Likewise.
3393 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3396 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3397 rs6000_cpu to the given -mcpu=, or to the default processor.
3399 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3401 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3402 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3403 instead of rs6000_cpu_attr.
3404 (rs6000_adjust_cost): Ditto.
3405 (is_microcoded_insn): Ditto.
3406 (rs6000_adjust_priority): Ditto.
3407 (rs6000_issue_rate): Ditto.
3408 (rs6000_use_sched_lookahead): Ditto.
3409 (rs6000_use_sched_lookahead_guard): Ditto.
3410 (rs6000_sched_reorder): Ditto.
3411 (force_new_group): Ditto.
3412 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3413 (group_ending_nop): Ditto.
3415 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3417 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3418 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3419 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3420 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3422 (rs6000_reassociation_width): Ditto.
3423 (rs6000_emit_epilogue): Ditto.
3424 (rs6000_adjust_cost): Ditto.
3425 (is_microcoded_insn): Ditto.
3426 (is_cracked_insn): Ditto.
3427 (rs6000_adjust_priority): Ditto.
3428 (rs6000_sched_reorder): Ditto.
3429 (rs6000_sched_reorder2): Ditto.
3430 (insn_must_be_first_in_group): Ditto.
3431 (insn_must_be_last_in_group): Ditto.
3432 (rs6000_register_move_cost): Ditto.
3433 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3436 2017-12-07 Julia Koval <julia.koval@intel.com>
3438 * config.gcc: Add vaesintrin.h.
3439 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3441 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3442 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3444 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3445 * config/i386/immintrin.h: Include vaesintrin.h.
3446 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3447 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3448 _mm_aesdec_epi128): New intrinsics.
3450 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3452 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3453 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3454 c-family/c-spellcheck.cc.
3455 (best_macro_match::best_macro_match): Likewise.
3457 (struct edit_distance_traits<cpp_hashnode *>): Move to
3458 c-family/c-spellcheck.h.
3459 (class best_macro_match): Likewise.
3461 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3463 PR tree-optimization/83293
3464 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3465 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3466 might insert into empty bb.
3469 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3470 simplify for plus with :c added, and pointer_plus without that.
3471 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3472 with undefined overflow and the conversion is not widening,
3473 perform negation in utype and only convert to type afterwards.
3474 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3475 simplify for plus with :c added, and pointer_plus without that.
3476 If type is integral with undefined overflow and the conversion is
3477 not widening, perform minus in utype and only convert to type
3478 afterwards. Move the last pointer_diff_expr simplify into the
3481 2017-12-06 Martin Sebor <msebor@redhat.com>
3483 PR tree-optimization/82646
3484 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3485 strncpy, not maxlen.
3487 2017-12-06 Martin Sebor <msebor@redhat.com>
3489 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3492 PR tree-optimization/83075
3493 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3494 strncat/strncpy don't change length of source string.
3496 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3499 2017-11-29 Martin Aberg <maberg@gaisler.com>
3501 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3502 to prevent b2bst errata sequence.
3503 (sqrtdf2_fix): Likewise.
3505 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3507 PR tree-optimization/81945
3508 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3509 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3510 to dest_cfun has orig_loop_num set, either remap it to the new
3511 loop number if the loop got moved too, or clear it.
3513 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3515 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3517 (thunderx2t99_nothing): New insn reservation.
3518 (thunderx2t99_mrs): New insn reservation.
3519 (thunderx2t99_multiple): New insn reservation.
3520 (thunderx2t99_alu_basi): Add bfx to reservation.
3521 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3523 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3526 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3528 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3530 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3531 tree-ssa-pre.c::remove_dead_inserted_code.
3532 * tree-ssa-dce.h: New file.
3533 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3534 (remove_dead_inserted_code): Move and rename to function
3535 tree-ssa-dce.c::simple_dce_from_worklist.
3536 (pass_pre::execute): Update use.
3538 2017-12-05 Richard Biener <rguenther@suse.de>
3540 PR tree-optimization/83277
3541 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3542 to code-gen liveout vars.
3544 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3546 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3547 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3548 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3551 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3553 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3555 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3556 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3559 2017-12-05 Richard Biener <rguenther@suse.de>
3561 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3562 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3563 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3564 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3566 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3568 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3570 2017-12-05 Martin Liska <mliska@suse.cz>
3571 Jakub Jelinek <jakub@redhat.com>
3573 * doc/invoke.texi: Document the options.
3574 * flag-types.h (enum sanitize_code): Add
3575 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3576 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3577 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3578 * opts.c: Define new sanitizer options.
3579 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3580 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3582 2017-12-05 Julia Koval <julia.koval@intel.com>
3584 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3585 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3586 (ix86_handle_option): Handle -mavx512vnni.
3587 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3588 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3589 * config/i386/i386-c (__AVX512VNNI__): New.
3590 * config/i386/i386.c (ix86_target_string): Handle new option.
3591 (ix86_valid_target_attribute_inner_p): Handle new option.
3592 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3593 * config/i386/i386.opt (mavx512vnni): New option.
3595 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3598 * athlon.md: Disable for generic.
3599 * haswell.md: Enable for generic.
3600 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3601 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3603 (ix86_adjust_cost): Move generic to haswell path.
3605 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3607 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3608 instead of MEM_P in a couple more places. Fix formatting issues.
3610 2017-12-04 Jim Wilson <jimw@sifive.com>
3612 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3613 instead of GP_REG_LAST-1.
3614 (riscv_adjust_libcall_cfi_prologue): Likewise.
3615 (riscv_adjust_libcall_cri_epilogue): Likewise.
3616 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3619 2017-12-04 Luis Machado <luis.machado@linaro.org>
3621 * ipa-pure-const.c (check_decl): Add missing newline.
3622 (state_from_flags): Likewise.
3624 2017-12-04 Jeff Law <law@redhat.com>
3626 PR tree-optimizatin/78496
3627 * gimple-ssa-evrp-analyze.h
3628 (evrp_range_analyzer::get_vr_values): Simplify.
3629 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3630 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3631 and gimple-ssa-evrp-analyze.h.
3632 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3633 (simplify_stmt_for_jump_threading): Copy a blob of code from
3634 tree-vrp.c to use ranges to simplify statements.
3635 (dom_opt_dom_walker::before_dom_children): Call
3636 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3637 (dom_opt_dom_walker::after_dom_children): Similarly for
3638 evrp_range_analyzer::leave.
3639 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3642 * gimple-ssa-evrp-analyze.c
3643 (evrp_range_analyzer::extract_range_from_stmt): Always use
3644 vr_values::update_value_range so preexisting range info is
3645 medged with new range info, even if the new range is VR_VARYING.
3647 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3649 * combine.c: Adjust comment.
3650 (use_crosses_set_p): Delete.
3651 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3652 (try_combine): Ditto.
3654 2017-12-04 Richard Biener <rguenther@suse.de>
3656 PR tree-optimization/83255
3657 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3658 Re-add zero-iteration check.
3660 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3662 PR rtl-optimization/83245
3663 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3664 hard registers as earlyclobber, also if not in an asm.
3666 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3670 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3673 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3674 rs6000_cpu based on cpu_index, not tune_index.
3676 2017-12-04 Richard Biener <rguenther@suse.de>
3678 PR tree-optimization/83238
3679 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3680 code match comment, rejecting invalid SESE regions.
3682 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3684 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3685 require base operand is a REG_POINTER prior to reload on targets
3686 with non-equivalent space registers.
3688 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3690 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3691 (update_specialized_profile): Fix updating of counts.
3692 (perhaps_add_new_callers): Likewise.
3694 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3697 * x86-tune.def: Remove obsolette FIXMEs.
3698 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3699 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3700 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3702 (X86_TUNE_PAD_RETURNS): Disable for generic.
3704 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3706 PR tree-optimization/83170
3707 PR tree-optimization/83241
3708 * gimple-ssa-store-merging.c
3709 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3710 gimple_vuse (ins_stmt) in case it has changed.
3711 (imm_store_chain_info::output_merged_store): Likewise.
3713 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3717 * tree-cfg.c (pass_warn_function_return::execute): Handle
3718 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3719 with BUILTINS_LOCATION.
3723 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3724 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3725 instead of DECL_MODE.
3727 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3728 last argument to rtx pointer.
3729 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3730 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3731 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3732 depending on the chosen ISAs.
3733 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3734 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3735 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3737 * config/i386/sse.md (mov<mode>_internal): Likewise.
3738 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3740 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3742 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3744 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3746 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3747 parameters from prototype.
3748 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3749 parameters. Don't print a comment.
3750 (emit_fusion_gpr_load): Adjust.
3751 (emit_fusion_load_store): Adjust.
3752 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3753 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3754 comment on the second line.
3756 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3759 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3760 rs6000_cpu based on cpu_index, not tune_index.
3762 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3764 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3765 which_alternative instead of which_alternative + 1.
3766 (output_asm_insn): Print an extra tab if the template is short.
3768 2017-12-01 Jim Wilson <jimw@sifive.com>
3770 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3772 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3773 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3775 * doc/tm.texi: Regenerate.
3777 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3780 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3781 whether we can allocate pseudos before trying to fix an address.
3782 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3783 memory address is indexed or indirect.
3784 (floatuns_<mode>si2_hw2): Likewise.
3786 2017-12-01 Jason Merrill <jason@redhat.com>
3788 * Makefile.in (TAGS): Add c-family/*.cc.
3790 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3792 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3793 (call_value_insn): Likewise.
3794 (sibcall_insn): Likewise.
3795 (sibcall_value_insn): Likewise.
3796 (movsi_aarch64): Likewise.
3797 (movdi_aarch64): Likewise.
3798 (add_losym_): Likewise.
3799 (ldr_got_small_): Likewise.
3800 (ldr_got_small_sidi): Likewise.
3801 (ldr_got_small_28k_): Likewise.
3802 (ldr_got_small_28k_sidi): Likewise.
3803 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3804 Move output_addr_const to symbolic case. Add error check.
3806 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3808 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3809 * config/s390/s390.md: Disable MVC merging peephole if it would
3810 disable operand forwarding.
3811 (new peephole2): Split MVCs if it would turn them into up to 2
3814 2017-12-01 Richard Biener <rguenther@suse.de>
3816 PR tree-optimization/83232
3817 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3818 detection of same access. Instead of breaking the group here
3819 do not consider the duplicate. Add comment explaining real fix.
3821 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3823 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3825 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3827 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3829 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3831 * function.h (struct function): Remove cilk_frame_decl,
3832 is_cilk_function and calls_cilk_spawn fields.
3833 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3835 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3836 cilk_elemental field.
3837 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3838 * target.def: Adjust comment.
3839 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3840 Don't test cilk_elemental.
3842 PR tree-optimization/83233
3843 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3844 bswap_stat name for the struct.
3847 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3848 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3849 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3850 SWITCH_BREAK_LABEL_P set on the label.
3851 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3852 added for default case if it was missing and not all cases covered.
3853 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3854 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3856 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3857 to avoid -Wimplicit-fallthrough warning.
3858 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3859 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3861 PR tree-optimization/83221
3862 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3864 (init_reassoc): Formatting fix.
3867 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3868 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3872 * tree-cfg.c: Include asan.h.
3873 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3875 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3877 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3879 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3881 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3882 vpcmpeqd instruction.
3884 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3886 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3887 registers type for 128bit mode.
3889 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3891 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3893 * gengtype-state.c (read_state_pair): Likewise.
3894 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3895 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3896 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3897 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3898 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3899 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3900 * builtins.c (fold_builtin_3): Likewise.
3901 * graphite-scop-detection.c
3902 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3903 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3906 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3909 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3910 parameter. Return true if ignore_decl_rtl_set_p is true and other
3911 conditions are satisfied.
3912 * asan.h (asan_protect_global): Add new parameter.
3913 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3914 to asan_protect_global calls.
3916 2017-11-30 Jim Wilson <jimw@sifive.com>
3918 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3919 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3920 -mstrict-align, add info on default value. Delete redundant lines for
3921 -mabi. Add missing -mexplicit-relocs docs.
3923 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3925 * config/arc/arc.md (trap): New pattern.
3927 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3929 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3930 instruction to end into a delay slot.
3931 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3932 can be placed into a delay slot against reg_note.
3934 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3936 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3937 labels number of usages.
3939 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3941 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3943 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3945 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3948 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3949 multiplication by power of 2 constant into two shifts + comparison.
3951 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3954 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3956 2017-11-30 Richard Biener <rguenther@suse.de>
3958 PR tree-optimization/83202
3959 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3960 allow_peel argument and guard peeling.
3961 (canonicalize_loop_induction_variables): Likewise.
3962 (canonicalize_induction_variables): Pass false.
3963 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3964 peeling from cunrolli.
3966 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3968 * combine.c (try_combine): Print a message to dump file whenever
3969 I0, I1, or I2 cannot be combined into I3.
3971 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3973 PR rtl-optimization/83156
3974 PR rtl-optimization/82621
3975 * combine.c (try_combine): Don't split an I2 if one of the dests is
3976 set again before I3. Allow unused dests.
3978 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3980 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3982 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3984 PR rtl-optimization/80818
3985 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3986 recursively. Use insn code for clobber.
3987 (lra_set_insn_recog_data): Pass the new arg to
3988 collect_non_operand_hard_regs.
3989 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3991 (lra_update_insn_regno_info): Pass insn to
3992 add_regs_to_insn_regno_info.
3994 2017-11-29 Jim Wilson <jimw@sifive.com>
3995 Andrew Waterman <andrew@sifive.com>
3997 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3998 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3999 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4000 use SINGLE_SHIFT_COST.
4001 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4002 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4004 2017-11-29 Julia Koval <julia.koval@intel.com>
4006 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4007 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4008 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4009 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4010 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4011 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4012 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4013 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4014 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4015 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4016 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4017 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4018 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4019 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4020 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4021 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4022 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4023 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4024 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4025 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4026 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4027 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4028 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4029 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4030 __builtin_ia32_vpshldv_v2di_mask,
4031 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4032 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4033 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4035 2017-11-29 Julia Koval <julia.koval@intel.com>
4037 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4038 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4039 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4040 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4041 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4042 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4043 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4044 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4045 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4046 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4047 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4048 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4049 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4050 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4051 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4052 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4053 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4054 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4055 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4056 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4057 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4059 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4061 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4062 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4065 2017-11-29 Martin Aberg <maberg@gaisler.com>
4067 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4068 to prevent b2bst errata sequence.
4069 (sqrtdf2_fix): Likewise.
4071 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4073 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4074 (div_sqrt_insn_p): New function.
4075 (sparc_do_work_around_errata): Insert NOP instructions to
4076 prevent sequences that could trigger the TN-0013 errata for
4077 certain LEON3 processors.
4078 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4079 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4080 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4081 (in_branch_delay): Prevent div and sqrt in delay slot if
4083 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4085 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4087 * config/sparc/sparc.c (atomic_insn_p): New function.
4088 (sparc_do_work_around_errata): Insert NOP instructions to
4089 prevent sequences that could trigger the TN-0010 errata for
4091 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4092 instruction referable in atomic_insns_p.
4094 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4096 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4097 (atomic_compare_and_swap_leon3_1): Likewise.
4100 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4102 * config/sparc/sparc.c (fpop_insn_p): New function.
4103 (sparc_do_work_around_errata): Insert NOP instructions to
4104 prevent sequences that could trigger the TN-0012 errata for
4106 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4107 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4108 (in_branch_annul_delay): Prevent floating-point instructions
4109 in delay slot of annulled integer branch.
4111 2017-11-29 Richard Biener <rguenther@suse.de>
4113 PR tree-optimization/83202
4114 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4116 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4117 nodes vectorized to the same stmts multiple times.
4118 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4119 (vect_analyze_slp_instance): Adjust.
4120 (scalar_stmts_to_slp_tree_map_t): New typedef.
4121 (vect_schedule_slp_instance): Add a map recording the SLP node
4122 representing the vectorized stmts for a set of scalar stmts.
4123 Avoid code-generating redundancies.
4124 (vect_schedule_slp): Allocate map and pass it down.
4126 2017-11-29 Nathan Sidwell <nathan@acm.org>
4129 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4132 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4134 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4135 cases for printing LDP/STP memory addresses.
4136 (aarch64_print_address_internal): Renamed from
4137 aarch64_print_operand_address, added parameter, add Pmode check.
4138 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4139 (aarch64_print_operand_address): Indirect to
4140 aarch64_print_address_internal.
4141 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4142 'y' operand output specifier.
4144 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4147 * tree.c (build_simple_mem_ref_loc): Handle
4148 get_addr_base_and_unit_offset returning a MEM_REF.
4151 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4154 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4157 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4159 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4161 (layout::move_to_column): Likewise.
4163 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4166 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4167 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4169 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4170 Martin Jambor <mjambor@suse.cz>
4173 * tree.h (expr_type_first_operand_type_p): Declare
4174 * tree.c (expr_type_first_operand_type_p): New function.
4175 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4176 (ipa_value_from_jfunc): Adjust declaration.
4177 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4178 Use it as result type for arithmetics, unless it is NULL in which case
4179 be more conservative.
4180 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4181 ipa_get_jf_pass_through_result.
4182 (propagate_vals_across_pass_through): Likewise.
4183 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4184 is to propagate_vals_across_pass_through.
4185 (propagate_constants_across_call): Pass PARM_TYPE to
4186 propagate_scalar_across_jump_function.
4187 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4188 ipa_value_from_jfunc.
4189 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4190 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4191 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4192 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4193 target_type, pass it to ipa_value_from_jfunc.
4194 (update_indirect_edges_after_inlining): Pass parameter type to
4195 try_make_edge_direct_simple_call.
4197 2017-11-28 Jeff Law <law@redhat.com>
4199 * gimple-ssa-evrp-analyze.c
4200 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4201 refine ranges if scev_initialized_p returns true.
4202 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4204 2017-11-28 Julia Koval <julia.koval@intel.com>
4206 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4207 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4208 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4209 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4210 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4211 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4212 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4213 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4214 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4215 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4216 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4217 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4218 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4219 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4220 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4221 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4222 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4223 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4224 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4225 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4226 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4227 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4229 2017-11-28 Julia Koval <julia.koval@intel.com>
4231 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4232 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4233 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4234 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4235 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4236 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4237 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4238 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4239 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4240 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4241 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4242 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4243 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4244 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4245 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4246 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4247 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4248 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4249 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4250 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4251 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4252 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4253 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4254 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4255 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4256 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4257 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4258 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4259 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4260 __builtin_ia32_vpshld_v2di_mask): New builtins.
4261 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4262 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4264 2017-11-28 Richard Biener <rguenther@suse.de>
4266 PR tree-optimization/80776
4267 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4269 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4271 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4272 If the incoming edge is an effective fallthru because the other
4273 edge only reaches a __builtin_unreachable () then record ranges
4274 derived from the controlling condition in SSA info.
4275 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4276 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4278 2017-11-28 Olivier Hainque <hainque@adacore.com>
4280 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4281 on mingw build hosts.
4283 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4286 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4288 2017-11-28 Richard Biener <rguenther@suse.de>
4291 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4292 copies generated from memcpy use a character array as reference
4295 2017-11-28 Julia Koval <julia.koval@intel.com>
4296 Sebastian Peryt <sebastian.peryt@intel.com>
4298 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4299 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4300 cilk-common.o, cilk.h, cilk-common.c): Remove.
4302 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4303 * builtins.c (is_builtin_name): Remove cilkplus condition.
4304 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4305 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4306 cilk-builtins.def, cilkplus.def): Remove.
4307 * cif-code.def (CILK_SPAWN): Remove.
4308 * cilk-builtins.def: Delete.
4309 * cilk-common.c: Ditto.
4311 * cilkplus.def: Ditto.
4312 * config/darwin.h (fcilkplus): Delete.
4313 * cppbuiltin.c: Ditto.
4314 * doc/extend.texi: Remove cilkplus doc.
4315 * doc/generic.texi: Ditto.
4316 * doc/invoke.texi: Ditto.
4317 * doc/passes.texi: Ditto.
4318 * gcc.c (fcilkplus): Remove.
4319 * gengtype.c (cilk.h): Remove.
4320 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4322 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4324 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4326 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4327 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4328 cilkplus conditions.
4329 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4330 inline_read_section): Ditto.
4331 * ipa-inline-analysis.c (cilk.h): Remove.
4332 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4333 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4334 append_linker_options): Remove condition for fcilkplus.
4335 * lto/lto-lang.c (cilk.h): Remove.
4336 (lto_init): Remove condition for fcilkplus.
4337 * omp-expand.c (expand_cilk_for_call): Delete.
4338 (expand_omp_taskreg, expand_omp_for_static_chunk,
4339 expand_omp_for): Remove cilkplus
4341 (expand_cilk_for): Delete.
4342 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4343 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4344 execute_lower_omp, diagnose_sb_0): Ditto.
4345 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4346 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4347 * tree-nested.c: Ditto.
4348 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4349 (dump_generic_node): Ditto.
4350 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4351 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4353 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4355 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4357 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4358 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4359 for vector mode and !TARGET_SIMD.
4361 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4363 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4365 * tree.h (SWITCH_LABELS): Remove.
4366 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4367 assert SWITCH_BODY is non-NULL.
4368 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4370 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4372 PR tree-optimization/80788
4373 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4374 has TREE_OVERFLOW set, call drop_tree_overflow.
4376 2017-11-28 Richard Biener <rguenther@suse.de>
4378 PR tree-optimization/83158
4379 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4381 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4384 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4385 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4387 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4389 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4390 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4392 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4395 * varasm.c (compare_constant): Take the mode of the constants into
4396 account when comparing floating point constants.
4398 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4400 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4401 from explicit instantiation of debug_helper.
4402 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4404 2017-11-27 Richard Biener <rguenther@suse.de>
4406 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4409 2017-11-27 Richard Biener <rguenther@suse.de>
4411 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4412 (type_hash_canon): Also clear int_cst_hash_table entry for
4414 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4416 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4418 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4419 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4420 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4422 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4424 * hash-map.h (gt_cleare_cache): Avoid UB.
4426 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4428 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4429 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4430 (decide_unroll_runtime_iterations): Likewise.
4431 (decide_unroll_stupid): Likewise.
4433 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4436 * config/i386/i386.md: Add a loop with incssp.
4438 2017-11-27 Martin Jambor <mjambor@suse.cz>
4440 PR tree-optimization/81248
4441 * tree-sra.c (splice_param_accesses): Remove size check.
4442 (decide_one_param_reduction): Fix size check.
4443 * gimple-pretty-print.c (dump_profile): Silence warning.
4444 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4446 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4449 * dbxout.c (lastlineno): New variable.
4450 (dbx_debug_hooks): Use dbxout_switch_text_section as
4451 switch_text_section debug hook.
4452 (dbxout_function_end): Switch to current_function_section
4453 rather than function_section. If crtl->has_bb_partition,
4454 output just one N_FUN, depending on in_cold_section_p.
4455 (dbxout_source_line): Remember last lineno in lastlineno.
4456 (dbxout_switch_text_section): New function.
4457 (dbxout_function_decl): Adjust dbxout_block caller.
4458 (dbx_block_with_cold_children): New function.
4459 (dbxout_block): Return true if any LBRAC/RBRAC have been
4460 emitted. Use dbx_block_with_cold_children at depth == 0
4461 in second partition. Add PARENT_BLOCKNUM argument, pass
4462 it optionally adjusted to children. Output LBRAC/RBRAC
4463 around recursive call only if the block is in the current
4464 partition, if not and anything was output, emit empty
4466 * final.c (final_scan_insn): Compute cold_function_name
4467 before calling switch_text_section debug hook. Call
4468 that hook even if dwarf2out_do_frame if not emitting
4472 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4473 TREE_READONLY decls.
4475 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4477 PR rtl-optimization/82488
4478 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4480 2017-11-26 Julia Koval <julia.koval@intel.com>
4482 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4484 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4487 2017-11-26 Julia Koval <julia.koval@intel.com>
4489 * config/i386/driver-i386.c (host_detect_local_cpu):
4490 Detect skylake-avx512.
4492 2017-11-26 Julia Koval <julia.koval@intel.com>
4494 * config.gcc: Add -march=cannonlake.
4495 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4496 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4497 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4498 (PTA_CANNONLAKE): New.
4499 (processor_target_table): Add cannonlake.
4500 (ix86_option_override_internal): Ditto.
4501 (fold_builtin_cpu): Ditto.
4502 (get_builtin_code_for_version): Handle cannonlake.
4503 (M_INTEL_COREI7_CANNONLAKE): New.
4504 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4505 * doc/invoke.texi: Add -march=cannonlake.
4507 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4509 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4510 (try_init_one_plugin): Alternative implementation for MinGW.
4511 * Makefile.in (plugin_implib): New.
4512 (gengtype-lex.c): Fix broken AIX workaround.
4513 * configure: Regenerate.
4514 * doc/plugins.texi: Document support for MinGW.
4516 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4518 PR rtl-optimization/81553
4519 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4520 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4521 is a shift where C1 has different mode than the whole shift, use C1's
4522 mode for MULT rather than the shift's mode.
4525 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4526 builtins not enabled in the currently selected ISA.
4528 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4530 PR tree-optimization/71026
4531 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4532 (insert_reciprocals): Change to insert reciprocals before a division
4533 by a square and to insert the square of a reciprocal.
4534 (execute_cse_reciprocals_1): Change to consider division by a square.
4535 (register_division_in): Add importance parameter.
4537 2017-11-24 Richard Biener <rguenther@suse.de>
4539 PR tree-optimization/82402
4540 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4541 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4543 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4545 * match.pd (0-ptr): New transformation.
4547 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4550 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4551 when skipping checking.
4553 2017-11-24 Richard Biener <rguenther@suse.de>
4555 PR tree-optimization/83128
4556 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4557 (vn_reference_lookup_3): Likewise.
4559 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4562 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4563 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4566 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4567 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4568 check for the latter with gimple_call_builtin_p. Do not handle
4569 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4571 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4573 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4576 2017-11-23 Julia Koval <julia.koval@intel.com>
4578 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4579 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4580 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4581 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4582 _mm512_maskz_expandloadu_epi16): New intrinsics.
4583 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4584 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4585 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4586 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4587 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4588 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4589 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4590 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4591 _mm256_maskz_expandloadu_epi8): New intrinsics.
4592 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4593 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4594 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4595 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4596 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4597 * config/i386/sse.md (VI248_VLBW): New iterator.
4598 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4600 2017-11-23 Julia Koval <julia.koval@intel.com>
4602 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4603 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4604 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4605 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4606 _mm512_mask_compressstoreu_epi16): New.
4607 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4608 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4609 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4610 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4611 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4612 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4613 _mm256_mask_compressstoreu_epi8): New.
4614 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4615 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4616 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4617 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4618 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4619 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4620 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4621 __builtin_ia32_compresshi128_mask,
4622 __builtin_ia32_compressstoreuqi512_mask,
4623 __builtin_ia32_compressstoreuhi512_mask,
4624 __builtin_ia32_compressstoreuqi256_mask,
4625 __builtin_ia32_compressstoreuqi128_mask,
4626 __builtin_ia32_compressstoreuhi256_mask,
4627 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4628 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4630 (ix86_expand_special_args_builtin): Handle new types.
4631 (s4fma_expand): Handle new builtin array.
4632 * config/i386/immintrin.h: Include new headers.
4633 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4634 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4636 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4638 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4639 fixes. Declare temp and g variables at the top in order to avoid
4640 {} in most of the cases.
4642 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4644 * match.pd (ptr-0): New transformation.
4646 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4648 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4649 (struct addr_mode_cost_table): New.
4650 (struct tune_params): Add field addr_mode_costs.
4651 * config/arm/arm.c (generic_addr_mode_costs): New.
4652 (arm_slowmul_tune): Initialise addr_mode_costs field.
4653 (arm_fastmul_tune): Likewise.
4654 (arm_strongarm_tune): Likewise.
4655 (arm_xscale_tune): Likewise.
4656 (arm_9e_tune): Likewise.
4657 (arm_marvell_pj4_tune): Likewise.
4658 (arm_v6t2_tune): Likewise.
4659 (arm_cortex_tune): Likewise.
4660 (arm_cortex_a8_tune): Likewise.
4661 (arm_cortex_a7_tune): Likewise.
4662 (arm_cortex_a15_tune): Likewise.
4663 (arm_cortex_a35_tune): Likewise.
4664 (arm_cortex_a53_tune): Likewise.
4665 (arm_cortex_a57_tune): Likewise.
4666 (arm_exynosm1_tune): Likewise.
4667 (arm_xgene1_tune): Likewise.
4668 (arm_cortex_a5_tune): Likewise.
4669 (arm_cortex_a9_tune): Likewise.
4670 (arm_cortex_a12_tune): Likewise.
4671 (arm_cortex_a73_tune): Likewise.
4672 (arm_v7m_tune): Likewise.
4673 (arm_cortex_m7_tune): Likewise.
4674 (arm_v6m_tune): Likewise.
4675 (arm_fa726te_tune): Likewise.
4676 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4678 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4680 * config/arm/arm.c (arm_mem_costs): New function.
4681 (arm_rtx_costs_internal): Use arm_mem_costs.
4683 2017-11-23 Mark Wielaard <mark@klomp.org>
4685 * dwarf2out.c (init_sections_and_labels): Use generation to create
4686 unique ranges_section_label and ranges_base_label. Return generation.
4687 (output_rnglists): Add generation argument. Use generation to create
4688 unique ranges labels.
4689 (dwarf2out_finish): Get generation from init_sections_and_labels
4690 and pass generation to output_rnglists.
4692 2017-11-23 Mike Stump <mikestump@comcast.net>
4693 Eric Botcazou <ebotcazou@adacore.com>
4695 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4696 * cfgloop.h (struct loop): Add unroll field.
4697 * function.h (struct function): Add has_unroll bitfield.
4698 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4699 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4700 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4701 (pass_rtl_unroll_loops::gate): Likewise.
4702 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4703 for which loop->unroll==1.
4704 (decide_unroll_constant_iterations): Use note for consistency and
4705 take loop->unroll into account. Return early if loop->unroll is set.
4706 Fix thinko in existing test.
4707 (decide_unroll_runtime_iterations): Use note for consistency and
4708 take loop->unroll into account.
4709 (decide_unroll_stupid): Likewise.
4710 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4711 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4712 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4714 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4715 (print_loop): Print loop->unroll if set.
4716 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4717 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4718 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4720 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4721 loop->unroll is set and smaller than the trip count. Otherwise bypass
4722 entirely the heuristics if loop->unroll is set. Remove dead note.
4723 Fix off-by-one bug in other note.
4724 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4725 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4727 (tree_unroll_loops_completely): Make static.
4728 (pass_complete_unroll::execute): Use correct type for variable.
4729 (pass_complete_unrolli::execute): Fix formatting.
4730 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4732 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4734 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4735 enable when TARGET_PREFER_AVX128 is set.
4737 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4739 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4741 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4742 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4744 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4746 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4747 frequencies. Fix estimation of aggregate parameters.
4749 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4751 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4754 2017-11-23 Tom de Vries <tom@codesourcery.com>
4756 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4758 2017-11-23 Tom de Vries <tom@codesourcery.com>
4760 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4763 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4766 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4767 sibcall_value_pcrel_fdpic): Use local variable instead of
4769 (calli_tbr_rel): Add missing operand 2.
4770 (call_valuei_tbr_rel): Add missing operand 3.
4772 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4775 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4776 bitpos/bitsize covering the whole destination, use store_expr only if
4777 the complex mode is the same. Otherwise, use expand_normal and if
4778 it returns CONCAT, subreg each part separately instead of trying to
4779 subreg the whole result.
4781 2017-11-23 Richard Biener <rguenther@suse.de>
4783 PR tree-optimization/23094
4784 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4785 come here from walking over backedges in the first iteration.
4786 (vn_reference_lookup_3): Skip clobbers that store the same value.
4788 2017-11-23 Richard Biener <rguenther@suse.de>
4790 PR tree-optimization/81403
4791 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4792 a block we need a leader relative to.
4793 (phi_translate_1): For nary processing require a leader from
4794 get_representative_for given we run expression simplification
4795 using match-and-simplify. Remove previous fix.
4797 2017-11-22 Jeff Law <law@redhat.com>
4799 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4800 Use new method allocate_value_range rather than accessing the
4801 vrp_value_range_pool data member directly.
4802 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4803 to use extract_range_from_stmt method to avoid need for
4804 extract_range_from_assignment method.
4805 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4806 method rather than setting values_propgated data member directly.
4807 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4808 and values propagated data members and extract_range_from_assignment
4809 method. Reorder private data members to conform to standards.
4810 Add new methods set_lattice_propagation_complete and
4811 allocate_value_range.
4813 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4815 PR rtl-optimization/83030
4816 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4817 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4818 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4819 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4820 CROSSING_JUMP_P flag is already set before setting it.
4821 * cfgrtl.c (fixup_partition_crossing): Likewise.
4822 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4825 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4827 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4828 Handle the case where both arguments are using gen_const_vec_series.
4830 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4833 * pretty-print.c (pp_format): Move quoting implementation to
4834 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4835 to pass address of "quote" local.
4836 (pp_begin_quote): New function.
4837 (pp_end_quote): New function.
4838 * pretty-print.h (printer_fn): Convert penultimate param from bool
4840 (pp_begin_quote): New decl.
4841 (pp_end_quote): New decl.
4842 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4843 param from bool to bool *.
4844 * tree-diagnostic.h (default_tree_printer): Likewise.
4846 2017-11-22 Jeff Law <law@redhat.com>
4848 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4849 Initialize vr_values.
4850 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4851 vr_values via vr_values class instance rather than delegators.
4852 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4853 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4854 (evrp_range_analyzer::push_value_range): Likewise.
4855 (evrp_range_analyzer::pop_value_range): Likewise.
4856 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4857 most delegators. Those remaining are exposed as public interfaces.
4858 Make vr_values a pointer and private.
4859 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4861 (evrp_range_analyzer::get_vr_value): New method.
4862 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4863 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4864 (class evrp_dom_walker): Attach evrp_folder class, initialize
4865 it in the ctor. Remove temporary delegators.
4866 (evrp_dom_walker::before_dom_children): Call methods in attached
4867 evrp_range_analyzer class via class instance pointer. Use
4868 free value_range_constant_singleton to remove need for
4869 op_with_constant_singleton_value delegator method. Do not
4870 create a vrp_prop class instance for every call! Narrow
4871 scope of a couple variables.
4872 (evrp_dom_walker::cleanup): Call methods in attached
4873 evrp_range_analyzer class via class instance pointer.
4874 * vr-values.h (class vr_values): Privatize many methods and
4877 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4879 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4880 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4881 * expr.c (expand_expr_real_2): Likewise.
4882 * fold-const.c (const_unop): Likewise.
4883 * optabs-tree.c (optab_for_tree_code): Likewise.
4884 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4885 * tree-inline.c (estimate_operator_cost): Likewise.
4886 * tree-pretty-print.c (dump_generic_node): Likewise.
4887 (op_code_prio): Likewise.
4888 (op_symbol_code): Likewise.
4889 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4890 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4891 * internal-fn.c (direct_internal_fn_optab): New function.
4892 (direct_internal_fn_array, direct_internal_fn_supported_p
4893 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4894 * fold-const-call.c (fold_const_reduction): New function.
4895 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4897 * tree-vect-loop.c: Include internal-fn.h.
4898 (reduction_code_for_scalar_code): Rename to...
4899 (reduction_fn_for_scalar_code): ...this and return an internal
4901 (vect_model_reduction_cost): Take an internal_fn rather than
4903 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4905 (vectorizable_reduction): Use internal functions rather than tree
4906 codes for the reduction operation. Update calls to the functions
4908 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4909 Use calls to internal functions rather than REDUC tree codes.
4910 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4912 2017-11-22 Olivier Hainque <hainque@adacore.com>
4914 * config/vxworks.c (vxworks_override_options): Pick default
4915 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4916 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4917 DWARF_GNAT_ENCODINGS_DEFAULT.
4918 * config/vxworksae.h: Likewise.
4920 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4922 PR tree-optimization/83104
4923 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4926 2017-11-22 Marek Polacek <polacek@redhat.com>
4927 H.J. Lu <hongjiu.lu@intel.com>
4928 Jason Merrill <jason@redhat.com>
4933 * calls.c (initialize_argument_information): Call
4934 warn_parameter_passing_abi target hook.
4935 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4936 argument onto stack.
4937 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4938 * common.opt: Update -fabi-version description.
4939 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4940 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4942 (ix86_is_empty_record): New function.
4943 (ix86_warn_parameter_passing_abi): New function.
4944 (TARGET_EMPTY_RECORD_P): Redefine.
4945 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4946 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4947 * doc/tm.texi: Regenerated.
4948 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4949 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4950 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4951 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4952 instead of int_size_in_bytes.
4953 * expr.c (copy_blkmode_to_reg): Likewise.
4954 * function.c (aggregate_value_p): Return 0 for empty types.
4955 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4957 (locate_and_pad_parm): Call arg size_in_bytes instead
4959 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4960 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4961 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4963 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4964 (std_gimplify_va_arg_expr): Skip empty records. Call
4965 arg_size_in_bytes instead size_in_bytes.
4966 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4967 * tree-core.h (tree_type_common): Add empty_flag.
4968 (tree_decl_common): Update comments.
4969 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4971 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4972 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4974 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4975 * tree.c (default_is_empty_type): New function.
4976 (default_is_empty_record): New function.
4977 (arg_int_size_in_bytes): New function.
4978 (arg_size_in_bytes): New function.
4979 (get_ultimate_context): New function.
4980 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4981 TRANSLATION_UNIT_WARN_EMPTY_P.
4982 (default_is_empty_record, arg_int_size_in_bytes,
4983 arg_size_in_bytes, get_ultimate_context): Declare.
4985 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4987 * config/arm/arm.c (cmse_clear_registers): New function.
4988 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4989 code by call to cmse_clear_registers.
4990 (cmse_nonsecure_entry_clear_before_return): Likewise.
4992 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4994 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4995 (vdot_s32, vdotq_s32): New.
4996 (vdot_lane_u32, vdotq_lane_u32): New.
4997 (vdot_lane_s32, vdotq_lane_s32): New.
5000 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5003 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5004 for unsigned values with fewer HWIs than the precision.
5005 (test_overflow): New function.
5006 (wide_int_cc_tests): Call it.
5008 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5010 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5011 has the same mode class as Pmode.
5013 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5015 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5016 padding_bits_to_clear_ptr.
5017 (cmse_nonsecure_entry_clear_before_return): Likewise.
5019 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5021 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5022 auto_sbitap instead of integer bitfield to control register needing
5025 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5027 PR tree-optimization/83044
5028 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5029 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5030 that need the upper bound. Subtract offset from
5031 get_addr_base_and_unit_offset only if positive and subtract it
5032 before division by eltsize rather than after it.
5035 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5036 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5037 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5041 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5042 expand_binop with *mul_widen_optab, make sure at least one of the
5043 operands doesn't have VOIDmode.
5046 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5048 PR rtl-optimization/82044
5049 PR tree-optimization/82042
5050 * dse.c (record_store): Check for overflow.
5051 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5052 add_wild_read instead of clear_rhs_from_active_local_stores on
5053 overflow and log it into dump_file.
5055 2017-11-22 Richard Biener <rguenther@suse.de>
5057 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5058 fake edges to exit when looking for a place to insert.
5059 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5061 (insert_into_preds_of_block): Commit edge insertion immediately,
5062 assert that doesn't require new BBs.
5063 (fini_pre): Release expressions.
5064 (pass_pre::execute): Shuffle things around a bit, if the fn
5065 is too large do not compute AVAIL either as this is really the
5068 2017-11-22 Richard Biener <rguenther@suse.de>
5070 PR tree-optimization/83089
5071 * tree-if-conv.c (pass_if_conversion::execute): If anything
5072 changed reset SCEV and free the number of iteration estimates.
5074 2017-11-21 Martin Sebor <msebor@redhat.com>
5076 PR tree-optimization/82945
5077 * calls.h (warn_nonstring_bound): Remove unused function.
5079 2017-11-21 Martin Sebor <msebor@redhat.com>
5081 PR tree-optimization/82945
5082 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5083 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5084 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5086 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5087 * calls.h (get_attr_nonstring_decl): Declare new function.
5088 * doc/extend.texi (attribute nonstring): Update.
5089 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5090 get_attr_nonstring_decl and handle it.
5091 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5092 detection of nul-termination.
5093 (strlen_to_stridx): Change to a pointer.
5094 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5095 (pass_strlen::execute): Same.
5097 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5099 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5100 for the new option -mprefer-vector-width=[none|128|256|512].
5101 * config/i386/i386.c (ix86_target_string): remove old style options
5102 -mprefer-avx256 and make -mprefer-avx128 as alias.
5103 (ix86_option_override_internal): Apply defaults for the
5104 -mprefer-vector-width=[128|256] option.
5105 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5106 Implement macros to work with -mprefer-vector-width=.
5107 * config/i386/i386.opt: Implemented option
5108 -mprefer-vector-width=[none|128|256|512].
5109 * doc/invoke.texi: Documentation for
5110 -mprefer-vector-width=[none|128|256|512].
5112 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5114 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5116 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5118 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5120 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5122 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5124 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5125 * expr.c (expand_expr_real_2): Likewise.
5126 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5127 fold_binary_loc): Likewise.
5128 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5129 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5130 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5131 MINUS_EXPR transformations.
5132 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5133 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5134 * tree-inline.c (estimate_operator_cost): Likewise.
5135 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5136 op_symbol_code): Likewise.
5137 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5138 * vr-values.c (extract_range_from_binary_expr): Likewise.
5139 * varasm.c (initializer_constant_valid_p_1): Likewise.
5140 * tree.def: New tree code POINTER_DIFF_EXPR.
5142 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5144 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5145 integer suffix to movbe mnemonic.
5146 (*bswaphi2_movbe): Ditto.
5147 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5149 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5152 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5153 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5154 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5156 2017-11-21 Martin Liska <mliska@suse.cz>
5158 * tree-inline.c (expand_call_inline): Remove not needed
5161 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5162 Jakub Jelinek <jakub@redhat.com>
5165 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5166 Remove static keyword from f variable.
5168 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5170 PR tree-optimization/83086
5171 * gimple-ssa-store-merging.c
5172 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5173 rather than n.base_addr.
5175 2017-11-21 Martin Liska <mliska@suse.cz>
5177 PR rtl-optimization/82044
5178 PR tree-optimization/82042
5179 * dse.c (check_mem_read_rtx): Check for overflow.
5181 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5183 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5186 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5188 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5189 and its corresponding call arg location note.
5191 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5192 Andrew Burgess <andrew.burgess@embecosm.com>
5194 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5196 (arc_return_slot_offset): Likewise.
5197 (arc_eh_return_address_location): New declaration.
5198 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5199 (MUST_SAVE_REGISTER): Add exception handler case.
5200 (MUST_SAVE_RETURN_ADDR): Likewise.
5201 (arc_frame_pointer_required): Likewise.
5202 (arc_frame_pointer_needed): New function.
5203 (arc_compute_frame_size): Changed.
5204 (arc_expand_prologue): Likewise.
5205 (arc_expand_epilogue): Likewise.
5206 (arc_initial_elimination_offset): Likewise.
5207 (arc_return_slot_offset): Delete.
5208 (arc_eh_return_address_location): New function.
5209 (arc_builtin_setjmp_frame_value): Likewise.
5210 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5211 (EH_RETURN_STACKADJ_RTX): Define.
5212 (EH_RETURN_HANDLER_RTX): Likewise.
5213 * config/arc/arc.md (eh_return): Delete.
5215 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5217 * print-tree.h (debug_vec_tree): Remove prototype.
5218 * gdbinit.in (pvt): Remove macro.
5220 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5222 PR tree-optimization/83047
5223 * gimple-ssa-store-merging.c
5224 (imm_store_chain_info::output_merged_store): If the loads with the
5225 same vuse are in different basic blocks, for load_gsi pick a load
5226 location that is dominated by the other loads.
5229 * config/i386/i386.c (ix86_memmodel_check): Start
5230 -Winvalid-memory-model diagnostics with lowercase letter.
5233 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5234 set in_cold_section_p to the partition containing loc_list->first.
5235 When seeing loc_list->last_before_switch node, update secname and
5236 perform range_across_switch second partition handling only after that.
5239 * run-rtl-passes.c: Include debug.h.
5240 (run_rtl_passes): Call debug_hooks->assembly_start.
5241 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5245 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5246 OPTAB_DIRECT in calls to expand_simple_binop.
5248 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5251 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5253 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5256 * spellcheck.h (best_match::blithely_get_best_candidate): New
5259 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5261 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5262 with lower case letter.
5264 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5266 * config/i386/i386.md (bswaphi2): New expander.
5267 (*bswaphi2_movbe): New insn pattern.
5268 (bswaphi -> rorhi pepehole2): New peephole pattern.
5270 2017-11-20 Jeff Law <law@redhat.com>
5272 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5273 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5274 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5275 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5276 gimple-ssa-evrp-analyze.h.
5278 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5279 set BB_VISITED here.
5280 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5282 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5284 * tree-predcom.c: Add general comment on Store-Store chains.
5285 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5287 (get_chain_last_ref_at): Rename into...
5288 (get_chain_last_write_at): ...this.
5289 (get_chain_last_write_before_load): New function.
5290 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5291 CT_STORE_STORE when write reference is added.
5292 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5293 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5294 (initialize_root_vars_store_elim_1): Ditto.
5295 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5296 definition is created.
5297 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5298 chain by replacing it with dominant stored value.
5300 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5302 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5304 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5306 * vr-values.c (extract_range_from_binary_expr): Use a full range
5309 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5311 * config/arm/arm.md (R4_REGNUM): Define constant.
5312 (nonsecure_call_internal): Remove r4 clobber.
5313 (nonsecure_call_value_internal): Likewise.
5314 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5315 clobber and resequence match_operands.
5316 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5317 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5318 (nonsecure_call_value_reg_thumb2): Likewise.
5320 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5322 PR tree-optimization/78821
5323 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5324 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5325 to the address of the base rather than the base itself.
5326 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5327 (find_bswap_or_nop_finalize): New function.
5328 (find_bswap_or_nop): Use it.
5329 (bswap_replace): Return a tree rather than bool, change first
5330 argument from gimple * to gimple_stmt_iterator, allow inserting
5331 into an empty sequence, allow ins_stmt to be NULL - then emit
5332 all stmts into gsi. Fix up MEM_REF address gimplification.
5333 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5334 (struct store_immediate_info): Add N and INS_STMT non-static
5336 (store_immediate_info::store_immediate_info): Initialize them
5337 from newly added ctor args.
5338 (merged_store_group::apply_stores): Formatting fixes. Sort by
5340 (stmts_may_clobber_ref_p): For stores call also
5341 refs_anti_dependent_p.
5342 (gather_bswap_load_refs): New function.
5343 (imm_store_chain_info::try_coalesce_bswap): New method.
5344 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5345 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5346 (imm_store_chain_info::output_merged_store): Fail if number of
5347 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5348 and NOP_EXPR rhs_code.
5349 (pass_store_merging::process_store): Compute n and ins_stmt, if
5350 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5351 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5353 (pass_store_merging::execute): Calculate dominators.
5355 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5356 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5357 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5358 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5359 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5360 class pass_optimize_bswap, bswap_replace,
5361 pass_optimize_bswap::execute): Moved to ...
5362 * gimple-ssa-store-merging.c: ... this file.
5363 Include optabs-tree.h.
5364 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5365 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5366 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5367 anonymous namespace, remove static keywords.
5368 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5369 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5371 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5374 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5376 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5378 * vec.h (debug_helper): New function.
5379 (DEFINE_DEBUG_VEC): New macro.
5380 * hash-set.h (debug_helper): New function.
5381 (DEFINE_DEBUG_HASH_SET): New macro.
5382 * cfg.c (debug_slim (edge)): New function.
5383 Call DEFINE_DEBUG_VEC for edges.
5384 Call DEFINE_DEBUG_HASH_SET for edges.
5385 * cfghooks.c (debug_slim (basic_block)): New function.
5386 Call DEFINE_DEBUG_VEC for basic blocks.
5387 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5388 * print-tree.c (debug_slim): New function to handle trees.
5389 Call DEFINE_DEBUG_VEC for trees.
5390 Call DEFINE_DEBUG_HASH_SET for trees.
5391 (debug (vec<tree, va_gc>) &): Remove.
5392 (debug (<vec<tree, va_gc>) *): Remove.
5393 * print-rtl.c (debug_slim): New function to handle const_rtx.
5394 Call DEFINE_DEBUG_VEC for rtx_def.
5395 Call DEFINE_DEBUG_VEC for rtx_insn.
5396 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5397 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5398 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5399 (debug (vec<rtx_insn *> *ptr): Remove.
5400 (debug_insn_vector): Remove.
5401 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5403 2017-11-20 Tom de Vries <tom@codesourcery.com>
5405 PR rtl-optimization/82020
5406 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5407 IF_THEN_ELSE condition.
5409 2017-11-19 Jeff Law <law@redhat.com>
5411 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5412 of degenerates resulting from ignoring an edge.
5414 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5417 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5419 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5422 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5423 for uninitialied counts.
5425 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5428 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5430 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5433 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5435 2017-11-19 Tom de Vries <tom@codesourcery.com>
5437 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5440 2017-11-19 Tom de Vries <tom@codesourcery.com>
5442 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5445 2017-11-19 Tom de Vries <tom@codesourcery.com>
5447 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5450 2017-11-19 Tom de Vries <tom@codesourcery.com>
5452 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5454 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5456 2017-11-19 Tom de Vries <tom@codesourcery.com>
5458 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5460 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5461 semicolon after MCORE_EXPORT_NAME call.
5463 2017-11-19 Tom de Vries <tom@codesourcery.com>
5466 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5467 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5469 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5471 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5472 (cgraph_edge::clone): Cleanup updating of profile.
5473 * ipa-cp.c (update_profiling_info): Likewise.
5474 * ipa-inline-transform.c (inline_transform): Likewise.
5475 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5476 * ipa-split.c (execute_split_functions): Do not split when function
5478 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5479 * profile-count.c (profile_count::dump): Add global0.
5480 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5482 (profile_count::to_sreal_scale): Likewise.
5483 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5484 (profile_count::combine_with_ipa_count): New function.
5485 * profile-count.h (profile_guessed_global0adjusted): New.
5486 (profile_count::adjusted_zero): New.
5487 (profile_count::global0adjusted): New.
5488 (profile_count::combine_with_ipa_count): New.
5489 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5490 correct profile of return block of split functions.
5491 (copy_cfg_body): Remove unused profile_count.
5492 (copy_body): Likewise.
5493 (expand_call_inline): Update.
5494 (tree_function_versioning): Update.
5496 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5498 * hash-set.h (hash_set::empty): New.
5499 * tree-ssa-threadbackward.h: Delete.
5500 * tree-ssa-threadbackward.c (class thread_jumps): New.
5501 Move max_threaded_paths into class.
5502 (fsm_find_thread_path): Remove arguments that are now in class.
5503 (profitable_jump_thread_path): Rename to...
5504 (thread_jumps::profitable_jump_thread_path): ...this.
5505 (convert_and_register_jump_thread_path): Rename to...
5506 (thread_jumps::convert_and_register_current_path): ...this.
5507 (check_subpath_and_update_thread_path): Rename to...
5508 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5509 (register_jump_thread_path_if_profitable): Rename to...
5510 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5511 (handle_phi): Rename to...
5512 (thread_jumps::handle_phi): ...this.
5513 (handle_assignment): Rename to...
5514 (thread_jumps::handle_assignment): ...this.
5515 (fsm_find_control_statement_thread_paths): Rename to...
5516 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5517 (find_jump_threads_backwards): Rename to...
5518 (thread_jumps::find_jump_threads_backwards): ...this.
5519 Initialize path local data.
5520 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5521 from within thread_jumps class.
5522 (pass_early_thread_jumps::execute): Same.
5524 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5526 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5528 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5530 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5531 to canonical location.
5533 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5536 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5538 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5540 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5542 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5543 rcrt1.o%s/grcrt1.o%s for -static-pie.
5545 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5547 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5548 ix86_shift_rotate_cost): Break out from ...
5549 (ix86_rtx_costs): ... here.
5550 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5553 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5555 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5556 when entry block was promoted unlikely.
5557 (estimate_bb_frequencies): Increase frequency scale.
5558 * profile-count.h (profile_count): Export precision info.
5560 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5562 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5563 disturbin profile of entry block.
5565 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5567 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5570 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5572 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5573 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5575 2017-11-17 Jeff Law <law@redhat.com>
5577 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5578 from evrp_dom_walker class. Various methods moved into new class.
5579 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5580 (evrp_range_analyzer::enter): New method.
5581 (evrp_range_analyzer::leave): New method.
5582 (evrp_dom_walker): Remove delegators no longer needed by this class.
5583 Replace vr_values data member with evrp_range_analyzer
5585 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5586 method extracted from evrp_dom_walker::before_dom_children.
5587 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5588 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5590 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5591 Add private copy constructor and move assignment operators.
5592 Privatize methods and class data where trivially possible.
5593 (evrp_dom_walker::cleanup): New function, extracted from
5594 execute_early_vrp. Simplify access to class data.
5596 * vr-values.h (get_output_for_vrp): Prototype.
5597 * vr-values.c (get_output_for_vrp): New function extracted from
5598 vrp_visit_assignment_or_call and extract_range_from_stmt.
5599 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5601 2017-11-17 Luis Machado <luis.machado@linaro.org>
5603 * config/aarch64/aarch64.c
5604 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5605 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5606 tune_params::AUTOPREFETCHER_WEAK.
5608 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5611 * config/arm/arm.c (arm_valid_target_attribute_rec):
5612 Parse "arch=" and "+<ext>".
5613 (arm_valid_target_attribute_tree): Re-init global options.
5614 (arm_option_override): Make non-static.
5615 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5616 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5617 (__ARM_FEATURE_CRC32): Support undef.
5618 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5619 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5621 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5623 * gdbinit.in (break-on-diagnostic): New command.
5625 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5627 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5628 return to indirect jump for EH return if control-flow
5629 protection is enabled. Change explicit 'false' argument in
5630 pro_epilogue_adjust_stack with a value of
5632 * config/i386/i386.md (simple_return_indirect_internal):
5633 Remove SImode restriction to support 64-bit.
5635 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5637 * combine.c (added_notes_insn): New.
5638 (try_combine): Handle added_notes_insn like added_links_insn.
5639 Rewrite return value code.
5640 (distribute_notes): Set added_notes_insn to the earliest insn we added
5643 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5645 PR rtl-optimization/82621
5646 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5647 dest of one of those SETs is unused.
5649 2017-11-17 Richard Biener <rguenther@suse.de>
5652 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5653 * tree-pretty-print.c (dump_generic_node): Handle
5654 annot_expr_parallel_kind.
5655 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5656 * gimplify.c (gimple_boolify): Likewise.
5658 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5660 * config.gcc (extra_headers): Add cet.h for x86 targets.
5661 * config/i386/cet.h: New file.
5662 * doc/install.texi: Add --enable-cet/--disable-cet.
5664 2017-11-17 Richard Biener <rguenther@suse.de>
5666 PR tree-optimization/83017
5667 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5668 (gen_parallel_loop): Properly count iterations.
5669 (parallelize_loops): Handle loop->can_be_parallel independent
5670 of flag_loop_parallelize_all. Make static profitability test match
5672 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5673 * invoke.texi (parloops-min-per-thread): Document.
5675 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5677 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5678 upstreaming review comments.
5680 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5682 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5683 with fast unaligned access.
5684 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5686 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5688 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5689 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5691 2017-11-17 Richard Biener <rguenther@suse.de>
5693 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5694 folding of references.
5696 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5699 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5700 of replacing call to strncmp with corresponding call to strcmp when
5703 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5705 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5706 option prefer-avx256 for skylake-avx512 configuration.
5707 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5708 (get_builtin_code_for_version): Ditto.
5710 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5711 Monk Chiang <sh.chiang04@gmail.com>
5713 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5714 (FIXED_REGISTERS): Reserve more register numbers.
5715 (CALL_USED_REGISTERS): Likewise.
5716 (REG_ALLOC_ORDER): Likewise.
5717 (REG_CLASS_CONTENTS): Likewise.
5718 (REGISTER_NAMES): Likewise.
5720 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5721 Kito Cheng <kito.cheng@gmail.com>
5723 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5725 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5727 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5729 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5730 parameter list for vec_splats.
5732 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5734 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5736 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5738 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5741 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5742 with the required version of automake 1.11.6.
5744 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5746 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5748 (DU_C2_3_power9): Correct reservation combinations.
5749 (FP_DIV_power9, VEC_DIV_power9): New.
5750 (power9-alu): Split out rotate/shift...
5751 (power9-rot): ...to here, correct dispatch resource.
5752 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5754 (power9-fp): Correct latency.
5755 (power9-sdiv): Add div/sqrt resource.
5756 (power9-ddiv): Correct latency, add div/sqrt resource.
5757 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5758 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5760 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5762 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5764 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5765 switch statement mapping KF built-ins to TF built-ins if we don't
5766 have the proper ISA 3.0 assembler support.
5768 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5770 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5771 (lower_emutls_function_body): Do not compute it.
5773 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5775 * ipa-split.c (split_bb_info): Turn time to sreal.
5776 (split_point): Likewise.
5777 (dump_split_point): Likewise.
5778 (fine_split_points): Likewise.
5779 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5781 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5783 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5786 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5788 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5790 (expensive_function_p): Remove useless assert.
5791 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5793 2017-11-16 Martin Sebor <msebor@redhat.com>
5795 PR tree-optimization/82588
5796 PR tree-optimization/82583
5797 * tree-vrp.c (check_array_ref): Handle flexible array members,
5798 string literals, and inner indices.
5799 (search_for_addr_array): Add detail to diagnostics.
5801 2017-11-16 Nathan Sidwell <nathan@acm.org>
5805 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5806 (SET_DECL_ASSEMBLER_NAME): Forward to
5807 overwrite_decl_assembler_name.
5808 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5809 (overwrite_decl_assembler_name): Declare.
5810 * tree.c (overwrite_decl_assembler_name): New.
5811 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5812 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5813 (LANG_HOOKS_INITIALIZER): Add it.
5814 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5815 * langhooks.c (lhd_set_decl_assembler_name): Use
5816 SET_DECL_ASSEMBLER_NAME.
5817 (lhd_overwrite_decl_assembler_name): Default implementation.
5819 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5820 Jackson Woodruff <jackson.woodruff@arm.com>
5822 PR tree-optimization/71026
5823 * match.pd: Canonicalize constant multiplies in division.
5825 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5827 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5828 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5829 Remove OPT_fomit_frame_pointer entry.
5830 * common/config/arc/arc-common.c: Likewise.
5831 * common/config/arm/arm-common.c: Likewise.
5832 * common/config/avr/avr-common.c: Likewise.
5833 * common/config/c6x/c6x-common.c: Likewise.
5834 * common/config/cr16/cr16-common.c: Likewise.
5835 * common/config/cris/cris-common.c: Likewise.
5836 * common/config/epiphany/epiphany-common.c: Likewise.
5837 * common/config/fr30/fr30-common.c: Likewise.
5838 * common/config/frv/frv-common.c: Likewise.
5839 * common/config/ia64/ia64-common.c: Likewise.
5840 * common/config/iq2000/iq2000-common.c: Likewise.
5841 * common/config/lm32/lm32-common.c: Likewise.
5842 * common/config/m32r/m32r-common.c: Likewise.
5843 * common/config/mcore/mcore-common.c: Likewise.
5844 * common/config/microblaze/microblaze-common.c: Likewise.
5845 * common/config/mips/mips-common.c: Likewise.
5846 * common/config/mmix/mmix-common.c: Likewise.
5847 * common/config/mn10300/mn10300-common.c: Likewise.
5848 * common/config/nios2/nios2-common.c: Likewise.
5849 * common/config/pa/pa-common.c: Likewise.
5850 * common/config/pdp11/pdp11-common.c: Likewise.
5851 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5852 * common/config/riscv/riscv-common.c: Likewise.
5853 * common/config/rs6000/rs6000-common.c: Likewise.
5854 * common/config/rx/rx-common.c: Likewise.
5855 * common/config/s390/s390-common.c: Likewise.
5856 * common/config/sh/sh-common.c: Likewise.
5857 * common/config/sparc/sparc-common.c: Likewise.
5858 * common/config/tilegx/tilegx-common.c: Likewise.
5859 * common/config/tilepro/tilepro-common.c: Likewise.
5860 * common/config/v850/v850-common.c: Likewise.
5861 * common/config/visium/visium-common.c: Likewise.
5862 * common/config/xstormy16/xstormy16-common.c: Likewise.
5863 * common/config/xtensa/xtensa-common.c: Likewise.
5864 * invoke.texi (-fomit-frame-pointer): Update documentation.
5866 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5868 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5870 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5872 * cfg.c (scale_bbs_frequencies_int,
5873 cale_bbs_frequencies_gcov_type): Remove.
5874 * cfg.h (scale_bbs_frequencies_int,
5875 cale_bbs_frequencies_gcov_type): Remove.
5877 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5879 * tree-ssa-loop-manip.c
5880 (scale_dominated_blocks_in_loop): Update to profile counts.
5881 (tree_transform_and_unroll_loop): Likewise.
5883 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5885 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5886 scale_bbs_frequencies_int.
5888 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5890 * final.c (compute_alignments): Use counts rather than frequencies.
5892 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5894 * cfgloopanal.c: Include sreal.h
5895 (average_num_loop_insns): Use counts and sreal for accounting.
5897 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5899 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5902 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5903 Kito Cheng <kito.cheng@gmail.com>
5905 * config/nds32/constraints.md: Provide more constraints.
5906 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5907 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5908 support constraints usage.
5910 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5912 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5914 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5915 Kito Cheng <kito.cheng@gmail.com>
5917 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5918 * config/nds32/nds32.opt: Refine the layout.
5919 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5920 TARGET_EXT_STRING): Support new options.
5921 * config/nds32/nds32.h: Likewise.
5922 * config/nds32/nds32.md: Likewise.
5923 * config/nds32/nds32-predicates.c: Likewise.
5924 * config/nds32/constraints.md: Likewise.
5925 * common/config/nds32/nds32-common.c: Likewise.
5927 2017-11-16 Julia Koval <julia.koval@intel.com>
5930 * config/i386/gfniintrin.h: Add sse check.
5931 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5933 2017-11-16 Julia Koval <julia.koval@intel.com>
5935 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5936 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5937 (ix86_handle_option): Handle -mavx512vbmi2.
5938 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5939 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5940 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5941 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5942 (ix86_valid_target_attribute_inner_p): Ditto.
5943 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5944 * config/i386/i386.opt (mavx512vbmi2): New option.
5945 * doc/invoke.texi: Add new option.
5947 2017-11-16 Julia Koval <julia.koval@intel.com>
5949 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5950 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5951 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5952 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5953 _mm512_gf2p8mul_epi8): New intrinsics.
5954 * config/i386/i386-builtin-types.def
5955 (V64QI_FTYPE_V64QI_V64QI): New type.
5956 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5957 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5958 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5959 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5960 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5961 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5963 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5965 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5966 explicitly as a stream of bytes.
5968 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5970 * config/rs6000/altivec.h (vec_xst_be): New #define.
5971 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5972 and externalize from *altivec_vperm_<mode>_internal.
5973 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5975 (XL_BE_V8HI): Likewise.
5976 (XL_BE_V4SI): Likewise.
5977 (XL_BE_V4SI): Likewise.
5978 (XL_BE_V2DI): Likewise.
5979 (XL_BE_V4SF): Likewise.
5980 (XL_BE_V2DF): Likewise.
5981 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5982 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5983 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5984 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5985 VSX_BUILTIN_VEC_XST_BE.
5986 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5987 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5989 (altivec_init_builtins): Replace conditional calls to def_builtin
5990 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5991 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5992 with unconditional calls. Remove calls to def_builtin for
5993 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5994 __builtin_vec_xst_be.
5995 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5996 to define_expand, and add alternate RTL generation for P8.
5997 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5998 vsx_ld_elemrev_v8hi.
5999 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6000 add alternate RTL generation for P8.
6001 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6002 vsx_ld_elemrev_v16qi.
6003 (vsx_st_elemrev_v8hi): Convert define_insn
6004 to define_expand, and add alternate RTL generation for P8.
6005 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6006 vsx_st_elemrev_v8hi.
6007 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6008 add alternate RTL generation for P8.
6009 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6010 vsx_st_elemrev_v16qi.
6012 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6015 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6016 TARGET_AVX512ER check.
6017 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6018 neither -mzeroupper nor -mno-zeroupper is used and
6019 TARGET_EMIT_VZEROUPPER is set.
6020 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6021 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6023 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6025 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6026 folding of vector compares.
6027 (fold_build_vec_cmp): New helper function.
6028 (fold_compare_helper): New helper function.
6029 (builtin_function_type): Add compare builtins to the list of functions
6030 having unsigned arguments. Cosmetic updates to comment indentation.
6031 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6032 the not+eq combination.
6034 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6036 PR tree-optimization/82726
6037 PR tree-optimization/70754
6038 * tree-predcom.c (order_drefs_by_pos): New function.
6039 (combine_chains): Move code setting has_max_use_after to...
6040 (try_combine_chains): ...here. New parameter. Sort combined chains
6041 according to position information.
6042 (tree_predictive_commoning_loop): Update call to above function.
6043 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6045 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6047 PR tree-optimization/82726
6049 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6051 PR tree-optimization/70754
6052 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6053 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6054 combined stmt before it if not NULL.
6055 (combine_chains): Process refs reversely and compute dominance point
6059 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6061 PR tree-optimization/79663
6062 * tree-predcom.c (combine_chains): Process refs in reverse order
6063 only for ZERO length chains, and add explaining comment.
6065 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6067 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6068 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6069 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6071 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6073 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6075 2017-11-15 Martin Liska <mliska@suse.cz>
6077 * tree-cfg.c (pass_warn_function_return::execute):
6078 Compare warn_return_type for greater than zero.
6080 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6084 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6085 to return true on Xeon and not on Xeon Phi.
6086 (ix86_check_avx256_register): Changed to ...
6087 (ix86_check_avx_upper_register): ... this. Add extra check for
6088 VALID_AVX512F_REG_OR_XI_MODE.
6089 (ix86_avx_u128_mode_needed): Changed
6090 ix86_check_avx256_register to ix86_check_avx_upper_register.
6091 (ix86_check_avx256_stores): Changed to ...
6092 (ix86_check_avx_upper_stores): ... this. Changed
6093 ix86_check_avx256_register to ix86_check_avx_upper_register.
6094 (ix86_avx_u128_mode_after): Changed
6095 avx_reg256_found to avx_upper_reg_found. Changed
6096 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6097 (ix86_avx_u128_mode_entry): Changed
6098 ix86_check_avx256_register to ix86_check_avx_upper_register.
6099 (ix86_avx_u128_mode_exit): Ditto.
6100 * config/i386/i386.h: (host_detect_local_cpu): New define.
6102 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6104 * config/arm/xgene1.md (xgene1): Split into automatons
6105 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6106 (xgene1_f_load): Adjust reservations and/or types.
6107 (xgene1_f_store): Likewise.
6108 (xgene1_load_pair): Likewise.
6109 (xgene1_store_pair): Likewise.
6110 (xgene1_fp_load1): Likewise.
6111 (xgene1_load1): Likewise.
6112 (xgene1_store1): Likewise.
6113 (xgene1_move): Likewise.
6114 (xgene1_alu): Likewise.
6115 (xgene1_simd): Likewise.
6116 (xgene1_bfm): Likewise.
6117 (xgene1_neon_load1): Likewise.
6118 (xgene1_neon_store1): Likewise.
6119 (xgene1_neon_logic): Likewise.
6120 (xgene1_neon_st1): Likewise.
6121 (xgene1_neon_ld1r): Likewise.
6122 (xgene1_alu_cond): Added.
6123 (xgene1_shift_reg): Likwise.
6124 (xgene1_bfx): Likewise.
6125 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6127 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6130 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6132 (can_widen_mult_without_libcall): New function.
6133 (expand_mul_overflow): If only checking unsigned mul overflow,
6134 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6135 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6136 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6137 (expand_DIVMOD): Formatting fix.
6138 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6139 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6140 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6142 PR tree-optimization/82977
6143 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6144 constructed temporary to strlen_to_stridx.put.
6146 2017-11-15 Martin Liska <mliska@suse.cz>
6148 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6149 * configure: Regenerate.
6151 2017-11-15 Martin Liska <mliska@suse.cz>
6154 * config/sh/sh-mem.cc: Use proper probability for
6157 2017-11-14 Jeff Law <law@redhat.com>
6159 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6160 the red zone for stack_clash_protection_final_dynamic_probe targets
6161 when the total dynamic stack size is zero bytes.
6163 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6164 blocks is post order.
6166 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6168 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6169 * final.c (rest_of_clean_state): Set it for the
6170 -fcompare-debug dump.
6171 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6172 class when TDF_COMPARE_DEBUG is set.
6174 * dwarf2out.c (gen_producer_string): Discard
6177 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6180 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6181 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6182 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6183 (__TGMATH_REAL_2_3): Remove macros.
6184 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6185 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6187 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6189 (remquo): Define using __TGMATH_REAL_3.
6191 2017-11-14 Jeff Law <law@redhat.com>
6193 * vr-values.c: New file with contents extracted from tree-vrp.c.
6194 * Makefile.in (OBJS): Add vr-values.o
6195 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6196 (set_value_range, set_and_canonicalize_value_range): Likewise.
6197 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6198 (value_range_constant_singleton, symbolic_range_p): Likewise.
6199 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6200 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6201 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6202 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6203 (find_case_label_range, find_case_label_index): Likewise.
6204 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6205 (range_int_cst_singleton_p, value_inside_range): Likewise.
6206 (get_single_symbol): Likewise.
6207 (switch_update): Move structure definition here.
6208 (to_remove_edges, to_update_switch_stmts): Provide externs.
6209 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6210 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6211 (vrp_val_is_min, set_value_range): Likewise.
6212 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6213 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6214 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6215 (range_is_nonnull, range_int_cst_p): Likewwise.
6216 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6217 (get_single_symbol, operand_less_p): Likewise
6218 (compare_values_warnv, compare_values): Likewise.
6219 (value_inside_range, value_range_constant_singleton): Likewise.
6220 (zero_nonzero_bitgs_from_vr): Likewise.
6221 (extract_range_from_binary_expr_1): Likewise.
6222 (overflow_comparison_p): Likewise.
6223 (to_remove_edges, to_update_switch_stmts): Likewise.
6224 (find_case_label-index, find_case_label_range): Likewise.
6225 (switch_update, set_value_range_to_nonnegative): Remove.
6226 (set_value_range_to_truthvalue): Likewise.
6227 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6228 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6229 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6230 (find_case_label_ranges, test_for_singularity): Likewise.
6231 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6232 (x_vr_values): Move to its remaining use site.
6234 2017-11-10 Jeff Law <law@redhat.com>
6236 * vr-values.h (VR_INITIALIZER): Move #define here.
6237 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6238 * Makefile.in (OBJS): Add tree-evrp.o
6239 * tree-vrp.h (assert_info): Move structure definition here.
6240 (set_value_range_to_varying): Prototype.
6241 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6242 (infer_value_range, register_edge_assert_for): Likewise.
6243 (stmt_interesting_for_vrp): Likewise.
6244 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6245 (set_value_range_to_varying): No longer static.
6246 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6247 (infer_value_range, register_edge_assert_for): Likewise.
6249 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6251 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6252 generate the XXBRD instruction.
6254 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6255 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6256 as being compatible if -mabi=ieeelongdouble.
6257 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6258 to setup float128 built-ins with hardware support.
6259 (BU_FLOAT128_HW_2): Likewise.
6260 (BU_FLOAT128_HW_3): Likewise.
6261 (BU_FLOAT128_HW_VSX_1): Likewise.
6262 (BU_FLOAT128_HW_VSX_2): Likewise.
6263 (scalar_extract_expq): Change float128 built-in functions to
6264 accommodate having both KFmode and TFmode functions. Use the
6265 KFmode variant as the default.
6266 (scalar_extract_sigq): Likewise.
6267 (scalar_test_neg_qp): Likewise.
6268 (scalar_insert_exp_q): Likewise.
6269 (scalar_insert_exp_qp): Likewise.
6270 (scalar_test_data_class_qp): Likewise.
6271 (sqrtf128_round_to_odd): Delete processing the round to odd
6272 built-in functions as special built-in functions, and define them
6273 as float128 built-ins. Use the KFmode variant as the default.
6274 (truncf128_round_to_odd): Likewise.
6275 (addf128_round_to_odd): Likewise.
6276 (subf128_round_to_odd): Likewise.
6277 (mulf128_round_to_odd): Likewise.
6278 (divf128_round_to_odd): Likewise.
6279 (fmaf128_round_to_odd): Likewise.
6280 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6281 support for KFmode and TFmode xststdcqp calls.
6282 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6283 point, switch the built-in handlers for the get/set float128
6284 exponent, get float128 mantissa, float128 test built-ins, and the
6285 float128 round to odd built-in functions. Eliminate creating the
6286 float128 round to odd built-in functions as special built-ins.
6287 (rs6000_init_builtins): Eliminate special creation of the float128
6288 round to odd built-in functions.
6289 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6290 function insns to support both TFmode and KFmode variants.
6291 (xsxsigqp_<mode>): Likewise.
6292 (xsiexpqpf_<mode>): Likewise.
6293 (xsiexpqp_<mode>): Likewise.
6294 (xststdcqp_<mode>): Likewise.
6295 (xststdcnegqp_<mode>): Likewise.
6296 (xststdcqp_<mode>): Likewise.
6298 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6300 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6301 unused path_in_freq_ptr parameter.
6302 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6304 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6306 * ipa-inline.c (edge_badness): Dump sreal frequency.
6307 (compute_inlined_call_time): Match natural implementaiton ...
6308 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6309 forgotten division by CGRAPH_FREQ_BASE.
6311 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6313 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6314 Solaris 11. Update comment.
6315 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6317 * config/sol2.h (STARTFILE_SPEC): Likewise.
6318 * configure: Regenerate.
6320 2017-11-14 Carl Love <cel@us.ibm.com>
6322 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6323 le_ and be_ prefixes to swap* variables. Remove
6324 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6327 2017-11-14 Jason Merrill <jason@redhat.com>
6329 Support GTY((cache)) on hash_map.
6330 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6331 (ggc_cache_remove): Override it instead of ggc_mx.
6332 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6333 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6334 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6335 (simple_cache_map_traits): Override maybe_mx.
6336 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6337 (hash_map): Friend gt_cleare_cache.
6338 (gt_cleare_cache): New.
6339 * tree.h (tree_cache_traits): New hash_map traits class.
6340 (tree_cache_map): New typedef.
6342 2017-11-14 Richard Biener <rguenther@suse.de>
6344 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6345 paramter and handling.
6346 (cleanup_control_flow_bb): Likewise.
6347 (cleanup_control_flow_pre): New helper performing a DFS walk
6348 to call cleanup_control_flow_bb in PRE order.
6349 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6350 via cleanup_control_flow_pre.
6352 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6354 * config/aarch64/aarch64-simd.md
6355 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6356 (*aarch64_simd_bsl<mode>_alt): Likewise.
6357 (aarch64_simd_bsldi_internal): New.
6358 (aarch64_simd_bsldi_alt): Likewise.
6360 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6362 * tracer.c (better_p): Do not compare frequencies.
6363 * reg-stack.c (better_edge): Likewise.
6364 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6367 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6369 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6370 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6371 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6372 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6373 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6374 * predict.c (maybe_hot_frequency_p): Inline to ...
6375 (maybe_hot_count_p): ... here; rewrite to counts.
6376 (counts_to_freqs): Rename to ...
6377 (update_max_bb_count): ... this one.
6378 (expensive_function_p): Use counts.
6379 (estimate_bb_frequencies): Update.
6380 (rebuild_frequencies): Update.
6381 * predict.h (counts_to_freqs): Rename to ...
6382 (update_max_bb_count): ... this one.
6383 * profile.c (compute_branch_probabilities): Add debug info
6384 * tree-inline.c (expand_call_inline): Update debug info.
6385 (optimize_inline_calls): Use update_max_bb_count..
6386 (tree_function_versioning): Use update_max_bb_count..
6387 * value-prof.c (gimple_value_profile_transformations):
6388 Do not use update_max_bb_count.
6390 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6392 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6393 always use frequencies.
6395 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6397 * bb-reorder.c: Remove frequencies from comments.
6398 (better_edge_p): Use profile counts.
6399 (find_traces): Dump profile counts.
6400 (rotate_loop): Use profile counts.
6401 (find_traces_1_round): Likewise.
6402 (connect_better_edge_p): Use counts instead of probabilities for
6404 (copy_bb_p): Drop early check for non-0 frequency.
6405 (sanitize_hot_paths): Update comments.
6407 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6409 * ipa-split.c (struct split_point): Add count.
6410 (consider_split): Do not compute incoming frequency; compute incoming
6411 count and store it to split_point.
6412 (split_function): Set count of the call to split part correctly.
6414 2017-11-13 Carl Love <cel@us.ibm.com>
6416 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6418 2017-11-13 Tom Tromey <tom@tromey.com>
6420 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6422 2017-11-13 Carl Love <cel@us.ibm.com>
6424 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6425 Add support for builtins:
6426 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6427 vector {un,}signed {char,int,short},
6428 vector {un,}signed {char,int,short}) arguments.
6429 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6430 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6431 Add BU_P9V_AV_2 expansions for the builtins.
6432 * config/rs6000/altivec.h (vec_first_match_index,
6433 vec_first_mismatch_index, vec_first_match_or_eos_index,
6434 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6435 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6436 new extern declaration.
6437 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6438 * config/rs6000/vsx.md (first_match_index_<mode>,
6439 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6440 first_mismatch_or_eos_index_<mode>): Add define expand.
6441 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6442 * doc/extend.texi: Update the built-in documenation file for the new
6445 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6447 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6448 into the min/max operations for _Float<N> and _Float<N>X types.
6450 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6453 * dwarf2out.c (do_eh_frame): New static variable.
6454 (dwarf2out_begin_prologue): Set it.
6455 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6457 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6459 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6461 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6463 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6466 * debug.h (dwarf2out_do_eh_frame): Declare.
6467 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6468 (dwarf2out_do_frame): Use it.
6469 (dwarf2out_do_cfi_asm): Likewise.
6470 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6471 (dwarf2out_assembly_start): Likewise.
6472 (dwarf2out_begin_prologue): Fix comment.
6473 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6474 if the target needs either debug or unwind DWARF2 info.
6475 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6476 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6477 -ftrapping-math, -ftrapv and -fwrapv.
6479 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6481 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6482 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6483 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6484 (estimate_edge_size_and_time): Likewise.
6485 (ipa_merge_fn_summary_after_inlining): Likewise.
6486 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6487 (compute_uninlined_call_time): Use sreal_frequency.
6488 (compute_inlined_call_time): Likewise.
6489 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6490 * profile-count.c: Include sreal.h.
6491 (profile_count::to_sreal_scale): New.
6492 * profile-count.h: Forward declare sreal.
6493 (profile_count::to_sreal_scale): Declare.
6495 2017-11-13 Nathan Sidwell <nathan@acm.org>
6497 * diagnostic.c (maybe_line_and_column): New.
6498 (diagnostic_get_location_text): Use it.
6499 (diagnostic_report_current_module): Likewise.
6500 (test_diagnostic_get_location_text): Add tests.
6502 2017-11-13 Luis Machado <luis.machado@linaro.org>
6504 * doc/md.texi (Specifying processor pipeline description): Fix
6505 incorrect latency for the div instruction example.
6507 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6509 PR tree-optimization/78821
6510 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6511 that bit_not_p is the same.
6512 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6513 (split_group): Count precisely bit_not_p bits in each statement.
6514 (invert_op): New function.
6515 (imm_store_chain_info::output_merged_store): Use invert_op to
6516 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6517 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6519 2017-11-13 Martin Liska <mliska@suse.cz>
6521 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6522 (struct source_info): Likewise.
6523 (add_branch_counts): Likewise.
6524 (add_line_counts): Likewise.
6525 (function_summary): Likewise.
6526 (output_intermediate_line): Likewise.
6527 (generate_results): Likewise.
6529 2017-11-13 Martin Liska <mliska@suse.cz>
6531 * gcov.c (struct block_info): Remove typedef for block_t.
6532 (struct line_info): Likewise.
6533 (line_info::has_block): Likewise.
6534 (EXIT_BLOCK): Likewise.
6535 (unblock): Likewise.
6536 (circuit): Likewise.
6537 (get_cycles_count): Likewise.
6538 (process_file): Likewise.
6539 (read_graph_file): Likewise.
6540 (solve_flow_graph): Likewise.
6541 (find_exception_blocks): Likewise.
6542 (add_line_counts): Likewise.
6543 (accumulate_line_info): Likewise.
6544 (output_line_details): Likewise.
6546 2017-11-13 Martin Liska <mliska@suse.cz>
6548 * gcov.c (struct arc_info): Remove typedef for arc_t.
6549 (struct line_info): Likewise.
6550 (add_branch_counts): Likewise.
6551 (output_branch_count): Likewise.
6552 (function_info::~function_info): Likewise.
6553 (circuit): Likewise.
6554 (output_intermediate_line): Likewise.
6555 (read_graph_file): Likewise.
6556 (solve_flow_graph): Likewise.
6557 (find_exception_blocks): Likewise.
6558 (add_line_counts): Likewise.
6559 (accumulate_line_info): Likewise.
6560 (output_line_details): Likewise.
6561 (output_function_details): Likewise.
6563 2017-11-13 Martin Liska <mliska@suse.cz>
6565 * gcov.c (struct function_info): Remove typedef for function_t.
6566 (struct source_info): Likewise.
6567 (source_info::get_functions_at_location): Likewise.
6568 (solve_flow_graph): Likewise.
6569 (find_exception_blocks): Likewise.
6570 (add_line_counts): Likewise.
6571 (output_intermediate_file): Likewise.
6572 (process_file): Likewise.
6573 (generate_results): Likewise.
6574 (release_structures): Likewise.
6575 (read_graph_file): Likewise.
6576 (read_count_file): Likewise.
6577 (accumulate_line_counts): Likewise.
6578 (output_lines): Likewise.
6580 2017-11-13 Martin Liska <mliska@suse.cz>
6582 * gcov.c (function_info::function_info): Remove num_counts
6583 and add vector<gcov_type>.
6584 (function_info::~function_info): Use the vector.
6585 (process_file): Likewise.
6586 (read_graph_file): Likewise.
6587 (read_count_file): Likewise.
6588 (solve_flow_graph): Likewise.
6590 2017-11-13 Martin Liska <mliska@suse.cz>
6592 * gcov.c (function_info::is_artificial): New function.
6593 (process_file): Erase all artificial early.
6594 (generate_results): Skip as all artificial are already
6597 2017-11-13 Martin Liska <mliska@suse.cz>
6599 * gcov.c (read_graph_file): Store to global vector of functions.
6600 (read_count_file): Iterate the vector.
6601 (process_file): Likewise.
6602 (generate_results): Likewise.
6603 (release_structures): Likewise.
6605 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6607 PR tree-optimization/82954
6608 * gimple-ssa-store-merging.c
6609 (imm_store_chain_info::coalesce_immediate_stores): If
6610 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6612 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6614 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6615 Upddate call to ENDIAN_LANE_N.
6616 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6617 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6618 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6619 and use aarch64_endian_lane_rtx.
6620 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6622 2017-11-12 Tom de Vries <tom@codesourcery.com>
6624 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6626 2017-11-12 Tom de Vries <tom@codesourcery.com>
6628 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6630 2017-11-12 Tom de Vries <tom@codesourcery.com>
6632 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6634 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6635 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6636 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6638 2017-11-11 Martin Sebor <msebor@redhat.com>
6641 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6644 * prefic.c (translate_name): Replace strncpy with memcpy to
6645 avoid -Wstringop-truncation.
6647 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6649 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6651 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6653 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6655 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6657 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6660 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6662 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6663 enable generating XXBRH if the value is in a vector register.
6664 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6665 value is in a vector register.
6666 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6667 register to register bswap64's instead of doing the GPR sequence
6668 used on previous machines.
6669 (bswapdi2_xxbrd): New insn.
6670 (bswapdi2_reg): Disallow on ISA 3.0.
6671 (register to register bswap64 splitter): Do not split the insn on
6672 ISA 3.0 systems that use XXBRD.
6674 2017-11-10 Martin Sebor <msebor@redhat.com>
6677 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6679 (find_subframework_file): Same.
6681 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6683 * auto-profile.c (afdo_indirect_call): Drop frequency.
6684 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6685 (cgraph_node::create_edge): Drop frequency argument.
6686 (cgraph_node::create_indirect_edge): Drop frequency argument.
6687 (cgraph_edge::make_speculative): Drop frequency arguments.
6688 (cgraph_edge::resolve_speculation): Do not update frequencies
6689 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6690 (cgraph_node::dump): Check consistency in IPA mode.
6691 (cgraph_edge::maybe_hot_p): Use IPA counter.
6692 (cgraph_edge::verify_count_and_frequency): Rename to ...
6693 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6694 (cgraph_node::verify_node): Update.
6695 * cgraph.h (struct cgraph_edge): Drop frequency.
6696 (cgraph_edge::frequency): New function.
6697 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6699 (cgraph_edge::rebuild_edges): Likewise.
6700 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6701 (duplicate_thunk_for_node): Do not pass frequency.
6702 (cgraph_node::create_clone): Scale only counts.
6703 (cgraph_node::create_virtual_clone): Do not pass frequency.
6704 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6705 (cgraph_node::create_version_clone): Do not pass frequency.
6706 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6707 (cgraph_node::expand_thunk): Do not pass frequency.
6708 (cgraph_node::create_wrapper): Do not pass frequency.
6709 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6711 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6712 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6713 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6714 (gather_caller_stats): Use frequency function.
6715 (ipcp_cloning_candidate_p): Use frequency function.
6716 (ipcp_propagate_stage): Use frequency function.
6717 (get_info_about_necessary_edges): Use frequency function.
6718 (update_profiling_info): Update only IPA profile.
6719 (update_specialized_profile): Use frequency functoin.
6720 (perhaps_add_new_callers): Update only IPA profile.
6721 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6722 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6723 (dump_ipa_call_summary): Use frequency function.
6724 (estimate_edge_size_and_time): Use frequency function.
6725 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6726 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6727 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6728 (update_noncloned_counts): ... ths one; scale counts only.
6729 (clone_inlined_nodes): Do not scale frequency.
6730 (inline_call): Do not pass frequency.
6731 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6732 (compute_inlined_call_time): Use IPA profile.
6733 (want_inline_small_function_p): Use IPA profile.
6734 (want_inline_self_recursive_call_p): Use IPA profile.
6735 (edge_badness): Use IPA profile.
6736 (lookup_recursive_calls): Use IPA profile.
6737 (recursive_inlining): Do not pass frequency.
6738 (resolve_noninline_speculation): Do not update frequency.
6739 (inline_small_functions): Collect max of IPA profile.
6740 (dump_overall_stats): Dump IPA porfile.
6741 (dump_inline_stats): Dump IPA porfile.
6742 (ipa_inline): Collect IPA stats.
6743 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6744 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6745 (ipa_propagate_frequency): Use frequency function.
6746 (ipa_profile): Cleanup.
6747 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6748 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6749 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6750 (input_node): Do not stream frequency.
6751 (input_edge): Do not stream frequency.
6752 (merge_profile_summaries): Scale only IPA profiles.
6753 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6754 * predict.c (drop_profile): Do not recompute frequency.
6755 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6756 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6757 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6758 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6759 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6760 * tree-inline.c (copy_bb): Do not scale frequency.
6761 (expand_call_inline): Do not scale frequency.
6762 (tree_function_versioning): Do not scale frequency.
6763 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6765 2017-11-10 Julia Koval <julia.koval@intel.com>
6767 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6768 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6769 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6770 (_mm256_maskz_gf2p8affine_epi64_epi8)
6771 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6772 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6773 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6774 (__builtin_ia32_vgf2p8affineqb_v32qi)
6775 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6776 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6778 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6782 (arm_option_override): Refactor.
6783 (arm_option_reconfigure_globals): New.
6784 (arm_options_perform_arch_sanity_checks): New.
6785 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6787 (arm_options_perform_arch_sanity_checks): Likewise
6789 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6791 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6792 (power9-qpmul): New.
6793 * rs6000/rs6000.md ("type" attr): Add qmul.
6794 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6795 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6796 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6798 2017-11-10 Martin Sebor <msebor@redhat.com>
6801 * builtins.c (compute_objsize): Handle arrays that
6802 compute_builtin_object_size likes to fail for. Make extern.
6803 * builtins.h (compute_objsize): Declare.
6804 (check_strncpy_sizes): New function.
6805 (expand_builtin_strncpy): Call check_strncpy_sizes.
6806 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6807 -Wstringop-truncation.
6808 (gimple_fold_builtin_strncat): Same.
6809 * gimple.c (gimple_build_call_from_tree): Set call location.
6810 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6811 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6812 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6813 (handle_builtin_strlen): Use strlen_to_stridx.
6814 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6816 Use strlen_to_stridx.
6817 (pass_strlen::execute): Release strlen_to_stridx.
6818 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6819 (-Wstringop-truncation): Document new option.
6821 2017-11-10 Martin Liska <mliska@suse.cz>
6823 PR gcov-profile/82702
6824 * gcov.c (main): Handle intermediate files in a different
6826 (get_gcov_intermediate_filename): New function.
6827 (output_gcov_file): Remove support of intermediate files.
6828 (generate_results): Allocate intermediate file.
6829 (release_structures): Clean-up properly fn_end.
6830 (output_intermediate_file): Start iterating with line 1.
6832 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6834 PR tree-optimization/82929
6835 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6836 ops_swapped_p non-static data member.
6837 (store_immediate_info::store_immediate_info): Clear it.
6838 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6839 ops set ops_swapped_p.
6840 (count_multiple_uses): Handle ops_swapped_p.
6842 2017-11-10 Martin Liska <mliska@suse.cz>
6844 * coverage.c (coverage_init): Stream information about
6845 support of has_unexecuted_blocks.
6846 * doc/gcov.texi: Document that.
6847 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6848 * gcov.c (read_graph_file): Likewise.
6849 (output_line_beginning): Fix a small issue with
6852 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6854 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6855 reference of trivial component.
6857 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6860 * gimple-ssa-store-merging.c
6861 (pass_store_merging::terminate_all_aliasing_chains): For
6862 gimple_store_p stmts also call refs_output_dependent_p.
6864 PR rtl-optimization/82913
6865 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6868 2017-11-09 Jeff Law <law@redhat.com>
6870 * vr-values.h: New file with vr_values class.
6871 * tree-vrp.c: Include vr-values.h
6872 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6873 data objects into the vr_values class.
6874 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6875 (get_value_range): Make it a member function within vr_values class.
6876 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6877 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6878 (op_with_constant_singleton_value_range): Likewise.
6879 (extract_range_for_var_from_comparison_expr): Likewise.
6880 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6881 (extract_range_from_binary_expr): Likewise.
6882 (extract_range_from_unary_expr): Likewise.
6883 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6884 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6885 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6886 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6887 (compare_name_with_value, compare_names): Likewise.
6888 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6889 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6890 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6891 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6892 (extract_range_from_phi_node): Likewise.
6893 (simplify_truth_ops_using_ranges): Likewise.
6894 (simplify_div_or_mod_using_ranges): Likewise.
6895 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6896 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6897 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6898 (simplify_float_conversion_using_ranges): Likewise.
6899 (simplify_internal_call_using_ranges): Likewise.
6900 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6901 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6902 poitner via x_vr_values for calls into gimple folder.
6903 (vrp_initialize_lattice): Make this the vr_values ctor.
6904 (vrp_free_lattice): Make this the vr_values dtor.
6905 (set_vr_value): New function.
6906 (class vrp_prop): Add vr_values data member. Add various member
6907 functions as well as member functions that delegate to vr_values.
6908 (check_array_ref): Make a member function within vrp_prop class.
6909 (search_for_addr_array, vrp_initialize): Likewise.
6910 (vrp_finalize): Likewise. Revamp to avoid direct access to
6911 vr_value, values_propagated, etc.
6912 (check_array_bounds): Extract vrp_prop class instance pointer from
6913 walk info structure. Use it to call member functions.
6914 (check_all_array_refs): Make a member function within vrp_prop class.
6915 Smuggle class instance pointer via walk info structure.
6916 (x_vr_values): New local static.
6917 (vrp_valueize): Use x_vr_values to get class instance.
6918 (vr_valueize_1): Likewise.
6919 (class vrp_folder): Add vr_values data member. Add various member
6920 functions as well as member functions that delegate to vr_values.
6921 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6922 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6923 class instance from vr_values. Use it to call member functions.
6924 (vrp_dom_walker): Add vr_values data member.
6925 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6926 instance via x_vr_values.
6927 (identify_jump_threads): Accept vr_values as argument. Store
6928 it into the walker structure.
6929 (evrp_dom_walker): Add vr_values class data member. Add various
6931 (evrp_dom_walker::try_find_new_range): Use vr_values data
6932 member to access the memory allocator.
6933 (evrp_dom_walker::before_dom_children): Store vr_values class
6934 instance into the vrp_folder class.
6935 (evrp_dom_walker::push_value_range): Rework to avoid direct
6936 access to num_vr_values and vr_value.
6937 (evrp_dom_walker::pop_value_range): Likewise.
6938 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6939 Use vr_values to get to dump_all_value_ranges member function.
6940 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6941 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6942 Pass vr_values class instance down to identify_jump_threads.
6943 Remove call to vrp_free_lattice.
6944 (debug_all_value_ranges): Remove.
6946 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6947 (vrp_folder): Likewise.
6949 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6950 Get it from the existing bitmap instead.
6951 (vrp_intersect_ranges_1): Likewise.
6953 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6955 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6957 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6958 set bit_not_p to it.
6959 (imm_store_chain_info::coalesce_immediate_stores): Break group
6960 if bit_not_p is different.
6961 (count_multiple_uses, split_group,
6962 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6963 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6964 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6965 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6966 to store_immediate_info ctor.
6968 2017-11-09 Jim Wilson <jimw@sifive.com>
6970 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6971 (scan_prog_file): Likewise.
6973 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6975 * bb-reorder.c (max_entry_frequency): Remove.
6976 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6977 connect_traces, push_to_next_round_p): Remove prototypes.
6978 (find_traces_1_round): Use counts only.
6979 (push_to_next_round_p): Likewise.
6980 (find_traces): Likewise.
6981 (rotate_loop): Likewise.
6982 (find_traces_1_round): Likewise.
6983 (connect_traces): Likewise.
6984 (edge_order): Likewise.
6986 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6988 * config/arm/arm.c (output_return_instruction): Add comments to
6989 indicate requirement for cmse_nonsecure_entry return to account
6990 for the size of clearing instruction output here.
6991 (thumb_exit): Likewise.
6992 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6993 return in hardfloat mode.
6995 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6997 * config/rs6000/rs6000.c (machine_function): Add a bool,
6998 "toc_is_wrapped_separately".
6999 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7000 if it wasn't explicitly set or unset, we are optimizing for speed, and
7001 doing separate shrink-wrapping.
7002 (rs6000_get_separate_components): Enable the TOC component if
7003 saving the TOC register in the prologue.
7004 (rs6000_components_for_bb): Handle the TOC component.
7005 (rs6000_emit_prologue_components): Store the TOC register where needed.
7006 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7007 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7009 2017-11-09 Martin Jambor <mjambor@suse.cz>
7011 * ipa-param-manipulation.c: New file.
7012 * ipa-param-manipulation.h: Likewise.
7013 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7014 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7015 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7016 (ipa_parm_adjustment): Likewise.
7017 (ipa_parm_adjustment_vec): Likewise.
7018 (ipa_get_vector_of_formal_parms): Moved declaration to
7019 ipa-param-manipulation.h.
7020 (ipa_get_vector_of_formal_parm_types): Likewise.
7021 (ipa_modify_formal_parameters): Likewise.
7022 (ipa_modify_call_arguments): Likewise.
7023 (ipa_combine_adjustments): Likewise.
7024 (ipa_dump_param_adjustments): Likewise.
7025 (ipa_modify_expr): Likewise.
7026 (ipa_get_adjustment_candidate): Likewise.
7027 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7028 ipa-param-manipulation.c.
7029 (ipa_get_vector_of_formal_parm_types): Likewise.
7030 (ipa_modify_formal_parameters): Likewise.
7031 (ipa_modify_call_arguments): Likewise.
7032 (ipa_modify_expr): Likewise.
7033 (get_ssa_base_param): Likewise.
7034 (ipa_get_adjustment_candidate): Likewise.
7035 (index_in_adjustments_multiple_times_p): Likewise.
7036 (ipa_combine_adjustments): Likewise.
7037 (ipa_dump_param_adjustments): Likewise.
7038 * tree-sra.c: Also include ipa-param-manipulation.h
7039 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7042 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7043 Alan Hayward <alan.hayward@arm.com>
7044 David Sherwood <david.sherwood@arm.com>
7046 * doc/sourcebuild.texi (vect_masked_store): Document.
7048 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7049 Alan Hayward <alan.hayward@arm.com>
7050 David Sherwood <david.sherwood@arm.com>
7052 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7054 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7055 Alan Hayward <alan.hayward@arm.com>
7056 David Sherwood <david.sherwood@arm.com>
7058 * doc/sourcebuild.texi (vect_variable_length): Document.
7060 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7061 Alan Hayward <alan.hayward@arm.com>
7062 David Sherwood <david.sherwood@arm.com>
7064 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7066 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7067 Alan Hayward <alan.hayward@arm.com>
7068 David Sherwood <david.sherwood@arm.com>
7070 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7072 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7073 Alan Hayward <alan.hayward@arm.com>
7074 David Sherwood <david.sherwood@arm.com>
7076 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7077 previously undocumented selectors.
7078 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7080 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7082 * doc/rtl.texi (const_vector): Say that elements can be
7083 const_wide_ints too.
7084 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7085 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7086 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7087 * optabs.c (expand_vector_broadcast): Likewise.
7089 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7090 Alan Hayward <alan.hayward@arm.com>
7091 David Sherwood <david.sherwood@arm.com>
7093 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7094 scaled index even if the unscaled address was invalid.
7095 Don't increase the complexity of using a scale in that case.
7097 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7098 Alan Hayward <alan.hayward@arm.com>
7099 David Sherwood <david.sherwood@arm.com>
7101 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7102 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7103 UNITS_PER_WORD bytes.
7104 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7105 based on REGMODE_NATURAL_SIZE of the inner mode.
7106 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7107 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7108 * expmed.c (lowpart_bit_field_p): Divide the value up into
7109 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7110 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7111 whether something is likely to occupy more than one register.
7113 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7116 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7117 profile_count::adjust_for_ipa_scaling.
7118 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7119 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7121 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7123 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7125 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7126 (split_group): Add total_orig and total_new arguments, estimate the
7127 number of statements related to the store group without store merging
7128 and with store merging.
7129 (imm_store_chain_info::output_merged_store): Adjust split_group
7130 callers, punt if estimated number of statements with store merging
7131 is not smaller than estimated number of statements without it.
7133 (handled_load): Remove has_single_use checks.
7134 (pass_store_merging::process_store): Likewise.
7136 2017-11-09 Richard Biener <rguenther@suse.de>
7138 PR tree-optimization/82902
7139 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7141 2017-11-09 Martin Liska <mliska@suse.cz>
7144 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7147 2017-11-09 Martin Liska <mliska@suse.cz>
7149 PR tree-optimization/82669
7150 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7152 2017-11-09 Martin Liska <mliska@suse.cz>
7154 PR gcov-profile/48463
7155 * coverage.c (coverage_begin_function): Output also end locus
7156 of a function and information whether the function is
7158 * gcov-dump.c (tag_function): Parse and print the information.
7159 * gcov.c (INCLUDE_MAP): Add include.
7160 (INCLUDE_SET): Likewise.
7161 (struct line_info): Move earlier in the source file because
7162 of vector<line_info> in function_info structure.
7163 (line_info::line_info): Likewise.
7164 (line_info::has_block): Likewise.
7165 (struct source_info): Add new member index.
7166 (source_info::get_functions_at_location): New function.
7167 (function_info::group_line_p): New function.
7168 (output_intermediate_line): New function.
7169 (output_intermediate_file): Use the mentioned function.
7170 (struct function_start): New.
7171 (struct function_start_pair_hash): Likewise.
7172 (process_file): Add code that identifies group functions.
7173 Assign lines either to global or function scope.
7174 (generate_results): Skip artificial functions.
7175 (find_source): Assign index for each source file.
7176 (read_graph_file): Read new flag artificial and end_line.
7177 (add_line_counts): Assign it either to global of function scope.
7178 (accumulate_line_counts): Isolate core of the function to
7179 accumulate_line_info and call it for both function and global
7181 (accumulate_line_info): New function.
7182 (output_line_beginning): Fix GNU coding style.
7183 (print_source_line): New function.
7184 (output_line_details): Likewise.
7185 (output_function_details): Likewise.
7186 (output_lines): Iterate both source (global) scope and function
7188 (struct function_line_start_cmp): New class.
7189 * doc/gcov.texi: Reflect changes in documentation.
7191 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7194 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7195 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7196 and similarly for not instead of neg.
7198 2017-11-08 Andi Kleen <ak@linux.intel.com>
7200 * config/i386/i386.opt: Add -mforce-indirect-call.
7201 * config/i386/predicates.md: Check for flag_force_indirect_call.
7202 * doc/invoke.texi: Document -mforce-indirect-call
7204 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7206 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7208 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7209 config/riscv/riscv.c (predict.h): New include.
7210 (riscv_slow_unaligned_access_p): No longer static.
7211 (riscv_block_move_straight): Add require.
7212 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7214 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7217 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7218 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7219 nonimmediate_operand predicate for operand 1 instead of
7222 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7224 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7226 * config/aarch64/constraints.md (Uml): New constraint.
7227 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7230 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7232 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7233 of two vec_duplicates into a vec_concat.
7235 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7237 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7238 Simplify vec_merge of vec_duplicate and vec_concat.
7239 * config/aarch64/constraints.md (Utq): New constraint.
7240 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7243 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7245 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7246 Simplify vec_merge of vec_duplicate and const_vector.
7247 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7249 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7250 mode iterator. Update predicate on operand 1 to
7251 handle non-const_vec constants. Delete constraints.
7252 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7254 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7256 PR tree-optimization/78821
7257 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7259 (store_operand_info::store_operand_info): Initialize it to false.
7260 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7261 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7262 store in the group, and if chain_info is non-NULL, to ignore altogether
7264 (compatible_load_p): Fail if bit_not_p does not match.
7265 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7266 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7267 (pass_store_merging::process_store): Adjust
7268 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7269 call terminate_all_aliasing_chains newly when adding a store into
7270 a chain with non-NULL chain_info.
7272 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7274 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7276 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7278 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7280 (aarch64_layout_frame): Initialise emit_frame_chain.
7281 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7282 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7284 2017-11-08 Martin Liska <mliska@suse.cz>
7286 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7287 of gimple_call_internal_p.
7289 2017-11-07 Tom de Vries <tom@codesourcery.com>
7291 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7293 2017-11-07 Tom de Vries <tom@codesourcery.com>
7295 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7298 2017-11-08 Martin Liska <mliska@suse.cz>
7301 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7303 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7305 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7307 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7310 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7311 (*cmp<mode>_ccz_1): New insn with $k alternative.
7314 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7315 RTX_COMM_COMPARE as commutative as well.
7316 (ix86_binary_operator_ok): Formatting fix.
7317 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7318 *<code><mode>3<mask_name><round_saeonly_name>,
7319 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7320 *<s>mul<mode>3_highpart<mask_name>,
7321 *vec_widen_umult_even_v16si<mask_name>,
7322 *vec_widen_umult_even_v8si<mask_name>,
7323 *vec_widen_umult_even_v4si<mask_name>,
7324 *vec_widen_smult_even_v16si<mask_name>,
7325 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7326 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7327 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7328 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7329 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7330 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7331 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7332 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7333 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7334 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7335 ix86_binary_operator_ok. Formatting fixes.
7336 (*<plusminus_insn><mode>3<mask_name><round_name>,
7337 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7340 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7342 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7343 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7344 eq and ne if TARGET_ISEL.
7345 (cmp): New code_iterator.
7346 (UNS, UNSU_, UNSIK): New code_attrs.
7347 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7348 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7350 ("eq<mode>3"): ... this.
7351 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7353 ("ne<mode>3"): ... this.
7355 2017-11-07 Julia Koval <julia.koval@intel.com>
7358 * common/config/i386/i386-common.c
7359 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7360 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7361 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7362 * config/i386/i386.opt: Ditto.
7363 * config/i386/i386.c (ix86_target_string): Ditto.
7364 (ix86_option_override_internal): Ditto.
7365 (ix86_init_mpx_builtins): Move MPX to args2.
7366 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7367 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7368 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7369 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7370 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7371 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7372 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7374 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7377 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7378 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7379 (zero-extendsidi peephole2): Remove peephole.
7381 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7384 * stor-layout.c: Include attribs.h.
7385 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7386 explicit lookup of "aligned" attribute.
7388 2017-11-07 Andrew Waterman <andrew@sifive.com>
7390 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7391 (riscv_expand_block_move): Likewise.
7392 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7394 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7395 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7396 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7397 (riscv_adjust_block_mem): Likewise.
7398 (riscv_block_move_loop): Likewise.
7399 (riscv_expand_block_move): Likewise.
7400 * config/riscv/riscv.md (movmemsi): New pattern.
7402 2017-11-07 Michael Clark <michaeljclark@mac.com>
7404 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7405 (MUSL_DYNAMIC_LINKER): Likewise.
7407 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7409 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7410 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7413 2017-11-07 Richard Biener <rguenther@suse.de>
7415 * match.pd: Fix build.
7417 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7418 Jackson Woodruff <jackson.woodruff@arm.com>
7420 PR tree-optimization/71026
7421 * match.pd: Canonicalize negate in division.
7423 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7426 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7428 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7430 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7431 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7433 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7435 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7436 non-scalar integral types.
7437 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7438 (-(A-B), -(~A)): New transformations.
7440 2017-11-07 Tom de Vries <tom@codesourcery.com>
7442 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7443 semicolon after "do {} while (0)".
7444 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7445 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7446 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7447 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7448 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7450 2017-11-07 Tom de Vries <tom@codesourcery.com>
7452 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7453 after "do {} while (0)".
7454 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7455 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7456 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7457 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7458 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7460 2017-11-07 Tom de Vries <tom@codesourcery.com>
7463 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7465 (arm_rtx_costs_internal): Add missing semicolon after
7466 HANDLE_NARROW_SHIFT_ARITH call.
7468 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7470 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7471 disable isel if it was not set explicitly.
7473 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7475 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7476 (add_type_attribute) likewise.
7478 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7480 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7481 of struct ix86_frame.
7482 (ix86_initial_elimination_offset): Likewise.
7483 (ix86_expand_split_stack_prologue): Likewise.
7485 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7487 * tree-vrp.h (enum value_range_type): Update stale comment.
7489 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7490 Alan Hayward <alan.hayward@arm.com>
7491 David Sherwood <david.sherwood@arm.com>
7493 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7494 (aarch64_expand_vec_perm_const): Take the number of units too.
7495 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7496 (aarch64_expand_vec_perm_const): Likewise.
7497 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7498 (vec_perm<mode>): Update accordingly.
7500 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7501 Alan Hayward <alan.hayward@arm.com>
7502 David Sherwood <david.sherwood@arm.com>
7504 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7505 Take the number of units too.
7506 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7507 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7508 but check for a vector mode before rather than after the call.
7509 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7510 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7511 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7512 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7513 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7514 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7515 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7516 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7517 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7518 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7519 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7520 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7521 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7522 (aarch64_sqdmull2_n<mode>): Update accordingly.
7524 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7525 Alan Hayward <alan.hayward@arm.com>
7526 David Sherwood <david.sherwood@arm.com>
7528 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7529 the number of units too.
7530 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7531 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7532 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7533 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7534 (vec_store_lanesxi<mode>): Update accordingly.
7536 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7537 Alan Hayward <alan.hayward@arm.com>
7538 David Sherwood <david.sherwood@arm.com>
7540 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7541 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7542 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7543 of units rather than the mode.
7544 * config/aarch64/iterators.md (nunits): New mode attribute.
7545 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7546 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7547 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7548 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7549 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7550 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7551 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7552 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7553 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7554 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7555 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7556 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7557 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7558 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7559 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7560 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7561 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7562 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7563 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7564 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7565 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7566 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7567 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7568 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7569 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7570 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7571 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7572 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7573 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7574 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7575 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7576 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7577 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7578 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7579 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7580 (aarch64_simd_vec_setv2di): Likewise.
7582 2017-11-06 Carl Love <cel@us.ibm.com>
7584 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7586 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7587 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7589 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7591 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7592 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7593 Add power 8 macro expansions.
7594 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7595 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7596 power 8 instructions. (VSX_XXBR): Add iterator.
7598 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7600 * config/arm/arm.md (predicable_short_it): Change default to "no",
7601 improve documentation, remove uses that are identical to the default.
7602 (enabled_for_depr_it): Rename to enabled_for_short_it.
7603 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7604 Remove default uses.
7605 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7606 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7607 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7608 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7610 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7613 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7614 float128 helper macros, which are no longer used after deleting
7615 the old 'q' built-in functions, and moving the round to odd
7616 built-in functions to being special built-in functions.
7617 (BU_FLOAT128_2): Likewise.
7618 (BU_FLOAT128_1_HW): Likewise.
7619 (BU_FLOAT128_2_HW): Likewise.
7620 (BU_FLOAT128_3_HW): Likewise.
7621 (FABSQ): Delete old 'q' built-in functions.
7622 (COPYSIGNQ): Likewise.
7623 (SQRTF128_ODD): Move round to odd built-in functions to be
7624 special built-in functions, so that we can handle
7625 -mabi=ieeelongdouble.
7626 (TRUNCF128_ODD): Likewise.
7627 (ADDF128_ODD): Likewise.
7628 (SUBF128_ODD): Likewise.
7629 (MULF128_ODD): Likewise.
7630 (DIVF128_ODD): Likewise.
7631 (FMAF128_ODD): Likewise.
7632 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7633 built-in names to 'f128'.
7634 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7635 old 'q' built-in functions, as the machine independent code for
7636 'f128' built-in functions handles this.
7637 (rs6000_expand_builtin): Add expansion for float128 round to odd
7638 functions, keying off on -mabi=ieeelongdouble of whether to use
7639 the KFmode or TFmode variant.
7640 (rs6000_init_builtins): Initialize the _Float128 round to odd
7642 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7643 _Float128 'q' built-in functions are now mapped into the new
7644 'f128' built-in functions.
7646 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7648 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7649 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7651 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7654 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7655 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7656 * toplev.c: Include "ipa-fnsummary.h".
7657 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7659 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7661 PR tree-optimization/82838
7662 * gimple-ssa-store-merging.c
7663 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7664 on a separate gimple_seq which is then appended to seq.
7666 2017-11-06 Jeff Law <law@redhat.com>
7669 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7670 (get_probe_interval): New functions.
7671 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7672 (ix86_adjust_stack_and_probe): Likewise.
7673 (output_adjust_stack_and_probe): Likewise.
7674 (ix86_emit_probe_stack_range): Likewise.
7675 (ix86_expand_prologue): Likewise.
7677 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7679 PR tree-optimization/82816
7680 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7681 if the modes of the two types are the same.
7682 (convert_plusminus_to_widen): Likewise.
7684 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7686 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7688 (usadv16qi): New define_expand.
7689 (usadv8hi): New define_expand.
7691 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7694 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7696 (inline_transform): Likewise.
7697 * predict.c (counts_to_freqs): Remove useless conditional.
7698 * profile-count.h (profile_count::apply_scale): Move sanity check.
7699 * tree-inline.c (copy_bb): Always scale.
7700 (copy_cfg_body): Likewise.
7702 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7705 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7707 (*cmp_ite0): Add enabled_for_depr_it attribute.
7708 (*cmp_ite1): Likewise.
7710 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7712 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7715 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7717 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7718 return the value by pointer.
7719 (extract_range_from_multiplicative_op_1): Update accordingly.
7720 Return as soon as an operation fails.
7722 2017-11-05 Tom de Vries <tom@codesourcery.com>
7725 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7726 (DEF_SANITIZER_BUILTIN): ... here.
7727 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7728 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7730 2017-11-05 Tom de Vries <tom@codesourcery.com>
7733 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7735 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7736 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7737 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7739 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7740 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7741 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7743 2017-11-05 Tom de Vries <tom@codesourcery.com>
7746 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7749 2017-11-04 Michael Clark <michaeljclark@mac.com>
7751 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7752 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7754 (*addsi3_extended): Likewise.
7755 (*addsi3_extended2): Likewise.
7756 (<optab>si3): Likewise.
7757 (<optab>di3): Likewise.
7758 (<optab><mode>3): Likewise.
7759 (<*optabe>si3_internal): Likewise.
7760 (zero_extendqi<SUPERQI:mode>2): Likewise.
7761 (*add<mode>hi3): Likewise.
7762 (*xor<mode>hi3): Likewise.
7763 (<optab>di3): Likewise.
7764 (*<optab>si3_extend): Likewise.
7765 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7766 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7767 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7769 2017-11-04 Andrew Waterman <andrew@sifive.com>
7771 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7772 TARGET_STRICT_ALIGN based upon -mtune argument.
7774 2017-11-04 Andrew Waterman <andrew@sifive.com>
7776 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7778 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7780 * config/i386/i386.c (choose_basereg): Use optional scratch
7781 register and add assertion.
7782 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7783 needed, and don't allocate stack.
7784 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7785 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7786 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7788 2017-11-03 Jeff Law <law@redhat.com>
7790 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7791 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7792 to probe at the start of a noreturn function.
7794 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7796 PR tree-optimization/78821
7797 * gimple-ssa-store-merging.c: Update the file comment.
7798 (MAX_STORE_ALIAS_CHECKS): Define.
7799 (struct store_operand_info): New type.
7800 (store_operand_info::store_operand_info): New constructor.
7801 (struct store_immediate_info): Add rhs_code and ops data members.
7802 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7803 and op1r arguments to the ctor, initialize corresponding data members.
7804 (struct merged_store_group): Add load_align_base and load_align
7806 (merged_store_group::merged_store_group): Initialize them.
7807 (merged_store_group::do_merge): Update them.
7808 (merged_store_group::apply_stores): Pick the constant for
7809 encode_tree_to_bitpos from one of the two operands, or skip
7810 encode_tree_to_bitpos if neither operand is a constant.
7811 (class pass_store_merging): Add process_store method decl. Remove
7812 bool argument from terminate_all_aliasing_chains method decl.
7813 (pass_store_merging::terminate_all_aliasing_chains): Remove
7814 var_offset_p argument and corresponding handling.
7815 (stmts_may_clobber_ref_p): New function.
7816 (compatible_load_p): New function.
7817 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7818 if there is overlap and rhs_code is not INTEGER_CST. For
7819 non-overlapping stores terminate group if rhs is not mergeable.
7820 (get_alias_type_for_stmts): Change first argument from
7821 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7822 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7823 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7825 (get_location_for_stmts): Change first argument from
7826 auto_vec<gimple *> & to vec<gimple *> &.
7827 (struct split_store): Remove orig_stmts data member, add orig_stores.
7828 (split_store::split_store): Create orig_stores rather than orig_stmts.
7829 (find_constituent_stmts): Renamed to ...
7830 (find_constituent_stores): ... this. Change second argument from
7831 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7832 to info structures rather than the statements.
7833 (split_group): Rename ALLOW_UNALIGNED argument to
7834 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7835 it. Adjust find_constituent_stores caller.
7836 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7837 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7838 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7839 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7840 (mem_valid_for_store_merging): New function.
7841 (handled_load): New function.
7842 (pass_store_merging::process_store): New method.
7843 (pass_store_merging::execute): Use process_store method. Adjust
7844 terminate_all_aliasing_chains caller.
7846 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7848 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7849 Return true for more constants, symbols and label references.
7850 (aarch64_valid_floating_const): Remove unused function.
7852 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7855 * config/aarch64/aarch64.c (aarch64_layout_frame):
7856 Undo forcing of LR at bottom of frame.
7858 2017-11-03 Jeff Law <law@redhat.com>
7861 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7862 for int_registers_saved.
7864 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7865 extracted from tree-ssa-dom.c.
7866 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7867 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7868 (record_equivalences_from_incoming_edge): Add additional argument
7869 to single_pred_edge_ignoring_loop_edges call.
7870 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7871 (uncprop_dom_walker::before_dom_children): Add additional argument
7872 to single_pred_edge_ignoring_loop_edges call.
7873 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7874 single_pred_edge_ignoring_loop_edges rather than open coding.
7875 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7877 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7879 * match.pd (-(-A)): Rewrite.
7881 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7883 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7884 (rs6000_emit_int_cmove): New declaration.
7885 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7886 (rs6000_emit_sISEL): Delete.
7887 (rs6000_emit_int_cmove): Make non-static.
7888 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7889 instead of rs6000_emit_sISEL.
7891 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7893 * asan.c (create_cond_insert_point): Maintain profile.
7894 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7896 * basic-block.h (struct basic_block_def): Remove frequency.
7897 (EDGE_FREQUENCY): Use to_frequency
7898 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7900 (find_traces): Update to use to_frequency.
7901 (find_traces_1_round): Likewise; use only IPA counts.
7902 (bb_to_key): Likewise.
7903 (connect_traces): Use IPA counts only.
7904 (copy_bb_p): Update to use to_frequency.
7905 (fix_up_crossing_landing_pad): Likewise.
7906 (sanitize_hot_paths): Likewise.
7907 * bt-load.c (basic_block_freq): Likewise.
7908 * cfg.c (init_flow): Set count_max to uninitialized.
7909 (check_bb_profile): Remove frequencies; check counts.
7910 (dump_bb_info): Do not dump frequencies.
7911 (update_bb_profile_for_threading): Update counts only.
7912 (scale_bbs_frequencies_int): Likewise.
7913 (MAX_SAFE_MULTIPLIER): Remove.
7914 (scale_bbs_frequencies_gcov_type): Update counts only.
7915 (scale_bbs_frequencies_profile_count): Update counts only.
7916 (scale_bbs_frequencies): Update counts only.
7917 * cfg.h (struct control_flow_graph): Add count-max.
7918 (update_bb_profile_for_threading): Update prototype.
7919 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7920 (find_many_sub_basic_blocks): Likewise.
7921 * cfgcleanup.c (try_forward_edges): Likewise.
7922 (try_crossjump_to_edge): Likewise.
7923 * cfgexpand.c (expand_gimple_cond): Likewise.
7924 (expand_gimple_tailcall): Likewise.
7925 (construct_init_block): Likewise.
7926 (construct_exit_block): Likewise.
7927 * cfghooks.c (verify_flow_info): Check consistency of counts.
7928 (dump_bb_for_graph): Do not dump frequencies.
7929 (split_block_1): Do not update frequencies.
7930 (split_edge): Do not update frequencies.
7931 (make_forwarder_block): Do not update frequencies.
7932 (duplicate_block): Do not update frequencies.
7933 (account_profile_record): Do not update frequencies.
7934 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7935 for global heuristics.
7936 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7937 (expected_loop_iterations_unbounded): Use counts only.
7938 * cfgloopmanip.c (scale_loop_profile): Simplify.
7939 (create_empty_loop_on_edge): Simplify
7941 (duplicate_loop_to_header_edge): Simplify
7942 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7943 (update_br_prob_note): Take care of removing note when profile
7945 (relink_block_chain): Do not dump frequency.
7946 (rtl_account_profile_record): Use to_frequency.
7947 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7948 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7949 (cgraph_update_edges_for_call_stmt_node): Likewise.
7950 (cgraph_edge::verify_count_and_frequency): Update.
7951 (cgraph_node::verify_node): Temporarily disable frequency verification.
7952 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7953 to_cgraph_frequency.
7954 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7955 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7957 (cgraph_node::expand_thunk): Update profile.
7958 * except.c (dw2_build_landing_pads): Do not update frequency.
7959 * final.c (compute_alignments): Use to_frequency.
7960 (dump_basic_block_info): Do not dump frequency.
7961 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7962 (dump_gimple_bb_header): Do not dump frequency.
7963 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7965 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7966 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7967 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7968 (init_before_recovery): Do not update frequency.
7969 (sched_create_recovery_edges): Do not update frequency.
7970 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7971 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7972 (ipa_cp_c_finalize): Set max_count to uninitialized.
7973 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7974 (param_change_prob): Use counts.
7975 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7977 * ipa-split.c (consider_split): Use to_frequency.
7978 (split_function): Use to_frequency.
7979 * ira-build.c (loop_compare_func): Likewise.
7980 (mark_loops_for_removal): Likewise.
7981 (mark_all_loops_for_removal): Likewise.
7982 * loop-doloop.c (doloop_modify): Do not update frequency.
7983 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7985 * lto-streamer-in.c (input_function): Update count_max.
7986 * omp-expand.c (expand_omp_taskreg): Update count_max.
7987 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7988 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7989 (maybe_hot_count_p): Use ipa counts only.
7990 (maybe_hot_bb_p): Simplify.
7991 (maybe_hot_edge_p): Simplify.
7992 (probably_never_executed): Do not take frequency argument.
7993 (probably_never_executed_bb_p): Do not pass frequency.
7994 (probably_never_executed_edge_p): Likewise.
7995 (combine_predictions_for_bb): Check that profile is nonzero.
7996 (propagate_freq): Do not set frequency.
7997 (drop_profile): Simplify.
7998 (counts_to_freqs): Simplify.
7999 (expensive_function_p): Use to_frequency.
8000 (propagate_unlikely_bbs_forward): Simplify.
8001 (determine_unlikely_bbs): Simplify.
8002 (estimate_bb_frequencies): Add hack to silence graphite issues.
8003 (compute_function_frequency): Use ipa counts.
8004 (pass_profile::execute): Update.
8005 (rebuild_frequencies): Use counts only.
8006 (force_edge_cold): Use counts only.
8007 * profile-count.c (profile_count::dump): Dump new count types.
8008 (profile_count::differs_from_p): Check compatiblity.
8009 (profile_count::to_frequency): New function.
8010 (profile_count::to_cgraph_frequency): New function.
8011 * profile-count.h (struct function): Declare.
8012 (enum profile_quality): Add profile_guessed_local and
8013 profile_guessed_global0.
8014 (class profile_proability): Decrease number of bits to 29;
8015 update from_reg_br_prob_note and to_reg_br_prob_note.
8016 (class profile_count: Update comment; decrease number of bits
8017 to 61. Check compatibility.
8018 (profile_count::compatible_p): New private member function.
8019 (profile_count::ipa_p): New member function.
8020 (profile_count::operator<): Handle global zero correctly.
8021 (profile_count::operator>): Handle global zero correctly.
8022 (profile_count::operator<=): Handle global zero correctly.
8023 (profile_count::operator>=): Handle global zero correctly.
8024 (profile_count::nonzero_p): New member function.
8025 (profile_count::force_nonzero): New member function.
8026 (profile_count::max): New member function.
8027 (profile_count::apply_scale): Handle IPA scalling.
8028 (profile_count::guessed_local): New member function.
8029 (profile_count::global0): New member function.
8030 (profile_count::ipa): New member function.
8031 (profile_count::to_frequency): Declare.
8032 (profile_count::to_cgraph_frequency): Declare.
8033 * profile.c (OVERLAP_BASE): Delete.
8034 (compute_frequency_overlap): Delete.
8035 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8036 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8037 * sched-ebb.c (rank): Use counts only.
8038 * shrink-wrap.c (handle_simple_exit): Use counts only.
8039 (try_shrink_wrapping): Use counts only.
8040 (place_prologue_for_one_component): Use counts only.
8041 * tracer.c (find_best_predecessor): Use to_frequency.
8042 (find_trace): Use to_frequency.
8043 (tail_duplicate): Use to_frequency.
8044 * trans-mem.c (expand_transaction): Do not update frequency.
8045 * tree-call-cdce.c: Do not update frequency.
8046 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8047 (gimple_merge_blocks): Likewise.
8048 (gimple_split_edge): Likewise.
8049 (gimple_duplicate_sese_region): Likewise.
8050 (gimple_duplicate_sese_tail): Likewise.
8051 (move_sese_region_to_fn): Likewise.
8052 (gimple_account_profile_record): Likewise.
8053 (insert_cond_bb): Likewise.
8054 * tree-complex.c (expand_complex_div_wide): Likewise.
8055 * tree-eh.c (lower_resx): Update profile.
8056 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8058 (initialize_cfun): Do not initialize frequencies
8059 (freqs_to_counts): Delete.
8060 (copy_cfg_body): Ignore count parameter.
8061 (copy_body): Update.
8062 (expand_call_inline): Update count_max.
8063 (optimize_inline_calls): Update count_max.
8064 (tree_function_versioning): Update count_max.
8065 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8066 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8068 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8069 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8070 (try_peel_loop): Likewise.
8071 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8073 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8074 (tree_transform_and_unroll_loop): Do not use frequencies
8075 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8076 Use reliable prediction only.
8077 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8078 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8079 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8080 probability scaling.
8081 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8082 not update frequency
8083 (any_remaining_duplicated_blocks): Likewise.
8084 (update_profile): Likewise.
8085 (estimated_freqs_path): Delete.
8086 (freqs_to_counts_path): Delete.
8087 (clear_counts_path): Delete.
8088 (ssa_fix_duplicate_block_edges): Likewise.
8089 (duplicate_thread_path): Likewise.
8090 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8091 * tree-tailcall.c (decrease_profile): Do not update frequency.
8092 (eliminate_tail_call): Likewise.
8093 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8094 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8095 (optimize_mask_stores): Likewise.
8096 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8097 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8098 (ubsan_expand_ptr_ifn): Update profile.
8099 * value-prof.c (gimple_ic): Simplify.
8100 * value-prof.h (gimple_ic): Update prototype.
8101 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8102 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8104 (want_inline_self_recursive_call_p): Likewise.
8105 (resolve_noninline_speculation): Only cummulate defined counts.
8106 (inline_small_functions): Use nonzero_p.
8107 (ipa_inline): Do not access freed node.
8109 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8111 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8112 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8114 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8116 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8117 non-legitimate address.
8119 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8121 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8122 (*lt0_<mode>di, *lt0_<mode>si): New.
8124 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8126 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8127 TARGET_PAIRED_FLOAT.
8129 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8130 Jim Wilson <jim.wilson@linaro.org>
8132 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8133 * config/aarch64/aarch64-tune.md: Regenerated.
8134 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8135 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8137 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8139 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8140 (arc_expand_prologue): Restore blink for millicode.
8141 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8143 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8146 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8147 gen_vec_duplicate after forcing the scalar into a register.
8149 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8151 * combine (try_combine): Print the insns input to try_combine to the
8154 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8157 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8158 Remove second argument from aarch64_process_target_attr call.
8159 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8161 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8163 (aarch64_handle_attr_arch): Remove second argument.
8164 (aarch64_handle_attr_cpu): Ditto.
8165 (aarch64_handle_attr_tune): Ditto.
8166 (aarch64_handle_attr_isa_flags): Ditto.
8167 (aarch64_process_one_target_attr): Ditto.
8168 (aarch64_process_target_attr): Ditto.
8169 (aarch64_option_valid_attribute_p): Remove second argument.
8170 on aarch64_process_target_attr call.
8172 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8174 * diagnostic.c: Include "selftest-diagnostic.h".
8175 (selftest::assert_location_text): New function.
8176 (selftest::test_diagnostic_get_location_text): New function.
8177 (selftest::diagnostic_c_tests): Call it.
8179 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8181 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8182 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8183 (class selftest::test_diagnostic_context): Move to...
8184 * selftest-diagnostic.c: New file.
8185 * selftest-diagnostic.h: New file.
8187 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8189 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8190 offset range for FT32B.
8191 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8192 * config/ft32/ft32.md: Add TARGET_NOPM.
8193 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8194 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8196 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8198 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8200 2017-11-02 Jeff Law <law@redhat.com>
8202 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8203 virtual keyword on FINAL OVERRIDE members.
8205 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8206 virtuals together. Add virtual destructor.
8207 (substitute_and_fold_engine): Similarly.
8209 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8211 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8213 2017-11-02 Richard Biener <rguenther@suse.de>
8215 PR tree-optimization/82795
8216 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8218 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8220 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8222 (gcc_GAS_CHECK_FEATURE): Remove.
8223 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8224 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8225 * configure: Regenerate.
8227 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8229 * config/arc/arc.c (hwloop_optimize): Account for empty
8232 2017-11-02 Richard Biener <rguenther@suse.de>
8235 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8236 Truncate ARRAY_REF index and element size.
8238 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8240 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8242 2017-11-01 Jeff Law <law@redhat.com>
8244 * tree-ssa-ccp.c (ccp_folder): New class derived from
8245 substitute_and_fold_engine.
8246 (ccp_folder::get_value): New member function.
8247 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8248 (ccp_fold_stmt): Remove prototype.
8249 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8250 * tree-ssa-copy.c (copy_folder): New class derived from
8251 substitute_and_fold_engine.
8252 (copy_folder::get_value): Renamed from get_value.
8253 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8254 * tree-vrp.c (vrp_folder): New class derived from
8255 substitute_and_fold_engine.
8256 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8257 (vrp_folder::get_value): New member function.
8258 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8259 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8260 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8261 provide a class interface to folder/substitute routines.
8262 (ssa_prop_fold_stmt_fn): Remove typedef.
8263 (ssa_prop_get_value_fn): Likewise.
8264 (subsitute_and_fold): Remove prototype.
8265 (replace_uses_in): Likewise.
8266 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8267 Renamed from replace_uses_in. Call the virtual member function
8268 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8269 (substitute_and_fold_dom_walker): Remove initialization of
8270 data member entries for calbacks. Add substitute_and_fold_engine
8271 member and initialize it.
8272 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8273 member functions for get_value, replace_phi_args_in c
8274 replace_uses_in, and fold_stmt calls.
8275 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8276 substitute_and_fold. Remove assert. Update ctor call.
8278 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8279 (ssa_prop_visit_phi_fn): Likewise.
8280 (class ssa_propagation_engine): New class to provide an interface
8282 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8284 (ssa_prop_visit_phi): Likewise.
8285 (ssa_propagation_engine::simulate_stmt): Moved into class.
8286 Call visit_phi/visit_stmt from the class rather than via
8287 file scoped static variables.
8288 (ssa_propagation_engine::simulate_block): Moved into class.
8289 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8290 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8291 set file scoped statics for the visit_stmt/visit_phi callbacks.
8292 * tree-complex.c (complex_propagate): New class derived from
8293 ssa_propagation_engine.
8294 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8295 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8296 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8298 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8299 ssa_propagation_engine.
8300 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8301 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8302 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8303 * tree-ssa-copy.c (copy_prop): New class derived from
8304 ssa_propagation_engine.
8305 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8306 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8307 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8308 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8309 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8310 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8311 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8313 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8315 PR rtl-optimization/82778
8316 PR rtl-optimization/82597
8317 * compare-elim.c (struct comparison): Add in_a_setter field.
8318 (find_comparison_dom_walker::before_dom_children): Remove killed
8319 bitmap and df_simulate_find_defs call, instead walk the defs.
8320 Compute last_setter and initialize in_a_setter. Merge definitions
8321 with first initialization for a few variables.
8322 (try_validate_parallel): Use insn_invalid_p instead of
8323 recog_memoized. Return insn rather than just the pattern.
8324 (try_merge_compare): Fix up comment. Don't uselessly test if
8325 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8327 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8330 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8331 Alan Hayward <alan.hayward@arm.com>
8332 David Sherwood <david.sherwood@arm.com>
8334 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8335 aarch64_hard_regno_nregs to get the number of registers
8338 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8339 Alan Hayward <alan.hayward@arm.com>
8340 David Sherwood <david.sherwood@arm.com>
8342 * config/aarch64/constraints.md (Upl): Rename to...
8344 * config/aarch64/aarch64.md
8345 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8348 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8349 Alan Hayward <alan.hayward@arm.com>
8350 David Sherwood <david.sherwood@arm.com>
8352 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8353 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8356 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8357 Alan Hayward <alan.hayward@arm.com>
8358 David Sherwood <david.sherwood@arm.com>
8360 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8361 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8362 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8364 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8365 of a single element.
8366 * config/aarch64/iterators.md: Add a comment above the permute
8367 unspecs to say that they are generated directly by
8368 aarch64_expand_vec_perm_const.
8369 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8371 2017-11-01 Nathan Sidwell <nathan@acm.org>
8373 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8375 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8377 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8378 and medany code models, and describe what they do.
8380 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8382 Revert accidental duplicate:
8384 * combine.c (can_change_dest_mode): Reject changes in
8385 REGMODE_NATURAL_SIZE.
8387 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8389 PR rtl-optimization/64682
8390 PR rtl-optimization/69567
8391 PR rtl-optimization/69737
8392 PR rtl-optimization/82683
8393 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8394 register mentioned in the note, drop the note, unless it came from I3,
8395 in which case it should go to I3 again.
8397 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8399 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8400 and return false if not.
8401 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8403 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8405 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8406 range is known to be empty.
8408 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8409 Alan Hayward <alan.hayward@arm.com>
8410 David Sherwood <david.sherwood@arm.com>
8412 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8413 and CONST_VECTOR_NUNITS instead of computing the number of units from
8414 the byte sizes of the vector and element.
8415 (simplify_binary_operation_1): Likewise.
8416 (simplify_const_binary_operation): Likewise.
8417 (simplify_ternary_operation): Likewise.
8419 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8420 Alan Hayward <alan.hayward@arm.com>
8421 David Sherwood <david.sherwood@arm.com>
8423 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8424 (int_mem_offset): ...this new function.
8425 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8426 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8427 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8430 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8431 Alan Hayward <alan.hayward@arm.com>
8432 David Sherwood <david.sherwood@arm.com>
8434 * lower-subreg.c (interesting_mode_p): New function.
8435 (compute_costs, find_decomposable_subregs, decompose_register)
8436 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8437 (resolve_clobber, dump_choices): Use it.
8439 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8440 Alan Hayward <alan.hayward@arm.com>
8441 David Sherwood <david.sherwood@arm.com>
8443 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8445 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8446 Alan Hayward <alan.hayward@arm.com>
8447 David Sherwood <david.sherwood@arm.com>
8449 * alias.c (find_base_value, find_base_term): Only process integer
8450 truncations. Check the precision rather than the size.
8452 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8453 Alan Hayward <alan.hayward@arm.com>
8454 David Sherwood <david.sherwood@arm.com>
8456 * machmode.h (is_narrower_int_mode): New function
8457 * optabs.c (expand_float, expand_fix): Use it.
8458 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8460 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8461 Alan Hayward <alan.hayward@arm.com>
8462 David Sherwood <david.sherwood@arm.com>
8464 * rtl.h (narrower_subreg_mode): New function.
8465 * ira-color.c (update_costs_from_allocno): Use it.
8467 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8468 Alan Hayward <alan.hayward@arm.com>
8469 David Sherwood <david.sherwood@arm.com>
8471 * optabs-query.h (convert_optab_p): New function, split out from...
8472 (convert_optab_handler): ...here.
8473 (widening_optab_handler): Delete.
8474 (find_widening_optab_handler): Remove permit_non_widening parameter.
8475 (find_widening_optab_handler_and_mode): Likewise. Provide an
8476 override that operates on mode class wrappers.
8477 * optabs-query.c (widening_optab_handler): Delete.
8478 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8479 parameter. Assert that the two modes are the same class and that
8480 the "from" mode is narrower than the "to" mode. Use
8481 convert_optab_handler instead of widening_optab_handler.
8482 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8483 instead of widening_optab_handler.
8484 * expr.c (expand_expr_real_2): Update calls to
8485 find_widening_optab_handler.
8486 * optabs.c (expand_widen_pattern_expr): Likewise.
8487 (expand_binop_directly): Take the insn_code as a parameter.
8488 (expand_binop): Only call find_widening_optab_handler for
8489 conversion optabs; use optab_handler otherwise. Update calls
8490 to find_widening_optab_handler and expand_binop_directly.
8491 Use convert_optab_handler instead of widening_optab_handler.
8492 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8493 find_widening_optab_handler and use scalar_mode rather than
8495 (convert_plusminus_to_widen): Likewise.
8497 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8498 Alan Hayward <alan.hayward@arm.com>
8499 David Sherwood <david.sherwood@arm.com>
8501 * machmode.h (fixed_size_mode): New class.
8502 * rtl.h (get_pool_mode): Return fixed_size_mode.
8503 * gengtype.c (main): Add fixed_size_mode.
8504 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8505 (get_raw_arg_mode): Likewise.
8506 * doc/tm.texi: Regenerate.
8507 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8508 * targhooks.c (default_get_reg_raw_mode): Likewise.
8509 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8510 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8511 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8512 (msp430_get_raw_result_mode): Likewise.
8513 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8514 * dbxout.c (dbxout_parms): Require fixed-size modes.
8515 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8516 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8517 * omp-low.c (lower_oacc_reductions): Likewise.
8518 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8519 (simplify_subreg): Update accordingly.
8520 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8521 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8522 that aren't fixed-size.
8523 (get_pool_mode): Return a fixed_size_mode.
8524 (output_constant_pool_2): Take a fixed_size_mode.
8526 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8527 Alan Hayward <alan.hayward@arm.com>
8528 David Sherwood <david.sherwood@arm.com>
8530 * doc/rtl.texi (vec_series): Document.
8531 (const): Say that the operand can be a vec_series.
8532 * rtl.def (VEC_SERIES): New rtx code.
8533 * rtl.h (const_vec_series_p_1): Declare.
8534 (const_vec_series_p): New function.
8535 * emit-rtl.h (gen_const_vec_series): Declare.
8536 (gen_vec_series): Likewise.
8537 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8538 (gen_vec_series): Likewise.
8539 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8540 * simplify-rtx.c (simplify_unary_operation): Handle negations
8542 (simplify_binary_operation_series): New function.
8543 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8544 (test_vector_ops_series): New function.
8545 (test_vector_ops): Call it.
8546 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8547 gen_const_vec_series.
8548 (altivec_lvsr): Likewise.
8549 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8551 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8552 Alan Hayward <alan.hayward@arm.com>
8553 David Sherwood <david.sherwood@arm.com>
8555 * doc/rtl.texi (const): Update description of address constants.
8556 Say that vector constants are allowed too.
8557 * common.md (E, F): Use CONSTANT_P instead of checking for
8559 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8560 checking for CONST_VECTOR.
8561 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8563 * expr.c (expand_expr_real_2): Check for vector modes instead
8564 of checking for CONST_VECTOR.
8565 * rtl.h (const_vec_p): New function.
8566 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8567 (unwrap_const_vec_duplicate): Handle them here too.
8569 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8570 David Malcolm <dmalcolm@redhat.com>
8571 Alan Hayward <alan.hayward@arm.com>
8572 David Sherwood <david.sherwood@arm.com>
8574 * rtl.h (vec_duplicate_p): New function.
8575 * selftest-rtl.c (assert_rtx_eq_at): New function.
8576 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8577 (assert_rtx_eq_at): Declare.
8578 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8579 * selftest-run-tests.c (selftest::run_tests): Call it.
8580 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8581 (simplify_unary_operation_1): Recursively handle vector duplicates.
8582 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8584 (simplify_subreg): Handle subregs of vector duplicates.
8585 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8586 (selftest::simplify_rtx_c_tests): New functions.
8588 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8589 Alan Hayward <alan.hayward@arm.com>
8590 David Sherwood <david.sherwood@arm.com>
8592 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8593 (gen_vec_duplicate): Likewise.
8594 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8596 (gen_const_vector): ...here.
8597 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8598 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8599 whose elements are all equal.
8600 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8601 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8602 (simplify_relational_operation): Likewise.
8603 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8605 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8606 (aarch64_expand_vector_init): Likewise.
8607 * config/arm/arm.c (neon_vdup_constant): Likewise.
8608 (neon_expand_vector_init): Likewise.
8609 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8610 (arm_block_set_unaligned_vect): Likewise.
8611 (arm_block_set_aligned_vect): Likewise.
8612 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8613 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8614 (expand_vec_perm_even_odd_pack): Likewise.
8615 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8616 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8617 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8618 gen_const_vec_duplicate.
8619 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8620 * config/mips/mips.c (mips_gen_const_int_vector): Use
8621 gen_const_vec_duplicate.
8622 (mips_expand_vector_init): Use CONST0_RTX.
8623 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8624 (define_split): Use gen_const_vec_duplicate.
8625 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8626 (define_split): Use gen_const_vec_duplicate.
8627 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8628 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8629 * config/spu/spu.c (spu_const): Likewise.
8631 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8632 Alan Hayward <alan.hayward@arm.com>
8633 David Sherwood <david.sherwood@arm.com>
8635 * combine.c (can_change_dest_mode): Reject changes in
8636 REGMODE_NATURAL_SIZE.
8638 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8640 * configure.ac (--enable-libssp): New.
8641 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8642 trying to determine target-specific default. Adjust indentation.
8643 * configure: Regenerated.
8644 * doc/install.texi (Configuration): Expand --disable-libssp
8647 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8649 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8652 2017-10-31 Martin Jambor <mjambor@suse.cz>
8655 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8657 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8659 * auto-profile.c (autofdo_source_profile::read): Use
8660 UNKNOWN_LOCATION rather than 0.
8661 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8662 (warning_at): ...this overload.
8663 (warning_at_rich_loc_n): Rename to...
8664 (warning_n): ...this overload.
8665 (error_at_rich_loc): Rename to...
8666 (error_at): ...this overload.
8667 (pedwarn_at_rich_loc): Rename to...
8668 (pedwarn): ...this overload.
8669 (permerror_at_rich_loc): Rename to...
8670 (permerror): ...this overload.
8671 (inform_at_rich_loc): Rename to...
8672 (inform): ...this overload.
8673 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8674 (diagnostic_n_impl_richloc): Rename to...
8675 (diagnostic_n_impl): ...this rich_location *-based decl.
8676 (inform_at_rich_loc): Rename to...
8677 (inform): ...this, and add an assertion.
8678 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8679 (warning_at_rich_loc): Rename to...
8680 (warning_at): ...this, and add an assertion.
8681 (warning_at_rich_loc_n): Rename to...
8682 (warning_n): ...this, and add an assertion.
8683 (warning_n): Update location_t-based implementation for removal of
8684 location_t-based diagnostic_n_impl.
8685 (pedwarn_at_rich_loc): Rename to...
8686 (pedwarn): ...this, and add an assertion.
8687 (permerror_at_rich_loc): Rename to...
8688 (permerror): ...this, and add an assertion.
8689 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8690 (error_at_rich_loc): Rename to...
8691 (error_at): ...this, and add an assertion.
8692 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8693 (driver::do_spec_on_infiles): Likewise.
8694 * substring-locations.c (format_warning_va): Update for renaming
8695 of inform_at_rich_loc.
8697 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8699 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8700 _Float<N>X built-in functions so that the variant without the
8701 "__builtin_" prefix is only enabled for the GNU C and Objective C
8702 languages when they are in non-strict ANSI/ISO mode.
8703 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8704 * target.def (floatn_builtin_p): Add a target hook to control
8705 whether _Float<N> and _Float<N>X built-in functions without the
8706 "__builtin_" prefix are enabled, and return true for C and
8707 Objective C in the default hook. Include langhooks.h in
8709 * targhooks.h (default_floatn_builtin_p): Likewise.
8710 * targhooks.c (default_floatn_builtin_p): Likewise.
8711 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8712 floatn_builtin_p target hook.
8713 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8715 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8716 Eric Botcazou <ebotcazou@adacore.com>
8718 PR rtl-optimization/81803
8719 * lra-constraints.c (curr_insn_transform): Also reload the whole
8720 register for a strict subreg no wider than a word if this is for
8721 a WORD_REGISTER_OPERATIONS target.
8723 2017-10-31 Jason Merrill <jason@redhat.com>
8725 * gdbinit.in: Skip over inlines from timevar.h.
8727 2017-10-31 Martin Liska <mliska@suse.cz>
8729 * doc/gcov.texi: Document new option.
8730 * gcov.c (print_usage): Likewise print it.
8731 (process_args): Support the argument.
8732 (format_count): New function.
8733 (format_gcov): Use the function.
8735 2017-10-31 Martin Liska <mliska@suse.cz>
8737 * gcov.c (struct name_map): do not use typedef.
8738 Define operator== and operator<.
8739 (name_search): Remove.
8740 (name_sort): Remove.
8741 (main): Do not allocate names.
8742 (process_file): Add vertical space.
8743 (generate_results): Use std::find.
8744 (release_structures): Do not release memory.
8745 (find_source): Use std::find.
8747 2017-10-31 Martin Liska <mliska@suse.cz>
8749 * gcov.c (struct line_info): Remove it's typedef.
8750 (line_info::line_info): Add proper ctor.
8751 (line_info::has_block): Do not use a typedef.
8752 (struct source_info): Do not use typedef.
8753 (circuit): Likewise.
8754 (get_cycles_count): Likewise.
8755 (output_intermediate_file): Iterate via vector iterator.
8756 (add_line_counts): Use std::vector methods.
8757 (accumulate_line_counts): Likewise.
8758 (output_lines): Likewise.
8760 2017-10-31 Martin Liska <mliska@suse.cz>
8762 * gcov.c (struct source_info): Remove typedef.
8763 (source_info::source_info): Add proper ctor.
8764 (accumulate_line_counts): Use struct, not it's typedef.
8765 (output_gcov_file): Likewise.
8766 (output_lines): Likewise.
8767 (main): Do not allocate an array.
8768 (output_intermediate_file): Use size of vector container.
8769 (process_file): Resize the vector.
8770 (generate_results): Do not preallocate, use newly added vector
8772 (release_structures): Do not release sources.
8773 (find_source): Use vector methods.
8774 (add_line_counts): Do not use typedef.
8776 2017-10-31 Martin Liska <mliska@suse.cz>
8778 * doc/gcov.texi: Document that.
8779 * gcov.c (add_line_counts): Mark lines with a non-executed
8781 (output_line_beginning): Handle such lines.
8782 (output_lines): Pass new argument.
8783 (output_intermediate_file): Print it in intermediate format.
8785 2017-10-31 Martin Liska <mliska@suse.cz>
8787 * color-macros.h: New file.
8788 * diagnostic-color.c: Factor out color related to macros to
8790 * doc/gcov.texi: Document -k option.
8791 * gcov.c (INCLUDE_STRING): Include string.h.
8792 (print_usage): Add -k option.
8793 (process_args): Parse it.
8794 (pad_count_string): New function.
8795 (output_line_beginning): Likewise.
8796 (DEFAULT_LINE_START): New macro.
8797 (output_lines): Support color output.
8799 2017-10-31 Martin Liska <mliska@suse.cz>
8801 PR gcov-profile/82633
8802 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8803 their interaction with GCOV infrastructure.
8804 * configure.ac: Add -fkeep-{inline,static}-functions to
8806 * configure: Regenerate.
8808 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8811 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8813 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8816 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8817 into a register if it does not fit into an immediate offset field.
8819 2017-10-31 Olivier Hainque <hainque@adacore.com>
8821 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8823 2017-10-31 Julia Koval <julia.koval@intel.com>
8825 * config.gcc: Add gfniintrin.h.
8826 * config/i386/gfniintrin.h: New.
8827 * config/i386/i386-builtin-types.def
8828 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8829 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8830 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8831 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8832 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8833 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8834 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8835 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8836 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8837 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8838 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8839 * config/i386/immintrin.h: Include gfniintrin.h.
8840 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8842 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8844 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8846 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8848 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8849 (arm_ashldi3_1bit): Remove pattern.
8850 (ashrdi3): Remove shift by 1 expansion.
8851 (arm_ashrdi3_1bit): Remove pattern.
8852 (lshrdi3): Remove shift by 1 expansion.
8853 (arm_lshrdi3_1bit): Remove pattern.
8854 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8855 cost of ashldi3 by 1.
8856 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8857 (<shift>di3_neon): Likewise.
8859 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8861 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8862 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8863 and (*aarch64_simd_mov<VQ:mode>).
8864 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8865 pattern alternative.
8866 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8867 attributes to match pattern alternative.
8869 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8871 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8872 * config/rs6000/emmintrin.h: New file.
8873 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8875 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8877 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8878 * (movdi_vfp_cortexa8): Remove pattern.
8880 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8882 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8884 (Specific, alpha*-dec-osf5.1): Remove.
8885 (Specific, mips-sgi-irix5): Remove.
8886 (Specific, mips-sgi-irix6): Remove.
8888 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8891 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8892 arguments to clear_bit_region_be.
8894 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8896 * gimplify.c: Include memmodel.h.
8898 2017-10-30 Martin Jambor <mjambor@suse.cz>
8900 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8901 condition whether loop should be executed at all.
8903 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8905 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8906 gimple folding of vec_madd() intrinsics.
8907 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8908 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8909 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8911 2017-10-30 Richard Biener <rguenther@suse.de>
8913 PR tree-optimization/82762
8915 2017-10-23 Richard Biener <rguenther@suse.de>
8917 PR tree-optimization/82129
8919 2017-08-01 Richard Biener <rguenther@suse.de>
8921 PR tree-optimization/81181
8922 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8923 (compute_antic): ... end of iteration here.
8925 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8927 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8928 -std=iso9899:2017 and -std=gnu17.
8929 * doc/standards.texi (C Language): Document C17 support.
8930 * doc/cpp.texi (Overview): Mention -std=c17.
8931 (Standard Predefined Macros): Document C11 and C17 values of
8932 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8933 * doc/extend.texi (Inline): Do not list individual options for
8934 standards newer than C99.
8935 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8937 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8940 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8942 * asan.c (asan_finish_file): Align asan globals array by shadow
8945 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8948 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8949 (struct store_immediate_info): Add bitregion_start and bitregion_end
8951 (store_immediate_info::store_immediate_info): Add brs and bre
8952 arguments and initialize bitregion_{start,end} from those.
8953 (struct merged_store_group): Add bitregion_start, bitregion_end,
8954 align_base and mask fields. Drop unnecessary struct keyword from
8955 struct store_immediate_info. Add do_merge method.
8956 (clear_bit_region_be): Use memset instead of loop storing zeros.
8957 (merged_store_group::do_merge): New method.
8958 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8959 stores as long as the surrounding bitregions have no gaps.
8960 (merged_store_group::merge_overlapping): Use do_merge.
8961 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8962 is byte aligned, rather than requiring that start and width are
8963 byte aligned. Drop unnecessary struct keyword from
8964 struct store_immediate_info. Allocate and populate also mask array.
8965 Make start of the arrays relative to bitregion_start rather than
8966 start and size them according to bitregion_{end,start} difference.
8967 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8968 struct store_immediate_info.
8969 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8970 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8971 struct keyword from struct store_immediate_info.
8972 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8973 between stores as long as the surrounding bitregions have no gaps.
8975 (struct split_store): Add orig non-static data member.
8976 (split_store::split_store): Initialize orig to false.
8977 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8978 if there is exactly a single original stmt. Change stmts argument
8979 to pointer from reference, if NULL, don't push anything to it. Add
8980 first argument, use it to optimize skipping over orig stmts that
8981 are known to be before bitpos already. Simplify.
8982 (split_group): Return unsigned int count how many stores are or
8983 would be needed rather than a bool. Add allow_unaligned argument.
8984 Change split_stores argument from reference to pointer, if NULL,
8985 only do a dry run computing how many stores would be produced.
8986 Rewritten algorithm to use both alignment and misalign if
8987 !allow_unaligned and handle bitfield stores with gaps.
8988 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8989 from bitregion_start instead of start. Compute allow_unaligned
8990 here, if true, do 2 split_group dry runs to compute which one
8991 produces fewer stores and prefer aligned if equal. Punt if
8992 new count is bigger or equal than original before emitting any
8993 statements, rather than during that. Remove no longer needed
8994 new_ssa_names tracking. Replace num_stmts with
8995 split_stores.length (). Use 32-bit stack allocated entries
8996 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8997 if possible. Handle bitfields with gaps.
8998 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8999 Compute bitregion_{start,end} for the stores and construct
9000 store_immediate_info with that. Formatting fixes.
9002 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9005 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9006 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9008 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9010 * gimplify.c: Include tm_p.h.
9012 * common.opt (gcoff): Re-add as ignored option.
9013 (gcoff1, gcoff2, gcoff3): Likewise.
9015 * Makefile.in (OBJS): Delete sdbout.o.
9016 (GTFILES): Delete $(srcdir)/sdbout.c.
9017 * debug.h: Delete sdb_debug_hooks.
9018 * final.c: Delete sdbout.h include.
9019 (final_scan_insn): Delete SDB_DEBUG check.
9020 (rest_of_clean_state): Likewise.
9021 * output.h: Delete sdb_begin_function_line.
9024 * toplev.c: Delete sdbout.h include.
9025 (process_options): Delete SDB_DEBUG check.
9026 * tree-core.h (tree_type_common): Delete pointer field of
9028 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9029 TYPE_SYMTAB_POINTER.
9030 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9031 (TYPE_SYMTAB_IS_POINTER): Delete.
9032 (TYPE_SYMTAB_IS_DIE): Renumber.
9033 * xcoffout.c: Refer to former sdbout.c file.
9034 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9036 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9037 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9038 (Debugging Options): Delete -gcoff.
9039 (-gstabs): Delete SDB reference.
9041 (-gcoff@var{level}): Delete.
9042 * doc/passes.texi (Debugging information output): Delete SDB and
9043 sdbout.c references.
9044 * doc/tm.texi: Regenerate.
9045 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9046 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9048 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9049 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9050 (SDB_DEBUGGING_INFO): Delete.
9051 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9052 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9053 * target.def (output_source_filename): Delete COFF reference.
9055 * common.opt (gcoff): Delete.
9056 (gxcoff+): Update Negative chain.
9057 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9059 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9060 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9061 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9063 (expand_function_start): Change sdb reference to past tense.
9064 (expand_function_end): Change sdb reference to past tense.
9065 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9066 * opts.c (debug_type_names): Delete coff entry.
9067 (common_handle_option): Delete OPT_gcoff case.
9068 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9070 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9071 * config/cris/cris.h: Delete SDB reference in comment.
9072 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9073 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9074 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9075 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9077 (ix86_expand_prologue): Likewise.
9078 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9079 * config/ia64/ia64.h: Likewise.
9080 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9081 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9083 * config/mmix/mmix.h: Likewise.
9084 * config/nds32/nds32.c: Likewise.
9085 * config/stormy/storym16.h: Likewise.
9086 * config/visium/visium.h: Likewise.
9087 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9089 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9091 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9092 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9093 FRAME_POINTER_REGNUM point at high end of local var area.
9095 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9097 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9098 Move comment around. Do not reset best_edge for a copiable
9099 destination if the copy would cause a partition change.
9100 (better_edge_p): Remove redundant check.
9102 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9104 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9106 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9108 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9109 for math functions that have _Float<N> and _Float<N>X variants.
9110 (mathfn_built_in_2): Add support for math functions that have
9111 _Float<N> and _Float<N>X variants.
9112 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9113 (expand_builtin_mathfn_ternary): Add support for fma with
9114 _Float<N> and _Float<N>X variants.
9115 (expand_builtin): Likewise.
9116 (fold_builtin_3): Likewise.
9117 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9118 create math function _Float<N> and _Float<N>X variants as external
9120 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9121 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9122 the __builtin_ prefix and if not strict ansi, without the prefix.
9123 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9124 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9125 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9126 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9127 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9128 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9129 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9130 function signatures for fma _Float<N> and _Float<N>X variants.
9131 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9132 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9133 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9134 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9135 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9136 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9137 * gencfn-macros.c (print_case_cfn): Add support for math functions
9138 that have _Float<N> and _Float<N>X variants.
9139 (print_define_operator_list): Likewise.
9140 (fltfn_suffixes): Likewise.
9142 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9143 for math functions that have _Float<N> and _Float<N>X variants.
9144 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9145 and _Float<N>X variants.
9146 (COPYSIGN): Likewise.
9149 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9150 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9152 (integer_valued_read_call_p): Likewise.
9153 * fold-const-call.c (fold_const_call_ss): Likewise.
9154 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9155 _Float<N> and _Float<N>X variants.
9156 (fold_const_call_ssss): Add support for fma _Float<N> and
9157 _Float<N>X variants.
9158 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9159 support for copysign and fma _Float<N> and _Float<N>X variants.
9160 (backprop::process_builtin_call_use): Likewise.
9161 * tree-call-cdce.c (can_test_argument_range); Add support for
9162 sqrt _Float<N> and _Float<N>X variants.
9163 (edom_only_function): Likewise.
9164 (get_no_error_domain): Likewise.
9165 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9166 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9167 copysign _Float<N> and _Float<N>X variants.
9168 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9169 handled by machine independent code.
9170 (FMAF128): Likewise.
9171 * doc/cpp.texi (Common Predefined Macros): Document defining
9172 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9173 fma _Float<N> and _Float<N>X variants.
9175 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9178 * config/i386/i386-modes.def (CCFPU): Remove definition.
9179 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9180 (ix86_cc_modes_compatible): Ditto.
9181 (ix86_expand_carry_flag_compare): Ditto.
9182 (ix86_expand_int_movcc): Ditto.
9183 (ix86_expand_int_addcc): Ditto.
9184 (ix86_reverse_condition): Ditto.
9185 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9186 Return true/false for unordered/ordered fp comparisons.
9187 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9188 (ix86_prepare_fp_compare_args): Update for rename.
9189 (ix86_expand_fp_compare): Update for rename. Generate unordered
9190 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9191 (ix86_expand_sse_compare_and_jump): Ditto.
9192 * config/i386/predicates.md (fcmov_comparison_operator):
9193 Remove CCFPU mode handling.
9194 (ix86_comparison_operator): Ditto.
9195 (ix86_carry_flag_operator): Ditto.
9196 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9197 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9198 (*cmpu<mode>_cc_i387): Ditto.
9199 (FPCMP): Remove mode iterator.
9200 (unord): Remove mode attribute.
9201 (unord_subst): New define_subst transformation
9202 (unord): New define_subst attribute.
9204 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9205 (*cmpi<unord>xf_i387): Ditto.
9206 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9207 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9208 using unord_subst transformation.
9209 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9210 (round_saeonly): Also handle CCFP mode.
9211 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9212 Remove UNSPEC_SAHF unspec handling.
9214 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9216 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9218 2017-10-27 Jeff Law <law@redhat.com>
9220 * gimple-ssa-sprintf.c: Include domwalk.h.
9221 (class sprintf_dom_walker): New class, derived from dom_walker.
9222 (sprintf_dom_walker::before_dom_children): New function.
9223 (struct call_info): Moved into sprintf_dom_walker class
9224 (compute_formath_length, handle_gimple_call): Likewise.
9225 (sprintf_length::execute): Call the dominator walker rather
9226 than walking the statements.
9228 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9229 gimple statement locations.
9230 (check_array_bounds): Corresponding changes. Get the statement's
9231 location directly from wi->stmt.
9233 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9236 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9238 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9240 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9241 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9243 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9246 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9247 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9248 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9249 maybe_get_pool_constant.
9250 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9253 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9255 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9256 2.26 caveat. Update gas and gld versions.
9257 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9260 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9262 * cgraph.h (set_malloc_flag): Declare.
9263 * cgraph.c (set_malloc_flag_1): New function.
9264 (set_malloc_flag): Likewise.
9265 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9266 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9268 (read_ipa_call_summary): Add support for reading is_return_callee.
9269 (write_ipa_call_summary): Stream is_return_callee.
9270 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9271 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9272 ipa-prop.h, ipa-fnsummary.h.
9273 (pure_const_names): Change to static.
9274 (malloc_state_e): Define.
9275 (malloc_state_names): Define.
9276 (funct_state_d): Add field malloc_state.
9277 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9278 (check_retval_uses): New function.
9279 (malloc_candidate_p): Likewise.
9280 (analyze_function): Add support for malloc attribute.
9281 (pure_const_write_summary): Stream malloc_state.
9282 (pure_const_read_summary): Add support for reading malloc_state.
9283 (dump_malloc_lattice): New function.
9284 (propagate_malloc): New function.
9285 (warn_function_malloc): New function.
9286 (ipa_pure_const::execute): Call propagate_malloc and
9287 ipa_free_fn_summary.
9288 (pass_local_pure_const::execute): Add support for malloc attribute.
9289 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9290 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9292 2017-10-27 Martin Liska <mliska@suse.cz>
9294 PR gcov-profile/82457
9295 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9296 for fork-like functions to be properly instrumented.
9298 2017-10-27 Richard Biener <rguenther@suse.de>
9301 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9302 info when we redirected EH.
9304 2017-10-26 Michael Collison <michael.collison@arm.com>
9306 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9308 (<optab>_trunchf<GPI:mode>2: New pattern.
9309 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9310 * config/aarch64/iterators.md (wv): New mode attribute.
9311 (vf, VF): New mode attributes.
9312 (vgp, VGP): New mode attributes.
9313 (s): Update attribute with SImode and DImode prefixes.
9315 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9317 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9318 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9319 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9320 (nios2_option_overide): Initialize it. Don't allow R0-relative
9321 addressing with PIC.
9322 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9323 (nios2_symbolic_constant_p): Likewise.
9324 (nios2_legitimate_address_p): Likewise.
9325 (nios2_r0rel_section_name_p): New.
9326 (nios2_symbol_ref_in_r0rel_data_p): New.
9327 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9328 (r0rel_constant_p): New.
9329 (nios2_print_operand_address): Handle r0rel_constant_p.
9330 (nios2_cdx_narrow_form_p): Likewise.
9331 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9332 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9333 (Nios II Options): Document -mr0rel-sec.
9335 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9337 * config/nios2/nios2.c: Include xregex.h.
9338 (nios2_gprel_sec_regex): New.
9339 (nios2_option_overide): Initialize it. Don't allow GP-relative
9340 addressing with PIC.
9341 (nios2_small_section_name_p): Check for regex match.
9342 * config/nios2/nios2.opt (mgprel-sec=): New option.
9343 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9344 (Nios II Options): Document -mgprel-sec.
9346 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9348 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9350 2017-10-26 Tom de Vries <tom@codesourcery.com>
9352 PR tree-optimization/82707
9353 * gimple.c (gimple_copy): Fix unsharing of
9354 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9356 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9358 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9359 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9360 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9361 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9362 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9363 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9364 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9365 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9366 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9367 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9368 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9369 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9370 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9371 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9372 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9373 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9374 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9376 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9378 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9380 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9381 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9382 warning to rs6000.c. Remove the Undocumented flag, since it has
9384 (-mabi=ibmlongdouble): Likewise.
9385 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9386 already set, set the default format for long double.
9387 (rs6000_debug_reg_global): Print whether long double is IBM or
9389 (rs6000_option_override_internal): Rework setting long double
9390 format. Only warn if the user is changing the long double default
9391 and they did not use -Wno-psabi.
9392 * doc/invoke.texi (PowerPC options): Update the documentation for
9393 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9395 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9396 Alan Hayward <alan.hayward@arm.com>
9397 David Sherwood <david.sherwood@arm.com>
9399 * rtl.h (wider_subreg_mode): New function.
9400 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9401 rather than an unsigned int *.
9402 * ira-color.c (regno_max_ref_width): Replace with...
9403 (regno_max_ref_mode): ...this new variable.
9404 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9405 Use wider_subreg_mode.
9406 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9407 rather than an unsigned int *.
9408 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9409 (process_alt_operands): Likewise.
9410 (invariant_p): Likewise.
9411 * lra-spills.c (assign_mem_slot): Likewise.
9412 (add_pseudo_to_slot): Likewise.
9413 * lra.c (collect_non_operand_hard_regs): Likewise.
9414 (add_regs_to_insn_regno_info): Likewise.
9415 * reload1.c (regno_max_ref_width): Replace with...
9416 (regno_max_ref_mode): ...this new variable.
9417 (reload): Update accordingly. Update call to
9418 ira_sort_regnos_for_alter_reg.
9419 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9420 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9421 (scan_paradoxical_subregs): Likewise.
9423 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9425 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9426 (aarch64_frame): Add emit_frame_chain boolean.
9427 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9428 Move eh_return case to aarch64_layout_frame.
9429 (aarch64_layout_frame): Initialize emit_frame_chain.
9430 (aarch64_expand_prologue): Use emit_frame_chain.
9432 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9434 * config/aarch64/aarch64.c (aarch64_layout_frame):
9435 Ensure LR is always stored at the bottom of the callee-saves.
9436 Remove rarely used frame layout which saves callee-saves at top of
9437 frame, so the store of LR can be used as a valid probe in all cases.
9439 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9441 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9442 Improve unaligned TImode/TFmode base/offset split.
9444 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9445 Alan Hayward <alan.hayward@arm.com>
9446 David Sherwood <david.sherwood@arm.com>
9448 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9449 * combine.c (find_single_use_1): Likewise.
9450 (expand_field_assignment): Likewise.
9451 (move_deaths): Likewise.
9452 * lra-constraints.c (simplify_operand_subreg): Likewise.
9453 (curr_insn_transform): Likewise.
9454 * lra.c (collect_non_operand_hard_regs): Likewise.
9455 (add_regs_to_insn_regno_info): Likewise.
9456 * rtlanal.c (reg_referenced_p): Likewise.
9457 (covers_regno_no_parallel_p): Likewise.
9459 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9461 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9462 Don't print any bits outside the precision of the value.
9463 * wide-int.cc (test_printing): Add some new tests.
9465 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9467 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9468 supports -xbrace_comment option.
9469 * configure: Regenerate.
9470 * config.in: Regenerate.
9471 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9472 (ASM_CPU_SPEC): Use it.
9474 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9476 * target.def (static_rtx_alignment): New hook.
9477 * targhooks.h (default_static_rtx_alignment): Declare.
9478 * targhooks.c (default_static_rtx_alignment): New function.
9479 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9480 * doc/tm.texi: Regenerate.
9481 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9482 instead of targetm.constant_alignment. Remove call to
9484 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9485 (cris_preferred_mininum_alignment): New function, split out from...
9486 (cris_constant_alignment): ...here.
9487 (cris_static_rtx_alignment): New function.
9488 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9490 (ix86_constant_alignment): ...here.
9491 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9492 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9493 (mmix_static_rtx_alignment): New function.
9494 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9495 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9497 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9500 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9501 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9503 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9505 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9506 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9507 mark as a sign-extending load.
9508 (local_pic_load_u): Define.
9510 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9513 * fold-const.c (operand_equal_for_comparison_p): Also return true
9514 if ARG0 is a simple variant of ARG1 with narrower precision.
9515 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9517 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9519 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9521 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9522 scatter_static, scatter_per_elt.
9523 * x86-tune-costs.h: Add new cost entries.
9525 2017-10-25 Richard Biener <rguenther@suse.de>
9527 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9528 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9529 class pass_fre): Move to ...
9530 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9531 class pass_fre): ... here and adjust for statistics.
9533 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9536 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9537 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9538 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9539 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9542 2017-10-25 Richard Biener <rguenther@suse.de>
9544 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9545 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9546 eliminate_push_avail, eliminate_insert): Move inside...
9547 (class eliminate_dom_walker): ... this class in preparation
9549 (fini_eliminate): Remove by merging with ...
9550 (eliminate): ... this function. Adjust for class changes.
9551 (pass_pre::execute): Remove fini_eliminate call.
9552 (pass_fre::execute): Likewise.
9554 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9557 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9558 (VPERMI2, VPERMI2I): New mode iterators.
9559 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9560 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9562 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9563 mode iterator. Remove 3 old define_insn patterns.
9564 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9565 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9566 VPERMI2 mode iterator, remove the other two expanders.
9567 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9568 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9569 instructions, remove the other two patterns.
9570 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9571 mode iterator, remove the other two patterns.
9572 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9573 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9574 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9575 and adjust argument order accordingly.
9576 (ix86_expand_vec_perm): Adjust caller.
9577 (expand_vec_perm_1): Likewise.
9578 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9579 (expand_vec_perm_vpermt2_vpshub2): ... this.
9580 (ix86_expand_vec_perm_const_1): Adjust caller.
9581 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9584 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9585 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9586 (vec_shl_<mode>): Remove unused expander.
9587 (avx512bw_<shift_insn><mode>3): New define_insn.
9588 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9589 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9591 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9594 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9597 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9599 PR rtl-optimization/82396
9600 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9601 (autopref_multipass_init): Simplify initialization.
9602 (autopref_rank_data): Simplify sort order.
9603 * gcc/sched-int.h (autopref_multipass_data_): Remove
9604 multi_mem_insn_p, min_offset and max_offset.
9606 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9609 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9610 Check special value of flag_omit_frame_pointer.
9611 (aarch64_can_eliminate): Likewise.
9612 (aarch64_override_options_after_change_1): Simplify handling of
9613 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9615 2017-10-24 Richard Biener <rguenther@suse.de>
9617 PR tree-optimization/82697
9618 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9619 zero for conditional load and unconditional store.
9621 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9623 * doc/install.texi: Document bootstrap-cet.
9625 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9628 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9629 ENDBR instruction at function entrance if function is only
9632 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9635 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9636 patterns to better describe from which operation the CF is computed.
9637 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9638 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9639 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9640 is 0, use _0 suffixed expanders instead of emitting a comparison
9643 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9645 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9646 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9648 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9651 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9652 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9653 * loop-iv.c (iv_get_reaching_def): Likewise.
9654 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9655 variable is promoted and the partition contains undefined values.
9657 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9659 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9661 (nios2_address_cost): Define.
9662 (nios2_legitimize_address): Recognize (exp + constant) directly.
9663 (TARGET_ADDRESS_COST): Define.
9665 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9667 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9668 (nios2_symbolic_memory_operand_p): Declare.
9669 (nios2_split_large_constant): Declare.
9670 (nios2_split_symbolic_memory_operand): Declare.
9671 * config/nios2/nios2.c: Adjust includes.
9672 (nios2_symbolic_constant_allowed): New.
9673 (nios2_symbolic_constant_p): New.
9674 (nios2_plus_symbolic_constant_p): New.
9675 (nios2_valid_addr_expr_p): Recognize addresses involving
9677 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9678 (nios2_symbolic_memory_operand_p): New.
9679 (nios2_large_constant_p): New.
9680 (nios2_split_large_constant): New.
9681 (nios2_split_plus_large_constant): New.
9682 (nios2_split_symbolic_memory_operand): New.
9683 (nios2_legitimize_address): Code refactoring. Handle addresses
9684 involving symbolic constants.
9685 (nios2_emit_move_sequence): Likewise.
9686 (nios2_print_operand): Improve error output.
9687 (nios2_print_operand_address): Handle LO_SUM.
9688 (nios2_cdx_narrow_form_p): Likewise.
9689 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9690 operands involving symbolic constants.
9691 (movhi_internal, movsi_internal): Likewise.
9692 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9693 (extendhisi2, extendqi<mode>2): Likewise.
9695 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9697 * tree-pass.h (PROP_rtl_split_insns): Define.
9698 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9700 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9702 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9704 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9707 * target.def (const_not_ok_for_debug_p): Default to
9708 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9709 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9710 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9711 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9712 which targetm.const_not_ok_for_debug_p returned true.
9713 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9715 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9717 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9718 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9719 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9721 (ix86_const_not_ok_for_debug_p): New function.
9722 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9723 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9725 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9728 * system.h: Conditionally include "unique-ptr.h" if
9729 INCLUDE_UNIQUE_PTR is defined.
9730 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9731 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9733 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9735 * config/rl78/rl78.md: New define_expand "subdi3".
9737 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9740 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9741 DF_REF_INSN if DF_REF_INSN_INFO is false.
9743 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9745 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9746 xmm_move instead of sse_move.
9747 (sse_store_index): New function.
9748 (ix86_register_move_cost): Be more sensible about mismatch stall;
9749 model AVX moves correctly; make difference between sse->integer and
9751 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9752 moves; make difference between SSE and AVX.
9753 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9754 and zmm_move. Increase size of sse load and store tables;
9755 add unaligned load and store tables; add ssemmx_to_integer.
9756 * x86-tune-costs.h: Update all entries according to real
9757 move latencies from Agner Fog's manual and chip documentation.
9759 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9762 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9763 * config/i386/constraints.md (Wf): New constraint.
9764 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9765 (cmp<dwi>_doubleword): Removed.
9766 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9767 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9768 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9769 expand with cmp<dwi>_doubleword. For LTU and GEU use
9770 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9772 * common.opt (gcolumn-info): Enable by default.
9773 * doc/invoke.texi (gcolumn-info): Document new default.
9775 2017-10-23 Richard Biener <rguenther@suse.de>
9777 PR tree-optimization/82672
9778 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9779 Fold the stmt if we propagated into it.
9781 2017-10-23 Richard Biener <rguenther@suse.de>
9783 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9784 (bitmap_remove_expr_from_set): ... this. All callers call this
9785 for non-constant values.
9786 (bitmap_set_subtract): Rename to...
9787 (bitmap_set_subtract_expressions): ... this. Adjust and
9789 (bitmap_set_contains_value): Remove superfluous check.
9790 (bitmap_set_replace_value): Inline into single caller ...
9791 (bitmap_value_replace_in_set): ... here and simplify.
9792 (dependent_clean): Merge into ...
9793 (clean): ... this using an overload. Adjust.
9794 (prune_clobbered_mems): Adjust.
9795 (compute_antic_aux): Likewise.
9796 (compute_partial_antic_aux): Likewise.
9798 2017-10-23 Richard Biener <rguenther@suse.de>
9800 PR tree-optimization/82129
9802 2017-08-01 Richard Biener <rguenther@suse.de>
9804 PR tree-optimization/81181
9805 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9806 (compute_antic): ... end of iteration here.
9808 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9810 * target.def (starting_frame_offset): New hook.
9811 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9812 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9813 * doc/tm.texi.in: Regenerate.
9814 * hooks.h (hook_hwi_void_0): Declare.
9815 * hooks.c (hook_hwi_void_0): New function.
9816 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9817 STARTING_FRAME_OFFSET.
9818 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9819 * reload1.c (reload): Likewise.
9820 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9821 instead of STARTING_FRAME_OFFSET.
9822 * function.c (try_fit_stack_local): Likewise.
9823 (assign_stack_local_1): Likewise
9824 (instantiate_virtual_regs): Likewise.
9825 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9826 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9827 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9828 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9829 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9830 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9831 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9832 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9833 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9834 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9835 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9836 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9837 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9838 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9839 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9840 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9841 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9842 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9843 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9844 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9845 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9846 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9847 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9848 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9849 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9850 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9851 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9852 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9853 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9854 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9855 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9856 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9857 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9858 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9859 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9860 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9861 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9862 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9863 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9864 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9865 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9866 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9867 return a HOST_WIDE_INT.
9868 (avr_builtin_setjmp_frame_value): Use it instead of
9869 STARTING_FRAME_OFFSET.
9870 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9871 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9872 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9874 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9875 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9876 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9877 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9878 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9879 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9880 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9881 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9882 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9883 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9884 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9885 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9887 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9888 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9889 * config/mips/mips.c (mips_compute_frame_info): Refer to
9890 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9891 (mips_starting_frame_offset): New function.
9892 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9893 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9894 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9895 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9896 and return a HOST_WIDE_INT.
9897 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9898 (mmix_initial_elimination_offset): Refer to
9899 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9900 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9901 * config/pa/pa.c (pa_starting_frame_offset): New function.
9902 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9903 (pa_expand_prologue): Likewise.
9904 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9905 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9906 !FRAME_GROWS_DOWNWARD handling to...
9907 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9908 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9909 !FRAME_GROWS_DOWNWARD handling to...
9910 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9911 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9912 !FRAME_GROWS_DOWNWARD handling to...
9913 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9914 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9916 (rs6000_starting_frame_offset): New function.
9917 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9918 !FRAME_GROWS_DOWNWARD handling to...
9919 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9920 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9921 !FRAME_GROWS_DOWNWARD handling to...
9922 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9923 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9924 !FRAME_GROWS_DOWNWARD handling to...
9925 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9926 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9927 (rs6000_starting_frame_offset): New function.
9928 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9929 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9930 * config/vax/vax.c (vax_starting_frame_offset): New function.
9931 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9932 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9933 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9934 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9935 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9936 * system.h (STARTING_FRAME_OFFSET): Poison.
9938 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9940 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9941 SCALAR_TYPE_MODE instead of TYPE_MODE.
9943 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9944 Alan Hayward <alan.hayward@arm.com>
9945 David Sherwood <david.sherwood@arm.com>
9947 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9949 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9950 Alan Hayward <alan.hayward@arm.com>
9951 David Sherwood <david.sherwood@arm.com>
9953 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9955 2017-10-23 Richard Biener <rguenther@suse.de>
9957 PR tree-optimization/82129
9958 * tree-ssa-pre.c (bitmap_set_and): Remove.
9959 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9960 canonicalizing expressions in the set to those with lowest
9961 ID rather than taking that from the first edge.
9963 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9965 * combine.c (rtx_equal_for_field_assignment_p): Use
9966 byte_lowpart_offset.
9968 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9969 Alan Hayward <alan.hayward@arm.com>
9970 David Sherwood <david.sherwood@arm.com>
9972 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9973 to a SUBREG_PROMOTED_VAR.
9975 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9976 Alan Hayward <alan.hayward@arm.com>
9977 David Sherwood <david.sherwood@arm.com>
9979 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9980 (expand_debug_source_expr): Likewise.
9981 * combine.c (combine_simplify_rtx): Likewise.
9982 * cse.c (fold_rtx): Likewise.
9983 * optabs.c (expand_float): Likewise.
9984 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9985 (simplify_binary_operation_1): Likewise.
9987 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9988 Alan Hayward <alan.hayward@arm.com>
9989 David Sherwood <david.sherwood@arm.com>
9991 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9992 (record_promoted_value): Likewise.
9993 * expr.c (expand_expr_real_2): Likewise.
9994 * ree.c (update_reg_equal_equiv_notes): Likewise.
9995 (combine_set_extension): Likewise.
9996 * rtlanal.c (low_bitmask_len): Likewise.
9997 * simplify-rtx.c (neg_const_int): Likewise.
9998 (simplify_binary_operation_1): Likewise.
10000 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10001 Alan Hayward <alan.hayward@arm.com>
10002 David Sherwood <david.sherwood@arm.com>
10004 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10005 * regcprop.c (maybe_mode_change): Likewise.
10006 * reload1.c (alter_reg): Likewise.
10008 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10010 * inchash.h (inchash::hash::add_wide_int): New function.
10011 * lto-streamer-out.c (hash_tree): Use it.
10013 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10015 * inchash.h (inchash::hash::add_wide_int): Rename to...
10016 (inchash::hash::add_hwi): ...this.
10017 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10018 (polymorphic_call_target_hasher::hash): Likewise.
10019 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10020 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10021 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10022 * lto-streamer-out.c (hash_tree): Likewise.
10023 * optc-save-gen.awk: Likewise.
10024 * tree.c (add_expr): Likewise.
10026 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10029 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10030 for ordered inequality comparisons even with TARGET_IEEE_FP.
10032 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10035 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10036 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10037 Expand with cmp<dwi>_doubleword.
10039 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10041 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10042 List CET intrinsics.
10043 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10044 specific to -fcf-protection option.
10046 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10048 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10049 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10050 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10051 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10052 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10053 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10054 (extra_objs): Add cet.o for Linux/x86 targets.
10055 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10056 * config/i386/cet.c: New file.
10057 * config/i386/cetintrin.h: Likewise.
10058 * config/i386/t-cet: Likewise.
10059 * config/i386/cpuid.h (bit_SHSTK): New.
10060 (bit_IBT): Likewise.
10061 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10062 pass IBT and SHSTK bits.
10063 * config/i386/i386-builtin-types.def
10064 (VOID_FTYPE_UNSIGNED_PVOID): New.
10065 (VOID_FTYPE_UINT64_PVOID): Likewise.
10066 * config/i386/i386-builtin.def: Add CET intrinsics.
10067 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10068 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10069 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10070 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10072 * config/i386/i386.c (rest_of_insert_endbranch): New.
10073 (pass_data_insert_endbranch): Likewise.
10074 (pass_insert_endbranch): Likewise.
10075 (make_pass_insert_endbranch): Likewise.
10076 (ix86_notrack_prefixed_insn_p): Likewise.
10077 (ix86_target_string): Add -mibt, -mshstk flags.
10078 (ix86_option_override_internal): Add flag_cf_protection
10080 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10081 (ix86_print_operand): Add 'notrack' prefix output.
10082 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10083 (ix86_expand_builtin): Expand CET intrinsics.
10084 (x86_output_mi_thunk): Add 'endbranch' instruction.
10085 * config/i386/i386.h (TARGET_IBT): New.
10086 (TARGET_IBT_P): Likewise.
10087 (TARGET_SHSTK): Likewise.
10088 (TARGET_SHSTK_P): Likewise.
10089 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10090 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10091 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10092 (builtin_setjmp_setup): New pattern.
10093 (builtin_longjmp): Likewise.
10094 (rdssp<mode>): Likewise.
10095 (incssp<mode>): Likewise.
10096 (saveprevssp): Likewise.
10097 (rstorssp): Likewise.
10098 (wrss<mode>): Likewise.
10099 (wruss<mode>): Likewise.
10100 (setssbsy): Likewise.
10101 (clrssbsy): Likewise.
10102 (nop_endbr): Likewise.
10103 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10105 * config/i386/immintrin.h: Include <cetintrin.h>.
10106 * config/i386/linux-common.h
10107 (file_end_indicate_exec_stack_and_cet): New prototype.
10108 (TARGET_ASM_FILE_END): New.
10110 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10112 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10113 latencies instead of having separate table; make difference between
10114 integer and float costs.
10115 * i386.h (processor_costs): Remove scalar_stmt_cost,
10116 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10117 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10119 * x86-tune-costs.h: Remove entries which has been removed in
10120 procesor_costs from all tables; make cond_taken_branch_cost
10121 and cond_not_taken_branch_cost COST_N_INSNS based.
10123 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10125 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10127 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10129 * config/i386/i386.md (isa): Remove fma_avx512f.
10130 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10131 <avx512>_fmadd_<mode>_mask3<round_name>,
10132 <avx512>_fmsub_<mode>_mask<round_name>,
10133 <avx512>_fmsub_<mode>_mask3<round_name>,
10134 <avx512>_fnmadd_<mode>_mask<round_name>,
10135 <avx512>_fnmadd_<mode>_mask3<round_name>,
10136 <avx512>_fnmsub_<mode>_mask<round_name>,
10137 <avx512>_fnmsub_<mode>_mask3<round_name>,
10138 <avx512>_fmaddsub_<mode>_mask<round_name>,
10139 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10140 <avx512>_fmsubadd_<mode>_mask<round_name>,
10141 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10142 (*vec_widen_umult_even_v16si<mask_name>,
10143 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10144 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10146 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10148 * extend.texi: Add 'nocf_check' documentation.
10149 * gimple.texi: Add second parameter to
10150 gimple_build_call_from_tree.
10151 * invoke.texi: Add -fcf-protection documentation.
10152 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10154 2017-10-20 Richard Biener <rguenther@suse.de>
10156 PR tree-optimization/82473
10157 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10158 the largest input type.
10160 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10162 * c-attribs.c (handle_nocf_check_attribute): New function.
10163 (c_common_attribute_table): Add 'nocf_check' handling.
10164 * gimple-parser.c: Add second argument NULL to
10165 gimple_build_call_from_tree.
10166 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10167 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10169 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10170 * common.opt: Add fcf-protection flag.
10171 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10172 * flag-types.h: Add enum cf_protection_level.
10173 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10174 Add 'nocf_check' attribute propagation to gimple call.
10175 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10176 (gimple_build_call_from_tree): Update prototype.
10177 (gimple_call_nocf_check_p): New function.
10178 (gimple_call_set_nocf_check): Likewise.
10179 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10180 * ipa-icf.c: Add nocf_check attribute in statement hash.
10181 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10182 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10183 * toplev.c (process_options): Add flag_cf_protection handling.
10185 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10187 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10189 2017-10-20 Richard Biener <rguenther@suse.de>
10191 PR tree-optimization/82603
10192 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10193 remove false predicated stores.
10195 2017-10-20 Richard Biener <rguenther@suse.de>
10197 * graphite-isl-ast-to-gimple.c
10198 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10199 Remove return value and simplify, dump copied stmt after lhs
10201 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10202 Reduce dump verbosity.
10203 (gsi_insert_earliest): Likewise.
10204 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10205 * graphite.c (print_global_statistics): Adjust dumping.
10206 (print_graphite_scop_statistics): Likewise.
10207 (print_graphite_statistics): Do not dump loops here.
10208 (graphite_transform_loops): But here.
10210 2017-10-20 Nicolas Roche <roche@adacore.com>
10212 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10213 * configure: Regenerate.
10215 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10218 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10219 functions when optimizing replace GIMPLE_RETURN stmts with
10220 calls to __builtin_unreachable ().
10223 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10224 for -fsanitize=thread link of executables.
10225 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10226 link of executables.
10229 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10230 New mode iterators.
10231 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10232 define_insns for logical vector shifts to use VI248_AVX512BW
10233 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10234 condition, useless isa and prefix attributes. Change the first
10235 2 of these define_insns to ...
10236 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10237 define_insn for avx512vl.
10238 (<shift_insn><mode>3): ... and this, new define_insn without
10239 masking for non-avx512vl.
10242 * config/i386/sse.md (*andnot<mode>3,
10243 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10244 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10245 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10246 not applied use empty suffix even for TARGET_AVX512VL.
10247 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10248 is applied, supply evex,evex or evex,evex,evex instead of just
10251 2017-10-20 Julia Koval <julia.koval@intel.com>
10253 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10254 (OPTION_MASK_ISA_GFNI_UNSET): New.
10255 (ix86_handle_option): Handle OPT_mgfni.
10256 * config/i386/cpuid.h (bit_GFNI): New.
10257 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10258 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10259 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10260 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10261 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10262 * config/i386/i386.opt: Add mgfni.
10264 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10266 * config/msp430/msp430.c (msp430_option_override): Disable
10267 -fdelete-null-pointer-checks.
10268 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10270 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10272 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10273 of x87 and SSE instructions.
10275 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10277 * asan.c (create_cond_insert_point): Do not update edge count.
10278 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10279 (afdo_propagate_circuit): Likewise.
10280 (afdo_calculate_branch_prob): Likewise.
10281 (afdo_annotate_cfg): Likewise.
10282 * basic-block.h (struct edge_def): Remove count.
10283 (edge_def::count): New accessor.
10284 * bb-reorder.c (rotate_loop): Update.
10285 (find_traces_1_round): Update.
10286 (connect_traces): Update.
10287 (sanitize_hot_paths): Update.
10288 * cfg.c (unchecked_make_edge): Update.
10289 (make_single_succ_edge): Update.
10290 (check_bb_profile): Update.
10291 (dump_edge_info): Update.
10292 (update_bb_profile_for_threading): Update.
10293 (scale_bbs_frequencies_int): Update.
10294 (scale_bbs_frequencies_gcov_type): Update.
10295 (scale_bbs_frequencies_profile_count): Update.
10296 (scale_bbs_frequencies): Update.
10297 * cfganal.c (connect_infinite_loops_to_exit): Update.
10298 * cfgbuild.c (compute_outgoing_frequencies): Update.
10299 (find_many_sub_basic_blocks): Update.
10300 * cfgcleanup.c (try_forward_edges): Update.
10301 (try_crossjump_to_edge): Update
10302 * cfgexpand.c (expand_gimple_cond): Update
10303 (expand_gimple_tailcall): Update
10304 (construct_exit_block): Update
10305 * cfghooks.c (verify_flow_info): Update
10306 (redirect_edge_succ_nodup): Update
10307 (split_edge): Update
10308 (make_forwarder_block): Update
10309 (duplicate_block): Update
10310 (account_profile_record): Update
10311 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10312 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10313 * cfgloopmanip.c (scale_loop_profile): Update.
10315 (lv_adjust_loop_entry_edge): Update.
10316 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10317 (force_nonfallthru_and_redirect): Update.
10318 (purge_dead_edges): Update.
10319 (rtl_flow_call_edges_add): Update.
10320 * cgraphunit.c (init_lowered_empty_function): Update.
10321 (cgraph_node::expand_thunk): Update.
10322 * gimple-pretty-print.c (dump_probability): Update.
10323 (dump_edge_probability): Update.
10324 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10325 * haifa-sched.c (sched_create_recovery_edges): Update.
10326 * hsa-gen.c (convert_switch_statements): Update.
10327 * ifcvt.c (dead_or_predicable): Update.
10328 * ipa-inline-transform.c (inline_transform): Update.
10329 * ipa-split.c (split_function): Update.
10330 * ipa-utils.c (ipa_merge_profiles): Update.
10331 * loop-doloop.c (add_test): Update.
10332 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10333 * lto-streamer-in.c (input_cfg): Update.
10334 (input_function): Update.
10335 * lto-streamer-out.c (output_cfg): Update.
10336 * modulo-sched.c (sms_schedule): Update.
10337 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10338 * predict.c (maybe_hot_edge_p): Update.
10339 (unlikely_executed_edge_p): Update.
10340 (probably_never_executed_edge_p): Update.
10341 (dump_prediction): Update.
10342 (drop_profile): Update.
10343 (propagate_unlikely_bbs_forward): Update.
10344 (determine_unlikely_bbs): Update.
10345 (force_edge_cold): Update.
10346 * profile.c (compute_branch_probabilities): Update.
10347 * reg-stack.c (better_edge): Update.
10348 * shrink-wrap.c (handle_simple_exit): Update.
10349 * tracer.c (better_p): Update.
10350 * trans-mem.c (expand_transaction): Update.
10351 (split_bb_make_tm_edge): Update.
10352 * tree-call-cdce.c: Update.
10353 * tree-cfg.c (gimple_find_sub_bbs): Update.
10354 (gimple_split_edge): Update.
10355 (gimple_duplicate_sese_region): Update.
10356 (gimple_duplicate_sese_tail): Update.
10357 (gimple_flow_call_edges_add): Update.
10358 (insert_cond_bb): Update.
10359 (execute_fixup_cfg): Update.
10360 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10361 * tree-complex.c (expand_complex_div_wide): Update.
10362 * tree-eh.c (lower_resx): Update.
10363 (unsplit_eh): Update.
10364 (cleanup_empty_eh_move_lp): Update.
10365 * tree-inline.c (copy_edges_for_bb): Update.
10366 (freqs_to_counts): Update.
10367 (copy_cfg_body): Update.
10368 * tree-ssa-dce.c (remove_dead_stmt): Update.
10369 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10370 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10371 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10372 (unloop_loops): Update.
10373 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10374 * tree-ssa-loop-split.c (connect_loops): Update.
10375 (split_loop): Update.
10376 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10377 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10378 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10379 * tree-ssa-reassoc.c (branch_fixup): Update.
10380 * tree-ssa-tail-merge.c (replace_block_by): Update.
10381 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10382 (compute_path_counts): Update.
10383 (update_profile): Update.
10384 (recompute_probabilities): Update.
10385 (update_joiner_offpath_counts): Update.
10386 (estimated_freqs_path): Update.
10387 (freqs_to_counts_path): Update.
10388 (clear_counts_path): Update.
10389 (ssa_fix_duplicate_block_edges): Update.
10390 (duplicate_thread_path): Update.
10391 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10392 (case_bit_test_cmp): Update.
10393 (collect_switch_conv_info): Update.
10394 (gen_inbound_check): Update.
10395 (do_jump_if_equal): Update.
10396 (emit_cmp_and_jump_insns): Update.
10397 * tree-tailcall.c (decrease_profile): Update.
10398 (eliminate_tail_call): Update.
10399 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10400 (vect_do_peeling): Update.
10401 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10402 * ubsan.c (ubsan_expand_null_ifn): Update.
10403 (ubsan_expand_ptr_ifn): Update.
10404 * value-prof.c (gimple_divmod_fixed_value): Update.
10405 (gimple_mod_pow2): Update.
10406 (gimple_mod_subtract): Update.
10407 (gimple_ic): Update.
10408 (gimple_stringop_fixed_value): Update.
10410 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10413 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10415 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10417 PR rtl-optimization/82395
10418 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10419 based on non_spilled_static_chain_regno_p.
10421 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10423 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10424 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10425 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10427 2017-10-19 Martin Sebor <msebor@redhat.com>
10429 PR tree-optimization/82596
10430 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10432 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10434 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10435 (get_mem_refs_of_builtin_call): Likewise.
10436 * builtins.c (expand_builtin_apply): Adjust call to
10437 allocate_dynamic_stack_space.
10438 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10439 the third argument to allocate_dynamic_stack_space, otherwise -1.
10440 (expand_builtin): Deal with all alloca variants.
10441 (is_inexpensive_builtin): Likewise.
10442 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10443 * calls.c (special_function_p): Deal with all alloca variants.
10444 (initialize_argument_information): Adjust call to
10445 allocate_dynamic_stack_space.
10446 (expand_call): Likewise.
10447 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10448 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10449 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10450 use it for the stack usage computation.
10451 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10452 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10453 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10454 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10455 (in_loop_p): Remove first argument and useless check.
10456 (pass_walloca::execute): Remove useless test and adjust call to above.
10457 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10458 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10459 (gimplify_call_expr): Deal with all alloca variants.
10460 * hsa-gen.c (gen_hsa_alloca): Likewise.
10461 (gen_hsa_insns_for_call): Likewise.
10462 * ipa-pure-const.c (special_builtin_state): Likewise.
10463 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10464 * tree-object-size.c (alloc_object_size): Likewise.
10465 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10466 (call_may_clobber_ref_p_1): Likewise.
10467 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10468 (ccp_fold_stmt): Likewise.
10469 (optimize_stack_restore): Likewise.
10470 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10471 (mark_all_reaching_defs_necessary_1): Likewise.
10472 (propagate_necessity): Likewise.
10473 (eliminate_unnecessary_stmts): Likewise.
10474 * tree.c (build_common_builtin_nodes): Build
10475 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10476 (build_alloca_call_expr): New function.
10477 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10478 (CASE_BUILT_IN_ALLOCA): Likewise.
10479 (build_alloca_call_expr): Declare.
10480 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10482 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10485 * dwarf2out.c (new_die_raw): New static inline function.
10486 (new_die): Use it to create the DIE.
10487 (add_AT_external_die_ref): Likewise.
10488 (clone_die): Likewise.
10489 (clone_as_declaration): Likewise.
10490 (dwarf2out_vms_debug_main_pointer): Likewise.
10491 (base_type_die): Likewise. Remove early return for corner cases.
10492 Do not call add_pubtype on the DIE here.
10493 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10494 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10495 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10496 native order exists for base types, attach the DIE manually and call
10497 add_pubtype on it. Do not equate a reverse order DIE to the type.
10499 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10501 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10502 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10504 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10506 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10508 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10509 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10510 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10511 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10512 * opts.c (sanitizer_opts): Add builtin.
10513 * ubsan.c (instrument_builtin): New function.
10514 (pass_ubsan::execute): Call it.
10515 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10516 * doc/invoke.texi: Document -fsanitize=builtin.
10518 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10519 builtins, store max (log2 (align), 0) into uchar field instead of
10520 align into uptr field.
10521 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10522 store uchar 0 field instead of uptr 0 field.
10523 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10524 instead of passing one address of struct with 2 locations pass
10525 two addresses of structs with 1 location each.
10526 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10527 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10528 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10529 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10530 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10531 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10532 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10533 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10535 2017-10-19 Martin Liska <mliska@suse.cz>
10538 * file-find.c (remove_prefix): Remove.
10539 * file-find.h (remove_prefix): Likewise.
10540 * gcc-ar.c: Remove smartness of lookup.
10542 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10544 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10545 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10546 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10548 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10551 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10552 (setcc + and to xor + setcc): New peephole2.
10554 2017-10-19 Tom de Vries <tom@codesourcery.com>
10556 * doc/sourcebuild.texi (Test Directives, Variants of
10557 dg-require-support): Add dg-require-stack-size.
10559 2017-10-19 Martin Liska <mliska@suse.cz>
10562 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10563 that have a large alignment.
10564 (gimplify_target_expr): Likewise.
10566 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10568 PR rtl-optimization/82602
10569 * ira.c (rtx_moveable_p): Return false for volatile asm.
10571 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10574 * config/i386/i386-modes.def (CCGZ): New CC mode.
10575 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10576 * config/i386/predicates.md (ix86_comparison_operator):
10578 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10579 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10580 with double-word subtraction.
10581 (put_condition_code): Handle CCGZmode.
10583 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10585 * wide-int.cc (debug (const wide_int &)): New.
10586 (debug (const wide_int *)): New.
10587 (debug (const widest_int &)): New.
10588 (debug (const widest_int *)): New.
10590 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10592 PR middle-end/82556
10593 * lra-constraints.c (curr_insn_transform): Use non-input operand
10594 instead of output one for matched reload.
10596 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10598 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10599 (tree-ssa-loop-ivopts.h): New header file.
10600 (struct builtin_info): New fields.
10601 (classify_builtin_1): Compute and record base and offset parts for
10602 memset builtin partition by calling strip_offset.
10603 (offset_cmp, fuse_memset_builtins): New functions.
10604 (finalize_partitions): Fuse adjacent memset partitions by calling
10606 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10607 Expose the interface.
10608 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10610 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10612 PR tree-optimization/82574
10613 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10614 that data reference must be executed exactly once per iteration
10615 against the outermost loop in nest.
10616 (classify_partition): Update call to above function.
10618 2017-10-18 Richard Biener <rguenther@suse.de>
10620 PR tree-optimization/82591
10621 * graphite.c (graphite_transform_loops): Move code gen message
10623 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10624 Here. Handle scop_to_isl_ast failing.
10625 (scop_to_isl_ast): Limit the number of ISL operations.
10627 2017-10-18 Richard Biener <rguenther@suse.de>
10629 * graphite-isl-ast-to-gimple.c
10630 (translate_isl_ast_to_gimple::set_rename): Simplify.
10631 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10632 (graphite_copy_stmts_from_block): ... here.
10633 (copy_bb_and_scalar_dependences): Simplify.
10634 (add_parameters_to_ivs_params): Canonicalize.
10635 (generate_entry_out_of_ssa_copies): Simplify.
10636 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10637 by passing in ISL dimension.
10638 (parameter_index_in_region_1): Rename to ...
10639 (parameter_index_in_region): ... this.
10640 (extract_affine): Adjust assert, pass down parameter index.
10641 (add_param_constraints): Use range-info when available.
10642 (build_scop_context): Adjust.
10643 * sese.c (new_sese_info): Adjust.
10644 (free_sese_info): Likewise.
10645 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10646 Remove unused typedefs.
10647 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10649 2017-10-18 Martin Liska <mliska@suse.cz>
10651 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10653 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10655 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10656 (s390_sched_init): Do not reset s390_sched_state if we entered the
10657 current basic block via a fallthru edge and all others are unlikely.
10659 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10661 * config/s390/s390.c (NUM_SIDES): New variable.
10662 (LONGRUNNING_THRESHOLD): New variable.
10663 (LATENCY_FACTOR): New variable.
10664 (s390_sched_score): Decrease score for long-running instructions on
10666 (s390_sched_variable_issue): Perform bookkeeping for long-running
10669 2017-10-18 Richard Biener <rguenther@suse.de>
10671 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10672 Simplify with removal of the parameter rename map.
10673 (set_rename): Likewise.
10674 (should_copy_to_new_region): Likewise.
10675 (graphite_copy_stmts_from_block): Likewise.
10676 (copy_bb_and_scalar_dependences): Remove initialization of
10677 unused copied_bb_map.
10678 (copy_def): Remove.
10679 (copy_internal_parameters): Likewise.
10680 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10681 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10682 Use INTEGRAL_TYPE_P.
10683 (parameter_index_in_region_1): Rename to ...
10684 (assign_parameter_index_in_region): ... this. Assert we have
10685 a parameter we handle.
10686 (scan_tree_for_params): Adjust.
10687 * sese.h (parameter_rename_map_t): Remove.
10688 (struct sese_info_t): Remove unused parameter_rename_map and
10689 copied_bb_map members.
10690 * sese.c (new_sese_info): Adjust.
10691 (free_sese_info): Likewise.
10693 2017-10-18 Martin Liska <mliska@suse.cz>
10696 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10697 on an abnormal edge.
10699 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10701 * doc/invoke.texi (ffunction-sections and fdata-sections):
10704 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10706 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10707 the use statement can throw internally.
10709 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10711 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10712 any RTX present on the RHS of a SET.
10713 * compare-elim.c (try_eliminate_compare): Restore comment.
10715 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10717 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10718 may be also called on tcc_type nodes.
10719 * langhooks.c (lhd_tree_size): Likewise.
10721 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10723 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10724 format_warning_at_substring.
10725 (maybe_warn): Convert source_range * param to a location_t. Pass
10726 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10727 (format_directive): Remove code to extract source_ranges and
10728 source_range * in favor of just a location_t.
10729 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10731 * substring-locations.c (format_warning_va): Convert
10732 source_range * param to a location_t.
10733 (format_warning_at_substring): Likewise.
10734 * substring-locations.h (format_warning_va): Likewise.
10735 (format_warning_at_substring): Likewise.
10737 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10739 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10741 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10742 and scatter/gather ops.
10744 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10745 vec_gather_load and vec_scatter_store.
10746 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10747 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10748 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10749 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10750 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10751 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10753 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10755 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10756 Detect FTST instruction and handle its register pops. Only pop
10757 second operand if can_pop_second_op is true.
10758 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10759 set can_pop_second_op to false in the compare_for_stack_reg call.
10761 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10762 output_fp_compare for stack register operands.
10763 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10764 instructions here. Do not emit stack register pops here. Assert
10765 that FCOMPP pops next to top stack register. Rewrite function.
10767 2017-10-17 Nathan Sidwell <nathan@acm.org>
10769 PR middle-end/82577
10770 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10771 use DECL_ASSEMBLER_NAME_RAW.
10773 PR middle-end/82546
10774 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10777 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10778 Wilco Dijkstra <wilco.dijkstra@arm.com>
10780 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10781 converstion to Pmode from the buf_addr.
10783 2017-10-17 Richard Biener <rguenther@suse.de>
10785 * graphite-dependences.c (scop_get_reads_and_writes): Change
10786 output parameters to references.
10788 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10790 PR 71026/tree-optimization
10791 * fold-const.c (distribute_real_division): Removed.
10792 (fold_binary_loc): Remove calls to distribute_real_divison.
10794 2017-10-17 Richard Biener <rguenther@suse.de>
10796 * graphite-scop-detection.c
10797 (scop_detection::stmt_has_simple_data_refs_p): Always use
10798 the full nest as region.
10799 (try_generate_gimple_bb): Likewise.
10800 * sese.c (scalar_evolution_in_region): Simplify now that
10801 SCEV can handle instantiation in regions.
10802 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10803 in the non-loop part of a function if requested.
10805 2017-10-17 Richard Biener <rguenther@suse.de>
10807 PR tree-optimization/82563
10808 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10810 (graphite_regenerate_ast_isl): Call it.
10811 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10813 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10815 PR tree-optimization/82549
10816 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10817 Formatting fixes. Instead of calling make_bit_field_ref with negative
10820 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10822 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10823 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10824 _mm_maskz_reduce_ss): New.
10825 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10826 __builtin_ia32_reducess_mask): Ditto..
10827 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10828 * config/i386/sse.md (reduces<mode>): Renamed to ...
10829 (reduces<mode><mask_scalar_name>): ... this.
10830 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10832 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10833 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10835 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10837 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10838 * selftest-run-tests.c (selftest::run_tests): Call
10839 selftest::unique_ptr_tests_cc_tests.
10840 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10841 * unique-ptr-tests.cc: New file.
10843 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10846 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10848 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10849 (make_hard_regno_born, make_hard_regno_dead): Update
10850 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10852 2017-10-16 Jeff Law <law@redhat.com>
10854 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10856 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10858 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10860 2017-10-16 Olivier Hainque <hainque@adacore.com>
10862 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10863 with_cpu if we were configured for an e500v2 target cpu name.
10865 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10867 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10868 * doc/invoke.texi: Document +nodsp as a valid extension for
10871 2017-10-16 Martin Liska <mliska@suse.cz>
10873 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10874 (test_set_range): Likewise.
10875 (test_range_functions): Rename to ...
10876 (test_bit_in_range): ... this.
10877 (sbitmap_c_tests): Add new test.
10879 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10881 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10883 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10884 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10886 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10888 * config/aarch64/aarch64-builtins.c
10889 (aarch64_types_quadopu_lane_qualifiers): New.
10890 (TYPES_QUADOPU_LANE): New.
10891 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10892 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10893 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10894 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10895 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10896 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10897 (Vdottype, DOTPROD): New.
10898 (sur): Add SDOT and UDOT.
10900 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10902 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10903 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10904 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10905 Add TARGET_DOTPROD.
10906 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10907 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10908 Enable TARGET_DOTPROD.
10909 (cortex-a75.cortex-a55): Likewise.
10910 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10912 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10914 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10915 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10917 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10919 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10920 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10921 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10922 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10923 * config/arm/types.md (neon_dot, neon_dot_q): New.
10924 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10926 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10928 * config/arm/arm.h (TARGET_DOTPROD): New.
10929 * config/arm/arm.c (arm_arch_dotprod): New.
10930 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10931 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10932 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10933 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10934 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10935 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10936 * doc/invoke.texi (armv8.2-a): Document dotprod
10938 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10940 * i386.c (ix86_vec_cost): New function.
10941 (ix86_rtx_costs): Handle vector operations better.
10942 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10943 * x86-tune-costs.h: Add new costs to all tables.
10945 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10947 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10949 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10950 divsd, sqrtss and sqrtsd
10951 * x86-tune-costs.h: Add new entries to all costs.
10952 (znver1_cost): Fix to match real instruction latencies.
10954 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10955 Michael Collison <michael.collison@arm.com>
10957 * compare-elim.c: Include emit-rtl.h.
10958 (can_merge_compare_into_arith): New function.
10959 (try_validate_parallel): Likewise.
10960 (try_merge_compare): Likewise.
10961 (try_eliminate_compare): Call the above when no previous clobber
10963 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10966 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10968 PR middle-end/62263
10969 PR middle-end/82498
10970 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10971 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10973 PR middle-end/62263
10974 PR middle-end/82498
10975 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10976 to be any operand_equal_p operands. For & (B - 1) require
10977 B to be power of 2. Recognize
10978 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10980 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10983 * optabs.c (expand_memory_blockage): Fix call of
10984 targetm.have_memory_blockage.
10986 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10989 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10990 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10991 objects to extra_objs instead of overwriting it.
10993 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10995 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10996 Use any_fp_register_operand as operand[3] predicate. Simplify
10997 equality test for operands[2] and operands[4] memory location.
10998 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10999 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11000 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11001 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11002 any_fp_register_operand as operand[1] predicate. Simplify
11003 equality test for operands[0] and operands[3] memory location.
11004 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11005 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11006 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11008 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11010 * target-insns.def: Add memory_blockage.
11011 * optabs.c (expand_memory_blockage): New function.
11012 (expand_asm_memory_barrier): Rename ...
11013 (expand_asm_memory_blockage): ... to this.
11014 (expand_mem_thread_fence): Call expand_memory_blockage
11015 instead of expand_asm_memory_barrier.
11016 (expand_mem_singnal_fence): Ditto.
11017 (expand_atomic_load): Ditto.
11018 (expand_atomic_store): Ditto.
11019 * doc/md.texi (Standard Pattern Names For Generation):
11020 Document memory_blockage instruction pattern.
11022 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11024 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11025 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11026 * config/rl78/rl78.md: New define_expand "adddi3".
11028 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11030 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11033 2017-10-13 Jeff Law <law@redhat.com>
11035 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11037 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11040 * internal-fn.c (expand_mul_overflow): If both operands have
11041 the same highpart of -1 or 0 and the topmost bit of lowpart
11042 is different, overflow is if res <= 0 rather than res < 0.
11044 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11046 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11047 TARGET_P9_VECTOR code for unaligned_load case.
11049 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11051 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11053 2017-10-13 Nathan Sidwell <nathan@acm.org>
11055 * tree-core.h (tree_contains_struct): Make bool.
11056 * tree.c (tree_contains_struct): Likewise.
11057 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11058 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11059 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11060 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11061 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11063 2017-10-13 Richard Biener <rguenther@suse.de>
11065 * graphite-isl-ast-to-gimple.c
11066 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11067 parameters and dominance check.
11068 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11069 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11070 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11071 Do not update SSA form here or do intermediate IL verification.
11072 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11073 (graphite_initialize): Remove check on the number of loops in
11074 the function and inline into graphite_transform_loops.
11075 (graphite_finalize): Inline into graphite_transform_loops.
11076 (graphite_transform_loops): Perform SSA update and IL verification
11078 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11080 2017-10-13 Richard Biener <rguenther@suse.de>
11082 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11083 graphite_expression_type_precision): Avoid global constructor
11085 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11086 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11087 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11088 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11089 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11090 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11092 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11095 * config/i386/i386.h (ix86_red_zone_size): New.
11096 * config/i386/i386.md (push peephole2s): Replace
11097 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11099 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11100 Alan Hayward <alan.hayward@arm.com>
11101 David Sherwood <david.sherwood@arm.com>
11103 * combine.c (can_change_dest_mode): Reject changes in
11104 REGMODE_NATURAL_SIZE.
11106 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11107 Alan Hayward <alan.hayward@arm.com>
11108 David Sherwood <david.sherwood@arm.com>
11110 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11111 (expand_debug_source_expr): Likewise.
11112 * combine.c (combine_simplify_rtx): Likewise.
11113 * cse.c (fold_rtx): Likewise.
11114 * fwprop.c (canonicalize_address): Likewise.
11115 * targhooks.c (default_shift_truncation_mask): Likewise.
11117 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11118 Alan Hayward <alan.hayward@arm.com>
11119 David Sherwood <david.sherwood@arm.com>
11121 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11122 (widened_mode): Likewise.
11123 (expand_unop): Likewise.
11124 * ree.c (transform_ifelse): Likewise.
11125 (merge_def_and_ext): Likewise.
11126 (combine_reaching_defs): Likewise.
11127 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11129 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11130 Alan Hayward <alan.hayward@arm.com>
11131 David Sherwood <david.sherwood@arm.com>
11133 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11134 * combine.c (gen_lowpart_for_combine): Likewise.
11135 * dwarf2out.c (rtl_for_decl_location): Likewise.
11136 * final.c (alter_subreg): Likewise.
11137 * rtlhooks.c (gen_lowpart_general): Likewise.
11138 (gen_lowpart_if_possible): Likewise.
11140 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11141 Alan Hayward <alan.hayward@arm.com>
11142 David Sherwood <david.sherwood@arm.com>
11144 * calls.c (expand_call): Use subreg_lowpart_offset.
11145 * cse.c (cse_insn): Likewise.
11146 * regcprop.c (copy_value): Likewise.
11147 (copyprop_hardreg_forward_1): Likewise.
11149 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11152 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11153 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11154 =Q constraints to +Q and into insn condition add check
11155 that operands[0] and operands[1] are equal.
11156 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11157 =Q constraints to +Q and into insn condition add check
11158 that operands[0] is equal to either operands[1] or operands[2].
11161 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11162 instead of handling MINUS_EXPR twice (once for each argument),
11163 canonicalize operand order and handle just once, use rtype where
11164 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11167 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11168 any values of __C while still being pattern recognizable as a simple
11169 rotate instruction.
11171 2017-10-13 Richard Biener <rguenther@suse.de>
11173 PR tree-optimization/82451
11175 2017-10-02 Richard Biener <rguenther@suse.de>
11177 PR tree-optimization/82355
11178 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11179 a mapping for the enclosing loop but avoid generating one for
11180 the loop tree root.
11181 (copy_bb_and_scalar_dependences): Remove premature codegen
11182 error on PHIs in blocks duplicated into multiple places.
11183 * graphite-scop-detection.c
11184 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11185 in the region use it as loop and nest to analyze the DR in.
11186 (try_generate_gimple_bb): Likewise.
11187 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11188 (add_loop_constraints): For blocks in a loop not in the region
11189 create a dimension with a single iteration.
11190 * sese.h (gbb_loop_at_index): Remove assert.
11192 * cfgloop.c (loop_preheader_edge): For the loop tree root
11193 return the single successor of the entry block.
11194 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11195 Reset the SCEV hashtable and niters.
11196 * graphite-scop-detection.c
11197 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11198 assert that we only have POLYNOMIAL_CHREC that vary in loops
11199 contained in the region.
11200 (scop_detection::graphite_can_represent_expr): Adjust.
11201 (scop_detection::stmt_has_simple_data_refs_p): For loops
11202 not in the region set loop to NULL. The nest is now the
11203 entry edge to the region.
11204 (try_generate_gimple_bb): Likewise.
11205 * sese.c (scalar_evolution_in_region): Adjust for
11206 instantiate_scev change.
11207 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11208 Make nest parameter the edge into the region.
11209 (create_data_ref): Likewise.
11210 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11211 entry edge into a region and adjust instantiate_scev calls.
11212 (create_data_ref): Likewise.
11213 (graphite_find_data_references_in_stmt): Likewise.
11214 (find_data_references_in_stmt): Pass the loop preheader edge
11215 from the nest argument.
11216 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11217 parameter the edge into the region.
11218 (instantiate_parameters): Use the loop preheader edge as entry.
11219 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11221 (get_instantiated_value_entry): Make instantiate_below parameter
11222 the edge into the region.
11223 (instantiate_scev_name): Likewise. Adjust dominance checks,
11224 when we cannot use loop-based instantiation instantiate by
11225 walking use-def chains.
11226 (instantiate_scev_poly): Adjust.
11227 (instantiate_scev_binary): Likewise.
11228 (instantiate_scev_convert): Likewise.
11229 (instantiate_scev_not): Likewise.
11230 (instantiate_array_ref): Remove.
11231 (instantiate_scev_3): Likewise.
11232 (instantiate_scev_2): Likewise.
11233 (instantiate_scev_1): Likewise.
11234 (instantiate_scev_r): Do not blindly handle N-operand trees.
11235 Do not instantiate array-refs. Handle all constants and invariants.
11236 (instantiate_scev): Make instantiate_below parameter
11237 the edge into the region.
11238 (resolve_mixers): Use the loop preheader edge for the region
11239 parameter to instantiate_scev_r.
11240 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11242 2017-10-13 Richard Biener <rguenther@suse.de>
11244 PR tree-optimization/82525
11245 * graphite-isl-ast-to-gimple.c
11246 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11248 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11249 Fail code generation when we cannot represent the isl integer.
11250 (binary_op_to_tree): Elide modulo operations that are no-ops
11251 in the type we code generate. Remove now superfluous code
11254 2017-10-13 Richard Biener <rguenther@suse.de>
11256 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11257 (scop_detection::harmful_loop_in_region): Remove premature
11258 IV type restriction.
11259 (scop_detection::graphite_can_represent_scev): We can handle
11260 pointer IVs just fine.
11262 2017-10-13 Alan Modra <amodra@gmail.com>
11264 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11265 "Clobbers and Scratch Registers". Add paragraph on
11266 alternative to clobbers for scratch registers and OpenBLAS
11269 2017-10-13 Alan Modra <amodra@gmail.com>
11271 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11272 example of a memory input for a string of known length. Move
11273 commentary out of table. Add a number of new examples
11274 covering array memory inputs.
11276 2017-10-12 Martin Liska <mliska@suse.cz>
11278 PR tree-optimization/82493
11279 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11280 (test_range_functions): New function.
11281 (sbitmap_c_tests): Likewise.
11282 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11283 * selftest.h (sbitmap_c_tests): New function.
11285 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11287 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11289 * config/rs6000/amo.h: Fix spacing issue.
11291 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11294 * config/i386/i386.md (*ashl<mode>3_mask_1,
11295 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11296 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11299 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11301 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11302 (profile_probability): Set max_probability
11303 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11305 * profile-count.c (profile_probability::differs_from_p): Do not
11306 rely on max_probaiblity == 10000
11308 2017-10-12 Jeff Law <law@redhat.com>
11310 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11313 2017-10-12 Martin Sebor <msebor@redhat.com>
11317 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11318 (handle_alias_pairs): Call it.
11319 * common.opt (-Wattribute-alias): New option.
11320 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11321 * doc/invoke.texi (-Wattribute-alias): Document.
11323 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11326 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11328 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11330 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11331 (make_hard_regno_born, make_hard_regno_dead): Update
11332 bb_killed_pseudos and bb_gen_pseudos.
11334 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11336 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11338 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11340 * config/alpha/alpha.c (alpha_split_conditional_move):
11341 Use std::swap instead of manually swapping.
11342 (alpha_stdarg_optimize_hook): Ditto.
11343 (alpha_canonicalize_comparison): Ditto.
11345 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11347 * tree-loop-distribution.c (struct builtin_info): New struct.
11348 (struct partition): Refactor fields into struct builtin_info.
11349 (partition_free): Free struct builtin_info.
11350 (build_size_arg_loc, build_addr_arg_loc): Delete.
11351 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11352 information from struct builtin_info.
11353 (find_single_drs): New function refactored from classify_partition.
11354 Also moved builtin validity checks to this function.
11355 (compute_access_range, alloc_builtin): New functions.
11356 (classify_builtin_st, classify_builtin_ldst): New functions.
11357 (classify_partition): Refactor code into functions find_single_drs,
11358 classify_builtin_st and classify_builtin_ldst.
11359 (distribute_loop): Don't do runtime alias check when distributing
11361 (find_seed_stmts_for_distribution): New function.
11362 (pass_loop_distribution::execute): Refactor code finding seed
11363 stmts into above function. Support distribution for the innermost
11364 two-level loop nest. Adjust dump information.
11366 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11368 * tree-loop-distribution.c: Adjust the general comment.
11369 (NUM_PARTITION_THRESHOLD): New macro.
11370 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11371 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11372 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11373 in call to build_partition_graph.
11374 (finalize_partitions): New parameter. Make loop distribution more
11375 conservative by fusing more partitions.
11376 (distribute_loop): Don't do runtime alias check in case of loop nest
11378 (find_seed_stmts_for_distribution): New function.
11379 (prepare_perfect_loop_nest): New function.
11380 (pass_loop_distribution::execute): Refactor code finding seed stmts
11381 and loop nest into above functions. Support loop nest distribution.
11382 Adjust dump information accordingly.
11384 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11386 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11387 and set PTYPE_SEQUENTIAL for merged partition.
11389 2017-10-12 Richard Biener <rguenther@suse.de>
11391 PR tree-optimization/69728
11393 2017-09-19 Richard Biener <rguenther@suse.de>
11395 PR tree-optimization/69728
11396 * graphite-sese-to-poly.c (schedule_error): New global.
11397 (add_loop_schedule): Handle empty domain by failing the
11399 (build_original_schedule): Handle schedule_error.
11401 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11402 domain by returning an unchanged schedule.
11404 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11406 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11407 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11409 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11411 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11414 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11417 * expr.c (store_field): Don't optimize away bitsize == 0 store
11418 from CALL_EXPR with addressable return type.
11420 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11422 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11423 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11424 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11425 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11426 TARGET_ISEL instead of TARGET_ISEL<sel>.
11428 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11430 * config/rs6000/rs6000.c
11431 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11433 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11435 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11437 (reg_or_cint_operand): Fix comment.
11438 (reg_or_zero_operand): New predicate.
11439 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11440 * config/rs6000/rs6000.c (output_isel): Delete.
11441 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11442 instead of reg_or_cint_operand. Output instruction directly (not via
11444 (isel_unsigned_<mode>): Ditto.
11445 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11446 gpc_reg_operand. Add an instruction alternative for this. Output
11447 instruction directly.
11448 (*isel_reversed_unsigned_<mode>): Ditto.
11450 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11452 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11453 (TARGET_CANONICALIZE_COMPARISON): Define.
11455 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11458 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11459 Check whether the dest is REG before adding REG_EQUIV note.
11461 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11464 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11466 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11467 (make_hard_regno_born, make_hard_regno_dead): Update
11468 bb_killed_pseudos and bb_gen_pseudos.
11470 2017-10-11 Nathan Sidwell <nathan@acm.org>
11472 * incpath.h (enum incpath_kind): Name enum, prefix values.
11473 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11474 * incpath.c (heads, tails): Use INC_MAX.
11475 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11476 (merge_include_chains, split_quote_chain,
11477 register_include_chains): Update incpath_kind names.
11478 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11479 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11481 (add_framework_path, darwin_register_objc_includes): Likewise.
11482 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11484 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11486 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11487 Do not use float_operator operator predicate.
11488 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11489 * config/i386/predicates.md (float_operator): Remove predicate.
11491 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11493 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11494 (*jccxf_i387): Ditto.
11495 (*jcc<mode>_i387): Ditto.
11496 (*jccu<mode>_i387): Ditto.
11497 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11498 (*jcc_*_i387 splitters): Remove.
11499 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11500 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11501 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11504 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11506 * profile-count.h (slow_safe_scale_64bit): New function.
11507 (safe_scale_64bit): New inline.
11508 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11509 * profile-count.c: Include wide-int.h
11510 (slow_safe_scale_64bit): New.
11512 2017-10-11 Nathan Sidwell <nathan@acm.org>
11514 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11515 HAS_DECL_ASSEMBLER_NAME_P.
11516 * gimple-expr.c (gimple_decl_printable_name: Check
11517 HAS_DECL_ASSEMBLER_NAME_P too.
11518 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11519 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11520 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11521 * passes.c (rest_of_decl_compilation): Check
11522 HAS_DECL_ASSEMBLER_NAME_P too.
11523 * recog.c (verify_changes): Likewise.
11524 * tree-pretty-print.c (dump_decl_name): Likewise.
11525 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11527 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11528 (SET_DECL_ASSEMBLER_NAME): Use it.
11529 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11530 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11531 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11533 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11535 * config.gcc (i386, x86_64): Add extra objects.
11536 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11537 (ix86_min_insn_size): Declare.
11538 (ix86_issue_rate): Declare.
11539 (ix86_adjust_cost): Declare.
11540 (ia32_multipass_dfa_lookahead): Declare.
11541 (ix86_macro_fusion_p): Declare.
11542 (ix86_macro_fusion_pair_p): Declare.
11543 (ix86_bd_has_dispatch): Declare.
11544 (ix86_bd_do_dispatch): Declare.
11545 (ix86_core2i7_init_hooks): Declare.
11546 (ix86_atom_sched_reorder): Declare.
11547 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11548 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11549 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11550 (rip_relative_addr_p): Rename to ...
11551 (ix86_rip_relative_addr_p): ... this one; export.
11552 (memory_address_length): Update.
11553 (ix86_issue_rate): Move to x86-tune-sched.c.
11554 (ix86_flags_dependent): Move to x86-tune-sched.c.
11555 (ix86_agi_dependent): Move to x86-tune-sched.c.
11556 (exact_dependency_1): Move to x86-tune-sched.c.
11557 (exact_store_load_dependency): Move to x86-tune-sched.c.
11558 (ix86_adjust_cost): Move to x86-tune-sched.c.
11559 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11560 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11561 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11562 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11563 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11564 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11565 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11566 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11567 (min_insn_size): Rename to ...
11568 (ix86_min_insn_size): ... this one; export.
11569 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11570 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11571 (core2i7_first_cycle_multipass_backtrack): Move to
11572 x86-tune-sched-core.c.
11573 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11574 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11575 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11576 (ix86_avoid_jump_mispredicts): Update.
11577 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11578 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11579 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11580 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11581 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11582 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11583 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11584 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11585 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11586 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11587 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11588 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11589 (BIG): Move to ix86-tune-sched-bd.c.
11590 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11591 (enum insn_path): Move to ix86-tune-sched-bd.c.
11592 (get_mem_group): Move to ix86-tune-sched-bd.c.
11593 (is_cmp): Move to ix86-tune-sched-bd.c.
11594 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11595 (is_branch): Move to ix86-tune-sched-bd.c.
11596 (is_prefetch): Move to ix86-tune-sched-bd.c.
11597 (init_window): Move to ix86-tune-sched-bd.c.
11598 (allocate_window): Move to ix86-tune-sched-bd.c.
11599 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11600 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11601 (process_end_window): Move to ix86-tune-sched-bd.c.
11602 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11603 (find_constant): Move to ix86-tune-sched-bd.c.
11604 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11605 (has_immediate): Move to ix86-tune-sched-bd.c.
11606 (get_insn_path): Move to ix86-tune-sched-bd.c.
11607 (get_insn_group): Move to ix86-tune-sched-bd.c.
11608 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11609 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11610 (add_insn_window): Move to ix86-tune-sched-bd.c.
11611 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11612 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11613 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11614 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11615 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11616 (do_dispatch): Move to ix86-tune-sched-bd.c.
11617 (has_dispatch): Move to ix86-tune-sched-bd.c.
11618 * config/i386/t-i386: Add new object files.
11619 * config/i386/x86-tune-costs.h: New file.
11620 * config/i386/x86-tune-sched-atom.c: New file.
11621 * config/i386/x86-tune-sched-bd.c: New file.
11622 * config/i386/x86-tune-sched-core.c: New file.
11623 * config/i386/x86-tune-sched.c: New file.
11625 2017-10-11 Liu Hao <lh_mouse@126.com>
11627 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11628 the generic version below instead.
11629 (should_colorize): Recognize Windows consoles as terminals
11631 * pretty-print.c [__MINGW32__] (write_all): New function.
11632 [__MINGW32__] (find_esc_head): Likewise.
11633 [__MINGW32__] (find_esc_terminator): Likewise.
11634 [__MINGW32__] (eat_esc_sequence): Likewise.
11635 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11637 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11640 2017-10-11 Richard Biener <rguenther@suse.de>
11642 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11643 Properly call analyze_scalar_evolution with the loop of the stmt.
11645 2017-10-11 Richard Biener <rguenther@suse.de>
11647 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11648 * tree-core.h (tree_base): Add chrec_var union member.
11649 * tree.h (CHREC_VAR): Remove.
11650 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11651 * tree-chrec.h (build_polynomial_chrec): Adjust.
11652 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11653 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11655 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11657 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11658 * match.pd: ... here.
11659 ((T) X == (T) Y): Relax condition.
11661 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11663 PR tree-optimization/82472
11664 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11666 (break_alias_scc_partitions): Update postorder number.
11668 2017-10-11 Martin Liska <mliska@suse.cz>
11671 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11673 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11675 2017-10-11 Martin Liska <mliska@suse.cz>
11677 * print-rtl.c (print_insn): Move declaration of idbuf
11678 to same scope as name.
11680 2017-10-11 Martin Liska <mliska@suse.cz>
11685 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11686 volatile arguments.
11688 2017-10-11 Martin Liska <mliska@suse.cz>
11691 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11692 volatile arguments.
11694 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11696 * config.gcc (default_gnu_indirect_function): Default to yes for
11697 arm*-*-linux* with glibc.
11699 2017-10-11 Richard Biener <rguenther@suse.de>
11701 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11702 default-defs and types we do not want to analyze.
11703 (interpret_loop_phi): Replace unreachable code with an assert.
11704 (compute_scalar_evolution_in_loop): Remove and inline ...
11705 (analyze_scalar_evolution_1): ... here, replacing condition with
11706 what makes the intent clearer. Remove handling of cases
11707 get_scalar_evolution now handles.
11709 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11711 PR rtl-optimization/81434
11712 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11713 comment for main loop. In sched_group_found if, also add checks for
11714 pass and min_cost_group.
11716 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11718 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11719 (rs6000_insn_cost): New function.
11720 * config/rs6000/rs6000.md (cost): New attribute.
11722 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11723 H.J. Lu <hongjiu.lu@intel.com>
11727 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11728 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11729 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11730 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11731 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11732 OPTION_MASK_ISA_MMX and some other bit set require both
11733 mmx and the ISAs without the mmx bit.
11734 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11735 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11736 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11737 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11738 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11739 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11740 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11741 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11742 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11743 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11744 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11745 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11746 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11747 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11748 Add OPTION_MASK_ISA_MMX.
11750 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11752 * config.gcc (armv7*-*-freebsd*): New target.
11753 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11755 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11757 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11758 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11761 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11763 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11764 a template parameter.
11765 (WIDE_INT_REF_FOR): Update accordingly.
11766 * tree.h (wi::int_traits <const_tree>): Delete.
11767 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11768 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11769 (wi::tree_to_wide_ref): New typedef.
11770 (wi::to_wide): New function.
11771 * calls.c (get_size_range): Use wi::to_wide when operating on
11772 trees as wide_ints.
11773 * cgraph.c (cgraph_node::create_thunk): Likewise.
11774 * config/i386/i386.c (ix86_data_alignment): Likewise.
11775 (ix86_local_alignment): Likewise.
11776 * dbxout.c (stabstr_O): Likewise.
11777 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11778 * expr.c (const_vector_from_tree): Likewise.
11779 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11780 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11781 (fold_negate_expr_1, int_const_binop_1, const_binop)
11782 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11783 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11784 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11785 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11786 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11787 (fold_not_const, round_up_loc): Likewise.
11788 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11789 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11790 (alloca_call_type): Likewise.
11791 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11792 * godump.c (go_output_typedef): Likewise.
11793 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11794 * internal-fn.c (get_min_precision): Likewise.
11795 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11796 * ipa-polymorphic-call.c
11797 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11798 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11799 (ipa_modify_call_arguments): Likewise.
11800 * match.pd: Likewise.
11801 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11802 * print-tree.c (print_node_brief, print_node): Likewise.
11803 * stmt.c (expand_case): Likewise.
11804 * stor-layout.c (layout_type): Likewise.
11805 * tree-affine.c (tree_to_aff_combination): Likewise.
11806 * tree-cfg.c (group_case_labels_stmt): Likewise.
11807 * tree-data-ref.c (dr_analyze_indices): Likewise.
11808 (prune_runtime_alias_test_list): Likewise.
11809 * tree-dump.c (dequeue_and_dump): Likewise.
11810 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11811 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11812 * tree-pretty-print.c (dump_generic_node): Likewise.
11813 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11814 (simple_iv_with_niters): Likewise.
11815 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11816 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11817 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11818 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11819 (refine_value_range_using_guard, number_of_iterations_ne_max)
11820 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11821 (get_cst_init_from_scev, record_nonwrapping_iv)
11822 (scev_var_range_cant_overflow): Likewise.
11823 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11824 * tree-ssa-pre.c (compute_avail): Likewise.
11825 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11826 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11827 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11828 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11829 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11830 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11831 (dump_case_nodes, try_switch_expansion): Likewise.
11832 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11833 (vect_do_peeling): Likewise.
11834 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11835 * tree-vect-stmts.c (vectorizable_load): Likewise.
11836 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11837 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11838 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11839 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11840 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11841 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11842 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11843 (evrp_dom_walker::before_dom_children): Likewise.
11844 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11845 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11846 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11847 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11848 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11849 (get_range_pos_neg): Likewise.
11850 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11851 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11852 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11853 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11854 * config/avr/avr.c (avr_fold_builtin): Likewise.
11855 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11856 * config/msp430/msp430.c (msp430_attr): Likewise.
11857 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11858 * config/powerpcspe/powerpcspe-c.c
11859 (altivec_resolve_overloaded_builtin): Likewise.
11860 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11861 (rs6000_expand_ternop_builtin): Likewise.
11862 * config/rs6000/rs6000-c.c
11863 (altivec_resolve_overloaded_builtin): Likewise.
11864 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11865 (rs6000_expand_ternop_builtin): Likewise.
11866 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11868 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11870 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11871 when copying loop nest with only one inner loop.
11873 2017-10-10 Richard Biener <rguenther@suse.de>
11875 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11876 blocks if SCEV is active.
11877 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11879 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11880 (scev_initialize): Assert we are not yet initialized.
11882 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11884 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11885 inner loop's exit stmt by making it always exit the loop, otherwise
11886 we would generate an infinite empty loop.
11888 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11890 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11891 renaming variables in new preheader if it's deleted.
11893 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11895 * tree-loop-distribution.c (struct partition): Remove unused field
11896 loops of the structure.
11897 (partition_alloc, partition_free): Ditto.
11898 (build_rdg_partition_for_vertex): Ditto.
11900 2017-10-09 Jeff Law <law@redhat.com>
11902 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11903 return type to match prototype and documentation.
11905 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11907 * config/rs6000/rs6000.c (processor_costs): Move to ...
11908 * config/rs6000/rs6000.h: ... here.
11909 (rs6000_cost): Declare.
11911 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11913 * except.c (setjmp_fn): New global variable.
11914 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11915 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11916 if DONT_USE_BUILTIN_SETJMP is defined.
11918 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11920 * target.def (insn_cost): New hook.
11921 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11922 * doc/tm.texi: Regenerate.
11923 * rtlanal.c (insn_cost): Use the new hook.
11925 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11927 * combine.c (combine_validate_cost): Compute the new insn_cost,
11928 not just pattern_cost.
11929 (try_combine): Adjust comment.
11931 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11933 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11935 * combine.c (uid_insn_cost): Adjust comment.
11936 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11938 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11939 * dse.c (find_shift_sequence): Ditto.
11940 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11941 (bb_valid_for_noce_process_p): Use pattern_cost.
11942 * rtl.h (insn_rtx_cost): Delete.
11943 (pattern_cost): New prototype.
11944 (insn_cost): New prototype.
11945 * rtlanal.c (insn_rtx_cost): Rename to...
11946 (pattern_cost): ... this.
11949 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11951 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11952 (*jcc<mode>_0_r_i387): Ditto.
11953 (*jccxf_r_i387): Ditto.
11954 (*jcc<mode>_r_i387): Ditto.
11955 (*jccu<mode>_r_i387): Ditto.
11956 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11957 (*jcc): Rename from *jcc_1.
11959 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11961 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11962 deferred rescans after the lvx/stvx recombination pre-pass.
11964 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11966 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11967 memory operation instruction support.
11968 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11969 (rs6000-ibm-aix[789]*): Likewise.
11970 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11971 Document new functions.
11973 2017-10-09 Richard Biener <rguenther@suse.de>
11975 PR tree-optimization/82397
11976 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11977 equality only for semantically equal trees.
11979 2017-10-09 Richard Biener <rguenther@suse.de>
11981 PR tree-optimization/82449
11982 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11983 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11984 allow constant addresses.
11985 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11988 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11990 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11993 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11996 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11999 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12002 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12004 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12007 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12008 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12010 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12012 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12013 (WI_BINARY_PREDICATE_RESULT): Likewise.
12014 (wi::binary_traits::operator_result): New type.
12015 (wi::binary_traits::predicate_result): Likewise.
12016 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12017 (generic_wide_int::operator==, generic_wide_int::operator!=)
12018 (generic_wide_int::operator&, generic_wide_int::and_not)
12019 (generic_wide_int::operator|, generic_wide_int::or_not)
12020 (generic_wide_int::operator^, generic_wide_int::operator+
12021 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12023 (operator~, unary operator-, operator==, operator!=, operator&)
12024 (operator|, operator^, operator+, binary operator-, operator*): New
12026 * expr.c (get_inner_reference): Use wi::bit_and_not.
12027 * fold-const.c (fold_binary_loc): Likewise.
12028 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12029 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12030 (bit_value_binop): Likewise.
12031 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12032 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12033 (extract_range_from_binary_expr_1): Likewise.
12034 (masked_increment): Likewise.
12035 (simplify_bit_ops_using_ranges): Likewise.
12037 2017-10-09 Martin Jambor <mjambor@suse.cz>
12040 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12041 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12042 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12044 (hsa_fixup_mov_insn_type): New function.
12045 (hsa_op_with_type::get_in_type): Use it.
12046 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12047 immediates in an assert.
12048 (hsa_op_with_type::extend_int_to_32bit): New method.
12049 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12050 types. Convert to dest type if necessary.
12051 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12052 (reg_for_gimple_ssa): Pass false as min32int to
12053 hsa_type_for_scalar_tree_type.
12054 (gen_hsa_addr): Fixup type when creating addresable temporary.
12055 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12056 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12057 necessary. Call hsa_fixup_mov_insn_type.
12058 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12059 extend operands and convert to dest type if necessary.
12060 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12061 to dest type if necessary.
12062 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12063 if conversion nt necessary and size matches.
12064 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12065 to dest type if necessary.
12066 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12067 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12069 (gen_hsa_clrsb): Likewise.
12070 (gen_hsa_ffs): Likewise.
12071 (gen_hsa_divmod): Extend operands and convert to dest type if
12073 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12075 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12077 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12078 Remove empty default arguments. Use a brace block as output
12080 (conditional return): Ditto.
12082 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12083 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12085 (group_ending_nop): Ditto.
12086 (doloop_end): Ditto.
12087 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12088 (splitters for those): Ditto.
12090 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12092 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12093 a conditional jump (and the compare for it) so that pc_rtx is the
12095 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12096 for the deleted and renamed ctr<mode>_internal[234] patterns.
12097 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12098 Delete second conditional return pattern.
12099 (ctr<mode>_internal2): Delete this second bdnz pattern.
12100 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12101 (ctr<mode>_internal4): Delete this second bdz pattern.
12103 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12105 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12106 (always_initialized_rtx_for_ssa_name_p): New predicate.
12107 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12108 (finish_out_of_ssa): Free new field of SA.
12109 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12110 * tree-ssa-coalesce.c: Include tree-ssa.h.
12111 (get_parm_default_def_partitions): Remove extern keyword.
12112 (get_undefined_value_partitions): New function.
12113 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12114 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12115 uninitialized bits.
12116 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12118 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12120 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12122 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12124 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12125 for targets that preffer 128bit.
12127 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12129 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12131 2017-10-08 Olivier Hainque <hainque@adacore.com>
12133 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12134 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12136 (thumb_set_return_address): Likewise.
12138 2017-10-08 Olivier Hainque <hainque@adacore.com>
12140 * common/config/arm/arm-common.c (arm_except_unwind_info):
12141 Handle DWARF2_UNWIND_INFO.
12143 2017-10-07 Michael Collison <michael.collison@arm.com>
12145 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12148 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12150 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12151 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12152 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12153 defined, force the creation of a new block for a dispatch label.
12155 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12157 * invoke.texi (Wsuggest-attribute=cold): Document.
12158 * common.opt (Wsuggest-attribute=cold): New
12159 * ipa-pure-const.c (warn_function_cold): New function.
12160 * predict.c (compute_function_frequency): Use it.
12161 * predict.h (warn_function_cold): Declare.
12163 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12165 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12168 2017-10-06 Martin Liska <mliska@suse.cz>
12170 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12171 keyword for member functions.
12172 (struct sanopt_tree_couple): New struct.
12173 (struct sanopt_tree_couple_hash): New function.
12174 (struct sanopt_ctx): Add new hash_map.
12175 (has_dominating_ubsan_ptr_check): New function.
12176 (record_ubsan_ptr_check_stmt): Likewise.
12177 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12178 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12179 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12181 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12184 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12185 aarch64_simd_valid_immediate on CONST_VECTORs.
12186 (aarch64_reg_or_bic_imm): Likewise.
12188 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12190 PR rtl-optimization/82396
12191 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12193 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12195 * graphite-dependences.c (scop_get_reads): Move code to...
12196 (scop_get_must_writes): Move code to...
12197 (scop_get_may_writes): Move code to...
12198 (scop_get_reads_and_writes): ... here.
12199 (scop_get_dependences): Call scop_get_reads_and_writes.
12201 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12203 PR tree-optimization/82434
12204 * fold-const.h (can_native_encode_type_p,
12205 can_native_encode_string_p): Remove.
12206 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12207 don't encode anything, just return what would be otherwise returned.
12208 (native_encode_fixed, native_encode_complex, native_encode_vector):
12210 (native_encode_string): Likewise. Inline by hand
12211 can_native_encode_string_p.
12212 (can_native_encode_type_p): Remove.
12213 (can_native_encode_string_p): Remove.
12214 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12215 STRING_CSTs using can_native_encode_string_p, test all
12216 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12217 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12218 argument from native_encode_expr.
12219 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12220 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12223 2017-10-06 Richard Biener <rguenther@suse.de>
12225 PR tree-optimization/82397
12226 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12227 operand_equal_p but rely on data_ref_compare_tree for detecting
12229 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12230 to match up with dr_group_sort_cmp.
12232 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12235 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12237 * config/s390/s390-builtin-types.def: Regenerate.
12239 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12242 * config/s390/s390-builtin-types.def: Regenerate.
12243 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12244 Change flag from B_VXE to B_VX.
12245 (s390_vec_min_dbl): Remove B_VXE flag.
12247 2017-10-06 Richard Biener <rguenther@suse.de>
12249 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12250 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12251 translate_isl_ast_to_gimple::is_valid_rename,
12252 translate_isl_ast_to_gimple::get_rename,
12253 translate_isl_ast_to_gimple::get_def_bb_for_const,
12254 translate_isl_ast_to_gimple::get_new_name,
12255 translate_isl_ast_to_gimple::collect_all_ssa_names,
12256 translate_isl_ast_to_gimple::copy_loop_phi_args,
12257 translate_isl_ast_to_gimple::collect_all_ssa_names,
12258 translate_isl_ast_to_gimple::copy_loop_phi_args,
12259 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12260 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12261 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12262 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12263 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12264 translate_isl_ast_to_gimple::copy_cond_phi_args,
12265 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12266 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12267 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12268 translate_isl_ast_to_gimple::rename_uses,
12269 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12270 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12271 (set_rename_for_each_def): Likewise.
12272 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12273 here. Handle rewriting SCEV analyzable uses here.
12274 (copy_bb_and_scalar_dependences): Generate code for PHI
12276 (graphite_regenerate_ast_isl): Adjust.
12277 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12278 (add_write, add_read): New functions.
12279 (build_cross_bb_scalars_def): Use it and simplify.
12280 (build_cross_bb_scalars_use): Likewise.
12281 (graphite_find_cross_bb_scalar_vars): Inline into...
12282 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12283 simulating out-of-SSA. Compute liveout and add dependencies.
12284 (build_scops): Force an empty entry block.
12285 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12287 (sese_build_liveouts): Declare.
12288 (sese_trivially_empty_bb_p): Likewise.
12289 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12290 compute liveout and debug_liveout.
12291 (sese_bad_liveouts_use): Remove.
12292 (sese_reset_debug_liveouts_bb): Likewise.
12293 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12294 (sese_build_liveouts): Build liveout and debug_liveout and store
12296 (new_sese_info): Adjust.
12297 (free_sese_info): Likewise.
12298 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12299 do not build liveout here.
12300 (move_sese_in_condition): Adjust region entry.
12301 (scev_analyzable_p): Match up with chrec_apply requirements.
12302 (sese_trivially_empty_bb_p): New.
12303 * tree-into-ssa.c (get_reaching_def): Properly support generating
12304 default-defs for incremental rewrite of anonymous names.
12306 2017-10-06 Richard Biener <rguenther@suse.de>
12308 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12309 precision do not perform modulo reduction.
12311 2017-10-06 Richard Biener <rguenther@suse.de>
12313 PR tree-optimization/82436
12314 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12315 conservatively choose the vectorization factor when checking
12316 whether we can perform the required load permutation.
12317 (vect_transform_slp_perm_load): Assert when we may not fail.
12319 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12321 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12322 message for incompatible -msdata=* and -mcall-* options.
12324 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12326 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12327 rate for post-reload scheduling.
12329 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12331 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12333 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12335 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12336 to improve monte carlo in scimark.
12338 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12340 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12341 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12342 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12343 pentium4_cost, nocona_cost): Set reassociation width to 1.
12344 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12345 width to 2 for fp operations and 1 otherwise.
12346 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12348 (atom_cost): Set reassociation width to 2.
12349 (slm_cost, generic_cost): Set fp reassociation width
12350 to 2 and 1 otherwise.
12351 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12352 (core_cost): Set fp reassociation width to 4 and vector to 2.
12353 (ix86_reassociation_width): Rewrite using cost table; special case
12354 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12355 and TARGET_AVX128_OPTIMAL.
12356 * config/i386/i386.h (processor_costs): Add
12357 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12358 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12359 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12360 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12361 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12362 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12364 2017-10-05 Nathan Sidwell <nathan@acm.org>
12366 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12368 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12370 * config/arm/arm.c (arm_test_fpu_data): New.
12371 (arm_run_selftests): Call arm_test_fpu_data.
12373 2017-10-04 Nathan Sidwell <nathan@acm.org>
12375 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12377 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12379 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12381 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12382 check type for aarch64_simd_valid_immediate.
12383 (aarch64_output_simd_mov_immediate): Update prototype.
12384 (aarch64_simd_valid_immediate): Update prototype.
12385 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12386 support for ORR-immediate.
12387 (and<mode>3): modified pattern to add support for BIC-immediate.
12388 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12389 now checks for valid immediate for BIC and ORR based on new enum
12391 (aarch64_output_simd_mov_immediate): Function now used to output
12392 BIC/ORR imm as well based on new enum argument.
12393 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12395 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12396 (aarch64_reg_or_bic_imm): Likewise.
12398 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12400 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12401 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12403 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12407 PR rtl-optimization/82396
12408 * haifa-sched.c (autopref_multipass_init): Simplify
12410 (autopref_rank_data): Simplify sort order.
12411 * sched-int.h (autopref_multipass_data_): Remove
12412 multi_mem_insn_p, min_offset and max_offset.
12414 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12416 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12418 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12420 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12421 vect_doubleint_cvt.
12423 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12425 * doc/sourcebuild.texi: Document vect_long_mult.
12427 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12429 PR tree-optimization/82413
12430 * fold-const.c (build_range_check): Use widest_int when comparing
12431 the maximum ETYPE value with HIGH.
12433 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12435 PR rtl-optimization/82396
12436 * haifa-sched.c (autopref_multipass_init): Simplify
12438 (autopref_rank_data): Simplify sort order.
12439 * sched-int.h (autopref_multipass_data_): Remove
12440 multi_mem_insn_p, min_offset and max_offset.
12442 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12444 PR tree-optimization/82381
12445 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12446 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12449 PR tree-optimization/82374
12450 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12451 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12452 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12453 current_function_decl to the new decl.
12455 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12457 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12458 helper macro for IEEE float128 hardware built-in functions.
12459 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12461 (TRUNCF128_ODD): Likewise.
12462 (ADDF128_ODD): Likewise.
12463 (SUBF128_ODD): Likewise.
12464 (MULF128_ODD): Likewise.
12465 (DIVF128_ODD): Likewise.
12466 (FMAF128_ODD): Likewise.
12467 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12468 UNSPEC_TRUNC_ROUND_TO_ODD.
12469 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12470 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12471 floating point round to odd instructions.
12472 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12473 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12474 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12475 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12476 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12477 (trunc<mode>sf2_hw): Change the truncate with round to odd
12478 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12479 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12480 to odd hardware instructions.
12481 (sub<mode>3_odd): Likewise.
12482 (mul<mode>3_odd): Likewise.
12483 (div<mode>3_odd): Likewise.
12484 (sqrt<mode>2_odd): Likewise.
12485 (fma<mode>4_odd): Likewise.
12486 (fms<mode>4_odd): Likewise.
12487 (nfma<mode>4_odd): Likewise.
12488 (nfms<mode>4_odd): Likewise.
12489 (trunc<mode>df2_odd): Change the truncate with round to odd
12490 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12492 * doc/extend.texi (PowerPC built-in functions): Update documentation
12493 for existing IEEE float128-bit built-in functions. Add built-in
12494 functions that generate the IEEE 128-bit floating point round to
12497 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12499 PR rtl-optimization/77729
12500 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12501 to (X&(C1&~C2))|C2 transformations.
12503 2017-10-03 Martin Jambor <mjambor@suse.cz>
12505 PR tree-optimization/82363
12506 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12507 mismatch, mark lacc written regardless of racc.
12509 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12511 PR tree-optimization/82381
12512 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12513 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12514 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12515 fallthrough into reassoc_stmt_dominates_stmt_p.
12518 * combine.c (combine_instructions): Don't combine in unreachable
12521 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12524 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12525 function to not use the have_cpu variable. Do not set cpu_index,
12526 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12527 or the default cpu.
12528 (rs6000_valid_attribute_p): Remove duplicate initializations of
12529 old_optimize and func_optimize.
12530 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12531 (rs6000_activate_target_options): Make global.
12532 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12535 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12537 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12538 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12539 Set *poffset to 0 and *psize and *pmax_size to -1 if
12540 *poffset + *psize overflows in HOST_WIDE_INT.
12542 PR tree-optimization/82387
12543 PR tree-optimization/82388
12544 PR tree-optimization/82389
12545 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12546 instead of live_bytes non-NULL.
12548 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12551 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12554 2017-10-02 Richard Biener <rguenther@suse.de>
12556 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12557 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12558 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12560 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12562 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12563 requested precision matches the type's.
12564 * calls.c (alloc_max_size): Calculate the new candidate size as
12565 a widest_int and use wi::to_widest when comparing it with the
12566 current candidate size.
12567 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12568 zero rather than integer_zero_node.
12569 * match.pd: Check for a no-op conversion before using wi::add
12570 rather than after. Use tree_to_uhwi when summing small shift
12571 counts into an unsigned int.
12573 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12574 Alan Hayward <alan.hayward@arm.com>
12575 David Sherwood <david.sherwood@arm.com>
12578 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12579 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12580 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12581 POINTER_AND_FP_REGS.
12583 2017-10-02 Richard Biener <rguenther@suse.de>
12585 PR tree-optimization/82355
12586 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12587 a mapping for the enclosing loop but avoid generating one for
12588 the loop tree root.
12589 (copy_bb_and_scalar_dependences): Remove premature codegen
12590 error on PHIs in blocks duplicated into multiple places.
12591 * graphite-scop-detection.c
12592 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12593 in the region use it as loop and nest to analyze the DR in.
12594 (try_generate_gimple_bb): Likewise.
12595 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12596 (add_loop_constraints): For blocks in a loop not in the region
12597 create a dimension with a single iteration.
12598 * sese.h (gbb_loop_at_index): Remove assert.
12600 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12602 * omp-expand.c (adjust_context_scope): New function.
12603 (expand_parallel_call): Call adjust_context_scope.
12605 2017-10-01 Jeff Law <law@redhat.com>
12607 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12608 dom_opt_dom_walker class with direct access to private members.
12609 Add comments. Call test_for_singularity.
12610 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12611 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12612 m_dummy_cond anymore.
12613 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12615 (pass_dominator:execute): Build the dummy_cond here and pass it
12616 to the dom_opt_dom_walker ctor.
12617 (test_for_singularity): New function.
12619 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12620 Maya Rashish <coypu@sdf.org>
12622 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12623 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12624 (alpha*-*-netbsd*) Use nbsd_tm_file.
12625 (arm*-*-netbsdelf*) Likewise.
12626 (i[34567]86-*-netbsdelf*) Likewise.
12627 (x86_64-*-netbsd*) Likewise.
12628 (mips*-*-netbsd*) Likewise.
12629 (powerpc-*-netbsd*) Likewise.
12630 (sh*-*-netbsd*) Likewise.
12631 (sparc-*-netbsdelf*) Likewise.
12632 (sparc64-*-netbsd*) Likewise.
12633 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12635 (vax-*-netbsdelf*) Likewise.
12636 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12637 (UINT_FAST8_TYPE) Likewise.
12638 (INT_FAST16_TYPE) Check CHAR_FAST16.
12639 (UINT_FAST16_TYPE) Likewise.
12641 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12644 * config/i386/i386.md
12645 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12646 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12647 *divmodsi4_zext_2): New define_insn_and_split.
12648 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12649 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12650 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12651 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12652 New define_insn_and_split.
12653 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12654 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12655 operands[1] having DImode when mode is SImode.
12657 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12658 always SImode for DIV and MOD in REG_EQUAL notes.
12660 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12662 PR middle-end/82319
12663 * match.pd: Fix handling of NaNs in pattern.
12665 2017-09-29 Jeff Law <law@redhat.com>
12667 * sbitmap.c (bitmap_bit_in_range_p): New function.
12668 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12669 * tree-ssa-dse.c (live_bytes_read): New function.
12670 (dse_classify_store): Ignore reads of dead bytes.
12672 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12673 typos and whitespace errors.
12674 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12675 * config/s390/s390.c (s390_emit_prologue): Likewise.
12677 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12680 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12681 with a symbol for LRA.
12683 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12685 PR rtl-optimization/82338
12686 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12688 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12690 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12691 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12692 (qsort_chk): Declare.
12693 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12694 (qsort_chk): New function.
12696 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12698 PR tree-optimization/82337
12699 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12700 phi definition if the PHI result appears in an abnormal PHI.
12701 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12702 basis appears in an abnormal PHI.
12704 2017-09-29 Richard Biener <rguenther@suse.de>
12706 * graphite-isl-ast-to-gimple.c
12707 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12708 (binary_op_to_tree): Use it.
12709 (get_rename_from_scev): Likewise.
12710 (copy_loop_phi_nodes): Likewise.
12711 (copy_bb_and_scalar_dependences): Likewise.
12712 (translate_pending_phi_nodes): Likewise.
12714 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12717 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12718 for movabsq $(i32 << shift), r64.
12720 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12722 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12723 index when encoding %esp as %rsp to avoid 0x67 prefix.
12725 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12727 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12728 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12730 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12732 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12733 Extensions with more than 16 double VFP registers.
12734 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12735 to_clear_mask and all code related to it. Replace the remaining
12736 entry by a sbitmap and adapt code accordingly.
12738 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12740 * brig-builtins.def: Change pure attributes to const.
12742 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12744 * config.gcc (default_gnu_indirect_function): Default to yes for
12745 sparc*-*-linux* with glibc.
12747 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12749 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12750 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12751 when creating .init_array and .fini_array sections with priority
12754 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12757 * config/aarch64/aarch64.c
12758 (aarch64_builtin_support_vector_misalignment): Always return false
12759 when misalignment is unknown.
12761 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12763 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12764 this function to return false if the definition used by the swap
12765 instruction is artificial, or if the memory address from which the
12766 constant value is loaded is not represented by a base address held
12767 in a register or if the base address register is a frame or stack
12768 pointer. Additionally, return false if the base address of the
12769 loaded constant is a SYMBOL_REF but is not considered to be a
12771 (replace_swapped_load_constant): New function.
12772 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12773 loaded constant vector with a load of a swapped constant vector.
12775 2017-09-27 Carl Love <cel@us.ibm.com>
12777 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12778 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12779 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12782 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12784 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12785 first, always call autopref_rank_data otherwise.
12787 2017-09-27 Richard Biener <rguenther@suse.de>
12789 * graphite-scop-detection.c (find_scop_parameters): Move
12790 loop bound handling ...
12791 (gather_bbs::before_dom_children): ... here, avoiding the need
12792 to build scop_info->loop_nest.
12793 (record_loop_in_sese): Remove.
12794 * sese.h (sese_info_t::loop_nest): Remove.
12795 * sese.c (new_sese_info): Do not allocate loop_nest.
12796 (free_sese_info): Do not free loop_nest.
12798 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12801 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12802 lhs from calls if the lhs has addressable type.
12804 2017-09-27 Richard Biener <rguenther@suse.de>
12806 * graphite.h (scop::max_alias_set): New member.
12807 * graphite-scop-detection.c: Remove references to non-existing
12808 --param in comments.
12809 (build_alias_sets): Record the maximum alias set used for drs.
12810 (build_scops): Support zero as unlimited for
12811 --param graphite-max-arrays-per-scop.
12812 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12813 and inline into ...
12814 (build_poly_sr_1): ... here. Compute alias set based on the
12815 maximum alias set used for drs rather than
12816 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12818 2017-09-27 Richard Biener <rguenther@suse.de>
12820 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12821 --param loop-block-tile-size=0 to disable tiling.
12823 2017-09-27 Richard Biener <rguenther@suse.de>
12825 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12826 (graphite-max-nb-scop-params): Document special value zero.
12827 * domwalk.h (dom_walker::STOP): New symbolical constant.
12828 (dom_walker::dom_walker): Add optional parameter for bb to
12830 (dom_walker::~dom_walker): Declare.
12831 (dom_walker::before_dom_children): Document STOP return value.
12832 (dom_walker::m_user_bb_to_rpo): New member.
12833 (dom_walker::m_bb_to_rpo): Likewise.
12834 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12835 mapping here if not provided by the user.
12836 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12837 provided by the user.
12838 (dom_walker::STOP): Define.
12839 (dom_walker::walk): Do not compute bb to RPO mapping here.
12840 Support STOP return value from before_dom_children to stop
12842 * graphite-optimize-isl.c (optimize_isl): If the schedule
12843 is the same still generate code if -fgraphite-identity
12844 or -floop-parallelize-all are given.
12845 * graphite-scop-detection.c: Include cfganal.h.
12846 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12848 (gather_bbs::before_dom_children): Return STOP for BBs
12850 (build_scops): Compute bb to RPO mapping and pass it to
12851 the domwalk. Treat --param graphite-max-nb-scop-params=0
12852 as not limiting the number of params.
12853 * graphite.c (graphite_initialize): Remove limit on the
12854 number of basic-blocks in a function.
12855 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12856 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12857 default value of 10.
12859 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12861 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12862 Adjust code to eliminate needing to do the shift right 32-bits
12863 operation after XSCVDPSPN.
12865 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12867 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12868 ((X / Y) != 0 -> X >= Y): Likewise.
12870 2017-09-26 Carl Love <cel@us.ibm.com>
12872 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12873 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12874 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12875 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12876 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12877 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12878 definitions and overloading.
12879 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12880 statement for P9V_BUILTIN_XST_LEN_R.
12881 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12882 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12883 define_expand and define_insn for the instructions and builtins.
12884 * doc/extend.texi: Update the built-in documentation file for the new
12885 built-in functions.
12886 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12887 define_insn for the instructions
12889 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12892 * gcc/config/netbsd-protos.h: New file.
12893 * gcc/config/netbsd.c: New file.
12894 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12895 * gcc/config/t-netbsd: New file.
12896 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12897 (tmake_file) Add t-netbsd.
12898 (extra_objs) Add netbsd.o.
12900 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12904 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12906 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12908 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12909 sign extension from a vector register to a GPR by doing a 32-bit
12910 direct move and then an EXTSW.
12911 (extendsi<mode>2 splitter): Likewise.
12912 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12913 right or vector extract after doing XSCVDPSPN. Use
12914 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12916 (movdi_from_sf_zero_ext): Likewise.
12917 (reload_gpr_from_vsxsf): Likewise.
12918 (p8_mfvsrd_4_disf): Delete, no longer used.
12919 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12920 then needing to move the SFmode to a GPR to use the XSCVDPSP
12921 instruction instead of FRSP and XSCVDPSPN.
12922 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12923 it is adjacent to the other XSCVSPDP insns.
12924 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12925 SFmode to be in traditional Altivec registers.
12926 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12927 (vsx_xscvspdpn): Likewise.
12928 (vsx_xscvspdpn_scalar): Likewise.
12930 2017-09-26 Martin Jambor <mjambor@suse.cz>
12932 * tree-sra.c (compare_access_positions): Put integral types first,
12933 stabilize sorting of integral types, remove conditions putting
12934 non-full-precision integers last.
12935 (sort_and_splice_var_accesses): Disable scalarization if a
12936 non-integert would be represented by a non-full-precision integer.
12938 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12940 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12941 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12942 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12943 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12944 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12945 conditionals inside the function instead of around it. Call
12946 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12947 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12949 2017-09-26 Richard Biener <rguenther@suse.de>
12951 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12953 (scop_detection::build_scop_breadth): ... this. Removed.
12954 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12955 (scop_detection::harmful_stmt_in_bb): Likewise.
12956 (scop_detection::graphite_can_represent_stmt): Likewise.
12957 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12958 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12959 (scop_detection::can_represent_loop_1): ... this. Removed.
12960 (scop_detection::harmful_loop_in_region): Simplify after inlining
12961 the above and remove more quadraticness.
12962 (build_scops): Adjust.
12963 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12966 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12969 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12970 REGNO (base) == SP_REG if base is a REG.
12972 PR middle-end/35691
12973 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12974 if it is different SSA_NAME.
12975 (optimize_range_tests_cmp_bitwise): New function.
12976 (optimize_range_tests): Call it.
12978 2017-09-26 Richard Biener <rguenther@suse.de>
12980 PR tree-optimization/82321
12981 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12982 for the def being inside the loop.
12984 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12986 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12988 * config/s390/s390-builtins.def: Fix constraint on op4.
12990 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12992 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12993 independent expanders.
12994 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12995 ("vec_ordered", "vec_unordered"): New expanders.
12997 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12999 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13002 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13004 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13005 vec_unpacks_lo_v16qi.
13006 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13008 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13010 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13011 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13012 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13014 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13016 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13018 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13020 ("*vec_slb<mode>"): New insn pattern.
13021 ("vec_shr_<mode>"): New expander.
13022 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13023 and force the shift count operand to V16QImode.
13024 ("vec_srb<mode>"): Set shift count mode to V16QI.
13026 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13028 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13029 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13030 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13032 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13035 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13037 2017-09-26 Richard Biener <rguenther@suse.de>
13039 PR tree-optimization/82320
13040 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13043 2017-09-25 Jeff Law <law@redhat.com>
13045 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13046 prototype for new argument.
13047 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13048 mostly extracted from rs6000_emit_allocate_stack.
13049 (rs6000_emit_probe_stack_range_stack_clash): New function.
13050 (rs6000_emit_allocate_stack): Call
13051 rs6000_emit_probe_stack_range_stack_clash as needed.
13052 (rs6000_emit_probe_stack_range): Add additional argument
13053 to call to gen_probe_stack_range{si,di}.
13054 (output_probe_stack_range): New.
13055 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13056 (output_probe_stack_range_stack_clash): New.
13057 (rs6000_emit_prologue): Emit notes into dump file as requested.
13058 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13059 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13060 Add additional operand and pass it to output_probe_stack_range.
13062 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13064 PR tree-optimization/82163
13065 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13066 (checking_verify_loop_closed_ssa): New parameter.
13067 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13068 (check_loop_closed_ssa_stmt): Delete.
13069 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13070 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13071 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13074 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13076 * brig-builtins.def: Treat HSAIL barrier builtins as
13077 setjmp/longjump style functions.
13079 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13081 * target.def (constant_alignment): New hook.
13082 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13083 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13084 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13085 * doc/tm.texi: Regenerate.
13086 * targhooks.h (default_constant_alignment): Declare.
13087 (constant_alignment_word_strings): Likewise.
13088 * targhooks.c (default_constant_alignment): New function.
13089 (constant_alignment_word_strings): Likewise.
13090 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13091 instead of CONSTANT_ALIGNMENT.
13092 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13093 (force_const_mem): Likewise.
13094 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13095 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13096 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13097 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13098 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13100 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13101 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13102 constant_alignment_word_strings.
13103 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13104 (CONSTANT_ALIGNMENT): Likewise.
13105 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13106 (arm_constant_alignment): New function.
13107 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13108 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13109 constant_alignment_word_strings.
13110 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13111 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13112 constant_alignment_word_strings.
13113 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13114 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13115 (cris_constant_alignment): New function.
13116 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13117 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13118 (epiphany_constant_alignment): New function.
13119 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13120 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13121 constant_alignment_word_strings.
13122 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13123 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13124 constant_alignment_word_strings.
13125 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13126 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13127 constant_alignment_word_strings.
13128 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13129 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13130 * config/i386/i386.c (ix86_constant_alignment): Make static.
13131 Use the same interface as the target hook.
13132 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13133 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13134 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13135 constant_alignment_word_strings.
13136 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13137 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13138 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13139 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13140 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13141 constant_alignment_word_strings.
13142 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13143 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13144 constant_alignment_word_strings.
13145 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13146 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13147 constant_alignment_word_strings.
13148 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13149 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13151 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13152 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13153 * config/mips/mips.c (mips_constant_alignment): New function.
13154 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13155 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13156 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13157 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13158 (mmix_constant_alignment): Make static. Use the same interface
13159 as the target hook.
13160 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13161 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13162 constant_alignment_word_strings.
13163 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13164 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13165 constant_alignment_word_strings.
13166 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13167 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13168 constant_alignment_word_strings.
13169 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13170 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13171 (rs6000_constant_alignment): New function.
13172 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13173 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13174 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13175 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13176 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13177 (rs6000_constant_alignment): New function.
13178 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13179 * config/s390/s390.c (s390_constant_alignment): New function.
13180 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13181 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13182 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13183 constant_alignment_word_strings.
13184 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13185 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13186 (sparc_constant_alignment): New function.
13187 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13188 * config/spu/spu.c (spu_constant_alignment): New function.
13189 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13190 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13191 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13192 constant_alignment_word_strings.
13193 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13194 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13195 constant_alignment_word_strings.
13196 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13197 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13198 constant_alignment_word_strings.
13199 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13200 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13201 (visium_constant_alignment): New function.
13202 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13203 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13204 (xtensa_constant_alignment): New function.
13205 * system.h (CONSTANT_ALIGNMENT): Poison.
13207 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13209 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13210 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13211 (rs6000_builtin_valid_without_lhs): New helper function.
13212 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13213 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13215 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13217 * target.h (vec_perm_indices): Use unsigned short rather than
13219 (auto_vec_perm_indices): Likewise.
13220 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13221 Use unsigned int rather than unsigned char.
13222 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13224 2017-09-25 Richard Biener <rguenther@suse.de>
13226 * cfgloop.h (sort_sibling_loops): Declare.
13227 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13228 (sort_sibling_loops): New function sorting the sibling loop list
13230 * graphite.c (graphite_transform_loops): Sort sibling loops.
13232 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13234 * target.def (vec_perm_const_ok): Change sel parameter to
13236 * optabs-query.c (can_vec_perm_p): Update accordingly.
13237 * doc/tm.texi: Regenerate.
13238 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13239 auto_vec_perm_indices and remove separate nelt field.
13240 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13241 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13242 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13243 (aarch64_expand_vec_perm_const): Update accordingly.
13244 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13245 to vec_perm_indices.
13246 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13247 auto_vec_perm_indices and remove separate nelt field.
13248 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13249 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13250 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13252 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13253 to vec_perm_indices.
13254 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13255 sel to vec_perm_indices.
13256 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13257 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13258 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13260 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13263 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13266 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13267 on the FUNCTION_DECL function context if it has a DIE that is a
13270 2017-09-25 Richard Biener <rguenther@suse.de>
13272 PR tree-optimization/82285
13273 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13276 2017-09-25 Tom de Vries <tom@codesourcery.com>
13280 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13283 2017-09-25 Richard Biener <rguenther@suse.de>
13285 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13286 ISL errors other than isl_error_quota happen. Dump if the
13287 schedule is the same.
13288 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13289 errors instead of aborting inside ISL.
13291 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13294 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13295 of libgcc_eh for m64.
13296 * config/i386/darwin64.h: Likewise.
13298 2017-09-25 Richard Biener <rguenther@suse.de>
13300 PR middle-end/82144
13301 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13302 attribute for incomplete types nor twice for complete ones.
13304 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13307 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13308 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13311 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13314 * config/i386/i386.opt (mprefer-avx256): Use
13315 ix86_target_flags variable.
13316 * config/i386/i386.c (ix86_target_string): Move
13317 -mprefer-avx256 to flag2_opts.
13319 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13321 PR middle-end/35691
13322 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13323 and x != -1 | y != -1 into (x & y) != -1.
13325 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13327 * config.gcc: Add new case statement to set
13328 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13329 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13330 s390x-*-linux* case statements. Added aarch64 to the list of
13331 supported architectures.
13333 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13335 PR tree-optimization/82289
13336 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13337 STMT_VINFO_RELEVANT_P.
13339 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13340 Alan Hayward <alan.hayward@arm.com>
13341 David Sherwood <david.sherwood@arm.com>
13343 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13344 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13345 (extract_range_from_binary_expr_1): Don't call
13346 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13348 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13349 Alan Hayward <alan.hayward@arm.com>
13350 David Sherwood <david.sherwood@arm.com>
13352 * target.def (preferred_vector_alignment): New hook.
13353 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13355 * doc/tm.texi: Regenerate.
13356 * targhooks.h (default_preferred_vector_alignment): Declare.
13357 * targhooks.c (default_preferred_vector_alignment): New function.
13358 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13360 (DR_TARGET_ALIGNMENT): New macro.
13361 (aligned_access_p): Update commentary.
13362 (vect_known_alignment_in_bytes): New function.
13363 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13365 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13366 Calculate the misalignment based on the target alignment rather than
13368 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13369 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13370 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13371 the target alignment, rather than masking the element misalignment
13372 with the number of elements in a vector. Also use the target
13373 alignment when calculating the maximum number of peels.
13374 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13375 instead of TYPE_ALIGN_UNIT.
13376 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13377 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13378 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13379 (vect_create_data_ref_ptr): Likewise.
13380 (vect_setup_realignment): Realign by ANDing with
13381 -DR_TARGET_MISALIGNMENT.
13382 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13383 the number of peels based on DR_TARGET_ALIGNMENT.
13384 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13385 with the guaranteed alignment boundary when deciding whether
13387 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13388 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13389 (ensure_base_align): Remove stmt_info parameter. Get the
13390 target base alignment from DR_TARGET_ALIGNMENT.
13391 (vectorizable_store): Update call accordingly. Interpret
13392 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13394 (vectorizable_load): Likewise.
13396 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13397 Alan Hayward <alan.hayward@arm.com>
13398 David Sherwood <david.sherwood@arm.com>
13400 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13401 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13402 (vect_enhance_data_refs_alignment): Likewise.
13404 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13406 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13407 error. Only quit immediately if parsing is complete.
13408 (BEGIN): Initialize fatal_err and parse_done.
13409 (begin fpu, end fpu): Check number of arguments.
13410 (begin arch, end arch): Likewise.
13411 (begin cpu, end cpu): Likewise.
13412 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13413 (optalias): Likewise.
13415 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13417 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13418 * config/arm/arm-isa.h: Delete. Move definitions to ...
13419 * arm-cpus.in: ... here. Use new feature and fgroup values.
13420 * config/arm/arm.c (arm_option_override): Use lower case for feature
13422 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13423 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13424 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13426 (print_isa_bits_for): New function.
13427 (gen_isa): New function.
13428 (gen_comm_data): Use print_isa_bits_for.
13429 (define feature): New keyword.
13430 (define fgroup): New keyword.
13431 * config/arm/t-arm (TM_H): Remove.
13432 (GTM_H): Add arm-isa.h.
13433 (arm-isa.h): Add rule to generate file.
13434 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13435 case for feature bit names.
13437 2017-09-22 Richard Biener <rguenther@suse.de>
13439 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13441 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13442 print of no dependency loops ...
13443 * graphite.c (graphite_transform_loops): ... here.
13444 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13446 (same_close_phi_node, remove_duplicate_close_phi,
13447 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13448 (canonicalize_loop_closed_ssa): ... here and simplify.
13449 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13450 (optimize_isl): Use dump_printf_loc to tell when we stopped
13451 optimizing because of an ISL timeout.
13453 2017-09-22 Richard Biener <rguenther@suse.de>
13455 PR tree-optimization/82291
13456 * tree-if-conv.c (predicate_mem_writes): Make sure to
13457 remove writes in blocks predicated with false.
13459 2017-09-22 Richard Biener <rguenther@suse.de>
13461 * sese.c: Include cfganal.h.
13462 (if_region_set_false_region): Remove.
13463 (create_if_region_on_edge): Likewise.
13464 (move_sese_in_condition): Re-implement without destroying
13467 2017-09-22 Richard Biener <rguenther@suse.de>
13469 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13470 Verify both BBs contain loop PHI nodes before dispatching to
13471 copy_loop_phi_args.
13472 (graphite_regenerate_ast_isl): Do not recompute dominators,
13473 do not verify three times. Restructure for clarity.
13474 * graphite-scop-detection.c (same_close_phi_node,
13475 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13476 defined_in_loop_p, canonicalize_loop_closed_ssa,
13477 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13478 checking and SSA rewrite, move to ...
13479 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13480 (graphite_initialize): Do not pass in ctx, do not reset the
13481 SCEV cache, compute only dominators.
13482 (graphite_transform_loops): Allocate ISL ctx after
13483 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13484 Maintain post-dominators only around build_scops.
13485 * sese.c (if_region_set_false_region): Make static. Free
13486 and recompute dominators.
13487 (move_sese_in_condition): Assert we don't get called with
13488 post-dominators computed.
13489 * sese.h (if_region_set_false_region): Remove.
13491 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13493 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13494 mode attribute for TARGET_AVX512VL.
13496 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13498 * config/i386/i386.opt (mprefer-avx256): New option.
13499 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13501 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13502 for TARGET_PREFER_AVX256.
13503 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13505 2017-09-21 Jeff Law <law@redhat.com>
13507 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13508 Fix dump output if the only stack space is for pushed registers.
13510 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13512 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13515 2017-09-21 Martin Sebor <msebor@redhat.com>
13518 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13519 code (in C++) or code that triggers warnings.
13521 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13523 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13525 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13527 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13528 * sched-rgn.c: Ditto.
13529 * sel-sched-ir.c: Ditto.
13531 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13533 * toplev.h (set_random_seed): Adjust return type.
13534 * toplev.c (init_local_tick): Move eager initialization of random_seed
13535 to get_random_seed. Adjust comment.
13536 (init_random_seed): Inline to get_random_seed, delete.
13537 (get_random_seed): Initialize random_seed lazily.
13538 (set_random_seed): Do not return previous value.
13539 (print_switch_value): Do not call get_random_seed.
13541 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13543 * cgraph.c (delete_function_version): New, broken out from...
13544 (cgraph_node::delete_function_version): ...here. Rename to
13545 cgraph_node::delete_function_version_by_decl. Update all uses.
13546 (cgraph_node::remove): Call delete_function_version.
13548 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13551 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13552 VAR_DECLs to which addressable non-volatile parameters are mapped
13553 and for id->retvar after the return value assignment. Clear
13554 id->retval and id->retbnd after inlining.
13556 2017-09-21 Richard Biener <rguenther@suse.de>
13558 PR tree-optimization/82276
13559 PR tree-optimization/82244
13560 * tree-vrp.c (build_assert_expr_for): Set
13561 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13563 (remove_range_assertions): Revert earlier change.
13565 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13568 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13570 2017-09-21 Richard Biener <rguenther@suse.de>
13572 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13573 Restore valid IL after code generation errors.
13574 * graphite.c (graphite_transform_loops): Diagnose code
13575 generation issues as MSG_MISSED_OPTIMIZATION and continue
13576 with processing SCOPs.
13578 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13579 Alan Hayward <alan.hayward@arm.com>
13580 David Sherwood <david.sherwood@arm.com>
13582 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13583 rather than choosing between plus_constant and gen_rtx_<CODE>.
13584 * expr.c (emit_push_insn): Likewise.
13585 (expand_expr_real_2): Likewise.
13587 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13588 Alan Hayward <alan.hayward@arm.com>
13589 David Sherwood <david.sherwood@arm.com>
13591 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13593 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13594 Alan Hayward <alan.hayward@arm.com>
13595 David Sherwood <david.sherwood@arm.com>
13597 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13598 calling tree_to_uhwi.
13600 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13601 Alan Hayward <alan.hayward@arm.com>
13602 David Sherwood <david.sherwood@arm.com>
13604 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13605 INTEGER_CST rather than a negative test for ADDR_EXPR.
13607 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13608 Alan Hayward <alan.hayward@arm.com>
13609 David Sherwood <david.sherwood@arm.com>
13611 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13612 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13614 2017-09-21 Richard Biener <rguenther@suse.de>
13616 PR tree-optimization/71351
13617 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13618 graphite_create_new_loop_guard): Remove, fold remaining parts
13620 (translate_isl_ast_node_for): ... here and simplify.
13622 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13625 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13626 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13627 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13628 alternative always use QI mode, for -Os imov (=R,R) alternative
13629 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13632 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13633 Jeff Law <law@redhat.com>
13635 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13636 (allocate_stack_space): New function, partially extracted from
13637 s390_emit_prologue.
13638 (s390_emit_prologue): Track offset to most recent stack probe.
13639 Code to allocate space moved into allocate_stack_space.
13640 Dump actions when no stack is allocated.
13641 (s390_prologue_plus_offset): New function.
13642 (s390_emit_stack_probe): Likewise.
13644 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13646 * common.opt (Wa, Wl, Wp, g, gz=): Add
13648 (gno-column-info): Remove.
13649 (gcolumn-info): Drop RejectNegative.
13650 (gno-): New prefix.
13651 (gno-record-gcc-switches): Remove.
13652 (grecord-gcc-switches): Drop RejectNegative.
13653 (gno-split-dwarf): Remove.
13654 (gsplit-dwarf): Drop RejectNegative.
13655 (gno-strict-dwarf): Remove.
13656 (gstrict-dwarf): Drop RejectNegative.
13657 * config/darwin.opt (gfull, gused): Add RejectNegative.
13658 * dwarf2out.c (gen_producer_string): Drop
13659 gno-record-gcc-switches handler.
13660 * optc-gen.awk: Add g to prefixes with negative forms.
13661 * opts-common.c (remapping_prefix_p): New.
13662 (find_opt): Check it.
13663 (generate_canonical_option): Test g prefix.
13664 (option_map): Add -gno- mapping.
13665 (add_misspelling_candidates): Check remapping_prefix_p.
13667 2017-09-20 Jeff Law <law@redhat.com>
13669 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13670 thinko in stack clash protection support.
13672 * explow.c (compute_stack_clash_protection_loop_data): Use
13673 CONST_INT_P instead of explicit test. Verify object is a
13674 CONST_INT_P before looking at INTVAL.
13675 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13676 instead of explicit test.
13678 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13681 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13682 address instead of to r1 and r11.
13684 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13686 * config.gcc: Support "knm".
13687 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13688 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13690 * config/i386/i386.c (m_KNM): Define.
13691 (processor_target_table): Add "knm".
13693 (ix86_option_override_internal): Add "knm".
13694 (ix86_issue_rate): Add PROCESSOR_KNM.
13695 (ix86_adjust_cost): Ditto.
13696 (ia32_multipass_dfa_lookahead): Ditto.
13697 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13698 (fold_builtin_cpu): Add M_INTEL_KNM.
13699 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13700 (processor_type): Add PROCESSOR_KNM.
13701 * config/i386/x86-tune.def: Add m_KNM.
13702 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13704 2017-09-20 Richard Biener <rguenther@suse.de>
13706 PR tree-optimization/80213
13707 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13708 are allowed in empty BBs as well.
13709 (canonicalize_loop_closed_ssa): Also look for other complex
13711 (scop_detection::get_sese): Include the loop-closed PHI block
13713 (scop_detection::merge_sese): Remove code adding extra blocks.
13714 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13715 (build_scops): Assert the final returned scop is invalid.
13717 2017-09-20 Richard Biener <rguenther@suse.de>
13719 PR tree-optimization/82264
13720 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13722 (vn_phi_lookup): Likewise.
13723 (vn_phi_insert): Likewise.
13725 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13727 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13728 that fits into uhwi or shwi, add DW_AT_const_value regardless
13729 of early_dwarf without going through RTL, using add_AT_unsigned
13732 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13733 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13734 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13735 (DEBUG_MACRO_SECTION): Likewise.
13736 (DEBUG_LTO_MACRO_SECTION): Likewise.
13737 (DEBUG_STR_DWO_SECTION): Likewise.
13738 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13739 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13740 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13741 (DEBUG_LTO_LINE_STR_SECTION): Define.
13742 (init_sections_and_labels): Initialize debug_line_str_section
13743 variable. Initialize debug_loc_section for -gdwarf-5 to
13744 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13746 2017-09-20 Richard Biener <rguenther@suse.de>
13748 * graphite-sese-to-poly.c (extract_affine): Properly handle
13749 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13751 2017-09-20 Richard Biener <rguenther@suse.de>
13753 PR tree-optimization/81373
13754 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13755 Force SESE live-out defs to be handled even if they are
13758 2017-09-19 Jeff Law <law@redhat.com>
13760 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13761 nothing for stack adjustments with REG_STACK_CHECK.
13762 * sched-deps.c (parse_add_or_inc): Reject insns with
13763 REG_STACK_CHECK from dependency breaking.
13764 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13765 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13766 * reg-notes.def (STACK_CHECK): New note.
13768 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13769 (ix86_expand_prologue): Dump stack clash info as needed.
13770 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13772 * function.c (dump_stack_clash_frame_info): New function.
13773 * function.h (dump_stack_clash_frame_info): Prototype.
13774 (enum stack_clash_probes): New enum.
13776 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13777 flag_stack_clash_protection.
13778 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13779 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13780 (arm_frame_pointer_required): Likewise.
13781 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13782 (ia64_expand_prologue): Likewise.
13783 * config/mips/mips.c (mips_expand_prologue): Likewise.
13784 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13785 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13786 (sparc_flat_expand_prologue): Likewise.
13787 * config/spu/spu.c (spu_expand_prologue): Likewise.
13789 * explow.c: Include "params.h".
13790 (anti_adjust_stack_and_probe_stack_clash): New function.
13791 (get_stack_check_protect): Likewise.
13792 (compute_stack_clash_protection_loop_data): Likewise.
13793 (emit_stack_clash_protection_loop_start): Likewise.
13794 (emit_stack_clash_protection_loop_end): Likewise.
13795 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13796 Use anti_adjust_stack_and_probe_stack_clash.
13797 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13798 (emit_stack_clash_protection_loop_start): Likewise.
13799 (emit_stack_clash_protection_loop_end): Likewise.
13800 * rtl.h (get_stack_check_protect): Prototype.
13801 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13802 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13803 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13805 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13807 * doc/tm.texi: Rebuilt.
13808 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13809 get_stack_check_protect.
13810 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13811 * config/arm/arm.c (arm_expand_prologue): Likewise.
13812 (arm_frame_pointer_required): Likewise.
13813 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13814 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13815 * config/mips/mips.c (mips_expand_prologue): Likewise.
13816 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13817 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13818 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13819 (sparc_flat_expand_prologue): Likewise.
13821 * common.opt (-fstack-clash-protection): New option.
13822 * flag-types.h (enum stack_check_type): Note difference between
13823 -fstack-check= and -fstack-clash-protection.
13824 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13825 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13826 * toplev.c (process_options): Issue warnings/errors for cases
13827 not handled with -fstack-clash-protection.
13828 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13829 (-fstack-check): Note additional problem with -fstack-check=generic.
13830 Note that -fstack-check is primarily for Ada and refer users
13831 to -fstack-clash-protection for stack-clash-protection.
13832 Document new params for stack clash protection.
13834 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13836 * config/i386/i386.c (ix86_split_long_move): Do not handle
13837 address used for LEA in a special way.
13839 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13841 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13843 2017-09-19 Martin Sebor <msebor@redhat.com>
13846 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13847 of incompatible types.
13849 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13851 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13852 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13853 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13854 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13856 2017-09-19 Richard Biener <rguenther@suse.de>
13858 PR tree-optimization/82244
13859 * tree-vrp.c (remove_range_assertions): Do not propagate
13860 a constant to abnormals but replace the assert with a copy.
13862 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13864 PR rtl-optimization/57878
13865 PR rtl-optimization/68988
13866 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13867 avoidance test involving non_reload_pseudos. Move frequency test
13868 below the general fragmentation avoidance test.
13870 2017-09-19 Richard Biener <rguenther@suse.de>
13872 PR tree-optimization/69728
13873 * graphite-sese-to-poly.c (schedule_error): New global.
13874 (add_loop_schedule): Handle empty domain by failing the
13876 (build_original_schedule): Handle schedule_error.
13878 2017-09-19 Richard Biener <rguenther@suse.de>
13880 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13881 Do not iterate to sibling loops but only to siblings of inner
13884 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13887 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13888 that the registers are different.
13890 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13892 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13893 to processor_model and "amdfam17h" to arch_names_table.
13894 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13896 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13899 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13901 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13902 Alan Hayward <alan.hayward@arm.com>
13903 David Sherwood <david.sherwood@arm.com>
13905 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13907 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13909 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13910 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13911 vect_schedule_slp_instance.
13912 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13913 Update call to vect_slp_analyze_node_operations. Simplify return
13915 (vect_slp_analyze_bb_1): Update call accordingly.
13916 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13917 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13918 (vect_schedule_slp): Update call accordingly.
13920 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13921 Alan Hayward <alan.hayward@arm.com>
13922 David Sherwood <david.sherwood@arm.com>
13924 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13925 with types that aren't in fact scalar.
13927 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13929 * tree-vect-slp.c (vect_record_max_nunits): New function,
13931 (vect_build_slp_tree_1): ...here.
13932 (vect_build_slp_tree_2): Call it for phis too.
13934 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13936 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13937 to vect_get_vec_def_for_operand when getting the mask operand.
13939 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13940 Alan Hayward <alan.hayward@arm.com>
13941 David Sherwood <david.sherwood@arm.com>
13943 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13946 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13947 Alan Hayward <alan.hayward@arm.com>
13948 David Sherwood <david.sherwood@arm.com>
13950 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13951 calculation for vector booleans.
13953 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13954 Alan Hayward <alan.hayward@arm.com>
13955 David Sherwood <david.sherwood@arm.com>
13957 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13959 (vect_transform_stmt): ...here.
13960 (vect_analyze_stmt): Use it instead of calling
13961 vectorizable_live_operation directly.
13963 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13965 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13966 non-SIMT targets in acc vector loops.
13968 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13970 * configure.ac: Add arc and check if assembler supports gdwarf2.
13971 * configure: Regenerate.
13973 2017-09-18 Richard Biener <rguenther@suse.de>
13975 PR tree-optimization/82220
13976 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13977 epilogue niters from the min_profitable_iters compute.
13979 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13982 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13983 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13984 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13986 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13989 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13990 switching to a new text section.
13992 2017-09-18 Richard Biener <rguenther@suse.de>
13994 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13996 (build_alias_set): Reject aliases with no access function.
13998 2017-09-18 Richard Biener <rguenther@suse.de>
14000 PR tree-optimization/79622
14001 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14003 (build_cross_bb_scalars_use): Likewise.
14005 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14007 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14009 2017-09-18 Alan Modra <amodra@gmail.com>
14012 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14013 stack_pointer_rtx for count 0. Update comments. Break up
14014 large rtl expression.
14016 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14018 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14019 Increase to 20 bytes.
14020 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14021 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14022 or avx version of the stub.
14024 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14027 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14028 compute the minimum stack alignment. Also update preferred stack
14029 boundary for leaf functions.
14031 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14033 PR tree-optimization/82228
14034 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14037 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14039 * common/config/nds32/nds32-common.c
14040 (nds32_option_optimization_table): Refine formatting.
14041 (nds32_option_optimization_table): Use -fsched-pressure and
14042 -fomit-frame-pointer for specific optimization level.
14044 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14046 * config/nds32/nds32.c: Refine formatting and comments.
14047 * config/nds32/nds32.h: Likewise.
14048 * config/nds32/nds32.md: Likewise.
14049 * config/nds32/nds32-cost.c: Likewise.
14050 * config/nds32/nds32-isr.c: Likewise.
14051 * config/nds32/nds32-md-auxiliary.c: Likewise.
14052 * config/nds32/nds32-multiple.md: Likewise.
14053 * config/nds32/nds32-predicates.c: Likewise.
14055 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14056 Jakub Jelinek <jakub@redhat.com>
14058 Add support for -std=c++2a.
14059 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14061 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14063 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14066 * doc/extend.texi (Common Function Attributes): Add
14067 references to ARM, AArch64, and S/390 specific attributes.
14068 (Function Specific Option Pragmas): Add AArch64 and S/390
14069 to list of back ends that support the target pragma.
14071 2017-09-15 Nathan Sidwell <nathan@acm.org>
14073 * doc/standards.texi: Fix C++17 description. Update URLs for
14076 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14078 * common.opt (Wcast-align=strict): New warning option.
14079 * doc/invoke.texi: Document -Wcast-align=strict.
14081 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14083 * cgraph.h (cgraph_thunk_info): Add comments.
14084 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14085 assert for VIRTUAL_* arguments stricter.
14087 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14089 PR tree-optimization/71026
14090 * match.pd: Move RDIV patterns from fold-const.c
14091 * fold-const.c (distribute_real_division): Removed.
14092 (fold_binary_loc): Remove calls to distribute_real_divison.
14094 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14096 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14097 c++1z and gnu++1z as deprecated. Change other references to
14098 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14099 Change -Wc++1z-compat to -Wc++17-compat.
14100 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14101 * dwarf2out.c (highest_c_language): Handle C++17.
14102 (gen_compile_unit_die): Likewise.
14104 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14106 PR rtl-optimization/82192
14107 * combine.c (make_extraction): Don't look through non-paradoxical
14108 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14111 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14112 Alan Hayward <alan.hayward@arm.com>
14113 David Sherwood <david.sherwood@arm.com>
14115 * target.def (function_arg_offset): New hook.
14116 * targhooks.h (default_function_arg_offset): Declare.
14117 * targhooks.c (default_function_arg_offset): New function.
14118 * function.c (locate_and_pad_parm): Use
14119 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14120 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14121 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14122 * doc/tm.texi: Regenerate.
14123 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14124 * config/spu/spu.c (spu_function_arg_offset): New function.
14125 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14126 * system.h (FUNCTION_ARG_OFFSET): Poison.
14128 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14129 Alan Hayard <alan.hayward@arm.com>
14130 David Sherwood <david.sherwood@arm.com>
14132 * target.def (truly_noop_truncation): New hook.
14133 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14134 than TRULY_NOOP_TRUNCATION.
14135 * hooks.h (hook_bool_uint_uint_true): Declare.
14136 * hooks.c (hook_bool_uint_uint_true): New function.
14137 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14138 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14139 * doc/tm.texi: Regenerate.
14140 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14141 rather than TRULY_NOOP_TRUNCATION in comments.
14142 (simplify_comparison): Likewise.
14143 (record_truncated_value): Likewise.
14144 * expmed.c (extract_bit_field_1): Likewise.
14145 (extract_split_bit_field): Likewise.
14146 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14147 instead of TRULY_NOOP_TRUNCATION.
14148 * function.c (assign_parm_setup_block): Likewise.
14149 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14150 * rtlhooks.c: Include target.h.
14151 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14152 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14153 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14154 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14155 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14156 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14157 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14158 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14159 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14160 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14161 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14162 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14163 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14164 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14165 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14166 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14167 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14168 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14169 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14170 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14171 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14172 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14173 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14174 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14175 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14176 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14177 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14178 rather than TRULY_NOOP_TRUNCATION in comments.
14179 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14180 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14181 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14182 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14183 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14184 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14185 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14186 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14187 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14188 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14189 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14190 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14191 rather than TRULY_NOOP_TRUNCATION in comments.
14192 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14193 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14194 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14195 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14196 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14197 TRULY_NOOP_TRUNCATION condition.
14198 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14199 (TRULY_NOOP_TRUNCATION): Delete.
14200 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14201 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14202 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14203 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14204 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14205 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14206 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14207 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14208 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14209 rather than TRULY_NOOP_TRUNCATION in comments.
14210 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14211 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14212 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14213 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14214 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14215 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14217 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14220 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14221 (*cmp_ior): Likewise.
14222 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14223 (*ior_scc_scc_cmp): Likewise.
14224 (*and_scc_scc): Likewise.
14225 (*and_scc_scc_cmp): Likewise.
14227 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14228 Alan Hayard <alan.hayward@arm.com>
14229 David Sherwood <david.sherwood@arm.com>
14231 * target.def (can_change_mode_class): New hook.
14232 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14233 (hard_regno_nregs): Likewise.
14234 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14235 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14236 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14237 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14238 (LOAD_EXTEND_OP): Update accordingly.
14239 * doc/tm.texi: Regenerate.
14240 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14241 CANNOT_CHANGE_MODE_CLASS.
14242 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14243 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14244 * combine.c (simplify_set): Update accordingly.
14245 * emit-rtl.c (validate_subreg): Likewise.
14246 * recog.c (general_operand): Likewise.
14247 * regcprop.c (mode_change_ok): Likewise.
14248 * reload1.c (choose_reload_regs): Likewise.
14249 (inherit_piecemeal_p): Likewise.
14250 * rtlanal.c (simplify_subreg_regno): Likewise.
14251 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14252 instead of CANNOT_CHANGE_MODE_CLASS.
14253 (reload_cse_simplify_operands): Likewise.
14254 * reload.c (push_reload): Use targetm.can_change_mode_class
14255 instead of CANNOT_CHANGE_MODE_CLASS.
14256 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14257 REG_CANNOT_CHANGE_MODE_P.
14258 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14259 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14260 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14261 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14262 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14263 (arm_can_change_mode_class): New function.
14264 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14265 than CANNOT_CHANGE_MODE_CLASS in comments.
14266 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14267 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14268 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14269 (ix86_can_change_mode_class): ...this new function, inverting the
14270 sense of the return value.
14271 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14272 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14273 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14274 (ia64_can_change_mode_class): New function.
14275 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14276 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14277 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14278 (m32c_can_change_mode_class): ...this new function, inverting the
14279 sense of the return value.
14280 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14281 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14282 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14283 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14284 (mips_can_change_mode_class): ...this new function, inverting the
14285 sense of the return value.
14286 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14287 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14288 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14289 (msp430_can_change_mode_class): New function.
14290 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14291 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14292 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14293 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14294 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14295 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14296 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14297 (pa_cannot_change_mode_class): Replace with...
14298 (pa_can_change_mode_class): ...this new function, inverting the
14299 sense of the return value.
14300 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14301 than CANNOT_CHANGE_MODE_CLASS in comments.
14302 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14303 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14304 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14305 (pdp11_cannot_change_mode_class): Replace with...
14306 (pdp11_can_change_mode_class): ...this new function, inverting the
14307 sense of the return value.
14308 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14309 * config/powerpcspe/powerpcspe-protos.h
14310 (rs6000_cannot_change_mode_class_ptr): Delete.
14311 * config/powerpcspe/powerpcspe.c
14312 (rs6000_cannot_change_mode_class_ptr): Delete.
14313 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14314 (rs6000_option_override_internal): Assign to
14315 targetm.can_change_mode_class instead of
14316 rs6000_cannot_change_mode_class_ptr.
14317 (rs6000_cannot_change_mode_class): Replace with...
14318 (rs6000_can_change_mode_class): ...this new function, inverting the
14319 sense of the return value.
14320 (rs6000_debug_cannot_change_mode_class): Replace with...
14321 (rs6000_debug_can_change_mode_class): ...this new function.
14322 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14323 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14324 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14325 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14326 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14328 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14329 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14330 (rs6000_option_override_internal): Assign to
14331 targetm.can_change_mode_class instead of
14332 rs6000_cannot_change_mode_class_ptr.
14333 (rs6000_cannot_change_mode_class): Replace with...
14334 (rs6000_can_change_mode_class): ...this new function, inverting the
14335 sense of the return value.
14336 (rs6000_debug_cannot_change_mode_class): Replace with...
14337 (rs6000_debug_can_change_mode_class): ...this new function.
14338 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14339 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14340 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14341 (s390_can_change_mode_class): ...this new function, inverting the
14342 sense of the return value.
14343 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14344 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14345 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14346 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14347 (sh_cannot_change_mode_class): Replace with...
14348 (sh_can_change_mode_class): ...this new function, inverting the
14349 sense of the return value.
14350 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14351 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14352 (sparc_can_change_mode_class): New function.
14353 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14354 * config/spu/spu.c (spu_can_change_mode_class): New function.
14355 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14356 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14357 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14358 (visium_can_change_mode_class): New function.
14359 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14361 2017-09-15 Richard Biener <rguenther@suse.de>
14363 PR tree-optimization/82217
14364 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14365 but not undefined case.
14367 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14370 * postreload.c (reload_cse_simplify_operands): Skip
14371 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14373 2017-09-15 Richard Biener <rguenther@suse.de>
14375 PR tree-optimization/68823
14376 * graphite-scop-detection.c (build_alias_set): If we have a
14377 possible dependence check whether we can handle them by just
14378 looking at the DRs DR_ACCESS_FNs.
14379 (build_scops): If build_alias_set fails, fail the SCOP.
14381 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14383 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14384 to support float128 built-in functions that require the ISA 3.0
14386 (BU_FLOAT128_3_HW): Likewise.
14387 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14388 built-in functions.
14389 (FMAF128): Likewise.
14391 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14392 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14393 floating point instructions.
14394 (rs6000_invalid_builtin): Likewise.
14395 (rs6000_builtin_mask_names): Likewise.
14396 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14397 (RS6000_BTM_FLOAT128_HW): Likewise.
14398 (RS6000_BTM_COMMON): Likewise.
14399 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14401 * doc/extend.texi (RS/6000 built-in functions): Document the
14402 IEEE 128-bit floating point square root and fused multiply-add
14403 built-in functions.
14405 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14407 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14408 reg (r2) isn't in the set of registers defined in the prologue.
14410 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14411 Alan Hayward <alan.hayward@arm.com>
14412 David Sherwood <david.sherwood@arm.com>
14414 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14415 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14416 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14417 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14418 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14420 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14421 max_vectorization_factor.
14422 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14424 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14425 Alan Hayward <alan.hayward@arm.com>
14426 David Sherwood <david.sherwood@arm.com>
14428 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14429 (vect_worthwhile_without_simd_p): Declare.
14430 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14431 (vectorizable_reduction): Use it.
14432 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14433 (vectorizable_operation): Likewise.
14435 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14436 Alan Hayward <alan.hayward@arm.com>
14437 David Sherwood <david.sherwood@arm.com>
14439 * tree-vectorizer.h (vect_get_num_copies): New function.
14440 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14441 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14442 (vectorizable_induction): Likewise.
14443 (vectorizable_live_operation): Likewise.
14444 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14445 (vectorizable_bswap): Likewise.
14446 (vectorizable_call): Likewise.
14447 (vectorizable_conversion): Likewise.
14448 (vectorizable_assignment): Likewise.
14449 (vectorizable_shift): Likewise.
14450 (vectorizable_operation): Likewise.
14451 (vectorizable_store): Likewise.
14452 (vectorizable_load): Likewise.
14453 (vectorizable_condition): Likewise.
14454 (vectorizable_comparison): Likewise.
14455 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14457 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14458 Alan Hayward <alan.hayward@arm.com>
14459 David Sherwood <david.sherwood@arm.com>
14461 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14462 of vect_init_vector.
14464 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14465 Alan Hayward <alan.hayward@arm.com>
14466 David Sherwood <david.sherwood@arm.com>
14468 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14469 an inline wrapper that provides a location.
14470 (gimple_build_vector): Likewise.
14471 * gimple-fold.c (gimple_build_vector_from_val): New function.
14472 (gimple_build_vector): Likewise.
14473 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14474 functions to build the initial value. Always return a gimple value.
14475 (get_initial_defs_for_reduction): Likewise. Only compute
14477 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14478 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14479 (vectorizable_induction): Use gimple_build_vector rather than
14482 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14483 Alan Hayward <alan.hayward@arm.com>
14484 David Sherwood <david.sherwood@arm.com>
14486 * target.h (vec_perm_indices): New typedef.
14487 (auto_vec_perm_indices): Likewise.
14488 * optabs-query.h: Include target.h
14489 (can_vec_perm_p): Take a vec_perm_indices *.
14490 * optabs-query.c (can_vec_perm_p): Likewise.
14491 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14492 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14493 * tree-vect-generic.c (lower_vec_perm): Likewise.
14494 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14495 (vect_grouped_load_supported): Likewise.
14496 (vect_shift_permute_load_chain): Likewise.
14497 (vect_permute_store_chain): Use auto_vec_perm_indices.
14498 (vect_permute_load_chain): Likewise.
14499 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14500 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14501 Update uses of can_vec_perm_p.
14502 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14503 mode with a number of elements. Take a vec_perm_indices *.
14504 (vect_create_epilog_for_reduction): Update accordingly.
14505 Use auto_vec_perm_indices.
14506 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14507 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14508 (vect_transform_slp_perm_load): Likewise.
14509 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14510 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14511 (vect_gen_perm_mask_checked): Likewise.
14512 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14513 (vect_gen_perm_mask_checked): Likewise.
14514 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14515 (vectorizable_store): Likewise.
14516 (vectorizable_load): Likewise.
14517 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14518 (vectorizable_bswap): Likewise.
14520 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14521 Alan Hayward <alan.hayward@arm.com>
14522 David Sherwood <david.sherwood@arm.com>
14524 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14525 * tree.c (build_vector): Likewise.
14526 (build_vector_from_ctor): Update accordingly.
14527 (build_vector_from_val): Likewise.
14528 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14529 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14530 * tree-vect-generic.c (add_rshift): Likewise.
14531 (expand_vector_divmod): Likewise.
14532 (optimize_vector_constructor): Likewise.
14533 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14534 (vect_transform_slp_perm_load): Likewise.
14535 (vect_schedule_slp_instance): Likewise.
14536 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14537 (vectorizable_call): Likewise.
14538 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14539 * expmed.c (make_tree): Likewise.
14540 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14541 a vector passed to build_vector.
14542 (fold_convert_const): Likewise.
14543 (exact_inverse): Likewise.
14544 (fold_ternary_loc): Likewise.
14545 (fold_relational_const): Likewise.
14546 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14547 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14548 (const_unop): Likewise. Store the reduction accumulator in a
14549 variable rather than an array.
14550 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14551 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14552 the new vector, rather than constructing it after the input arrays.
14553 (native_interpret_vector): Use auto_vec<tree> when building
14554 a vector passed to build_vector. Add elements in order.
14555 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14556 auto_vec<tree> when building a vector passed to build_vector.
14557 (vect_create_epilog_for_reduction): Likewise.
14558 (vectorizable_induction): Likewise.
14559 (get_initial_def_for_reduction): Likewise. Fix indentation of
14561 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14563 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14564 passed to build_vector.
14566 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14567 Alan Hayward <alan.hayward@arm.com>
14568 David Sherwood <david.sherwood@arm.com>
14570 * tree-core.h (tree_base::u): Add an "nelts" field.
14571 (tree_vector): Use VECTOR_CST_NELTS as the length.
14572 * tree.c (tree_size): Likewise.
14573 (make_vector): Initialize VECTOR_CST_NELTS.
14574 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14575 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14576 TYPE_VECTOR_SUBPARTS.
14577 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14578 as the number of units, setting it from VECTOR_CST_NELTS.
14579 (const_vector_from_tree): Likewise.
14580 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14581 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14582 (fold_negate_expr_1): Likewise.
14583 (fold_convert_const): Likewise.
14584 (const_binop): Likewise. Differentiate the number of output and
14586 (const_unop): Likewise.
14587 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14588 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14589 in cases that did the opposite.
14591 2017-09-14 Richard Biener <rguenther@suse.de>
14593 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14596 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14598 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14600 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14603 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14604 if and where to split a bb, except for splitting before debug insn
14605 sequences followed by non-label real insn. Delete debug insns
14606 in between basic blocks.
14608 * combine.c (make_compound_operation_int): Formatting fixes.
14610 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14611 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14612 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14613 * config/sol2.h (LINK_EH_SPEC): Likewise.
14614 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14615 * config/s390/linux.h (LINK_SPEC): Likewise.
14616 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14617 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14618 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14619 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14620 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14621 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14622 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14623 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14625 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14627 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14629 (ENDFILE_LINUX_SPEC): Likewise.
14630 (LINK_EH_SPEC): Likewise.
14631 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14632 (LINK_OS_LINUX_SPEC32): Likewise.
14633 (LINK_OS_LINUX_SPEC64): Likewise.
14634 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14635 (LINK_OS_LINUX_SPEC): Likewise.
14637 2017-09-13 Martin Liska <mliska@suse.cz>
14639 PR middle-end/82154
14640 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14641 CASE_HIGH is NULL_TREE.
14643 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14644 Alan Hayward <alan.hayward@arm.com>
14645 David Sherwood <david.sherwood@arm.com>
14647 * target.def (secondary_memory_needed): New hook.
14648 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14649 instead of SECONDARY_MEMORY_NEEDED.
14650 (secondary_memory_needed_mode): Likewise.
14651 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14652 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14653 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14654 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14655 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14656 * doc/tm.texi: Regenerate.
14657 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14658 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14659 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14660 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14661 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14662 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14663 mode argument first and change the reg_class arguments to reg_class_t.
14664 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14665 Make static. Update the call to inline_secondary_memory_needed.
14666 (ix86_register_move_cost): Update the call to
14667 inline_secondary_memory_needed.
14668 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14669 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14671 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14672 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14674 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14675 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14676 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14677 and match hook interface. Add comment from mips.h.
14678 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14679 * config/mmix/mmix.md (truncdfsf2): Refer to
14680 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14682 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14683 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14684 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14685 (pa_secondary_memory_needed): New function.
14686 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14687 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14688 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14689 (pdp11_secondary_memory_needed): Make static and match hook interface.
14690 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14691 * config/powerpcspe/powerpcspe-protos.h
14692 (rs6000_secondary_memory_needed_ptr): Delete.
14693 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14695 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14696 (rs6000_option_override_internal): Assign to
14697 targetm.secondary_memory_needed rather than
14698 rs6000_secondary_memory_needed_ptr.
14699 (rs6000_secondary_memory_needed): Match hook interface.
14700 (rs6000_debug_secondary_memory_needed): Likewise.
14701 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14702 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14703 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14704 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14705 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14706 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14708 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14709 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14710 (rs6000_option_override_internal): Assign to
14711 targetm.secondary_memory_needed rather than
14712 rs6000_secondary_memory_needed_ptr.
14713 (rs6000_secondary_memory_needed): Match hook interface.
14714 (rs6000_debug_secondary_memory_needed): Likewise.
14715 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14716 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14717 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14718 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14719 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14720 (sparc_secondary_memory_needed): New function.
14721 * lra-constraints.c (check_and_process_move): Refer to
14722 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14724 (curr_insn_transform): Likewise.
14725 (process_alt_operands): Use targetm.secondary_memory_needed
14726 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14727 (check_secondary_memory_needed_p): Likewise.
14728 (choose_split_class): Likewise.
14729 * reload.c: Unconditionally include code that was previously
14730 conditional on SECONDARY_MEMORY_NEEDED.
14731 (push_secondary_reload): Use targetm.secondary_memory_needed
14732 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14733 (push_reload): Likewise.
14734 * reload1.c: Unconditionally include code that was previously
14735 conditional on SECONDARY_MEMORY_NEEDED.
14736 (choose_reload_regs): Use targetm.secondary_memory_needed
14737 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14738 (gen_reload): Likewise.
14739 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14741 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14742 Alan Hayward <alan.hayward@arm.com>
14743 David Sherwood <david.sherwood@arm.com>
14745 * target.def (secondary_memory_needed_mode): New hook:
14746 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14747 * targhooks.h (default_secondary_memory_needed_mode): New function.
14748 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14749 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14750 * doc/tm.texi: Regenerate.
14751 * lra-constraints.c (check_and_process_move): Use
14752 targetm.secondary_memory_needed_mode instead of
14753 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14754 (curr_insn_transform): Likewise.
14755 * reload.c (get_secondary_mem): Likewise.
14756 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14757 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14759 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14760 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14761 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14762 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14763 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14765 * config/powerpcspe/powerpcspe-protos.h
14766 (rs6000_secondary_memory_needed_mode): Delete.
14767 * config/powerpcspe/powerpcspe.c
14768 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14769 (rs6000_secondary_memory_needed_mode): Make static.
14770 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14771 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14773 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14775 (rs6000_secondary_memory_needed_mode): Make static.
14776 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14777 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14778 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14779 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14780 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14782 (sparc_secondary_memory_needed_mode): New function.
14783 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14785 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14787 * config/aarch64/constraints.md (Umq): New constraint.
14788 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14790 (mov<mode>): Update condition.
14792 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14794 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14795 when bitposition is the same.
14797 2017-09-13 Richard Biener <rguenther@suse.de>
14799 * dwarf2out.c (output_die_symbol): Remove.
14800 (output_die): Do not output a DIEs symbol.
14802 2017-09-13 Richard Biener <rguenther@suse.de>
14804 PR middle-end/82128
14805 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14806 default-def to avoid breaking iterator update with the weird
14807 interaction with cgraph_update_edges_for_call_stmt_node.
14809 2017-09-13 Richard Biener <rguenther@suse.de>
14811 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14812 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14813 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14814 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14816 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14818 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14819 Disable pc relative literal load irrespective of
14820 TARGET_FIX_ERR_A53_84341 for default.
14822 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14824 * config/sparc/sparc.c (output_return): Output the source location of
14825 the insn in the delay slot, if any.
14826 (output_sibcall): Likewise.
14828 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14831 * common.opt (-static-pie): New alias.
14832 (shared): Negate static-pie.
14833 (-no-pie): Update help text.
14835 (static-pie): New option.
14836 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14837 -static-pie support.
14838 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14839 (LINK_EH_SPEC): Likewise.
14840 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14841 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14842 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14843 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14844 (init_gcc_specs): Likewise.
14845 (init_spec): Likewise.
14846 (display_help): Update help message for -pie.
14847 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14850 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14852 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14853 (movdi_aarch64): Likewise.
14854 (movti_aarch64): Likewise.
14856 2017-09-12 Simon Wright <simon@pushface.org>
14859 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14860 calculation of the minor version, always output as 0.
14862 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14865 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14866 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14867 on it early, rather than manual conversion late. For
14868 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14869 instead of performing manual conversion.
14871 2017-09-12 Carl Love <cel@us.ibm.com>
14873 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14874 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14876 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14877 VMULOSW): Add definitions.
14878 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14879 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14880 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14881 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14882 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14884 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14886 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14888 (movti_aarch64): Likewise.
14889 (movdf_aarch64): Likewise.
14890 (movtf_aarch64): Likewise.
14891 (load_pairdi): Likewise.
14892 (store_pairdi): Likewise.
14893 (load_pairdf): Likewise.
14894 (store_pairdf): Likewise.
14895 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14896 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14897 (ldr_got_small_<mode>): Likewise.
14898 (ldr_got_small_28k_<mode>): Likewise.
14899 (ldr_got_tiny): Likewise.
14900 * config/aarch64/iterators.md (ldst_sz): New.
14901 (ldpstp_sz): Likewise.
14902 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14904 (thunderx_load): Split load_8 to load_16.
14905 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14907 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14908 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14909 (xgene1_store_pair): Split store_8 to store_16.
14910 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14911 (falkor_st_0_st_sd): Split store_8 to store_16.
14913 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14915 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14916 and store1/2/3/4 to store_4/8/12/16.
14917 * config/aarch64/aarch64.md: Update for rename.
14918 * config/arm/arm.md: Likewise.: Likewise.
14919 * config/arm/arm.c: Likewise.
14920 * config/arm/thumb1.md: Likewise.
14921 * config/arm/thumb2.md: Likewise.
14922 * config/arm/vfp.md: Likewise.
14923 * config/arm/arm-generic.md: Likewise.
14924 * config/arm/arm1020e.md: Likewise.
14925 * config/arm/arm1026ejs.md: Likewise.
14926 * config/arm/arm1136jfs.md: Likewise.
14927 * config/arm/arm926ejs.md: Likewise.
14928 * config/arm/cortex-a15.md: Likewise.
14929 * config/arm/cortex-a17.md: Likewise.
14930 * config/arm/cortex-a5.md: Likewise.
14931 * config/arm/cortex-a53.md: Likewise.
14932 * config/arm/cortex-a57.md: Likewise.
14933 * config/arm/cortex-a7.md: Likewise.
14934 * config/arm/cortex-a8.md: Likewise.
14935 * config/arm/cortex-a9.md: Likewise.
14936 * config/arm/cortex-m4.md: Likewise.
14937 * config/arm/cortex-m7.md: Likewise.
14938 * config/arm/cortex-r4.md: Likewise.
14939 * config/arm/exynos-m1.md: Likewise.
14940 * config/arm/fa526.md: Likewise.
14941 * config/arm/fa606te.md: Likewise.
14942 * config/arm/fa626te.md: Likewise.
14943 * config/arm/fa726te.md: Likewise.
14944 * config/arm/fmp626.md: Likewise.
14945 * config/arm/iwmmxt.md: Likewise.
14946 * config/arm/ldmstm.md: Likewise.
14947 * config/arm/marvell-pj4.md: Likewise.
14948 * config/arm/xgene1.md: Likewise.
14949 * config/aarch64/thunderx.md: Likewise.
14950 * config/aarch64/thunderx2t99.md: Likewise.
14951 * config/aarch64/falkor.md: Likewise.
14953 2017-09-12 Martin Liska <mliska@suse.cz>
14955 * attribs.c (private_lookup_attribute): New function.
14956 * attribs.h (private_lookup_attribute): Declared here.
14957 (lookup_attribute): Called from this place.
14959 2017-09-12 Richard Biener <rguenther@suse.de>
14961 PR tree-optimization/82157
14962 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14963 stmts with side-effects.
14965 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14966 Alan Hayward <alan.hayward@arm.com>
14967 David Sherwood <david.sherwood@arm.com>
14969 * target.def (hard_regno_nregs): New hook.
14970 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14971 * targhooks.h (default_hard_regno_nregs): Declare.
14972 * targhooks.c (default_hard_regno_nregs): New function.
14973 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14974 (TARGET_HARD_REGNO_NREGS): ...this hook.
14975 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14976 (CLASS_MAX_NREGS): Likewise.
14977 * doc/tm.texi: Regenerate.
14978 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14979 instead of HARD_REGNO_NREGS.
14980 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14981 HARD_REGNO_NREGS in the comment.
14982 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14983 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14984 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14985 Return an unsigned int.
14986 (TARGET_HARD_REGNO_NREGS): Redefine.
14987 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14988 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14989 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14990 (arc_hard_regno_nregs): New function.
14991 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14992 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14993 (arm_hard_regno_nregs): New function.
14994 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14995 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14996 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14997 (TARGET_HARD_REGNO_NREGS): Redefine.
14998 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14999 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15000 (HARD_REGNO_NREGS): Delete.
15001 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15002 (cr16_hard_regno_nregs): New function.
15003 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15004 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15005 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15006 (cris_hard_regno_nregs): New function.
15007 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15008 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15009 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15010 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15011 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15012 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15013 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15014 (frv_hard_regno_nregs): Make static. Take and return an
15016 (frv_class_max_nregs): Remove outdated copy of documentation.
15017 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15018 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15019 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15020 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15021 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15022 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15023 (TARGET_HARD_REGNO_NREGS): Redefine.
15024 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15025 (CLASS_MAX_NREGS): Update comment.
15026 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15027 (ia64_hard_regno_nregs): New function.
15028 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15029 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15030 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15031 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15032 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15034 (m32c_hard_regno_nregs): Likewise. Make static.
15035 (TARGET_HARD_REGNO_NREGS): Redefine.
15036 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15037 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15038 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15039 (m68k_hard_regno_nregs): New function.
15040 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15041 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15042 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15043 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15044 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15045 Take and return an unsigned int.
15046 (TARGET_HARD_REGNO_NREGS): Redefine.
15047 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15048 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15049 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15050 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15051 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15052 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15053 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15054 (msp430_hard_regno_nregs): Make static. Take and return an
15056 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15057 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15058 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15059 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15060 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15061 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15062 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15063 (TARGET_HARD_REGNO_NREGS): Redefine.
15064 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15065 (PA_HARD_REGNO_NREGS): ...this.
15066 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15067 (PA_HARD_REGNO_NREGS): ...this.
15068 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15069 (pa_hard_regno_nregs): New function.
15070 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15071 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15072 (pdp11_hard_regno_nregs): New function.
15073 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15074 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15075 (rs6000_hard_regno_nregs_hook): New function.
15076 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15077 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15078 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15079 Take and return an unsigned int. Move earlier in file.
15080 (TARGET_HARD_REGNO_NREGS): Redefine.
15081 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15082 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15083 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15084 (rl78_hard_regno_nregs): Make static. Take and return an
15086 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15087 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15088 (rs6000_hard_regno_nregs_hook): New function.
15089 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15090 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15091 (TARGET_HARD_REGNO_NREGS): Redefine.
15092 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15093 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15094 instead of HARD_REGNO_NREGS.
15095 (s390_hard_regno_nregs): New function.
15096 (s390_hard_regno_mode_ok): Add comment from s390.h.
15097 (TARGET_HARD_REGNO_NREGS): Redefine.
15098 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15099 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15100 (sh_hard_regno_nregs): New function.
15101 (sh_pass_in_reg_p): Use it.
15102 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15103 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15104 (sparc_hard_regno_nregs): New function.
15105 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15106 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15107 (spu_function_arg_advance): Use it, supplying a valid register number.
15108 (TARGET_HARD_REGNO_NREGS): Redefine.
15109 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15110 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15111 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15112 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15113 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15114 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15115 (CLASS_MAX_NREGS): Remove copy of old documentation.
15116 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15117 (visium_hard_regno_nregs): New function.
15118 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15119 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15120 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15121 xtensa_hard_regno_nregs): New function.
15122 * system.h (HARD_REGNO_NREGS): Poison.
15124 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15126 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15127 hard_regno_nregs instead of HARD_REGNO_NREGS.
15128 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15129 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15130 (c6x_expand_epilogue): Likewise.
15131 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15132 (frv_read_iacc_argument): Likewise.
15133 * config/sh/sh.c: Include regs.h.
15134 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15135 (regs_used): Likewise.
15136 (output_stack_adjust): Likewise.
15137 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15138 * expmed.c: Include regs.h.
15139 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15140 * ree.c: Include regs.h.
15141 (combine_reaching_defs): Use hard_regno_nregs instead of
15143 (add_removable_extension): Likewise.
15145 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15147 * regs.h (hard_regno_nregs): Turn into a function.
15148 (end_hard_regno): Update accordingly.
15149 * caller-save.c (setup_save_areas): Likewise.
15150 (save_call_clobbered_regs): Likewise.
15151 (replace_reg_with_saved_mem): Likewise.
15152 (insert_restore): Likewise.
15153 (insert_save): Likewise.
15154 * combine.c (can_change_dest_mode): Likewise.
15155 (move_deaths): Likewise.
15156 (distribute_notes): Likewise.
15157 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15158 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15159 (rs6000_split_multireg_move): Likewise.
15160 (rs6000_register_move_cost): Likewise.
15161 (rs6000_memory_move_cost): Likewise.
15162 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15163 (rs6000_split_multireg_move): Likewise.
15164 (rs6000_register_move_cost): Likewise.
15165 (rs6000_memory_move_cost): Likewise.
15166 * cselib.c (cselib_reset_table): Likewise.
15167 (cselib_lookup_1): Likewise.
15168 * emit-rtl.c (set_mode_and_regno): Likewise.
15169 * function.c (aggregate_value_p): Likewise.
15170 * ira-color.c (setup_profitable_hard_regs): Likewise.
15171 (check_hard_reg_p): Likewise.
15172 (calculate_saved_nregs): Likewise.
15173 (assign_hard_reg): Likewise.
15174 (improve_allocation): Likewise.
15175 (calculate_spill_cost): Likewise.
15176 * ira-emit.c (modify_move_list): Likewise.
15177 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15178 (ira_hard_reg_in_set_p): Likewise.
15179 * ira.c (setup_reg_mode_hard_regset): Likewise.
15180 (clarify_prohibited_class_mode_regs): Likewise.
15181 (check_allocation): Likewise.
15182 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15183 (lra_setup_reg_renumber): Likewise.
15184 (setup_try_hard_regno_pseudos): Likewise.
15185 (spill_for): Likewise.
15186 (assign_hard_regno): Likewise.
15187 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15188 * lra-constraints.c (in_class_p): Likewise.
15189 (lra_constraint_offset): Likewise.
15190 (simplify_operand_subreg): Likewise.
15191 (lra_constraints): Likewise.
15192 (split_reg): Likewise.
15193 (split_if_necessary): Likewise.
15194 (invariant_p): Likewise.
15195 (inherit_in_ebb): Likewise.
15196 * lra-lives.c (process_bb_lives): Likewise.
15197 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15198 (get_hard_regs): Likewise.
15199 (do_remat): Likewise.
15200 * lra-spills.c (assign_spill_hard_regs): Likewise.
15201 * mode-switching.c (create_pre_exit): Likewise.
15202 * postreload.c (reload_combine_recognize_pattern): Likewise.
15203 * recog.c (peep2_find_free_register): Likewise.
15204 * regcprop.c (kill_value_regno): Likewise.
15205 (set_value_regno): Likewise.
15206 (copy_value): Likewise.
15207 (maybe_mode_change): Likewise.
15208 (find_oldest_value_reg): Likewise.
15209 (copyprop_hardreg_forward_1): Likewise.
15210 * regrename.c (check_new_reg_p): Likewise.
15211 (regrename_do_replace): Likewise.
15212 * reload.c (push_reload): Likewise.
15213 (combine_reloads): Likewise.
15214 (find_dummy_reload): Likewise.
15215 (operands_match_p): Likewise.
15216 (find_reloads): Likewise.
15217 (find_equiv_reg): Likewise.
15218 (reload_adjust_reg_for_mode): Likewise.
15219 * reload1.c (count_pseudo): Likewise.
15220 (count_spilled_pseudo): Likewise.
15221 (find_reg): Likewise.
15222 (clear_reload_reg_in_use): Likewise.
15223 (free_for_value_p): Likewise.
15224 (allocate_reload_reg): Likewise.
15225 (choose_reload_regs): Likewise.
15226 (reload_adjust_reg_for_temp): Likewise.
15227 (emit_reload_insns): Likewise.
15228 (delete_output_reload): Likewise.
15229 * rtlanal.c (subreg_get_info): Likewise.
15230 * sched-deps.c (sched_analyze_reg): Likewise.
15231 * sel-sched.c (init_regs_for_mode): Likewise.
15232 (mark_unavailable_hard_regs): Likewise.
15233 (choose_best_reg_1): Likewise.
15234 (verify_target_availability): Likewise.
15235 * valtrack.c (dead_debug_insert_temp): Likewise.
15236 * var-tracking.c (track_loc_p): Likewise.
15237 (emit_note_insn_var_location): Likewise.
15238 * varasm.c (make_decl_rtl): Likewise.
15239 * reginfo.c (choose_hard_reg_mode): Likewise.
15240 (init_reg_modes_target): Refer directly to
15241 this_target_regs->x_hard_regno_nregs.
15243 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15245 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15246 instead of hard_regno_nregs.
15248 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15250 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15251 end_hard_regno instead of hard_regno_nregs.
15252 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15253 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15254 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15255 * ira-color.c (improve_allocation): Likewise.
15256 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15257 * lra-lives.c (mark_regno_live): Likewise.
15258 (mark_regno_dead): Likewise.
15259 * lra-remat.c (operand_to_remat): Likewise.
15260 * lra.c (collect_non_operand_hard_regs): Likewise.
15261 * postreload.c (reload_combine_note_store): Likewise.
15262 (move2add_valid_value_p): Likewise.
15263 * reload.c (regno_clobbered_p): Likewise.
15265 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15267 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15269 * config/v850/v850.c (v850_reorg): Likewise.
15270 * reload.c (refers_to_regno_for_reload_p): Likewise.
15271 (find_equiv_reg): Likewise.
15272 * reload1.c (reload_reg_reaches_end_p): Likewise.
15274 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15276 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15278 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15279 * config/arm/arm.c (output_move_neon): Likewise.
15280 (arm_attr_length_move_neon): Likewise.
15281 (neon_split_vcombine): Likewise.
15282 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15283 (c6x_mark_reg_written): Likewise.
15284 (c6x_dwarf_register_span): Likewise.
15285 * config/i386/i386.c (ix86_save_reg): Likewise.
15286 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15287 (rws_access_reg): Likewise.
15288 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15289 * mode-switching.c (create_pre_exit): Likewise.
15290 * ree.c (combine_reaching_defs): Likewise.
15291 (add_removable_extension): Likewise.
15292 * regcprop.c (find_oldest_value_reg): Likewise.
15293 (copyprop_hardreg_forward_1): Likewise.
15294 * reload.c (reload_inner_reg_of_subreg): Likewise.
15295 (push_reload): Likewise.
15296 (combine_reloads): Likewise.
15297 (find_dummy_reload): Likewise.
15298 (reload_adjust_reg_for_mode): Likewise.
15299 * reload1.c (find_reload_regs): Likewise.
15300 (forget_old_reloads_1): Likewise.
15301 (reload_reg_free_for_value_p): Likewise.
15302 (reload_adjust_reg_for_temp): Likewise.
15303 (emit_reload_insns): Likewise.
15304 (delete_output_reload): Likewise.
15305 * sel-sched.c (choose_best_reg_1): Likewise.
15306 (choose_best_pseudo_reg): Likewise.
15308 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15309 Alan Hayward <alan.hayward@arm.com>
15310 David Sherwood <david.sherwood@arm.com>
15312 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15313 * target.def (slow_unaligned_access): New hook.
15314 * targhooks.h (default_slow_unaligned_access): Declare.
15315 * targhooks.c (default_slow_unaligned_access): New function.
15316 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15317 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15318 * doc/tm.texi: Regenerate.
15319 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15320 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15321 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15323 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15324 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15326 (rs6000_slow_unaligned_access): New function.
15327 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15328 (expand_block_compare): Likewise.
15329 (expand_strn_compare): Likewise.
15330 (rs6000_rtx_costs): Likewise.
15331 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15332 (riscv_slow_unaligned_access): Likewise.
15333 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15334 (riscv_slow_unaligned_access_p): ...this and make static.
15335 (riscv_option_override): Update accordingly.
15336 (riscv_slow_unaligned_access): New function.
15337 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15338 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15339 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15340 (rs6000_slow_unaligned_access): New function.
15341 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15342 (rs6000_rtx_costs): Likewise.
15343 * config/rs6000/rs6000-string.c (expand_block_compare)
15344 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15345 of SLOW_UNALIGNED_ACCESS.
15346 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15347 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15348 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15349 of SLOW_UNALIGNED_ACCESS.
15350 * expmed.c (simple_mem_bitfield_p): Likewise.
15351 * expr.c (alignment_for_piecewise_move): Likewise.
15352 (emit_group_load_1): Likewise.
15353 (emit_group_store): Likewise.
15354 (copy_blkmode_from_reg): Likewise.
15355 (emit_push_insn): Likewise.
15356 (expand_assignment): Likewise.
15357 (store_field): Likewise.
15358 (expand_expr_real_1): Likewise.
15359 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15360 * lra-constraints.c (simplify_operand_subreg): Likewise.
15361 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15362 * gimple-ssa-store-merging.c: Likewise in block comment at start
15364 * tree-ssa-strlen.c: Include target.h.
15365 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15366 of SLOW_UNALIGNED_ACCESS.
15367 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15369 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15371 PR rtl-optimization/82185
15372 * expmed.c (emit_store_flag_int): Only test tem if it has been
15375 2017-09-12 Richard Biener <rguenther@suse.de>
15377 PR middle-end/82149
15378 * match.pd ((FTYPE) N CMP CST): Fix typo.
15380 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15382 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15384 (mips_near_type_p): Add 'short_call' attribute as a synonym
15386 * doc/extend.texi (short_call): Document new function attribute.
15388 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15391 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15392 assertion check that in the condition.
15393 (get_atomic_generic_size): Likewise. Before testing if parameter
15394 has pointer type, if it has array type, call for C++
15395 default_conversion to perform array-to-pointer conversion.
15397 2017-09-12 Richard Biener <rguenther@suse.de>
15399 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15400 for operations we cannot scalarize.
15402 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15404 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15405 vectors heap vectors. Clean up comments.
15406 Make visited_bbs a reference.
15407 (profitable_jump_thread_path): Make GC
15408 vectors heap vectors. Clean up comments.
15410 (convert_and_register_jump_thread_path): Make GC vectors heap
15412 (check_subpath_and_update_thread_path): Same. Clean up comments.
15413 Make visited_bbs a reference.
15414 (handle_phi): Abstract common code to to
15415 register_jump_thread_path_if_profitable.
15416 Rename VAR_BB to DEF_BB.
15418 Make GC vectors heap vectors.
15419 Make visited_bbs a reference.
15420 (handle_assignment): Same.
15421 (register_jump_thread_path_if_profitable): New.
15422 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15424 Make GC vectors heap vectors. Clean up comments.
15425 Make visited_bbs a reference.
15426 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15427 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15430 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15433 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15434 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15436 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15438 Revert r251800 and r251799.
15440 2017-09-11 Martin Jambor <mjambor@suse.cz>
15443 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15444 arguments in advance.
15445 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15446 use it to find predecessor edges.
15447 (naive_outof_ssa): Collect vector of predecessors.
15449 2017-09-08 Jason Merrill <jason@redhat.com>
15451 PR c++/70029 - ICE with ref-qualifier and -flto
15452 * langhooks.h (struct lang_hooks_for_types): Add
15453 copy_lang_qualifiers.
15454 * attribs.c (build_type_attribute_qual_variant): Use it.
15455 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15457 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15458 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15460 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15463 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15464 (*mulsi3_sp64): New instruction.
15465 (mulsi3): New expander.
15467 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15469 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15471 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15473 * sancov.c: Include memmodel.h.
15475 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15478 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15481 2017-09-07 Carl Love <cel@us.ibm.com>
15483 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15484 the sldi instruction.
15486 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15488 * sancov.c: Include tm_p.h.
15490 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15493 * output.h (switch_to_other_text_partition): New declaration.
15494 * varasm.c (switch_to_other_text_partition): New function.
15495 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15496 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15497 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15498 to the other text partition before emitting LCL label and switch back
15499 after emitting the word after it.
15501 2017-09-07 Richard Biener <rguenther@suse.de>
15503 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15504 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15505 critical edges here, after loop init.
15506 (pass_data_pre): Remove PROP_no_crit_edges flags.
15507 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15508 for valueization of call args to avoid leaking VN_TOP.
15509 (visit_use): Assert we do not visit default defs.
15510 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15511 Use error_mark_node to more easily detect leaking VN_TOP.
15512 All default-defs are varying, not VN_TOP. Mark them visited.
15513 (run_scc_vn): Make code match comment.
15515 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15517 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15518 OPTION_MASK_FLOAT128_KEYWORD.
15519 (POWERPC_MASKS): Likewise.
15520 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15521 support for the -mfloat128-type option, and make -mfloat128
15522 default on PowerPC Linux systems. Define or undefine
15523 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15524 Define __float128 to be __ieee128 if IEEE 128-bit support is
15525 enabled, or undefine it.
15526 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15527 Delete defining __FLOAT128_TYPE__.
15528 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15529 -mfloat128-type option and make -mfloat128 default on PowerPC
15531 (TARGET_FLOAT128_TYPE): Likewise.
15532 (-mfloat128-type): Likewise.
15533 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15534 Delete the -mfloat128-type option and make -mfloat128 default on
15535 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15536 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15537 128-bit floating point is enabled. Change tests from using
15538 -mfloat128-type to -mfloat128.
15539 (rs6000_mangle_type): Use the correct mangling for the __float128
15540 type even if normal long double is restricted to 64-bits.
15541 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15543 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15544 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15545 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15547 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15548 documentation for -mfloat128.
15550 2017-09-06 Olivier Hainque <hainque@adacore.com>
15552 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15554 2017-09-06 Wish Wu <wishwu007@gmail.com>
15555 Jakub Jelinek <jakub@redhat.com>
15557 * asan.c (initialize_sanitizer_builtins): Add
15558 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15559 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15560 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15561 BT_FN_VOID_UINT64_PTR variables.
15562 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15563 (BT_FN_VOID_UINT16_UINT16): Likewise.
15564 (BT_FN_VOID_UINT32_UINT32): Likewise.
15565 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15566 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15567 (BT_FN_VOID_UINT64_PTR): Likewise.
15568 * common.opt (flag_sanitize_coverage): New variable.
15569 (fsanitize-coverage=trace-pc): Remove.
15570 (fsanitize-coverage=): Add.
15571 * flag-types.h (enum sanitize_coverage_code): New enum.
15572 * fold-const.c (fold_range_test): Disable non-short-circuit
15573 optimization if flag_sanitize_coverage.
15574 (fold_truth_andor): Likewise.
15575 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15576 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15577 (coverage_sanitizer_opts): New array.
15578 (get_closest_sanitizer_option): Add OPTS argument, handle also
15579 OPT_fsanitize_coverage_.
15580 (parse_sanitizer_options): Adjusted to also handle
15581 OPT_fsanitize_coverage_.
15582 (common_handle_option): Add OPT_fsanitize_coverage_.
15583 * sancov.c (instrument_comparison, instrument_switch): New function.
15584 (sancov_pass): Add trace-cmp support.
15585 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15586 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15587 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15588 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15589 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15590 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15591 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15592 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15593 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15594 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15596 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15598 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15599 error. Only quit immediately if parsing is complete.
15600 (BEGIN): Initialize fatal_err and parse_done.
15601 (begin fpu, end fpu): Check number of arguments.
15602 (begin arch, end arch): Likewise.
15603 (begin cpu, end cpu): Likewise.
15604 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15605 (optalias): Likewise.
15607 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15609 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15610 * config/arm/arm-isa.h: Delete. Move definitions to ...
15611 * arm-cpus.in: ... here. Use new feature and fgroup values.
15612 * config/arm/arm.c (arm_option_override): Use lower case for feature
15614 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15615 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15616 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15618 (print_isa_bits_for): New function.
15619 (gen_isa): New function.
15620 (gen_comm_data): Use print_isa_bits_for.
15621 (define feature): New keyword.
15622 (define fgroup): New keyword.
15623 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15624 (arm-isa.h): Add rule to generate file.
15625 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15626 case for feature bit names.
15628 2017-09-06 Richard Biener <rguenther@suse.de>
15630 * tree-ssa-pre.c (NECESSARY): Remove.
15631 (create_expression_by_pieces): Do not touch pass-local flags.
15632 (insert_into_preds_of_block): Likewise.
15633 (do_pre_regular_insertion): Likewise.
15634 (eliminate_insert): Likewise.
15635 (eliminate_dom_walker::before_dom_children): Likewise.
15636 (fini_eliminate): Do not look at inserted_exprs.
15637 (mark_operand_necessary): Remove.
15638 (remove_dead_inserted_code): Replace with simple work-list
15639 algorithm based on inserted_exprs and SSA uses.
15640 (pass_pre::execute): Re-order fini_eliminate and
15641 remove_dead_inserted_code.
15643 2017-09-06 Olivier Hainque <hainque@adacore.com>
15645 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15646 for VxWorks 7. Adjust surrounding comments.
15648 2017-09-06 Richard Biener <rguenther@suse.de>
15650 * gimple-ssa-strength-reduction.c
15651 (find_candidates_dom_walker::before_dom_children): Also allow
15654 2017-09-06 Richard Biener <rguenther@suse.de>
15656 PR tree-optimization/82108
15657 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15658 for gap in the non-permutation SLP case.
15660 2017-09-06 Martin Jambor <mjambor@suse.cz>
15662 PR tree-optimization/82078
15663 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15664 add_access_to_work_queue...
15665 (build_accesses_from_assign): ...here.
15666 (propagate_all_subaccesses): Make sure racc is the group
15667 representative, if there is one.
15669 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15671 PR middle-end/82095
15672 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15675 2017-09-06 Richard Biener <rguenther@suse.de>
15677 * gimple-ssa-strength-reduction.c
15678 (find_candidates_dom_walker::before_doom_children): Use a
15679 type and not a mode check.
15681 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15684 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15686 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15688 2017-09-05 Jeff Law <law@redhat.com>
15690 PR tree-optimization/64910
15691 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15692 cases where we have 3 or more operands.
15694 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15696 PR middle-end/81768
15697 * omp-low.c (lower_omp_for): Recompute tree invariant if
15698 gimple_omp_for_initial/final is ADDR_EXPR.
15700 PR middle-end/81768
15701 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15702 into gimple val before gimplification fo the COND_EXPR.
15704 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15706 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15707 REGION_COPY argument.
15708 (thread_through_all_blocks): Remove unused argument to
15709 duplicate_thread_path.
15711 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15712 Alan Hayward <alan.hayward@arm.com>
15713 David Sherwood <david.sherwood@arm.com>
15715 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15716 Take a scalar_mode rather than a machine_mode.
15717 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15718 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15719 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15720 (aarch64_gen_adjusted_ldpstp): Likewise.
15721 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15723 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15724 Alan Hayward <alan.hayward@arm.com>
15725 David Sherwood <david.sherwood@arm.com>
15727 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15728 Take a scalar_int_mode instead of a machine_mode.
15729 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15730 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15731 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15732 (aarch64_simd_attr_length_rglist): Delete.
15733 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15734 a scalar_int_mode instead of a machine_mode.
15735 (aarch64_add_offset): Likewise.
15736 (aarch64_internal_mov_immediate): Likewise
15737 (aarch64_add_constant_internal): Likewise.
15738 (aarch64_add_constant): Likewise.
15739 (aarch64_movw_imm): Likewise.
15740 (aarch64_rtx_arith_op_extract_p): Likewise.
15741 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15742 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15743 Remove assert that the mode isn't a vector.
15744 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15745 (aarch64_expand_mov_immediate): Update calls after above changes.
15746 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15747 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15748 (aarch64_move_imm): Likewise.
15749 (aarch64_can_const_movi_rtx_p): Likewise.
15750 (aarch64_strip_extend): Likewise.
15751 (aarch64_extr_rtx_p): Likewise.
15752 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15753 a CONST_INT when the mode parameter is VOIDmode.
15754 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15756 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15758 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15759 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15760 (bitwise_type_for_mode): Update accordingly.
15762 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15764 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15765 * stor-layout.c (mode_for_size_tree): Likewise.
15766 (mode_for_array): Update accordingly.
15767 (layout_decl): Likewise.
15768 (compute_record_mode): Likewise. Only set the mode once.
15770 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15772 * target.def (get_mask_mode): Change return type to opt_mode.
15774 * doc/tm.texi: Regenerate.
15775 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15776 * targhooks.c (default_get_mask_mode): Likewise.
15777 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15778 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15779 targetm.get_mask_mode.
15780 * tree.c (build_truth_vector_type): Likewise.
15782 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15784 * machmode.h (mode_for_vector): Return an opt_mode.
15785 * stor-layout.c (mode_for_vector): Likewise.
15786 (mode_for_int_vector): Update accordingly.
15787 (layout_type): Likewise.
15788 * config/i386/i386.c (emit_memmov): Likewise.
15789 (ix86_expand_set_or_movmem): Likewise.
15790 (ix86_expand_vector_init): Likewise.
15791 (ix86_get_mask_mode): Likewise.
15792 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15794 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15795 * expmed.c (extract_bit_field_1): Likewise.
15796 * expr.c (expand_expr_real_2): Likewise.
15797 * optabs-query.c (can_vec_perm_p): Likewise.
15798 (can_vec_mask_load_store_p): Likewise.
15799 * optabs.c (expand_vec_perm): Likewise.
15800 * targhooks.c (default_get_mask_mode): Likewise.
15801 * tree-vect-stmts.c (vectorizable_store): Likewise.
15802 (vectorizable_load): Likewise.
15803 (get_vectype_for_scalar_type_and_size): Likewise.
15805 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15807 * machmode.h (mode_for_int_vector): New function.
15808 * stor-layout.c (mode_for_int_vector): Likewise.
15809 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15810 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15811 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15812 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15813 (s390_expand_vcond): Likewise.
15815 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15817 * machmode.h (opt_machine_mode): New type.
15818 (opt_mode<T>): Allow construction from anything that can be
15820 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15821 (mode_for_size): Return an opt_machine_mode.
15822 * stor-layout.c (mode_for_size): Likewise.
15823 (mode_for_size_tree): Update call accordingly.
15824 (bitwise_mode_for_mode): Likewise.
15825 (make_fract_type): Likewise.
15826 (make_accum_type): Likewise.
15827 * caller-save.c (replace_reg_with_saved_mem): Update call
15829 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15830 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15831 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15832 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15833 * expmed.c (extract_bit_field_1): Likewise.
15834 * reload.c (get_secondary_mem): Likewise.
15835 * varasm.c (assemble_integer): Likewise.
15836 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15839 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15841 * machmode.h (decimal_float_mode_for_size): New function.
15842 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15843 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15844 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15845 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15846 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15847 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15849 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15851 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15852 (get_builtin_sync_mode): Likewise.
15853 (expand_ifn_atomic_compare_exchange): Likewise.
15854 (expand_builtin_atomic_clear): Likewise.
15855 (expand_builtin_atomic_test_and_set): Likewise.
15856 (fold_builtin_atomic_always_lock_free): Likewise.
15857 * calls.c (compute_argument_addresses): Likewise.
15858 (emit_library_call_value_1): Likewise.
15859 (store_one_arg): Likewise.
15860 * combine.c (combine_instructions): Likewise.
15861 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15862 * config/arm/arm.c (arm_function_value): Likewise.
15863 (aapcs_allocate_return_reg): Likewise.
15864 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15865 * config/i386/i386.c (construct_container): Likewise.
15866 (ix86_gimplify_va_arg): Likewise.
15867 (ix86_expand_sse_cmp): Likewise.
15868 (emit_memmov): Likewise.
15869 (emit_memset): Likewise.
15870 (expand_small_movmem_or_setmem): Likewise.
15871 (ix86_expand_pextr): Likewise.
15872 (ix86_expand_pinsr): Likewise.
15873 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15874 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15876 * config/mips/mips.c (mips_function_value_1) Likewise.
15877 (mips_block_move_straight): Likewise.
15878 (mips_expand_ins_as_unaligned_store): Likewise.
15879 * config/powerpcspe/powerpcspe.c
15880 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15881 (rs6000_darwin64_record_arg_flush): Likewise.
15882 * config/rs6000/rs6000.c
15883 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15884 (rs6000_darwin64_record_arg_flush): Likewise.
15885 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15886 (sparc_function_value_1): Likewise.
15887 * config/spu/spu.c (adjust_operand): Likewise.
15888 (spu_emit_branch_or_set): Likewise.
15889 (arith_immediate_p): Likewise.
15890 * emit-rtl.c (gen_lowpart_common): Likewise.
15891 * expr.c (expand_expr_real_1): Likewise.
15892 * function.c (assign_parm_setup_block): Likewise.
15893 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15894 * reload1.c (alter_reg): Likewise.
15895 * stor-layout.c (mode_for_vector): Likewise.
15896 (layout_type): Likewise.
15898 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15900 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15901 (spu_convert_move): Likewise.
15902 * lower-subreg.c (resolve_simple_move): Likewise.
15904 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15907 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15908 define_insn to a define_expand.
15909 (altivec_vsum2sws_direct): New define_insn.
15910 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15912 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15914 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15917 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15922 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15923 realign stack if stack alignment needed is less than incoming
15926 2017-09-05 Marek Polacek <polacek@redhat.com>
15929 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15932 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15934 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15936 2017-09-05 Richard Biener <rguenther@suse.de>
15938 PR tree-optimization/82084
15939 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15941 2017-09-05 Richard Biener <rguenther@suse.de>
15943 PR tree-optimization/82102
15944 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15946 2017-09-05 Martin Liska <mliska@suse.cz>
15948 PR tree-optimization/82032
15949 * tree-cfg.c (generate_range_test): New function.
15950 * tree-cfg.h (generate_range_test): Declared here.
15951 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15952 (cleanup_control_expr_graph): Use it.
15953 * tree-switch-conversion.c (try_switch_expansion): Remove
15955 (emit_case_nodes): Use generate_range_test.
15957 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15960 * config/i386/i386.md (*<btsc><mode>_mask): Add
15961 TARGET_USE_BT to insn constraint.
15962 (*btr<mode>_mask): Ditto.
15964 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15966 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15967 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15969 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15972 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15973 TARGET_NEON and TARGET_IWMMXT.
15974 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15975 TARGET_NEON and TARGET_IWMMXT.
15976 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15978 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15980 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15981 (ix86_rewrite_tls_address): Ditto.
15982 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15983 (ix86_rewrite_tls_address_1): Ditto.
15984 (ix86_rewrite_tls_address): Ditto.
15985 * config/i386/predicates.md (tls_address_pattern): New predicate.
15986 * config/i386/i386.md (TLS address splitter): New splitter.
15988 2017-09-04 Richard Biener <rguenther@suse.de>
15990 PR tree-optimization/82084
15991 * fold-const.h (can_native_encode_string_p): Declare.
15992 * fold-const.c (can_native_encode_string_p): Factor out from ...
15993 (native_encode_string): ... here.
15994 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15995 vectorizing stores from constants we later cannot handle.
15997 2017-09-04 Marek Polacek <polacek@redhat.com>
16000 * doc/invoke.texi: Update -Wtautological-compare documentation.
16002 2017-09-04 Jeff Law <law@redhat.com>
16004 PR tree-optimization/64910
16005 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16006 swap the first and last operand if the last is a constant.
16008 2017-09-04 Marek Polacek <polacek@redhat.com>
16011 * convert.c (do_narrow): When sanitizing signed integer overflows,
16012 bail out for signed types.
16013 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16015 2017-09-04 Richard Biener <rguenther@suse.de>
16017 PR tree-optimization/82060
16018 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16019 Move devirtualization after stmt folding and before EH/AB/noreturn
16020 cleanup to get the stmt refs canonicalized. Use a bool instead
16021 of gimple_modified_p since that doesn't work for NOPs. Schedule
16022 NOPs generated by folding for removal.
16024 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16025 Alan Hayward <alan.hayward@arm.com>
16026 David Sherwood <david.sherwood@arm.com>
16028 * coretypes.h (pad_direction): New enum.
16029 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16030 (FUNCTION_ARG_PADDING): Likewise.
16031 * target.def (function_arg_padding): New hook.
16032 * targhooks.h (default_function_arg_padding): Declare.
16033 * targhooks.c (default_function_arg_padding): New function.
16034 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16035 (TARGET_FUNCTION_ARG_PADDING): ...this.
16036 * doc/tm.texi: Regenerate.
16037 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16038 instead of direction.
16039 (compute_argument_addresses): Likewise.
16040 (load_register_parameters): Likewise.
16041 (emit_library_call_value_1): Likewise.
16042 (store_one_arg): Use targetm.calls.function_arg_padding instead
16043 of FUNCTION_ARG_PADDING.
16044 (must_pass_in_stack_var_size_or_pad): Likewise.
16045 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16046 (emit_group_store): Likewise.
16047 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16048 instead of FUNCTION_ARG_PADDING.
16049 (emit_push_insn): Likewise, and propagate enum change throughout
16051 * function.h (direction): Delete.
16052 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16054 * function.c (assign_parm_find_stack_rtl): Likewise.
16055 (assign_parm_setup_block_p): Likewise.
16056 (assign_parm_setup_block): Likewise.
16057 (gimplify_parameters): Likewise.
16058 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16059 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16061 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16062 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16063 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16064 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16065 (aarch64_function_arg_padding): ...this new function.
16066 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16067 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16068 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16069 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16070 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16071 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16072 (arm_pad_arg_upward): Replace with...
16073 (arm_function_arg_padding): ...this new function.
16074 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16076 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16077 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16078 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16079 (ia64_hpux_function_arg_padding): Replace with...
16080 (ia64_function_arg_padding): ...this new function. Use pad_direction
16081 instead of direction. Check for TARGET_HPUX.
16082 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16083 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16084 (iq2000_function_arg_padding): New function.
16085 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16086 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16087 (mips_function_arg_padding): ...this new function.
16088 (mips_pad_reg_upward): Update accordingly.
16089 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16090 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16091 targetm.calls.function_arg_padding.
16092 (FUNCTION_ARG_PADDING): Delete.
16093 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16094 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16095 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16096 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16097 (nios2_block_reg_padding): Return pad_direction instead of direction.
16098 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16099 instead of direction.
16100 (nios2_function_arg_padding): Likewise. Make static.
16101 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16102 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16103 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16104 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16105 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16106 (pa_function_arg_padding): Make static. Return pad_direction instead
16108 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16109 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16110 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16111 instead of direction. Use targetm.calls.function_arg_padding.
16112 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16113 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16114 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16115 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16116 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16118 (function_arg_padding): Rename to...
16119 (rs6000_function_arg_padding): ...this. Make static. Return
16120 pad_direction instead of direction.
16121 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16122 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16123 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16124 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16125 instead of direction. Use targetm.calls.function_arg_padding.
16126 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16127 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16128 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16129 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16130 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16131 (function_arg_padding): Rename to...
16132 (rs6000_function_arg_padding): ...this. Make static. Return
16133 pad_direction instead of direction.
16134 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16135 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16136 * config/s390/s390.c (s390_function_arg_padding): New function.
16137 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16138 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16139 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16140 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16141 (function_arg_padding): Rename to...
16142 (sparc_function_arg_padding): ...this. Make static. Return
16143 pad_direction instead of direction.
16144 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16145 * config/spu/spu.c (spu_function_arg_padding): New function.
16146 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16147 * system.h (FUNCTION_ARG_PADDING): Poison.
16149 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16150 Alan Hayward <alan.hayward@arm.com>
16151 David Sherwood <david.sherwood@arm.com>
16153 * target.def (modes_tieable_p): New hook.
16154 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16155 (TARGET_MODES_TIEABLE_P): ...this.
16156 * doc/tm.texi.in: Regenerate.
16157 * hooks.h (hook_bool_mode_mode_true): Declare.
16158 * hooks.c (hook_bool_mode_mode_true): New function.
16159 * combine.c (subst): Use targetm.modes_tieable_p instead of
16161 * dse.c (find_shift_sequence): Likewise.
16162 * expmed.c (extract_low_bits): Likewise.
16163 * lower-subreg.c: Include target.h.
16164 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16166 * rtlanal.c (rtx_cost): Likewise.
16167 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16168 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16169 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16170 (TARGET_MODES_TIEABLE_P): Redefine.
16171 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16172 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16173 (TARGET_MODES_TIEABLE_P): Redefine.
16174 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16175 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16176 (arc_modes_tieable_p): New function.
16177 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16178 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16179 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16180 (arm_modes_tieable_p): Make static.
16181 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16182 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16183 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16184 (TARGET_MODES_TIEABLE_P): Redefine.
16185 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16186 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16187 (TARGET_MODES_TIEABLE_P): Redefine.
16188 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16189 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16190 (cr16_modes_tieable_p): New function.
16191 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16192 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16193 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16194 (TRULY_NOOP_TRUNCATION): Update comment.
16195 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16196 (TRULY_NOOP_TRUNCATION): Update comment.
16197 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16198 (frv_modes_tieable_p): New function.
16199 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16200 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16201 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16202 (TARGET_MODES_TIEABLE_P): Redefine.
16203 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16204 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16205 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16206 (TARGET_MODES_TIEABLE_P): Redefine.
16207 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16208 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16209 (ia64_modes_tieable_p): New function.
16210 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16211 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16212 (iq2000_modes_tieable_p): New function.
16213 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16214 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16215 (lm32_modes_tieable_p): New function.
16216 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16217 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16218 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16219 (TARGET_MODES_TIEABLE_P): Redefine.
16220 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16221 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16222 (m32r_modes_tieable_p): New function.
16223 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16224 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16225 (m68k_modes_tieable_p): New function.
16226 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16227 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16228 (mcore_modes_tieable_p): New function.
16229 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16230 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16232 (TARGET_MODES_TIEABLE_P): Redefine.
16233 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16234 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16235 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16236 (TARGET_MODES_TIEABLE_P): Redefine.
16237 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16238 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16239 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16240 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16241 (mn10300_modes_tieable_p): ...this and make static.
16242 (TARGET_MODES_TIEABLE_P): Redefine.
16243 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16244 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16245 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16246 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16247 (msp430_modes_tieable_p): Make static.
16248 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16249 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16250 (TARGET_MODES_TIEABLE_P): Redefine.
16251 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16252 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16253 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16254 (TARGET_MODES_TIEABLE_P): Redefine.
16255 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16256 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16257 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16258 (TARGET_MODES_TIEABLE_P): Redefine.
16259 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16260 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16261 (pdp11_modes_tieable_p): New function.
16262 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16263 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16264 (rs6000_modes_tieable_p): New function.
16265 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16266 * config/powerpcspe/powerpcspe.md: Update comment.
16267 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16268 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16269 (TARGET_MODES_TIEABLE_P): Redefine.
16270 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16271 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16272 (rl78_modes_tieable_p): New function.
16273 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16274 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16275 (rs6000_modes_tieable_p): New function.
16276 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16277 * config/rs6000/rs6000.md: Update comment.
16278 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16279 * config/rx/rx.c (rx_modes_tieable_p): New function.
16280 (TARGET_MODES_TIEABLE_P): Redefine.
16281 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16282 * config/s390/s390.c (s390_modes_tieable_p): New function.
16283 (TARGET_MODES_TIEABLE_P): Redefine.
16284 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16285 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16286 (sh_modes_tieable_p): New function.
16287 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16288 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16289 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16290 (sparc_modes_tieable_p): Make static.
16291 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16292 * config/spu/spu.c (spu_modes_tieable_p): New function.
16293 (TARGET_MODES_TIEABLE_P): Redefine.
16294 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16295 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16296 (TARGET_MODES_TIEABLE_P): Redefine.
16297 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16298 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16299 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16300 * config/v850/v850.c (v850_modes_tieable_p): New function.
16301 (TARGET_MODES_TIEABLE_P): Redefine.
16302 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16303 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16304 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16305 (visium_modes_tieable_p): New function.
16306 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16307 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16308 (xtensa_modes_tieable_p): New function.
16309 * system.h (MODES_TIEABLE_P): Poison.
16311 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16312 Alan Hayward <alan.hayward@arm.com>
16313 David Sherwood <david.sherwood@arm.com>
16315 * target.def (hard_regno_mode_ok): New hook.
16316 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16317 (TARGET_HARD_REGNO_MODE_OK): ...this.
16318 * doc/tm.texi.in: Regenerate.
16319 * hooks.h (hook_bool_uint_mode_true): Declare.
16320 * hooks.c (hook_bool_uint_mode_true): New function.
16321 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16322 HARD_REGNO_MODE_OK.
16323 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16324 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16325 instead of HARD_REGNO_MODE_OK.
16326 * caller-save.c: Include target.h.
16327 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16328 HARD_REGNO_MODE_OK.
16329 * combine.c (can_combine_p): Likewise.
16330 (combinable_i3pat): Likewise.
16331 (can_change_dest_mode): Likewise.
16332 * expr.c (init_expr_target): Likewise.
16333 (convert_move): Likewise.
16334 (convert_modes): Likewise.
16335 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16336 (setup_prohibited_mode_move_regs): Likewise.
16337 * ira.h (target_ira): Likewise.
16338 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16339 * lra-constraints.c (process_alt_operands): Likewise.
16340 (split_reg): Likewise.
16341 * recog.c (peep2_find_free_register): Likewise.
16342 * ree.c (combine_reaching_defs): Likewise.
16343 * regcprop.c (maybe_mode_change): Likewise.
16344 * reginfo.c (init_reg_sets_1): Likewise.
16345 (choose_hard_reg_mode): Likewise.
16346 (simplifiable_subregs): Likewise.
16347 * regrename.c (check_new_reg_p): Likewise.
16348 * reload.c (find_valid_class): Likewise.
16349 (find_valid_class_1): Likewise.
16350 (reload_inner_reg_of_subreg): Likewise.
16351 (push_reload): Likewise.
16352 (combine_reloads): Likewise.
16353 (find_dummy_reload): Likewise.
16354 (find_reloads): Likewise.
16355 * reload1.c (find_reg): Likewise.
16356 (set_reload_reg): Likewise.
16357 (allocate_reload_reg): Likewise.
16358 (choose_reload_regs): Likewise.
16359 (reload_adjust_reg_for_temp): Likewise.
16360 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16361 (simplify_subreg_regno): Likewise.
16362 * sel-sched.c (init_regs_for_mode): Likewise.
16363 * varasm.c (make_decl_rtl): Likewise.
16364 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16365 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16366 HARD_REGNO_MODE_OK.
16367 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16368 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16369 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16370 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16371 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16372 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16373 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16374 (arc_mode_class): Delete.
16375 (HARD_REGNO_MODE_OK): Delete.
16376 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16377 (arc_hard_regno_mode_ok): Rename old array to...
16378 (arc_hard_regno_mode_ok_modes): ...this.
16379 (arc_conditional_register_usage): Update accordingly.
16380 (arc_mode_class): Make static.
16381 (arc_hard_regno_mode_ok): New function.
16382 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16383 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16384 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16385 (arm_hard_regno_mode_ok): Make static.
16386 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16387 HARD_REGNO_MODE_OK.
16388 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16389 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16390 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16392 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16393 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16394 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16395 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16396 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16397 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16398 * config/bfin/predicates.md (valid_reg_operand): Use
16399 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16400 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16401 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16402 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16403 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16404 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16405 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16406 (cr16_hard_regno_mode_ok): Make static and return a bool.
16407 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16408 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16409 (cris_hard_regno_mode_ok): New function.
16410 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16411 (epiphany_mode_class): Delete.
16412 (HARD_REGNO_MODE_OK): Delete.
16413 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16414 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16415 (hard_regno_mode_ok): Rename to...
16416 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16417 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16418 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16419 HARD_REGNO_MODE_OK.
16420 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16421 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16422 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16423 (frv_hard_regno_mode_ok): Make static and return a bool.
16424 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16425 HARD_REGNO_MODE_OK.
16426 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16427 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16428 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16429 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16431 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16432 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16433 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16434 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16436 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16437 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16438 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16439 (ia64_hard_regno_mode_ok): New function.
16440 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16441 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16442 (iq2000_hard_regno_mode_ok): New function.
16443 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16444 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16445 (lm32_hard_regno_mode_ok): New function.
16446 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16447 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16448 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16449 instead of HARD_REGNO_MODE_OK.
16450 (m32c_hard_regno_ok): Rename to...
16451 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16452 (m32c_cannot_change_mode_class): Update accordingly.
16453 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16454 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16455 (m32r_mode_class): Delete.
16456 (HARD_REGNO_MODE_OK): Delete.
16457 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16458 (m32r_hard_regno_mode_ok): Rename to...
16459 (m32r_hard_regno_modes): ...this.
16460 (m32r_mode_class): Make static.
16461 (m32r_hard_regno_mode_ok): New function.
16462 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16463 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16464 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16465 (m68k_hard_regno_mode_ok): Make static.
16466 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16467 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16468 (mcore_hard_regno_mode_ok): New function.
16469 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16470 (HARD_REGNO_MODE_OK): Delete.
16471 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16473 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16474 (microblaze_hard_regno_mode_ok): New function.
16475 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16476 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16477 (mips_hard_regno_mode_ok): Delete.
16478 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16479 (mips_hard_regno_mode_ok_p): ...this and make static.
16480 (mips_hard_regno_mode_ok_p): Rename to...
16481 (mips_hard_regno_mode_ok_uncached): ...this.
16482 (mips_hard_regno_mode_ok): New function.
16483 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16484 of HARD_REGNO_MODE_OK.
16485 (mips_option_override): Update after above name changes.
16486 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16487 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16488 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16489 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16490 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16491 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16492 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16493 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16494 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16495 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16496 (msp430_hard_regno_mode_ok): Make static and return a bool.
16497 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16498 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16499 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16501 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16502 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16503 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16504 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16505 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16506 (PA_HARD_REGNO_MODE_OK): ...this
16507 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16508 (PA_HARD_REGNO_MODE_OK): ...this.
16509 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16510 (pa_hard_regno_mode_ok): New function.
16511 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16512 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16513 (pdp11_hard_regno_mode_ok): New function.
16514 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16515 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16517 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16519 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16520 (rs6000_hard_regno_mode_ok): Rename to...
16521 (rs6000_hard_regno_mode_ok_uncached): ...this.
16522 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16523 (rs6000_hard_regno_mode_ok): New function.
16524 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16525 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16526 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16527 (riscv_hard_regno_mode_ok): ...this and make static.
16528 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16529 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16530 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16531 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16532 (rl78_hard_regno_mode_ok): Make static and return bool.
16533 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16534 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16536 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16537 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16538 (rs6000_hard_regno_mode_ok): Rename to...
16539 (rs6000_hard_regno_mode_ok_uncached): ...this.
16540 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16541 (rs6000_hard_regno_mode_ok): New function.
16542 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16543 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16544 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16545 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16546 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16547 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16548 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16549 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16550 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16551 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16552 (sh_hard_regno_mode_ok): Make static.
16553 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16554 instead of HARD_REGNO_MODE_OK.
16555 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16556 (sparc_mode_class): Delete.
16557 (HARD_REGNO_MODE_OK): Delete.
16558 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16559 (hard_regno_mode_classes): Make static.
16560 (sparc_mode_class): Likewise.
16561 (sparc_hard_regno_mode_ok): New function.
16562 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16563 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16564 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16566 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16567 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16568 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16569 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16570 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16571 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16572 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16573 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16574 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16575 (visium_hard_regno_mode_ok): New function.
16576 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16577 instead of HARD_REGNO_MODE_OK.
16578 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16579 (HARD_REGNO_MODE_OK): Delete.
16580 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16581 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16582 (xtensa_option_override): Update accordingly.
16583 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16584 (xtensa_hard_regno_mode_ok): New function.
16585 * system.h (HARD_REGNO_MODE_OK): Poison.
16587 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16588 Alan Hayward <alan.hayward@arm.com>
16589 David Sherwood <david.sherwood@arm.com>
16591 * target.def (hard_regno_call_part_clobbered): New hook.
16592 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16593 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16594 * doc/tm.texi: Regenerate.
16595 * hooks.h (hook_bool_uint_mode_false): Declare.
16596 * hooks.c (hook_bool_uint_mode_false): New function.
16597 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16598 * cselib.c (cselib_process_insn): Use
16599 targetm.hard_regno_call_part_clobbered instead of
16600 HARD_REGNO_CALL_PART_CLOBBERED.
16601 * ira-conflicts.c (ira_build_conflicts): Likewise.
16602 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16603 * lra-constraints.c (need_for_call_save_p): Likewise.
16604 * lra-lives.c: Include target.h.
16605 (check_pseudos_live_through_calls): Use
16606 targetm.hard_regno_call_part_clobbered instead of
16607 HARD_REGNO_CALL_PART_CLOBBERED.
16608 * regcprop.c: Include target.h.
16609 (copyprop_hardreg_forward_1): Use
16610 targetm.hard_regno_call_part_clobbered instead of
16611 HARD_REGNO_CALL_PART_CLOBBERED.
16612 * reginfo.c (choose_hard_reg_mode): Likewise.
16613 * regrename.c (check_new_reg_p): Likewise.
16614 * reload.c (find_equiv_reg): Likewise.
16615 * reload1.c (emit_reload_insns): Likewise.
16616 * sched-deps.c (deps_analyze_insn): Likewise.
16617 * sel-sched.c (init_regs_for_mode): Likewise.
16618 (mark_unavailable_hard_regs): Likewise.
16619 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16620 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16621 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16623 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16624 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16625 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16627 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16629 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16630 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16631 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16633 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16634 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16635 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16637 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16638 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16640 * config/powerpcspe/powerpcspe.c
16641 (rs6000_hard_regno_call_part_clobbered): New function.
16642 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16643 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16644 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16646 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16647 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16648 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16650 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16651 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16652 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16654 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16655 Alan Hayward <alan.hayward@arm.com>
16656 David Sherwood <david.sherwood@arm.com>
16658 * rtl.h (subreg_memory_offset): Declare.
16659 * emit-rtl.c (subreg_memory_offset): New function.
16660 * expmed.c (store_bit_field_1): Use it.
16661 * expr.c (undefined_operand_subword_p): Likewise.
16662 * simplify-rtx.c (simplify_subreg): Likewise.
16664 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16666 PR rtl-optimization/57448
16669 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16670 using atomic_load pattern.
16671 (expand_atomic_store): Likewise.
16673 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16676 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16677 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16678 handling. Use replace_call_with_value with NULL instead of
16679 gsi_replace, unlink_stmt_vdef and release_defs.
16681 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16684 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16686 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16689 * rtl.h (emit_library_call_value_1): Declare.
16690 (emit_library_call): Replace declaration with a series of overloads.
16691 Remove the parameter count argument.
16692 (emit_library_call_value): Likewise.
16693 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16694 with an "rtx_mode_t *".
16695 (emit_library_call_value): Delete.
16696 (emit_library_call): Likewise.
16697 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16698 (asan_emit_allocas_unpoison): Likewise.
16699 * builtins.c (expand_builtin_powi): Likewise.
16700 (expand_asan_emit_allocas_unpoison): Likewise.
16701 * cfgexpand.c (expand_main_function): Likewise.
16702 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16703 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16704 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16705 * config/arm/arm.c (arm_trampoline_init): Likewise.
16706 (arm_call_tls_get_addr): Likewise.
16707 (arm_expand_divmod_libfunc): Likewise.
16708 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16709 (smulsi3_highpart): Likewise.
16710 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16711 (c6x_expand_compare): Likewise.
16712 (c6x_expand_movmem): Likewise.
16713 * config/frv/frv.c (frv_trampoline_init): Likewise.
16714 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16715 (ix86_expand_divmod_libfunc): Likewise.
16716 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16717 (ia64_expand_compare): Likewise.
16718 (ia64_profile_hook): Likewise.
16719 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16720 (nonlocal_goto): Likewise.
16721 (restore_stack_nonlocal): Likewise.
16722 * config/m32r/m32r.c (block_move_call): Likewise.
16723 (m32r_trampoline_init): Likewise.
16724 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16725 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16726 (m68k_call_m68k_read_tp): Likewise.
16727 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16728 (microblaze_expand_divide): Likewise.
16729 * config/mips/mips.h (mips_args): Likewise.
16730 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16731 (MIPS_ICACHE_SYNC): Likewise.
16732 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16733 (nios2_trampoline_init): Likewise.
16734 * config/pa/pa.c (hppa_tls_call): Likewise.
16735 (pa_trampoline_init): Likewise.
16736 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16737 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16738 (expand_strn_compare): Likewise.
16739 (rs6000_generate_compare): Likewise.
16740 (rs6000_expand_float128_convert): Likewise.
16741 (output_profile_hook): Likewise.
16742 (rs6000_trampoline_init): Likewise.
16743 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16744 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16745 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16746 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16747 (rs6000_generate_compare): Likewise.
16748 (rs6000_expand_float128_convert): Likewise.
16749 (output_profile_hook): Likewise.
16750 (rs6000_trampoline_init): Likewise.
16751 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16752 * config/sh/sh.c (sh_trampoline_init): Likewise.
16753 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16754 (sparc_emit_float_lib_cmp): Likewise.
16755 (sparc32_initialize_trampoline): Likewise.
16756 (sparc64_initialize_trampoline): Likewise.
16757 (sparc_profile_hook): Likewise.
16758 * config/spu/spu.c (ea_load_store): Likewise.
16759 * config/spu/spu.md (floatunssidf2): Likewise.
16760 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16761 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16762 * config/visium/visium.c (expand_block_move_4): Likewise.
16763 (expand_block_move_2): Likewise.
16764 (expand_block_move_1): Likewise.
16765 (expand_block_set_4): Likewise.
16766 (expand_block_set_2): Likewise.
16767 (expand_block_set_1): Likewise.
16768 (visium_trampoline_init): Likewise.
16769 (visium_profile_hook): Likewise.
16770 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16771 (xtensa_setup_frame_addresses): Likewise.
16772 (xtensa_trampoline_init): Likewise.
16773 * except.c (sjlj_emit_function_enter): Likewise.
16774 (sjlj_emit_function_exit): Likewise.
16775 * explow.c (allocate_dynamic_stack_space): Likewise.
16776 (probe_stack_range): Likewise.
16777 * expr.c (convert_mode_scalar): Likewise.
16778 * optabs.c (expand_binop): Likewise.
16779 (expand_twoval_binop_libfunc): Likewise.
16780 (expand_unop): Likewise.
16781 (prepare_cmp_insn): Likewise.
16782 (prepare_float_lib_cmp): Likewise.
16783 (expand_float): Likewise.
16784 (expand_fix): Likewise.
16785 (expand_fixed_convert): Likewise.
16786 (maybe_emit_sync_lock_test_and_set): Likewise.
16787 (expand_atomic_compare_and_swap): Likewise.
16788 (expand_mem_thread_fence): Likewise.
16789 (expand_atomic_fetch_op): Likewise.
16791 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16793 * doc/generic.texi (OpenACC): Adjust URL.
16794 * doc/invoke.texi (C Dialect Options): Ditto.
16796 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16798 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16799 predicate for operand 1. Add (m,<S>) constraint.
16800 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16801 Prevent memory operand 1 with register operand 2.
16803 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16805 PR rtl-optimization/82024
16806 * combine.c (try_combine): If the combination result is a PARALLEL,
16807 and we only need to retain the SET in there that would be placed
16808 at I2, check that we can place that at I3 instead, before doing so.
16810 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16813 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16815 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16816 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16819 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16820 Jeff Law <law@redhat.com>
16822 * varasm.c (bss_initializer_p): Do not put constants into .bss
16823 (categorize_decl_for_section): Handle bss_initializer_p returning
16824 false when DECL_INITIAL is NULL.
16826 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16829 * config/s390/s390.c (s390_can_inline_p): New function.
16831 2017-09-01 Jeff Law <law@redhat.com>
16833 PR tree-optimization/82052
16834 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16835 Always initialize the returned slot after a hash table miss
16836 when INSERT is true.
16838 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16840 * config/s390/s390.md (mem_signal_fence): Remove.
16841 * doc/md.texi (mem_signal_fence): Remove.
16842 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16844 * target-insns.def (mem_signal_fence): Remove.
16846 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16849 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16852 * asan.c (create_odr_indicator): Strip name encoding from assembler
16853 name before appending it after __odr_asan_.
16855 2017-09-01 Martin Liska <mliska@suse.cz>
16857 PR tree-optimization/82059
16858 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16859 frequency only when an edge is redirected.
16861 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16863 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16864 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16865 (arc_conditional_register_usage): Remove ARC600 lp_count
16867 (arc_file_start): Emit Tag_ARC_CPU_variation.
16868 (arc_can_use_doloop_p): New conditions to use ZOLs.
16869 (hwloop_fail): New function.
16870 (hwloop_optimize): Likewise.
16871 (hwloop_pattern_reg): Likewise.
16872 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16873 (arc_reorg_loops): New function, calls reorg_loops.
16874 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16875 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16877 (arc_loop_hazard): Remove function, functionality moved into
16879 (arc_hazard): Remove arc_loop_hazard call.
16880 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16881 into hwloop_optimize.
16882 (arc_label_align): Remove ZOL handling.
16883 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16884 * config/arc/arc.md (doloop_begin): Remove pattern.
16885 (doloop_begin_i): Likewise.
16886 (doloop_end_i): Likewise.
16887 (doloop_fallback): Likewise.
16888 (doloop_fallback_m): Likewise.
16889 (doloop_end): Reimplement expand.
16890 (arc_lp): New pattern for LP instruction.
16891 (loop_end): New pattern.
16892 (loop_fail): Likewise.
16893 (decrement_and_branch_until_zero): Likewise.
16894 * config/arc/arc.opt (mlpc-width): New option.
16895 * doc/invoke.texi (mlpc-width): Document option.
16897 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16899 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16900 (arc_ccfsm_advance): Fix checking for delay slots.
16901 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16903 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16905 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16907 (movhi_insn): Update store instruction constraint which are saving
16908 6-bit short immediates.
16909 (movsi_insn): Consider also short scaled load operations.
16910 (zero_extendhisi2_i): Use Usd constraint instead of T.
16911 (extendhisi2_i): Add q constraint.
16912 (arc_clzsi2): Add type and length attributes.
16913 (arc_ctzsi2): Likewise.
16914 * config/arc/constraints.md (Usc): Update constraint, the
16915 assembler can parse two relocations for a single instruction.
16917 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16919 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16920 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16922 2017-08-31 Olivier Hainque <hainque@adacore.com>
16924 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16925 match as powerpc-wrs-vxworks*.
16927 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16929 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16930 register constraint for by-element operand.
16931 (aarch64_mls_elt_merge<mode>): Likewise.
16933 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16935 * config/arc/arc.c (arc_can_follow_jump): Check for short
16938 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16940 * config.gcc: Use g.opt for arc.
16941 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16942 functionality moved to ...
16943 (legitimate_scaled_address_p): New function, ...here.
16944 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16945 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16946 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16948 (arc_override_options): Handle G option.
16949 (arc_output_pic_addr_const): Correct function definition.
16950 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16951 (arc_decl_anon_ns_mem_p): Delete.
16952 (arc_in_small_data_p): Overhaul this function to take into
16953 consideration the value given via G option.
16954 (arc_rewrite_small_data_1): Renamed and corrected old
16955 arc_rewrite_small_data function.
16956 (arc_rewrite_small_data): New function.
16957 (small_data_pattern): Don't use pic_offset_table_rtx.
16958 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16959 * config/arc/simdext.md (movmisalignv2hi): Use
16960 prepare_move_operands function.
16962 (movmisalign*): Likewise.
16963 * doc/invoke.texi (ARC options): Document -G option.
16965 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16967 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16969 * config/arc/arc.c (arc_print_operand): Output scalled address for
16970 sdata whenever is possible.
16971 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16972 load/stores are available.
16973 (compact_sda_memory_operand): Check for the alignment required by
16974 code density instructions.
16975 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16977 * config/arc/constraints.md (Usd): Update constraint.
16978 (Us0): New constraint.
16979 (Usc): Update constraint.
16981 2017-08-31 Richard Biener <rguenther@suse.de>
16983 PR middle-end/82054
16984 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16985 function only once.
16987 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16989 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16990 Resize type_signature.
16992 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16993 Alan Hayward <alan.hayward@arm.com>
16994 David Sherwood <david.sherwood@arm.com>
16996 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16997 subregs whose inner modes can be stored in GPRs.
16998 (aarch64_classify_index): Likewise.
17000 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17001 Alan Hayward <alan.hayward@arm.com>
17002 David Sherwood <david.sherwood@arm.com>
17004 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17005 (V_INT_EQUIV): ...this.
17006 (v_cmp_result): Rename to...
17007 (v_int_equiv): ...this.
17008 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17009 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17010 (copysign<mode>3): Likewise.
17011 (aarch64_simd_bsl<mode>_internal): Likewise.
17012 (aarch64_simd_bsl<mode>): Likewise.
17013 (vec_cmp<mode><mode>): Likewise.
17014 (vcond<mode><mode>): Likewise.
17015 (vcond<v_cmp_mixed><mode>): Likewise.
17016 (vcondu<mode><v_cmp_mixed>): Likewise.
17017 (aarch64_cm<optab><mode>): Likewise.
17018 (aarch64_cmtst<mode>): Likewise.
17019 (aarch64_fac<optab><mode>): Likewise.
17020 (vec_perm_const<mode>): Likewise.
17021 (vcond_mask_<mode><v_cmp_result>): Rename to...
17022 (vcond_mask_<mode><v_int_equiv>): ...this.
17023 (vec_cmp<mode><v_cmp_result>): Rename to...
17024 (vec_cmp<mode><v_int_equiv>): ...this.
17026 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17027 Alan Hayward <alan.hayward@arm.com>
17028 David Sherwood <david.sherwood@arm.com>
17030 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17032 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17033 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17034 (UNSPEC_LD4_DREG): New unspecs.
17035 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17036 (aarch64_ld2<mode>_dreg_be): Replace with...
17037 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17039 (aarch64_ld3<mode>_dreg_le)
17040 (aarch64_ld3<mode>_dreg_be): Replace with...
17041 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17043 (aarch64_ld4<mode>_dreg_le)
17044 (aarch64_ld4<mode>_dreg_be): Replace with...
17045 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17048 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17050 PR tree-optimization/81987
17051 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17052 insert an initializer in a location not dominated by the stride
17055 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17057 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17058 on the entire header of the finally block in the fallthru case.
17060 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17062 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17064 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17066 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17067 rs6000_emit_move_from_cr and call renamed function.
17068 (rs6000_emit_prologue): Call renamed functions.
17069 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17070 movesi_from_cr, remove volatile CRs.
17072 2017-08-30 Jon Beniston <jon@beniston.com>
17073 Richard Biener <rguenther@suse.de>
17075 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17076 of VECTOR_MODE_P check.
17077 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17078 element vector types.
17080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17082 * df.h (df_read_modify_subreg_p): Remove in favor of...
17083 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17084 const_rtx instead of an rtx.
17085 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17086 * df-problems.c (df_word_lr_mark_ref): Likewise.
17087 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17088 (mark_pseudo_reg_dead): Likewise.
17089 (mark_ref_dead): Likewise.
17090 * reginfo.c (init_subregs_of_mode): Likewise.
17091 * sched-deps.c (sched_analyze_1): Likewise.
17092 * df-scan.c (df_def_record_1): Likewise.
17093 (df_uses_record): Likewise.
17094 (df_read_modify_subreg_p): Remove in favor of...
17095 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17096 const_rtx instead of an rtx.
17098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17099 Alan Hayward <alan.hayward@arm.com>
17100 David Sherwood <david.sherwood@arm.com>
17102 * rtl.h (partial_subreg_p): New function.
17103 * caller-save.c (save_call_clobbered_regs): Use it.
17104 * calls.c (expand_call): Likewise.
17105 * combine.c (combinable_i3pat): Likewise.
17106 (simplify_set): Likewise.
17107 (make_extraction): Likewise.
17108 (make_compound_operation_int): Likewise.
17109 (gen_lowpart_or_truncate): Likewise.
17110 (force_to_mode): Likewise.
17111 (make_field_assignment): Likewise.
17112 (reg_truncated_to_mode): Likewise.
17113 (record_truncated_value): Likewise.
17114 (move_deaths): Likewise.
17115 * cse.c (record_jump_cond): Likewise.
17116 (cse_insn): Likewise.
17117 * cselib.c (cselib_lookup_1): Likewise.
17118 * expmed.c (extract_bit_field_using_extv): Likewise.
17119 * function.c (assign_parm_setup_reg): Likewise.
17120 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17121 * ira-build.c (create_insn_allocnos): Likewise.
17122 * lra-coalesce.c (merge_pseudos): Likewise.
17123 * lra-constraints.c (match_reload): Likewise.
17124 (simplify_operand_subreg): Likewise.
17125 (curr_insn_transform): Likewise.
17126 * lra-lives.c (process_bb_lives): Likewise.
17127 * lra.c (new_insn_reg): Likewise.
17128 (lra_substitute_pseudo): Likewise.
17129 * regcprop.c (mode_change_ok): Likewise.
17130 (maybe_mode_change): Likewise.
17131 (copyprop_hardreg_forward_1): Likewise.
17132 * reload.c (push_reload): Likewise.
17133 (find_reloads): Likewise.
17134 (find_reloads_subreg_address): Likewise.
17135 * reload1.c (alter_reg): Likewise.
17136 (eliminate_regs_1): Likewise.
17137 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17139 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17141 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17142 back to if statements, including unpack.
17144 2017-08-30 Martin Liska <mliska@suse.cz>
17146 PR inline-asm/82001
17147 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17149 (func_checker::compare_asm_inputs_outputs): ... this function.
17150 (func_checker::compare_gimple_asm): Use the function to compare
17151 also ASM constrains.
17152 * ipa-icf-gimple.h: Rename the function.
17154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17155 Alan Hayward <alan.hayward@arm.com>
17156 David Sherwood <david.sherwood@arm.com>
17158 * coretypes.h (complex_mode): New type.
17159 * gdbhooks.py (build_pretty_printer): Handle it.
17160 * machmode.h (complex_mode): New class.
17161 (complex_mode::includes_p): New function.
17162 (is_complex_int_mode): Likewise.
17163 (is_complex_float_mode): Likewise.
17164 * genmodes.c (get_mode_class): Handle complex mode classes.
17165 * function.c (expand_function_end): Use is_complex_int_mode.
17167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17168 Alan Hayward <alan.hayward@arm.com>
17169 David Sherwood <david.sherwood@arm.com>
17171 * coretypes.h (scalar_mode_pod): New typedef.
17172 * gdbhooks.py (build_pretty_printer): Handle it.
17173 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17174 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17175 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17176 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17177 as_a <scalar_mode>.
17179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17180 Alan Hayward <alan.hayward@arm.com>
17181 David Sherwood <david.sherwood@arm.com>
17183 * machmode.h (mode_for_vector): Take a scalar_mode instead
17185 * stor-layout.c (mode_for_vector): Likewise.
17186 * explow.c (promote_mode): Use as_a <scalar_mode>.
17187 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17190 Alan Hayward <alan.hayward@arm.com>
17191 David Sherwood <david.sherwood@arm.com>
17193 * target.def (preferred_simd_mode): Take a scalar_mode
17194 instead of a machine_mode.
17195 * targhooks.h (default_preferred_simd_mode): Likewise.
17196 * targhooks.c (default_preferred_simd_mode): Likewise.
17197 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17198 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17199 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17200 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17201 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17202 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17203 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17204 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17206 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17207 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17208 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17209 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17210 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17211 * doc/tm.texi: Regenerate.
17212 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17216 Alan Hayward <alan.hayward@arm.com>
17217 David Sherwood <david.sherwood@arm.com>
17219 * target.def (scalar_mode_supported_p): Take a scalar_mode
17220 instead of a machine_mode.
17221 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17222 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17223 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17224 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17225 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17226 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17227 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17228 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17229 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17230 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17231 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17232 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17233 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17234 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17236 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17237 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17238 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17239 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17240 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17242 * doc/tm.texi: Regenerate.
17244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17245 Alan Hayward <alan.hayward@arm.com>
17246 David Sherwood <david.sherwood@arm.com>
17248 * coretypes.h (opt_scalar_mode): New typedef.
17249 * gdbhooks.py (build_pretty_printers): Handle it.
17250 * machmode.h (mode_iterator::get_2xwider): Add overload for
17252 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17254 * expr.c (convert_mode_scalar): Likewise.
17255 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17256 * optabs.c (expand_float): Likewise.
17257 (expand_fix): Likewise.
17258 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17261 Alan Hayward <alan.hayward@arm.com>
17262 David Sherwood <david.sherwood@arm.com>
17264 * optabs.c (expand_float): Explicitly check for scalars before
17265 using a branching expansion.
17266 (expand_fix): Likewise.
17268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17269 Alan Hayward <alan.hayward@arm.com>
17270 David Sherwood <david.sherwood@arm.com>
17272 * expr.c (convert_mode): Split scalar handling out into...
17273 (convert_mode_scalar): ...this new function. Treat the modes
17276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17277 Alan Hayward <alan.hayward@arm.com>
17278 David Sherwood <david.sherwood@arm.com>
17280 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17282 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17285 Alan Hayward <alan.hayward@arm.com>
17286 David Sherwood <david.sherwood@arm.com>
17288 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17289 rather than a machine_mode.
17290 (fixed_from_string): Likewise.
17291 (fixed_convert): Likewise.
17292 (fixed_convert_from_int): Likewise.
17293 (fixed_convert_from_real): Likewise.
17294 (real_convert_from_fixed): Likewise.
17295 * fixed-value.c (fixed_from_double_int): Likewise.
17296 (fixed_from_string): Likewise.
17297 (fixed_convert): Likewise.
17298 (fixed_convert_from_int): Likewise.
17299 (fixed_convert_from_real): Likewise.
17300 (real_convert_from_fixed): Likewise.
17301 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17304 Alan Hayward <alan.hayward@arm.com>
17305 David Sherwood <david.sherwood@arm.com>
17307 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17308 of separate mode class checks. Do not allow vector modes here.
17309 (immed_wide_int_const): Use as_a <scalar_mode>.
17310 * explow.c (trunc_int_for_mode): Likewise.
17311 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17312 (wi::shwi): Likewise.
17313 (wi::min_value): Likewise.
17314 (wi::max_value): Likewise.
17315 * dwarf2out.c (loc_descriptor): Likewise.
17316 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17317 for CONST_WIDE_INT.
17319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17320 Alan Hayward <alan.hayward@arm.com>
17321 David Sherwood <david.sherwood@arm.com>
17323 * tree.h (SCALAR_TYPE_MODE): New macro.
17324 * expr.c (expand_expr_addr_expr_1): Use it.
17325 (expand_expr_real_2): Likewise.
17326 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17327 (fold_convert_const_fixed_from_int): Likewise.
17328 (fold_convert_const_fixed_from_real): Likewise.
17329 (native_encode_fixed): Likewise
17330 (native_encode_complex): Likewise
17331 (native_encode_vector): Likewise.
17332 (native_interpret_fixed): Likewise.
17333 (native_interpret_real): Likewise.
17334 (native_interpret_complex): Likewise.
17335 (native_interpret_vector): Likewise.
17336 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17337 (simd_clone_adjust_argument_types): Likewise.
17338 (simd_clone_init_simd_arrays): Likewise.
17339 (simd_clone_adjust): Likewise.
17340 * stor-layout.c (layout_type): Likewise.
17341 * tree.c (build_minus_one_cst): Likewise.
17342 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17343 * tree-inline.c (estimate_move_cost): Likewise.
17344 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17345 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17346 (vectorizable_reduction): Likewise.
17347 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17348 (vect_recog_mixed_size_cond_pattern): Likewise.
17349 (check_bool_pattern): Likewise.
17350 (adjust_bool_pattern): Likewise.
17351 (search_type_for_mask_1): Likewise.
17352 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17353 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17354 (vectorizable_load): Likewise.
17355 (vectorizable_store): Likewise.
17356 * ubsan.c (ubsan_encode_value): Likewise.
17357 * varasm.c (output_constant): Likewise.
17359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17360 Alan Hayward <alan.hayward@arm.com>
17361 David Sherwood <david.sherwood@arm.com>
17363 * coretypes.h (scalar_mode): New class.
17364 * machmode.h (scalar_mode): Likewise.
17365 (scalar_mode::includes_p): New function.
17366 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17367 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17368 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17369 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17370 * expmed.c (store_bit_field_1): Likewise.
17371 (extract_bit_field_1): Likewise.
17372 * expr.c (write_complex_part): Likewise.
17373 (read_complex_part): Likewise.
17374 (emit_move_complex_push): Likewise.
17375 (expand_expr_real_2): Likewise.
17376 * function.c (assign_parm_setup_reg): Likewise.
17377 (assign_parms_unsplit_complex): Likewise.
17378 * optabs.c (expand_binop): Likewise.
17379 * rtlanal.c (subreg_get_info): Likewise.
17380 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17381 * varasm.c (output_constant_pool_2): Likewise.
17383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17384 Alan Hayward <alan.hayward@arm.com>
17385 David Sherwood <david.sherwood@arm.com>
17387 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17389 (expmed_mult_highpart_optab): Likewise.
17390 (expmed_mult_highpart): Likewise.
17392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17393 Alan Hayward <alan.hayward@arm.com>
17394 David Sherwood <david.sherwood@arm.com>
17396 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17397 instead of a machine_mode.
17398 (builtin_memset_read_str): Likewise.
17399 * builtins.c (c_readstr): Likewise.
17400 (builtin_memcpy_read_str): Likewise.
17401 (builtin_strncpy_read_str): Likewise.
17402 (builtin_memset_read_str): Likewise.
17403 (builtin_memset_gen_str): Likewise.
17404 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17405 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17406 instead of a machine_mode.
17407 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17409 (make_extraction): Likewise.
17410 (try_widen_shift_mode): Take and return scalar_int_modes instead
17412 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17413 a scalar_int_mode instead of a machine_mode.
17414 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17415 (avr_addr_space_pointer_mode): Likewise.
17416 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17417 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17418 (msp430_unwind_word_mode): Likewise.
17419 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17420 (spu_addr_space_pointer_mode): Likewise.
17421 (spu_addr_space_address_mode): Likewise.
17422 (spu_libgcc_cmp_return_mode): Likewise.
17423 (spu_libgcc_shift_count_mode): Likewise.
17424 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17425 (rl78_addr_space_pointer_mode): Likewise.
17426 (fl78_unwind_word_mode): Likewise.
17427 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17429 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17430 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17431 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17432 (mips_valid_pointer_mode): Likewise.
17433 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17434 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17435 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17437 (ft32_addr_space_address_mode): Likewise.
17438 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17439 scalar_int_mode instead of a machine_mode.
17440 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17442 (m32c_addr_space_address_mode): Likewise.
17443 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17444 (rs6000_eh_return_filter_mode): Likewise.
17445 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17446 (rs6000_eh_return_filter_mode): Likewise.
17447 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17448 (s390_libgcc_shift_count_mode): Likewise.
17449 (s390_unwind_word_mode): Likewise.
17450 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17452 * target.def (mode_rep_extended): Likewise.
17453 (valid_pointer_mode): Likewise.
17454 (addr_space.valid_pointer_mode): Likewise.
17455 (eh_return_filter_mode): Return a scalar_int_mode rather than
17457 (libgcc_cmp_return_mode): Likewise.
17458 (libgcc_shift_count_mode): Likewise.
17459 (unwind_word_mode): Likewise.
17460 (addr_space.pointer_mode): Likewise.
17461 (addr_space.address_mode): Likewise.
17462 * doc/tm.texi: Regenerate.
17463 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17465 (do_jump): Use scalar_int_mode for local variables.
17466 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17467 rather than a machine_mode.
17468 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17469 (scompare_loc_descriptor_wide): Likewise.
17470 (scompare_loc_descriptor_narrow): Likewise.
17471 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17473 * except.c (sjlj_emit_dispatch_table): Likewise.
17474 (expand_builtin_eh_copy_values): Likewise.
17475 * explow.c (convert_memory_address_addr_space_1): Likewise.
17476 Take a scalar_int_mode rather than a machine_mode.
17477 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17478 than a machine_mode.
17479 (memory_address_addr_space): Use scalar_int_mode for local variables.
17480 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17481 rather than a machine_mode.
17482 * expmed.c (mask_rtx): Likewise.
17483 (init_expmed_one_conv): Likewise.
17484 (expand_mult_highpart_adjust): Likewise.
17485 (extract_high_half): Likewise.
17486 (expmed_mult_highpart_optab): Likewise.
17487 (expmed_mult_highpart): Likewise.
17488 (expand_smod_pow2): Likewise.
17489 (expand_sdiv_pow2): Likewise.
17490 (emit_store_flag_int): Likewise.
17491 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17493 (extract_low_bits): Likewise.
17494 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17496 * expr.c (pieces_addr::adjust): Likewise.
17497 (can_store_by_pieces): Likewise.
17498 (store_by_pieces): Likewise.
17499 (clear_by_pieces_1): Likewise.
17500 (expand_expr_addr_expr_1): Likewise.
17501 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17502 (expand_expr_real_1): Likewise.
17503 (try_casesi): Likewise.
17504 * final.c (shorten_branches): Likewise.
17505 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17506 type of "mode" to machine_mode.
17507 * internal-fn.c (expand_arith_overflow_result_store): Take a
17508 scalar_int_mode rather than a machine_mode.
17509 (expand_mul_overflow): Use scalar_int_mode for local variables.
17510 * loop-doloop.c (doloop_modify): Likewise.
17511 (doloop_optimize): Likewise.
17512 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17513 than a machine_mode.
17514 (expand_doubleword_shift_condmove): Likewise.
17515 (expand_doubleword_shift): Likewise.
17516 (expand_doubleword_clz): Likewise.
17517 (expand_doubleword_popcount): Likewise.
17518 (expand_doubleword_parity): Likewise.
17519 (expand_absneg_bit): Use scalar_int_mode for local variables.
17520 (prepare_float_lib_cmp): Likewise.
17521 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17522 rather than a machine_mode.
17523 (convert_memory_address_addr_space): Likewise.
17524 (get_mode_bounds): Likewise.
17525 (get_address_mode): Return a scalar_int_mode rather than a
17527 * rtlanal.c (get_address_mode): Likewise.
17528 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17529 than a machine_mode.
17530 * targhooks.c (default_mode_rep_extended): Likewise.
17531 (default_valid_pointer_mode): Likewise.
17532 (default_addr_space_valid_pointer_mode): Likewise.
17533 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17534 than a machine_mode.
17535 (default_libgcc_cmp_return_mode): Likewise.
17536 (default_libgcc_shift_count_mode): Likewise.
17537 (default_unwind_word_mode): Likewise.
17538 (default_addr_space_pointer_mode): Likewise.
17539 (default_addr_space_address_mode): Likewise.
17540 * targhooks.h (default_eh_return_filter_mode): Likewise.
17541 (default_libgcc_cmp_return_mode): Likewise.
17542 (default_libgcc_shift_count_mode): Likewise.
17543 (default_unwind_word_mode): Likewise.
17544 (default_addr_space_pointer_mode): Likewise.
17545 (default_addr_space_address_mode): Likewise.
17546 (default_mode_rep_extended): Take a scalar_int_mode rather than
17548 (default_valid_pointer_mode): Likewise.
17549 (default_addr_space_valid_pointer_mode): Likewise.
17550 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17552 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17553 rather than a machine_mode.
17554 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17555 for local variables.
17556 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17557 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17558 than a machine_mode.
17560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17561 Alan Hayward <alan.hayward@arm.com>
17562 David Sherwood <david.sherwood@arm.com>
17564 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17565 the mode argument to scalar_int_mode.
17566 (do_jump_by_parts_zero_rtx): Likewise.
17567 (do_jump_by_parts_equality_rtx): Likewise.
17568 (do_jump_by_parts_greater): Take a mode argument.
17569 (do_jump_by_parts_equality): Likewise.
17570 (do_jump_1): Update calls accordingly.
17572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17573 Alan Hayward <alan.hayward@arm.com>
17574 David Sherwood <david.sherwood@arm.com>
17576 * is-a.h (safe_dyn_cast): New function.
17577 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17578 (jump_table_for_label): Likewise.
17579 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17580 instead of an rtx_insn *.
17581 (shorten_branches): Use dyn_cast instead of LABEL_P and
17582 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17583 rtx_jump_table_data::get_data_mode.
17584 (final_scan_insn): Likewise.
17586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17587 Alan Hayward <alan.hayward@arm.com>
17588 David Sherwood <david.sherwood@arm.com>
17590 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17591 trying to combine a full-register integer set with a subreg
17594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17595 Alan Hayward <alan.hayward@arm.com>
17596 David Sherwood <david.sherwood@arm.com>
17598 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17599 that is always either address_mode or pointer_mode.
17601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17602 Alan Hayward <alan.hayward@arm.com>
17603 David Sherwood <david.sherwood@arm.com>
17605 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17606 when the two are known to be equal.
17608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17609 Alan Hayward <alan.hayward@arm.com>
17610 David Sherwood <david.sherwood@arm.com>
17612 * simplify-rtx.c (simplify_const_unary_operation): Use
17613 is_a <scalar_int_mode> instead of checking for a nonzero
17614 precision. Forcibly convert op_mode to a scalar_int_mode
17615 in that case. More clearly differentiate the operand and
17616 result modes and use the former when deciding what the value
17617 of a count-bits operation should be. Use is_int_mode instead
17618 of checking for a MODE_INT. Remove redundant check for whether
17619 this mode has a zero precision.
17621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17622 Alan Hayward <alan.hayward@arm.com>
17623 David Sherwood <david.sherwood@arm.com>
17625 * optabs.c (widen_leading): Change the type of the mode argument
17626 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17627 (widen_bswap): Likewise.
17628 (expand_parity): Likewise.
17629 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17630 (expand_ffs): Likewise.
17631 (epand_unop): Check for scalar integer modes before calling the
17634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17635 Alan Hayward <alan.hayward@arm.com>
17636 David Sherwood <david.sherwood@arm.com>
17638 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17640 (expand_expr_real_1): Update call accordingly.
17642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17643 Alan Hayward <alan.hayward@arm.com>
17644 David Sherwood <david.sherwood@arm.com>
17646 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17647 value_mode arguments. Use scalar_int_mode internally.
17648 (store_bit_field_1): Rename the new integer mode from imode
17649 to op0_mode and use it instead of GET_MODE (op0). Update calls
17650 to store_split_bit_field, store_bit_field_using_insv and
17651 store_fixed_bit_field.
17652 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17653 Use scalar_int_mode internally. Use a bit count rather than a mode
17654 when calculating the largest bit size for get_best_mode.
17655 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17656 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17657 Remove assertion that OP0 has a scalar integer mode.
17658 (store_split_bit_field): Add op0_mode and value_mode arguments.
17659 Update calls to extract_fixed_bit_field.
17660 (extract_bit_field_using_extv): Add an op0_mode argument.
17661 Use scalar_int_mode internally.
17662 (extract_bit_field_1): Rename the new integer mode from imode to
17663 op0_mode and use it instead of GET_MODE (op0). Update calls to
17664 extract_split_bit_field, extract_bit_field_using_extv and
17665 extract_fixed_bit_field.
17666 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17667 to extract_split_bit_field and extract_fixed_bit_field_1.
17668 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17669 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17670 on the target mode.
17671 (extract_split_bit_field): Add an op0_mode argument. Update call
17672 to extract_fixed_bit_field.
17674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17675 Alan Hayward <alan.hayward@arm.com>
17676 David Sherwood <david.sherwood@arm.com>
17678 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17679 * explow.c (hard_function_value): Likewise.
17680 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17681 convert_to_mode call outside the loop.
17682 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17683 for the mode iterator. Require the mode specified by max_pieces
17685 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17687 (copy_blkmode_to_reg): Likewise.
17688 (set_storage_via_setmem): Likewise.
17689 * optabs.c (prepare_cmp_insn): Likewise.
17690 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17691 * stor-layout.c (finish_bitfield_representative): Likewise.
17693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17694 Alan Hayward <alan.hayward@arm.com>
17695 David Sherwood <david.sherwood@arm.com>
17697 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17698 * expr.c (convert_move): Use them.
17699 (convert_modes): Likewise.
17700 (store_expr_with_bounds): Likewise.
17702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17703 Alan Hayward <alan.hayward@arm.com>
17704 David Sherwood <david.sherwood@arm.com>
17706 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17707 parameter for the mode of "x". Remove the "known_x", "known_mode"
17708 and "known_ret" arguments. Change the type of the mode argument
17709 to scalar_int_mode.
17710 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17711 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17712 (reg_num_sign_bit_copies_for_combine): Likewise.
17713 * rtlanal.c (nonzero_bits1): Likewise.
17714 (num_sign_bit_copies1): Likewise.
17715 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17716 (reg_num_sign_bit_copies_general): Likewise.
17717 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17718 (reg_nonzero_bits_general): Likewise.
17720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17721 Alan Hayward <alan.hayward@arm.com>
17722 David Sherwood <david.sherwood@arm.com>
17724 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17725 than in subroutines. Return 1 for non-integer modes.
17726 (cached_num_sign_bit_copies): Change the type of the mode parameter
17727 to scalar_int_mode.
17728 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17729 classes. Handle CONST_INT_P first and then check whether X also
17730 has a scalar integer mode. Check the same thing for inner registers
17731 of a SUBREG and for values that are being extended or truncated.
17733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17734 Alan Hayward <alan.hayward@arm.com>
17735 David Sherwood <david.sherwood@arm.com>
17737 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17738 in subroutines. Return the mode mask for non-integer modes.
17739 (cached_nonzero_bits): Change the type of the mode parameter
17740 to scalar_int_mode.
17741 (nonzero_bits1): Likewise. Remove early exit for other mode
17742 classes. Handle CONST_INT_P first and then check whether X
17743 also has a scalar integer mode.
17745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17746 Alan Hayward <alan.hayward@arm.com>
17747 David Sherwood <david.sherwood@arm.com>
17749 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17750 Return a scalar_int_mode and assert that the size is greater than
17752 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17753 (op_by_pieces_d::op_by_pieces_d): Likewise.
17754 (op_by_pieces_d::run): Likewise.
17755 (can_store_by_pieces): Likewise.
17757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17758 Alan Hayward <alan.hayward@arm.com>
17759 David Sherwood <david.sherwood@arm.com>
17761 * combine.c (extract_left_shift): Add a mode argument and update
17763 (make_compound_operation_int): Change the type of the mode parameter
17764 to scalar_int_mode and update the call to extract_left_shift.
17766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17767 Alan Hayward <alan.hayward@arm.com>
17768 David Sherwood <david.sherwood@arm.com>
17770 * combine.c (simplify_and_const_int): Change the type of the mode
17771 parameter to scalar_int_mode.
17772 (simplify_and_const_int_1): Likewise. Update recursive call.
17774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17775 Alan Hayward <alan.hayward@arm.com>
17776 David Sherwood <david.sherwood@arm.com>
17778 * combine.c (simplify_compare_const): Check that the mode is a
17779 scalar_int_mode (rather than VOIDmode) before testing its
17781 (simplify_comparison): Move COMPARISON_P handling out of the
17782 loop and restrict the latter part of the loop to scalar_int_modes.
17783 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17784 and when considering SUBREG_REGs. Use is_int_mode instead of
17785 checking GET_MODE_CLASS against MODE_INT.
17787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17788 Alan Hayward <alan.hayward@arm.com>
17789 David Sherwood <david.sherwood@arm.com>
17791 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17792 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17793 shift_unit_mode and for modes involved in scalar shifts.
17795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17796 Alan Hayward <alan.hayward@arm.com>
17797 David Sherwood <david.sherwood@arm.com>
17799 * combine.c (force_int_to_mode): New function, split out from...
17800 (force_to_mode): ...here. Keep xmode up-to-date and use it
17801 instead of GET_MODE (x).
17803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17804 Alan Hayward <alan.hayward@arm.com>
17805 David Sherwood <david.sherwood@arm.com>
17807 * optabs-query.h (extraction_insn::struct_mode): Change type to
17808 opt_scalar_int_mode and update comment.
17809 (extraction_insn::field_mode): Change type to scalar_int_mode.
17810 (extraction_insn::pos_mode): Likewise.
17811 * combine.c (make_extraction): Update accordingly.
17812 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17813 (get_optab_extraction_insn): Likewise.
17814 * recog.c (simplify_while_replacing): Likewise.
17815 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17816 parameter to opt_scalar_int_mode.
17818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17819 Alan Hayward <alan.hayward@arm.com>
17820 David Sherwood <david.sherwood@arm.com>
17822 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17823 to a scalar_int_mode instead of a machine_mode.
17824 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17825 (get_best_mode): Return a boolean and use a pointer argument to store
17826 the selected mode. Replace the limit mode parameter with a bit limit.
17827 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17828 for the values returned by bit_field_mode_iterator::next_mode.
17829 (store_bit_field): Update call to get_best_mode.
17830 (store_fixed_bit_field): Likewise.
17831 (extract_fixed_bit_field): Likewise.
17832 * expr.c (optimize_bitfield_assignment_op): Likewise.
17833 * fold-const.c (optimize_bit_field_compare): Likewise.
17834 (fold_truth_andor_1): Likewise.
17835 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17836 Update for new type of m_mode.
17837 (get_best_mode): As above.
17839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17840 Alan Hayward <alan.hayward@arm.com>
17841 David Sherwood <david.sherwood@arm.com>
17843 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17844 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17845 (store_bit_field): Check is_a <scalar_int_mode> before calling
17846 strict_volatile_bitfield_p.
17847 (extract_bit_field): Likewise.
17849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17850 Alan Hayward <alan.hayward@arm.com>
17851 David Sherwood <david.sherwood@arm.com>
17853 * target.def (cstore_mode): Return a scalar_int_mode.
17854 * doc/tm.texi: Regenerate.
17855 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17856 * targhooks.h (default_cstore_mode): Likewise.
17857 * targhooks.c (default_cstore_mode): Likewise, using a forced
17859 * expmed.c (emit_cstore): Expect the target of the cstore to be
17862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17863 Alan Hayward <alan.hayward@arm.com>
17864 David Sherwood <david.sherwood@arm.com>
17866 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17868 (niter_desc): Likewise mode.
17869 (iv_analyze): Add a mode parameter.
17871 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17872 and change its type to scalar_int_mode.
17873 * loop-iv.c: Update commentary at head of file.
17874 (iv_constant): Pass the mode paraeter before the rtx it describes
17875 and change its type to scalar_int_mode. Remove VOIDmode handling.
17876 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17877 (iv_extend): Likewise.
17878 (shorten_into_mode): Likewise.
17879 (iv_add): Use scalar_int_mode.
17880 (iv_mult): Likewise.
17881 (iv_shift): Likewise.
17882 (canonicalize_iv_subregs): Likewise.
17883 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17884 it describes and change its mode to scalar_int_mode. Also change
17885 the type of the returned inner_mode to scalar_int_mode.
17886 (get_biv_step): Likewise, turning outer_mode from a pointer
17887 into a direct parameter. Update call to get_biv_step_1.
17888 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17889 iv_constant and get_biv_step.
17890 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17891 and change its type to scalar_int_mode. Don't initialise iv->mode
17892 to VOIDmode and remove later checks for its still being VOIDmode.
17893 Update calls to iv_analyze_op and iv_analyze_expr. Check
17894 is_a <scalar_int_mode> when changing the mode under consideration.
17895 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17896 Update call to iv_analyze_expr.
17897 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17898 inner register is not also a scalar_int_mode. Update call to
17900 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17901 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17902 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17903 separate mode class checks. Update calls to iv_analyze. Remove
17904 fix-up of VOIDmodes after iv_analyze_biv.
17905 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17906 don't have a scalar_int_mode. Update call to biv_p.
17908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17909 Alan Hayward <alan.hayward@arm.com>
17910 David Sherwood <david.sherwood@arm.com>
17912 * cfgexpand.c (convert_debug_memory_address): Use
17913 as_a <scalar_int_mode>.
17914 * combine.c (expand_compound_operation): Likewise.
17915 (make_extraction): Likewise.
17916 (change_zero_ext): Likewise.
17917 (simplify_comparison): Likewise.
17918 * cse.c (cse_insn): Likewise.
17919 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17920 (mem_loc_descriptor): Likewise.
17921 (loc_descriptor): Likewise.
17922 * expmed.c (init_expmed_one_mode): Likewise.
17923 (synth_mult): Likewise.
17924 (emit_store_flag_1): Likewise.
17925 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17926 of a comparison with size.
17927 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17928 (reduce_to_bit_field_precision): Likewise.
17929 * function.c (expand_function_end): Likewise.
17930 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17931 * loop-doloop.c (doloop_modify): Likewise.
17932 * optabs.c (expand_binop): Likewise.
17933 (expand_unop): Likewise.
17934 (expand_copysign_absneg): Likewise.
17935 (prepare_cmp_insn): Likewise.
17936 (maybe_legitimize_operand): Likewise.
17937 * recog.c (const_scalar_int_operand): Likewise.
17938 * rtlanal.c (get_address_mode): Likewise.
17939 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17940 (simplify_cond_clz_ctz): Likewise.
17941 * tree-nested.c (get_nl_goto_field): Likewise.
17942 * tree.c (build_vector_type_for_mode): Likewise.
17943 * var-tracking.c (use_narrower_mode): Likewise.
17945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17946 Alan Hayward <alan.hayward@arm.com>
17947 David Sherwood <david.sherwood@arm.com>
17949 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17950 * builtins.c (expand_builtin_signbit): Use it.
17951 * cfgexpand.c (expand_debug_expr): Likewise.
17952 * dojump.c (do_jump): Likewise.
17953 (do_compare_and_jump): Likewise.
17954 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17955 * expmed.c (make_tree): Likewise.
17956 * expr.c (expand_expr_real_2): Likewise.
17957 (expand_expr_real_1): Likewise.
17958 (try_casesi): Likewise.
17959 * fold-const-call.c (fold_const_call_ss): Likewise.
17960 * fold-const.c (unextend): Likewise.
17961 (extract_muldiv_1): Likewise.
17962 (fold_single_bit_test): Likewise.
17963 (native_encode_int): Likewise.
17964 (native_encode_string): Likewise.
17965 (native_interpret_int): Likewise.
17966 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17967 * internal-fn.c (expand_addsub_overflow): Likewise.
17968 (expand_neg_overflow): Likewise.
17969 (expand_mul_overflow): Likewise.
17970 (expand_arith_overflow): Likewise.
17971 * match.pd: Likewise.
17972 * stor-layout.c (layout_type): Likewise.
17973 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17974 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17975 * tree-ssanames.c (get_range_info): Likewise.
17976 * tree-switch-conversion.c (array_value_type) Likewise.
17977 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17978 (vect_recog_divmod_pattern): Likewise.
17979 (vect_recog_mixed_size_cond_pattern): Likewise.
17980 * tree-vrp.c (extract_range_basic): Likewise.
17981 (simplify_float_conversion_using_ranges): Likewise.
17982 * tree.c (int_fits_type_p): Likewise.
17983 * ubsan.c (instrument_bool_enum_load): Likewise.
17984 * varasm.c (mergeable_string_section): Likewise.
17985 (narrowing_initializer_constant_valid_p): Likewise.
17986 (output_constant): Likewise.
17988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17989 Alan Hayward <alan.hayward@arm.com>
17990 David Sherwood <david.sherwood@arm.com>
17992 * machmode.h (NARROWEST_INT_MODE): New macro.
17993 * expr.c (alignment_for_piecewise_move): Use it instead of
17994 GET_CLASS_NARROWEST_MODE (MODE_INT).
17995 (push_block): Likewise.
17996 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17998 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18001 Alan Hayward <alan.hayward@arm.com>
18002 David Sherwood <david.sherwood@arm.com>
18004 * postreload.c (move2add_valid_value_p): Change the type of the
18005 mode parameter to scalar_int_mode.
18006 (move2add_use_add2_insn): Add a mode parameter and use it instead
18008 (move2add_use_add3_insn): Likewise.
18009 (reload_cse_move2add): Update accordingly.
18011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18012 Alan Hayward <alan.hayward@arm.com>
18013 David Sherwood <david.sherwood@arm.com>
18015 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18017 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18018 * optabs.c (expand_unop): Likewise.
18020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18021 Alan Hayward <alan.hayward@arm.com>
18022 David Sherwood <david.sherwood@arm.com>
18024 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18025 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18026 (popcount_loc_descriptor): Likewise.
18027 (bswap_loc_descriptor): Likewise.
18028 (rotate_loc_descriptor): Likewise.
18029 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18030 calling the functions above.
18032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18033 Alan Hayward <alan.hayward@arm.com>
18034 David Sherwood <david.sherwood@arm.com>
18036 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18038 (try_combine): Likewise.
18039 (simplify_if_then_else): Likewise.
18040 * cse.c (cse_insn): Likewise.
18041 * dwarf2out.c (mem_loc_descriptor): Likewise.
18042 * emit-rtl.c (gen_lowpart_common): Likewise.
18043 * simplify-rtx.c (simplify_truncation): Likewise.
18044 (simplify_binary_operation_1): Likewise.
18045 (simplify_const_relational_operation): Likewise.
18046 (simplify_ternary_operation): Likewise.
18047 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18050 Alan Hayward <alan.hayward@arm.com>
18051 David Sherwood <david.sherwood@arm.com>
18053 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18054 * reload.c (push_reload): Likewise.
18055 (find_reloads): Likewise.
18057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18058 Alan Hayward <alan.hayward@arm.com>
18059 David Sherwood <david.sherwood@arm.com>
18061 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18062 (make_compound_operation_int): Likewise.
18063 (change_zero_ext): Likewise.
18064 * expr.c (convert_move): Likewise.
18065 (convert_modes): Likewise.
18066 * fwprop.c (forward_propagate_subreg): Likewise.
18067 * loop-iv.c (get_biv_step_1): Likewise.
18068 * optabs.c (widen_operand): Likewise.
18069 * postreload.c (move2add_valid_value_p): Likewise.
18070 * recog.c (simplify_while_replacing): Likewise.
18071 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18072 (simplify_binary_operation_1): Likewise. Remove redundant
18073 mode equality check.
18075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18076 Alan Hayward <alan.hayward@arm.com>
18077 David Sherwood <david.sherwood@arm.com>
18079 * combine.c (combine_simplify_rtx): Add checks for
18080 is_a <scalar_int_mode>.
18081 (simplify_if_then_else): Likewise.
18082 (make_field_assignment): Likewise.
18083 (simplify_comparison): Likewise.
18084 * ifcvt.c (noce_try_bitop): Likewise.
18085 * loop-invariant.c (canonicalize_address_mult): Likewise.
18086 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18089 Alan Hayward <alan.hayward@arm.com>
18090 David Sherwood <david.sherwood@arm.com>
18092 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18093 is_a <scalar_int_mode> instead of != BLKmode.
18095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18096 Alan Hayward <alan.hayward@arm.com>
18097 David Sherwood <david.sherwood@arm.com>
18099 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18100 instead of != VOIDmode.
18101 * combine.c (if_then_else_cond): Likewise.
18102 (change_zero_ext): Likewise.
18103 * dwarf2out.c (mem_loc_descriptor): Likewise.
18104 (loc_descriptor): Likewise.
18105 * rtlanal.c (canonicalize_condition): Likewise.
18106 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18109 Alan Hayward <alan.hayward@arm.com>
18110 David Sherwood <david.sherwood@arm.com>
18112 * simplify-rtx.c (simplify_binary_operation_1): Use
18113 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18116 Alan Hayward <alan.hayward@arm.com>
18117 David Sherwood <david.sherwood@arm.com>
18119 * wide-int.h (int_traits<unsigned char>) New class.
18120 (int_traits<unsigned short>) Likewise.
18121 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18122 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18124 * combine.c (set_nonzero_bits_and_sign_copies): Use
18125 is_a <scalar_int_mode>.
18126 (find_split_point): Likewise.
18127 (combine_simplify_rtx): Likewise.
18128 (simplify_logical): Likewise.
18129 (expand_compound_operation): Likewise.
18130 (expand_field_assignment): Likewise.
18131 (make_compound_operation): Likewise.
18132 (extended_count): Likewise.
18133 (change_zero_ext): Likewise.
18134 (simplify_comparison): Likewise.
18135 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18136 (ucompare_loc_descriptor): Likewise.
18137 (minmax_loc_descriptor): Likewise.
18138 (mem_loc_descriptor): Likewise.
18139 (loc_descriptor): Likewise.
18140 * expmed.c (init_expmed_one_mode): Likewise.
18141 * lra-constraints.c (lra_constraint_offset): Likewise.
18142 * optabs.c (prepare_libcall_arg): Likewise.
18143 * postreload.c (move2add_note_store): Likewise.
18144 * reload.c (operands_match_p): Likewise.
18145 * rtl.h (load_extend_op): Likewise.
18146 * rtlhooks.c (gen_lowpart_general): Likewise.
18147 * simplify-rtx.c (simplify_truncation): Likewise.
18148 (simplify_unary_operation_1): Likewise.
18149 (simplify_binary_operation_1): Likewise.
18150 (simplify_const_binary_operation): Likewise.
18151 (simplify_const_relational_operation): Likewise.
18152 (simplify_subreg): Likewise.
18153 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18154 * var-tracking.c (adjust_mems): Likewise.
18155 (prepare_call_arguments): Likewise.
18157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18158 Alan Hayward <alan.hayward@arm.com>
18159 David Sherwood <david.sherwood@arm.com>
18161 * machmode.h (is_int_mode): New fuction.
18162 * combine.c (find_split_point): Use it.
18163 (combine_simplify_rtx): Likewise.
18164 (simplify_if_then_else): Likewise.
18165 (simplify_set): Likewise.
18166 (simplify_shift_const_1): Likewise.
18167 (simplify_comparison): Likewise.
18168 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18169 * cse.c (notreg_cost): Likewise.
18170 (cse_insn): Likewise.
18171 * cselib.c (cselib_lookup_1): Likewise.
18172 * dojump.c (do_jump_1): Likewise.
18173 (do_compare_rtx_and_jump): Likewise.
18174 * dse.c (get_call_args): Likewise.
18175 * dwarf2out.c (rtl_for_decl_init): Likewise.
18176 (native_encode_initializer): Likewise.
18177 * expmed.c (emit_store_flag_1): Likewise.
18178 (emit_store_flag): Likewise.
18179 * expr.c (convert_modes): Likewise.
18180 (store_field): Likewise.
18181 (expand_expr_real_1): Likewise.
18182 * fold-const.c (fold_read_from_constant_string): Likewise.
18183 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18184 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18185 * optabs.c (expand_binop): Likewise.
18186 (expand_unop): Likewise.
18187 (expand_abs_nojump): Likewise.
18188 (expand_one_cmpl_abs_nojump): Likewise.
18189 * simplify-rtx.c (mode_signbit_p): Likewise.
18190 (val_signbit_p): Likewise.
18191 (val_signbit_known_set_p): Likewise.
18192 (val_signbit_known_clear_p): Likewise.
18193 (simplify_relational_operation_1): Likewise.
18194 * tree.c (vector_type_mode): Likewise.
18196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18197 Alan Hayward <alan.hayward@arm.com>
18198 David Sherwood <david.sherwood@arm.com>
18200 * machmode.h (smallest_mode_for_size): Fix formatting.
18201 (smallest_int_mode_for_size): New function.
18202 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18203 instead of smallest_mode_for_size.
18204 * combine.c (make_extraction): Likewise.
18205 * config/arc/arc.c (arc_expand_movmem): Likewise.
18206 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18207 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18208 * config/s390/s390.c (s390_expand_insv): Likewise.
18209 * config/sparc/sparc.c (assign_int_registers): Likewise.
18210 * config/spu/spu.c (spu_function_value): Likewise.
18211 (spu_function_arg): Likewise.
18212 * coverage.c (get_gcov_type): Likewise.
18213 (get_gcov_unsigned_t): Likewise.
18214 * dse.c (find_shift_sequence): Likewise.
18215 * expmed.c (store_bit_field_1): Likewise.
18216 * expr.c (convert_move): Likewise.
18217 (store_field): Likewise.
18218 * internal-fn.c (expand_arith_overflow): Likewise.
18219 * optabs-query.c (get_best_extraction_insn): Likewise.
18220 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18221 * stor-layout.c (layout_type): Likewise.
18222 (initialize_sizetypes): Likewise.
18223 * targhooks.c (default_get_mask_mode): Likewise.
18224 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18227 Alan Hayward <alan.hayward@arm.com>
18228 David Sherwood <david.sherwood@arm.com>
18230 * machmode.h (opt_mode::else_blk): New function.
18231 (int_mode_for_mode): Declare.
18232 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18233 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18235 * cfgexpand.c (expand_debug_expr): Likewise.
18236 * combine.c (gen_lowpart_or_truncate): Likewise.
18237 (gen_lowpart_for_combine): Likewise.
18238 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18239 * config/avr/avr.c (avr_to_int_mode): Likewise.
18240 (avr_out_plus_1): Likewise.
18241 (avr_out_plus): Likewise.
18242 (avr_out_round): Likewise.
18243 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18244 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18245 (s390_expand_vcond): Likewise.
18246 * config/spu/spu.c (spu_split_immediate): Likewise.
18247 (spu_expand_mov): Likewise.
18248 * dse.c (get_stored_val): Likewise.
18249 * expmed.c (store_bit_field_1): Likewise.
18250 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18252 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18253 (extract_low_bits): Likewise.
18254 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18255 handling rather than repeating the check.
18256 (emit_group_store): Likewise.
18257 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18258 * optabs.c (expand_absneg_bit): Likewise.
18259 (expand_copysign_absneg): Likewise.
18260 (expand_copysign_bit): Likewise.
18261 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18262 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18263 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18264 * var-tracking.c (prepare_call_arguments): Likewise.
18265 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18266 int_mode_for_mode instead of mode_for_size.
18267 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18270 Alan Hayward <alan.hayward@arm.com>
18271 David Sherwood <david.sherwood@arm.com>
18273 * machmode.h (int_mode_for_size): New function.
18274 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18275 instead of mode_for_size.
18276 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18278 * combine.c (expand_field_assignment): Use int_mode_for_size
18279 instead of mode_for_size.
18280 (make_extraction): Likewise.
18281 (simplify_shift_const_1): Likewise.
18282 (simplify_comparison): Likewise.
18283 * dojump.c (do_jump): Likewise.
18284 * dwarf2out.c (mem_loc_descriptor): Likewise.
18285 * emit-rtl.c (init_derived_machine_modes): Likewise.
18286 * expmed.c (flip_storage_order): Likewise.
18287 (convert_extracted_bit_field): Likewise.
18288 * expr.c (copy_blkmode_from_reg): Likewise.
18289 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18290 * internal-fn.c (expand_mul_overflow): Likewise.
18291 * lower-subreg.c (simple_move): Likewise.
18292 * optabs-libfuncs.c (init_optabs): Likewise.
18293 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18294 * tree.c (vector_type_mode): Likewise.
18295 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18296 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18297 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18298 * tree-vect-stmts.c (vectorizable_load): Likewise.
18299 (vectorizable_store): Likewise.
18301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18302 Alan Hayward <alan.hayward@arm.com>
18303 David Sherwood <david.sherwood@arm.com>
18305 * coretypes.h (pod_mode): New type.
18306 (scalar_int_mode_pod): New typedef.
18307 * machmode.h (pod_mode): New class.
18308 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18309 * genmodes.c (emit_mode_int_n): Update accordingly.
18310 * lower-subreg.h (target_lower_subreg): Change type to
18311 scalar_int_mode_pod.
18312 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18313 scalar_int_mode_pod.
18315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18316 Alan Hayward <alan.hayward@arm.com>
18317 David Sherwood <david.sherwood@arm.com>
18319 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18320 machine_mode to scalar_int_mode.
18321 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18322 (rs6000_option_override_internal): Remove cast to int.
18323 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18324 machine_mode to scalar_int_mode.
18325 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18326 (rs6000_option_override_internal): Remove cast to int.
18327 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18328 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18330 * config/s390/s390.c (s390_expand_builtin): Likewise.
18331 * coretypes.h (scalar_int_mode): New type.
18332 (opt_scalar_int_mode): New typedef.
18333 * machmode.h (scalar_int_mode): New class.
18334 (scalar_int_mode::includes_p): New function.
18335 (byte_mode): Change type to scalar_int_mode.
18336 (word_mode): Likewise.
18337 (ptr_mode): Likewise.
18338 * emit-rtl.c (byte_mode): Likewise.
18339 (word_mode): Likewise.
18340 (ptr_mode): Likewise.
18341 (init_derived_machine_modes): Update accordingly.
18342 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18343 and MODE_PARTIAL_INT.
18344 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18345 opt_scalar_int_mode.
18347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18348 Alan Hayward <alan.hayward@arm.com>
18349 David Sherwood <david.sherwood@arm.com>
18351 * target.def (libgcc_floating_mode_supported_p): Take a
18353 * doc/tm.texi: Regenerate.
18354 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18356 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18357 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18361 Alan Hayward <alan.hayward@arm.com>
18362 David Sherwood <david.sherwood@arm.com>
18364 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18365 * doc/tm.texi: Regenerate.
18366 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18367 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18368 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18369 * targhooks.h (default_floatn_mode): Likewise.
18370 * targhooks.c (default_floatn_mode): Likewise.
18371 * tree.c (build_common_tree_nodes): Update accordingly.
18373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18374 Alan Hayward <alan.hayward@arm.com>
18375 David Sherwood <david.sherwood@arm.com>
18377 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18378 (mode_iterator::iterate_p): Likewise.
18379 (mode_iterator::get_wider): Likewise.
18380 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18383 Alan Hayward <alan.hayward@arm.com>
18384 David Sherwood <david.sherwood@arm.com>
18386 * coretypes.h (opt_scalar_float_mode): New typedef.
18387 * machmode.h (float_mode_for_size): New function.
18388 * emit-rtl.c (double_mode): Delete.
18389 (init_emit_once): Use float_mode_for_size.
18390 * stor-layout.c (layout_type): Likewise.
18391 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18394 Alan Hayward <alan.hayward@arm.com>
18395 David Sherwood <david.sherwood@arm.com>
18397 * output.h (assemble_real): Take a scalar_float_mode.
18398 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18399 * config/arm/arm.md (consttable_4): Likewise.
18400 (consttable_8): Likewise.
18401 (consttable_16): Likewise.
18402 * config/mips/mips.md (consttable_float): Likewise.
18403 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18404 * varasm.c (assemble_real): Take a scalar_float_mode.
18405 (output_constant_pool_2): Update accordingly.
18406 (output_constant): Likewise.
18408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18409 Alan Hayward <alan.hayward@arm.com>
18410 David Sherwood <david.sherwood@arm.com>
18412 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18413 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18414 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18415 (native_encode_real): Likewise.
18416 (native_interpret_real): Likewise.
18417 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18418 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18421 Alan Hayward <alan.hayward@arm.com>
18422 David Sherwood <david.sherwood@arm.com>
18424 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18425 <scalar_float_mode>. Simplify.
18426 (gen_extend_conv_libfunc): Likewise.
18428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18429 Alan Hayward <alan.hayward@arm.com>
18430 David Sherwood <david.sherwood@arm.com>
18432 * coretypes.h (scalar_float_mode): New type.
18433 * machmode.h (mode_traits::from_int): Use machine_mode if
18434 USE_ENUM_MODES is defined.
18435 (is_a): New function.
18437 (dyn_cast): Likewise.
18438 (scalar_float_mode): New class.
18439 (scalar_float_mode::includes_p): New function.
18440 (is_float_mode): Likewise.
18441 * gdbhooks.py (MachineModePrinter): New class.
18442 (build_pretty_printer): Use it for scalar_float_mode.
18443 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18444 (format_helper::format_helper): Turn into a template.
18445 * genmodes.c (get_mode_class): New function.
18446 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18447 or machine_mode if none.
18448 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18449 as_a <scalar_float_mode>.
18450 * dwarf2out.c (mem_loc_descriptor): Likewise.
18451 (insert_float): Likewise.
18452 (add_const_value_attribute): Likewise.
18453 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18454 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18455 (expand_unop): Update accordingly.
18456 (expand_abs_nojump): Likewise.
18457 (expand_copysign_absneg): Take a scalar_float_mode.
18458 (expand_copysign_bit): Likewise.
18459 (expand_copysign): Update accordingly.
18461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18462 Alan Hayward <alan.hayward@arm.com>
18463 David Sherwood <david.sherwood@arm.com>
18465 * coretypes.h (opt_mode): New class.
18466 * machmode.h (opt_mode): Likewise.
18467 (opt_mode::else_void): New function.
18468 (opt_mode::require): Likewise.
18469 (opt_mode::exists): Likewise.
18470 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18471 (GET_MODE_2XWIDER_MODE): Likewise.
18472 (mode_iterator::get_wider): Update accordingly.
18473 (mode_iterator::get_2xwider): Likewise.
18474 (mode_iterator::get_known_wider): Likewise, turning into a template.
18475 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18476 forcing a wider mode to exist.
18477 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18478 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18479 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18480 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18481 * lower-subreg.c (init_lower_subreg): Likewise.
18482 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18483 on the final iteration.
18484 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18485 a wider mode exists before asking for a move pattern.
18486 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18487 forcing a wider mode to exist.
18488 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18489 returning false if no such mode exists.
18490 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18491 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18492 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18493 Avoid checking for a MODE_INT if we already know the mode is not a
18495 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18496 forcing a wider mode to exist.
18497 (expmed_mult_highpart_optab): Likewise.
18498 (expmed_mult_highpart): Likewise.
18499 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18501 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18502 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18503 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18504 * internal-fn.c (expand_mul_overflow): Update use of
18505 GET_MODE_2XWIDER_MODE.
18506 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18507 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18508 GET_MODE_WIDER_MODE.
18509 (convert_plusminus_to_widen): Likewise.
18510 * tree-switch-conversion.c (array_value_type): Likewise.
18511 * var-tracking.c (emit_note_insn_var_location): Likewise.
18512 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18513 Return false inside rather than outside the loop if no wider mode
18515 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18516 and GET_MODE_2XWIDER_MODE
18517 (can_compare_p): Use else_void.
18518 * gdbhooks.py (OptMachineModePrinter): New class.
18519 (build_pretty_printer): Use it for opt_mode.
18521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18522 Alan Hayward <alan.hayward@arm.com>
18523 David Sherwood <david.sherwood@arm.com>
18525 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18526 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18529 Alan Hayward <alan.hayward@arm.com>
18530 David Sherwood <david.sherwood@arm.com>
18532 * machmode.h (mode_traits): New structure.
18533 (get_narrowest_mode): New function.
18534 (mode_iterator::start): Likewise.
18535 (mode_iterator::iterate_p): Likewise.
18536 (mode_iterator::get_wider): Likewise.
18537 (mode_iterator::get_known_wider): Likewise.
18538 (mode_iterator::get_2xwider): Likewise.
18539 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18540 (FOR_EACH_MODE): Likewise.
18541 (FOR_EACH_MODE_FROM): Likewise.
18542 (FOR_EACH_MODE_UNTIL): Likewise.
18543 (FOR_EACH_WIDER_MODE): Likewise.
18544 (FOR_EACH_2XWIDER_MODE): Likewise.
18545 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18546 * combine.c (simplify_comparison): Likewise
18547 * config/i386/i386.c (type_natural_mode): Likewise.
18548 * cse.c (cse_insn): Likewise.
18549 * dse.c (find_shift_sequence): Likewise.
18550 * emit-rtl.c (init_derived_machine_modes): Likewise.
18551 (init_emit_once): Likewise.
18552 * explow.c (hard_function_value): Likewise.
18553 * expmed.c (extract_fixed_bit_field_1): Likewise.
18554 (extract_bit_field_1): Likewise.
18555 (expand_divmod): Likewise.
18556 (emit_store_flag_1): Likewise.
18557 * expr.c (init_expr_target): Likewise.
18558 (convert_move): Likewise.
18559 (alignment_for_piecewise_move): Likewise.
18560 (widest_int_mode_for_size): Likewise.
18561 (emit_block_move_via_movmem): Likewise.
18562 (copy_blkmode_to_reg): Likewise.
18563 (set_storage_via_setmem): Likewise.
18564 (compress_float_constant): Likewise.
18565 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18566 * optabs-query.c (get_best_extraction_insn): Likewise.
18567 * optabs.c (expand_binop): Likewise.
18568 (expand_twoval_unop): Likewise.
18569 (expand_twoval_binop): Likewise.
18570 (widen_leading): Likewise.
18571 (widen_bswap): Likewise.
18572 (expand_parity): Likewise.
18573 (expand_unop): Likewise.
18574 (prepare_cmp_insn): Likewise.
18575 (prepare_float_lib_cmp): Likewise.
18576 (expand_float): Likewise.
18577 (expand_fix): Likewise.
18578 (expand_sfix_optab): Likewise.
18579 * postreload.c (move2add_use_add2_insn): Likewise.
18580 * reg-stack.c (reg_to_stack): Likewise.
18581 * reginfo.c (choose_hard_reg_mode): Likewise.
18582 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18583 * stor-layout.c (mode_for_size): Likewise.
18584 (smallest_mode_for_size): Likewise.
18585 (mode_for_vector): Likewise.
18586 (finish_bitfield_representative): Likewise.
18587 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18588 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18589 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18590 * var-tracking.c (prepare_call_arguments): Likewise.
18592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18593 Alan Hayward <alan.hayward@arm.com>
18594 David Sherwood <david.sherwood@arm.com>
18596 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18597 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18598 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18599 * machmode.h (mode_size): Move earlier in file.
18600 (mode_precision): Likewise.
18601 (mode_inner): Likewise.
18602 (mode_nunits): Likewise.
18603 (mode_unit_size): Likewise.
18604 (unit_unit_precision): Likewise.
18605 (mode_wider): Likewise.
18606 (mode_2xwider): Likewise.
18607 (machine_mode): New class.
18608 (mode_to_bytes): New function.
18609 (mode_to_bits): Likewise.
18610 (mode_to_precision): Likewise.
18611 (mode_to_inner): Likewise.
18612 (mode_to_unit_size): Likewise.
18613 (mode_to_unit_precision): Likewise.
18614 (mode_to_nunits): Likewise.
18615 (GET_MODE_SIZE): Use mode_to_bytes.
18616 (GET_MODE_BITSIZE): Use mode_to_bits.
18617 (GET_MODE_PRECISION): Use mode_to_precision.
18618 (GET_MODE_INNER): Use mode_to_inner.
18619 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18620 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18621 (GET_MODE_NUNITS): Use mode_to_nunits.
18622 * system.h (ALWAYS_INLINE): New macro.
18623 * config/powerpcspe/powerpcspe-c.c
18624 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18625 int for arg1_mode and arg2_mode.
18627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18628 Alan Hayward <alan.hayward@arm.com>
18629 David Sherwood <david.sherwood@arm.com>
18631 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18632 Prefix mode names with E_ in case statements.
18633 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18634 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18635 (aarch64_split_simd_move): Likewise.
18636 (aarch64_gen_storewb_pair): Likewise.
18637 (aarch64_gen_loadwb_pair): Likewise.
18638 (aarch64_gen_store_pair): Likewise.
18639 (aarch64_gen_load_pair): Likewise.
18640 (aarch64_get_condition_code_1): Likewise.
18641 (aarch64_constant_pool_reload_icode): Likewise.
18642 (get_rsqrte_type): Likewise.
18643 (get_rsqrts_type): Likewise.
18644 (get_recpe_type): Likewise.
18645 (get_recps_type): Likewise.
18646 (aarch64_gimplify_va_arg_expr): Likewise.
18647 (aarch64_simd_container_mode): Likewise.
18648 (aarch64_emit_load_exclusive): Likewise.
18649 (aarch64_emit_store_exclusive): Likewise.
18650 (aarch64_expand_compare_and_swap): Likewise.
18651 (aarch64_gen_atomic_cas): Likewise.
18652 (aarch64_emit_bic): Likewise.
18653 (aarch64_emit_atomic_swap): Likewise.
18654 (aarch64_emit_atomic_load_op): Likewise.
18655 (aarch64_evpc_trn): Likewise.
18656 (aarch64_evpc_uzp): Likewise.
18657 (aarch64_evpc_zip): Likewise.
18658 (aarch64_evpc_ext): Likewise.
18659 (aarch64_evpc_rev): Likewise.
18660 (aarch64_evpc_dup): Likewise.
18661 (aarch64_gen_ccmp_first): Likewise.
18662 (aarch64_gen_ccmp_next): Likewise.
18663 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18664 (alpha_emit_xfloating_libcall): Likewise.
18665 (emit_insxl): Likewise.
18666 (alpha_arg_type): Likewise.
18667 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18668 (arc_preferred_simd_mode): Likewise.
18669 (arc_secondary_reload): Likewise.
18670 (get_arc_condition_code): Likewise.
18671 (arc_print_operand): Likewise.
18672 (arc_legitimate_constant_p): Likewise.
18673 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18674 * config/arc/arc.md (casesi_load): Likewise.
18675 (casesi_compact_jump): Likewise.
18676 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18677 (cc_use_register): Likewise.
18678 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18679 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18680 (arm_init_iwmmxt_builtins): Likewise.
18681 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18682 (neon_expand_vector_init): Likewise.
18683 (arm_attr_length_move_neon): Likewise.
18684 (maybe_get_arm_condition_code): Likewise.
18685 (arm_emit_vector_const): Likewise.
18686 (arm_preferred_simd_mode): Likewise.
18687 (arm_output_iwmmxt_tinsr): Likewise.
18688 (thumb1_output_casesi): Likewise.
18689 (thumb2_output_casesi): Likewise.
18690 (arm_emit_load_exclusive): Likewise.
18691 (arm_emit_store_exclusive): Likewise.
18692 (arm_expand_compare_and_swap): Likewise.
18693 (arm_evpc_neon_vuzp): Likewise.
18694 (arm_evpc_neon_vzip): Likewise.
18695 (arm_evpc_neon_vrev): Likewise.
18696 (arm_evpc_neon_vtrn): Likewise.
18697 (arm_evpc_neon_vext): Likewise.
18698 (arm_validize_comparison): Likewise.
18699 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18700 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18701 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18702 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18703 (c6x_preferred_simd_mode): Likewise.
18704 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18705 (epiphany_rtx_costs): Likewise.
18706 * config/epiphany/predicates.md (proper_comparison_operator):
18708 * config/frv/frv.c (condexec_memory_operand): Likewise.
18709 (frv_emit_move): Likewise.
18710 (output_move_single): Likewise.
18711 (output_condmove_single): Likewise.
18712 (frv_hard_regno_mode_ok): Likewise.
18713 (frv_matching_accg_mode): Likewise.
18714 * config/h8300/h8300.c (split_adds_subs): Likewise.
18715 (h8300_rtx_costs): Likewise.
18716 (h8300_print_operand): Likewise.
18717 (compute_mov_length): Likewise.
18718 (output_logical_op): Likewise.
18719 (compute_logical_op_length): Likewise.
18720 (compute_logical_op_cc): Likewise.
18721 (h8300_shift_needs_scratch_p): Likewise.
18722 (output_a_shift): Likewise.
18723 (compute_a_shift_length): Likewise.
18724 (compute_a_shift_cc): Likewise.
18725 (expand_a_rotate): Likewise.
18726 (output_a_rotate): Likewise.
18727 * config/i386/i386.c (classify_argument): Likewise.
18728 (function_arg_advance_32): Likewise.
18729 (function_arg_32): Likewise.
18730 (function_arg_64): Likewise.
18731 (function_value_64): Likewise.
18732 (ix86_gimplify_va_arg): Likewise.
18733 (ix86_legitimate_constant_p): Likewise.
18734 (put_condition_code): Likewise.
18735 (split_double_mode): Likewise.
18736 (ix86_avx256_split_vector_move_misalign): Likewise.
18737 (ix86_expand_vector_logical_operator): Likewise.
18738 (ix86_split_idivmod): Likewise.
18739 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18740 (ix86_build_const_vector): Likewise.
18741 (ix86_build_signbit_mask): Likewise.
18742 (ix86_match_ccmode): Likewise.
18743 (ix86_cc_modes_compatible): Likewise.
18744 (ix86_expand_branch): Likewise.
18745 (ix86_expand_sse_cmp): Likewise.
18746 (ix86_expand_sse_movcc): Likewise.
18747 (ix86_expand_int_sse_cmp): Likewise.
18748 (ix86_expand_vec_perm_vpermi2): Likewise.
18749 (ix86_expand_vec_perm): Likewise.
18750 (ix86_expand_sse_unpack): Likewise.
18751 (ix86_expand_int_addcc): Likewise.
18752 (ix86_split_to_parts): Likewise.
18753 (ix86_vectorize_builtin_gather): Likewise.
18754 (ix86_vectorize_builtin_scatter): Likewise.
18755 (avx_vpermilp_parallel): Likewise.
18756 (inline_memory_move_cost): Likewise.
18757 (ix86_tieable_integer_mode_p): Likewise.
18758 (x86_maybe_negate_const_int): Likewise.
18759 (ix86_expand_vector_init_duplicate): Likewise.
18760 (ix86_expand_vector_init_one_nonzero): Likewise.
18761 (ix86_expand_vector_init_one_var): Likewise.
18762 (ix86_expand_vector_init_concat): Likewise.
18763 (ix86_expand_vector_init_interleave): Likewise.
18764 (ix86_expand_vector_init_general): Likewise.
18765 (ix86_expand_vector_set): Likewise.
18766 (ix86_expand_vector_extract): Likewise.
18767 (emit_reduc_half): Likewise.
18768 (ix86_emit_i387_round): Likewise.
18769 (ix86_mangle_type): Likewise.
18770 (ix86_expand_round_sse4): Likewise.
18771 (expand_vec_perm_blend): Likewise.
18772 (canonicalize_vector_int_perm): Likewise.
18773 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18774 (expand_vec_perm_1): Likewise.
18775 (expand_vec_perm_interleave3): Likewise.
18776 (expand_vec_perm_even_odd_pack): Likewise.
18777 (expand_vec_perm_even_odd_1): Likewise.
18778 (expand_vec_perm_broadcast_1): Likewise.
18779 (ix86_vectorize_vec_perm_const_ok): Likewise.
18780 (ix86_expand_vecop_qihi): Likewise.
18781 (ix86_expand_mul_widen_hilo): Likewise.
18782 (ix86_expand_sse2_abs): Likewise.
18783 (ix86_expand_pextr): Likewise.
18784 (ix86_expand_pinsr): Likewise.
18785 (ix86_preferred_simd_mode): Likewise.
18786 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18787 * config/i386/sse.md (*andnot<mode>3): Likewise.
18788 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18789 (*<code><mode>3): Likewise.
18790 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18791 (ia64_expand_atomic_op): Likewise.
18792 (ia64_arg_type): Likewise.
18793 (ia64_mode_to_int): Likewise.
18794 (ia64_scalar_mode_supported_p): Likewise.
18795 (ia64_vector_mode_supported_p): Likewise.
18796 (expand_vec_perm_broadcast): Likewise.
18797 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18798 (iq2000_function_arg_advance): Likewise.
18799 (iq2000_function_arg): Likewise.
18800 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18801 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18802 (m68k_libcall_value): Likewise.
18803 (m68k_function_value): Likewise.
18804 (sched_attr_op_type): Likewise.
18805 * config/mcore/mcore.c (mcore_output_move): Likewise.
18806 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18808 (microblaze_function_arg): Likewise.
18809 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18810 (mips_print_operand): Likewise.
18811 (mips_mode_ok_for_mov_fmt_p): Likewise.
18812 (mips_vector_mode_supported_p): Likewise.
18813 (mips_preferred_simd_mode): Likewise.
18814 (mips_expand_vpc_loongson_even_odd): Likewise.
18815 (mips_expand_vec_unpack): Likewise.
18816 (mips_expand_vi_broadcast): Likewise.
18817 (mips_expand_vector_init): Likewise.
18818 (mips_expand_vec_reduc): Likewise.
18819 (mips_expand_msa_cmp): Likewise.
18820 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18821 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18822 (cc_flags_for_mode): Likewise.
18823 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18824 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18825 (nds32_output_casesi_pc_relative): Likewise.
18826 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18827 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18828 (nvptx_gen_unpack): Likewise.
18829 (nvptx_gen_pack): Likewise.
18830 (nvptx_gen_shuffle): Likewise.
18831 (nvptx_gen_wcast): Likewise.
18832 (nvptx_preferred_simd_mode): Likewise.
18833 * config/pa/pa.c (pa_secondary_reload): Likewise.
18834 * config/pa/predicates.md (base14_operand): Likewise.
18835 * config/powerpcspe/powerpcspe-c.c
18836 (altivec_resolve_overloaded_builtin): Likewise.
18837 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18839 (rs6000_preferred_simd_mode): Likewise.
18840 (output_vec_const_move): Likewise.
18841 (rs6000_expand_vector_extract): Likewise.
18842 (rs6000_split_vec_extract_var): Likewise.
18843 (reg_offset_addressing_ok_p): Likewise.
18844 (rs6000_legitimate_offset_address_p): Likewise.
18845 (rs6000_legitimize_address): Likewise.
18846 (rs6000_emit_set_const): Likewise.
18847 (rs6000_const_vec): Likewise.
18848 (rs6000_emit_move): Likewise.
18849 (spe_build_register_parallel): Likewise.
18850 (rs6000_darwin64_record_arg_recurse): Likewise.
18851 (swap_selector_for_mode): Likewise.
18852 (spe_init_builtins): Likewise.
18853 (paired_init_builtins): Likewise.
18854 (altivec_init_builtins): Likewise.
18855 (do_load_for_compare): Likewise.
18856 (rs6000_generate_compare): Likewise.
18857 (rs6000_expand_float128_convert): Likewise.
18858 (emit_load_locked): Likewise.
18859 (emit_store_conditional): Likewise.
18860 (rs6000_output_function_epilogue): Likewise.
18861 (rs6000_handle_altivec_attribute): Likewise.
18862 (rs6000_function_value): Likewise.
18863 (emit_fusion_gpr_load): Likewise.
18864 (emit_fusion_p9_load): Likewise.
18865 (emit_fusion_p9_store): Likewise.
18866 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18867 (fusion_gpr_mem_load): Likewise.
18868 (fusion_addis_mem_combo_load): Likewise.
18869 (fusion_addis_mem_combo_store): Likewise.
18870 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18871 (fusion_gpr_mem_load): Likewise.
18872 (fusion_addis_mem_combo_load): Likewise.
18873 (fusion_addis_mem_combo_store): Likewise.
18874 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18876 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18877 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18878 (rs6000_preferred_simd_mode): Likewise.
18879 (output_vec_const_move): Likewise.
18880 (rs6000_expand_vector_extract): Likewise.
18881 (rs6000_split_vec_extract_var): Likewise.
18882 (reg_offset_addressing_ok_p): Likewise.
18883 (rs6000_legitimate_offset_address_p): Likewise.
18884 (rs6000_legitimize_address): Likewise.
18885 (rs6000_emit_set_const): Likewise.
18886 (rs6000_const_vec): Likewise.
18887 (rs6000_emit_move): Likewise.
18888 (rs6000_darwin64_record_arg_recurse): Likewise.
18889 (swap_selector_for_mode): Likewise.
18890 (paired_init_builtins): Likewise.
18891 (altivec_init_builtins): Likewise.
18892 (rs6000_expand_float128_convert): Likewise.
18893 (emit_load_locked): Likewise.
18894 (emit_store_conditional): Likewise.
18895 (rs6000_output_function_epilogue): Likewise.
18896 (rs6000_handle_altivec_attribute): Likewise.
18897 (rs6000_function_value): Likewise.
18898 (emit_fusion_gpr_load): Likewise.
18899 (emit_fusion_p9_load): Likewise.
18900 (emit_fusion_p9_store): Likewise.
18901 * config/rx/rx.c (rx_gen_move_template): Likewise.
18902 (flags_from_mode): Likewise.
18903 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18904 (s390_slb_comparison): Likewise.
18905 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18906 (s390_vector_mode_supported_p): Likewise.
18907 (s390_cc_modes_compatible): Likewise.
18908 (s390_match_ccmode_set): Likewise.
18909 (s390_canonicalize_comparison): Likewise.
18910 (s390_emit_compare_and_swap): Likewise.
18911 (s390_branch_condition_mask): Likewise.
18912 (s390_rtx_costs): Likewise.
18913 (s390_secondary_reload): Likewise.
18914 (__SECONDARY_RELOAD_CASE): Likewise.
18915 (s390_expand_cs): Likewise.
18916 (s390_preferred_simd_mode): Likewise.
18917 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18918 * config/sh/sh.c (sh_print_operand): Likewise.
18919 (dump_table): Likewise.
18920 (sh_secondary_reload): Likewise.
18921 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18922 * config/sh/sh.md (casesi_worker_1): Likewise.
18923 (casesi_worker_2): Likewise.
18924 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18925 (fcc_comparison_operator): Likewise.
18926 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18927 (emit_soft_tfmode_cvt): Likewise.
18928 (sparc_preferred_simd_mode): Likewise.
18929 (output_cbranch): Likewise.
18930 (sparc_print_operand): Likewise.
18931 (sparc_expand_vec_perm_bmask): Likewise.
18932 (vector_init_bshuffle): Likewise.
18933 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18934 (spu_vector_mode_supported_p): Likewise.
18935 (spu_expand_insv): Likewise.
18936 (spu_emit_branch_or_set): Likewise.
18937 (spu_handle_vector_attribute): Likewise.
18938 (spu_builtin_splats): Likewise.
18939 (spu_builtin_extract): Likewise.
18940 (spu_builtin_promote): Likewise.
18941 (spu_expand_sign_extend): Likewise.
18942 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18943 (tilegx_simd_int): Likewise.
18944 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18945 (tilepro_simd_int): Likewise.
18946 * config/v850/v850.c (const_double_split): Likewise.
18947 (v850_print_operand): Likewise.
18948 (ep_memory_offset): Likewise.
18949 * config/vax/vax.c (vax_rtx_costs): Likewise.
18950 (vax_output_int_move): Likewise.
18951 (vax_output_int_add): Likewise.
18952 (vax_output_int_subtract): Likewise.
18953 * config/visium/predicates.md (visium_branch_operator): Likewise.
18954 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18955 (visium_print_operand_address): Likewise.
18956 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18957 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18958 (xtensa_expand_conditional_branch): Likewise.
18959 (xtensa_copy_incoming_a7): Likewise.
18960 (xtensa_output_literal): Likewise.
18961 * dfp.c (decimal_real_maxval): Likewise.
18962 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18965 Alan Hayward <alan.hayward@arm.com>
18966 David Sherwood <david.sherwood@arm.com>
18968 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18969 (mode_nunits_inline): Likewise.
18970 (mode_inner_inline): Likewise.
18971 (mode_unit_size_inline): Likewise.
18972 (mode_unit_precision_inline): Likewise.
18973 (emit_insn_modes_h): Likewise. Also emit a #define of the
18975 (emit_mode_wider): Add an E_ prefix to mode names.
18976 (emit_mode_complex): Likewise.
18977 (emit_mode_inner): Likewise.
18978 (emit_mode_adjustments): Likewise.
18979 (emit_mode_int_n): Likewise.
18980 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18981 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18982 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18983 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18984 (CRC32_BUILTIN, ENTRY): Likewise.
18985 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18986 (aarch64_pop_regs): Likewise.
18987 (aarch64_process_components): Likewise.
18988 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18989 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18990 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18991 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18992 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18993 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18995 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18996 (pcmpistr): Likewise.
18997 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18998 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18999 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19001 * config/rl78/rl78.c (mduc_regs): Likewise.
19002 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19003 (htm_expand_builtin): Likewise.
19004 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19005 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19006 * config/xtensa/xtensa.c (print_operand): Likewise.
19007 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19008 (NUM_MODE_VECTOR_INT): Likewise.
19009 * genoutput.c (null_operand): Likewise.
19010 (output_operand_data): Likewise.
19011 * genrecog.c (print_parameter_value): Likewise.
19012 * lra.c (debug_operand_data): Likewise.
19014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19016 * dwarf2out.c (scompare_loc_descriptor_wide)
19017 (scompare_loc_descriptor_narrow): New functions, split out from...
19018 (scompare_loc_descriptor): ...here.
19019 * expmed.c (emit_store_flag_int): New function, split out from...
19020 (emit_store_flag): ...here.
19022 2017-08-30 Richard Biener <rguenther@suse.de>
19024 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19025 (dwarf2out_early_finish): Move setting of AT_pubnames from
19026 early debug output to early finish.
19028 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19030 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19031 and -mdata-region to the assembler.
19033 2017-08-30 Richard Biener <rguenther@suse.de>
19035 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19037 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19038 (dwarf2out_early_global_decl): Only generate a DIE for the
19039 abstract origin if it doesn't already exist or is a declaration DIE.
19040 (resolve_addr): Do not add the linkage name twice when
19041 generating a stub DIE for the DW_TAG_GNU_call_site target.
19043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19045 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19046 Use machine_mode rather than int for arg1_mode.
19048 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19051 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19052 that the second argument of the built-in functions to unpack
19053 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19054 switch statement instead a lot of if statements.
19055 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19056 Allow 64-bit values to be in Altivec registers as well as
19057 traditional floating point registers.
19058 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19060 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19062 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19063 MAX_REGS_PER_ADDRESS == 1.
19065 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19067 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19068 * config/i386/i386.c (ix86_option_override_internal): Simplify
19069 setting of opts->x_flag_entry.
19071 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19072 Jakub Jelinek <jakub@redhat.com>
19073 Richard Biener <rguenther@suse.de>
19075 PR tree-optimization/81503
19076 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19077 folded constant fits in the target type; reorder tests for clarity.
19079 2017-08-29 Martin Liska <mliska@suse.cz>
19081 * passes.def: Include pass_lower_switch.
19082 * stmt.c (dump_case_nodes): Remove and move to
19083 tree-switch-conversion.
19084 (case_values_threshold): Likewise.
19085 (expand_switch_as_decision_tree_p): Likewise.
19086 (emit_case_decision_tree): Likewise.
19087 (expand_case): Likewise.
19088 (balance_case_nodes): Likewise.
19089 (node_has_low_bound): Likewise.
19090 (node_has_high_bound): Likewise.
19091 (node_is_bounded): Likewise.
19092 (emit_case_nodes): Likewise.
19093 (struct simple_case_node): New struct.
19094 (add_case_node): Remove.
19095 (emit_case_dispatch_table): Use vector instead of case_list.
19096 (reset_out_edges_aux): Remove.
19097 (compute_cases_per_edge): Likewise.
19098 (expand_case): Build list of simple_case_node.
19099 (expand_sjlj_dispatch_table): Use it.
19100 * tree-switch-conversion.c (struct case_node): Moved from
19101 stmt.c and adjusted.
19102 (emit_case_nodes): Likewise.
19103 (node_has_low_bound): Likewise.
19104 (node_has_high_bound): Likewise.
19105 (node_is_bounded): Likewise.
19106 (case_values_threshold): Likewise.
19107 (reset_out_edges_aux): Likewise.
19108 (compute_cases_per_edge): Likewise.
19109 (add_case_node): Likewise.
19110 (dump_case_nodes): Likewise.
19111 (balance_case_nodes): Likewise.
19112 (expand_switch_as_decision_tree_p): Likewise.
19113 (emit_jump): Likewise.
19114 (emit_case_decision_tree): Likewise.
19115 (try_switch_expansion): Likewise.
19116 (do_jump_if_equal): Likewise.
19117 (emit_cmp_and_jump_insns): Likewise.
19118 (fix_phi_operands_for_edge): New function.
19119 (record_phi_operand_mapping): Likewise.
19120 (class pass_lower_switch): New pass.
19121 (pass_lower_switch::execute): New function.
19122 (make_pass_lower_switch): Likewise.
19123 (conditional_probability):
19124 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19125 * tree-pass.h: Add make_pass_lower_switch.
19127 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19130 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19133 2017-08-29 Richard Biener <rguenther@suse.de>
19135 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19136 we do not add a DW_AT_inline attribute twice.
19137 (gen_subprogram_die): Remove code setting DW_AT_inline on
19138 DECL_ABSTRACT_P nodes.
19140 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19142 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19143 calls to internal functions.
19144 (gimplify_modify_expr): Likewise.
19145 * tree-call-cdce.c (use_internal_fn): Likewise.
19146 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19147 (convert_to_divmod): Set the nothrow flag.
19148 * tree-if-conv.c (predicate_mem_writes): Likewise.
19149 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19150 (vectorizable_call): Likewise.
19151 (vectorizable_store): Likewise.
19152 (vectorizable_load): Likewise.
19153 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19154 (vect_recog_mask_conversion_pattern): Likewise.
19156 2017-08-29 Martin Liska <mliska@suse.cz>
19159 * gcc.c (driver_handle_option): Add new argument.
19160 * opts-common.c (handle_option): Pass
19161 target_option_override_hook.
19162 * opts-global.c (lang_handle_option): Add new option.
19163 (set_default_handlers): Add new argument.
19164 (decode_options): Likewise.
19165 * opts.c (target_handle_option): Likewise.
19166 (common_handle_option): Call target_option_override_hook.
19167 * opts.h (struct cl_option_handler_func): Add hook for
19168 target option override.
19169 (struct cl_option_handlers): Likewise.
19170 (set_default_handlers): Add new argument.
19171 (decode_options): Likewise.
19172 (common_handle_option): Likewise.
19173 (target_handle_option): Likewise.
19174 * toplev.c (toplev::main): Pass targetm.target_option.override
19177 2017-08-29 Richard Biener <rguenther@suse.de>
19178 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19180 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19181 life to the active subtree.
19183 2017-08-28 Jeff Law <law@redhat.com>
19185 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19186 derive_equivalences.
19187 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19188 Code moved into....
19189 (edge_info::derive_equivalences): New private member function
19191 * tree-ssa-dom.c (class edge_info): Changed from a struct
19192 to a class. Add ctor/dtor, methods and data members.
19193 (edge_info::edge_info): Renamed from allocate_edge_info.
19194 Initialize additional members.
19195 (edge_info::~edge_info): New.
19196 (free_dom_edge_info): Delete the edge info.
19197 (record_edge_info): Use new class & associated member functions.
19198 Tighten forms for testing for edge equivalences.
19199 (record_temporary_equivalences): Iterate over the simple
19200 equivalences rather than assuming there's only one per edge.
19201 (cprop_into_successor_phis): Iterate over the simple
19202 equivalences rather than assuming there's only one per edge.
19203 (optimize_stmt): Use operand_equal_p rather than pointer
19204 equality for mini-DSE code.
19206 2017-08-28 Nathan Sidwell <nathan@acm.org>
19208 * gcc.c (execute): Fold SIGPIPE handling into switch
19209 statement. Adjust internal error message.
19211 2017-08-28 Richard Biener <rguenther@suse.de>
19214 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19215 Do nothing for removed DIEs.
19217 2017-08-28 Richard Biener <rguenther@suse.de>
19219 PR tree-optimization/81977
19220 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19223 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19226 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19227 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19228 using targetm.gen_mem_thread_fence.
19230 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19233 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19234 predicate to register_operand. Reorder operands.
19235 (*btr<mode>): Ditto.
19236 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19237 (*btr<mode>_mask): Ditto.
19239 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19241 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19242 * config/rs6000/xmmintrin.h: New file.
19243 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19245 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19248 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19249 parameter and_insn and return it.
19250 (recombine_lvx_pattern): Insert a copy to ensure availability of
19251 the base register of the copied masking operation at the point of
19252 the instruction replacement.
19253 (recombine_stvx_pattern): Likewise.
19255 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19257 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19258 undocumented switches.
19259 (-mpower9-dform-vector): Likewise.
19260 (-mpower9-dform): Likewise.
19261 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19262 comments to delete references to -mpower9-dform* switches.
19263 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19264 Delete reference to -mpower9-dform* switches, test for
19265 -mpower9-vector instead.
19266 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19267 (OTHER_P9_VECTOR_MASKS): Likewise.
19268 (POWERPC_MASKS): Likewise.
19269 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19270 tests against -mpower9-dform* to -mpower9-vector. Delete code
19271 that checked for -mpower9-dform* consistancy with other options.
19272 Add test for -mpower9-misc to enable other power9 switches.
19273 (rs6000_init_hard_regno_mode_ok): Likewise.
19274 (rs6000_option_override_internal): Likewise.
19275 (rs6000_emit_prologue): Likewise.
19276 (rs6000_emit_epilogue): Likewise.
19277 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19278 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19279 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19281 (emit_fusion_p9_store): Likewise.
19282 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19283 resetting these macros if the assembler does not support ISA 3.0
19285 (TARGET_P9_DFORM_VECTOR): Likewise.
19286 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19287 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19289 2017-08-25 Alan Modra <amodra@gmail.com>
19291 PR rtl-optimization/81747
19292 * cse.c (cse_extended_basic_block): Don't attempt to record
19293 equivalences for degenerate conditional branches that branch
19294 to their fall-through.
19296 2017-08-24 Martin Sebor <msebor@redhat.com>
19298 PR middle-end/81908
19299 * gimple-fold.c (size_must_be_zero_p): New function.
19300 (gimple_fold_builtin_memory_op): Call it.
19302 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19304 * config/rs6000/mm_malloc.h: New file.
19306 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19308 PR tree-optimization/81913
19309 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19310 analysis when either IVs in condition can wrap.
19312 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19314 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19315 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19317 2017-08-24 Richard Biener <rguenther@suse.de>
19320 * targhooks.c (default_target_can_inline_p): Properly
19321 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19322 is present and always compare.
19323 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19324 infer -mfpmath=sse from TARGET_SSE_P.
19325 (ix86_can_inline_p): Properly use target_option_default_node when
19326 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19328 2017-08-24 Richard Biener <rguenther@suse.de>
19331 * dwarf2out.c (output_die): Handle flag_generate_offload like
19333 (output_comp_unit): Likewise.
19334 (gen_array_type_die): Likewise.
19335 (dwarf2out_early_finish): Likewise.
19336 (note_variable_value_in_expr): Likewise.
19337 (dwarf2out_finish): Likewise. Adjust assert.
19338 * cgraphunit.c (symbol_table::compile): Move setting of
19339 flag_generate_offload earlier ...
19340 (symbol_table::finalize_compilation_unit): ... here, before
19341 early debug finalization.
19343 2017-08-24 Richard Biener <rguenther@suse.de>
19345 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19346 and ipa-fnsummary.h.
19347 (ix86_can_inline_p): When ix86_fpmath flags do not match
19348 check whether the callee uses FP math at all.
19350 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19352 PR middle-end/81931
19353 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19354 instead of TYPE_PRECISION.
19356 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19357 Alan Hayward <alan.hayward@arm.com>
19358 David Sherwood <david.sherwood@arm.com>
19360 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19362 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19364 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19366 (ix86_frame::stack_realign_allocate): New field.
19367 (struct machine_frame_state): Modify comments.
19368 (machine_frame_state::sp_realigned_fp_end): New field.
19369 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19370 layout calculation.
19371 (sp_valid_at): Add assertion to assure no attempt to access invalid
19372 offset of a realigned stack.
19373 (fp_valid_at): Likewise.
19374 (choose_baseaddr): Modify comments.
19375 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19376 ix86_expand_prologue.
19377 (ix86_expand_prologue): Modify stack realignment and allocation.
19378 (ix86_expand_epilogue): Modify comments.
19379 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19380 avx2_runtime, avx512f, and avx512f_runtime.
19382 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19384 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19385 (mstackrealign): Do not init to -1.
19386 * config/i386/i386.c (ix86_option_override_internal):
19387 Check opts_set, not opts when setting default value of
19388 opts->x_ix86_force_align_arg_pointer.
19390 2017-08-23 Richard Biener <rguenther@suse.de>
19392 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19393 lang_hooks.decl_printable_name.
19394 * print-rtl-function.c (print_rtx_function): Likewise.
19395 * tree-pretty-print.c (dump_function_header): Likewise.
19397 2017-08-23 Richard Biener <rguenther@suse.de>
19400 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19401 -g0 at compile-time.
19403 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19405 PR middle-end/19706
19406 * doc/sourcebuild.texi (Other hardware attributes):
19409 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19411 PR middle-end/19706
19412 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19413 Fix single-use check.
19415 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19417 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19419 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19421 * doc/install.texi: Modify to add more details on running selected
19424 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19426 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19427 is combined with -mabi=ms.
19428 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19431 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19433 PR tree-optimization/81488
19434 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19435 and cached_basis fields.
19436 (MAX_SPREAD): New constant.
19437 (alloc_cand_and_find_basis): Initialize new fields.
19438 (clear_visited): New function.
19439 (create_phi_basis_1): Rename from create_phi_basis, set visited
19440 and cached_basis fields.
19441 (create_phi_basis): New wrapper function.
19442 (phi_add_costs_1): Rename from phi_add_costs, add spread
19443 parameter, set visited field, short-circuit when limits reached.
19444 (phi_add_costs): New wrapper function.
19445 (record_phi_increments_1): Rename from record_phi_increments, set
19447 (record_phi_increments): New wrapper function.
19448 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19449 (phi_incr_cost): New wrapper function.
19450 (all_phi_incrs_profitable_1): Rename from
19451 all_phi_incrs_profitable, set visited field.
19452 (all_phi_incrs_profitable): New wrapper function.
19454 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19455 Alan Hayward <alan.hayward@arm.com>
19456 David Sherwood <david.sherwood@arm.com>
19458 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19459 that takes the outer and inner modes.
19460 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19461 comparison as the canonical test for a paradoxical subreg.
19462 * combine.c (simplify_set): Use paradoxical_subreg_p.
19463 (make_extraction): Likewise.
19464 (force_to_mode): Likewise.
19465 (rtx_equal_for_field_assignment_p): Likewise.
19466 (gen_lowpart_for_combine): Likewise.
19467 (simplify_comparison): Likewise.
19468 * cse.c (equiv_constant): Likewise.
19469 * expmed.c (store_bit_field_1): Likewise.
19470 * final.c (alter_subreg): Likewise.
19471 * fwprop.c (propagate_rtx): Likewise.
19472 (forward_propagate_subreg): Likewise.
19473 * ira-conflicts.c (ira_build_conflicts): Likewise.
19474 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19475 * lra-constraints.c (curr_insn_transform): Likewise.
19476 (split_reg): Likewise.
19477 * lra-eliminations.c (move_plus_up): Likewise.
19478 (lra_eliminate_regs_1): Likewise.
19479 * recog.c (general_operand): Likewise.
19480 * ree.c (combine_reaching_defs): Likewise.
19481 * reload.c (push_reload): Likewise.
19482 (find_reloads): Likewise.
19483 * reload1.c (elimination_effects): Likewise.
19484 (compute_reload_subreg_offset): Likewise.
19485 (choose_reload_regs): Likewise.
19486 * rtlanal.c (subreg_lsb_1): Likewise.
19487 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19488 (simplify_subreg): Likewise.
19489 * var-tracking.c (track_loc_p): Likewise.
19490 * emit-rtl.c (byte_lowpart_offset): Likewise.
19491 (paradoxical_subreg_p): Delete out-of-line definition.
19493 2017-08-22 Jeff Law <law@redhat.com>
19495 PR tree-optimization/81741
19496 PR tree-optimization/71947
19497 * tree-ssa-dom.c: Include tree-inline.h.
19498 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19499 equivalences if one is more expensive to compute than the other.
19500 * tree-ssa-scopedtables.h (class const_or_copies): Make
19501 record_const_or_copy_raw method private.
19502 (class avail_exprs_stack): New method simplify_binary_operation.
19503 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19504 avail_exprs_stack::simplify_binary_operation as needed.
19505 (avail_exprs_stack::simplify_binary_operation): New function.
19507 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19509 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19510 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19511 (DOT_SYMBOLS): Likewise.
19512 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19513 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19514 (RS6000_ABI_NAME): Likewise.
19515 (TARGET_CMODEL): Likewise.
19516 (TOC_SECTION_ASM_OP): Likewise.
19517 (SET_CMODEL): New macro.
19518 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19520 2017-08-22 Richard Biener <rguenther@suse.de>
19522 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19523 to allow for free-lang-data replacements similar to verify_type_variant.
19525 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19526 Alan Hayward <alan.hayward@arm.com>
19527 David Sherwood <david.sherwood@arm.com>
19529 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19530 VOIDmode for the LABEL_REF.
19532 2017-08-22 Richard Biener <rguenther@suse.de>
19534 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19536 2017-08-22 Richard Biener <rguenther@suse.de>
19538 * common.opt (feliminate-dwarf2-dups): Ignore.
19539 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19540 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19541 same_die_p_wrap, compute_section_prefix,
19542 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19543 (comdat_symbol_id, comdat_symbol_number): Likewise.
19544 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19546 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19547 (output_die): Mark unreachable path unreachable.
19548 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19549 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19550 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19551 (dwarf2out_early_finish): Likewise.
19553 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19555 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19557 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19560 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19561 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19562 (avr_attribute_table) <io, io_low, address>: Initialize
19563 .decl_required with true.
19565 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19567 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19568 undocumented debugging options.
19569 (-mvsx-scalar-double): Likewise.
19570 (-mallow-df-permute): Likewise.
19571 (-mvectorize-builtins): Likewise.
19572 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19573 (rs6000_builtin_vectorized_function): Likewise.
19574 (rs6000_builtin_md_vectorized_function): Likewise.
19575 (rs6000_opt_vars): Likewise.
19577 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19580 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19581 (*btrq_imm): Rename from *btrq.
19582 (*btcq_imm): Rename from *btcq.
19583 (btsc): New code attribute.
19584 (*<btsc><mode>): New insn pattern.
19585 (*btr<mode>): Ditto.
19586 (*<btsc><mode>_mask): New insn_and_split pattern.
19587 (*btr<mode>_mask): Ditto.
19589 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19590 Alan Hayward <alan.hayward@arm.com>
19591 David Sherwood <david.sherwood@arm.com>
19593 * function.c (pad_below): Simplify padding calculation.
19595 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19596 Alan Hayward <alan.hayward@arm.com>
19597 David Sherwood <david.sherwood@arm.com>
19599 * target.def (function_prologue): Remove frame size argument.
19600 (function_epilogue): Likewise.
19601 * doc/tm.texi: Regenerate.
19602 * final.c (final_start_function): Update call to function_prologue.
19603 (final_end_function): Update call to function_epilogue.
19604 (default_function_pro_epilogue): Remove frame size argument.
19605 * output.h (default_function_pro_epilogue): Likewise.
19606 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19607 (arm_output_function_prologue): Likewise.
19608 * config/frv/frv.c (frv_function_prologue): Likewise.
19609 (frv_function_epilogue): Likewise.
19610 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19611 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19612 (ia64_output_function_epilogue): Likewise.
19613 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19614 (m32r_output_function_epilogue): Likewise.
19615 * config/microblaze/microblaze.c (microblaze_function_prologue)
19616 (microblaze_function_epilogue): Likewise.
19617 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19618 (mips_output_function_epilogue): Likewise.
19619 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19620 (mmix_target_asm_function_epilogue): Likewise.
19621 * config/msp430/msp430.c (msp430_start_function): Likewise.
19622 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19623 (nds32_asm_function_epilogue): Likewise.
19624 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19625 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19626 (pa_output_function_epilogue): Likewise.
19627 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19628 (rs6000_output_function_epilogue): Likewise.
19629 * config/rl78/rl78.c (rl78_start_function): Likewise.
19630 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19631 (rs6000_output_function_epilogue): Likewise.
19632 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19633 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19634 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19635 (sparc_asm_function_epilogue): Likewise.
19637 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19639 * tree.h (type_has_mode_precision_p): New function.
19640 * convert.c (convert_to_integer_1): Use it.
19641 * expr.c (expand_expr_real_2): Likewise.
19642 (expand_expr_real_1): Likewise.
19643 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19644 * match.pd: Likewise.
19645 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19646 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19647 * tree-tailcall.c (process_assignment): Likewise.
19648 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19649 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19650 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19651 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19652 (vectorizable_assignment): Likewise.
19653 (vectorizable_shift): Likewise.
19654 (vectorizable_operation): Likewise.
19655 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19657 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19659 * match.pd: Add pow (C, x) simplification.
19661 2017-08-21 Richard Biener <rguenther@suse.de>
19663 PR tree-optimization/81900
19664 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19665 for blocks with abnormal predecessors.
19666 (compute_antic): Do not set visited flag prematurely.
19668 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19671 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19673 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19675 * stor-layout.h (vector_type_mode): Move to...
19676 * tree.h (vector_type_mode): ...here.
19677 * stor-layout.c (vector_type_mode): Move to...
19678 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19680 2017-08-21 Richard Biener <rguenther@suse.de>
19682 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19683 register_external_die hooks.
19684 (debug_false_tree_charstarstar_uhwistar): Declare.
19685 (debug_nothing_tree_charstar_uhwi): Likewise.
19686 * debug.c (do_nothing_debug_hooks): Adjust.
19687 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19688 (debug_nothing_tree_charstar_uhwi): Likewise.
19689 * dbxout.c (dbx_debug_hooks): Adjust.
19690 (xcoff_debug_hooks): Likewise.
19691 * sdbout.c (sdb_debug_hooks): Likewise.
19692 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19693 * dwarf2out.c (macinfo_label_base): New global.
19694 (dwarf2out_register_external_die): New function for the
19695 register_external_die hook.
19696 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19697 (dwarf2_debug_hooks): Use them.
19698 (dwarf2_lineno_debug_hooks): Adjust.
19699 (struct die_struct): Add with_offset flag.
19700 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19701 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19702 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19703 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19704 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19705 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19706 defining section names for the early LTO debug variants.
19707 (reset_indirect_string): New helper.
19708 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19709 (print_dw_val): Add support for offsetted symbol references.
19710 (get_ultimate_context): Split out from is_cxx.
19711 (is_cxx): Use get_ultimate_context.
19712 (is_fortran): Add decl overload.
19713 (compute_comp_unit_symbol): Split out worker from
19714 compute_section_prefix.
19715 (compute_section_prefix): Call compute_comp_unit_symbol and
19716 set comdat_type_p here.
19717 (output_die): Skip DIE symbol output for the LTO added one.
19718 Handle DIE symbol references with offset.
19719 (output_comp_unit): Guard section name mangling properly.
19720 For LTO debug sections emit a symbol at the section beginning
19721 which we use to refer to its DIEs.
19722 (add_abstract_origin_attribute): For DIEs registered via
19723 dwarf2out_register_external_die directly refer to the early
19724 DIE rather than indirectly through the shadow one we created.
19725 Remove obsolete call to dwarf2out_abstract_function for
19726 non-function/block origins.
19727 (gen_array_type_die): When generating early LTO debug do
19728 not emit DW_AT_string_length.
19729 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19730 late when in LTO. As suggested place a gcc_unreachable for
19731 the DECL_ABSTRACT_P case.
19732 (gen_subprogram_die): Avoid another specification DIE
19733 for early built declarations/definitions for the late LTO case.
19734 (gen_variable_die): Add type references for late duplicated VLA dies
19736 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19737 we have the abstract instance already.
19738 (process_scope_var): Adjust decl DIE contexts in LTO which
19739 first puts them in limbo.
19740 (gen_decl_die): Do not generate type DIEs late apart from
19741 types for VLAs or for decls we do not yet have a DIE. Do not
19742 call dwarf2out_abstract_function late.
19743 (dwarf2out_early_global_decl): Make sure to create DIEs
19744 for abstract instances of a decl first.
19745 (dwarf2out_late_global_decl): Adjust comment.
19746 (output_macinfo_op): With multiple macro sections use
19747 macinfo_label_base to distinguish labels.
19748 (output_macinfo): Likewise. Update macinfo_label_base.
19749 Pass in the line info label.
19750 (note_variable_value_in_expr): When generating LTO resolve
19751 all variable values here by generating DIEs as needed.
19752 (init_sections_and_labels): Add early LTO debug flag parameter
19753 and generate different sections and names if set. Add generation
19754 counter for the labels so we can have multiple of them.
19755 (reset_dies): Helper to allow DIEs to be output multiple times.
19756 (dwarf2out_finish): When outputting DIEs to the fat part of an
19757 LTO object first reset DIEs.
19758 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19759 (modified_type_die): Check for decl_ultimate_origin being self
19761 (gen_type_die_with_usage): Likewise.
19762 (gen_typedef_die): Allow decl_ultimate_origin being self.
19763 (set_decl_abstract_flags): Remove.
19764 (set_block_abstract_flags): Likewise.
19765 (dwarf2out_abstract_function): Treat the early generated DIEs
19766 as the abstract copy and only add DW_AT_inline and
19767 DW_AT_artificial here and call set_decl_origin_self.
19768 If the DIE has an abstract origin don't do anything.
19769 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19770 if we have none yet (Go fails to build one, PR78628).
19771 (variably_modified_type_p): Prevent endless recursion for Ada
19772 cyclic pointer types.
19773 * lto-streamer-in.c: Include debug.h.
19774 (dref_queue): New global.
19775 (lto_read_tree_1): Stream in DIE references.
19776 (lto_input_tree): Register DIE references.
19777 (input_function): Stream DECL_DEBUG_ARGS.
19778 * lto-streamer-out.c: Include debug.h.
19779 (lto_write_tree_1): Output DIE references.
19780 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19781 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19782 (output_function): Stream DECL_DEBUG_ARGS.
19783 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19784 Stream DECL_ABSTRACT_ORIGIN.
19785 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19786 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19787 DECL_CONTEXT for file-scope decls.
19788 * lto-streamer.h (struct dref_entry): Declare.
19789 (dref_queue): Likewise.
19790 * cfgexpand.c (pass_expand::execute): Do not call the
19791 outlining_inline_function hook here.
19792 * lto-wrapper.c (debug_obj): New global.
19793 (tool_cleanup): Unlink it if required.
19794 (debug_objcopy): New function.
19795 (run_gcc): Handle early debug sections in the IL files by
19796 extracting them to separate files, partially linkin them and
19797 feeding the result back as result to the linker.
19798 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19799 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19800 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19801 sections into a separate segment.
19802 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19804 (darwin_asm_dwarf_section): Likewise.
19805 (darwin_asm_output_dwarf_offset): Likewise.
19806 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19808 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19809 Alan Hayward <alan.hayward@arm.com>
19810 David Sherwood <david.sherwood@arm.com>
19812 * read-md.h (md_reader::record_potential_iterator_use): Replace
19813 pointer argument with an rtx and an index.
19814 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19815 (apply_mode_iterator): Likewise.
19816 (apply_code_iterator): Likewise.
19817 (apply_int_iterator): Likewise.
19818 (apply_subst_iterator): Likewise.
19819 (record_iterator_use): Likewise.
19820 (record_attribute_use): Likewise.
19821 (md_reader::record_potential_iterator_use): Likewise. Update calls
19822 to record_iterator_use and apply_iterator.
19823 (iterator_use): Replace ptr with x and index.
19824 (attribute_use): Likewise.
19825 (apply_attribute_uses): Update calls to apply_iterator.
19826 (apply_iterators): Likewise. Update initialization of iterator_use.
19827 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19828 and record_potential_iterator_use.
19829 (rtx_reader::read_rtx_operand): Likewise.
19831 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19832 Alan Hayward <alan.hayward@arm.com>
19833 David Sherwood <david.sherwood@arm.com>
19835 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19838 2017-08-21 Richard Biener <rguenther@suse.de>
19840 PR middle-end/81884
19841 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19842 at struct end conservatively when comparing common bases.
19844 2017-08-21 Richard Biener <rguenther@suse.de>
19846 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19847 (mem_ref_in_stmt): Remove.
19848 (determine_max_movement): Use ref index to get at the reference.
19849 (invariantness_dom_walker::before_dom_children): Deal with
19850 lim data already initialized.
19851 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19853 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19855 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19856 (TARGET_ISA_ROUND): Ditto.
19857 (TARGET_ROUND): Ditto.
19858 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19859 * config/i386/i386.md: Ditto.
19860 * config/i386/sse.md: Ditto.
19861 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19862 with OPTION_MASK_ISA_SSE4_1.
19864 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19867 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19868 __builtin_ia32_lzcnt_u16.
19870 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19873 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19874 (rs6000_set_current_function): Rewrite function to use it.
19876 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19879 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19880 and TYPE_WARN_IF_NOT_ALIGN.
19881 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19882 (handle_warn_if_not_align): New.
19883 (place_union_field): Call handle_warn_if_not_align.
19884 (place_field): Call handle_warn_if_not_align.
19885 Copy TYPE_WARN_IF_NOT_ALIGN.
19886 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19887 (layout_type): Likewise.
19888 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19890 (tree_decl_common): Add warn_if_not_align.
19891 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19892 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19893 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19894 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19895 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19896 * doc/extend.texi: Document warn_if_not_aligned attribute.
19897 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19899 2017-08-17 Martin Liska <mliska@suse.cz>
19902 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19903 (get_data_dependence): Use it as pointer type.
19904 (distribute_loop): Likewise.
19906 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19908 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19909 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19910 (p8_vmrgow_<mode>_direct): New define_insn.
19911 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19912 handle endianness for vmrgew and vmrgow permute patterns.
19914 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19916 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19917 * config/rs6000/rs6000-cpus.def: Remove comment.
19918 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19919 (POWERPC_MASKS): Likewise.
19920 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19921 use of TARGET_VSX_TIMODE.
19922 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19923 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19924 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19925 (rs6000_option_override_internal): Remove dead code.
19926 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19927 (rs6000_legitimize_reload_address): Likewise.
19928 (rs6000_legitimate_address_p): Likewise.
19929 (rs6000_opt_masks): Delete "vsx-timode".
19930 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19931 from function comment.
19932 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19933 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19934 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19936 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19937 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19938 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19939 (VSX_TI): Likewise.
19941 (define_peephole2): Likewise.
19943 2017-08-17 Martin Sebor <msebor@redhat.com>
19946 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19947 past the end of an array.
19948 (test_pp_format): Add test cases.
19950 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19952 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19953 missing ECF_NOTHROW flags.
19955 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19958 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19959 operands residing in integer registers.
19960 (*vsx_le_perm_load_<mode>): Likewise.
19961 (*vsx_le_perm_store_<mode>): Likewise.
19962 (define_peephole2): Add peepholes to optimize the above.
19964 2017-08-17 Marek Polacek <polacek@redhat.com>
19966 PR middle-end/81814
19967 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19968 to mimic what shorten_compare did. Change the return type to bool.
19969 (fold_cond_expr_with_comparison): Update call to
19970 operand_equal_for_comparison_p.
19971 (fold_ternary_loc): Likewise.
19973 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19975 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19977 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19979 2017-08-17 Richard Biener <rguenther@suse.de>
19981 * tree-ssa-structalias.c (solve_graph): When propagating
19982 to successors update the graphs succ edges and avoid duplicate work.
19984 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19987 * config/i386/i386.c (ix86_option_override_internal): Save target
19988 specific options after ix86_stack_protector_guard_reg was changed.
19990 2017-08-17 Richard Biener <rguenther@suse.de>
19992 PR tree-optimization/81827
19993 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19994 (new_var_info): Initialize it conservatively.
19995 (get_call_vi): Mark register vars.
19996 (new_scalar_tmp_constraint_exp): Likewise.
19997 (handle_rhs_call): Likewise.
19998 (handle_const_call): Likewise.
19999 (create_function_info_for): Likewise.
20000 (solve_constraints): Sort varinfos to separate register from
20001 non-register vars to pack points-to solution bitmaps during
20004 2017-08-17 Marek Polacek <polacek@redhat.com>
20006 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20008 2017-08-17 Richard Biener <rguenther@suse.de>
20010 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20011 to true when overflow is undefined and we saturated the result.
20013 2017-08-17 Alan Modra <amodra@gmail.com>
20016 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20017 Don't use store multiple if only one reg needs saving.
20018 (interesting_frame_related_regno): New function.
20019 (rs6000_frame_related): Don't emit frame info for regs that
20021 (rs6000_emit_epilogue): Likewise.
20023 2017-08-16 Nathan Sidwell <nathan@acm.org>
20025 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20026 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20027 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20028 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20029 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20030 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20031 (verify_type): Adjust for TYPE_BINFO move.
20032 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20033 process TYPE_BINFO directly.
20034 (hash_tree): Likewise.
20035 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20037 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20040 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20042 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20044 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20047 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20048 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20049 Add ix86_binary_operator_ok to insn constraint.
20050 (*iordi_1_bts): Ditto.
20051 (*xordi_1_btc): Ditto.
20052 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20053 Update corresponding peephole2 pattern.
20057 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20059 PR tree-optimization/81832
20060 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20061 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20063 2017-08-16 Marek Polacek <polacek@redhat.com>
20066 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20067 regarding size_zero_node.
20069 2017-08-16 Martin Liska <mliska@suse.cz>
20072 * config.gcc: Respect previously set extra_objs in case
20075 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20077 PR tree-optimization/81835
20078 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20079 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20080 not depend on the phi.
20082 2017-08-16 Alan Modra <amodra@gmail.com>
20084 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20087 2017-08-16 Alan Modra <amodra@gmail.com>
20089 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20090 (save_reg_p): ..into this. Update all callers.
20091 (first_reg_to_save): Simplify.
20093 2017-08-16 Alan Modra <amodra@gmail.com>
20095 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20098 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20102 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20103 constant count if that count is less than 32.
20105 2017-08-15 Nathan Sidwell <nathan@acm.org>
20107 * gcc.c (execute): Emit friendlier message if inferior is killed
20108 by an external cause.
20110 2017-08-15 Richard Biener <rguenther@suse.de>
20112 PR tree-optimization/81790
20113 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20114 CONSTRUCTORs from simplifying and VN.
20116 2017-08-14 Martin Sebor <msebor@redhat.com>
20118 * builtin-attrs.def: Add comments.
20120 2017-08-14 Martin Sebor <msebor@redhat.com>
20123 * doc/extend.texi (attribute nonstring): Document new attribute.
20125 2017-08-14 Martin Sebor <msebor@redhat.com>
20128 * tree-diagnostic.c (default_tree_printer): Handle %G.
20129 * gimple-pretty-print.h (percent_G_format): Declare new function.
20130 * gimple-pretty-print.c (percent_G_format): Define.
20131 * tree-pretty-print.c (percent_K_format): Add argument.
20133 2017-08-14 Martin Sebor <msebor@redhat.com>
20135 PR translation/79998
20136 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20137 Remove a stray space.
20139 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20142 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20143 (*iordi_1_bts): Ditto.
20144 (*xordi_1_btc): Ditto.
20146 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20149 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20150 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20152 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20153 quoted strings, and make more translator-friendly.
20154 (darwin_rs6000_override_options): Likewise.
20155 (rs6000_option_override_internal): Likewise.
20156 (rs6000_return_in_memory): Fix overlong line.
20157 (init_cmulative_args): Use quoted strings, and make more
20158 translator-friendly.
20159 (rs6000_pass_by_reference): Fix overlong line.
20160 (def_builtin): Use quoted strings.
20161 (altivec_expand_predicate_builtin): Use quoted strings, and make
20162 more translator-friendly.
20163 (htm_expand_builtin): Use quoted strings.
20164 (cpu_expand_builtin): Use quoted strings, and make more
20165 translator-friendly.
20166 (altivec_expand_builtin): Likewise.
20167 (paired_expand_predicate_builtin): Likewise.
20168 (rs6000_invalid_builtin): Likewise.
20169 (builtin_function_type): Use quoted strings.
20170 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20171 more translator-friendly.
20172 (rs6000_trampoline_init): Likewise.
20173 (rs6000_handle_altivec_attribute): Likewise.
20174 (rs6000_inner_target_options): Use quoted strings.
20175 (rs6000_disable_incompatible_switches): Likewise.
20176 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20177 strings, and make more translator-friendly.
20178 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20180 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20182 PR tree-optimization/81799
20183 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20184 cond_expr to simple gimple operand.
20186 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20188 PR middle-end/46932
20189 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20191 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20195 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20196 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20197 TARGET_GASISR_PROLOGUES.
20198 * config/avr/avr.c (avr_option_override): Same.
20199 (avr_pass_pre_proep::execute): Same.
20201 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20204 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20205 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20207 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20209 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20210 define_insn_and_split. Split to a memory load from 0 in
20211 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20212 using PTR mode iterator.
20213 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20214 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20215 (*add_tp_<mode>): Redefine as define_insn_and_split.
20216 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20217 address space. Merge with *add_tp_x32 using PTR mode iterator.
20218 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20219 Split to an add with a memory load from 0 in
20220 DEFAULT_TLS_SEG_REG address space.
20222 2017-08-12 Andrew Pinski <apinski@cavium.com>
20224 * config/aarch64/aarch64-option-extensions.def (rdma):
20225 Fix feature string to what Linux prints out in /proc/cpuinfo.
20227 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20230 * dwarf2out.c (modified_type_die): For C typedef types that have
20231 an ultimate origin, process the ultimate origin instead of the
20233 (gen_typedef_die): Assert that input DECLs have no ultimate
20235 (gen_type_die_with_usage): For typedef variants that have an
20236 ultimate origin, just call gen_decl_die on the original DECL.
20237 (process_scope_var): Avoid creating DIEs for local typedefs and
20238 concrete static variables.
20240 2017-08-12 Alan Modra <amodra@gmail.com>
20244 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20245 match gnu-user.h startfile.
20246 (ENDFILE_LINUX_SPEC): Similarly.
20248 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20251 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20253 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20255 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20256 * config/aarch64/aarch64.md (mov<mode>): Change.
20257 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20258 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20259 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20261 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20263 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20264 for storage order barriers.
20266 2017-08-11 Martin Liska <mliska@suse.cz>
20269 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20270 variables of void type.
20272 2017-08-11 Martin Liska <mliska@suse.cz>
20274 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20275 TARGET_SUPPORTS_ALIASES.
20276 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20277 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20278 (optimize_weakref): Likewise.
20279 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20280 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20281 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20283 2017-08-11 Martin Liska <mliska@suse.cz>
20286 * config/i386/i386.c (make_resolver_func): Do complete
20287 refactoring of the function.
20289 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20292 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20293 * config/i386/i386.c (ix86_stack_protect_guard): Use
20294 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20295 * doc/invoke.texi (x86 Options): Document
20296 -mstack-protector-guard-symbol= option.
20298 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20300 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20301 * config/i386/i386.c (ix86_split_stack_guard): New function.
20302 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20303 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20304 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20305 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20306 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20307 (split_stack_space_check): Call ix86_split_stack_guard.
20309 2017-08-10 Martin Sebor <msebor@redhat.com>
20311 * print-tree.c (print_node): Print location using the established
20313 Replace spaces with colons.
20314 (debug_raw, debug): Ditto.
20316 2017-08-10 Martin Sebor <msebor@redhat.com>
20319 * pretty-print.c (pp_format): Correct the handling of %s precision.
20321 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20324 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20326 (ix86_finalize_stack_frame_flags): This. Also clear
20327 frame_pointer_needed if -fno-omit-frame-pointer is used without
20329 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20330 with ix86_finalize_stack_frame_flags.
20331 (ix86_expand_epilogue): Likewise.
20332 (ix86_expand_split_stack_prologue): Likewise.
20333 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20335 2017-08-10 Martin Liska <mliska@suse.cz>
20338 * c-attribs.c (handle_target_attribute):
20339 Report warning for an empty string argument of target attribute.
20341 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20344 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20346 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20347 or DECL_NONLOCAL labels.
20348 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20349 or DECL_NONLOCAL labels here.
20351 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20353 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20354 to indicate when early gimple folding has been disabled.
20355 (rs6000_gimple_fold_builtin): Add debug content.
20356 (rs6000_invalid_builtin): Fix whitespace.
20357 (rs6000_expand_builtin): Fix whitespace.
20358 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20360 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20363 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20364 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20366 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20368 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20369 (qdf24xx): Likewise.
20370 * config/aarch64/aarch64.md: Include falkor.md.
20371 * config/aarch64/falkor.md: New.
20373 2017-08-09 Marek Polacek <polacek@redhat.com>
20376 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20377 * diagnostic.c (emit_diagnostic): Add a comment.
20378 (emit_diagnostic_valist): New function.
20380 2017-08-09 Marek Polacek <polacek@redhat.com>
20383 * input.c (make_location): New overload.
20384 * input.h (make_location): Declare.
20386 2017-08-08 Alan Modra <amodra@gmail.com>
20387 H.J. Lu <hongjiu.lu@intel.com>
20390 * gcc.c (NO_PIE_SPEC): Delete.
20391 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20393 (LINK_PIE_SPEC): ..to here.
20394 (LINK_COMMAND_SPEC): Support -no-pie.
20395 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20396 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20398 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20399 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20400 (ENDFILE_CRTEND_SPEC): Similarly.
20402 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20405 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20406 (mstack-protector-guard-offset=): Ditto.
20407 * config/i386/i386.c (ix86_option_override): Handle
20408 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20410 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20411 ix86_stack_protect_guard_offset variables.
20412 (TARGET_STACK_PROTECT_GUARD): Always define.
20413 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20414 and -mstack-protector-guard-offset= options.
20416 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20418 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20419 boundary case for the last candidate.
20421 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20423 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20424 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20426 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20428 PR middle-end/19706
20429 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20430 * config/aarch64/aarch64-builtins.c
20431 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20432 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20433 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20435 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20436 Andrew Pinski <pinskia@gmail.com>
20438 PR middle-end/19706
20439 * internal-fn.def (XORSIGN): New.
20440 * optabs.def (xorsign_optab): New.
20441 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20442 (convert_expand_mult_copysign): New.
20443 (pass_optimize_widening_mul::execute): Call
20444 convert_expand_mult_copysign.
20446 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20448 PR tree-optimization/81354
20449 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20450 Insert on edges rather than explicitly creating landing pads.
20451 (analyze_candidates_and_replace): Commit edge inserts.
20453 2017-08-08 Richard Biener <rguenther@suse.de>
20455 PR middle-end/81719
20456 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20457 (expand_simple_operations): Also look through ADDR_EXPRs with
20458 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20460 2017-08-08 Richard Biener <rguenther@suse.de>
20462 PR tree-optimization/81723
20463 * tree-vect-slp.c (struct bst_traits): New hash traits.
20464 (bst_fail): New global.
20465 (vect_build_slp_tree_2): New worker, split out from ...
20466 (vect_build_slp_tree): ... this now wrapping it with using
20467 bst_fail set to cache SLP tree build fails. Properly handle
20469 (vect_analyze_slp_instance): Allocate and free bst_fail.
20471 2017-08-08 Martin Liska <mliska@suse.cz>
20474 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20475 LABEL_DECLs that can be from a different function.
20477 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20479 PR tree-optimization/81744
20480 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20481 loop's number of iterations.
20483 2017-08-08 Martin Liska <mliska@suse.cz>
20485 * asan.c: Include header files.
20486 * attribs.c (build_decl_attribute_variant): New function moved
20488 (build_type_attribute_qual_variant): Likewise.
20489 (cmp_attrib_identifiers): Likewise.
20490 (simple_cst_list_equal): Likewise.
20491 (omp_declare_simd_clauses_equal): Likewise.
20492 (attribute_value_equal): Likewise.
20493 (comp_type_attributes): Likewise.
20494 (build_type_attribute_variant): Likewise.
20495 (lookup_ident_attribute): Likewise.
20496 (remove_attribute): Likewise.
20497 (merge_attributes): Likewise.
20498 (merge_type_attributes): Likewise.
20499 (merge_decl_attributes): Likewise.
20500 (merge_dllimport_decl_attributes): Likewise.
20501 (handle_dll_attribute): Likewise.
20502 (attribute_list_equal): Likewise.
20503 (attribute_list_contained): Likewise.
20504 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20505 (lookup_attribute_by_prefix): Likewise.
20506 * bb-reorder.c: Include header files.
20507 * builtins.c: Likewise.
20508 * calls.c: Likewise.
20509 * cfgexpand.c: Likewise.
20510 * cgraph.c: Likewise.
20511 * cgraphunit.c: Likewise.
20512 * convert.c: Likewise.
20513 * dwarf2out.c: Likewise.
20514 * final.c: Likewise.
20515 * fold-const.c: Likewise.
20516 * function.c: Likewise.
20517 * gimple-expr.c: Likewise.
20518 * gimple-fold.c: Likewise.
20519 * gimple-pretty-print.c: Likewise.
20520 * gimple.c: Likewise.
20521 * gimplify.c: Likewise.
20522 * hsa-common.c: Likewise.
20523 * hsa-gen.c: Likewise.
20524 * internal-fn.c: Likewise.
20525 * ipa-chkp.c: Likewise.
20526 * ipa-cp.c: Likewise.
20527 * ipa-devirt.c: Likewise.
20528 * ipa-fnsummary.c: Likewise.
20529 * ipa-inline.c: Likewise.
20530 * ipa-visibility.c: Likewise.
20532 * lto-cgraph.c: Likewise.
20533 * omp-expand.c: Likewise.
20534 * omp-general.c: Likewise.
20535 * omp-low.c: Likewise.
20536 * omp-offload.c: Likewise.
20537 * omp-simd-clone.c: Likewise.
20538 * opts-global.c: Likewise.
20539 * passes.c: Likewise.
20540 * predict.c: Likewise.
20541 * sancov.c: Likewise.
20542 * sanopt.c: Likewise.
20543 * symtab.c: Likewise.
20544 * toplev.c: Likewise.
20545 * trans-mem.c: Likewise.
20546 * tree-chkp.c: Likewise.
20547 * tree-eh.c: Likewise.
20548 * tree-into-ssa.c: Likewise.
20549 * tree-object-size.c: Likewise.
20550 * tree-parloops.c: Likewise.
20551 * tree-profile.c: Likewise.
20552 * tree-ssa-ccp.c: Likewise.
20553 * tree-ssa-live.c: Likewise.
20554 * tree-ssa-loop.c: Likewise.
20555 * tree-ssa-sccvn.c: Likewise.
20556 * tree-ssa-structalias.c: Likewise.
20557 * tree-ssa.c: Likewise.
20558 * tree-streamer-in.c: Likewise.
20559 * tree-vectorizer.c: Likewise.
20560 * tree-vrp.c: Likewise.
20561 * tsan.c: Likewise.
20562 * ubsan.c: Likewise.
20563 * varasm.c: Likewise.
20564 * varpool.c: Likewise.
20565 * tree.c: Remove functions moved to attribs.[ch].
20566 * tree.h: Likewise.
20567 * config/aarch64/aarch64.c: Add attrs.h header file.
20568 * config/alpha/alpha.c: Likewise.
20569 * config/arc/arc.c: Likewise.
20570 * config/arm/arm.c: Likewise.
20571 * config/avr/avr.c: Likewise.
20572 * config/bfin/bfin.c: Likewise.
20573 * config/c6x/c6x.c: Likewise.
20574 * config/cr16/cr16.c: Likewise.
20575 * config/cris/cris.c: Likewise.
20576 * config/darwin.c: Likewise.
20577 * config/epiphany/epiphany.c: Likewise.
20578 * config/fr30/fr30.c: Likewise.
20579 * config/frv/frv.c: Likewise.
20580 * config/ft32/ft32.c: Likewise.
20581 * config/h8300/h8300.c: Likewise.
20582 * config/i386/winnt.c: Likewise.
20583 * config/ia64/ia64.c: Likewise.
20584 * config/iq2000/iq2000.c: Likewise.
20585 * config/lm32/lm32.c: Likewise.
20586 * config/m32c/m32c.c: Likewise.
20587 * config/m32r/m32r.c: Likewise.
20588 * config/m68k/m68k.c: Likewise.
20589 * config/mcore/mcore.c: Likewise.
20590 * config/microblaze/microblaze.c: Likewise.
20591 * config/mips/mips.c: Likewise.
20592 * config/mmix/mmix.c: Likewise.
20593 * config/mn10300/mn10300.c: Likewise.
20594 * config/moxie/moxie.c: Likewise.
20595 * config/msp430/msp430.c: Likewise.
20596 * config/nds32/nds32-isr.c: Likewise.
20597 * config/nds32/nds32.c: Likewise.
20598 * config/nios2/nios2.c: Likewise.
20599 * config/nvptx/nvptx.c: Likewise.
20600 * config/pa/pa.c: Likewise.
20601 * config/pdp11/pdp11.c: Likewise.
20602 * config/powerpcspe/powerpcspe.c: Likewise.
20603 * config/riscv/riscv.c: Likewise.
20604 * config/rl78/rl78.c: Likewise.
20605 * config/rx/rx.c: Likewise.
20606 * config/s390/s390.c: Likewise.
20607 * config/sh/sh.c: Likewise.
20608 * config/sol2.c: Likewise.
20609 * config/sparc/sparc.c: Likewise.
20610 * config/spu/spu.c: Likewise.
20611 * config/stormy16/stormy16.c: Likewise.
20612 * config/tilegx/tilegx.c: Likewise.
20613 * config/tilepro/tilepro.c: Likewise.
20614 * config/v850/v850.c: Likewise.
20615 * config/vax/vax.c: Likewise.
20616 * config/visium/visium.c: Likewise.
20617 * config/xtensa/xtensa.c: Likewise.
20619 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20622 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20623 constraints since the -mupper-regs-* switches have been
20625 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20626 into a vector from a double word element that was extracted from
20627 another vector, and eliminate extra XXPERMDI instructions.
20628 (vsx_concat_<mode>_2): Likewise.
20629 (vsx_concat_<mode>_3): Likewise.
20630 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20631 concat to allow optimizing inserts from previous extracts.
20633 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20635 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20636 memory reference to a SSP offset in TLS address space.
20637 (ix86_print_operand) <case '@'>: Remove.
20638 (ix86_print_operand_punct_valid_p): Remove '@' code.
20639 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20640 UNSPEC_SP_TLS_TEST.
20641 (stack_tls_protect_set_<mode>): Remove.
20642 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20643 (stack_tls_protect_test_<mode>): Remove.
20644 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20646 2017-08-07 Olivier Hainque <hainque@adacore.com>
20649 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20651 2017-08-07 Douglas Rupp <rupp@adacore.com>
20653 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20654 variable was referenced as multidir in command.
20656 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20659 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20662 2017-08-07 Martin Liska <mliska@suse.cz>
20664 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20665 * config/rl78/rl78.c: Add include of attribs.h.
20666 * config/sh/sh.c: Likewise.
20667 * config/v850/v850.c: Likewise.
20669 2017-08-07 Tom de Vries <tom@codesourcery.com>
20671 PR middle-end/78266
20672 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20674 2017-08-07 Martin Liska <mliska@suse.cz>
20676 * config/mips/mips.c: Include attribs.h.
20678 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20681 * doc/invoke.texi: Document change in behvaior for -Ofast for
20684 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20686 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20688 (aarch64_pop_regs): Likewise.
20689 (aarch64_gen_load_pair): Likewise.
20690 (aarch64_save_callee_saves): Likewise.
20691 (aarch64_restore_callee_saves): Likewise.
20693 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20695 * config/i386/i386.c: Revert the last change.
20697 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20700 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20702 (ix86_finalize_stack_frame_flags): This. Also clear
20703 frame_pointer_needed if -fno-omit-frame-pointer is used without
20705 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20706 with ix86_finalize_stack_frame_flags.
20707 (ix86_expand_epilogue): Likewise.
20708 (ix86_expand_split_stack_prologue): Likewise.
20710 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20713 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20714 to P_AES for Westmere.
20716 2017-08-07 Jonathan Yong <10walls@gmail.com>
20718 * config/i386/mingw.opt (fset-stack-executable): Removed.
20719 * config/i386/cygming.opt (fset-stack-executable): Moved
20721 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20723 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20725 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20727 2017-08-07 Marek Polacek <polacek@redhat.com>
20729 PR middle-end/81737
20730 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20732 2017-08-07 Martin Liska <mliska@suse.cz>
20734 * attribs.h (canonicalize_attr_name): New function.
20735 (cmp_attribs): Move from c-format.c and adjusted.
20736 (is_attribute_p): Moved from tree.h.
20737 * tree-inline.c: Add new includes.
20738 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20739 (private_is_attribute_p): Remove.
20740 (private_lookup_attribute): Likewise.
20741 (private_lookup_attribute_by_prefix): Simplify.
20742 (remove_attribute): Use is_attribute_p.
20743 * tree.h: Remove removed declarations.
20745 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20747 PR middle-end/81698
20748 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20749 instead of computing it in the function. Formatting fix.
20750 (expand_case): Don't rely on default_edge being the first edge,
20751 clear it if removing it, pass default_edge to
20752 emit_case_dispatch_table.
20753 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20756 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20758 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20759 insn in the function, emit NOP after the insn.
20761 2017-08-06 Tom de Vries <tom@codesourcery.com>
20763 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20766 2017-08-06 Tom de Vries <tom@codesourcery.com>
20768 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20771 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20773 PR tree-optimization/57371
20774 * match.pd: New pattern.
20776 2017-08-04 Marek Polacek <polacek@redhat.com>
20778 PR middle-end/81695
20779 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20780 perform the computation in offset_int.
20782 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20784 PR tree-optimization/81136
20785 * tree-vectorizer.h: Include tree-hash-traits.h.
20786 (vec_base_alignments): New typedef.
20787 (vec_info): Add a base_alignments field.
20788 (vect_record_base_alignments): Declare.
20789 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20791 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20792 (create_data_ref): Add an is_conditional_in_stmt argument.
20793 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20794 the is_conditional_in_stmt field.
20795 (data_ref_loc): Add an is_conditional_in_stmt field.
20796 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20797 (find_data_references_in_stmt): Update call to create_data_ref.
20798 (graphite_find_data_references_in_stmt): Likewise.
20799 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20800 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20801 (vect_record_base_alignment): New function.
20802 (vect_record_base_alignments): Likewise.
20803 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20804 for nested statements even if we fail to compute a misalignment.
20805 Use pooled base alignments for unconditional references.
20806 (vect_find_same_alignment_drs): Compare base addresses instead
20808 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20809 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20811 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20813 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20814 Add an explicit name for the enum. Use auto_vec for slp_instances
20815 and grouped_stores.
20816 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20818 (_bb_vec_info): Add a constructor and destructor.
20819 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20820 (destroy_loop_vec_info): Delete.
20821 (vect_destroy_datarefs): Likewise.
20822 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20823 (vec_info::vec_info): New function.
20824 (vec_info::~vec_info): Likewise.
20825 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20826 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20827 destroy_loop_vec_info.
20828 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20829 (_loop_vec_info::_loop_vec_info): ...this.
20830 (destroy_loop_vec_info): Replace with...
20831 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20832 the stmt_vec_infos. Leave handling of vec_info information to its
20833 destructor. Remove explicit vector releases.
20834 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20835 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20836 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20837 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20838 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20839 (destroy_bb_vec_info): Replace with...
20840 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20841 information to its destructor.
20842 (vect_slp_analyze_bb_1): Use new and delete instead of
20843 new_bb_vec_info and destroy_bb_vec_info.
20844 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20847 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20849 * tree-data-ref.h (subscript): Add access_fn field.
20850 (data_dependence_relation): Add could_be_independent_p.
20851 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20852 (same_access_functions): Move to tree-data-ref.c.
20853 * tree-data-ref.c (ref_contains_union_access_p): New function.
20854 (access_fn_component_p): Likewise.
20855 (access_fn_components_comparable_p): Likewise.
20856 (dr_analyze_indices): Add a reference to access_fn_component_p.
20857 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20859 (constant_access_functions): Likewise.
20860 (add_other_self_distances): Likewise.
20861 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20862 (initialize_data_dependence_relation): Use XCNEW and remove
20863 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20864 of access functions that have the same type. Allow the
20865 subsequence to end with different bases in some circumstances.
20866 Record the chosen access functions in SUB_ACCESS_FN.
20867 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20868 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20869 (subscript_dependence_tester_1): Likewise dra and drb.
20870 (build_classic_dist_vector): Update calls accordingly.
20871 (subscript_dependence_tester): Likewise.
20872 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20873 DDR_COULD_BE_INDEPENDENT_P.
20874 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20875 comp_alias_ddrs instead of may_alias_ddrs.
20876 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20878 (vect_analyze_data_ref_dependence): Use it if
20879 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20880 distance vectors if that fails.
20881 (dependence_distance_ge_vf): New function.
20882 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20883 LOOP_VINFO_MAY_ALIAS_DDRS.
20885 2017-08-04 Richard Biener <rguenther@suse.de>
20887 PR middle-end/81705
20888 * fold-const.c (fold_binary_loc): Properly restrict
20889 minus_var0 && minus_var1 case when associating undefined overflow
20892 2017-08-04 Tom de Vries <tom@codesourcery.com>
20894 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20896 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20898 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20899 Don't start diagnostic messages with a capital letter.
20900 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20902 (rs6000_invalid_builtin): Likewise.
20903 (rs6000_trampoline_init): Likewise.
20905 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20908 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20909 after setting changeable df flags.
20911 2017-08-03 Richard Biener <rguenther@suse.de>
20913 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20914 up if the use is in USE - X.
20916 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20918 * toplev.c (dumpfile.h): New include.
20919 (internal_error_reentered): New static function. Use it...
20920 (internal_error_function): ...here to handle reentered internal_error.
20922 2017-08-03 Richard Biener <rguenther@suse.de>
20924 PR middle-end/81148
20925 * fold-const.c (split_tree): Add minus_var and minus_con
20926 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20927 here but always use minus_*.
20928 (associate_trees): Assert we never associate with MINUS_EXPR
20929 and NULL first operand. Do not recurse for PLUS_EXPR operands
20930 when associating as MINUS_EXPR either.
20931 (fold_binary_loc): Track minus_var and minus_con.
20933 2017-08-03 Tom de Vries <tom@codesourcery.com>
20936 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20937 ACCEL_COMPILER, apply finish_options on
20938 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20940 2017-08-03 Tom de Vries <tom@codesourcery.com>
20943 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20944 function_entry_patch_area_size > 0.
20946 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20949 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20950 instead of 10??LU, perform unit multiplication in wide_int,
20951 don't change alloc_object_size_limit if the limit is larger
20954 PR tree-optimization/81655
20955 PR tree-optimization/81588
20956 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20957 the case when ranges[i].low and high are 1 for unsigned type with
20960 PR middle-end/81052
20961 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20962 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20964 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20966 * tree-vrp.h: Add include guard.
20968 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20971 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20972 (ud2): New insn pattern.
20973 * config/i386/i386.c (ix86_expand_epilogue):
20974 For naked functions, generate ud2 instead of trap insn.
20976 2017-08-02 Marek Polacek <polacek@redhat.com>
20979 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20981 2017-08-02 Tom de Vries <tom@codesourcery.com>
20982 Cesar Philippidis <cesar@codesourcery.com>
20984 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20985 Add missing edge probabilities.
20987 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20989 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20990 Correct endianness.
20992 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20994 PR middle-end/79499
20995 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20996 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20997 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20998 prologue_seq sequences - if any.
21000 2017-08-02 Richard Biener <rguenther@suse.de>
21002 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21003 via vectors if supported, integer extracts via punning if supported
21004 or otherwise vector extracts.
21006 2017-08-02 Richard Biener <rguenther@suse.de>
21008 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21010 (bitmap_insert_into_set): ... this.
21012 2017-08-02 Richard Biener <rguenther@suse.de>
21014 PR tree-optimization/81633
21016 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21018 PR tree-optimization/71752
21019 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21021 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21023 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21024 (machine_function::call_ms2sysv_pad_out): Remove field.
21025 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21026 (ix86_compute_frame_layout): Likewise.
21028 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21031 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21032 attribute with interrupt attribute.
21034 2017-08-01 Andrew Pinski <apinski@cavium.com>
21036 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21037 BIT_INSERT_EXPR's operand 1
21038 to see if the types precision matches.
21040 2017-08-01 Martin Liska <mliska@suse.cz>
21042 PR middle-end/70140
21043 * builtins.c (expand_builtin_memcpy_args): Remove.
21044 (expand_builtin_memcpy): Call newly added function
21045 expand_builtin_memory_copy_args.
21046 (expand_builtin_memcpy_with_bounds): Likewise.
21047 (expand_builtin_mempcpy): Remove last argument.
21048 (expand_builtin_mempcpy_with_bounds): Likewise.
21049 (expand_builtin_memory_copy_args): New function created from
21050 expand_builtin_mempcpy_args with small modifications.
21051 (expand_builtin_mempcpy_args): Remove.
21052 (expand_builtin_stpcpy): Remove unused argument.
21053 (expand_builtin): Likewise.
21054 (expand_builtin_with_bounds): Likewise.
21056 2017-08-01 Martin Liska <mliska@suse.cz>
21059 Make mempcpy more optimal (PR middle-end/70140).
21061 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21064 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21065 __builtin_vec_cmpne verify both arguments are compatible vectors
21066 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21067 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21068 move computation of aligned to after checking the argument types.
21072 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21073 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21076 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21077 Doug Rupp <rupp@adacore.com>
21078 Olivier Hainque <hainque@adacore.com>
21080 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21081 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21083 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21084 for TARGET_OS_CPP_BUILTIN.
21085 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21086 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21088 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21089 passing required abi options to the assembler for EABI configurations.
21090 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21091 of .text.hot and .text.unlikely sections for kernel modules when
21092 using ARM style exceptions.
21093 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21094 options. Add EXTRA_CC1_SPEC.
21095 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21097 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21099 (ARM_TARGET2_DWARF_FORMAT): Define.
21100 * config/arm/t-vxworks: Adjust multilib control to removal of the
21101 Diab command line options.
21103 2017-08-01 Martin Liska <mliska@suse.cz>
21105 PR gcov-profile/81561
21106 * gcov.c (unblock): Make unblocking safe as we need to preserve
21107 index correspondence of blocks and block_lists.
21109 2017-08-01 Richard Biener <rguenther@suse.de>
21111 PR tree-optimization/81181
21112 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21113 (compute_antic): ... end of iteration here.
21115 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21117 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21118 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21119 (ftree-slp-vectorize): Likewise.
21120 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21121 can no longer be set independent of flag_tree_loop_vectorize.
21122 * omp-general.c (emp_max_vf): Likewise.
21123 * opts.c (enable_fdo_optimizations): Remove references to
21124 flag_tree_vectorize, these are now implicit.
21125 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21126 and leave it for the options machinery.
21128 2017-08-01 Martin Liska <mliska@suse.cz>
21130 PR middle-end/70140
21131 * builtins.c (expand_builtin_memcpy_args): Remove.
21132 (expand_builtin_memcpy): Call newly added function
21133 expand_builtin_memory_copy_args.
21134 (expand_builtin_memcpy_with_bounds): Likewise.
21135 (expand_builtin_mempcpy): Remove last argument.
21136 (expand_builtin_mempcpy_with_bounds): Likewise.
21137 (expand_builtin_memory_copy_args): New function created from
21138 expand_builtin_mempcpy_args with small modifications.
21139 (expand_builtin_mempcpy_args): Remove.
21140 (expand_builtin_stpcpy): Remove unused argument.
21141 (expand_builtin): Likewise.
21142 (expand_builtin_with_bounds): Likewise.
21144 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21147 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21148 print "ds:" only for immediates in generic address space.
21150 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21153 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21154 (ix86_function_ok_for_sibcall): Return false for naked functions.
21156 2017-08-01 Richard Biener <rguenther@suse.de>
21158 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21159 (compute_antic): Seed worklist with exit block predecessors.
21160 * cfganal.c (dfs_find_deadend): For a cycle return the source
21161 of the edge closing it.
21163 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21165 * config/aarch64/aarch64.c
21166 (aarch64_can_const_movi_rtx_p): Move 0 check.
21168 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21170 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21171 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21173 * match.pd: Ditto in address comparison pattern.
21175 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21177 PR tree-optimization/81627
21178 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21179 closed ssa form for store-store chain.
21181 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21183 PR tree-optimization/81620
21184 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21185 for store-store chain.
21187 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21189 PR tree-optimization/81588
21190 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21191 ranges[i].in_p, invert comparison code ccode. For >/>=,
21192 swap rhs1 and rhs2 and comparison code unconditionally,
21193 for </<= don't do that. Don't swap rhs1/rhs2 again if
21194 ranges[i].in_p, instead invert comparison code ccode if
21195 opcode or oe->rank is BIT_IOR_EXPR.
21198 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21199 a direct optab to conversion optab.
21200 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21201 with GET_MODE_INNER as last argument instead of optab_handler.
21202 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21203 vector extraction if possible and optab is available.
21204 * expr.c (store_constructor): Use convert_optab_handler instead
21205 of optab_handler. Use vector initialization from smaller
21206 vectors if possible and optab is available.
21207 * tree-vect-stmts.c (vectorizable_load): Likewise.
21208 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21209 now have two modes.
21210 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21211 of vec_init from half-sized vectors with the same element mode.
21212 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21213 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21214 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21215 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21216 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21217 after mode in gen_vec_extract* calls.
21218 (vec_extract<mode>): Renamed to ...
21219 (vec_extract<mode><ssescalarmodelower>): ... this.
21220 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21221 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21222 element mode after mode in gen_vec_init* calls.
21223 (VEC_INIT_HALF_MODE): New mode iterator.
21224 (vec_init<mode>): Renamed to ...
21225 (vec_init<mode><ssescalarmodelower>): ... this.
21226 (vec_init<mode><ssehalfvecmodelower>): New expander.
21227 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21228 (vec_extractv2sfsf): ... this.
21229 (vec_initv2sf): Renamed to ...
21230 (vec_initv2sfsf): ... this.
21231 (vec_extractv2si): Renamed to ...
21232 (vec_extractv2sisi): ... this.
21233 (vec_initv2si): Renamed to ...
21234 (vec_initv2sisi): ... this.
21235 (vec_extractv4hi): Renamed to ...
21236 (vec_extractv4hihi): ... this.
21237 (vec_initv4hi): Renamed to ...
21238 (vec_initv4hihi): ... this.
21239 (vec_extractv8qi): Renamed to ...
21240 (vec_extractv8qiqi): ... this.
21241 (vec_initv8qi): Renamed to ...
21242 (vec_initv8qiqi): ... this.
21243 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21244 (vec_init<mode>): Renamed to ...
21245 (vec_init<mode><VEC_base_l>): ... this.
21246 (vec_extract<mode>): Renamed to ...
21247 (vec_extract<mode><VEC_base_l>): ... this.
21248 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21249 (vec_initv2sfsf): ... this.
21250 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21251 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21252 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21253 element mode after mode in gen_vec_init* calls.
21254 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21255 (vec_init<mode><Vel>): ... this.
21256 (vec_extract<mode>): Renamed to ...
21257 (vec_extract<mode><Vel>): ... this.
21258 * config/aarch64/iterators.md (Vel): New mode attribute.
21259 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21260 Add element mode after mode in gen_vec_extract* calls.
21261 * config/s390/vector.md (non_vec_l): New mode attribute.
21262 (vec_extract<mode>): Renamed to ...
21263 (vec_extract<mode><non_vec_l>): ... this.
21264 (vec_init<mode>): Renamed to ...
21265 (vec_init<mode><non_vec_l>): ... this.
21266 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21267 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21269 * config/arm/iterators.md (V_elem_l): New mode attribute.
21270 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21271 (vec_extract<mode><V_elem_l>): ... this.
21272 (vec_extractv2di): Renamed to ...
21273 (vec_extractv2didi): ... this.
21274 (vec_init<mode>): Renamed to ...
21275 (vec_init<mode><V_elem_l>): ... this.
21276 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21277 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21278 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21279 Add element mode after gen_vec_extract* calls.
21280 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21281 (vec_init<mode><unitmode>): ... this.
21282 (vec_extract<mode>): Renamed to ...
21283 (vec_extract<mode><unitmode>): ... this.
21284 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21285 (vec_init<mode><unitmode>): ... this.
21286 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21287 (vec_initv2sfsf): ... this.
21288 (vec_extractv2sf): Renamed to ...
21289 (vec_extractv2sfsf): ... this.
21290 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21291 Add element mode after gen_vec_extract* calls.
21292 * config/mips/mips.md (unitmode): New mode iterator.
21293 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21294 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21295 * config/spu/spu.md (inner_l): New mode attribute.
21296 (vec_init<mode>): Renamed to ...
21297 (vec_init<mode><inner_l>): ... this.
21298 (vec_extract<mode>): Renamed to ...
21299 (vec_extract<mode><inner_l>): ... this.
21300 * config/sparc/sparc.md (veltmode): New mode iterator.
21301 (vec_init<VMALL:mode>): Renamed to ...
21302 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21303 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21304 (vec_initv2sisi): ... this.
21305 (vec_initv2sf): Renamed to ...
21306 (vec_initv2sfsf): ... this.
21307 (vec_extractv2sf): Renamed to ...
21308 (vec_extractv2sfsf): ... this.
21309 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21310 (vec_init<mode>): Renamed to ...
21311 (vec_init<mode><VEC_base_l>): ... this.
21312 (vec_extract<mode>): Renamed to ...
21313 (vec_extract<mode><VEC_base_l>): ... this.
21314 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21315 (vec_initv2sfsf): ... this.
21316 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21317 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21318 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21319 gen_vec_init* calls.
21321 2017-08-01 Richard Biener <rguenther@suse.de>
21323 PR tree-optimization/81297
21324 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21325 TREE_OVERFLOW from INTEGER_CSTs.
21327 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21329 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21331 2017-07-31 Carl Love <cel@us.ibm.com>
21333 * config/rs6000/rs6000-c: Add support for built-in functions
21334 vector signed char vec_xl_be (signed long long, signed char *);
21335 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21336 vector signed int vec_xl_be (signed long long, signed int *);
21337 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21338 vector signed long long vec_xl_be (signed long long, signed long long *);
21339 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21340 vector signed short vec_xl_be (signed long long, signed short *);
21341 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21342 vector double vec_xl_be (signed long long, double *);
21343 vector float vec_xl_be (signed long long, float *);
21344 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21345 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21346 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21348 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21349 (altivec_expand_builtin): Add switch statement to call
21350 altivec_expand_xl_be for each builtin.
21351 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21352 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21353 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21354 __builtin_vsx_le_be_v16qi.
21355 * doc/extend.texi: Update the built-in documentation file for the
21356 new built-in functions.
21358 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21361 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21363 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21365 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21367 * config.gcc: Add z14.
21368 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21369 CPU model numbers for z13s and z14.
21370 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21372 * config/s390/s390-opts.h (enum processor_type): Rename
21373 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21374 * config/s390/s390.c (processor_table): Add field for CPU name to
21375 be passed to Binutils.
21376 (s390_asm_output_machine_for_arch): Use the new field in
21377 processor_table for Binutils.
21378 (s390_expand_builtin): Replace arch12 with z14.
21379 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21380 (s390_get_sched_attrmask): Likewise.
21381 (s390_get_unit_mask): Likewise.
21382 * config/s390/s390.opt: Add z14 to processor_type enum.
21384 2017-07-31 Martin Jambor <mjambor@suse.cz>
21387 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21388 regardless of optimization level.
21390 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21391 Martin Liska <mliska@suse.cz>
21393 * predict.def: Remove old comment and adjust probability.
21394 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21395 PREDICT statements.
21397 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21400 * config/i386/i386.c (ix86_function_naked): New function.
21401 (ix86_can_use_return_insn_p): Return false for naked functions.
21402 (ix86_expand_prologue): Skip prologue for naked functions.
21403 (ix86_expand_epilogue): Skip epilogue for naked functions
21404 and emit trap instruction.
21405 (ix86_warn_func_return): New function.
21406 (ix86_attribute_table): Add "naked" attribute specification.
21407 (TARGET_WARN_FUNC_RETURN): Define.
21408 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21410 2017-07-31 Martin Liska <mliska@suse.cz>
21412 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21413 (dump_gimple_bb_header): Always dump BB info.
21414 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21416 2017-07-31 Martin Liska <mliska@suse.cz>
21419 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21420 also with current_function_decl non-null equality.
21422 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21425 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21426 change type to the element type, instead add eltype variable and
21427 use it where we are interested in the element type.
21429 PR tree-optimization/81603
21430 * ipa-polymorphic-call.c
21431 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21432 offset arithmetic in offset_int, bail out if the resulting bit offset
21433 doesn't fit into shwi.
21435 2017-07-31 Martin Liska <mliska@suse.cz>
21437 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21438 (gimplify_save_expr): Fix comment.
21440 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21443 * config/i386/i386.c (ix86_function_arg): Update arguments for
21445 (ix86_compute_frame_layout): Set the initial stack offset to
21446 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21447 INCOMING_FRAME_SP_OFFSET.
21448 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21449 stack before exception handler returns.
21450 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21451 the 'ERROR_CODE' for exception handler.
21453 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21455 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21456 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21457 (ASM_OUTPUT_REG_POP): Ditto.
21458 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21459 instead of asm_fprintf to output pure string.
21461 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21463 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21464 to imported_module_or_decl hook.
21465 (debug_nothing_tree_tree_tree_bool): Remove.
21466 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21467 * debug.c (do_nothing_debug_hooks): Use
21468 debug_nothing_tree_tree_tree_bool_bool instead of
21469 debug_nothing_tree_tree_tree_bool.
21470 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21471 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21472 * sdbout.c (sdb_debug_hooks): Likewise.
21473 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21474 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21476 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21477 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21478 attribute, don't add anything.
21480 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21482 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21483 (fold_build2_stat_loc): Likewise.
21484 (fold_build3_stat_loc): Likewise.
21485 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21486 (fold_build1_loc): Remove macro.
21487 (fold_build2_loc): Likewise.
21488 (fold_build3_loc): Likewise.
21490 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21492 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21493 (gimple_build_debug_bind_source_stat): Likewise.
21494 * gimple.h (gimple_build_debug_bind): Remove macro.
21495 (gimple_build_debug_bind_source): Likewise.
21497 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21499 * bitmap.c (bitmap_alloc): Adjust.
21500 (bitmap_gc_alloc): Likewise.
21501 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21503 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21505 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21506 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21507 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21508 (bitmap_gc_alloc_stat): Likewise.
21509 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21511 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21513 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21514 * rtl.h (shallow_copy_rtx): Remove macro.
21516 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21518 * emit-rtl.c (gen_raw_REG): Adjust.
21519 * gengenrtl.c (gendef): Likewise.
21520 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21521 * rtl.h (rtx_alloc): Remove macro.
21523 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21525 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21526 (build_tree_list_stat): Likewise.
21527 * tree.h (build_tree_list): Remove macro.
21528 (build_tree_list_vec): Likewise.
21530 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21532 * tree.c (make_vector_stat): Remove _stat from name.
21533 (build_vector_stat): Likewise.
21534 * tree.h (make_vector_stat): Remove macro.
21535 (build_vector_stat): Likewise.
21537 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21539 * tree.h (build_var_debug_value): Remove prototype.
21541 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21543 * tree.c (tree_cons_stat): Remove _stat from name.
21544 * tree.h (tree_cons): Remove macro.
21546 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21548 * tree.c (build_vl_exp_stat): Remove _stat from name.
21549 * tree.h (build_vl_exp): Remove macro.
21551 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21553 * tree.c (build_decl_stat): Remove _stat from name.
21554 * tree.h (build_decl): Remove macro.
21556 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21558 * gimple.c (gimple_build_with_ops_stat): Adjust.
21559 (gimple_alloc_stat): Remove _stat from name.
21560 * gimple.h (gimple_alloc): Remove macro.
21562 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21564 * tree.c (make_tree_vec_stat): Remove _stat from name.
21565 (grow_tree_vec_stat): Likewise.
21566 * tree.h (make_tree_vec_stat): Adjust prototype.
21567 (grow_tree_vec_stat): Likewise.
21568 (make_tree_vec): Remove macro.
21569 (grow_tree_vec): Likewise.
21571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21573 * fold-const.c (fold_build1_stat_loc): Adjust.
21574 (fold_build2_stat_loc): Likewise.
21575 (fold_build3_stat_loc): Likewise.
21576 * tree.c (build0_stat): Remove _stat from name.
21577 (build1_stat): Likewise.
21578 (build2_stat): Likewise.
21579 (build3_stat): Likewise.
21580 (build4_stat): Likewise.
21581 (build5_stat): Likewise.
21582 * tree.h (build1_loc): Remove macro, and rename _stat function
21584 (build2_loc): Likewise.
21585 (build3_loc): Likewise.
21586 (build4_loc): Likewise.
21587 (build5_loc): Likewise.
21589 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21591 * tree.c (make_int_cst_stat): Remove _stat from name.
21592 * tree.h (make_int_cst_stat): Adjust prototype.
21593 (make_int_cst): Remove macro.
21595 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21597 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21598 * tree.h (make_tree_binfo_stat): Adjust prototype.
21599 (make_tree_binfo): Remove.
21601 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21603 * tree.c (copy_node_stat): Rename to copy_node.
21604 (build_distinct_type_copy): Adjust.
21605 * tree.h (copy_node_stat): Adjust prototype.
21606 (copy_node): Remove macro.
21608 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21610 * tree.c (make_node_stat): rename to make_node.
21611 (build_tree_list_stat): Adjust.
21612 (build0_stat): Likewise.
21613 (build2_stat): Likewise.
21614 (build3_stat): Likewise.
21615 (build4_stat): Likewise.
21616 (build5_stat): Likewise.
21617 (build_decl_stat): Likewise.
21618 * tree.h (make_node_stat): Adjust prototype.
21619 (make_node): remove macro.
21621 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21623 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21624 (PPC_FEATURE2_SCV): Likewise.
21625 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21627 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21629 * config/aarch64/aarch64.c
21630 (aarch64_internal_mov_immediate): Add new special pattern.
21631 * config/aarch64/aarch64.md (*movdi_aarch64):
21632 Add reg/32bit const mov case.
21634 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21635 Richard Sandiford <richard.sandiford@linaro.org>
21637 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21638 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21639 Add integer and movi cases.
21640 (movi-split-hf-df-sf split, fp16): New.
21641 (enabled): Added TARGET_FP_F16INST.
21642 * config/aarch64/iterators.md (GPF_HF): New.
21643 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21645 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21647 * config/aarch64/aarch64.c
21648 (aarch64_simd_container_mode): Add prototype.
21649 (aarch64_expand_mov_immediate): Add HI support.
21650 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21651 (aarch64_can_const_movi_rtx_p): New.
21652 (aarch64_preferred_reload_class):
21653 Remove restrictions of using FP registers for certain SIMD operations.
21654 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21655 (aarch64_valid_floating_const): Add integer move validation.
21656 (aarch64_simd_imm_scalar_p): Remove.
21657 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21658 (aarch64_legitimate_constant_p): Expand list of supported cases.
21659 * config/aarch64/aarch64-protos.h
21660 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21661 (aarch64_reinterpret_float_as_int): New.
21662 (aarch64_simd_imm_scalar_p): Remove.
21663 * config/aarch64/constraints.md (Uvi): New.
21664 (Dd): Split into Ds and new Dd.
21665 * config/aarch64/aarch64.md (*movsi_aarch64):
21667 (*movdi_aarch64): Add SIMD mov case.
21669 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21671 * tree-predcom.c: (struct chain): Handle store-store chain in which
21672 stores for elimination only store loop invariant values.
21673 (execute_pred_commoning_chain): Ditto.
21674 (prepare_initializers_chain_store_elim): Ditto.
21675 (prepare_finalizers): Ditto.
21676 (is_inv_store_elimination_chain): New function.
21677 (initialize_root_vars_store_elim_1): New function.
21679 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21681 * tree-predcom.c: Revise general description of the pass.
21682 (enum chain_type): New enum type for store elimination.
21683 (struct chain): New field supporting store elimination.
21684 (struct component): Ditto.
21685 (dump_chain): Dump store-stores chain.
21686 (release_chain): Release resources.
21687 (split_data_refs_to_components): Compute and create component
21688 contains only stores for elimination.
21689 (get_chain_last_ref_at): New function.
21690 (make_invariant_chain): Initialization.
21691 (make_rooted_chain): Specify chain type in parameter and record it.
21692 (add_looparound_copies): Skip for store-stores chain.
21693 (determine_roots_comp): Compute type of chain and pass it to
21695 (initialize_root_vars_store_elim_2): New function.
21696 (finalize_eliminated_stores): New function.
21697 (remove_stmt): Handle store for elimination.
21698 (execute_pred_commoning_chain): Execute predictive commoning on
21699 store-store chains.
21700 (determine_unroll_factor): Skip unroll for store-stores chain.
21701 (prepare_initializers_chain_store_elim): New function.
21702 (prepare_initializers_chain): Hanlde store-store chain.
21703 (prepare_finalizers_chain, prepare_finalizers): New function.
21704 (tree_predictive_commoning_loop): Return integer value indicating
21705 if loop is unrolled or lcssa form is corrupted.
21706 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21708 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21710 * tree-predcom.c (initialize_root): Delete.
21711 (execute_pred_commoning_chain): Initialize root vars and replace
21712 reference of non-combined chain directly, rather than call above
21715 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21717 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21718 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21720 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21722 * tree-predcom.c (struct chain): New field init_seq.
21723 (release_chain): Release init_seq.
21724 (prepare_initializers_chain): Record intialization stmts in above
21726 (insert_init_seqs): New function.
21727 (tree_predictive_commoning_loop): Call insert_init_seqs.
21729 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21731 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21733 2017-07-28 Richard Biener <rguenther@suse.de>
21735 * match.pd: Remove superfluous :c.
21736 * genmatch.c (simplify::id): Add member.
21737 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21739 (current_id): New global.
21740 (dt_node::parent): Move from ...
21741 (dt_operand::parent): ... here. Add for_id member.
21742 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21743 (decision_tree::find_node): Relax order requirement when
21744 merging DT_TRUE nodes to ones inbetween the current simplify
21745 and the one we try to merge with. Add diagnostic whenever
21746 we need to enforce pattern order by not merging.
21747 (decision_tree::insert): Set current_id.
21748 (decision_tree::print_node): Dump parent node and for_id.
21749 (parser::last_id): Add member.
21750 (parser::push_simplify): Assign unique id.
21751 (parser::parser): Initialize last_id.
21753 2017-07-28 Martin Liska <mliska@suse.cz>
21756 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21757 gimple_build_debug_bind.
21759 2017-07-28 Richard Biener <rguenther@suse.de>
21761 PR tree-optimization/81502
21762 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21764 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21766 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21767 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21768 for BIT_FIELD_REF args.
21769 * fold-const.c (make_bit_field_ref): Likewise.
21770 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21772 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21775 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21776 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21777 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21778 Or it into SANITIZER_UNDEFINED.
21779 * ubsan.c: Include gimple-fold.h and varasm.h.
21780 (ubsan_expand_ptr_ifn): New function.
21781 (instrument_pointer_overflow): New function.
21782 (maybe_instrument_pointer_overflow): New function.
21783 (instrument_object_size): Formatting fix.
21784 (pass_ubsan::execute): Call instrument_pointer_overflow
21785 and maybe_instrument_pointer_overflow.
21786 * internal-fn.c (expand_UBSAN_PTR): New function.
21787 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21788 * sanitizer.def (__ubsan_handle_pointer_overflow,
21789 __ubsan_handle_pointer_overflow_abort): New builtins.
21790 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21791 * internal-fn.def (UBSAN_PTR): New internal function.
21792 * opts.c (sanitizer_opts): Add pointer-overflow.
21793 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21794 * fold-const.c (build_range_check): Compute pointer range check in
21795 integral type if pointer arithmetics would be needed. Formatting
21798 2017-07-28 Martin Liska <mliska@suse.cz>
21801 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21802 parameters that are of a variable-length.
21804 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21806 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21808 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21809 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21810 (CRT_CALL_STATIC_FUNCTION): Likewise.
21811 (ASM_DEFAULT_SPEC): New define.
21812 (ASM_SPEC32): Likewise.
21813 (ASM_SPEC64): Likewise.
21814 (ASM_SPEC_COMMON): Likewise.
21815 (ASM_SPEC): Likewise.
21816 (INVALID_64BIT): Likewise.
21817 (LINK_OS_DEFAULT_SPEC): Likewise.
21818 (LINK_OS_SPEC32): Likewise.
21819 (LINK_OS_SPEC64): Likewise.
21820 (POWERPC_LINUX): Likewise.
21821 (PTRDIFF_TYPE): Likewise.
21822 (RESTORE_FP_PREFIX): Likewise.
21823 (RESTORE_FP_SUFFIX): Likewise.
21824 (SAVE_FP_PREFIX): Likewise.
21825 (SAVE_FP_SUFFIX): Likewise.
21826 (SIZE_TYPE): Likewise.
21827 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21828 (TARGET_64BIT): Likewise.
21829 (TARGET_64BIT): Likewise.
21830 (TARGET_AIX): Likewise.
21831 (WCHAR_TYPE_SIZE): Likewise.
21832 (WCHAR_TYPE): Undefine.
21833 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21834 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21835 (CPP_OS_RTEMS_SPEC): Delete.
21836 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21837 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21839 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21841 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21843 PR tree-optimization/81578
21844 * tree-parloops.c (build_new_reduction): Bail out if
21845 reduction_code isn't one of the standard OpenMP reductions.
21846 Move the details printing after that decision.
21848 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21850 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21851 related to reload_in_progress.
21852 (splat_input_operand): Likewise.
21853 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21855 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21857 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21858 (TARGET_INSTANTIATE_DECLS): Likewise.
21859 (legitimate_indexed_address_p): Delete reload_in_progress code.
21860 (rs6000_debug_legitimate_address_p): Likewise.
21861 (rs6000_eliminate_indexed_memrefs): Likewise.
21862 (rs6000_emit_le_vsx_store): Likewise.
21863 (rs6000_emit_move_si_sf_subreg): Likewise.
21864 (rs6000_emit_move): Likewise.
21865 (register_to_reg_type): Likewise.
21866 (rs6000_pre_atomic_barrier): Likewise.
21867 (rs6000_machopic_legitimize_pic_address): Likewise.
21868 (rs6000_allocate_stack_temp): Likewise.
21869 (rs6000_address_for_fpconvert): Likewise.
21870 (rs6000_address_for_altivec): Likewise.
21871 (rs6000_secondary_memory_needed_rtx): Delete function.
21872 (rs6000_check_sdmode): Likewise.
21873 (rs6000_alloc_sdmode_stack_slot): Likewise.
21874 (rs6000_instantiate_decls): Likewise.
21875 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21876 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21877 Delete reload_in_progress.
21878 (*vec_reload_and_plus_<mptrsize>): Likewise.
21879 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21880 (vsx_div_v2di): Likewise.
21881 (vsx_udiv_v2di): Likewise.
21883 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21885 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21886 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21887 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21888 (rs6000_debug_reg_global): Delete print of LRA status.
21889 (rs6000_option_override_internal): Delete dead LRA related code.
21890 (rs6000_lra_p): Delete function.
21891 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21893 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21895 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21896 * config/riscv/rtems.h: New file.
21898 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21899 Sudakshina Das <sudi.das@arm.com>
21901 * config/aarch64/aarch64.md
21902 (define_split for and<mode>3nr_compare): Move
21903 non aarch64_logical_operand to a register.
21904 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21905 register immediate operand to a register.
21906 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21908 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21910 PR middle-end/81564
21911 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21913 2017-07-27 Richard Biener <rguenther@suse.de>
21915 PR tree-optimization/81573
21916 PR tree-optimization/81494
21917 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21918 multi defuse cycle case.
21920 2017-07-27 Richard Biener <rguenther@suse.de>
21922 PR tree-optimization/81571
21923 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21926 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21928 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21929 earlier and only if MASK_FPU is set. Adjust formatting.
21931 2017-07-27 Martin Liska <mliska@suse.cz>
21933 * opt-functions.awk: Add validation of value of Init.
21934 * optc-gen.awk: Pass new argument.
21936 2017-07-27 Martin Liska <mliska@suse.cz>
21938 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21939 Fix wrong condition.
21941 2017-07-27 Martin Liska <mliska@suse.cz>
21943 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21944 BBs and edges seen by autoFDO.
21946 2017-07-27 Richard Biener <rguenther@suse.de>
21948 PR tree-optimization/81502
21949 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21950 with incompatible but same sized type.
21951 (execute_update_addresses_taken): Likewise.
21953 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21955 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21956 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21958 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21961 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21962 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21963 Change s_operand to memory_operand.
21965 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21967 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21968 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21969 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21970 Emit instructions rather than returning an expression. Handle TFmode
21971 and KFmode by casting to TImode.
21972 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21973 (rs6000_emit_le_vsx_store): Likewise.
21974 * config/rs6000/vsx.md (VSX_TI): New iterator.
21975 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21976 (*vsx_le_undo_permute_<mode>): Likewise.
21977 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21978 emit the split sequence.
21979 (*vsx_le_perm_store_<mode>): Likewise.
21981 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21983 PR tree-optimization/81555
21984 PR tree-optimization/81556
21985 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21986 if true, force CHANGED for the recursive invocation.
21987 (reassociate_bb): Remember original length of ops array, pass
21988 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21990 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21991 attributes for noipa attribute. For naked attribute use
21992 lookup_attribute first before lookup_attribute_spec.
21993 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21995 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21996 for functions with noipa attribute.
21997 (cgraph_externally_visible_p): Return true for functions with noipa
21999 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22000 for functions with noipa attribute.
22001 * doc/extend.texi: Document noipa function attribute.
22002 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22003 also for functions with noipa attribute.
22004 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22006 2017-07-26 Andrew Pinski <apinski@cavium.com>
22008 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22009 vec_unalign_load_cost and vec_unalign_store_cost.
22011 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22013 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22014 -mvsx-small-integer option.
22015 (ISA_3_0_MASKS_IEEE): Likewise.
22016 (OTHER_VSX_VECTOR_MASKS): Likewise.
22017 (POWERPC_MASKS): Likewise.
22018 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22019 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22020 code, only testing for DImode being allowed in non-VSX floating
22022 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22023 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22025 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22026 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22027 TARGET_P8_VECTOR test.
22028 (rs6000_secondary_reload_simple_move): Likewise.
22029 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22030 since TARGET_P9_VECTOR was already tested.
22031 (rs6000_opt_masks): Remove -mvsx-small-integer.
22032 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22033 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22035 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22036 test for TARGET_VEXTRACTUB was used, and that uses
22038 (p9 extract splitter): Likewise.
22039 (vsx_extract_<mode>_di_p9): Likewise.
22040 (vsx_extract_<mode>_store_p9): Likewise.
22041 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22042 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22043 the elimination of TARGET_VSX_SMALL_INTEGER.
22044 (vsx_extract_<mode>_p8): Likewise.
22045 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22046 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22047 (vsx_set_<mode>_p9): Likewise.
22048 (vsx_set_v4sf_p9): Likewise.
22049 (vsx_set_v4sf_p9_zero): Likewise.
22050 (vsx_insert_extract_v4sf_p9): Likewise.
22051 (vsx_insert_extract_v4sf_p9_2): Likewise.
22052 * config/rs6000/rs6000.md (sign extend splitter): Change
22053 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22054 (floatsi<mode>2_lfiwax_mem): Likewise.
22055 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22056 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22057 since a test for TARGET_P9_VECTOR was used.
22058 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22059 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22060 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22061 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22062 TARGET_P8_VECTOR test.
22063 (fix_trunc<mode>si2_stfiwx): Likewise.
22064 (fix_trunc<mode>si2_internal): Likewise.
22065 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22066 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22068 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22069 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22070 TARGET_P8_VECTOR test.
22071 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22072 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22073 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22075 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22076 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22077 since a test for TARGET_P9_VECTOR was used.
22078 (splitter for loading small constants): Likewise.
22080 2017-07-26 Andrew Pinski <apinski@cavium.com>
22082 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22085 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22088 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22089 (fp_valid_at): Likewise.
22091 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22093 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22094 (qdf24xx_addrcost_table): Likewise.
22095 (cortexa57_tunings): Update to use generic_branch_cost.
22096 (cortexa72_tunings): Likewise.
22097 (cortexa73_tunings): Likewise.
22098 (qdf24xx_tunings): Likewise.
22100 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22102 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22103 (thunderx2t99_branch_cost): Likewise.
22104 (cortexa35_tunings): Update to use generic_branch_cost.
22105 (cortexa53_tunings): Likewise.
22106 (cortexa57_tunings): Likewise.
22107 (cortexa72_tunings): Likewise.
22108 (cortexa73_tunings): Likewise.
22109 (thunderx2t99_tunings): Likewise.
22111 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22113 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22114 (sparc_option_override): Honour MASK_FSMULD.
22115 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22116 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22117 * config/sparc/sparc.opt (mfsmuld): New option.
22118 * doc/invoke.texi (mfsmuld): Document option.
22120 2017-07-26 Marek Polacek <polacek@redhat.com>
22122 PR middle-end/70992
22123 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22125 2017-07-26 Richard Biener <rguenther@suse.de>
22127 * gimple-match-head.c (do_valueize): Return OP if valueize
22129 (get_def): New helper to get at the def stmt of a SSA name
22130 if valueize allows.
22131 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22132 do_valueize to get at the def stmt.
22133 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22135 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22137 PR middle-end/46932
22138 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22140 2017-07-26 Martin Liska <mliska@suse.cz>
22143 * function.c (expand_function_start): Make expansion of
22144 nonlocal_goto_save_area after parm_birth_insn.
22146 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22148 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22149 from all CPU target flags enable members.
22151 2017-07-26 Richard Biener <rguenther@suse.de>
22153 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22154 (decision_tree::gen): Make 'type' const.
22155 (write_predicate): Likewise.
22157 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22159 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22160 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22161 (rs6000_option_override_internal): Likewise.
22162 (rs6000_expand_vector_set): Likewise.
22163 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22164 (TARGET_UPPER_REGS_SF): Likewise.
22165 (TARGET_UPPER_REGS_DI): Likewise.
22166 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22167 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22168 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22169 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22170 (Splitters for DI constants in Altivec registers): Likewise.
22171 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22172 (vsx_set_v4sf_p9): Likewise.
22173 (vsx_set_v4sf_p9_zero): Likewise.
22174 (vsx_insert_extract_v4sf_p9): Likewise.
22175 (vsx_insert_extract_v4sf_p9_2): Likewise.
22177 2017-07-25 Carl Love <cel@us.ibm.com>
22179 * doc/extend.texi: Update the built-in documentation file for the
22180 existing built-in functions
22181 vector signed char vec_cnttz (vector signed char);
22182 vector unsigned char vec_cnttz (vector unsigned char);
22183 vector signed short vec_cnttz (vector signed short);
22184 vector unsigned short vec_cnttz (vector unsigned short);
22185 vector signed int vec_cnttz (vector signed int);
22186 vector unsigned int vec_cnttz (vector unsigned int);
22187 vector signed long long vec_cnttz (vector signed long long);
22188 vector unsigned long long vec_cnttz (vector unsigned long long);
22190 2017-07-25 Andrew Pinski <apinski@cavium.com>
22192 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22193 accesses where the use is for the first operand of a BIT_INSERT.
22195 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22198 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22199 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22201 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22203 * config/i386/gstabs.h: Delete.
22204 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22206 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22208 * config/i386/i386.c (ix86_decompose_address): Do not check for
22209 register RTX when looking at index_reg or base_reg.
22210 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22212 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22214 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22215 to update EH info here.
22217 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22219 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22221 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22223 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22225 2017-07-25 Torsten Duwe <duwe@suse.de>
22227 * common.opt: Introduce -fpatchable-function-entry
22228 command line option, and its variables function_entry_patch_area_size
22229 and function_entry_patch_area_start.
22230 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22231 including a two-value parser.
22232 * target.def (print_patchable_function_entry): New target hook.
22233 * targhooks.h (default_print_patchable_function_entry): New function.
22234 * targhooks.c (default_print_patchable_function_entry): Likewise.
22235 * toplev.c (process_options): Switch off IPA-RA if
22236 patchable function entries are being generated.
22237 * varasm.c (assemble_start_function): Look at the
22238 patchable-function-entry command line switch and current
22239 function attributes and maybe generate NOP instructions by
22240 calling the print_patchable_function_entry hook.
22241 * doc/extend.texi: Document patchable_function_entry attribute.
22242 * doc/invoke.texi: Document -fpatchable_function_entry
22243 command line option.
22244 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22246 * doc/tm.texi: Re-generate.
22248 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22251 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22252 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22254 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22256 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22258 2017-07-25 Richard Biener <rguenther@suse.de>
22260 PR tree-optimization/81455
22261 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22262 not walk in cycles when looking for guards.
22264 2017-07-25 Richard Biener <rguenther@suse.de>
22266 PR tree-optimization/81529
22267 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22268 when optimizing backedge uses.
22270 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22272 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22274 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22275 to dl_section_ref. On AIX, append an expression to subtract
22276 the size of the section length to dl_section_ref.
22278 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22280 * configure.ac: If any of the config.* scripts fail, exit 1.
22281 * configure: Regenerate.
22283 2017-07-25 Richard Biener <rguenther@suse.de>
22285 PR middle-end/81546
22286 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22287 of immediate uses, be more verbose on errors.
22289 2017-07-25 Richard Biener <rguenther@suse.de>
22291 PR tree-optimization/81510
22292 * tree-vect-loop.c (vect_is_simple_reduction): When the
22293 reduction stmt is not inside the loop bail out.
22295 2017-07-25 Richard Biener <rguenther@suse.de>
22297 PR tree-optimization/81303
22298 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22299 profitability check against LOOP_VINFO_NITERSM1.
22301 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22303 * domwalk.c (cmp_bb_postorder): Simplify.
22304 (sort_bbs_postorder): New function. Use it...
22305 (dom_walker::walk): ...here to optimize common cases.
22307 2017-07-25 Martin Liska <mliska@suse.cz>
22310 * ipa-visibility.c (function_and_variable_visibility): Make the
22311 redirection just on target that supports aliasing.
22312 Fix GNU coding style.
22314 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22317 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22319 * config/arm/rtems.h: Likewise.
22320 * config/bfin/rtems.h: Likewise.
22321 * config/i386/rtemself.h: Likewise.
22322 * config/lm32/rtems.h: Likewise.
22323 * config/m32c/rtems.h: Likewise.
22324 * config/m68k/rtemself.h: Likewise.
22325 * config/microblaze/rtems.h: Likewise.
22326 * config/mips/rtems.h: Likewise.
22327 * config/moxie/rtems.h: Likewise.
22328 * config/nios2/rtems.h: Likewise.
22329 * config/powerpcspe/rtems.h: Likewise.
22330 * config/rs6000/rtems.h: Likewise.
22331 * config/rtems.h: Likewise.
22332 * config/sh/rtems.h: Likewise.
22333 * config/sh/rtemself.h: Likewise.
22334 * config/sparc/rtemself.h: Likewise.
22336 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22339 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22340 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22341 * tree-ssa-structalias.c (alias_get_name): Same.
22343 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22346 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22347 instructions if no du chain is found.
22349 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22351 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22353 2017-07-25 Richard Biener <rguenther@suse.de>
22355 PR middle-end/81505
22356 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22359 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22361 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22362 upper-regs options.
22363 (ISA_2_7_MASKS_SERVER): Likewise.
22364 (ISA_3_0_MASKS_IEEE): Likewise.
22365 (OTHER_P8_VECTOR_MASKS): Likewise.
22366 (OTHER_VSX_VECTOR_MASKS): Likewise.
22367 (POWERPC_MASKS): Likewise.
22368 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22369 duplicate list of options.
22370 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22371 explicit -mupper-regs options.
22372 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22373 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22374 alias for -mupper-regs-df.
22375 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22376 (rs6000_init_hard_regno_mode_ok): Likewise.
22377 (rs6000_option_override_internal): Likewise.
22378 (rs6000_opt_masks): Likewise.
22379 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22380 options in terms of whether -mvsx or -mpower8-vector was used.
22381 (TARGET_UPPER_REGS_DI): Likewise.
22382 (TARGET_UPPER_REGS_SF): Likewise.
22383 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22384 -mupper-regs-* options.
22386 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22388 * passes.c (emergency_dump_function): Print some empty lines and a
22389 header before the RTL dump.
22391 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22393 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22395 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22398 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22399 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22401 2017-07-24 Carl Love <cel@us.ibm.com>
22403 * config/rs6000/rs6000-c.c: Add support for built-in functions
22404 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22405 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22406 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22407 vec_extract_fp_from_shortl): Add defines for the two builtins.
22408 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22409 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22411 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22412 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22413 * doc/extend.texi: Update the built-in documentation file for the
22414 new built-in function.
22416 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22419 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22421 * doc/generic.texi: Likewise.
22422 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22423 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22425 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22427 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22428 (aarch64_mls_elt_merge<mode>): Likewise.
22430 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22432 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22433 having __cxa_atexit.
22435 2017-07-23 Michael Collison <michael.collison@arm.com>
22437 * config/arm/arm.c (arm_option_override): Deprecate
22438 use of -mstructure-size-boundary.
22439 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22440 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22442 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22445 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22446 Reduce cost estimate for direct moves.
22448 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22451 * config/i386/i386.c (ix86_option_override_internal): Disable
22452 BMI, BMI2 and TBM instructions for -m16.
22454 2017-07-21 Carl Love <cel@us.ibm.com>
22456 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22457 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22458 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22459 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22460 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22461 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22462 VMULOSW): New enum "unspec" values.
22463 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22464 altivec_vmulosw): New patterns.
22465 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22466 VMULOSW): Add definitions.
22468 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22470 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22471 (qdf24xx): Likewise.
22472 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22473 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22474 (AARCH64_FL_V8_1): Renumber.
22475 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22476 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22477 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22478 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22479 rdma to feature modifiers list.
22481 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22483 PR middle-end/56727
22484 * ipa-visibility (function_and_variable_visibility): Convert
22485 recursive PLT call to direct call if appropriate.
22487 2017-07-21 Andrew Pinski <apinski@cavium.com>
22489 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22490 operand 1 to see if the types precision matches.
22491 * fold-const.c (operand_equal_p): Likewise.
22493 2017-07-21 Richard Biener <rguenther@suse.de>
22495 PR tree-optimization/81303
22496 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22497 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22498 (vect_peeling_hash_get_lowest_cost): Adjust.
22499 (vect_enhance_data_refs_alignment): Likewise. Use
22500 vect_get_peeling_costs_all_drs to compute the penalty for no
22501 peeling to match up costs.
22503 2017-07-21 Richard Biener <rguenther@suse.de>
22505 PR tree-optimization/81500
22506 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22507 we didn't identify a reduction path.
22509 2017-07-21 Tom de Vries <tom@codesourcery.com>
22510 Cesar Philippidis <cesar@codesourcery.com>
22512 PR gcov-profile/81442
22513 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22516 2017-07-21 Tom de Vries <tom@codesourcery.com>
22519 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22521 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22522 nvptx_override_options_after_change.
22524 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22526 * dwarf2out.c (output_file_names): Avoid double testing for
22527 dwarf_version >= 5.
22529 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22531 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22533 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22535 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22537 (try_crossjump_to_edge): Do not punt on partitioned functions.
22539 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22541 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22542 Put all BBs reachable only via paths crossing cold region to cold
22544 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22546 2016-07-21 Richard Biener <rguenther@suse.de>
22548 PR tree-optimization/81303
22549 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22550 into account prologue and epilogue iterations when raising
22551 min_profitable_iters to sth at least covering one vector iteration.
22553 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22555 * config/arm/arm.c (arm_test_cpu_arch_dat):
22558 2017-07-20 Nathan Sidwell <nathan@acm.org>
22560 Remove TYPE_METHODS.
22561 * tree.h (TYPE_METHODS): Delete.
22562 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22563 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22564 (dbxout_type_methods): Scan TYPE_FIELDS.
22565 (dbxout_type): Don't check TYPE_METHODS here.
22566 * function.c (use_register_for_decl): Always ignore register for
22567 class types when not optimizing.
22568 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22569 * tree.c (free_lang_data_in_type): Stitch out member functions and
22570 templates from TYPE_FIELDS.
22571 (build_distinct_type_copy, verify_type_variant,
22572 verify_type): Member fns are on TYPE_FIELDS.
22573 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22574 * tree-pretty-print.c (dump_generic_node): Likewise.
22576 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22579 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22580 V2TImode and V4TImode.
22581 (ix86_expand_vector_extract): Likewise.
22582 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22583 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22584 (ssescalarmode): Handle V4TImode and V2TImode.
22585 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22586 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22587 (VEXTRACTI128_MODE): New mode iterator.
22588 (splitter for *vec_extractv?ti first element): New.
22589 (VEC_INIT_MODE): New mode iterator.
22590 (vec_init<mode>): Consolidate 3 expanders into one using
22591 VEC_INIT_MODE mode iterator.
22593 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22595 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22596 non_spilled_static_chain_regno_p.
22598 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22600 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22602 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22604 * bb-reorder.c (connect_traces): Allow copying of blocks within
22607 2017-07-20 Richard Biener <rguenther@suse.de>
22609 * gimple.h (gimple_phi_result): Add gphi * overload.
22610 (gimple_phi_result_ptr): Likewise.
22611 (gimple_phi_arg): Likewise. Adjust index assert to only
22612 allow actual argument accesses rather than all slots available
22614 (gimple_phi_arg_def): Add gphi * overload.
22615 * tree-phinodes.c (make_phi_node): Initialize only actual
22617 (resize_phi_node): Clear memory not covered by old node,
22618 do not initialize excess argument slots.
22619 (reserve_phi_args_for_new_edge): Initialize new argument slot
22622 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22624 PR tree-optimization/81388
22626 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22628 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22629 by removing computation of may_be_zero.
22631 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22632 Tom de Vries <tom@codesourcery.com>
22634 PR middle-end/81030
22635 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22636 when gimple level profile disagrees with what RTL expander did.
22638 2017-07-20 Richard Biener <rguenther@suse.de>
22640 PR tree-optimization/61171
22641 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22642 (vect_analyze_stmt): Add slp instance parameter.
22643 (vectorizable_reduction): Likewise.
22644 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22645 (vect_is_simple_reduction): Deal with chains not detected
22646 as SLP reduction chain, specifically not properly associated
22647 chains containing a mix of plus/minus.
22648 (get_reduction_op): Remove.
22649 (get_initial_defs_for_reduction): Simplify, pass in whether
22650 this is a reduction chain, pass in the SLP node for the PHIs.
22651 (vect_create_epilog_for_reduction): Get the SLP instance as
22653 (vectorizable_reduction): Get the SLP instance as arg.
22654 During analysis remember the SLP node with the PHIs in the
22655 instance. Simplify getting at the vectorized reduction PHIs.
22656 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22657 through SLP instance.
22658 (vect_slp_analyze_operations): Likewise.
22659 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22660 (vect_transform_stmt): Likewise.
22662 2017-07-20 Tom de Vries <tom@codesourcery.com>
22664 PR tree-optimization/81489
22665 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22666 read of phi arg location to before loop that modifies phi.
22668 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22670 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22673 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22675 PR middle-end/81331
22676 * except.c (execute): Fix ordering issue.
22678 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22680 PR rtl-optimization/81423
22681 * combine.c (make_compound_operation_int): Don't try to optimize
22682 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22684 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22686 PR rtl-optimization/81423
22687 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22688 with a constant that is -1 in the truncated to mode.
22690 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22692 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22693 (determine_unlikely_bbs): ... here.
22694 * predict.h (propagate_unlikely_bbs_forward): Declare.
22695 * cfgexpand.c (pass_expand::execute): Use it.
22696 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22698 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22699 propagate_unlikely_bbs_forward.
22701 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22703 PR middle-end/81331
22704 * except.c (maybe_add_nop_after_section_switch): New function.
22707 2017-07-19 Tom de Vries <tom@codesourcery.com>
22709 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22711 2017-07-19 Tom de Vries <tom@codesourcery.com>
22713 * gimple.h (gimple_phi_arg): Make assert more strict.
22715 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22717 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22718 * config/rs6000/mmintrin.h: New file.
22719 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22721 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22723 PR tree-optimization/81346
22724 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22726 2017-07-19 Tom de Vries <tom@codesourcery.com>
22728 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22730 2017-07-19 Tom de Vries <tom@codesourcery.com>
22732 * config/nvptx/nvptx-modes.def: Add V2DImode.
22733 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22734 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22735 (nvptx_output_mov_insn): Handle lack of mov.b128.
22736 (nvptx_print_operand): Handle 'H' and 'L' codes.
22737 (nvptx_vector_mode_supported): Allow V2DImode.
22738 (nvptx_preferred_simd_mode): New function.
22739 (nvptx_data_alignment): New function.
22740 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22741 nvptx_preferred_simd_mode.
22742 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22744 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22746 2017-07-19 Tom de Vries <tom@codesourcery.com>
22748 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22749 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22750 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22751 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22752 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22753 (mov<VECIM>_insn): New define_insn.
22754 (define_expand "mov<VECIM>): New define_expand.
22756 2017-07-19 Tom de Vries <tom@codesourcery.com>
22758 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22760 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22762 PR tree-optimization/81346
22763 * fold-const.h (fold_div_compare, range_check_type): Declare.
22764 * fold-const.c (range_check_type): New function.
22765 (build_range_check): Use range_check_type.
22766 (fold_div_compare): No longer static, rewritten into
22767 a match.pd helper function.
22768 (fold_comparison): Don't call fold_div_compare here.
22769 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22770 as helper function.
22772 2017-07-19 Nathan Sidwell <nathan@acm.org>
22774 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22775 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22776 * tree.c (find_decls_types_r, verify_type): Use
22777 TYPE_{MIN,MAX}_VALUE_RAW.
22778 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22779 (hash_tree): Likewise.
22780 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22782 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22785 2017-07-18 Tom de Vries <tom@codesourcery.com>
22787 PR middle-end/81464
22788 * omp-expand.c (expand_omp_for_static_chunk): Handle
22789 equal-argument loop exit phi.
22791 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22794 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22795 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22796 operand 2 predicate.
22797 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22798 operand 2 predicate.
22799 (ror,rol -> rorx splitters): Use const_int_operand as
22800 operand 2 predicate.
22802 2017-06-18 Richard Biener <rguenther@suse.de>
22804 PR tree-optimization/81410
22805 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22806 the gap in the ! slp_perm SLP case after each group.
22808 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22810 PR middle-end/81463
22811 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22814 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22816 PR middle-end/81462
22817 * predict.c (set_even_probabilities): Cleanup; do not affect
22818 probabilities that are already known.
22819 (combine_predictions_for_bb): Call even when count is set.
22821 2017-07-18 Nathan Sidwell <nathan@acm.org>
22823 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22826 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22829 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22830 optimization for loop niter analysis.
22832 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22835 * config/avr/avr.c (avr_optimize_casesi): Don't use
22836 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22838 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22840 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22841 body_cost_vec from _vect_peel_extended_info.
22842 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22843 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22846 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22848 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22850 2017-07-18 Richard Biener <rguenther@suse.de>
22852 PR tree-optimization/80620
22853 PR tree-optimization/81403
22854 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22855 info when re-using a VN table entry.
22857 2017-07-18 Richard Biener <rguenther@suse.de>
22859 PR tree-optimization/81418
22860 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22861 vectype_in. Verify that with lane-reducing reduction operations
22862 we have a single def-use cycle.
22864 2017-07-17 Carl Love <cel@us.ibm.com>
22866 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22868 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22869 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22870 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22872 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22873 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22874 VMULOSW): New enum "unspec" values.
22875 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22876 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22877 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22878 altivec_vmulosw): New patterns.
22879 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22880 VMULOSW): Add definitions.
22882 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22884 * config/alpha/alpha.c: Include predict.h.
22886 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22888 * tree-vrp.c (compare_assert_loc): Fix comparison function
22889 to return predictable results.
22891 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22893 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22895 (subdi3): Likewise.
22896 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22897 * doc/invoke.texi (mexpand-adddi): Update text.
22899 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22901 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22902 that also clobbers the CC register. The old expand code is moved
22904 (*arc_clzsi2): ... here.
22905 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22906 the CC register. The old expand code is moved to ...
22907 (arc_ctzsi2): ... here.
22909 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22911 * config/arc/arc.opt (mindexed-loads): Use initial value
22912 TARGET_INDEXED_LOADS_DEFAULT.
22913 (mauto-modify-reg): Use initial value
22914 TARGET_AUTO_MODIFY_REG_DEFAULT.
22915 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22916 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22917 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22918 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22920 2017-07-17 Martin Liska <mliska@suse.cz>
22923 * opts.c (finish_options): Do not allow -fgnu-tm
22924 w/ -fsanitize={kernel-,}address. Say sorry.
22926 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22929 * tree-loop-distribution.c (classify_partition): Only assert on
22930 numer of iterations.
22931 (merge_dep_scc_partitions): Delete prameter. Update function call.
22932 (distribute_loop): Remove code handling loop with unknown niters.
22933 (pass_loop_distribution::execute): Skip loop with unknown niters.
22935 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22938 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22939 function sort_partitions_by_post_order.
22941 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22943 PR tree-optimization/81374
22944 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22945 the max index of basic blocks, rather than number of basic blocks.
22947 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22949 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22951 (arc_legitimate_pic_operand_p): Likewise.
22952 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22954 (arc_needs_pcl_p): Likewise.
22955 (arc_legitimate_pc_offset_p): Likewise.
22956 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22957 function is also used in constrains.md.
22958 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22959 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22960 PLUS. Only return true/false in known cases, otherwise assert.
22961 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22962 is already called in arc_legitimate_constant_p.
22963 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22965 (LEGITIMATE_PIC_OPERAND_P): Use
22966 arc_raw_symbolic_reference_mentioned_p function.
22967 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22972 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22973 Andrew Burgess <andrew.burgess@embecosm.com>
22975 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22976 (arc_return_address_register): New function.
22977 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22978 (arc_handle_fndecl_attribute): Add naked attribute.
22979 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22980 (TARGET_WARN_FUNC_RETURN): Likewise.
22981 (arc_allocate_stack_slots_for_args): New function.
22982 (arc_warn_func_return): Likewise.
22983 (machine_function): Change type fn_type.
22984 (arc_compute_function_type): Consider new naked function type,
22985 change function return type.
22986 (arc_must_save_register): Adapt to handle new
22987 arc_compute_function_type's return type.
22988 (arc_expand_prologue): Likewise.
22989 (arc_expand_epilogue): Likewise.
22990 (arc_return_address_regs): Delete.
22991 (arc_return_address_register): New function.
22992 (arc_epilogue_uses): Use above function.
22993 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22994 (arc_function_type): Change encoding, add naked type.
22995 (ARC_INTERRUPT_P): Change to handle the new encoding.
22996 (ARC_FAST_INTERRUPT_P): Likewise.
22997 (ARC_NORMAL_P): Define.
22998 (ARC_NAKED_P): Likewise.
22999 (arc_compute_function_type): Delete prototype.
23000 * config/arc/arc.md (in_ret_delay_slot): Use
23001 arc_return_address_register function.
23002 (simple_return): Likewise.
23003 (p_return_i): Likewise.
23005 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23007 PR tree-optimization/81428
23008 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23009 can't be built for those types.
23011 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23013 Remove stuff dead since r239246.
23015 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23016 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23017 (avr_inform_devices): Remove dead stuff.
23019 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23021 * config/arm/arm_neon.h: Fix softp typo.
23023 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23025 PR tree-optimization/81365
23026 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23027 aggregate moves onto bb predecessor edges, make sure there are no
23028 loads that could alias the lhs in between the start of bb and the
23031 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23034 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23035 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23036 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23038 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23040 PR tree-optimization/81396
23041 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23042 (init_symbolic_number): Initialize it to 1.
23043 (perform_symbolic_merge): Add n_ops from both operands into the new
23045 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23046 without base_addr as useless if they need more than one operation.
23047 (bswap_replace): Handle !bswap case for NULL base_addr.
23049 2017-07-17 Tom de Vries <tom@codesourcery.com>
23052 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23055 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23057 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23058 conditional builtin define __FIX_LEON3FT_B2BST.
23060 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23062 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23063 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23066 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23068 PR rtl-optimization/81424
23069 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23070 to remove potential trapping from operands if -fnon-call-exceptions.
23072 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23074 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23075 profile_proability for scalling.
23076 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23078 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23080 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23082 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23084 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23085 fixpoint arithmetics.
23087 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23089 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23090 fixpoint arithmetics.
23092 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23094 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23095 fixpoint arithmetics.
23097 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23099 * profile-count.h (profile_probability::from_reg_br_prob_note,
23100 profile_probability::to_reg_br_prob_note): New functions.
23101 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23102 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23103 * predict.c (probability_reliable_p): Update.
23104 (edge_probability_reliable_p): Update.
23105 (br_prob_note_reliable_p): Update.
23106 (invert_br_probabilities): Update.
23107 (add_reg_br_prob_note): New function.
23108 (combine_predictions_for_insn): Update.
23109 * asan.c (asan_clear_shadow): Update.
23110 * cfgbuild.c (compute_outgoing_frequencies): Update.
23111 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23112 (update_br_prob_note): Update.
23113 (rtl_verify_edges): Update.
23114 (purge_dead_edges): Update.
23115 (fixup_reorder_chain): Update.
23116 * emit-rtl.c (try_split): Update.
23117 * ifcvt.c (cond_exec_process_insns): Update.
23118 (cond_exec_process_if_block): Update.
23119 (dead_or_predicable): Update.
23120 * internal-fn.c (expand_addsub_overflow): Update.
23121 (expand_neg_overflow): Update.
23122 (expand_mul_overflow): Update.
23123 * loop-doloop.c (doloop_modify): Update.
23124 * loop-unroll.c (compare_and_jump_seq): Update.
23125 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23126 * predict.h: Update.
23127 * reorg.c (mostly_true_jump): Update.
23129 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23130 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23131 * config/arc/arc.c: (emit_unlikely_jump): Update.
23132 * config/arm/arm.c: (emit_unlikely_jump): Update.
23133 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23134 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23135 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23136 (ix86_print_operand): Update.
23137 (ix86_split_fp_branch): Update.
23138 (predict_jump): Update.
23139 * config/ia64/ia64.c (ia64_print_operand): Update.
23140 * config/mmix/mmix.c (mmix_print_operand): Update.
23141 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23142 (rs6000_expand_split_stack_prologue): Update.
23143 * config/rs6000/rs6000.c: Update.
23144 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23145 (s390_expand_vec_movstr): Update.
23146 (s390_expand_cs_tdsi): Update.
23147 (s390_expand_split_stack_prologue): Update.
23148 * config/sh/sh.c (sh_print_operand): Update.
23149 (expand_cbranchsi4): Update.
23150 (expand_cbranchdi4): Update.
23151 * config/sparc/sparc.c (output_v9branch): Update.
23152 * config/spu/spu.c (get_branch_target): Update.
23153 (ea_load_store_inline): Update.
23154 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23155 * config/tilepro/tilepro.c: Update.
23157 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23159 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23160 (gimplify_save_expr): Likewise.
23162 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23164 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23166 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23168 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23170 * timevar.def (TV_IPA_FNSUMMARY): Define.
23172 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23174 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23175 to back store errata sensitive sequence from being generated.
23176 (sqrtdf2_fix): Likewise.
23178 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23180 * tree-ssa-threadupdate.c (compute_path_counts,
23181 update_joiner_offpath_counts): Use profile_probability.
23183 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23186 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23188 * config/arm/arm-c.c (arm_cpu_builtins): Define
23189 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23191 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23193 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23194 array entries to represent __ieee128 versions of the
23195 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23196 scalar_extract_sig, and scalar_insert_exp built-in functions.
23197 (altivec_resolve_overloaded_builtin): Add special case handling
23198 for the __builtin_scalar_insert_exp function, as represented by
23199 the P9V_BUILTIN_VEC_VSIEDP constant.
23200 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23201 exponent support for __ieee128 argument.
23202 (VSESQP): Add scalar extract signature support for __ieee128
23204 (VSTDCNQP): Add scalar test negative support for __ieee128
23206 (VSIEQP): Add scalar insert exponent support for __int128 argument
23207 with __ieee128 result.
23208 (VSIEQPF): Add scalar insert exponent support for __ieee128
23209 argument with __ieee128 result.
23210 (VSTDCQP): Add scalar test data class support for __ieee128
23212 (VSTDCNQP): Add overload support for scalar test negative with
23213 __ieee128 argument.
23214 (VSTDCQP): Add overload support for scalar test data class
23215 __ieee128 argument.
23216 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23217 UNSPEC_VSX_SXSIGDP.
23218 (UNSPEC_VSX_SIEXPQP): New constant.
23219 (xsxexpqp): New insn for VSX scalar extract exponent quad
23221 (xsxsigqp): New insn for VSX scalar extract significand quad
23223 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23224 precision with floating point argument.
23225 (xststdcqp): New expand for VSX scalar test data class quad
23227 (xststdcnegqp): New expand for VSX scalar test negative quad
23229 (xststdcqp): New insn to match expansions for VSX scalar test data
23230 class quad precision and VSX scalar test negative quad precision.
23231 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23232 special case operand checking to enforce that second operand of
23233 VSX scalar test data class with quad precision argument is a 7-bit
23235 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23236 prototypes and descriptions of __ieee128 versions of
23237 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23238 scalar_test_data_class, and scalar_test_neg built-in functions.
23240 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23242 PR tree-optimization/81162
23243 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23244 replace a negate with an add.
23246 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23248 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23250 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23252 * config/arm/arm-c.c (arm_cpu_builtins): Define
23253 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23255 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23257 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23258 (armv8-r): Set ARM Cortex-R52 as default CPU.
23259 * config/arm/arm-tables.opt: Regenerate.
23260 * config/arm/arm-tune.md: Regenerate.
23261 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23263 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23264 extension for -mcpu=cortex-r52.
23266 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23268 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23269 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23270 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23271 (fp-armv8): Define it as FP_ARMv8 only.
23272 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23273 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23275 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23276 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23277 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23278 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23279 than TARGET_FPU_ARMV8.
23280 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23281 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23282 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23284 * config/arm/neon.md (neon_vrint): Likewise.
23285 (neon_vcvt): Likewise.
23286 (neon_<fmaxmin_op><mode>): Likewise.
23287 (<fmaxmin><mode>3): Likewise.
23288 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23289 * config/arm/predicates.md (arm_cond_move_operator): Check against
23290 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23292 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23294 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23295 to top of function.
23297 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23299 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23300 loop in comment with memset.
23302 2017-07-14 Martin Liska <mliska@suse.cz>
23304 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23305 * dwarf2out.c (is_java): Remove the function.
23306 (output_pubname): Remove usage of the function.
23307 (lower_bound_default): Remove usage of DW_LANG_Java.
23308 (gen_compile_unit_die): Likewise.
23309 * gcc.c: Remove compiler defaults for .java and .zip files.
23310 * gimple-expr.c (remove_suffix): Change as there's no longer
23311 extension than 4-letter one.
23312 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23313 (gimplify_save_expr): Likewise.
23314 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23315 as it's possible even for other languages than Java.
23316 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23317 * lto-opts.c (lto_write_options): Remove reference to Java.
23318 * opts.c (strip_off_ending): Update file extension handling.
23319 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23320 * tree-eh.c (lower_resx): Likewise.
23321 * tree.c (free_lang_data_in_type): Remove dead code.
23322 (find_decls_types_r): Likewise.
23323 (build_common_builtin_nodes): Remove Java from a comment.
23324 (verify_type): Remove dead code.
23325 * varasm.c (assemble_external): Remove Java from a comment.
23327 2017-07-14 Martin Liska <mliska@suse.cz>
23329 * opts.c (finish_options): Add quotes.
23330 (common_handle_option): Likewise.
23332 2017-07-14 Martin Liska <mliska@suse.cz>
23334 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23335 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23336 Remove N_SO_PASCAL.
23337 * dwarf2out.c (lower_bound_default): Do not handle
23339 (gen_compile_unit_die): Likewise.
23340 * gcc.c: Remove default extension binding for GNU Pascal.
23341 * stmt.c: Remove Pascal language from a comment.
23342 * xcoffout.c: Likewise.
23344 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23347 * diagnostic-show-locus.c (fixit_cmp): New function.
23348 (layout::layout): Sort m_fixit_hints.
23349 (column_range::column_range): Assert that the values are valid.
23350 (struct char_span): New struct.
23351 (correction::overwrite): New method.
23352 (struct source_line): New struct.
23353 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23354 calls in terms of classes source_line and char_span, and
23355 correction::overwrite.
23356 (selftest::test_overlapped_fixit_printing_2): New function.
23357 (selftest::diagnostic_show_locus_c_tests): Call it.
23359 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23361 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23362 early if there is no lhs.
23364 2017-07-13 Martin Liska <mliska@suse.cz>
23366 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23367 (gen_reference_type_die): Likewise.
23368 * stor-layout.c: Remove Pascal-related comment.
23370 2017-07-13 Martin Liska <mliska@suse.cz>
23372 * opts.c (finish_options): Add quotes to error messages.
23373 (parse_sanitizer_options): Likewise.
23375 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23377 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23379 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23381 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23383 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23385 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23387 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23389 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23392 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23393 provides the hardware capability bits, define the macro
23394 __BUILTIN_CPU_SUPPORTS__.
23395 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23396 if GLIBC does not provide the hardware capability bits. Add a
23397 gcc_unreachable call if the built-in cpu function is neither
23398 __builtin_cpu_is nor __builtin_cpu_supports.
23399 (rs6000_get_function_versions_dispatcher): Change the warning
23400 that an old GLIBC is used which does not export the capability
23401 bits to be an error.
23402 * doc/extend.texi (target_clones attribute): Document the
23403 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23404 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23405 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23406 the macros defined by GCC if the newer GLIBC is available.
23408 2017-07-12 Jeff Law <law@redhat.com>
23410 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23411 remaining includes slightly.
23412 * config/riscv/riscv-builtins.c: Include profile-count.h.
23414 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23417 * config/avr/avr.c (avr_set_current_function): In diagnostic
23418 messages: Quote keywords and (parts of) identifiers.
23419 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23422 2017-07-12 Carl Love <cel@us.ibm.com>
23424 * config/rs6000/rs6000-c.c: Add support for built-in functions
23425 vector bool char vec_revb (vector bool char);
23426 vector bool short vec_revb (vector short char);
23427 vector bool int vec_revb (vector bool int);
23428 vector bool long long vec_revb (vector bool long long);
23429 * doc/extend.texi: Update the built-in documentation file for the
23430 new built-in functions.
23432 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23434 * config/s390/s390.md: Remove movcc splitter.
23436 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23438 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23439 load/store on condition.
23441 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23444 * config/avr/avr.c (avr_encode_section_info)
23445 [progmem && !TREE_READONLY]: Error if progmem object needs
23448 2017-07-11 Michael Collison <michael.collison@arm.com>
23450 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23453 2017-07-11 Carl Love <cel@us.ibm.com>
23455 * config/rs6000/rs6000-c.c: Add support for builtins
23456 vector unsigned int vec_parity_lsbb (vector signed int);
23457 vector unsigned int vec_parity_lsbb (vector unsigned int);
23458 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23459 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23460 vector unsigned long long vec_parity_lsbb (vector signed long long);
23461 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23462 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23463 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23464 * doc/extend.texi: Update the built-in documentation file for the
23465 new built-in functions.
23467 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23469 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23470 (layout::m_primary_loc): New field.
23471 (layout::layout): Initialize new field. Move location filtering
23472 logic from here to...
23473 (layout::maybe_add_location_range): ...this new method. Add
23474 support for filtering to just the lines already specified by other
23476 (layout::will_show_line_p): New method.
23477 (gcc_rich_location::add_location_if_nearby): New method.
23478 (selftest::test_add_location_if_nearby): New test function.
23479 (selftest::diagnostic_show_locus_c_tests): Call it.
23480 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23483 2017-07-11 Tom de Vries <tom@codesourcery.com>
23485 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23486 (bb_first_real_insn): New function.
23487 (nvptx_single): Add extra initialization of broadcasted condition
23490 2017-07-11 Nathan Sidwell <nathan@acm.org>
23492 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23494 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23496 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23497 Binutils doc as TEXI will mess them up.
23498 * doc/invoke.texi (AVR Options): Same here.
23500 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23502 * config/sparc/sparc.opt (mfix-ut700): New option.
23503 (mfix-gr712rc): Likewise.
23504 (sparc_fix_b2bst): New variable.
23505 * doc/invoke.texi (SPARC options): Document them.
23506 (ARM options): Fix warnings.
23507 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23508 instructions to prevent sequences that can trigger the store-store
23509 errata for certain LEON3FT processors.
23510 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23511 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23512 * config/sparc/sparc.md (fix_b2bst): New attribute.
23513 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23515 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23518 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23520 (*rsqrtsf2_sse): Ditto.
23522 (div<mode>3): Macroize insn from divdf3 and divsf3
23523 using MODEF mode iterator.
23525 2017-07-10 Martin Sebor <msebor@redhat.com>
23527 PR tree-optimization/80397
23528 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23529 instead of testing for equality to INTEGER_TYPE.
23531 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23533 * config.gcc: Remove uclibc from arc target spec.
23535 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23537 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23539 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23542 * lto-wrapper.c (remove_option): New function.
23543 (merge_and_complain): Merge PIC/PIE options more realistically.
23545 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23547 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23551 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23552 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23553 * config.in: Regenerate.
23554 * configure: Regenerate.
23555 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23556 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23557 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23558 (TARGET_GASISR_PROLOGUES): ...target mask.
23559 * common/config/avr/avr-common.c
23560 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23561 Set -mgas-isr-prologues.
23562 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23563 INSERT_PASS_BEFORE for it.
23564 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23565 * config/avr/avr.c (avr_option_override)
23566 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23567 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23568 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23569 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23570 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23572 (make_avr_pass_pre_proep): New function.
23573 (emit_push_sfr) <treg>: Add argument to function and use it
23574 instead of TMP_REG.
23575 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23576 and set machine->gasisr.yes.
23577 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23578 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23579 __gcc_isr.n_pushed to .L__stack_usage.
23580 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23581 (avr_asm_final_postscan_insn): ...this new static function.
23582 * config/avr/avr.h (machine_function)
23583 <is_no_gccisr, use_L__stack_usage>: New fields.
23584 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23585 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23586 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23587 (gasisr, *gasisr): New expander and insn.
23588 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23589 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23590 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23592 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23594 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23597 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23599 Move jump-tables out of .text again.
23602 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23603 (ASM_OUTPUT_ADDR_VEC): New function.
23604 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23605 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23606 INSN_ADDRESSes as asm comment.
23607 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23608 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23609 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23610 * config/avr/avr.md (*tablejump): Adjust comment.
23611 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23612 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23614 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23615 (avr_output_addr_vec): New proto.
23616 (avr_log_t) <insn_addresses>: New field.
23618 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23621 * config/i386/i386.c (ix86_function_arg_advance): Set
23622 outgoing_args_on_stack to true if there are outgoing arguments
23624 (ix86_function_arg): Likewise.
23625 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23626 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23627 * config/i386/i386.h (machine_function): Add
23628 outgoing_args_on_stack.
23630 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23632 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23633 supporting pthreds.
23634 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23636 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23638 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23639 (REAL_H): Remove $(MACHMODE_H).
23640 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23642 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23643 $(MACHMODE_H) and double-int.h.
23644 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23646 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23649 2017-07-07 Andrew Pinski <apinski@cavium.com>
23651 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23652 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23654 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23656 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23657 Add warning if GCC was not configured to link against a GLIBC that
23658 exports the hardware capability bits.
23659 (make_resolver_func): Make resolver function private and not a
23660 COMDAT function. Create the name with clone_function_name instead
23661 of make_unique_name.
23664 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23665 correct operand in doing the split.
23667 2017-07-07 Carl Love <cel@us.ibm.com>
23669 * config/rs6000/rs6000-c: Add support for built-in function
23670 vector unsigned short vec_pack_to_short_fp32 (vector float,
23672 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23673 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23674 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23675 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23676 (convert_4f32_8i16): Add define_expand.
23677 * doc/extend.texi: Update the built-in documentation file for the
23678 new built-in function.
23680 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23682 * config/sparc/m8.md: New file.
23683 * config/sparc/sparc.md: Include m8.md.
23685 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23687 * config/sparc/sparc.opt: New option -mvis4b.
23688 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23689 (sparc_option_override): Handle VIS4B.
23690 (enum sparc_builtins): Define
23691 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23692 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23693 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23694 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23695 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23696 (check_constant_argument): New function.
23697 (sparc_vis_init_builtins): Define builtins
23698 __builtin_vis_dictunpack{8,16,32},
23699 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23700 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23701 __builtin_vis_fpcmpde{8,16,32}shl and
23702 __builtin_vis_fpcmpur{8,16,32}shl.
23703 (sparc_expand_builtin): Check that the constant operands to
23704 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23705 constant and in range.
23706 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23708 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23709 (SPARC_IMM5_P): Likewise.
23710 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23711 (enabled): Handle vis4b.
23712 (UNSPEC_DICTUNPACK): New unspec.
23713 (UNSPEC_FPCMPSHL): Likewise.
23714 (UNSPEC_FPUCMPSHL): Likewise.
23715 (UNSPEC_FPCMPDESHL): Likewise.
23716 (UNSPEC_FPCMPURSHL): Likewise.
23717 (cpu_feature): New CPU feature `vis4b'.
23718 (dictunpack{8,16,32}): New insns.
23719 (FPCSMODE): New mode iterator.
23720 (fpcscond): New code iterator.
23721 (fpcsucond): Likewise.
23722 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23723 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23724 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23725 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23726 * config/sparc/constraints.md: Define constraints `q' for unsigned
23727 2-bit integer constants and `t' for unsigned 5-bit integer
23729 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23731 (imm5_operand_dictunpack16): Likewise.
23732 (imm5_operand_dictunpack32): Likewise.
23733 (imm2_operand): Likewise.
23734 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23735 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23736 ditunpack* and fpcmp*shl builtins.
23738 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23740 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23741 * config.in: Add HAVE_AS_SPARC6 define.
23742 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23744 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23746 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23747 (CPP_CPU_SPEC): Handle m8.
23748 (ASM_CPU_SPEC): Likewise.
23749 * config/sparc/sparc-opts.h (enum processor_type): Add
23751 * config/sparc/sparc.c (m8_costs): New struct.
23752 (sparc_option_override): Handle TARGET_CPU_m8.
23753 (sparc32_initialize_trampoline): Likewise.
23754 (sparc64_initialize_trampoline): Likewise.
23755 (sparc_issue_rate): Likewise.
23756 (sparc_register_move_cost): Likewise.
23757 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23758 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23759 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23760 (CPP_CPU_SPEC): Handle M8.
23761 (ASM_CPU_SPEC): Likewise.
23762 (AS_M8_FLAG): Define.
23763 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23764 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23765 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23767 * configure: Regenerate.
23768 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23771 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23773 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23775 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23776 ("*movdi_insn_sp32"): Do not set v3pipe.
23777 ("*movsi_insn"): Likewise.
23778 ("*movdi_insn_sp64"): Likewise.
23779 ("*movsf_insn"): Likewise.
23780 ("*movdf_insn_sp32"): Likewise.
23781 ("*movdf_insn_sp64"): Likewise.
23782 ("*zero_extendsidi2_insn_sp64"): Likewise.
23783 ("*sign_extendsidi2_insn"): Likewise.
23784 ("*mov<VM32:mode>_insn"): Likewise.
23785 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23786 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23787 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23788 ("<vlop:code><VL:mode>3"): Likewise.
23789 ("*not_<vlop:code><VL:mode>3"): Likewise.
23790 ("*nand<VL:mode>_vis"): Likewise.
23791 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23792 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23793 ("one_cmpl<VL:mode>2"): Likewise.
23794 ("faligndata<VM64:mode>_vis"): Likewise.
23795 ("alignaddrsi_vis"): Likewise.
23796 ("alignaddrdi_vis"): Likweise.
23797 ("alignaddrlsi_vis"): Likewise.
23798 ("alignaddrldi_vis"): Likewise.
23799 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23800 ("bmaskdi_vis"): Likewise.
23801 ("bmasksi_vis"): Likewise.
23802 ("bshuffle<VM64:mode>_vis"): Likewise.
23803 ("cmask8<P:mode>_vis"): Likewise.
23804 ("cmask16<P:mode>_vis"): Likewise.
23805 ("cmask32<P:mode>_vis"): Likewise.
23806 ("pdistn<P:mode>_vis"): Likewise.
23807 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23809 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23811 * config/sparc/sparc.md ("subtype"): New insn attribute.
23812 ("*wrgsr_sp64"): Set insn subtype.
23813 ("*rdgsr_sp64"): Likewise.
23814 ("alignaddrsi_vis"): Likewise.
23815 ("alignaddrdi_vis"): Likewise.
23816 ("alignaddrlsi_vis"): Likewise.
23817 ("alignaddrldi_vis"): Likewise.
23818 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23819 ("fexpand_vis"): Likewise.
23820 ("fpmerge_vis"): Likewise.
23821 ("faligndata<VM64:mode>_vis"): Likewise.
23822 ("bshuffle<VM64:mode>_vis"): Likewise.
23823 ("cmask8<P:mode>_vis"): Likewise.
23824 ("cmask16<P:mode>_vis"): Likewise.
23825 ("cmask32<P:mode>_vis"): Likewise.
23826 ("fchksm16_vis"): Likewise.
23827 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23828 ("fmean16_vis"): Likewise.
23829 ("fp<plusminus_insn>64_vis"): Likewise.
23830 ("<plusminus_insn>v8qi3"): Likewise.
23831 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23832 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23833 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23834 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23835 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23836 ("*movqi_insn"): Likewise.
23837 ("*movhi_insn"): Likewise.
23838 ("*movsi_insn"): Likewise.
23839 ("movsi_pic_gotdata_op"): Likewise.
23840 ("*movdi_insn_sp32"): Likewise.
23841 ("*movdi_insn_sp64"): Likewise.
23842 ("movdi_pic_gotdata_op"): Likewise.
23843 ("*movsf_insn"): Likewise.
23844 ("*movdf_insn_sp32"): Likewise.
23845 ("*movdf_insn_sp64"): Likewise.
23846 ("*zero_extendhisi2_insn"): Likewise.
23847 ("*zero_extendqihi2_insn"): Likewise.
23848 ("*zero_extendqisi2_insn"): Likewise.
23849 ("*zero_extendqidi2_insn"): Likewise.
23850 ("*zero_extendhidi2_insn"): Likewise.
23851 ("*zero_extendsidi2_insn_sp64"): Likewise.
23852 ("ldfsr"): Likewise.
23853 ("prefetch_64"): Likewise.
23854 ("prefetch_32"): Likewise.
23855 ("tie_ld32"): Likewise.
23856 ("tie_ld64"): Likewise.
23857 ("*tldo_ldub_sp32"): Likewise.
23858 ("*tldo_ldub1_sp32"): Likewise.
23859 ("*tldo_ldub2_sp32"): Likewise.
23860 ("*tldo_ldub_sp64"): Likewise.
23861 ("*tldo_ldub1_sp64"): Likewise.
23862 ("*tldo_ldub2_sp64"): Likewise.
23863 ("*tldo_ldub3_sp64"): Likewise.
23864 ("*tldo_lduh_sp32"): Likewise.
23865 ("*tldo_lduh1_sp32"): Likewise.
23866 ("*tldo_lduh_sp64"): Likewise.
23867 ("*tldo_lduh1_sp64"): Likewise.
23868 ("*tldo_lduh2_sp64"): Likewise.
23869 ("*tldo_lduw_sp32"): Likewise.
23870 ("*tldo_lduw_sp64"): Likewise.
23871 ("*tldo_lduw1_sp64"): Likewise.
23872 ("*tldo_ldx_sp64"): Likewise.
23873 ("*mov<VM32:mode>_insn"): Likewise.
23874 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23875 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23877 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23879 * config/sparc/sparc.md ("type"): New insn type viscmp.
23880 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23882 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23883 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23884 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23885 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23887 ("n7_vis_logical_11cycle"): Likewise.
23888 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23889 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23890 * config/sparc/niagara.md ("niag_vis"): Likewise.
23891 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23892 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23894 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23896 * config/sparc/sparc.md: New instruction type `bmask'.
23897 (bmaskdi_vis): Use the `bmask' type.
23898 (bmasksi_vis): Likewise.
23899 * config/sparc/ultra3.md (us3_array): Likewise.
23900 * config/sparc/niagara7.md (n7_array): Likewise.
23901 * config/sparc/niagara4.md (n4_array): Likewise.
23902 * config/sparc/niagara2.md (niag2_vis): Likewise.
23903 (niag3_vis): Likewise.
23904 * config/sparc/niagara.md (niag_vis): Likewise.
23906 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23908 * ipa-comdats.c: Remove optimize check from gate.
23909 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23910 for functions not optimized.
23911 (ipa_fn_summary_read): Skip optimize check.
23912 (ipa_fn_summary_write): Likewise.
23913 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23915 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23917 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23918 (check_callers): Give up on caller which is not optimized.
23919 (inline_small_functions): Likewise.
23920 (ipa_inline): Do not give up when not optimizing.
23921 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23922 away unoptimizes cdtors.
23923 (whole_program_function_and_variable_visibility): Do
23924 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23925 * ipa.c (process_references): Do not check optimize.
23926 (symbol_table::remove_unreachable_nodes): Update optimize check.
23927 (set_writeonly_bit): Update optimize check.
23928 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23929 (pass_ipa_single_use::gate): Remove.
23931 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23933 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23934 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23935 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23936 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23937 permute_load, permute_store, adjust_extract, adjust_splat,
23938 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23939 replace_swap_with_copy, dump_swap_insn_table,
23940 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23941 recombine_lvx_pattern, recombine_stvx_pattern,
23942 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23943 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23944 to file rs6000-p8swap.c.
23945 * config/rs6000/rs6000-p8swap.c: New file.
23946 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23947 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23948 and rs6000*-*-* targets.
23950 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23952 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23954 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23956 * lto-wrapper.c (merge_and_complain): Do not merge
23957 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23958 fsigned_zeros, ftrapping_math, fwrapv.
23959 (append_compiler_options): Do not track these options.
23960 (append_linker_options): Likewie
23962 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23964 * cgraphunit.c (cgraph_node::finalize_function): When
23965 !flag_toplevel_reorde set no_reorder flag.
23966 (varpool_node::finalize_decl): Likewise.
23967 (symbol_table::compile): Drop no toplevel reorder path.
23969 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23971 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23972 edges; zero probability is not better than uninitialized.
23974 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23976 * asan.h (asan_sanitize_allocas_p): Declare.
23977 * asan.c (asan_sanitize_allocas_p): New function.
23978 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23979 (handle_builtin_alloca): Likewise.
23980 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23981 if !asan_sanitize_allocas_p.
23982 * params.def (asan-instrument-allocas): Add new option.
23983 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23984 * opts.c (common_handle_option): Disable allocas sanitization for
23987 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23989 * asan.c: Include gimple-fold.h.
23990 (get_last_alloca_addr): New function.
23991 (handle_builtin_stackrestore): Likewise.
23992 (handle_builtin_alloca): Likewise.
23993 (asan_emit_allocas_unpoison): Likewise.
23994 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23995 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23996 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23997 (instrument_builtin_call): Pass gimple iterator to
23998 get_mem_refs_of_builtin_call.
23999 (last_alloca_addr): New global.
24000 * asan.h (asan_emit_allocas_unpoison): Declare.
24001 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24002 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24003 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24004 if function calls alloca.
24005 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24006 * gimple-fold.h (replace_call_with_value): Declare.
24007 * internal-fn.c: Include asan.h.
24008 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24009 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24011 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24013 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24014 (C_SELFTEST_FLAGS): New.
24015 (CPP_SELFTEST_FLAGS): New.
24016 (SELFTEST_DEPS): New, from deps of s-selftest.
24017 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24018 (CPP_SELFTEST_DEPS): New.
24019 (selftest): Add dependency on s-selftest-c++.
24020 (s-selftest): Rename to...
24021 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24022 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24023 than SELFTEST_FLAGS.
24024 (selftest-gdb): Rename to...
24025 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24027 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24028 (selftest-valgrind): Rename to...
24029 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24031 (selftest-valgrind): Reintroduce as an alias for
24032 selftest-c-valgrind.
24033 (s-selftest-c++): New.
24034 (selftest-c++-gdb): New.
24035 (selftest-c++-valgrind): New.
24037 2017-07-06 Olivier Hainque <hainque@adacore.com>
24039 * gcc.c (process_command): When deciding if undefined variables
24040 should be ignored when processing specs, accept "gcc -v" as well.
24042 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24044 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24045 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24047 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24049 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24050 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24051 * config/arm/arm-tables.opt: Regenerate.
24052 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24054 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24056 2017-07-06 Carl Love <cel@us.ibm.com>
24058 * ChangeLog: Clean up from mid air collision
24060 2017-07-06 Carl Love <cel@us.ibm.com>
24062 * config/rs6000/rs6000-c.c: Add support for built-in functions
24063 vector signed int vec_subc (vector signed int, vector signed int);
24064 vector signed __int128 vec_subc (vector signed __int128,
24065 vector signed __int128);
24066 vector unsigned __int128 vec_subc (vector unsigned __int128,
24067 vector unsigned __int128);
24068 vector signed int vec_sube (vector signed int, vector signed int,
24069 vector signed int);
24070 vector unsigned int vec_sube (vector unsigned int,
24071 vector unsigned int,
24072 vector unsigned int);
24073 vector signed __int128 vec_sube (vector signed __int128,
24074 vector signed __int128,
24075 vector signed__int128);
24076 vector unsigned __int128 vec_sube (vector unsigned __int128,
24077 vector unsigned __int128,
24078 vector unsigned __int128);
24079 vector signed int vec_subec (vector signed int, vector signed int,
24080 vector signed int);
24081 vector unsigned int vec_subec (vector unsigned int,
24082 vector unsigned int,
24083 vector unsigned int);
24084 vector signed __int128 vec_subec (vector signed __int128,
24085 vector signed __int128,
24086 vector signed__int128);
24087 vector unsigned __int128 vec_subec (vector unsigned __int128,
24088 vector unsigned __int128,
24089 vector unsigned __int128);
24090 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24091 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24092 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24093 BU_ALTIVEC_OVERLOAD_X definitions.
24094 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24095 * doc/extend.texi: Update the built-in documentation file for the new
24096 built-in functions.
24098 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24101 * diagnostic-show-locus.c (layout::layout): Use start and finish
24102 spelling location for the start and finish of each range.
24103 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24104 Add unused aspect param.
24105 * input.c (expand_location_1): Add "aspect" param, and use it
24106 to access the correct part of the location.
24107 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24109 (expand_location_to_spelling_point): Likewise.
24110 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24111 param, and pass it to expand_location_1.
24113 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24115 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24116 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24117 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24118 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24119 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24120 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24121 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24122 _mm_maskz_getmant_ss): New intrinsics.
24123 (__builtin_ia32_getexpss128_mask): Changed to ...
24124 __builtin_ia32_getexpss128_round ... this.
24125 (__builtin_ia32_getexpsd128_mask): Changed to ...
24126 __builtin_ia32_getexpsd128_round ... this.
24127 * config/i386/i386-builtin-types.def
24128 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24129 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24130 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24131 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24132 __builtin_ia32_getmantss_mask_round): New builtins.
24133 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24134 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24135 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24136 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24137 * config/i386/sse.md
24138 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24139 avx512f_sgetexp<mode><mask_scalar_name>
24140 <round_saeonly_scalar_name> ... this.
24141 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24142 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24143 vgetexp<ssescalarmodesuffix>
24144 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24145 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24146 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24147 avx512f_vgetmant<mode><mask_scalar_name>
24148 <round_saeonly_scalar_name> ... this.
24149 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24150 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24151 vgetmant<ssescalarmodesuffix>
24152 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24153 %0<mask_scalar_operand4>, %1, %2
24154 <round_saeonly_scalar_mask_op4>, %3} ... this.
24155 * config/i386/subst.md (mask_scalar_operand4,
24156 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24157 round_saeonly_scalar_nimm_predicate): New subst attributes.
24159 2017-07-06 Julia Koval <julia.koval@intel.com>
24161 * config/i386/i386.c (ix86_erase_embedded_rounding):
24162 Remove code for old rounding pattern.
24164 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24166 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24168 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24170 * doc/sourcebuild.texi (Test Directives, Variants of
24171 dg-require-support): Add documentation for dg-require-stack-check.
24173 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24175 * config/i386/subst.md (mask_scalar, round_scalar,
24176 round_saeonly_scalar): New meta-templates.
24177 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24178 round_scalar_mask_operand3, round_scalar_mask_op3,
24179 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24180 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24181 round_saeonly_scalar_constraint,
24182 round_saeonly_scalar_prefix): New subst attribute.
24183 * config/i386/sse.md
24184 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24185 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24186 <round_scalar_name> ... this.
24187 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24188 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24189 <round_scalar_name> ... this.
24190 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24191 <sse>_vm<code><mode>3<mask_scalar_name>
24192 <round_saeonly_scalar_name> ... this.
24193 (v<plusminus_mnemonic><ssescalarmodesuffix>
24194 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24195 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24196 v<plusminus_mnemonic><ssescalarmodesuffix>
24197 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24198 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24199 (v<multdiv_mnemonic><ssescalarmodesuffix>
24200 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24201 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24202 v<multdiv_mnemonic><ssescalarmodesuffix>
24203 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24204 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24205 (v<maxmin_float><ssescalarmodesuffix>
24206 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24207 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24208 v<maxmin_float><ssescalarmodesuffix>
24209 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24210 %0<mask_scalar_operand3>, %1, %<iptr>2
24211 <round_saeonly_scalar_mask_op3>} ... this.
24213 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24215 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24216 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24218 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24219 Alan Hayward <alan.hayward@arm.com>
24220 David Sherwood <david.sherwood@arm.com>
24222 * combine.c (simplify_if_then_else): Remove "enum" before
24224 * compare-elim.c (can_eliminate_compare): Likewise.
24225 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24227 (aarch64_lookup_simd_builtin_type): Likewise.
24228 (aarch64_simd_builtin_type): Likewise.
24229 (aarch64_init_simd_builtin_types): Likewise.
24230 (aarch64_simd_expand_args): Likewise.
24231 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24233 (aarch64_reverse_mask): Likewise.
24234 (aarch64_simd_emit_reg_reg_move): Likewise.
24235 (aarch64_gen_adjusted_ldpstp): Likewise.
24236 (aarch64_ccmp_mode_to_code): Likewise.
24237 (aarch64_operands_ok_for_ldpstp): Likewise.
24238 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24239 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24241 (aarch64_min_divisions_for_recip_mul): Likewise.
24242 (aarch64_reassociation_width): Likewise.
24243 (aarch64_get_condition_code_1): Likewise.
24244 (aarch64_simd_emit_reg_reg_move): Likewise.
24245 (aarch64_simd_attr_length_rglist): Likewise.
24246 (aarch64_reverse_mask): Likewise.
24247 (aarch64_operands_ok_for_ldpstp): Likewise.
24248 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24249 (aarch64_gen_adjusted_ldpstp): Likewise.
24250 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24252 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24253 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24254 (arm_lookup_simd_builtin_type): Likewise.
24255 (arm_simd_builtin_type): Likewise.
24256 (arm_init_simd_builtin_types): Likewise.
24257 (arm_expand_builtin_args): Likewise.
24258 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24259 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24260 (ft32_setup_incoming_varargs): Likewise.
24261 (ft32_function_arg): Likewise.
24262 (ft32_function_arg_advance): Likewise.
24263 (ft32_pass_by_reference): Likewise.
24264 (ft32_arg_partial_bytes): Likewise.
24265 (ft32_valid_pointer_mode): Likewise.
24266 (ft32_addr_space_pointer_mode): Likewise.
24267 (ft32_addr_space_legitimate_address_p): Likewise.
24268 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24270 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24271 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24272 (iamcu_alignment): Likewise.
24273 (canonicalize_vector_int_perm): Likewise.
24274 (ix86_noce_conversion_profitable_p): Likewise.
24275 (ix86_mpx_bound_mode): Likewise.
24276 (ix86_operands_ok_for_move_multiple): Likewise.
24277 * config/microblaze/microblaze-protos.h
24278 (microblaze_expand_conditional_branch_reg): Likewise.
24279 * config/microblaze/microblaze.c
24280 (microblaze_expand_conditional_branch_reg): Likewise.
24281 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24283 (rs6000_reassociation_width): Likewise.
24284 (rs6000_invalid_binary_op): Likewise.
24285 (fusion_p9_p): Likewise.
24286 (emit_fusion_p9_load): Likewise.
24287 (emit_fusion_p9_store): Likewise.
24288 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24290 (riscv_hard_regno_mode_ok_p): Likewise.
24291 (riscv_address_insns): Likewise.
24292 (riscv_split_symbol): Likewise.
24293 (riscv_legitimize_move): Likewise.
24294 (riscv_function_value): Likewise.
24295 (riscv_hard_regno_nregs): Likewise.
24296 (riscv_expand_builtin): Likewise.
24297 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24298 (riscv_build_integer): Likewise.
24299 (riscv_split_integer): Likewise.
24300 (riscv_legitimate_constant_p): Likewise.
24301 (riscv_cannot_force_const_mem): Likewise.
24302 (riscv_regno_mode_ok_for_base_p): Likewise.
24303 (riscv_valid_base_register_p): Likewise.
24304 (riscv_valid_offset_p): Likewise.
24305 (riscv_valid_lo_sum_p): Likewise.
24306 (riscv_classify_address): Likewise.
24307 (riscv_legitimate_address_p): Likewise.
24308 (riscv_address_insns): Likewise.
24309 (riscv_load_store_insns): Likewise.
24310 (riscv_force_binary): Likewise.
24311 (riscv_split_symbol): Likewise.
24312 (riscv_force_address): Likewise.
24313 (riscv_legitimize_address): Likewise.
24314 (riscv_move_integer): Likewise.
24315 (riscv_legitimize_const_move): Likewise.
24316 (riscv_legitimize_move): Likewise.
24317 (riscv_address_cost): Likewise.
24318 (riscv_subword): Likewise.
24319 (riscv_output_move): Likewise.
24320 (riscv_canonicalize_int_order_test): Likewise.
24321 (riscv_emit_int_order_test): Likewise.
24322 (riscv_function_arg_boundary): Likewise.
24323 (riscv_pass_mode_in_fpr_p): Likewise.
24324 (riscv_pass_fpr_single): Likewise.
24325 (riscv_pass_fpr_pair): Likewise.
24326 (riscv_get_arg_info): Likewise.
24327 (riscv_function_arg): Likewise.
24328 (riscv_function_arg_advance): Likewise.
24329 (riscv_arg_partial_bytes): Likewise.
24330 (riscv_function_value): Likewise.
24331 (riscv_pass_by_reference): Likewise.
24332 (riscv_setup_incoming_varargs): Likewise.
24333 (riscv_print_operand): Likewise.
24334 (riscv_elf_select_rtx_section): Likewise.
24335 (riscv_save_restore_reg): Likewise.
24336 (riscv_for_each_saved_reg): Likewise.
24337 (riscv_register_move_cost): Likewise.
24338 (riscv_hard_regno_mode_ok_p): Likewise.
24339 (riscv_hard_regno_nregs): Likewise.
24340 (riscv_class_max_nregs): Likewise.
24341 (riscv_memory_move_cost): Likewise.
24342 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24343 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24344 (rl78_addr_space_address_mode): Likewise.
24345 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24347 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24348 (rs6000_reassociation_width): Likewise.
24349 (rs6000_invalid_binary_op): Likewise.
24350 (fusion_p9_p): Likewise.
24351 (emit_fusion_p9_load): Likewise.
24352 (emit_fusion_p9_store): Likewise.
24353 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24354 (ok_for_simple_move_operands): Likewise.
24355 (ok_for_simple_move_strict_operands): Likewise.
24356 (ok_for_simple_arith_logic_operands): Likewise.
24357 (visium_legitimize_reload_address): Likewise.
24358 (visium_select_cc_mode): Likewise.
24359 (output_cbranch): Likewise.
24360 (visium_split_double_move): Likewise.
24361 (visium_expand_copysign): Likewise.
24362 (visium_expand_int_cstore): Likewise.
24363 (visium_expand_fp_cstore): Likewise.
24364 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24365 (visium_function_arg): Likewise.
24366 (visium_function_arg_advance): Likewise.
24367 (visium_libcall_value): Likewise.
24368 (visium_setup_incoming_varargs): Likewise.
24369 (visium_legitimate_constant_p): Likewise.
24370 (visium_legitimate_address_p): Likewise.
24371 (visium_legitimize_address): Likewise.
24372 (visium_secondary_reload): Likewise.
24373 (visium_register_move_cost): Likewise.
24374 (visium_memory_move_cost): Likewise.
24375 (prepare_move_operands): Likewise.
24376 (ok_for_simple_move_operands): Likewise.
24377 (ok_for_simple_move_strict_operands): Likewise.
24378 (ok_for_simple_arith_logic_operands): Likewise.
24379 (visium_function_value_1): Likewise.
24380 (rtx_ok_for_offset_p): Likewise.
24381 (visium_legitimize_reload_address): Likewise.
24382 (visium_split_double_move): Likewise.
24383 (visium_expand_copysign): Likewise.
24384 (visium_expand_int_cstore): Likewise.
24385 (visium_expand_fp_cstore): Likewise.
24386 (visium_split_cstore): Likewise.
24387 (visium_select_cc_mode): Likewise.
24388 (visium_split_cbranch): Likewise.
24389 (output_cbranch): Likewise.
24390 (visium_print_operand_address): Likewise.
24391 * expmed.c (flip_storage_order): Likewise.
24392 * expmed.h (emit_cstore): Likewise.
24393 (flip_storage_order): Likewise.
24394 * genrecog.c (validate_pattern): Likewise.
24395 * hsa-gen.c (gen_hsa_addr): Likewise.
24396 * internal-fn.c (expand_arith_overflow): Likewise.
24397 * ira-color.c (allocno_copy_cost_saving): Likewise.
24398 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24399 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24400 (process_invariant_for_inheritance): Likewise.
24401 * lra-eliminations.c (move_plus_up): Likewise.
24402 * omp-low.c (lower_oacc_reductions): Likewise.
24403 * simplify-rtx.c (simplify_subreg): Likewise.
24404 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24405 (TARGET_CHKP_BOUND_MODE): Likewise..
24406 * targhooks.c (default_chkp_bound_mode): Likewise.
24407 (default_setup_incoming_vararg_bounds): Likewise.
24408 * targhooks.h (default_chkp_bound_mode): Likewise.
24409 (default_setup_incoming_vararg_bounds): Likewise.
24410 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24411 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24412 (have_whole_vector_shift): Likewise.
24413 * tree-vect-stmts.c (vectorizable_load): Likewise.
24414 * doc/tm.texi: Regenerate.
24416 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24418 Graceful degrade if Binutils PR21472 is not available.
24421 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24422 .rodata in flash test fails.
24423 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24424 * confgure: Regenerate.
24425 * config.in: Regenerate.
24426 * config/avr/avr.c (avr_asm_named_section)
24427 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24428 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24429 (avr_asm_init_sections): Same.
24431 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24433 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24434 (fma<VH:mode>4_intrinsic): Likewise.
24435 (*fmsub<VCVTF:mode>4): Likewise.
24436 (*fmsub<VH:mode>4_intrinsic): Likewise.
24438 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24441 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24442 Don't depend on "optimize > 0".
24443 (out_movhi_r_mr, out_movqi_mr_r): Same.
24444 (out_movhi_mr_r, out_movqi_r_mr): Same.
24445 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24446 io_address_operand on "optimize > 0".
24448 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24450 * tree-loop-distribution.c: Add general explanantion on the pass.
24451 (generate_loops_for_partition): Mark distributed loop.
24452 (pg_add_dependence_edges): New parameter. Handle alias data
24453 dependence specially and record it in the parameter if asked.
24454 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24455 (init_partition_graph_vertices, add_partition_graph_edge): New.
24456 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24457 (free_partition_graph_vdata, build_partition_graph): New.
24458 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24459 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24460 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24461 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24462 (version_for_distribution_p, finalize_partitions): New.
24463 (distribute_loop): Handle alias data dependence specially. Factor
24464 out loop fusion code as functions and call these functions.
24466 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24468 * tree-loop-distribution.c (classify_partition): New parameter and
24469 better handle reduction statement.
24470 (rdg_build_partitions): Revise comment.
24471 (distribute_loop): Compute statements in all partitions and pass it
24472 to classify_partition.
24474 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24476 * tree-loop-distribution.c (enum partition_type): New.
24477 (struct partition): New field type.
24478 (partition_merge_into): Add parameter. Update partition type.
24479 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24480 (build_rdg_partition_for_vertex): Compute partition type.
24481 (rdg_build_partitions): Dump partition type.
24482 (distribute_loop): Update calls to partition_merge_into.
24484 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24486 * tree-loop-distribution.c (struct ddr_hasher): New.
24487 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24489 (classify_partition): Call get_data_dependence.
24490 (pg_add_dependence_edges): Ditto.
24491 (distribute_loop): Release data dependence hash table.
24493 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24495 * tree-loop-distribution.c (ref_base_address): Delete.
24496 (similar_memory_accesses): Rename ...
24497 (share_memory_accesses): ... to this. Check if partitions access
24498 the same memory reference.
24499 (distribute_loop): Call share_memory_accesses.
24501 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24503 * tree-loop-distribution.c (struct partition): New field recording
24504 its data reference.
24505 (partition_alloc, partition_free): Init and release data refs.
24506 (partition_merge_into): Merge data refs.
24507 (build_rdg_partition_for_vertex): Collect data refs for partition.
24508 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24510 (distribute_loop): Remve data refs from vertice data of partition
24513 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24515 * tree-loop-distribution.c (params.h): Include header file.
24516 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24517 (datarefs_vec): New global var.
24518 (create_rdg_vertices): Use datarefs_vec directly.
24519 (free_rdg): Don't free data references.
24520 (build_rdg): Update use. Don't free data references.
24521 (distribute_loop): Compute global variable for data references.
24522 Bail out if there are too many data references.
24524 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24526 * tree-loop-distribution.c (loop_nest): New global var.
24527 (build_rdg): Use loop directly, rather than loop nest.
24528 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24530 (distribute_loop): Compute global variable loop nest. Update use.
24532 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24534 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24535 (partition_merge_into): New parameter. Dump reason for fusion.
24536 (distribute_loop): Update use of partition_merge_into.
24538 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24540 * tree-loop-distribution.c (bb_top_order_index): New.
24541 (bb_top_order_index_size, bb_top_order_cmp): New.
24542 (stmts_from_loop): Use topological order.
24543 (pass_loop_distribution::execute): Compute and release topological
24544 order for basic blocks.
24546 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24548 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24551 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24553 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24554 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24555 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24556 * internal-fn.def (LOOP_DIST_ALIAS): New.
24557 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24558 (fold_loop_internal_call): ... this.
24559 (vect_loop_dist_alias_call): New function.
24560 (set_uid_loop_bbs): Call fold_loop_internal_call.
24561 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24564 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24567 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24568 Require dead FLAGS_REG at the beginning of a peephole.
24570 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24573 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24574 arguments in the call to __builtin_ia32_sbb_u32.
24575 (_subborrow_u64): Swap _X and _Y arguments in the call to
24576 __builtin_ia32_sbb_u64.
24578 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24581 * tree-vrp.c (compare_assert_loc): Turn into a function template
24582 with stable template parameter. Only test if a->e is NULL,
24583 !a->e == !b->e has been verified already. Use e == NULL or
24584 e != NULL instead of e or ! e tests. If stable is true, don't use
24585 iterative_hash_expr, on the other side allow a or b or both NULL
24586 and sort the NULLs last.
24587 (process_assert_insertions): Sort using compare_assert_loc<false>
24588 instead of compare_assert_loc, later sort using
24589 compare_assert_loc<true> before calling process_assert_insertions_for
24590 in a loop. Use break instead of continue once seen NULL pointer.
24592 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24594 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24595 Cortex-R7 and Cortex-R8 processors.
24597 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24599 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24600 uninitialized while src is not.
24602 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24604 * common/config/arm/arm-common.c: Adjust include path for
24606 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24607 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24608 (arm-cpu-data.h): Likewise.
24609 (arm-cpu-cdata.h): Likewise.
24610 * config/arm/arm-cpu.h: Delete.
24611 * config/arm/arm-cpu-cdata.h: Delete.
24612 * config/arm/arm-cpu-data.h: Delete.
24614 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24616 * config/arm/arm-cpus.in (cortex-a55): New.
24617 (cortex-a75): Likewise.
24618 (cortex-a75.cortex-a55): Likewise.
24619 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24621 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24622 * config/arm/arm-cpu-cdata.h: Regenerate.
24623 * config/arm/arm-cpu-data.h: Regenerate.
24624 * config/arm/arm-cpu.h: Regenerate.
24625 * config/arm/arm-tables.opt: Regenerate.
24626 * config/arm/arm-tune.md: Regenerate.
24628 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24630 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24632 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24634 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24637 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24639 PR tree-optimization/81292
24640 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24641 full_string_p, also call adjust_related_strinfos if the adjustment
24642 is simple, otherwise invalidate related strinfos.
24644 2017-07-04 Martin Liska <mliska@suse.cz>
24647 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24648 newly created variable as DECL_IGNORED_P.
24650 2017-07-04 Martin Liska <mliska@suse.cz>
24653 * ipa-inline.c (inline_small_functions):
24654 Use xstrdup_for_dump.
24656 2017-07-04 Tom de Vries <tom@codesourcery.com>
24658 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24660 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24663 * config/darwin.c (darwin_function_switched_text_sections):
24666 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24668 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24670 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24672 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24674 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24676 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24677 min_profitable_iters, and th as inclusive lower bounds.
24678 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24679 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24680 for min_profitable_iters and min_profitable_estimate.
24681 (vect_transform_loop): Treat th as an inclusive lower bound.
24682 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24684 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24687 * config/darwin.c (darwin_function_switched_text_sections):
24688 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24689 in two pieces, and suppress the use of buf.
24691 2017-07-03 Nathan Sidwell <nathan@acm.org>
24693 * hash-table.h (hash_table_mod1): Fix indentation.
24695 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24697 PR middle-end/81290
24698 * predict.c (force_edge_cold): Be more careful about propagation
24700 * profile-count.h (profile_probability::guessed,
24701 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24703 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24705 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24707 * doc/invoke.texi (rcpc architecture extension): Document it.
24709 2017-07-03 Richard Biener <rguenther@suse.de>
24711 PR tree-optimization/60510
24712 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24713 the scalar reduction PHI and use it.
24714 (vectorizable_reduction): Properly guard the single_defuse_cycle
24715 path for non-SLP reduction chains where we cannot use it.
24716 Rework reduc_def/index and vector type deduction. Rework
24717 vector operand gathering during reduction op code-gen.
24718 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24719 chains dissolve the chain and leave it to non-SLP reduction
24722 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24724 * tree-data-ref.h (dr_alignment): Declare.
24725 * tree-data-ref.c (dr_alignment): New function.
24726 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24727 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24729 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24731 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24733 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24734 and base_misalignment fields.
24735 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24736 * tree-data-ref.c: Include builtins.h.
24737 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24738 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24739 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24740 * tree-vect-data-refs.c: Include tree-cfg.h.
24741 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24742 fields instead of calculating an alignment here.
24743 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24744 innermost_loop_behavior fields.
24746 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24748 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24750 (DR_STEP_ALIGNMENT): New macro.
24751 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24752 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24753 (create_data_ref): Print it.
24754 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24755 to tell whether the step preserves vector (mis)alignment.
24756 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24757 Move the check for an integer step and generalise to all INTEGER_CST.
24758 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24759 Print the outer step alignment.
24761 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24763 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24764 with offset_alignment.
24765 (DR_ALIGNED_TO): Delete.
24766 (DR_OFFSET_ALIGNMENT): New macro.
24767 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24768 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24769 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24770 (create_data_ref): Likewise.
24771 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24772 (vect_analyze_data_refs): Likewise.
24773 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24774 creating dummy innermost behavior.
24776 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24778 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24779 with a "innermost_loop_behavior *" and refeence tree.
24780 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24781 (create_data_ref): Update call accordingly.
24782 * tree-predcom.c (find_looparound_phi): Likewise.
24784 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24786 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24787 fields with dr_wrt_vec_loop.
24788 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24789 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24790 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24791 (vect_dr_behavior): New function.
24792 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24793 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24794 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24795 track whether the step preserves the misalignment.
24796 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24797 Use vect_dr_behavior.
24798 (vect_setup_realignment): Update call accordingly.
24799 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24800 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24801 call to vect_create_addr_base_for_vector_ref.
24802 (vect_create_cond_for_align_checks): Likewise.
24803 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24804 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24805 (vect_recog_mask_conversion_pattern): Likewise.
24806 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24807 (new_stmt_vec_info): Remove redundant zeroing.
24809 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24811 * common/config/arm/arm-common.c (arm_be8_option): New function.
24812 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24813 (ISA_ARMv6): Add isa_bit_be8.
24814 * config/arm/arm.h (arm_be8_option): Add prototype.
24815 (BE8_SPEC_FUNCTION): New define.
24816 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24817 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24818 (mlittle-endian): Similarly.
24819 (mbe8, mbe32): New options.
24820 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24821 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24823 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24825 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24827 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24829 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24830 (cleanup_tree_cfg_bb): Use it.
24831 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24833 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24835 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24838 * loop-doloop.c (add_test): Update profile.
24840 2017-07-03 Martin Liska <mliska@suse.cz>
24843 * sanopt.c (rewrite_usage_of_param): New function.
24844 (sanitize_rewrite_addressable_params): Likewise.
24845 (pass_sanopt::execute): Call rewrite_usage_of_param.
24847 2017-07-03 Richard Biener <rguenther@suse.de>
24849 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24850 back to using VIEW_CONVERT_EXPR.
24852 2017-07-03 Martin Liska <mliska@suse.cz>
24855 * doc/extend.texi: Document when a resolver function is
24856 generated for target_clones.
24858 2017-07-03 Martin Liska <mliska@suse.cz>
24860 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24861 and shadow memory of auto variables which are subject of
24862 use-after-scope sanitization.
24863 (asan_expand_mark_ifn): Add do set only when is_poison.
24865 2016-07-03 Richard Biener <rguenther@suse.de>
24867 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24869 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24870 (vectorizable_reduction): Perform reduction PHI creation when
24871 visiting a reduction PHI and adjust and simplify code generation
24872 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24873 (vect_transform_loop): Visit reduction PHIs.
24874 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24875 defs into the SLP tree.
24876 (vect_build_slp_tree): Reduction defs terminate the recursion.
24877 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24879 (vect_get_vec_defs_for_stmt_copy): Export.
24880 (vect_get_vec_defs): Likewise.
24881 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24883 (vect_get_vec_defs_for_stmt_copy): Declare.
24884 (vect_get_vec_defs): Likewise.
24886 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24888 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24889 parameter with a "loop" parameter and use it instead of the
24890 loop containing DR_STMT. Don't check simple_iv when doing
24891 BB analysis. Describe the two analysis modes in the comment.
24893 2017-07-03 Tom de Vries <tom@codesourcery.com>
24895 PR tree-optimization/69468
24896 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24897 (find_same_succ_bb): Handle ignore_edge_flags.
24899 2017-07-03 Tom de Vries <tom@codesourcery.com>
24901 PR tree-optimization/81192
24902 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24904 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24906 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24908 2017-07-03 Tom de Vries <tom@codesourcery.com>
24910 PR tree-optimization/81192
24911 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24912 BB_SAME_SUCC (bb) == NULL.
24914 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24916 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24919 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24921 * dumpfile.c: Include profile-count.h
24922 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24924 (insert_cond_bb): Update profile.
24925 * tree-cfg.h (insert_cond_bb): Update prototype.
24926 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24927 * tree-dump.c: Do not include tree-cfg.
24929 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24931 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24933 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24935 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24938 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24940 * tree-complex.c (expand_complex_div_wide): update profile.
24942 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24943 Alan Hayward <alan.hayward@arm.com>
24944 David Sherwood <david.sherwood@arm.com>
24946 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24947 (CORETYPES_H): New define.
24948 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24949 (insn-modes-inline.h, s-modes-inline-h): New rules.
24950 (generated_files): Add insn-modes-inline.h.
24951 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24952 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24953 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24954 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24955 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24956 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24957 (build/gencodes.o, build/genconditions.o): Likewise.
24958 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24959 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24960 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24961 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24962 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24963 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24964 * coretypes.h: Include everything up to real.h for generators.
24965 Include insn-modes.h first. Include wide-int-print.h after
24966 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24967 * machmode.h: Don't include insn-modes.h here.
24968 * function-tests.c: Remove includes of signop.h, machmode.h,
24969 double-int.h and wide-int.h.
24971 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24973 * optc-save-gen.awk: Likewise.
24974 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24975 * godump.c: Remove include of wide-int-print.h.
24976 * pretty-print.h: Likewise.
24977 * wide-int-print.cc: Likewise.
24978 * wide-int.cc: Likewise.
24979 * hash-map-tests.c: Remove include of signop.h.
24980 * hash-set-tests.c: Likewise.
24981 * rtl-tests.c: Likewise.
24982 * mkconfig.sh: Remove include of machmode.h.
24983 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24985 (emit_insn_modes_inline_h): ...this new function. Emit the code
24986 into an insn-modes-inline.h header file, adding appropriate
24987 include guards and end comments.
24988 (emit_insn_modes_c_header): Remove include of machmode.h.
24989 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24991 (main): Handle -i flag and call emit_insn_modes_inline_h when
24994 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24996 * tree-ssa-strlen.c (strinfo): Rename the length field to
24997 nonzero_chars. Add a full_string_p field.
24998 (compare_nonzero_chars, zero_length_string_p): New functions.
24999 (get_addr_stridx): Add an offset_out parameter.
25000 Use compare_nonzero_chars.
25001 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25002 (new_strinfo): Update after above changes to strinfo.
25003 (set_endptr_and_length): Set full_string_p.
25004 (get_string_length): Update after above changes to strinfo.
25005 (unshare_strinfo): Update call to new_strinfo.
25006 (maybe_invalidate): Likewise.
25007 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25008 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25009 as a uhwi instead of an shwi. Update after above changes to
25010 strinfo and new_strinfo.
25011 (zero_length_string): Assert that chainsi contains full strings.
25012 Use zero_length_string_p. Update call to new_strinfo.
25013 (adjust_related_strinfos): Update after above changes to strinfo.
25014 Copy full_string_p from origsi.
25015 (adjust_last_stmt): Use zero_length_string_p.
25016 (handle_builtin_strlen): Update after above changes to strinfo and
25017 new_strinfo. Install the lhs as the string length if the previous
25018 entry didn't describe a full string.
25019 (handle_builtin_strchr): Update after above changes to strinfo
25021 (handle_builtin_strcpy): Likewise.
25022 (handle_builtin_strcat): Likewise.
25023 (handle_builtin_malloc): Likewise.
25024 (handle_pointer_plus): Likewise.
25025 (handle_builtin_memcpy): Likewise. Track nonzero characters
25026 that aren't necessarily followed by a nul terminator.
25027 (handle_char_store): Likewise.
25029 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25031 PR tree-optimization/80769
25032 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25033 for malloc and calloc. Document the new invariant that all related
25034 strinfos have delayed lengths or none do.
25035 (verify_related_strinfos): Move earlier in file.
25036 (set_endptr_and_length): New function, split out from...
25037 (get_string_length): ...here. Also set the lengths of related
25039 (zero_length_string): Assert that chainsi has known (rather than
25041 (adjust_related_strinfos): Likewise.
25043 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25045 PR tree-optimization/81136
25046 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25047 assert that two references with the same misalignment have the same
25048 compile-time misalignment if those compile-time misalignments
25051 2017-07-01 Andi Kleen <ak@linux.intel.com>
25053 * print-tree.c (print_node): Print all attributes.
25055 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25057 * cfg.c (scale_bbs_frequencies): New function.
25058 * cfg.h (scale_bbs_frequencies): Declare it.
25059 * cfgloopanal.c (single_likely_exit): Cleanup.
25060 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25062 (scale_loop_profile): Likewise.
25063 (loop_version): Likewise.
25064 (create_empty_loop_on_edge): Update.
25065 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25066 scale_loop_frequencies, scale_loop_profile, loopify,
25067 loop_version): Update prototypes.
25068 * modulo-sched.c (sms_schedule): Update.
25069 * predict.c (unlikely_executed_edge_p): Also check probability.
25070 (probably_never_executed_edge_p): Fix typo.
25071 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25072 * tree-parloops.c (gen_parallel_loop): Update.
25073 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25074 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25075 * tree-ssa-loop-split.c (split_loop): Update.
25076 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25077 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25078 (vect_loop_versioning): Update.
25079 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25081 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25083 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25085 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25087 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25088 to keep profile consistent.
25090 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25092 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25093 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25094 * profile-count.h (max_safe_multiplier): Make unsigned.
25095 (profile_count::guessed_zero): New.
25097 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25099 * bb-reorder.c (fix_up_crossing_landing_pad,
25100 fix_crossing_conditional_branches): Use make_single_succ_edge
25101 to keep profile consistent.
25103 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25105 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25108 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25111 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25112 the right scopes, make sure cond_jump isn't preserved between multiple
25113 iterations. Search for fallthru edge whenever there are 3+ edges and
25114 use find_fallthru_edge for it.
25116 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25118 Patch by Alexander Monakov <amonakov@ispras.ru>
25119 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25120 probabilities consistently.
25122 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25124 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25125 emit_cmp_and_jump_insns.
25127 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25130 * tree-inline.c (expand_call_inline): Combine profile statuses.
25132 2017-06-30 Andrew Pinski <apinski@cavium.com>
25134 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25135 fold_stmt returned true.
25137 2017-06-30 Nathan Sidwell <nathan@acm.org>
25139 * ggc.h (empty_string): Delete.
25140 * cfgexpand.c (expand_asm_stmt): Use plain "".
25141 * optabs.c (expand_asm_memory_barrier): Likewise.
25142 * stringpool.c (empty_string): Delete.
25143 (digit_vector, digit_string): Delete.
25144 (ggc_alloc_string): Use plain "", don't optimize single digit
25145 strings. Use ggc_alloc_atomic.
25147 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25149 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25150 comparison set and one other set, use the cost of the non-comparison
25153 2017-06-30 Nathan Sidwell <nathan@acm.org>
25155 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25158 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25160 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25161 loops. Remove now unneeded calls to gimple_switch_set_label() that
25162 just set removed labels to NULL_TREE.
25164 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25166 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25167 (set_range_info): ...here. Only call set_range_info_raw if domain
25169 (set_nonzero_bits): Call set_range_info_raw.
25170 * tree-ssanames.h (set_range_info_raw): New.
25172 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25175 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25176 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25177 of nonimmediate_operand and <store_mask_constraint> instead of m
25178 for the input operand. For V8FI iterator, always split if input
25179 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25180 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25181 <store_mask_predicate> instead of register_operand and
25182 <store_mask_constraint> instead of v for the input operand. Make
25183 sure both operands aren't MEMs for if not <mask_applied>.
25185 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25187 * lto-wrapper.c (copy_file) Close both file descriptors before
25190 2017-06-30 Martin Liska <mliska@suse.cz>
25193 * multiple_target.c (create_dispatcher_calls): Make ifunc
25194 also for function that don't have calls or are not referenced.
25196 2017-06-30 Richard Biener <rguenther@suse.de>
25198 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25199 analyze the first scalar stmt. Move vector type computation
25200 for the BB case here from ...
25201 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25202 live operation processing in the SLP case properly.
25204 2017-06-30 Richard Biener <rguenther@suse.de>
25206 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25208 2017-06-30 Martin Liska <mliska@suse.cz>
25211 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25212 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25214 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25216 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25219 2017-06-30 Richard Biener <rguenther@suse.de>
25221 PR tree-optimization/81249
25222 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25223 condition reduction result to original scalar type.
25225 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25227 * profile-count.h (enum profile_quality): Fix typos and whitespace
25230 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25232 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25233 type for branch probabilities.
25235 2017-06-29 Julian Brown <julian@codesourcery.com>
25236 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25238 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25239 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25240 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25241 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25243 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25245 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25246 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25247 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25248 CC usage from generic code to here.
25249 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25250 CC usage into the target macros.
25252 2017-06-29 Maya Rashish <coypu@sdf.org>
25254 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25257 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25259 * arm/arm-builtins.c: Include profile-count.h
25260 * except.c (sjlj_emit_function_enter): Use
25261 profile_probability::unlikely.
25263 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25265 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25266 and tocrel_offset be pointer args rather than implicitly using
25268 (legitimate_constant_pool_address_p, rs6000_emit_move,
25269 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25270 tocrel_offset and use in toc_relative_expr_p call.
25271 (print_operand, print_operand_address): Use static tocrel_base_oac
25272 and tocrel_offset_oac.
25273 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25276 2017-06-29 Maya Rashish <coypu@sdf.org>
25278 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25280 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25282 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25283 objects, take into account only the alignment of 'op0' and 'mode1' if
25286 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25288 * ccmp.c (ccmp_tree_comparison_p): New function.
25289 (ccmp_candidate_p): Update to use above function.
25290 (get_compare_parts): New function.
25291 (expand_ccmp_next): Update to use new functions.
25292 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25294 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25295 take mode as argument.
25296 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25297 * expr.c (expand_expr_real_1): Pass mode as argument.
25299 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25301 * combine.c (combine_instructions): Print insns to dump_file, together
25304 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25306 * asan.c (asan_emit_stack_protection): Update.
25307 (create_cond_insert_point): Update.
25308 * auto-profile.c (afdo_propagate_circuit): Update.
25309 * basic-block.h (struct edge_def): Turn probability to
25310 profile_probability.
25311 (EDGE_FREQUENCY): Update.
25312 * bb-reorder.c (find_traces_1_round): Update.
25313 (better_edge_p): Update.
25314 (sanitize_hot_paths): Update.
25315 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25316 (make_single_succ_edge): Update.
25317 (check_bb_profile): Update.
25318 (dump_edge_info): Update.
25319 (update_bb_profile_for_threading): Update.
25320 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25321 probabilitycount to 0.
25322 * cfgbuild.c (compute_outgoing_frequencies): Update.
25323 * cfgcleanup.c (try_forward_edges): Update.
25324 (outgoing_edges_match): Update.
25325 (try_crossjump_to_edge): Update.
25326 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25327 (expand_gimple_tailcall): Update.
25328 (construct_init_block): Use make_single_succ_edge.
25329 (construct_exit_block): Use make_single_succ_edge.
25330 * cfghooks.c (verify_flow_info): Update.
25331 (redirect_edge_succ_nodup): Update.
25332 (split_edge): Update.
25333 (account_profile_record): Update.
25334 * cfgloopanal.c (single_likely_exit): Update.
25335 * cfgloopmanip.c (scale_loop_profile): Update.
25336 (set_zero_probability): Remove.
25337 (duplicate_loop_to_header_edge): Update.
25338 * cfgloopmanip.h (loop_version): Update prototype.
25339 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25340 (force_nonfallthru_and_redirect): Update.
25341 (update_br_prob_note): Update.
25342 (rtl_verify_edges): Update.
25343 (purge_dead_edges): Update.
25344 (rtl_lv_add_condition_to_bb): Update.
25345 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25346 * cgraphunit.c (init_lowered_empty_function): Update.
25347 (cgraph_node::expand_thunk): Update.
25348 * cilk-common.c: Include profile-count.h
25349 * dojump.c (inv): Remove.
25350 (jumpifnot): Update.
25351 (jumpifnot_1): Update.
25352 (do_jump_1): Update.
25354 (do_jump_by_parts_greater_rtx): Update.
25355 (do_compare_rtx_and_jump): Update.
25356 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25357 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25358 * dwarf2cfi.c: Include profile-count.h
25359 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25360 (sjlj_emit_dispatch_table): Likewise.
25361 * explow.c: Include profile-count.h
25362 * expmed.c (emit_store_flag_force): Update.
25363 (do_cmp_and_jump): Update.
25364 * expr.c (compare_by_pieces_d::generate): Update.
25365 (compare_by_pieces_d::finish_mode): Update.
25366 (emit_block_move_via_loop): Update.
25367 (store_expr_with_bounds): Update.
25368 (store_constructor): Update.
25369 (expand_expr_real_2): Update.
25370 (expand_expr_real_1): Update.
25371 * expr.h (try_casesi, try_tablejump): Update prototypes.
25372 * gimple-pretty-print.c (dump_probability): Update.
25373 (dump_profile): New.
25374 (dump_gimple_label): Update.
25375 (dump_gimple_bb_header): Update.
25376 * graph.c (draw_cfg_node_succ_edges): Update.
25377 * hsa-gen.c (convert_switch_statements): Update.
25378 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25379 (find_if_case_1): Update.
25380 (find_if_case_2): Update.
25381 * internal-fn.c (expand_arith_overflow_result_store): Update.
25382 (expand_addsub_overflow): Update.
25383 (expand_neg_overflow): Update.
25384 (expand_mul_overflow): Update.
25385 (expand_vector_ubsan_overflow): Update.
25386 * ipa-cp.c (good_cloning_opportunity_p): Update.
25387 * ipa-split.c (split_function): Use make_single_succ_edge.
25388 * ipa-utils.c (ipa_merge_profiles): Update.
25389 * loop-doloop.c (add_test): Update.
25390 (doloop_modify): Update.
25391 * loop-unroll.c (compare_and_jump_seq): Update.
25392 (unroll_loop_runtime_iterations): Update.
25393 * lra-constraints.c (lra_inheritance): Update.
25394 * lto-streamer-in.c (input_cfg): Update.
25395 * lto-streamer-out.c (output_cfg): Update.
25396 * mcf.c (adjust_cfg_counts): Update.
25397 * modulo-sched.c (sms_schedule): Update.
25398 * omp-expand.c (expand_omp_for_init_counts): Update.
25399 (extract_omp_for_update_vars): Update.
25400 (expand_omp_ordered_sink): Update.
25401 (expand_omp_for_ordered_loops): Update.
25402 (expand_omp_for_generic): Update.
25403 (expand_omp_for_static_nochunk): Update.
25404 (expand_omp_for_static_chunk): Update.
25405 (expand_cilk_for): Update.
25406 (expand_omp_simd): Update.
25407 (expand_omp_taskloop_for_outer): Update.
25408 (expand_omp_taskloop_for_inner): Update.
25409 * omp-simd-clone.c (simd_clone_adjust): Update.
25410 * optabs.c (expand_doubleword_shift): Update.
25411 (expand_abs): Update.
25412 (emit_cmp_and_jump_insn_1): Update.
25413 (expand_compare_and_swap_loop): Update.
25414 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25415 * predict.c (predictable_edge_p): Update.
25416 (edge_probability_reliable_p): Update.
25417 (set_even_probabilities): Update.
25418 (combine_predictions_for_insn): Update.
25419 (combine_predictions_for_bb): Update.
25420 (propagate_freq): Update.
25421 (estimate_bb_frequencies): Update.
25422 (force_edge_cold): Update.
25423 * profile-count.c (profile_count::dump): Add missing space into dump.
25424 (profile_count::debug): Add newline.
25425 (profile_count::differs_from_p): Explicitly convert to unsigned.
25426 (profile_count::stream_in): Update.
25427 (profile_probability::dump): New member function.
25428 (profile_probability::debug): New member function.
25429 (profile_probability::differs_from_p): New member function.
25430 (profile_probability::differs_lot_from_p): New member function.
25431 (profile_probability::stream_in): New member function.
25432 (profile_probability::stream_out): New member function.
25433 * profile-count.h (profile_count_quality): Rename to ...
25434 (profile_quality): ... this one.
25435 (profile_probability): New.
25436 (profile_count): Update.
25437 * profile.c (compute_branch_probabilities): Update.
25438 * recog.c (peep2_attempt): Update.
25439 * sched-ebb.c (schedule_ebbs): Update.
25440 * sched-rgn.c (find_single_block_region): Update.
25441 (compute_dom_prob_ps): Update.
25442 (schedule_region): Update.
25443 * sel-sched-ir.c (compute_succs_info): Update.
25444 * stmt.c (struct case_node): Update.
25445 (do_jump_if_equal): Update.
25446 (get_outgoing_edge_probs): Update.
25447 (conditional_probability): Update.
25448 (emit_case_dispatch_table): Update.
25449 (expand_case): Update.
25450 (expand_sjlj_dispatch_table): Update.
25451 (emit_case_nodes): Update.
25452 * targhooks.c: Update.
25453 * tracer.c (better_p): Update.
25454 (find_best_successor): Update.
25455 * trans-mem.c (expand_transaction): Update.
25456 * tree-call-cdce.c: Update.
25457 * tree-cfg.c (gimple_split_edge): Upate.
25458 (move_sese_region_to_fn): Upate.
25459 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25460 * tree-eh.c (lower_resx): Upate.
25461 (cleanup_empty_eh_move_lp): Upate.
25462 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25463 * tree-inline.c (copy_edges_for_bb): Update.
25464 (copy_cfg_body): Update.
25465 * tree-parloops.c (gen_parallel_loop): Update.
25466 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25467 (gimple_gen_time_profiler): Update.
25468 * tree-ssa-dce.c (remove_dead_stmt): Update.
25469 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25470 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25471 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25472 (unloop_loops): Update.
25473 (try_peel_loop): Update.
25474 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25475 * tree-ssa-loop-split.c (connect_loops): Update.
25476 (split_loop): Update.
25477 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25478 (hoist_guard): Update.
25479 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25480 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25481 (value_replacement): Update.
25482 * tree-ssa-reassoc.c (branch_fixup): Update.
25483 * tree-ssa-tail-merge.c (replace_block_by): Update.
25484 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25485 (create_edge_and_update_destination_phis): Update.
25486 (compute_path_counts): Update.
25487 (recompute_probabilities): Update.
25488 (update_joiner_offpath_counts): Update.
25489 (freqs_to_counts_path): Update.
25490 (duplicate_thread_path): Update.
25491 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25492 (struct switch_conv_info): Update.
25493 (gen_inbound_check): Update.
25494 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25495 (vect_do_peeling): Update.
25496 (vect_loop_versioning): Update.
25497 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25498 (optimize_mask_stores): Update.
25499 * ubsan.c (ubsan_expand_null_ifn): Update.
25500 * value-prof.c (gimple_divmod_fixed_value): Update.
25501 (gimple_divmod_fixed_value_transform): Update.
25502 (gimple_mod_pow2): Update.
25503 (gimple_mod_pow2_value_transform): Update.
25504 (gimple_mod_subtract): Update.
25505 (gimple_mod_subtract_transform): Update.
25506 (gimple_ic): Update.
25507 (gimple_stringop_fixed_value): Update.
25508 (gimple_stringops_transform): Update.
25509 * value-prof.h: Update.
25511 2017-06-29 Carl Love <cel@us.ibm.com>
25513 * config/rs6000/rs6000-c.c: Add support for built-in functions
25514 vector signed int vec_signed (vector float);
25515 vector signed long long vec_signed (vector double);
25516 vector signed int vec_signed2 (vector double, vector double);
25517 vector signed int vec_signede (vector double);
25518 vector signed int vec_signedo (vector double);
25519 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25520 instruction generator.
25521 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25522 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25523 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25525 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25526 vunsignede_v2df): Add define_expands.
25527 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25528 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25529 VEC_UNSIGNEDO): Add definitions.
25530 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25531 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25532 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25533 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25534 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25535 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25536 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25537 vec_unsignede, vec_unsignedo): Add builtin defines.
25538 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25540 * doc/extend.texi: Update the built-in documentation file for the
25541 new built-in functions.
25543 2017-06-29 Richard Biener <rguenther@suse.de>
25545 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25546 reduction chains to LOOP_VINFO_REDUCTIONS.
25547 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25548 SLP reductions after processing reduction chains.
25550 2017-06-29 Nathan Sidwell <nathan@acm.org>
25552 * builtins.c (fold_builtin_FUNCTION): Use
25553 lang_hooks.decl_printable_name.
25555 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25557 PR middle-end/81194
25558 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25559 with only one label.
25560 * stmt.c (expand_case): Assert NCASES is greater than one.
25562 2017-06-29 Richard Biener <rguenther@suse.de>
25564 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25566 (group_case_labels): Likewise.
25567 (find_taken_edge): Push sanity checking on val to workers...
25568 (find_taken_edge_cond_expr): ... here
25569 (find_taken_edge_switch_expr): ... and here, handle cases
25570 with just a default label.
25571 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25572 (group_case_labels): Likewise.
25573 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25574 group_case_labels does anything cleanup the CFG again.
25576 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25578 PR tree-optimization/81196
25579 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25580 exit condition comparing two IVs.
25582 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25584 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25585 profile to the dummy entry at the end of the list of architectures.
25586 * config/arm/arm-cpu-cdata.h: Regenerated.
25588 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25589 Michael Collison <michael.collison@arm.com>
25592 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25594 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25595 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25596 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25597 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25598 of shift when the shift amount is masked with constant equal to
25599 the size of the mode.
25600 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25603 2017-06-29 Martin Liska <mliska@suse.cz>
25605 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25607 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25610 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25613 (edge_growth_cache_entry::edge_growth_cache_entry): New
25615 (reset_edge_growth_cache): Update to use constructor.
25617 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25619 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25620 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25621 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25623 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25625 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25626 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25628 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25630 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25631 (*-linux-uclibc*): Add t-uclibc tmake_file.
25632 * config/t-musl: New.
25633 * config/t-uclibc: New.
25635 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25637 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25639 (gen_comm_data): Emit architectural setting of arch_prof.
25640 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25642 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25643 (armv8-m.base, armv8-m.main): Likewise.
25644 * arm-protos.h (arm_build_target): Add profile field.
25645 (arch_option): Likewise.
25646 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25648 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25649 arm_active_target.profile.
25651 2017-06-28 Richard Biener <rguenther@suse.de>
25653 PR middle-end/81227
25654 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25655 TYPE_OVERFLOW_WRAPS.
25656 * match.pd (negate_expr_p): Likewise.
25657 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25658 fold_build2, not fold_binary.
25660 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25662 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25663 Convert memory address to Pmode.
25664 (aarch64_print_operand): Assert MEM operands are always Pmode.
25666 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25669 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25670 Remove redundant if.
25671 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25672 * config/arm/aarch-common-protos.h
25673 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25674 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25676 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25679 * multiple_target.c (create_dispatcher_calls): Set the default
25680 clone to be static, not public.
25682 2017-06-28 Richard Biener <rguenther@suse.de>
25684 * tree-vect-loop.c (vectorizable_reduction): Move special
25685 cond reduction IV var creation ...
25686 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25687 parameter. Use STMT_VINFO_VECTYPE.
25688 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25691 2017-06-28 Martin Liska <mliska@suse.cz>
25694 * ipa-visibility.c (non_local_p): Handle visibility.
25696 2017-06-28 Martin Liska <mliska@suse.cz>
25699 * common.opt: Add IntegerRange to various options.
25700 * opt-functions.awk (integer_range_info): New function.
25701 * optc-gen.awk: Add integer_range_info to cl_options struct.
25702 * opts-common.c (decode_cmdline_option): Handle
25703 CL_ERR_INT_RANGE_ARG.
25704 (cmdline_handle_error): Likewise.
25705 * opts.c (print_filtered_help): Show valid interval in
25706 when --help is provided.
25707 * opts.h (struct cl_option): Add range_min and range_max fields.
25708 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25710 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25712 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25713 (x * C EQ/NE y * C): New transformation.
25715 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25717 * genmultilib (combination_space): Accept '+' in option names.
25719 2017-06-28 Martin Liska <mliska@suse.cz>
25722 * asan.c (instrument_derefs): Bail out inner references
25723 that are hard register variables.
25725 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25728 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25729 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25731 2017-06-28 Richard Biener <rguenther@suse.de>
25733 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25734 (vect_get_slp_defs): Adjust.
25735 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25737 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25739 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25740 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25741 (vectorizable_reduction): Adjust.
25742 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25744 (vect_get_slp_defs): Likewise.
25745 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25746 (vectorizable_bswap): Adjust.
25747 (vectorizable_call): Likewise.
25748 (vectorizable_conversion): Likewise.
25749 (vectorizable_assignment): Likewise.
25750 (vectorizable_shift): Likewise.
25751 (vectorizable_operation): Likewise.
25752 (vectorizable_store): Likewise.
25753 (vectorizable_condition): Likewise.
25754 (vectorizable_comparison): Likewise.
25756 2017-06-28 Michael Collison <michael.collison@arm.com>
25759 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25761 (arm_gen_movmemqi): Removed unused variable 'i'.
25762 Convert 'for' loop into 'while' loop.
25763 (arm_expand_prologue): Remove last unnecessary set of insn.
25764 (thumb_pop): Remove unused variable 'pushed_words'.
25765 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25767 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25769 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25770 * config/s390/s390-protos.h: Add prototype of
25771 s390_rel_address_ok_p.
25772 * config/s390/s390.c (s390_got_symbol): New function.
25773 (s390_rel_address_ok_p): New function.
25774 (legitimize_pic_address): Use s390_rel_address_ok_p.
25775 (s390_load_got): Use s390_got_symbol.
25776 (s390_option_override): Issue error if
25777 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25778 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25780 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25782 2017-06-27 Andrew Pinski <apinski@cavium.com>
25784 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25785 (X * copysign (1.0, X)): New pattern.
25786 (X * copysign (1.0, -X)): New pattern.
25787 (copysign (-1.0, CST)): New pattern.
25789 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25791 * genmultilib (combination_space): Remove variable.
25792 Validate reuse rules against regular expression for any sequence
25793 of multilib options in any order.
25795 2017-06-27 Michael Collison <michael.collison@arm.com>
25797 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25798 call aarch64_split_simd_combine.
25799 * (aarch64_combine_internal<mode>): Delete pattern.
25800 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25801 Allow register and subreg operands.
25803 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25805 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25806 specific need, just fallback on defaults.
25807 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25809 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25810 Olivier Hainque <hainque@adacore.com>
25812 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25814 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25815 targets. Pick a default if no particular attempt applied.
25816 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25819 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25821 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25822 (x86_64-wrs-vxworks7): Likewise.
25824 2017-06-27 Marek Polacek <polacek@redhat.com>
25827 * ubsan.c (instrument_null): Check get_base_address's result for null.
25829 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25831 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25833 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25835 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25836 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25837 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25838 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25839 New function types.
25840 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25841 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25842 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25843 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25844 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25845 BUILT_IN_FEUPDATEENV): New builtins.
25846 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25847 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25848 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25849 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25851 (builtin_structptr_types): Adjust size.
25852 * tree.c (builtin_structptr_types): Add four entries.
25854 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25855 Olivier Hainque <hainque@adacore.com>
25857 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25858 (TLS_SYM): New local macro, forcing reference to __tls__ on
25859 link command lines for VxWorks 7 RTPs, triggering initialization
25861 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25862 OS features TLS support, true for RTPs on VxWorks 7.
25863 * config/vxworks.c (vxworks_override_options): Setup emutls
25866 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25868 * predict.c (test_prediction_value_range): Use -1U instead of -1
25869 to avoid narrowing conversion warning.
25870 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25871 to avoid narrowing conversion warning.
25872 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25874 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25876 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25878 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25879 64bit configurations.
25880 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25881 (SIZE_TYPE): Likewise.
25882 * config/vxworks.c (vxworks_emutls_var_fields): Use
25883 long_unsigned_type_node instead of unsigned_type_node as the offset
25884 field type, which is "pointer" mode in emutls.c.
25886 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25889 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25891 PR middle-end/81207
25892 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25893 gimple_vuse copying separately from gimple_vdef copying.
25895 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25897 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25900 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25901 Olivier Hainque <hainque@adacore.com>
25903 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25904 for all vxworks7 targets.
25905 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25906 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25907 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25908 variations for VX6/VX7 and 32/64bits later on in ...
25909 (VXWORKS_LIB_SPEC): Leverage new macros.
25910 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25911 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25913 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25915 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25916 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25918 2017-06-26 Carl Love <cel@us.ibm.com>
25920 * config/rs6000/rs6000-c.c: Add support for built-in functions
25921 vector bool char vec_reve (vector bool char);
25922 vector signed char vec_reve (vector signed char);
25923 vector unsigned char vec_reve (vector unsigned char);
25924 vector bool int vec_reve (vector bool int);
25925 vector signed int vec_reve (vector signed int);
25926 vector unsigned int vec_reve (vector unsigned int);
25927 vector bool long long vec_reve (vector bool long long);
25928 vector signed long long vec_reve (vector signed long long);
25929 vector unsigned long long vec_reve (vector unsigned long long);
25930 vector bool short vec_reve (vector bool short);
25931 vector signed short vec_reve (vector signed short);
25932 vector double vec_reve (vector double);
25933 vector float vec_reve (vector float);
25934 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25935 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25936 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25937 (altivec_vreve): New pattern.
25938 * config/rs6000/altivec.h (vec_reve): New define.
25939 * doc/extend.texi (vec_rev): Update the built-in documentation file
25940 for the new built-in functions.
25942 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25944 PR tree-optimization/71815
25945 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25947 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25949 (slsr_process_phi): Likewise.
25950 (replace_uncond_cands_and_profitable_phis): Don't replace a
25951 multiply candidate with a stride of 1 (copy or cast).
25952 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25954 (lowest_cost_path): Likewise.
25955 (total_savings): Likewise.
25957 2017-06-26 Richard Biener <rguenther@suse.de>
25960 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25961 Use def_builtin_pure for all gather builtins.
25963 2017-06-26 Richard Biener <rguenther@suse.de>
25965 PR tree-optimization/81203
25966 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25967 non-dominating BBs.
25969 2017-06-26 Marek Polacek <polacek@redhat.com>
25972 * doc/invoke.texi: Document -Wmultistatement-macros.
25974 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25976 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25977 arm_neon_ok_no_float_abi effective target.
25979 2017-06-26 Richard Biener <rguenther@suse.de>
25981 PR tree-optimization/80928
25982 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25983 (copy_bbs): Set BB_DUPLICATED flag early.
25984 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25986 (execute_on_shrinking_pred): Likewise.
25987 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25988 BB_DUPLICATED blocks.
25989 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25990 iterate over all PHIs considering removal of *gsi.
25992 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25994 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25997 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25999 * config/rs6000/rs6000-string.c: (expand_block_clear,
26000 do_load_for_compare, select_block_compare_mode,
26001 compute_current_alignment, expand_block_compare,
26002 expand_strncmp_align_check, expand_strn_compare,
26003 expand_block_move, rs6000_output_load_multiple)
26004 Move functions related to string/block move/compare
26005 to a separate file.
26006 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26007 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26008 for this function which is now used in two files.
26009 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26010 * config.gcc: Add rs6000-string.o to extra_objs for
26011 targets powerpc*-*-* and rs6000*-*-*.
26013 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26016 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26017 32-bit, since indexed is not valid for DImode.
26018 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26019 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26020 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26021 (define_peephole2 for Altivec d-form store): Likewise.
26024 * multiple_target.c (create_dispatcher_calls): Only create the
26025 dispatcher call if the function is the default clone of a
26026 versioned function.
26028 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26030 PR middle-end/80902
26031 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26032 a call, force the call to not be a tail call.
26034 2017-06-23 Jeff Law <law@redhat.com>
26036 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26039 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26041 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26042 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26043 handling for early expansion of vector shifts (sl,sr,sra,rl).
26044 (builtin_function_type): Add vector shift right instructions
26045 to the unsigned argument list.
26047 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26049 rtl-optimizatoin/79286
26050 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26051 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26052 trap. PIC register plus a const unspec without offset can never trap.
26054 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26056 * tree.h (builtin_structptr_type): New type.
26057 (builtin_structptr_types): Declare new array.
26058 * tree.c (builtin_structptr_types): New array.
26059 (free_lang_data, build_common_tree_nodes): Use it.
26061 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26064 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26067 2017-06-22 Matt Turner <mattst88@gmail.com>
26069 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26070 Lake models to skylake case. Assume skylake for unknown
26071 models with clflushopt.
26073 2017-06-22 Jeff Law <law@redhat.com>
26075 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26076 frame sizes that do not satisfy aarch64_uimm12_shift.
26078 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26080 * profile-count.h (apply_probability,
26081 apply_scale, probability_in): Fix checks for zero.
26083 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26085 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26086 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26088 2016-06-22 Richard Biener <rguenther@suse.de>
26090 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26091 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26092 REDUC_MAX_EXPR support.
26093 (vectorizable_reduction): Likewise.
26094 (vect_create_epilog_for_reduction): Likewise.
26096 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26098 * match.pd (A / (1 << B) -> A >> B): New.
26099 * generic-match-head.c: Include optabs-tree.h.
26100 * gimple-match-head.c: Likewise.
26101 * optabs-tree.h (target_supports_op_p): New.
26102 * optabs-tree.c (target_supports_op_p): New.
26104 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26106 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26107 $gcc_cv_ld --help output.
26108 (gcc_cv_ld_demangle): Likewise.
26109 (gcc_cv_ld_eh_frame_hdr): Likewise.
26110 (gcc_cv_ld_pie): Likewise.
26111 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26112 (gcc_cv_ld_buildid): Likewise.
26113 (gcc_cv_ld_sysroot): Likewise.
26114 (ld_bndplt_support): Likewise.
26115 (ld_pushpopstate_support): Likewise.
26116 * configure: Regenerate.
26117 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26119 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26122 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26123 operands indexes to avoid gap between operands and match_dups.
26125 2017-06-21 Andrew Pinski <apinski@cavium.com>
26127 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26128 Increment Arith_shift and Arith_shift_reg by 1.
26129 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26131 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26132 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26133 (aarch64_strip_extend): Add new argument and test for it.
26134 (aarch64_cheap_mult_shift_p): New function.
26135 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26136 add a cost if it is true.
26137 Update calls to aarch64_strip_extend.
26138 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26140 2017-06-21 Andrew Pinski <apinski@cavium.com>
26142 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26144 (thunderxt88): Likewise.
26145 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26146 (thunderx_prefetch_tune): New variable.
26147 (thunderx2t99_prefetch_tune): Update for the correct values.
26148 (thunderxt88_tunings): New variable.
26149 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26150 generic_prefetch_tune.
26151 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26153 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26155 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26156 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26157 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26158 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26159 (aarch64_atomic_cas<mode>, GPI): Likewise.
26161 2017-06-21 Martin Liska <mliska@suse.cz>
26163 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26164 statements on cold and hot labels.
26165 * predict.c (tree_estimate_probability_bb): Remove the
26166 prediction from this place.
26168 2017-06-21 Martin Liska <mliska@suse.cz>
26170 PR tree-optimization/79489
26171 * gimplify.c (maybe_add_early_return_predict_stmt): New
26173 (gimplify_return_expr): Call the function.
26174 * predict.c (tree_estimate_probability_bb): Remove handling
26176 * predict.def: Update comment about early return predictor.
26177 * gimple-predict.h (is_gimple_predict): New function.
26178 * predict.def: Change default value of early return to 66.
26179 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26181 * passes.def: Put pass_strip_predict_hints to the beginning of
26184 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26186 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26187 FUNCTION_DECL declarations.
26188 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26190 (dwaf2out_decl): Likewise.
26191 * godump.c (go_early_global_decl): Skip call to the real debug hook
26192 for FUNCTION_DECL declarations.
26193 * passes.c (rest_of_decl_compilation): Skip call to the
26194 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26195 -fdump-go-spec is passed.
26197 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26199 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26200 Reorder for compactness.
26201 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26202 (def_builtin_pure, def_builtin_pure2): New functions.
26203 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26205 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26207 * match.pd (nop_convert): New predicate.
26208 ((A +- CST1) +- CST2): Allow some NOP conversions.
26210 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26213 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26214 with ctors/dtors if GOVD_SHARED is set.
26216 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26218 * config/aarch64/aarch64.md (movti_aarch64):
26219 Emit mov rather than orr.
26220 (movtf_aarch64): Likewise.
26221 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26222 Emit mov rather than orr.
26224 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26226 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26227 Swap alternatives, make integer dup more expensive.
26229 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26231 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26232 Return true for non-tls symbols.
26234 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26236 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26237 (cortex-a75): Likewise.
26238 (cortex-a75.cortex-a55): Likewise.
26239 * config/aarch64/aarch64-tune.md: Regenerate.
26240 * doc/invoke.texi (-mtune): Document new values for -mtune.
26242 2017-06-21 Tom de Vries <tom@codesourcery.com>
26244 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26245 stack_size feature.
26246 (Effective-Target Keywords, Other attributes): Suggest using
26247 dg-add-options stack_size feature to get stack limit in stack_size
26248 effective target documentation.
26250 2017-06-21 Julian Brown <julian@codesourcery.com>
26251 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26253 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26254 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26255 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26257 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26258 attribute type list for neon_multiply.
26259 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26260 attribute type list for neon_multiply.
26261 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26262 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26263 attribute type list for neon_multiply.
26264 * config/arm/types.md (crypto_pmull): Add.
26265 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26266 attribute type list.
26268 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26270 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26273 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26275 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26276 to make sure not to dereference a NULL cost_classes_ptr pointer.
26278 2017-06-20 Carl Love <cel@us.ibm.com>
26280 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26281 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26282 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26283 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26284 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26285 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26286 VMULOSW): New enum "unspec" values.
26287 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26288 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26289 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26290 altivec_vmulosw): New patterns.
26291 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26292 VMULOSW): Add definitions.
26294 2017-06-20 Julia Koval <julia.koval@intel.com>
26296 * config/i386/i386.c: Fix rounding expand for new pattern.
26297 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26299 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26301 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26302 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26304 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26306 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26309 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26311 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26312 architecture, then by implementer ID.
26313 * config/aarch64/aarch64-tune.md: Regenerate.
26315 2017-06-20 Richard Biener <rguenther@suse.de>
26317 PR middle-end/81097
26318 * fold-const.c (split_tree): Fold to type before negating.
26320 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26322 * diagnostic-show-locus.c
26323 (selftest::test_fixit_deletion_affecting_newline): New function.
26324 (selftest::diagnostic_show_locus_c_tests): Call it.
26326 2017-06-20 Andreas Schwab <schwab@suse.de>
26329 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26332 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26334 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26335 __ARM_FEATURE_COPROC according to support.
26337 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26339 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26340 Rewritten to avoid overflow for > 32-bit pointers.
26343 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26344 by removing enum keyword.
26345 (ubsan_type_descriptor): Likewise. Formatting fix.
26348 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26349 splitter): Require TARGET_SSE2 in the condition.
26351 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26354 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26355 for doing vector set of SFmode on ISA 3.0.
26356 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26357 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26359 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26360 SFmode value into a V4SF variable that was extracted from another
26361 V4SF variable without converting the element to double precision
26362 and back to single precision vector format.
26363 (vsx_insert_extract_v4sf_p9_2): Likewise.
26365 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26367 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26368 in UWHI to avoid undefined overflow.
26371 * ubsan.h (enum ubsan_encode_value_phase): New.
26372 (ubsan_encode_value): Change second argument to
26373 enum ubsan_encode_value_phase with default value of
26374 UBSAN_ENCODE_VALUE_GENERIC.
26375 * ubsan.c (ubsan_encode_value): Change second argument to
26376 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26377 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26378 create_tmp_var_raw instead of create_tmp_var and use a
26380 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26381 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26382 ubsan_encode_value callers.
26385 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26386 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26387 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26389 2017-06-19 Richard Biener <rguenther@suse.de>
26391 PR middle-end/81118
26392 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26393 estimates if we changed anything.
26395 2017-06-19 Richard Biener <rguenther@suse.de>
26397 PR tree-optimization/80887
26398 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26399 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26400 simplified lookups, then reset mprts_hook.
26401 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26403 (try_to_simplify): Likewise.
26405 2017-06-19 Martin Liska <mliska@suse.cz>
26408 * gimplify.c (gimplify_switch_expr):
26409 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26411 2017-06-19 Martin Liska <mliska@suse.cz>
26413 * doc/install.texi: Document that PGO runs in 4 stages.
26415 2017-06-19 Martin Liska <mliska@suse.cz>
26418 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26419 to dispatcher function name.
26420 * multiple_target.c (replace_function_decl): New function.
26421 (create_dispatcher_calls): Redirect both edges and references.
26423 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26425 * profile-count.c (profile_count::dump): Dump quality.
26426 (profile_count::differs_from_p): Update for unsigned val.
26427 * profile-count.h (profile_count_quality): New enum.
26428 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26430 2017-06-19 Richard Biener <rguenther@suse.de>
26432 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26433 struct function as arg.
26434 (estimate_numbers_of_iterations): Export overload with loop arg.
26435 (free_numbers_of_iterations_estimates_loop): Use an overload of
26436 free_numbers_of_iterations_estimates instead.
26437 * tree-cfg.c (remove_bb): Adjust.
26438 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26439 * tree-parloops.c (gen_parallel_loop): Likewise.
26440 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26442 (tree_unroll_loops_completely): Likewise.
26443 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26444 Use an overload instead and export.
26445 (estimated_loop_iterations): Adjust.
26446 (max_loop_iterations): Likewise.
26447 (likely_max_loop_iterations): Likewise.
26448 (estimate_numbers_of_iterations): Take struct function as arg
26450 (loop_exits_before_overflow): Adjust.
26451 (free_numbers_of_iterations_estimates_loop): Use an overload.
26452 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26453 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26455 2017-06-19 Richard Biener <rguenther@suse.de>
26458 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26459 RANGE_EXPR conservatively.
26461 2017-06-16 Carl Love <cel@us.ibm.com>
26463 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26464 definitions for vec_float, vec_float2, vec_floato,
26465 vec_floate built-ins.
26466 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26467 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26469 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26470 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26471 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26472 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26473 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26474 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26475 vec_floato): Add builtin defines.
26476 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26477 Update the built-in documentation file for the new built-in
26480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26482 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26483 (mthumb): Mark as the negative of -marm.
26485 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26487 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26489 (ARM Options, -mtune): Document that this accepts the same
26490 extension options as -mcpu.
26491 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26495 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26496 permitted extensions.
26498 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26500 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26501 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26502 (armv8-m.main): Add option +nodsp.
26503 * config/arm/arm-cpu-cdata.h: Regenerated.
26505 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26507 * config/arm/t-fuchsia: New file.
26508 * config.gcc (arm*-*-fuchsia*): Use it.
26510 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26512 * config/arm/t-symbian: Rewrite for new option infrastructure.
26514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26516 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26517 (MULTILIB_REQUIRED): Likewise.
26519 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26521 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26522 (MULTILIB_RESUE): Likewise.
26523 (MULTILIB_MATCHES): Likewise.
26524 (MULTLIB_REQUIRED): Likewise.
26526 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26528 * config/arm/t-rtems: Rewrite for new option framework.
26530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26532 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26533 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26534 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26535 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26536 * config/arm/t-multilib: ... here.
26537 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26538 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26539 armv7-a and armv8*-a when A-profile libraries have not been built.
26540 * config/arm/t-rmprofile: Rewrite.
26542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26544 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26545 with a backslash. Remove the backslash after substituting unescaped
26547 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26551 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26552 the driver to pass through the default setting of -mfloat-abi.
26553 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26555 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26556 (all_feat_combs): New rule.
26557 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26559 * config/arm/t-aprofile: Rewrite.
26561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26563 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26564 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26565 fpu is not specified by the user/command-line.
26566 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26567 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26568 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26569 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26570 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26571 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26575 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26576 * config/arm/t-arm-elf: Rewritten.
26578 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26580 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26581 have some floating-point instructions.
26582 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26583 (TARGET_MAYBE_HARD_FLOAT): New macro.
26584 * config/arm/arm-builtins.c (arm_init_builtins): Use
26585 TARGET_MAYBE_HARD_FLOAT.
26586 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26590 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26591 (configargs.h): Include it.
26592 (arm_print_hint_for_fpu_option): New function.
26593 (arm_parse_fpu_option): New function.
26594 (candidate_extension): New class.
26595 (arm_canon_for_multilib): New function.
26596 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26597 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26598 (ARCH_CANONICAL_SPECS): New macro.
26599 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26601 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26603 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26604 are set after handling multilib fragments. Set target_cpu_default2
26607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26609 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26611 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26612 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26614 (check_arch): Likewise.
26615 * config/arm/arm.c (arm_configure_build_target): Handle
26616 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26617 options in the default.
26619 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26621 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26622 when an option is an alias of another.
26623 * config/arm/parsecpu.awk (optalias): New parser token.
26624 (gen_comm_data): Mark non-alias options as such. Emit entries
26625 for extension aliases.
26626 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26627 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26628 (armv6kz, armv6zk, armv6t2): Likewise.
26629 (armv7): Make vfpv3-d16 an alias.
26630 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26632 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26633 Sort in canonical order.
26634 (armv8-a): Sort in canonical order.
26635 (armv8.1-a, armv8.2-a): Likewise.
26636 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26638 (cortex-a9): Sort in canonical order.
26639 * config/arm/arm.c (selftests.h): Include it.
26640 (arm_test_cpu_arch_data): New function.
26641 (arm_run_self_tests): New function.
26642 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26643 (targetm): Move declaration to the end of the file.
26644 * arm-cpu-cdata.h: Regenerated.
26646 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26648 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26649 call to target_mode_check describing the type of option passed.
26650 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26651 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26652 arm_parse_cpu_option_name to match parameters against list of
26654 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26655 arm_arch_core_flags data structure.
26656 * config/arm/arm-cpu_cdata.h: Regenerated.
26658 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26660 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26662 (arm_print_hint_for_cpu_option): Likewise.
26663 (arm_print_hint_for_arch_option): Likewise.
26664 (arm_parse_cpu_option_name): Likewise.
26665 (arm_parse_arch_option_name): Likewise.
26666 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26667 of entries in the all_fpus list.
26668 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26669 (arm_parse_cpu_option_name): Declare.
26670 (arm_parse_arch_option_name): Declare.
26671 (arm_parse_option_features): Declare.
26672 (arm_intialize_isa): Declare.
26673 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26675 (gen_comm_data): ... here. Make definitions non-static.
26676 * config/arm/arm-cpu-data.h: Regenerated.
26677 * config/arm/arm-cpu-cdata.h: Regenerated.
26679 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26681 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26682 (cpu_arch_extension): New structure.
26683 (cpu_arch_option, arch_option, cpu_option): New structures.
26684 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26685 architecture types.
26686 (gen_data): Generate new format data tables.
26687 * config/arm/arm.c (cpu_tune): New structure.
26688 (cpu_option, processors): Delete.
26689 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26690 (arm_print_hint_for_cpu_option): ... this and ...
26691 (arm_print_hint_for_arch_option): ... this.
26692 (arm_parse_arch_cpu_name): Delete. Replace with ...
26693 (arm_parse_cpu_option_name): ... this and ...
26694 (arm_parse_arch_option_name): ... this.
26695 (arm_unrecognized_feature): Change type of target parameter to
26697 (arm_parse_arch_cpu_features): Delete. Replace with ...
26698 (arm_parse_option_features): ... this.
26699 (arm_configure_build_target): Rework to use new configuration data
26701 (arm_print_tune_info): Rework for new configuration data tables.
26702 * config/arm/arm-cpu-data.h: Regenerated.
26703 * config/arm/arm-cpu.h: Regenerated.
26705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26707 * Makefile.in (OBJS): Move sbitmap.o from here ...
26708 (OBJS-libcommon): ... to here.
26710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26712 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26713 (ISA_ALL_CRYPTO): New macro.
26714 (ISA_ALL_SIMD): New macro
26715 (ISA_ALL_FP): New macro.
26716 * config/arm/arm.c (fpu_bitlist): Update initializer.
26717 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26719 (arm9e): Add fpu. Add option for nofp
26720 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26721 (arm926ej-s, arm1026ej-s): Likewise.
26722 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26723 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26724 neon-fp16, neon-vfpv4, nofp and nosimd.
26725 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26726 (cortex-a8): Add fpu. Add option for nofp.
26727 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26728 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26729 (cortex-r4f): Add fpu.
26730 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26731 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26733 (cortex-r8): Likewise.
26734 (cortex-m4): Add fpu. Add option for nofp.
26735 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26736 (cortex-a17.cortex-a7): Likewise.
26737 (cortex-a32): Add fpu. Add options for crypto and nofp.
26738 (cortex-a35, cortex-a53): Likewise.
26739 (cortex-a57): Add fpu. Add option for crypto.
26740 (cortex-a72, cortex-a73): Likewise.
26741 (exynos-m1): Likewise.
26742 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26743 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26744 (cortex-m33): Add fpu. Add option for nofp.
26745 * config/arm/arm-cpu-cdata.h: Regenerated
26746 * config/arm/arm-cpu-data.h: Regenerated.
26748 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26750 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26751 (armv5te, armv5tej): Likewise.
26752 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26753 (armv7): Add options fp and vfpv3-d16.
26754 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26755 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26757 (armv7ve): Likewise.
26758 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26759 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26760 (armv8-a): Add nocrypto option.
26761 (armv8.1-a, armv8.2-a): Likewise.
26762 (armv8-m.main): add options fp, fp.dp and nofp.
26764 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26766 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26768 (armv8-a+crc): Delete.
26769 (armv8.1-a): Add options simd, crypto and nofp.
26770 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26771 (armv8.2-a+fp16): Delete.
26772 (armv8-m.main): Add option dsp.
26773 (armv8-m.main+dsp): Delete.
26774 (cortex-a8): Add fpu. Add nofp option.
26775 (cortex-a9): Add fpu. Add nofp and nosimd options.
26776 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26777 link to main cpu and architecture data structures.
26778 (gen_comm_data): Only put isa attributes from the main architecture
26780 (option): New statement for architecture and CPU entries.
26781 * arm.c (struct cpu_option): New structure.
26782 (struct processors): Add entry for options.
26783 (arm_unrecognized_feature): New function.
26784 (arm_parse_arch_cpu_name): Ignore any characters after the first
26786 (arm_parse_arch_cpu_feature): New function.
26787 (arm_configure_build_target): Separate out any CPU and architecture
26788 features and parse separately. Don't error out if -mfpu=auto is
26789 used with only an architecture string.
26790 (arm_print_asm_arch_directives): New function.
26791 (arm_file_start): Call it.
26792 * config/arm/arm-cpu-cdata.h: Regenerated.
26793 * config/arm/arm-cpu-data.h: Likewise.
26794 * config/arm/arm-tables.opt: Likewise.
26796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26798 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26799 assembler when it is not -mfpu=auto.
26801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26803 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26804 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26805 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26806 (ASM_CPU_SPEC): Rewrite.
26807 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26808 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26809 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26810 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26811 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26812 copied string is NUL-terminated. Also strip any characters prefixed
26814 (arm_rewrite_selected_arch): New function.
26815 (arm_rewrite_march): New function.
26817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26819 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26820 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26821 (march, mcpu, mtune): Convert to string-based options.
26822 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26823 (arm_parse_arch_cpu_name): New function.
26824 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26825 identify selected architecture or CPU.
26826 (arm_option_save): New function.
26827 (TARGET_OPTION_SAVE): Redefine.
26828 (arm_option_restore): Restore string options.
26829 (arm_option_print): Print string options.
26831 2017-06-16 Martin Sebor <msebor@redhat.com>
26833 PR tree-optimization/80933
26834 PR tree-optimization/80934
26835 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26836 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26837 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26838 (gimple_fold_builtin): Call them.
26840 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26842 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26843 as unlikely; update profile.
26845 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26847 * predict.c (force_edge_cold): Handle declaring edges impossible
26850 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26852 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26854 (try_unroll_loop_completely): Fix reporting.
26856 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26858 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26860 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26863 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26864 if given a non-constant argument for an intrinsic which requires a
26867 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26869 * profile.c (compare_freqs): New function.
26870 (branch_prob): Sort edge list.
26871 (find_spanning_tree): Assume that the list is priority sorted.
26873 2017-06-16 Richard Biener <rguenther@suse.de>
26875 PR tree-optimization/81090
26876 * passes.def (pass_record_bounds): Remove.
26877 * tree-pass.h (make_pass_record_bounds): Likewise.
26878 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26879 make_pass_record_bounds): Likewise.
26880 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26881 not free niter estimates at the beginning but at the end.
26882 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26884 2017-06-16 Richard Biener <rguenther@suse.de>
26886 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26887 initializer to workaround ICE in host GCC 4.8.
26889 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26891 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26893 (clone_inlined_nodes): Update.
26895 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26897 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26898 prefetch settings, and enable prefetching by default at -O3.
26900 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26902 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26903 Set flag_prefetch_loop_arrays according to tuning data.
26905 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26907 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26908 New tune structure.
26909 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26910 [Unrelated to main purpose of the patch] Place the pointer field last
26911 to enable type checking errors when tune structure are wrongly merged.
26912 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26913 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26914 (thunderx2t99_prefetch_tune): New tune constants.
26915 (tune_params *_tunings): Update all tunings (no functional change).
26916 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26917 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26918 from tunings structures.
26920 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26923 * ubsan.c (instrument_null): Add T argument, use it instead
26924 of computing it based on IS_LHS.
26925 (instrument_object_size): Likewise.
26926 (pass_ubsan::execute): Adjust instrument_null and
26927 instrument_object_size callers to pass gimple_get_lhs or
26928 gimple_assign_rhs1 result to it. Use instrument_null instead of
26929 calling get_base_address and instrument_mem_ref. Handle
26930 aggregate call arguments for object-size sanitization.
26932 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26934 PR tree-optimization/81089
26935 * tree-vrp.c (is_masked_range_test): Validate operands of
26938 2017-06-15 Martin Sebor <msebor@redhat.com>
26941 * dumpfile.c (dump_register): Avoid calling memset to initialize
26942 a class with a default ctor.
26943 * gcc.c (struct compiler): Remove const qualification.
26944 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26945 * hash-table.h: Ditto.
26946 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26948 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26949 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26951 * params.h (struct param_info): Make struct members non-const.
26952 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26953 with default initialization.
26954 * vec.h (vec_copy_construct, vec_default_construct): New helper
26956 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26957 with vec_copy_construct.
26958 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26959 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26960 * doc/invoke.texi (-Wclass-memaccess): Document.
26962 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26964 * emit-rtl.h (is_leaf): Update comment about local
26965 register allocator.
26967 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26970 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26971 for a variable to have a section before checking if the section has a
26973 Set section to.persistent if persistent attribute is set.
26974 Warn if .persistent attribute is used on an automatic variable.
26976 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26978 PR rtl-optimization/80474
26979 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26981 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26983 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26986 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26988 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26989 (casesi): Emit bounds checking as RTL.
26990 (casesi_internal_mips16_<mode>): Remove bounds checking.
26992 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26994 * config/xtensa/xtensa.c (xtensa_option_override): Append
26995 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26996 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26997 xtensa_doloop_hooks): Define unconditionally.
26998 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27000 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27001 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27002 for it in xtensa_option_override.
27003 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27004 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27006 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27008 * doc/cppopts.texi: Document '-' special value to -MF.
27010 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27012 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27013 (cortex_a53_fconst): Likewise.
27014 (cortex_a53_fpmul): Likewise.
27015 (cortex_a53_f_load_64): Likewise.
27016 (cortex_a53_f_load_many): Likewise.
27017 (cortex_a53_advsimd_alu): Likewise.
27018 (cortex_a53_advsimd_alu_q): Likewise.
27019 (cortex_a53_advsimd_mul): Likewise.
27020 (cortex_a53_advsimd_mul_q): Likewise.
27021 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27022 Add missing fmul, r2f_cvt and fconst cases.
27024 2017-06-14 Richard Biener <rguenther@suse.de>
27026 PR middle-end/81088
27027 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27029 (fold_binary_loc): When associating do not treat pre-existing
27030 TREE_OVERFLOW on literal constants as a reason to allow
27031 TREE_OVERFLOW on associated literal constants.
27033 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27035 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27036 (MASK_FEATURES): New macro.
27037 * config/sparc/sparc.c (sparc_option_override): Remove the special
27038 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27040 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27042 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27043 a division of 0 if non-call exceptions are enabled.
27045 2017-06-14 Andrew Pinski <apinski@cavium.com>
27046 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27049 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27050 Improve vector initialization code gen for only variable case.
27052 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27054 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27056 2017-06-14 Richard Biener <rguenther@suse.de>
27058 PR tree-optimization/81083
27059 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27062 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27064 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27065 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27066 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27067 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27068 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27069 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27071 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27073 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27074 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27076 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27078 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27080 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27082 * config/rs6000/t-rtems: Don't handle SPE.
27084 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27086 * config/rs6000/t-linux: Don't handle SPE.
27088 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27090 * config/rs6000/eabispe.h: Delete file.
27092 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27094 * config/rs6000/t-spe: Delete file.
27096 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27098 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27099 (rs6000_legitimate_offset_address_p): Return false for anything in
27100 V2SImode or V2SFmode.
27102 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27104 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27105 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27106 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27108 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27109 (rs6000_legitimate_offset_address_p): Ditto.
27110 (rs6000_emit_move): Ditto.
27111 (rs6000_init_builtins): Remove V4HI_type_node.
27113 2017-06-13 Martin Liska <mliska@suse.cz>
27116 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27117 (gate_asan): Likewise.
27118 * asan.h (asan_no_sanitize_address_p): Remove the function.
27119 (sanitize_flags_p): New function.
27120 * builtins.def: Fix coding style.
27121 * common.opt: Use renamed enum value.
27122 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27123 * doc/extend.texi: Document no_sanitize attribute.
27124 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27125 to SANITIZE_UNDEFINED_NONDEFAULT.
27126 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27127 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27128 Use sanitize_flags_p.
27129 * gimplify.c (gimplify_function_tree): Likewise.
27130 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27131 * opts.c (parse_no_sanitize_attribute): New function.
27132 (common_handle_option): Use renamed enum value.
27133 * opts.h (parse_no_sanitize_attribute): Declare.
27134 * tree.c (sanitize_flags_p): New function.
27135 * tree.h: Declared here.
27136 * tsan.c: Use sanitize_flags_p.
27137 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27138 (instrument_mem_ref): Likewise.
27139 (instrument_bool_enum_load): Likewise.
27140 (do_ubsan_in_current_function): Remove the function.
27141 (pass_ubsan::execute): Use sanitize_flags_p.
27142 * ubsan.h: Remove do_ubsan_in_current_function
27143 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27144 (dump_function_to_file): Use it here.
27146 2017-06-13 Martin Jambor <mjambor@suse.cz>
27148 PR tree-optimization/80803
27149 PR tree-optimization/81063
27150 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27151 (propagate_subaccesses_across_link): Enqueue subtree whenever
27152 necessary instead of relying on the caller.
27154 2017-06-13 Martin Jambor <mjambor@suse.cz>
27156 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27157 that have a first_link.
27158 (sort_and_splice_var_accesses): Do not check first_link before
27160 (subtree_mark_written_and_enqueue): Likewise.
27161 (propagate_all_subaccesses): Likewise and do not stop at first
27162 parent with a first_link.
27164 2017-06-13 Martin Jambor <mjambor@suse.cz>
27166 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27169 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27171 * match.pd: New pattern.
27173 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27175 * tree-vrp.c (is_masked_range_test): New function.
27176 (register_edge_assert_for): Determine ranges for
27179 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27181 PR tree-optimization/67328
27182 * fold-const.c (maskable_range_p): New function.
27183 (build_range_check): Generate bittests if possible.
27185 2017-06-13 Martin Liska <mliska@suse.cz>
27187 * gimple-pretty-print.c (dump_probability): Add new argument.
27188 (dump_edge_probability): Dump both probability and count.
27189 (dump_gimple_label): Likewise.
27190 (dump_gimple_bb_header): Likewise.
27192 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27195 * config/avr/avr-devices.c: Fix indentation.
27196 * config/avr/gen-avr-mmcu-specs.c: Dito.
27198 2017-06-13 Richard Biener <rguenther@suse.de>
27200 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27201 instead get vector type from stmt_info.
27202 (vectorizable_reduction): Adjust. Remove dead code.
27204 2017-06-13 Richard Biener <rguenther@suse.de>
27206 PR middle-end/81065
27207 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27208 case of C * (x * C2 + C3).
27209 (fold_addr_of_array_ref_difference): Properly fold index difference.
27211 2017-06-12 David S. Miller <davem@davemloft.net>
27214 * config/sparc/sparc.md (return expander): Emit frame blockage if
27215 function uses alloca.
27217 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27219 * combine.c (make_field_assignment): Check len rather than the mode
27220 precision when calling force_to_mode.
27222 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27224 Support multilibs and devices that see flash in RAM address range.
27227 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27228 (avr_mcu_t) <flash_pm_offset>: New field.
27229 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27230 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27231 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27232 (AVR_TINY_PM_OFFSET): Remove macro.
27233 * config/avr/avr.opt (-mshort-calls): New option.
27234 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27235 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27236 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27237 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27238 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27239 instead of avr_arch->have_jmp_call.
27240 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27241 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27242 avr_arch->flash_pm_offset to define.
27243 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27244 new field flash_pm_offset. Add entry for avrxmega3.
27245 (avr_texinfo): Add entry for avrxmega3.
27246 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27247 attiny212, attiny214,
27248 attiny412, attiny414, attiny416, attiny417,
27249 attiny814, attiny816, attiny817,
27250 attiny1614, attiny1616, attiny1617,
27251 attiny3214, attiny3216, attiny3217.
27252 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27253 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27254 (avr_print_operand_address) [AVR_TINY]: Same.
27255 (avr_asm_init_sections) <readonly_data_section>: Only patch
27256 callback if avr_arch->flash_pm_offset = 0.
27257 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27258 for rodata if avr_arch->flash_pm_offset != 0.
27259 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27260 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27261 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27262 (m_options): Append opt_rcall.
27263 (m_dirnames): Append dir_rcall.
27264 * config/avr/t-multilib: Regenerate.
27266 * configure.ac [target=avr]: Check whether avrxmega3 default
27267 linker description file works as needed.
27268 * configure: Regenerate.
27269 * doc/avr-mmcu.texi: Regenerate.
27270 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27271 <__AVR_ARCH__>: Document avrxmega3 and 103.
27272 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27273 <__AVR_SHORT_CALLS__>: Document it.
27274 <__AVR_PM_BASE_ADDRESS__>: Document it.
27275 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27276 (AVR Variable Attributes) <progmem>: Document this is
27277 not needed for avrxmega3.
27278 (AVR Named Address Spaces) <__flash>: Dito.
27280 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27282 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27284 2017-06-12 Doug Rupp <rupp@adacore.com>
27286 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27287 Append vxworks-stdint.h to the tm_file list.
27288 * config/vxworks-stdint.h: New file.
27290 2017-06-12 Martin Liska <mliska@suse.cz>
27292 PR tree-optimization/81041
27293 * tree-profile.c (gimple_gen_ic_func_profiler):
27294 Create an extra BB in profile-generate
27295 (gimple_gen_time_profiler): Likewise.
27297 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27299 PR tree-optimization/81003
27300 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27301 (update_range_test): Use it instead of force_gimple_operand_gsi.
27303 2017-06-12 Richard Biener <rguenther@suse.de>
27305 PR tree-optimization/81053
27306 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27307 with backedge value not defined in loop. Simplify def stmt
27310 2017-06-11 Tom de Vries <tom@codesourcery.com>
27313 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27315 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27316 nvptx_cannot_force_const_mem.
27318 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27320 * opts.c (finish_options): Move test for flag_split_stack after
27321 it has been initialized.
27323 2017-06-11 Jason Merrill <jason@redhat.com>
27325 * tree.h (id_equal): New.
27326 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27327 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27328 instead of strcmp of IDENTIFIER_POINTER.
27330 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27332 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27333 (mark_all_inlined_calls_cdtor): Fix formating.
27334 (inline_transform): Rescale profile before inlining.
27336 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27338 * cgraph.h (cgraph_edge::clone): Update prototype.
27339 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27340 (cgraph_node::create_clone): Update.
27341 (cgraph_node::create_version_clone): Update.
27342 * tree-inline.c (copy_bb): Update.
27343 (expand_call_inline): Update.
27345 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27347 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27348 factored out from ...
27349 (rs6000_emit_prologue): ... here.
27351 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27353 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27354 factored out from ...
27355 (rs6000_emit_prologue): ... here.
27357 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27359 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27361 (handle_missing_profiles): Fix computation of tp_first_run.
27362 (counts_to_freqs): Do not touch freqs when count is 0.
27364 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27366 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27369 2017-06-10 Tom de Vries <tom@codesourcery.com>
27371 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27372 attributes): Document signal effective target.
27374 2017-06-10 Tom de Vries <tom@codesourcery.com>
27376 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27377 Document effective target stack_size.
27379 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27381 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27382 to the edit_context if they can be auto-applied.
27384 2017-06-9 Ian Lance Taylor <iant@golang.org>
27386 * opts.c (finish_options): If -fsplit-stack, disable implicit
27387 -forder-blocks-and-partition.
27388 * doc/invoke.texi (Optimize Options): Document that when using
27389 -fsplit-stack -forder-blocks-and-partition is not implicitly
27392 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27394 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27395 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27396 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27397 * builtins.def (abort, trap, unreachable): Declare cold.
27398 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27399 * tree-core.h (ECF_COLD): New.
27400 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27401 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27403 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27405 * predict.c (unlikely_executed_stmt_p): Cleanup.
27407 2017-06-09 Richard Biener <rguenther@suse.de>
27409 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27410 model if the ref is always written to.
27412 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27414 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27416 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27418 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27421 2017-06-09 Tom de Vries <tom@codesourcery.com>
27424 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27425 "target cannot support label values" when encountering LABEL_REF.
27427 2017-06-09 Martin Liska <mliska@suse.cz>
27429 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27430 (gimple_gen_ic_func_profiler): Emit direct comparison
27431 of __gcov_indirect_call_callee with NULL.
27432 (gimple_gen_time_profiler): Change probability from
27433 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27435 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27437 * profile.c (edge_gcov_counts): Turn to pointer.
27438 (compute_branch_probabilities, compute_branch_probabilities): Update.
27439 (branch_prob): Do not clear edge_gcov_count.
27440 * profile.h (edge_gcov_counts): Turn to pointer.
27441 (edge_gcov_count): Update.
27443 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27445 * gimple.h (gimple_check_failed): Mark cold.
27447 2017-06-09 Richard Biener <rguenther@suse.de>
27449 PR tree-optimization/66623
27450 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27451 refactor check_reduction into two parts, properly computing
27452 whether we have to check reduction validity for outer loop
27455 2017-06-09 Richard Biener <rguenther@suse.de>
27457 PR tree-optimization/79483
27458 * graphite-scop-detection.c (order): New global.
27459 (get_order): Compute bb to order mapping that satisfies code
27460 generation constraints.
27461 (cmp_pbbs): New helper.
27462 (build_scops): Start domwalk at entry block, sort generated
27465 2017-06-09 Richard Biener <rguenther@suse.de>
27467 PR middle-end/81007
27468 * ipa-polymorphic-call.c
27469 (ipa_polymorphic_call_context::restrict_to_inner_class):
27470 Skip FIELD_DECLs with error_mark_node type.
27471 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27474 2017-06-09 Martin Liska <mliska@suse.cz>
27476 * predict.c (struct branch_predictor): New struct.
27477 (test_prediction_value_range): New test.
27478 (predict_c_tests): New function.
27479 * selftest-run-tests.c (selftest::run_tests): Run the function.
27480 * selftest.h: Declare new tests.
27482 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27485 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27486 gen_add3_insn did not fail.
27487 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27488 r0, construct that number in a temporary reg and add that reg to r0.
27489 If asked to put the result in r0 as well, fail.
27491 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27493 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27494 for early expansion of vec_eqv.
27496 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27498 PR middle-end/81005
27499 * ubsan.c (instrument_null): Avoid pointless code temporary.
27500 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27502 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27506 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27509 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27512 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27514 * predict.c (unlikely_executed_edge_p): Move ahead.
27515 (probably_never_executed_edge_p): Use it.
27517 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27519 PR middle-end/79988
27520 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27521 gimple_call_builtin_p call.
27523 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27525 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27526 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27527 rtl_check_failed_type2, rtl_check_failed_code1,
27528 rtl_check_failed_code2, rtl_check_failed_code_mode,
27529 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27530 rtvec_check_failed_bounds, rtl_check_failed_flag,
27531 _fatal_insn_not_found, _fatal_insn): Likewise.
27532 * tree.h (tree_contains_struct_check_failed,
27533 tree_check_failed, tree_not_check_failed,
27534 tree_class_check_failed, tree_range_check_failed,
27535 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27536 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27537 tree_operand_check_failed, omp_clause_check_failed,
27538 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27541 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27543 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27544 flag_branch_probabilities.
27545 * ipa-inline.c (edge_badness): Likewise.
27546 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27547 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27548 * predict.c (maybe_hot_frequency_p): Likewise.
27549 (probably_never_executed): Likewise.
27550 * sched-ebb.c (schedule_ebbs): Likewise.
27551 * sched-rgn.c (find_single_block_region): Likewise.
27552 * tracer.c (tail_duplicate): Likewise.
27554 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27556 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27557 longer requires x_flag_profile_use.
27559 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27561 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27562 instead of flag_reorder_blocks_and_partition.
27563 * dbxout.c (dbxout_function_end): Likewise.
27564 * dwarf2out.c (gen_subprogram_die): Likewise.
27565 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27566 * hw-doloop.c (reorg_loops): Likewise.
27567 * varasm.c (assemble_start_function,
27568 assemble_end_function): Likewise.
27569 (decide_function_section): Do not check for
27570 flag_reorder_blocks_and_partition.
27572 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27574 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27576 (chkp_get_hard_register_fake_addr_expr): Ditto.
27577 (chkp_build_addr_expr): Add check for hard reg case.
27578 (chkp_parse_array_and_component_ref): Ditto.
27579 (chkp_find_bounds_1): Ditto.
27580 (chkp_process_stmt): Don't generate bounds store for
27583 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27585 * predict.c (maybe_hot_bb_p): Do not check profile status.
27586 (maybe_hot_edge_p): Likewise.
27587 (probably_never_executed): Check for zero counts even if profile
27589 (unlikely_executed_edge_p): New function.
27590 (unlikely_executed_stmt_p): New function.
27591 (unlikely_executed_bb_p): New function.
27592 (set_even_probabilities): Use unlikely predicates.
27593 (combine_predictions_for_bb): Likewise.
27594 (predict_paths_for_bb): Likewise.
27595 (predict_paths_leading_to_edge): Likewise.
27596 (determine_unlikely_bbs): New function.
27597 (estimate_bb_frequencies): Use it.
27598 (compute_function_frequency): Use zero counts even if profile is
27600 * profile-count.h: Fix typo.
27602 2017-08-08 Julia Koval <julia.koval@intel.com>
27604 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27605 _mm512_mask_cvtsepi16_storeu_epi8,
27606 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27607 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27608 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27609 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27610 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27611 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27612 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27613 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27614 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27615 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27616 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27617 __builtin_ia32_pmovuswb256mem_mask,
27618 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27619 __builtin_ia32_pmovwb512mem_mask): New builtins.
27621 2017-08-08 Julia Koval <julia.koval@intel.com>
27623 PR target/73350,80862
27624 * config/i386/subst.md (round): Fix round pattern.
27625 * config/i386/i386.c (ix86_erase_embedded_rounding):
27626 Fix erasing rounding for the fixed pattern.
27628 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27630 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27632 2017-06-08 Martin Liska <mliska@suse.cz>
27634 PR gcov-profile/80911
27635 * gcov.c (block_info::block_info): New constructor.
27637 2017-06-07 Carl Love <cel@us.ibm.com>
27639 * config/rs6000/rs6000-c: The return type of the following
27640 built-in functions was implemented as int not long long. Fix sign
27641 of return value for the unsigned version of vec_mulo and vec_mule.
27642 vector unsigned long long vec_bperm (vector unsigned long long,
27643 vector unsigned char)
27644 vector signed long long vec_mule (vector signed int,
27646 vector unsigned long long vec_mule (vector unsigned int,
27647 vector unsigned int)
27648 vector signed long long vec_mulo (vector signed int,
27650 vector unsigned long long vec_mulo (vector unsigned int,
27651 vector unsigned int)
27652 * doc/extend.texi: Fix the documentation for the built-in
27655 2017-06-07 Carl Love <cel@us.ibm.com>
27658 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27661 2017-06-07 Carl Love <cel@us.ibm.com>
27663 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27664 support, Generate doublehv for signed int/float for BE case only.
27666 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27668 * doc/invoke.texi (mcx16): Rewrite.
27670 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27672 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27673 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27674 *mov<mode>_softfloat, and an anonymous splitter): Use
27675 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27677 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27679 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27681 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27682 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27683 (rs6000_debug_reg_global): Adjust.
27684 (rs6000_init_hard_regno_mode_ok): Adjust.
27685 (rs6000_dbx_register_number): Adjust.
27686 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27687 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27688 Remove SPE_ACC and SPEFSCR.
27689 (REG_ALLOC_ORDER): Ditto.
27690 (FRAME_POINTER_REGNUM): Change to 111.
27691 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27692 (REG_CLASS_NAMES): Ditto.
27693 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27694 (REGISTER_NAMES): Ditto.
27695 (ADDITIONAL_REG_NAMES): Ditto.
27696 (rs6000_reg_names): Ditto.
27697 * config/rs6000/rs6000.md: Renumber some register number
27700 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27702 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27704 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27705 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27707 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27708 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27709 Delete the SPE high registers.
27710 (REG_ALLOC_ORDER): Ditto.
27711 (enum reg_class): Remove SPE_HIGH_REGS.
27712 (REG_CLASS_NAMES): Ditto.
27713 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27714 (REGISTER_NAMES): Ditto.
27715 (rs6000_reg_names): Ditto.
27716 * doc/tm.texi.in: Remove SPE as example.
27717 * doc/tm.texi: Regenerate.
27719 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27721 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27722 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27723 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27724 * config/rs6000/rs6000.md (type): Remove "brinc".
27726 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27728 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27729 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27730 * config/rs6000/linuxspe.h: Delete file.
27731 * config/rs6000/rs6000.md: Don't include spe.md.
27732 * config/rs6000/spe.h: Delete file.
27733 * config/rs6000/spe.md: Delete file.
27734 * config/rs6000/t-rs6000: Remove spe.md.
27736 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27738 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27739 (reg_or_none500mem_operand): Delete.
27740 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27741 instead of reg_or_none500mem_operand.
27743 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27745 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27746 handling of SPE flags.
27747 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27749 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27751 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27753 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27754 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27755 paired_divv2sf3): Similar.
27756 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27757 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27758 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27760 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27761 Rename the paired_* instruction patterns.
27762 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27764 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27765 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27766 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27767 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27768 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27769 PAIRED_VECTOR_MODE.
27770 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27771 (spe_func_has_64bit_regs_p): Delete.
27772 (spe_expand_predicate_builtin): Delete.
27773 (spe_expand_evsel_builtin): Delete.
27774 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27775 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27776 (invalid_e500_subreg): Delete.
27777 (rs6000_legitimize_address): Always force_reg op2 as well, for
27778 paired single memory accesses.
27779 (rs6000_member_type_forces_blk): Delete.
27780 (rs6000_spe_function_arg): Delete.
27781 (rs6000_expand_unop_builtin): Delete SPE handling.
27782 (rs6000_expand_binop_builtin): Ditto.
27783 (spe_expand_stv_builtin): Delete.
27784 (bdesc_2arg_spe): Delete.
27785 (spe_expand_builtin): Delete.
27786 (spe_expand_predicate_builtin): Delete.
27787 (spe_expand_evsel_builtin): Delete.
27788 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27789 (spe_init_builtins): Delete.
27790 (spe_func_has_64bit_regs_p): Delete.
27791 (savres_routine_name): Delete "info" parameter. Adjust callers.
27792 (rs6000_emit_stack_reset): Ditto.
27793 (rs6000_dwarf_register_span): Delete.
27794 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27795 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27796 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27798 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27800 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27801 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27802 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27803 mulv2sf3, divv2sf3): Delete expanders.
27805 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27807 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27809 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27811 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27812 * config/rs6000/rs6000.c: Ditto.
27814 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27816 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27817 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27818 comparison_operator.
27820 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27822 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27823 * config/rs6000/rs6000.opt: Ditto.
27824 * config/rs6000/t-rtems: Ditto.
27826 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27828 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27829 TARGET_E500_SINGLE by 0, simplify.
27830 * config/rs6000/rs6000.c: Ditto.
27831 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27832 (spe_build_register_parallel): Delete.
27833 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27834 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27835 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27836 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27837 (E500_CONVERT): Delete.
27838 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27840 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27842 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27843 * config/rs6000/dfp.md: Ditto.
27844 (negdd2, *negdd2_fpr): Merge.
27845 (absdd2, *absdd2_fpr): Merge.
27846 (negtd2, *negtd2_fpr): Merge.
27847 (abstd2, *abstd2_fpr): Merge.
27848 * config/rs6000/e500.h: Delete file.
27849 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27850 TARGET_FPRS by 1 and simplify.
27851 * config/rs6000/rs6000-c.c: Ditto.
27852 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27853 TARGET_DF_SPE by 0.
27854 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27856 * config/rs6000/rs6000.md: Ditto.
27857 (floatdidf2, *floatdidf2_fpr): Merge.
27858 (move_from_CR_gt_bit): Delete.
27859 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27860 (E500_CR_IOR_COMPARE): Delete.
27861 (All patterns that require !TARGET_FPRS): Delete.
27862 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27864 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27866 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27868 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27870 * graphds.c (add_edge): Intitialize edge's attached data.
27871 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27872 pointer parameter. Call pointed function on each edge during
27873 graph traversing. Skip traversing the edge when the function
27875 (graphds_dfs, graphds_scc): Ditto.
27876 (for_each_edge): New parameter. Pass the new parameter to callback
27878 * graphds.h (skip_edge_callback): New function pointer type.
27879 (graphds_dfs, graphds_scc): New function pointer parameter.
27880 (graphds_edge_callback, for_each_edge): New parameter.
27882 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27884 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27885 out code checking if runtime alias check is possible to below ...
27886 Call the new function.
27887 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27888 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27890 2017-06-07 Marek Polacek <polacek@redhat.com>
27893 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27894 TYPE_OVERFLOW_WRAPS check.
27896 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27898 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27899 if versioning is required.
27900 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27901 peeling with the check for versioning.
27903 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27905 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27906 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27907 Set true to new parameter if new ssa variable is defined.
27908 (vect_gen_vector_loop_niters): Refactor. Set range information
27909 for the new vector loop bound variable.
27910 (vect_do_peeling): Ditto.
27912 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27914 * tree-affine.c (ssa.h): Include header file.
27915 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27916 has wrapping overflow behavior.
27918 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27920 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27922 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27924 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27925 (tree_to_aff_combination): ... here.
27927 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27929 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27930 reg_pressure model function.
27931 (ivopts_global_cost_for_size): Delete.
27932 (determine_set_costs, iv_ca_recount_cost): Call new model function
27933 ivopts_estimate_reg_pressure.
27935 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27937 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27938 expensive than udiv. Remove floating point cases from mod.
27940 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27942 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27943 Increase idiv cost.
27945 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27947 * config/aarch64/aarch64.md
27948 (copysignsf3): Fix mask generation.
27950 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27952 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27954 (class dump_manager): Add register_dumps method.
27955 * dumpfile.c: Include langhooks.h.
27956 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27957 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27958 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27959 (dump_manager::dump_register): Start with 512 entries instead of 32.
27960 (dump_manager::register_dumps): New method.
27961 * toplev.c (general_init): Instead of invoking register_dumps
27962 langhook, invoke register_dumps method on the dump manager.
27963 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27966 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27968 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27969 Say that # requires an associated define_split to exist, and that
27970 the define_split must be suitable for use after register allocation.
27972 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27974 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27975 (compute_outgoing_frequencies): Also initialize zero counts.
27976 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27977 around loops; preserve more of profile when nothing changes.
27979 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27981 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27983 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27984 * config/arm/arm-cpu-cdata.h: Regenerate.
27985 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27986 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27987 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27988 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27989 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27991 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27993 * config/arm/t-rmprofile: Likewise.
27994 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27996 2017-06-06 David S. Miller <davem@davemloft.net>
27999 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28000 blockage if function uses alloca.
28002 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28004 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28005 New "uid" fields to hold pretty-print IDs of group and ref.
28006 Memory references are now identified as <group_id>:<ref_id>
28007 instead of using [random] addresses.
28008 (dump_mem_details): Simplify, no functional change.
28009 (dump_mem_ref): Simplify and make output more concise.
28010 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28011 (find_or_create_group): Initialize group uid.
28012 (record_ref): Initialize ref uid. Improve debug output.
28013 (prune_group_by_reuse, should_issue_prefetch_p,)
28014 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28015 (mark_nontemporal_store, determine_loop_nest_reuse):
28016 Improve debug output.
28018 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28020 * dbgcnt.def (prefetch): New debug counter.
28021 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28022 (schedule_prefetches): Stop issueing prefetches if debug counter
28025 2017-06-06 Tom de Vries <tom@codesourcery.com>
28027 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28028 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28030 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28032 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28033 Use aarch64_reg_or_zero predicate for operand 4.
28034 (aarch64_compare_and_swap<mode> define_insn_and_split):
28035 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28036 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28038 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28040 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28041 (arm_compute_save_core_reg_mask): This.
28042 (thumb1_compute_save_reg_mask): Rename into ...
28043 (thumb1_compute_save_core_reg_mask): This.
28044 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28045 (arm_compute_frame_layout): Likewise.
28047 2017-06-06 Richard Biener <rguenther@suse.de>
28049 PR tree-optimization/80974
28050 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28051 keep or clear leaders SSA info.
28053 2017-06-06 Tom de Vries <tom@codesourcery.com>
28055 * config/nvptx/nvptx.c (split_mode_p): New function.
28056 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28058 2017-06-06 Tom de Vries <tom@codesourcery.com>
28060 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28062 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28065 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28068 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28070 * shrink-wrap.c (handle_simple_exit): Update profile.
28071 (try_shrink_wrapping): Upate profile.
28073 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28075 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28076 (tree_guess_outgoing_edge_probabilities): New.
28077 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28078 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28080 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28082 * ipa-split.c (split_function): Initialize return bb profile.
28084 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28086 * profile.c (compute_branch_probabilities): Also initialize
28087 EXIT_BLOCK profile.
28089 2017-06-06 Richard Biener <rguenther@suse.de>
28091 PR tree-optimization/80928
28092 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28093 (vect_analyze_loop_operations): Properly guard analysis for
28095 (vect_transform_loop): Likewise.
28096 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28097 (vect_model_induction_cost): Do not cost for pure SLP.
28098 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28099 of induction in inner loop vectorization.
28100 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28101 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28102 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28104 (vect_analyze_slp_cost_1): Cost induction.
28105 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28106 (vect_get_slp_vect_defs): Likewise.
28107 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28108 (vect_transform_stmt): Handle SLP reductions.
28109 * tree-vectorizer.h (vectorizable_induction): Adjust.
28111 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28113 * config/rs6000/rs6000.c (make_resolver_func): Update
28114 init_lowered_empty_function call.
28116 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28118 * doc/invoke.texi: Document the -fprofile-abs-path option.
28119 * common.opt (fprofile-abs-path): New option.
28120 * gcov-io.h (gcov_write_filename): Declare.
28121 * gcov-io.c (gcov_write_filename): New function.
28122 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28123 * profile.c (output_location): Likewise.
28125 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28127 * shring-wrap.c: Revert accidental commit.
28129 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28131 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28133 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28135 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28137 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28138 profile in callgraph edge.
28139 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28140 (apply_scale): Likewise.
28141 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28142 Also scale profile when inlining function with zero profile.
28143 (initialize_cfun): Update exit block profile even when it is zero.
28144 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28145 when profile is read.
28147 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28149 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28150 (CLONE_*): New constants to define the processors we can generate
28151 code for with the target_clone attribute.
28152 (rs6000_clone_map): New array to identify which clone processors
28153 the current program is running on.
28154 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28155 target_clone attribute.
28156 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28157 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28158 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28159 (cpu_expand_builtin): Add support for target_clone attribute.
28160 (rs6000_valid_attribute_p): Allow "default" attribute.
28161 (get_decl_name): New debug function to simplify printing the
28162 current function name in debugging statements.
28163 (rs6000_clone_priority): New functions to support the target_clone
28164 attribute, and be able to generate code to switch between ISA 2.05
28165 through ISA 3.0 (power6 through power9).
28166 (rs6000_compare_version_priority): Likewise.
28167 (rs6000_get_function_versions_dispatcher): Likewise.
28168 (make_resolver_func): Likewise.
28169 (add_condition_to_bb): Likewise.
28170 (dispatch_function_versions): Likewise.
28171 (rs6000_generate_version_dispatcher_body): Likewise.
28172 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28173 (fusion_gpr_load_p): Fix a spacing issue.
28174 * doc/extend.texi (Common Function Attributes): Document that the
28175 PowerPC supports the target_clone attribute.
28177 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28179 * config/arm/arm.h: explain F symbol found in description of ARM
28180 register allocation in its legend.
28182 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28184 * config/mips/frame-header-opt.c: Include profile-count.h.
28185 * config/riscv/riscv.c: Include profile-count.h
28187 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28189 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28191 (sm_set_flag_if_changed): Add bbs field.
28192 (execute_sm_if_changed_flag_set): Pass BBS.
28193 (execute_sm): Update.
28195 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28197 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28200 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28202 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28203 (peephole2): New peephole2 to emit the above.
28204 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28206 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28208 * config/aarch64/aarch64.c (define_peephole2 above
28209 *sub_<shift>_<mode>): New peephole.
28211 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28213 * config/i386/i386.c (make_resolver_func): Update.
28214 * Makefile.in: Add profile-count.h and profile-count.o
28215 * auto-profile.c (afdo_indirect_call): Update to new API.
28216 (afdo_set_bb_count): Update.
28217 (afdo_propagate_edge): Update.
28218 (afdo_propagate_circuit): Update.
28219 (afdo_calculate_branch_prob): Update.
28220 (afdo_annotate_cfg): Update.
28221 * basic-block.h: Include profile-count.h
28222 (struct edge_def): Turn count to profile_count.
28223 (struct basic_block_def): Likewie.
28224 (REG_BR_PROB_BASE): Move to profile-count.h
28225 (RDIV): Move to profile-count.h
28226 * bb-reorder.c (max_entry_count): Turn to profile_count.
28227 (find_traces): Update.
28228 (rotate_loop):Update.
28229 (connect_traces):Update.
28230 (sanitize_hot_paths):Update.
28231 * bt-load.c (migrate_btr_defs): Update.
28232 * cfg.c (RDIV): Remove.
28233 (init_flow): Use alloc_block.
28234 (alloc_block): Uninitialize count.
28235 (unchecked_make_edge): Uninitialize count.
28236 (check_bb_profile): Update.
28237 (dump_edge_info): Update.
28238 (dump_bb_info): Update.
28239 (update_bb_profile_for_threading): Update.
28240 (scale_bbs_frequencies_int): Update.
28241 (scale_bbs_frequencies_gcov_type): Update.
28242 (scale_bbs_frequencies_profile_count): New.
28243 * cfg.h (update_bb_profile_for_threading): Update.
28244 (scale_bbs_frequencies_profile_count): Declare.
28245 * cfgbuild.c (compute_outgoing_frequencies): Update.
28246 (find_many_sub_basic_blocks): Update.
28247 * cfgcleanup.c (try_forward_edges): Update.
28248 (try_crossjump_to_edge): Update.
28249 * cfgexpand.c (expand_gimple_tailcall): Update.
28250 (construct_exit_block): Update.
28251 * cfghooks.c (verify_flow_info): Update.
28252 (dump_bb_for_graph): Update.
28253 (split_edge): Update.
28254 (make_forwarder_block): Update.
28255 (duplicate_block): Update.
28256 (account_profile_record): Update.
28257 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28258 (get_estimated_loop_iterations): Update.
28259 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28260 (single_likely_exit): Update.
28261 * cfgloopmanip.c (scale_loop_profile): Update.
28263 (set_zero_probability): Update.
28264 (lv_adjust_loop_entry_edge): Update.
28265 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28266 (purge_dead_edges): Update.
28267 (rtl_account_profile_record): Update.
28268 * cgraph.c (cgraph_node::create): Uninitialize count.
28269 (symbol_table::create_edge): Uninitialize count.
28270 (cgraph_update_edges_for_call_stmt_node): Update.
28271 (cgraph_edge::dump_edge_flags): Update.
28272 (cgraph_node::dump): Update.
28273 (cgraph_edge::maybe_hot_p): Update.
28274 * cgraph.h: Include profile-count.h
28275 (create_clone), create_edge, create_indirect_edge): Update.
28276 (cgraph_node): Turn count to profile_count.
28277 (cgraph_edge0: Likewise.
28278 (make_speculative, clone): Update.
28279 (create_edge): Update.
28280 (init_lowered_empty_function): Update.
28281 * cgraphclones.c (cgraph_edge::clone): Update.
28282 (duplicate_thunk_for_node): Update.
28283 (cgraph_node::create_clone): Update.
28284 * cgraphunit.c (cgraph_node::analyze): Update.
28285 (cgraph_node::expand_thunk): Update.
28286 * final.c (dump_basic_block_info): Update.
28287 * gimple-streamer-in.c (input_bb): Update.
28288 * gimple-streamer-out.c (output_bb): Update.
28289 * graphite.c (print_global_statistics): Update.
28290 (print_graphite_scop_statistics): Update.
28291 * hsa-brig.c: Include basic-block.h.
28292 * hsa-dump.c: Include basic-block.h.
28293 * hsa-gen.c (T sum_slice): Update.
28294 (convert_switch_statements):Update.
28295 * hsa-regalloc.c: Include basic-block.h.
28296 * ipa-chkp.c (chkp_produce_thunks): Update.
28297 * ipa-cp.c (struct caller_statistics): Update.
28298 (init_caller_stats): Update.
28299 (gather_caller_stats): Update.
28300 (ipcp_cloning_candidate_p): Update.
28301 (good_cloning_opportunity_p): Update.
28302 (get_info_about_necessary_edges): Update.
28303 (dump_profile_updates): Update.
28304 (update_profiling_info): Update.
28305 (update_specialized_profile): Update.
28306 (perhaps_add_new_callers): Update.
28307 (decide_about_value): Update.
28308 (ipa_cp_c_finalize): Update.
28309 * ipa-devirt.c (struct odr_type_warn_count): Update.
28310 (struct decl_warn_count): Update.
28311 (struct final_warning_record): Update.
28312 (possible_polymorphic_call_targets): Update.
28313 (ipa_devirt): Update.
28314 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28315 * ipa-icf.c (sem_function::merge): Update.
28316 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28317 * ipa-inline.c (compute_uninlined_call_time): Update.
28318 (compute_inlined_call_time): Update.
28319 (want_inline_small_function_p): Update.
28320 (want_inline_self_recursive_call_p): Update.
28321 (edge_badness): Update.
28322 (lookup_recursive_calls): Update.
28323 (recursive_inlining): Update.
28324 (inline_small_functions): Update.
28325 (dump_overall_stats): Update.
28326 (dump_inline_stats): Update.
28327 * ipa-profile.c (ipa_profile_generate_summary): Update.
28328 (ipa_propagate_frequency): Update.
28329 (ipa_profile): Update.
28330 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28331 * ipa-utils.c (ipa_merge_profiles): Update.
28332 * loop-doloop.c (doloop_modify): Update.
28333 * loop-unroll.c (report_unroll): Update.
28334 (unroll_loop_runtime_iterations): Update.
28335 * lto-cgraph.c (lto_output_edge): Update.
28336 (lto_output_node): Update.
28337 (input_node): Update.
28338 (input_edge): Update.
28339 (merge_profile_summaries): Update.
28340 * lto-streamer-in.c (input_cfg): Update.
28341 * lto-streamer-out.c (output_cfg): Update.
28342 * mcf.c (create_fixup_graph): Update.
28343 (adjust_cfg_counts): Update.
28344 (sum_edge_counts): Update.
28345 * modulo-sched.c (sms_schedule): Update.
28346 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28347 * predict.c (maybe_hot_count_p): Update.
28348 (probably_never_executed): Update.
28349 (dump_prediction): Update.
28350 (combine_predictions_for_bb): Update.
28351 (propagate_freq): Update.
28352 (handle_missing_profiles): Update.
28353 (counts_to_freqs): Update.
28354 (rebuild_frequencies): Update.
28355 (force_edge_cold): Update.
28356 * predict.h: Include profile-count.h
28357 (maybe_hot_count_p, counts_to_freqs): UPdate.
28358 * print-rtl-function.c: Do not include cfg.h
28359 * print-rtl.c: Include basic-block.h
28360 * profile-count.c: New file.
28361 * profile-count.h: New file.
28362 * profile.c (is_edge_inconsistent): Update.
28363 (correct_negative_edge_counts): Update.
28364 (is_inconsistent): Update.
28365 (set_bb_counts): Update.
28366 (read_profile_edge_counts): Update.
28367 (compute_frequency_overlap): Update.
28368 (compute_branch_probabilities): Update; Initialize and deinitialize
28370 (branch_prob): Update.
28371 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28372 (edge_gcov_count): New.
28373 (bb_gcov_count): New.
28374 * shrink-wrap.c (try_shrink_wrapping): Update.
28375 * tracer.c (better_p): Update.
28376 * trans-mem.c (expand_transaction): Update.
28377 (ipa_tm_insert_irr_call): Update.
28378 (ipa_tm_insert_gettmclone_call): Update.
28379 * tree-call-cdce.c: Update.
28380 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28381 (gimple_duplicate_sese_tail): Update.
28382 (gimple_account_profile_record): Update.
28383 (execute_fixup_cfg): Update.
28384 * tree-inline.c (copy_bb): Update.
28385 (copy_edges_for_bb): Update.
28386 (initialize_cfun): Update.
28387 (freqs_to_counts): Update.
28388 (copy_cfg_body): Update.
28389 (expand_call_inline): Update.
28390 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28391 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28392 (try_unroll_loop_completely): Update.
28393 (try_peel_loop): Update.
28394 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28395 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28396 * tree-ssa-loop-split.c (connect_loops): Update.
28397 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28398 * tree-ssa-reassoc.c (branch_fixup): Update.
28399 * tree-ssa-tail-merge.c (replace_block_by): Update.
28400 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28401 (compute_path_counts): Update.
28402 (update_profile): Update.
28403 (recompute_probabilities): Update.
28404 (update_joiner_offpath_counts): Update.
28405 (estimated_freqs_path): Update.
28406 (freqs_to_counts_path): Update.
28407 (clear_counts_path): Update.
28408 (ssa_fix_duplicate_block_edges): Update.
28409 (duplicate_thread_path): Update.
28410 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28411 (struct switch_conv_info): Update.
28412 * tree-tailcall.c (decrease_profile): Update.
28413 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28414 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28415 * value-prof.c (check_counter): Update.
28416 (gimple_divmod_fixed_value): Update.
28417 (gimple_mod_pow2): Update.
28418 (gimple_mod_subtract): Update.
28419 (gimple_ic_transform): Update.
28420 (gimple_stringop_fixed_value): Update.
28421 * value-prof.h (gimple_ic): Update.
28423 2017-06-02 Carl Love <cel@us.ibm.com>
28425 * config/rs6000/rs6000-c: Add support for built-in functions
28426 vector double vec_doublee (vector signed int);
28427 vector double vec_doublee (vector unsigned int);
28428 vector double vec_doublee (vector float);
28429 vector double vec_doubleh (vector signed int);
28430 vector double vec_doubleh (vector unsigned int);
28431 vector double vec_doubleh (vector float);
28432 vector double vec_doublel (vector signed int);
28433 vector double vec_doublel (vector unsigned int);
28434 vector double vec_doublel (vector float);
28435 vector double vec_doubleo (vector signed int);
28436 vector double vec_doubleo (vector unsigned int);
28437 vector double vec_doubleo (vector float);.
28438 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28439 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28441 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28442 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28443 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28445 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28446 vec_doublel, vec_doubleh.
28447 * doc/extend.texi: Update the built-in documentation file for the
28448 new built-in functions.
28450 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28453 * ipa-inline-analysis.c (free_growth_caches): Set
28454 edge_removal_hook_holder to NULL after removing it.
28456 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28458 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28459 comparision with zero.
28461 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28462 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28463 for early expansion of vec_min and vec_max builtins.
28464 (builtin_function_type): Add min/max unsigned variants to those
28465 identified as having unsigned arguments.
28467 2017-06-02 Olivier Hainque <hainque@adacore.com>
28469 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28471 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28473 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28474 Use VALL_F16 iterator rather than VALL.
28476 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28478 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28479 Emit CBNZ inside loop when doing a strong exchange and comparing
28480 against zero. Generate the CC flags after the loop.
28482 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28484 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28485 (dl_section_ref): New.
28486 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28487 On AIX, append an expression to subtract the size of the
28488 section length to dl_section_ref.
28490 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28492 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28493 for early expansion of vector absolute builtins.
28495 2017-06-02 Richard Biener <rguenther@suse.de>
28497 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28498 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28500 2017-06-02 Richard Biener <rguenther@suse.de>
28502 PR tree-optimization/80948
28503 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28504 stmt order as well.
28506 2017-06-02 Richard Biener <rguenther@suse.de>
28508 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28510 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28511 for inductions as relevant.
28513 2017-06-02 Richard Biener <rguenther@suse.de>
28515 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28516 (vectorizable_induction): ... this. Remove dead code.
28518 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28520 * builtins. (expand_builtin_alloca): Remove second parameter and
28521 infer its value from the first parameter instead.
28522 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28524 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28526 PR rtl-optimization/80903
28527 * loop-doloop.c (add_test): Unshare sequence.
28529 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28531 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28533 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28535 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28537 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28538 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28539 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28540 (xlogue_layout::get_stub_rtx): Make static.
28541 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28542 (xlogue_layout::compute_stub_managed_regs): Rename to...
28543 (xlogue_layout::count_stub_managed_regs): ...this.
28544 (xlogue_layout::is_stub_managed_reg): New function.
28545 (xlogue_layout::m_stub_names): Rename to...
28546 (xlogue_layout::s_stub_names): ...this, make static.
28547 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28548 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28549 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28550 xlogue_layout::s_stub_names): Instantiate statics.
28551 (stub_managed_regs): Remove.
28552 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28553 (disable_call_ms2sysv_xlogues): Rename to...
28554 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28555 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28557 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28558 change after reload_completed.
28559 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28561 (ix86_expand_prologue): Likewise.
28562 (ix86_expand_epilogue): Likewise.
28563 (ix86_expand_split_stack_prologue): Likewise.
28564 (ix86_compute_frame_layout): Remove frame parameter ...
28565 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28566 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28568 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28569 (ix86_frame): Move from here ...
28570 * config/i386/i386.h (ix86_frame): ... to here.
28571 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28572 complete ix86_frame data structure instead. Remove some_ld_name.
28574 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28576 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28577 symbols that hold a DECL_VALUE_EXPR.
28579 2017-06-01 Martin Jambor <mjambor@suse.cz>
28581 PR tree-optimization/80898
28582 * tree-sra.c (process_subtree_disqualification): Removed.
28583 (disqualify_candidate): Do not acll
28584 process_subtree_disqualification.
28585 (subtree_mark_written_and_enqueue): New function.
28586 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28587 RHS has been disqualified and re-queue LHS if necessary. Apart
28588 from that, ignore disqualified RHS.
28590 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28592 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28593 address fetch for z10 or later.
28595 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28597 * config/arc/arc.md (tst_movb): Add guard when splitting.
28599 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28601 * config/arc/arc.c (arc_can_eliminate): Test against
28602 arc_frame_pointer_needed.
28604 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28606 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28607 to prevent store reordering.
28608 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28609 (type): Add block type.
28610 (stack_tie): Define special instruction to be used in
28613 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28615 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28616 constraint. It is not valid for the pattern.
28617 (noncommutative_binary_comparison): Likewise.
28619 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28621 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28624 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28626 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28627 be used by the reg-alloc.
28629 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28631 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28632 reg-alloc when having mul64 or mul32x16 instructions.
28633 (mulsidi3): Likewise.
28634 (umulsidi3): Likewise.
28635 (mulsi32x16): New pattern.
28636 (mulsi64): Likewise.
28637 (mulsidi64): Likewise.
28638 (umulsidi64): Likewise.
28639 (MUL32x16_REG): Define.
28640 (mul64_600): Use MUL32x16_REG.
28641 (mac64_600): Likewise.
28642 (umul64_600): Likewise.
28643 (umac64_600): Likewise.
28645 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28647 * config/arc/arc.md (mulsi3_700): Make it commutative.
28649 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28651 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28653 (*sign_extendsidi2_insn): Likewise for movstosw.
28655 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28657 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28658 the type of the input discriminant value. Convert the
28659 discriminant value of signedness vary.
28661 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28663 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28664 Add to -Wall section.
28666 2017-06-01 Richard Biener <rguenther@suse.de>
28668 PR middle-end/66313
28669 * fold-const.c (fold_plusminus_mult_expr): If the factored
28670 factor may be zero use a wrapping type for the inner operation.
28671 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28672 and handle moved defs.
28673 (process_assignment): Properly guard the unary op case. Return a
28674 tri-state indicating that moving the stmt before the call may allow
28675 to continue. Pass through to_move.
28676 (find_tail_calls): Handle moving unrelated defs before
28679 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28682 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28683 splitter result in the canonical way.
28685 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28687 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28688 also for 32bit target. Update insn attributes.
28689 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28691 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28693 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28694 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28695 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28696 (_mm_maskz_min_ss): New intrinsics.
28698 2017-05-31 Martin Liska <mliska@suse.cz>
28700 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28701 Change comment style to one we normally use.
28702 (vectorizable_reduction): Likewise.
28703 (vectorizable_induction): Likewise.
28704 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28705 (vectorizable_call): Likewise.
28706 (vectorizable_simd_clone_call): Likewise.
28707 (vectorizable_conversion): Likewise.
28708 (vectorizable_assignment): Likewise.
28709 (vectorizable_shift): Likewise.
28710 (vectorizable_operation): Likewise.
28711 (vectorizable_store): Likewise.
28712 (vectorizable_load): Likewise.
28713 * tree-vectorizer.h: Likewise.
28715 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28717 * passes.c (emergency_dump_function): New.
28718 * tree-pass.h (emergency_dump_function): Declare.
28719 * plugin.c (plugins_internal_error_function): Remove.
28720 * plugin.h (plugins_internal_error_function): Remove declaration.
28721 * toplev.c (internal_error_function): New static function. Use it...
28722 (general_init): ...here.
28724 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28726 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28727 (arc_rtx_costs): Add costs for new patterns.
28728 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28729 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28731 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28733 * tree-ssa-strlen.c (get_next_strinfo): New function.
28734 (get_stridx_plus_constant): Use it.
28735 (zero_length_string): Likewise.
28736 (adjust_related_strinfos): Likewise.
28737 (adjust_last_stmt): Likewise.
28739 2017-05-31 Richard Biener <rguenther@suse.de>
28742 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28743 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28745 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28747 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28748 loop_vinfo argument and use of dependence distance vectors.
28749 Check instead whether the two references differ only in their
28750 initial value and assume that they have the same alignment if the
28751 difference is a multiple of the vector alignment.
28752 (vect_analyze_data_refs_alignment): Update call accordingly.
28754 2017-05-31 Martin Liska <mliska@suse.cz>
28757 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28759 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28761 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28762 (create_intersect_range_checks): Move from ...
28763 * tree-data-ref.c (create_intersect_range_checks_index)
28764 (create_intersect_range_checks): ... to here.
28765 (create_runtime_alias_checks): New function factored from ...
28766 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28767 here. Call above function.
28768 * tree-data-ref.h (create_runtime_alias_checks): New function.
28770 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28772 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28773 segment length for dr_b and compute it in wide_int.
28775 2017-05-31 Richard Biener <rguenther@suse.de>
28777 PR tree-optimization/80906
28778 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28779 and pass through iv_map.
28780 (copy_bb_and_scalar_dependences): Adjust.
28781 (translate_pending_phi_nodes): Likewise.
28782 (copy_loop_close_phi_args): Handle code-generating IVs instead
28785 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28787 * diagnostic-color.c (color_dict): Add "type-diff".
28788 (parse_gcc_colors): Update comment.
28789 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28790 -fdiagnostics-show-template-tree and -fno-elide-type.
28791 (GCC_COLORS): Add type-diff to example.
28793 (-fdiagnostics-show-template-tree): New.
28794 (-fno-elide-type): New.
28795 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28796 the pp_format_decoder callback. Call any m_format_postprocessor's
28798 (pretty_printer::pretty_printer): Initialize
28799 m_format_postprocessor.
28800 (pretty_printer::~pretty_printer): Delete any
28801 m_format_postprocessor.
28802 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28803 (class format_postprocessor): New class.
28804 (struct pretty_printer::format_decoder): Document the new parameters.
28805 (struct pretty_printer::m_format_postprocessor): New field.
28806 * tree-diagnostic.c (default_tree_printer): Update for new
28807 bool and const char ** params.
28808 * tree-diagnostic.h (default_tree_printer): Likewise.
28810 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28812 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28813 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28814 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28815 rs6000_gen_cell_microcode code.
28816 (rs6000_final_prescan_insn): Delete.
28817 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28818 "warn-cell-microcode" entries.
28819 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28820 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28821 throughout. Change cc_reg_not_micro_cr0_operand to
28822 cc_reg_not_cr0_operand throughout.
28823 (*extendhi<mode>2_noload): Delete.
28824 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28825 (mwarn-cell-microcode): Delete.
28826 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28827 -mgen-cell-microcode and -mwarn-cell-microcode.
28829 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28832 * config/i386/constraints.md (Yd): New constraint.
28834 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28835 and (?Yd, r) alternatives. Update insn attributes.
28836 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28837 and (?*Yd, r) alternatives. Update insn attributes.
28838 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28840 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28842 * gimplify.c (gimplify_modify_expr): Don't create a
28843 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28846 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28848 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28850 2017-05-30 Richard Biener <rguenther@suse.de>
28852 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28853 and reduc_def fields.
28854 (STMT_VINFO_REDUC_TYPE): New define.
28855 (STMT_VINFO_REDUC_DEF): Likewise.
28856 (vect_force_simple_reduction): Adjust prototype.
28857 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28858 (vect_is_simple_reduction): Remove check_reduction argument.
28859 (vect_force_simple_reduction): Adjust and set
28860 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28861 (vectorizable_reduction): Do not re-do reduction analysis
28862 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28863 * tree-parloops.c (gather_scalar_reductions): Adjust.
28865 2017-05-30 Richard Biener <rguenther@suse.de>
28867 PR middle-end/80901
28868 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28871 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28873 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28874 Introduce unknown_misalignment parameter and remove vf.
28875 (vect_peeling_hash_get_lowest_cost):
28876 Pass unknown_misalignment parameter.
28877 (vect_enhance_data_refs_alignment):
28878 Fix unsupportable data ref treatment.
28880 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28882 * tree-vect-data-refs.c (vect_get_data_access_cost):
28883 Workaround for SLP handling.
28884 (vect_enhance_data_refs_alignment):
28885 Compute costs for doing no peeling at all, compare to the best
28886 peeling costs so far and avoid peeling if cheaper.
28888 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28890 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28891 Return peeling info and set costs to zero for unlimited cost
28893 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28894 with unknown misalignment. Compute and costs for unknown
28895 misalignment, compare them to the costs for known misalignment
28896 and choose the cheapest for peeling.
28898 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28900 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28901 (vect_get_peeling_costs_all_drs): Create function.
28902 (vect_peeling_hash_get_lowest_cost):
28903 Use vect_get_peeling_costs_all_drs.
28904 (vect_peeling_supportable): Create function.
28905 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28907 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28909 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28910 DR_HAS_NEGATIVE_STEP.
28911 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28912 (vect_enhance_data_refs_alignment): Use.
28913 (vect_duplicate_ssa_name_ptr_info): Use.
28914 * tree-vectorizer.h (dr_misalignment): Use.
28915 (known_alignment_for_access_p): Use.
28917 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28920 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28922 (has_section_name): New function.
28924 2017-05-30 Martin Liska <mliska@suse.cz>
28927 * auto-profile.c (get_function_decl_from_block): Fix
28930 2017-05-30 Richard Biener <rguenther@suse.de>
28932 PR middle-end/80876
28933 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28935 2017-05-30 Martin Liska <mliska@suse.cz>
28937 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28938 * dumpfile.h (struct dump_file_info): Remove ctors.
28940 2017-05-30 Martin Liska <mliska@suse.cz>
28942 * predict.def: Fix GNU coding style.
28944 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28946 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28947 Mark 'to' argument with ATTRIBUTE_UNUSED.
28949 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28951 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28952 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28953 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28956 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28958 * doc/install.texi (Options specification): Restore entry of
28959 --enable-sjlj-exceptions.
28961 2017-05-27 Michael Eager <eager@eagercon.com>
28964 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28966 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28968 * config/microblaze/microblaze.h
28969 (FIXED_REGISTERS): Update in macro.
28970 (CALL_USED_REGISTERS): Update in macro.
28972 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28974 * doc/install.texi: Add links to macOS binary distributions.
28976 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28980 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28982 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28984 2017-05-26 Martin Liska <mliska@suse.cz>
28986 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28988 2017-05-26 Martin Liska <mliska@suse.cz>
28990 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28991 always leading ';; '.
28992 (dump_bb_info): Likewise.
28993 (brief_dump_cfg): Likewise.
28994 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28995 * dumpfile.c: Remove usage of TDF_VERBOSE.
28996 * dumpfile.h (enum dump_kind): Likewise.
28997 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28998 * print-tree.c (debug_verbose): Remove.
28999 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29000 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29001 * tree-diagnostic.c (default_tree_printer): Replace
29002 TDF_DIAGNOSTIC with TDF_SLIM.
29004 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29006 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29007 in parameter loop, rather than loop_vinfo.
29008 (create_intersect_range_checks): Ditto.
29009 (vect_create_cond_for_alias_checks): Update call to above functions.
29011 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29013 PR tree-optimization/80815
29014 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29015 for merging runtime alias checks. Handle negative DR_STEPs.
29017 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29019 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29021 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29022 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29023 out code pruning runtime alias checks.
29024 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29025 factored out from above.
29026 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29028 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29030 (prune_runtime_alias_test_list): New decalaration.
29032 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29034 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29035 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29036 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29037 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29038 (operator==, comp_dr_with_seg_len_pair): Ditto.
29039 (vect_prune_runtime_alias_test_list): Ditto.
29041 2017-05-26 Martin Liska <mliska@suse.cz>
29044 * params.def: Bound partial-inlining-entry-probability param.
29046 2017-05-26 Marek Polacek <polacek@redhat.com>
29049 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29052 2017-05-26 Richard Biener <rguenther@suse.de>
29054 PR tree-optimization/80842
29055 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29058 2017-05-26 Richard Biener <rguenther@suse.de>
29060 PR tree-optimization/80844
29061 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29063 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29065 * doc/md.texi (Machine Constraints): Update x86 family
29066 machine constraints section to match 'config/i386/constraints.md'.
29068 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29070 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29072 2017-05-25 Nathan Sidwell <nathan@acm.org>
29074 * doc/invoke.texi (--enable-languages): Update documentation.
29076 2017-05-25 Martin Liska <mliska@suse.cz>
29078 * dumpfile.c: Add TDF_FOLDING.
29079 * dumpfile.h (enum dump_kind): Likewise.
29080 * genmatch.c (dt_simplify::gen_1): Use it.
29082 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29084 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29086 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29088 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29089 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29091 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29093 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29094 * match.pd (X == C): Rewrite it here.
29095 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29096 with_certain_nonzero_bits2): New predicates.
29097 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29099 2017-05-24 Nathan Sidwell <nathan@acm.org>
29101 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29104 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29107 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29109 * config/powerpcspe: New port. Files are copied from the rs6000
29110 port, with "rs6000" in filenames replaced by "powerpcspe".
29112 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29114 PR rtl-optimization/80754
29115 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29117 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29119 * config/sparc/sparc.md (length): Return the correct value for -mflat
29120 sibcalls to match output_sibcall.
29122 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29126 * config/rs6000/rs6000.c (struct machine_function): Add new field
29128 (rs6000_get_separate_components): Init that field, use it.
29129 (rs6000_components_for_bb): Use the field.
29131 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29133 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29135 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29137 PR middle-end/80823
29138 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29140 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29143 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29144 address against address_operand predicate.
29145 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29147 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29149 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29150 parameters passed indirectly.
29152 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29154 * config/i386/i386.md (*movdi_internal): Remove SSE4
29155 alternative 18 (?r, *v). Update insn attributes.
29156 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29157 Update insn attributes.
29158 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29159 Update insn attributes.
29160 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29161 alternative 1 (r, v). Remove isa attribute.
29162 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29163 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29164 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29166 2017-05-23 Tom de Vries <tom@codesourcery.com>
29168 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29171 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29173 * cgraphunit.c (symbol_table::process_new_functions): Update.
29174 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29175 (inline_generate_summary): Rename to ...
29176 (ipa_fn_summary_generate): ... this one.
29177 (inline_read_summary): Rename to ...
29178 (ipa_fn_summary_read): ... this one.
29179 (inline_write_summary): Rename to ...
29180 (ipa_fn_summary_write): ... this one.
29181 (inline_free_summary): Rename to ...
29182 (ipa_free_fn_summary): ... this one.
29183 (pass_data_local_fn_summary, pass_local_fn_summary,
29184 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29185 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29186 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29187 make_pass_ipa_fn_summary): New.
29188 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29189 inline_write_summary, inline_free_summary): Remove.
29190 (ipa_free_fn_summary) : New.
29191 * ipa-inline.c (ipa_inline): Update.
29192 (pass_ipa_inline): Do not generate summaries.
29193 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29195 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29196 and add pass_ipa_fn_summary.
29197 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29199 (make_pass_inline_parameters): Remove.
29201 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29203 * omp-low.c (struct omp_context): Remove "default_kind" member.
29206 * omp-offload.c (execute_oacc_device_lower): Remove the
29207 parallelism dimensions function attributes for unparallelized
29208 OpenACC kernels constructs.
29210 2017-05-23 Martin Liska <mliska@suse.cz>
29212 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29214 (cgraph_edge::make_speculative): Likewise.
29215 (cgraph_edge::resolve_speculation): Likewise.
29216 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29217 (cgraph_node::dump): Likewise.
29218 * cgraph.h: Likewise.
29219 * cgraphunit.c (analyze_functions): Likewise.
29220 (symbol_table::compile): Likewise.
29221 * ipa-cp.c (print_all_lattices): Likewise.
29222 (determine_versionability): Likewise.
29223 (initialize_node_lattices): Likewise.
29224 (ipcp_verify_propagated_values): Likewise.
29225 (estimate_local_effects): Likewise.
29226 (update_profiling_info): Likewise.
29227 (create_specialized_node): Likewise.
29228 (perhaps_add_new_callers): Likewise.
29229 (decide_about_value): Likewise.
29230 (decide_whether_version_node): Likewise.
29231 (identify_dead_nodes): Likewise.
29232 (ipcp_store_bits_results): Likewise.
29233 * ipa-devirt.c (dump_targets): Likewise.
29234 (ipa_devirt): Likewise.
29235 * ipa-icf.c (sem_item::dump): Likewise.
29236 (sem_function::equals): Likewise.
29237 (sem_variable::equals): Likewise.
29238 (sem_item_optimizer::read_section): Likewise.
29239 (sem_item_optimizer::execute): Likewise.
29240 (congruence_class::dump): Likewise.
29241 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29242 (dump_inline_summary): Likewise.
29243 (estimate_node_size_and_time): Likewise.
29244 (inline_analyze_function): Likewise.
29245 * ipa-inline-transform.c (inline_call): Likewise.
29246 * ipa-inline.c (report_inline_failed_reason): Likewise.
29247 (want_early_inline_function_p): Likewise.
29248 (edge_badness): Likewise.
29249 (update_edge_key): Likewise.
29250 (inline_small_functions): Likewise.
29251 * ipa-profile.c (ipa_profile): Likewise.
29252 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29253 (ipa_make_edge_direct_to_target): Likewise.
29254 (remove_described_reference): Likewise.
29255 (ipa_impossible_devirt_target): Likewise.
29256 (propagate_controlled_uses): Likewise.
29257 (ipa_print_node_params): Likewise.
29258 (ipcp_transform_function): Likewise.
29259 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29260 (propagate_pure_const): Likewise.
29261 * ipa-reference.c (generate_summary): Likewise.
29262 (read_write_all_from_decl): Likewise.
29263 (propagate): Likewise.
29264 (ipa_reference_read_optimization_summary): Likewise.
29265 * ipa-utils.c (ipa_merge_profiles): Likewise.
29266 * ipa.c (walk_polymorphic_call_targets): Likewise.
29267 (symbol_table::remove_unreachable_nodes): Likewise.
29268 (ipa_single_use): Likewise.
29269 * passes.c (execute_todo): Likewise.
29270 * predict.c (drop_profile): Likewise.
29271 * symtab.c (symtab_node::get_dump_name): New function.
29272 (symtab_node::dump_name): Likewise.
29273 (symtab_node::dump_asm_name): Likewise.
29274 (symtab_node::dump_references): Likewise.
29275 (symtab_node::dump_referring): Likewise.
29276 (symtab_node::dump_base): Likewise.
29277 (symtab_node::debug_symtab): Likewise.
29278 * tree-sra.c (convert_callers_for_node): Likewise.
29279 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29280 * value-prof.c (init_node_map): Likewise.
29282 2017-05-23 Martin Liska <mliska@suse.cz>
29284 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29285 and symtab_node::debug_symtab to symbol_table::debug.
29286 * cgraphunit.c (analyze_functions): Use the renamed function.
29287 (symbol_table::compile): Likewise.
29288 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29289 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29290 * passes.c (execute_todo): Likewise.
29291 * symtab.c (symbol_table::dump): New function.
29292 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29294 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29296 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29297 that nonconst implies exec.
29299 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29301 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29302 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29303 (inline_edge_summary_vec): Turn into ...
29304 (ipa_call_summaries): ... this one.
29305 (redirect_to_unreachable, edge_set_predicate,
29306 evaluate_properties_for_edge, inline_summary_alloc,
29307 reset_ipa_call_summary, reset_inline_summary,
29308 inline_summary_t::duplicate): Update.
29309 (inline_edge_duplication_hook): Turn to ...
29310 (ipa_call_summary_t::duplicate): ... this one.
29311 (inline_edge_removal_hook): Turn to ...
29312 (ipa_call_summary_t::remove): ... this one.
29313 (dump_inline_edge_summary): Turn to ...
29314 (dump_ipa_call_summary): ... this one.
29315 (estimate_function_body_sizes): Update.
29316 (inline_update_callee_summaries): Update.
29317 (remap_edge_change_prob): Update.
29318 (remap_edge_summaries): Update.
29319 (inline_merge_summary): Update.
29320 (do_estimate_edge_time): Update.
29321 (inline_generate_summary): Update.
29322 (inline_read_section): Update.
29323 (inline_read_summary): Update.
29324 (inline_free_summary): Update.
29325 * ipa-inline.c (can_inline_edge_p): Update.
29326 (compute_inlined_call_time): Update.
29327 (want_inline_small_function_p): Update.
29328 (edge_badness): Update.
29329 (early_inliner): Update.
29330 * ipa-inline.h (inline_edge_summary): Turn to ...
29331 (ipa_call_summary): ... this one.
29332 (ipa_call_summary_t): New class.
29333 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29334 (ipa_call_summaries): New.
29335 (inline_edge_summary): Remove.
29336 (estimate_edge_growth): Update.
29337 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29338 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29339 * ipa-split.c (execute_split_functions): Update.
29340 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29342 2017-05-23 Tom de Vries <tom@codesourcery.com>
29344 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29345 attributes): Document rdrand effective target.
29347 2017-05-23 Tom de Vries <tom@codesourcery.com>
29349 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29350 attributes): Sort alphabetically.
29352 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29354 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29356 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29359 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29360 V2DF/V2DI splat into two separate patterns, one that handles
29361 registers, and the other that only handles memory. Drop support
29362 for splatting from a GPR on ISA 2.07 and then splitting the
29363 splat into direct move and splat.
29364 (vsx_splat_<mode>_reg): Likewise.
29365 (vsx_splat_<mode>_mem): Likewise.
29367 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29369 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29371 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29373 PR middle-end/80809
29374 * omp-low.c (finish_taskreg_remap): New function.
29375 (finish_taskreg_scan): If unit size of ctx->record_type
29376 is non-constant, unshare the size expression and replace
29377 decls in it with possible outer var refs.
29379 PR middle-end/80809
29380 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29381 GOVD_SHARED rather than GOVD_PRIVATE with it.
29382 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29383 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29385 PR middle-end/80853
29386 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29387 as last argument to build_outer_var_ref for pointer bases of array
29388 section reductions.
29390 2017-05-19 Martin Sebor <msebor@redhat.com>
29392 * print-tree.c (print_node): Print DECL_READ_P flag.
29394 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29396 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29397 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29398 * cgraph.c: Likewise.
29399 * cgraphunit.c: Likewise.
29400 * gengtype.c: Likewise.
29401 * ipa-cp.c: Likewise.
29402 * ipa-devirt.c: Likewise.
29403 * ipa-icf.c: Likewise.
29404 * ipa-predicate.c: Likewise.
29405 * ipa-profile.c: Likewise.
29406 * ipa-prop.c: Likewise.
29407 * ipa-split.c: Likewise.
29409 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29410 edge_predicate_pool, dump_inline_hints,
29411 inline_summary::account_size_time, redirect_to_unreachable,
29412 edge_set_predicate, set_hint_predicate,
29413 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29414 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29415 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29416 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29417 ipa_call_summary_t::remove, initialize_growth_caches,
29418 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29419 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29420 mark_modified, unmodified_parm_1, unmodified_parm,
29421 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29422 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29423 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29424 will_be_nonconstant_predicate, record_modified_bb_info,
29425 get_minimal_bb, record_modified, param_change_prob,
29426 phi_result_unknown_predicate, predicate_for_phi_result,
29427 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29428 estimate_function_body_sizes, compute_inline_parameters,
29429 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29430 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29431 inline_update_callee_summaries, remap_edge_change_prob,
29432 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29433 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29434 inline_analyze_function, inline_summary_t::insert,
29435 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29436 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29437 inline_free_summary): Move to ipa-fnsummary.h
29438 (predicate_t): Remove.
29439 * ipa-fnsummary.c: New file.
29440 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29441 (enum inline_hints_vals, inline_hints, agg_position_info,
29442 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29443 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29444 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29445 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29446 inline_read_summary, inline_write_summary, inline_free_summary,
29447 inline_analyze_function, initialize_inline_failed,
29448 inline_merge_summary, inline_update_overall_summary,
29449 compute_inline_parameters): Move to ipa-fnsummary.h
29450 * ipa-fnsummary.h: New file.
29451 * ipa-inline-transform.h: Include ipa-inline.h.
29452 * ipa-inline.c: LIkewise.
29454 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29456 * ipa-inline.c (edge_badness): Use inlined_time instead of
29457 inline_summaries->get.
29459 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29461 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29463 2017-05-22 Nathan Sidwell <nathan@acm.org>
29465 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29466 (fdump-lang): Document 'raw' option.
29467 * dumpfile.h (TDI_tu): Delete.
29468 * dumpfile.c (dump_files): Remove translation-unit.
29469 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29471 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29473 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29474 command option from $(AWK) call.
29475 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29476 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29477 [FORMAT]: Remove handling of variable.
29478 * config/avr/t-multilib: Regenerate.
29480 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29482 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29484 (dump_inline_summary): Do not print self_time.
29485 (estimate_function_body_sizes): Do not set self_time.
29486 (compute_inline_parameters): Likewise.
29487 (inline_read_section, inline_write_summary): Do not stream self_time.
29488 * ipa-inline.h (inline_summary): Drop self_time.
29490 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29492 * ipa-inline-analysis.c (account_size_time): Rename to ...
29493 (inline_summary::account_size_time): ... this one.
29494 (reset_ipa_call_summary): Turn to ...
29495 (ipa_call_summary::reset): ... this one.
29496 (reset_inline_summary): Turn to ...
29497 (inline_summary::reset): ... this one.
29498 (inline_summary_t::remove): Update.
29499 (inline_summary_t::duplicate): Update.
29500 (ipa_call_summary_t::remove): Update.
29501 (dump_inline_summary): Update.
29502 (estimate_function_body_sizes): Update.
29503 (compute_inline_parameters): Update.
29504 (estimate_node_size_and_time): Update.
29505 (inline_merge_summary): Update.
29506 (inline_update_overall_summary): Update.
29507 (inline_read_section): Update.
29508 (inline_write_summary): Update.
29509 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29510 add account_size_time and reset member functions.
29511 (ipa_call_summary): Add reset function.
29512 * ipa-predicate.h (predicate::operator &): Constify.
29514 2017-05-22 Richard Biener <rguenther@suse.de>
29516 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29518 2017-05-19 Jason Merrill <jason@redhat.com>
29520 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29522 2017-05-19 Marek Polacek <polacek@redhat.com>
29525 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29526 TYPE_OVERFLOW_WRAPS checks.
29528 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29530 * tree-core.h (enum omp_clause_default_kind): Add
29531 "OMP_CLAUSE_DEFAULT_PRESENT".
29532 * tree-pretty-print.c (dump_omp_clause): Handle it.
29533 * gimplify.c (enum gimplify_omp_var_data): Add
29534 "GOVD_MAP_FORCE_PRESENT".
29535 (gimplify_adjust_omp_clauses_1): Map it to
29536 "GOMP_MAP_FORCE_PRESENT".
29537 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29539 * gimplify.c (oacc_default_clause): Clarify.
29541 2017-05-19 Nathan Sidwell <nathan@acm.org>
29543 LANG_HOOK_REGISTER_DUMPS
29544 * toplev.c (general_init): Call register dump lang hook.
29545 * doc/invoke.texi: Document -fdump-lang option family.
29546 * dumpfile.c (dump_files): Remove class dump here.
29547 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29548 * dumpfile.h (tree_dump_index): Remove TDI_class.
29549 * langhooks-def.h (lhd_register_dumps): Declare.
29550 (LANG_HOOKS_REGISTER_DUMPS): Define.
29551 (LANG_HOOKS_INITIALIZER): Add it.
29552 * langhooks.c (lhd_register_dumps): Define.
29553 * langhooks.h (struct lang_hooks): Add register_dumps.
29555 2017-05-19 Nathan Sidwell <nathan@acm.org>
29557 * context.h (context::set_passes): New.
29558 * context.c (context::context): Do not create pass manager.
29559 * toplev.c (general_init): Create pass manager here.
29561 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29563 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29564 use this splitter if two add or or instructions would also work for
29565 the constant we want to generate.
29567 2017-05-19 Richard Biener <rguenther@suse.de>
29570 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29571 predicate evaluation.
29573 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29575 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29577 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29580 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29582 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29583 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29584 (inline_edge_summary_vec): Turn into ...
29585 (ipa_call_summaries): ... this one.
29586 (redirect_to_unreachable, edge_set_predicate,
29587 evaluate_properties_for_edge, inline_summary_alloc,
29588 reset_ipa_call_summary, reset_inline_summary,
29589 inline_summary_t::duplicate): Update.
29590 (inline_edge_duplication_hook): Turn to ...
29591 (ipa_call_summary_t::duplicate): ... this one.
29592 (inline_edge_removal_hook): Turn to ...
29593 (ipa_call_summary_t::remove): ... this one.
29594 (dump_inline_edge_summary): Turn to ...
29595 (dump_ipa_call_summary): ... this one.
29596 (estimate_function_body_sizes): Update.
29597 (inline_update_callee_summaries): Update.
29598 (remap_edge_change_prob): Update.
29599 (remap_edge_summaries): Update.
29600 (inline_merge_summary): Update.
29601 (do_estimate_edge_time): Update.
29602 (inline_generate_summary): Update.
29603 (inline_read_section): Update.
29604 (inline_read_summary): Update.
29605 (inline_free_summary): Update.
29606 * ipa-inline.c (can_inline_edge_p): Update.
29607 (compute_inlined_call_time): Update.
29608 (want_inline_small_function_p): Update.
29609 (edge_badness): Update.
29610 (early_inliner): Update.
29611 * ipa-inline.h (inline_edge_summary): Turn to ...
29612 (ipa_call_summary): ... this one.
29613 (ipa_call_summary_t): New class.
29614 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29615 (ipa_call_summaries): New.
29616 (inline_edge_summary): Remove.
29617 (estimate_edge_growth): Update.
29618 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29619 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29620 * ipa-split.c (execute_split_functions): Update.
29621 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29623 2017-05-19 Richard Biener <rguenther@suse.de>
29625 PR middle-end/80764
29626 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29628 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29630 * config/rs6000/rs6000.c (struct machine_function): Add field
29631 fpr_is_wrapped_separately.
29632 (rs6000_get_separate_components): Use 64 components. Handle the
29633 new FPR components.
29634 (rs6000_components_for_bb): Handle the FPR components.
29635 (rs6000_emit_prologue_components): Handle the FPR components.
29636 (rs6000_emit_epilogue_components): Handle the FPR components.
29637 (rs6000_set_handled_components): Handle the FPR components.
29638 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29639 that are already separately shrink-wrapped.
29640 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29641 that are already separately shrink-wrapped.
29643 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29646 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29649 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29650 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29651 cases where the register allocator uses a move and an offsettable
29652 memory operation to/from a FPR register on ISA 2.06/2.07.
29653 (define_peephole2 for Altivec d-form store): Likewise.
29655 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29658 * config/i386/mmx.md (*mov<mode>_internal): Enable
29659 alternatives 11, 12, 13 and 14 also for 32bit targets.
29660 Remove alternatives 15, 16, 17 and 18.
29661 * config/i386/sse.md (vec_concatv2di): Change
29662 alternative (!x, *y) to (x, ?!*Yn).
29664 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29666 * dumpfile.h (enum dump_kind): Remove stray comma.
29668 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29670 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29671 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29672 predicate::num_conditions
29673 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29674 (CHANGED): turn into predicate::changed.
29675 (agg_position_info): Move to ipa-predicate.h
29676 (add_condition, predicate::add_clause, predicate::operator &=,
29677 predicate::or_with, predicate::evaluate, predicate::probability,
29678 dump_condition, dump_clause, predicate::dump,
29679 predicate::remap_after_duplication, predicate::remap_after_inlining,
29680 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29681 (evaluate_conditions_for_known_args): Update.
29682 (set_cond_stmt_execution_predicate): Update.
29683 * ipa-inline.h: Include ipa-predicate.h
29684 (condition, inline_param_summary, conditions, agg_position_info,
29685 predicate): Move to ipa-predicate.h
29686 * ipa-predicate.c: New file.
29687 * ipa-predicate.h: New file.
29689 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29691 * final.c (leaf_function_p): Check we are not in a sequence.
29693 2017-05-18 Martin Liska <mliska@suse.cz>
29695 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29696 * dumpfile.c (dump_register): Use new enum dump_kind.
29697 (get_dump_file_name): Likewise.
29698 (dump_enable_all): Likewise.
29699 (dump_switch_p_1): Likewise.
29700 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29701 * dumpfile.h (enum dump_kind): New enum type.
29702 (struct dump_file_info): Create constructor and
29703 format fields and comments.
29704 * passes.c (pass_manager::register_one_dump_file):
29706 * statistics.c (statistics_early_init): Likewise.
29707 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29708 TDF_TREE with TDF_SLIM.
29709 (gather_memory_references_ref): Likewise.
29711 2017-05-18 Martin Liska <mliska@suse.cz>
29713 * vec.h (struct vnull): Use it.
29715 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29717 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29718 (true_predicate, false_predicate, true_predicate_p,
29719 false_predicate_p): Remove.
29720 (single_cond_predicate, not_inlined_predicate): Turn to member function
29722 (add_condition): Update.
29723 (add_clause): Turn to...
29724 (predicate::add_clause): ... this one; update; allow passing NULL
29726 (and_predicates): Turn to ...
29727 (predicate::operator &=): ... this one.
29728 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29729 (or_predicates): Turn to ...
29730 (predicate::or_with): ... this one.
29731 (evaluate_predicate): Turn to ...
29732 (predicate::evaluate): ... this one.
29733 (predicate_probability): Turn to ...
29734 (predicate::probability): ... this one.
29735 (dump_condition): Update.
29736 (dump_predicate): Turn to ...
29737 (predicate::dump): ... this one.
29738 (account_size_time): Update.
29739 (edge_set_predicate): Update.
29740 (set_hint_predicate): UPdate.
29741 (evaluate_conditions_for_known_args): Update.
29742 (evaluate_properties_for_edge): Update.
29743 (remap_predicate_after_duplication): Turn to...
29744 (predicate::remap_after_duplication): ... this one.
29745 (remap_hint_predicate_after_duplication): Update.
29746 (inline_summary_t::duplicate): UPdate.
29747 (dump_inline_edge_summary): Update.
29748 (dump_inline_summary): Update.
29749 (set_cond_stmt_execution_predicate): Update.
29750 (set_switch_stmt_execution_predicate): Update.
29751 (compute_bb_predicates): Update.
29752 (will_be_nonconstant_expr_predicate): Update.
29753 (will_be_nonconstant_predicate): Update.
29754 (phi_result_unknown_predicate): Update.
29755 (predicate_for_phi_result): Update.
29756 (array_index_predicate): Update.
29757 (estimate_function_body_sizes): Update.
29758 (estimate_node_size_and_time): Update.
29759 (estimate_ipcp_clone_size_and_time): Update.
29760 (remap_predicate): Rename to ...
29761 (predicate::remap_after_inlining): ... this one.
29762 (remap_hint_predicate): Update.
29763 (inline_merge_summary): Update.
29764 (inline_update_overall_summary): Update.
29765 (estimate_size_after_inlining): Update.
29766 (read_predicate): Rename to ...
29767 (predicate::stream_in): ... this one.
29768 (read_inline_edge_summary): Update.
29769 (write_predicate): Rename to ...
29770 (predicate::stream_out): ... this one.
29771 (write_inline_edge_summary): Update.
29772 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29773 (clause_t): Turn to uint32_t
29774 (predicate): Turn to class; implement constructor and operators
29776 (size_time_entry): Update.
29777 (inline_summary): Update.
29778 (inline_edge_summary): Update.
29780 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29782 * fold-const.c (fold_binary_loc): Move transformation...
29783 * match.pd (C - X CMP X): ... here.
29785 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29787 * config/sparc/sparc.c (sparc_option_override): Set function
29788 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29789 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29791 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29793 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29795 2017-05-18 Marek Polacek <polacek@redhat.com>
29798 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29799 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29800 gimple_assign_load_p.
29802 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29804 PR middle-end/80692
29805 * real.c (do_compare): Give decimal_do_compare preference over
29806 comparing just the signs.
29808 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29810 * doc/md.texi (Canonicalization of Instructions): Describe the
29811 canonical form of instructions that inherently set a condition
29814 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29816 PR middle-end/80775
29817 * tree-cfg.c: Move deletion of unreachable case statements to after
29818 the merging of consecutive case labels.
29820 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29822 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29823 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29824 restoring of callee-saved registers.
29826 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29828 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29829 * config/visium/visium.c (single_set_and_flags): Likewise.
29830 * config/visium/visium.md (Substitutions): Likewise.
29832 2017-05-17 Martin Liska <mliska@suse.cz>
29834 * cfg.c: Introduce dump_flags_t type and
29835 use it instead of int type.
29837 * cfghooks.c: Likewise.
29838 * cfghooks.h (struct cfg_hooks): Likewise.
29839 * cfgrtl.c: Likewise.
29840 * cfgrtl.h: Likewise.
29841 * cgraph.c (cgraph_node::get_body): Likewise.
29842 * coretypes.h: Likewise.
29843 * domwalk.c: Likewise.
29844 * domwalk.h: Likewise.
29845 * dumpfile.c (struct dump_option_value_info): Likewise.
29846 (dump_enable_all): Likewise.
29847 (dump_switch_p_1): Likewise.
29848 (opt_info_switch_p): Likewise.
29849 * dumpfile.h (enum tree_dump_index): Likewise.
29850 (struct dump_file_info): Likewise.
29851 * genemit.c: Likewise.
29852 * generic-match-head.c: Likewise.
29853 * gengtype.c (open_base_files): Likewise.
29854 * gimple-pretty-print.c: Likewise.
29855 * gimple-pretty-print.h: Likewise.
29856 * graph.c (print_graph_cfg): Likewise.
29857 * graphite-scop-detection.c (dot_all_sese): Likewise.
29858 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29859 * loop-unroll.c (report_unroll): Likewise.
29860 * passes.c (pass_manager::register_one_dump_file): Likewise.
29861 * print-tree.c: Likewise.
29862 * statistics.c: Likewise.
29863 * tree-cfg.c: Likewise.
29864 * tree-cfg.h: Likewise.
29865 * tree-dfa.c: Likewise.
29866 * tree-dfa.h: Likewise.
29867 * tree-dump.c (dump_function): Likewise.
29868 * tree-dump.h (struct dump_info): Likewise.
29869 * tree-pretty-print.c: Likewise.
29870 * tree-pretty-print.h: Likewise.
29871 * tree-ssa-live.c: Likewise.
29872 * tree-ssa-live.h: Likewise.
29873 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29874 * tree-vect-loop.c: Likewise.
29875 * tree-vect-slp.c: Likewise.
29877 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29878 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29880 PR tree-optimization/80457
29881 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29882 of all arguments to a statement as scalar_to_vec operations.
29883 (vectorizable_call): Adjust call to vect_model_simple_cost for
29885 (vectorizable_conversion): Likewise.
29886 (vectorizable_assignment): Likewise.
29887 (vectorizable_shift): Likewise.
29888 (vectorizable_operation): Likewise.
29889 (vectorizable_comparison): Likewise.
29890 (vect_is_simple_cond): Record the def types for operands.
29891 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29892 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29893 for statement argument count.
29895 2017-05-16 Carl Love <cel@us.ibm.com>
29897 * config/rs6000/rs6000-c: Add support for built-in functions
29898 vector unsigned long long vec_bperm (vector unsigned long long,
29899 vector unsigned char)
29900 vector signed long long vec_mule (vector signed int,
29902 vector unsigned long long vec_mule (vector unsigned int,
29903 vector unsigned int)
29904 vector signed long long vec_mulo (vector signed int,
29906 vector unsigned long long vec_mulo (vector unsigned int,
29907 vector unsigned int)
29908 vector signed char vec_sldw (vector signed char,
29909 vector signed char,
29911 vector unsigned char vec_sldw (vector unsigned char,
29912 vector unsigned char,
29914 vector signed short vec_sldw (vector signed short,
29915 vector signed short,
29917 vector unsigned short vec_sldw (vector unsigned short,
29918 vector unsigned short,
29920 vector signed int vec_sldw (vector signed int,
29923 vector unsigned int vec_sldw (vector unsigned int,
29924 vector unsigned int,
29926 vector signed long long vec_sldw (vector signed long long,
29927 vector signed long long,
29929 vector unsigned long long vec_sldw (vector unsigned long long,
29930 vector unsigned long long,
29932 * config/rs6000/rs6000-c: Add support for built-in functions
29933 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29934 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29935 * doc/extend.texi: Update the built-in documentation for the
29936 new built-in functions.
29938 2017-05-16 Marek Polacek <polacek@redhat.com>
29942 * tree.c (save_expr): Don't fold the expression.
29944 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29946 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29947 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29948 and (?*y,m). Update insn attributes.
29950 2017-05-16 Martin Liska <mliska@suse.cz>
29952 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29953 flags argument of print_gimple_stmt, print_gimple_expr,
29954 print_generic_stmt and print_generic_expr.
29955 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29956 * coretypes.h: Likewise.
29957 * except.c (dump_eh_tree): Likewise.
29958 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29959 * gimple-pretty-print.h: Likewise.
29960 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29961 (backprop::push_to_worklist): Likewise.
29962 (backprop::pop_from_worklist): Likewise.
29963 (backprop::process_use): Likewise.
29964 (backprop::intersect_uses): Likewise.
29965 (note_replacement): Likewise.
29966 * gimple-ssa-store-merging.c
29967 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29968 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29969 (pass_store_merging::execute): Likewise.
29970 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29971 (ssa_base_cand_dump_callback): Likewise.
29972 (dump_incr_vec): Likewise.
29973 (replace_refs): Likewise.
29974 (replace_mult_candidate): Likewise.
29975 (create_add_on_incoming_edge): Likewise.
29976 (create_phi_basis): Likewise.
29977 (insert_initializers): Likewise.
29978 (all_phi_incrs_profitable): Likewise.
29979 (introduce_cast_before_cand): Likewise.
29980 (replace_one_candidate): Likewise.
29981 * gimplify.c (gimplify_expr): Likewise.
29982 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29983 (set_rename): Likewise.
29984 (rename_uses): Likewise.
29985 (copy_loop_phi_nodes): Likewise.
29986 (add_close_phis_to_merge_points): Likewise.
29987 (copy_loop_close_phi_args): Likewise.
29988 (copy_cond_phi_args): Likewise.
29989 (graphite_copy_stmts_from_block): Likewise.
29990 (translate_pending_phi_nodes): Likewise.
29991 * graphite-poly.c (print_pdr): Likewise.
29992 (dump_gbb_cases): Likewise.
29993 (dump_gbb_conditions): Likewise.
29994 (print_scop_params): Likewise.
29995 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29996 (build_cross_bb_scalars_use): Likewise.
29997 (gather_bbs::before_dom_children): Likewise.
29998 * hsa-dump.c (dump_hsa_immed): Likewise.
29999 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30000 (get_replacement_map): Likewise.
30001 * ipa-inline-analysis.c (dump_condition): Likewise.
30002 (estimate_function_body_sizes): Likewise.
30003 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30004 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30005 * ipa-prop.c (ipa_dump_param): Likewise.
30006 (ipa_print_node_jump_functions_for_edge): Likewise.
30007 (ipa_modify_call_arguments): Likewise.
30008 (ipa_modify_expr): Likewise.
30009 (ipa_dump_param_adjustments): Likewise.
30010 (ipa_dump_agg_replacement_values): Likewise.
30011 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30012 * ipa-pure-const.c (check_stmt): Likewise.
30013 (pass_nothrow::execute): Likewise.
30014 * ipa-split.c (execute_split_functions): Likewise.
30015 * omp-offload.c (dump_oacc_loop_part): Likewise.
30016 (dump_oacc_loop): Likewise.
30017 * trans-mem.c (tm_log_emit): Likewise.
30018 (tm_memopt_accumulate_memops): Likewise.
30019 (dump_tm_memopt_set): Likewise.
30020 (dump_tm_memopt_transform): Likewise.
30021 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30022 (print_loop): Likewise.
30023 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30024 (chkp_gather_checks_info): Likewise.
30025 (chkp_get_check_result): Likewise.
30026 (chkp_remove_check_if_pass): Likewise.
30027 (chkp_use_outer_bounds_if_possible): Likewise.
30028 (chkp_reduce_bounds_lifetime): Likewise.
30029 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30030 (chkp_mark_completed_bounds): Likewise.
30031 (chkp_register_incomplete_bounds): Likewise.
30032 (chkp_mark_invalid_bounds): Likewise.
30033 (chkp_maybe_copy_and_register_bounds): Likewise.
30034 (chkp_build_returned_bound): Likewise.
30035 (chkp_get_bound_for_parm): Likewise.
30036 (chkp_build_bndldx): Likewise.
30037 (chkp_get_bounds_by_definition): Likewise.
30038 (chkp_generate_extern_var_bounds): Likewise.
30039 (chkp_get_bounds_for_decl_addr): Likewise.
30040 * tree-chrec.c (chrec_apply): Likewise.
30041 * tree-data-ref.c (dump_data_reference): Likewise.
30042 (dump_subscript): Likewise.
30043 (dump_data_dependence_relation): Likewise.
30044 (analyze_overlapping_iterations): Likewise.
30045 * tree-inline.c (expand_call_inline): Likewise.
30046 (tree_function_versioning): Likewise.
30047 * tree-into-ssa.c (dump_defs_stack): Likewise.
30048 (dump_currdefs): Likewise.
30049 (dump_names_replaced_by): Likewise.
30050 (dump_update_ssa): Likewise.
30051 (update_ssa): Likewise.
30052 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30053 * tree-parloops.c (build_new_reduction): Likewise.
30054 (try_create_reduction_list): Likewise.
30055 (ref_conflicts_with_region): Likewise.
30056 (oacc_entry_exit_ok_1): Likewise.
30057 (oacc_entry_exit_single_gang): Likewise.
30058 * tree-pretty-print.h: Likewise.
30059 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30060 (get_scalar_evolution): Likewise.
30061 (add_to_evolution): Likewise.
30062 (get_loop_exit_condition): Likewise.
30063 (analyze_evolution_in_loop): Likewise.
30064 (analyze_initial_condition): Likewise.
30065 (analyze_scalar_evolution): Likewise.
30066 (instantiate_scev): Likewise.
30067 (number_of_latch_executions): Likewise.
30068 (gather_chrec_stats): Likewise.
30069 (final_value_replacement_loop): Likewise.
30070 (scev_const_prop): Likewise.
30071 * tree-sra.c (dump_access): Likewise.
30072 (disqualify_candidate): Likewise.
30073 (create_access): Likewise.
30074 (reject): Likewise.
30075 (maybe_add_sra_candidate): Likewise.
30076 (create_access_replacement): Likewise.
30077 (analyze_access_subtree): Likewise.
30078 (analyze_all_variable_accesses): Likewise.
30079 (sra_modify_assign): Likewise.
30080 (initialize_constant_pool_replacements): Likewise.
30081 (find_param_candidates): Likewise.
30082 (decide_one_param_reduction): Likewise.
30083 (replace_removed_params_ssa_names): Likewise.
30084 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30085 * tree-ssa-copy.c (dump_copy_of): Likewise.
30086 (copy_prop_visit_cond_stmt): Likewise.
30087 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30088 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30089 (record_equivalences_from_stmt): Likewise.
30090 * tree-ssa-dse.c (compute_trims): Likewise.
30091 (delete_dead_call): Likewise.
30092 (delete_dead_assignment): Likewise.
30093 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30094 (forward_propagate_into_cond): Likewise.
30095 (pass_forwprop::execute): Likewise.
30096 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30097 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30099 (move_computations_worker): Likewise.
30100 (execute_sm): Likewise.
30101 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30102 (remove_exits_and_undefined_stmts): Likewise.
30103 (remove_redundant_iv_tests): Likewise.
30104 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30105 (adjust_iv_update_pos): Likewise.
30106 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30107 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30108 (value_replacement): Likewise.
30109 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30110 * tree-ssa-pre.c (print_pre_expr): Likewise.
30111 (get_representative_for): Likewise.
30112 (create_expression_by_pieces): Likewise.
30113 (insert_into_preds_of_block): Likewise.
30114 (eliminate_insert): Likewise.
30115 (eliminate_dom_walker::before_dom_children): Likewise.
30116 (eliminate): Likewise.
30117 (remove_dead_inserted_code): Likewise.
30118 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30119 * tree-ssa-reassoc.c (get_rank): Likewise.
30120 (eliminate_duplicate_pair): Likewise.
30121 (eliminate_plus_minus_pair): Likewise.
30122 (eliminate_not_pairs): Likewise.
30123 (undistribute_ops_list): Likewise.
30124 (eliminate_redundant_comparison): Likewise.
30125 (update_range_test): Likewise.
30126 (optimize_range_tests_var_bound): Likewise.
30127 (optimize_vec_cond_expr): Likewise.
30128 (rewrite_expr_tree): Likewise.
30129 (rewrite_expr_tree_parallel): Likewise.
30130 (linearize_expr): Likewise.
30131 (break_up_subtract): Likewise.
30132 (linearize_expr_tree): Likewise.
30133 (attempt_builtin_powi): Likewise.
30134 (attempt_builtin_copysign): Likewise.
30135 (transform_stmt_to_copy): Likewise.
30136 (transform_stmt_to_multiply): Likewise.
30137 (dump_ops_vector): Likewise.
30138 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30139 (print_scc): Likewise.
30140 (set_ssa_val_to): Likewise.
30141 (visit_reference_op_store): Likewise.
30142 (visit_use): Likewise.
30143 (sccvn_dom_walker::before_dom_children): Likewise.
30144 (run_scc_vn): Likewise.
30145 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30147 (expr_hash_elt::print): Likewise.
30148 (const_and_copies::pop_to_marker): Likewise.
30149 (const_and_copies::record_const_or_copy_raw): Likewise.
30150 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30151 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30152 (dump_predicates): Likewise.
30153 (find_uninit_use): Likewise.
30154 (warn_uninitialized_phi): Likewise.
30155 (pass_late_warn_uninitialized::execute): Likewise.
30156 * tree-ssa.c (verify_vssa): Likewise.
30157 (verify_ssa): Likewise.
30158 (maybe_optimize_var): Likewise.
30159 * tree-vrp.c (dump_value_range): Likewise.
30160 (dump_all_value_ranges): Likewise.
30161 (dump_asserts_for): Likewise.
30162 (register_edge_assert_for_2): Likewise.
30163 (vrp_visit_cond_stmt): Likewise.
30164 (vrp_visit_switch_stmt): Likewise.
30165 (vrp_visit_stmt): Likewise.
30166 (vrp_visit_phi_node): Likewise.
30167 (simplify_cond_using_ranges_1): Likewise.
30168 (fold_predicate_in): Likewise.
30169 (evrp_dom_walker::before_dom_children): Likewise.
30170 (evrp_dom_walker::push_value_range): Likewise.
30171 (evrp_dom_walker::pop_value_range): Likewise.
30172 (execute_early_vrp): Likewise.
30174 2017-05-16 Richard Biener <rguenther@suse.de>
30176 * dwarf2out.c (loc_list_from_tree_1): Do not create
30177 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30179 2017-05-16 Richard Biener <rguenther@suse.de>
30181 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30183 (note_variable_value_in_expr): If we resolved the decl ref
30184 do not push to the stack.
30186 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30188 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30189 operations in fast-math mode.
30190 (vaddq_f16): Likewise.
30191 (vmul_f16): Likewise.
30192 (vmulq_f16): Likewise.
30193 (vsub_f16): Likewise.
30194 (vsubq_f16): Likewise.
30195 * config/arm/neon.md (add<mode>3): New.
30197 (fma:<VH:mode>3): New. Also remove outdated comment.
30200 2017-05-16 Martin Liska <mliska@suse.cz>
30204 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30205 (odr_types_equivalent_p): Likewise.
30207 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30209 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30211 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30214 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30215 non-interunit SSE move alternatives with '?'.
30216 (zero-extendsidi peephole2): New peephole to skip intermediate
30217 general register in SSE zero-extend sequence.
30219 2017-05-15 Jeff Law <law@redhat.com>
30221 * reorg.c (relax_delay_slots): Create a new variable to hold
30222 the temporary target rather than clobbering TARGET_LABEL.
30224 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30225 missing argument to extract_bit_field call.
30226 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30228 2017-05-15 Martin Liska <mliska@suse.cz>
30231 * gcc.c (process_command): Do not allow empty argument of -o option.
30233 2017-05-15 Renlin Li <renlin.li@arm.com>
30235 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30236 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30237 * config/aarch64/constraints.md (Usf): Add long call check.
30238 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30239 (call_value): Likewise.
30240 (sibcall): Likewise.
30241 (sibcall_value): Likewise.
30243 (call_value_insn): New.
30244 (sibcall_insn): Update rtx pattern.
30245 (sibcall_value_insn): Likewise.
30246 (call_internal): Remove.
30247 (call_value_internal): Likewise.
30248 (sibcall_internal): Likewise.
30249 (sibcall_value_internal): Likewise.
30250 (call_reg): Likewise.
30251 (call_symbol): Likewise.
30252 (call_value_reg): Likewise.
30253 (call_value_symbol): Likewise.
30255 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30258 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30260 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30262 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30263 compatible with CCGOCmode and with CCZmode.
30265 2017-05-14 Martin Sebor <msebor@redhat.com>
30267 PR middle-end/77671
30268 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30269 (gimple_fold_builtin_snprintf): Same.
30270 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30271 (gimple_fold_builtin_snprintf): Same.
30272 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30273 of character types.
30274 (is_call_safe): New function.
30275 (try_substitute_return_value): Call it.
30276 (try_simplify_call): New function.
30277 (pass_sprintf_length::handle_gimple_call): Call it.
30279 2017-05-14 Martin Sebor <msebor@redhat.com>
30281 PR middle-end/80669
30282 * builtins.c (expand_builtin_stpncpy): Simplify.
30284 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30286 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30287 * config/i386/i386.h
30288 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30289 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30290 (struct machine_function): Add new members call_ms2sysv,
30291 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30292 (struct machine_frame_state): New fields sp_realigned and
30293 sp_realigned_offset.
30294 * config/i386/i386.c
30295 (enum xlogue_stub): New enum.
30296 (enum xlogue_stub_sets): New enum.
30297 (class xlogue_layout): New class.
30298 (struct ix86_frame): New fields stack_realign_allocate_offset,
30299 stack_realign_offset and outlined_save_offset. Modify comments to
30300 detail stack layout when using out-of-line stubs.
30301 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30302 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30303 -mcall-ms2sysv-xlogues.
30304 (stub_managed_regs): New static variable.
30305 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30306 registers managed by out-of-line stub.
30307 (disable_call_ms2sysv_xlogues): New function.
30308 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30309 m->call_ms2sysv when appropriate and compute frame layout for
30311 (sp_valid_at, fp_valid_at): New inline functions.
30312 (choose_basereg): New function.
30313 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30315 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30316 Use align parameter of choose_baseaddr to generated aligned SSE movs
30318 (pro_epilogue_adjust_stack): Modify to track
30319 machine_frame_state::sp_realigned.
30320 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30321 (ix86_nsaved_sseregs): Likewise.
30322 (ix86_emit_save_regs): Likewise.
30323 (ix86_emit_save_regs_using_mov): Likewise.
30324 (ix86_emit_save_sse_regs_using_mov): Likewise.
30325 (get_scratch_register_on_entry): Likewise.
30326 (gen_frame_set): New function.
30327 (gen_frame_load): Likewise.
30328 (gen_frame_store): Likewise.
30329 (emit_outlined_ms2sysv_save): Likewise.
30330 (emit_outlined_ms2sysv_restore): Likewise.
30331 (ix86_expand_prologue): Modify stack re-alignment code and call
30332 emit_outlined_ms2sysv_save when appropriate.
30333 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30334 parameter rtx_insn *insn, which allows the function to be used to only
30335 generate the notes.
30336 (ix86_expand_epilogue): Modify validity checks of frame and stack
30337 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30338 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30339 * config/i386/predicates.md
30340 (save_multiple): New predicate.
30341 (restore_multiple): Likewise.
30342 * config/i386/sse.md
30343 (save_multiple<mode>): New pattern.
30344 (save_multiple_realign<mode>): Likewise.
30345 (restore_multiple<mode>): Likewise.
30346 (restore_multiple_and_return<mode>): Likewise.
30347 (restore_multiple_leave_return<mode>): Likewise.
30348 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30350 2017-05-14 Julia Koval <julia.koval@intel.com>
30352 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30353 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30354 (__builtin_ia32_xsetbv): New builtins.
30355 * config/i386/i386.c (ix86_expand_special_args_builtin):
30357 (ix86_expand_builtin): Special expand for new intrinsics.
30358 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30359 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30360 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30362 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30364 * cfganal.c (inverted_post_order_compute): Change argument type
30366 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30367 * df-core.c (rest_of_handle_df_initialize): Adjust.
30368 (rest_of_handle_df_finish): Likewise.
30369 (df_analyze_1): Likewise.
30370 (df_analyze): Likewise.
30371 (loop_inverted_post_order_compute): Change argument to be a vec *.
30372 (df_analyze_loop): Adjust.
30373 (df_get_n_blocks): Likewise.
30374 (df_get_postorder): Likewise.
30375 * df.h (struct df_d): Change field to be a vec.
30376 * lcm.c (compute_laterin): Adjust.
30377 (compute_available): Likewise.
30378 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30379 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30380 * tree-ssa-pre.c (compute_antic): Likewise.
30382 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30384 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30385 (depth_first_search::depth_first_search): Change structure init
30386 function to this constructor.
30387 (depth_first_search::add_bb): Rename function to this member.
30388 (depth_first_search::execute): Likewise.
30389 (flow_dfs_compute_reverse_finish): Adjust.
30391 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30393 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30394 (longest_simple_path): Likewise.
30395 * shrink-wrap.c (spread_components): Likewise.
30396 (disqualify_problematic_components): Likewise.
30397 (emit_common_heads_for_components): Likewise.
30398 (emit_common_tails_for_components): Likewise.
30399 (insert_prologue_epilogue_for_components): Likewise.
30401 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30403 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30406 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30408 * df-core.c (df_set_blocks): Start using auto_bitmap.
30409 (df_compact_blocks): Likewise.
30410 * df-problems.c (df_rd_confluence_n): Likewise.
30411 * df-scan.c (df_insn_rescan_all): Likewise.
30412 (df_process_deferred_rescans): Likewise.
30413 (df_update_entry_block_defs): Likewise.
30414 (df_update_exit_block_uses): Likewise.
30415 (df_entry_block_bitmap_verify): Likewise.
30416 (df_exit_block_bitmap_verify): Likewise.
30417 (df_scan_verify): Likewise.
30418 * lra-constraints.c (lra_constraints): Likewise.
30419 (undo_optional_reloads): Likewise.
30420 (lra_undo_inheritance): Likewise.
30421 * lra-remat.c (calculate_gen_cands): Likewise.
30422 (do_remat): Likewise.
30423 * lra-spills.c (assign_spill_hard_regs): Likewise.
30424 (spill_pseudos): Likewise.
30425 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30426 (bitmap_set_subtract_values): Likewise.
30428 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30430 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30431 management with auto_bitmap.
30432 (fix_inter_tick): Likewise.
30433 (fix_recovery_deps): Likewise.
30434 * ira.c (add_store_equivs): Likewise.
30435 (find_moveable_pseudos): Likewise.
30436 (split_live_ranges_for_shrink_wrap): Likewise.
30437 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30438 (rtx_reuse_manager::seen_def_p): Likewise.
30439 (rtx_reuse_manager::set_seen_def): Likewise.
30440 * print-rtl.h (class rtx_reuse_manager): Likewise.
30442 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30444 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30446 (migrate_btr_def): Likewise.
30447 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30448 * df-core.c (loop_post_order_compute): Likewise.
30449 (loop_inverted_post_order_compute): Likewise.
30450 * hsa-common.h: Likewise.
30451 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30452 * init-regs.c (initialize_uninitialized_regs): Likewise.
30453 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30454 (inline_small_functions): Likewise.
30455 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30456 * ira.c (combine_and_move_insns): Likewise.
30457 (build_insn_chain): Likewise.
30458 * loop-invariant.c (find_invariants): Likewise.
30459 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30460 * predict.c (tree_predict_by_opcode): Likewise.
30461 (predict_paths_leading_to): Likewise.
30462 (predict_paths_leading_to_edge): Likewise.
30463 (estimate_loops_at_level): Likewise.
30464 (estimate_loops): Likewise.
30465 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30466 (spread_components): Likewise.
30467 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30468 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30469 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30470 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30471 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30472 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30473 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30474 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30475 (mark_threaded_blocks): Likewise.
30476 (thread_through_all_blocks): Likewise.
30477 * tree-ssa.c (verify_ssa): Likewise.
30478 (execute_update_addresses_taken): Likewise.
30479 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30481 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30483 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30485 (post_order_compute): Likewise.
30486 (inverted_post_order_compute): Likewise.
30487 (pre_and_rev_post_order_compute_fn): Likewise.
30489 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30491 * genrecog.c (int_set::int_set): Explicitly construct our
30492 auto_vec base class.
30493 * vec.h (auto_vec::auto_vec): New constructor.
30495 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30497 * bitmap.h (class auto_bitmap): New constructor taking
30498 bitmap_obstack * argument.
30500 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30502 * bitmap.h (class auto_bitmap): Change type of m_bits to
30503 bitmap_head, and adjust ctor / dtor and member operators.
30505 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30507 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30508 when returned register mode doesn't match original mode.
30510 2017-05-12 Jeff Law <law@redhat.com>
30511 Jakub Jelinek <jakub@redhat.com>
30513 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30514 we look for cc setter after the compare-elim changes.
30515 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30516 within the vector to match what compare-elim now expects.
30517 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30518 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30520 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30521 after the compare-elim changes.
30522 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30523 the vector to match what compare-elim now expects.
30524 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30525 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30526 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30527 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30528 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30530 * config/visium/visium.c (single_set_and_flags): Fix where
30531 we look for cc setter after the compare-elim changes.
30532 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30533 with the vector to match what compare-elim now expects.
30534 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30535 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30536 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30537 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30538 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30539 (neg<mode>2_insn_set_overflow): Likewise.
30541 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30543 PR middle-end/79794
30544 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30545 maybe_expand_insn call, set ops[0].target. If still set after call,
30546 set alt_rtl. Add extra arg to recursive calls.
30547 (extract_bit_field): Add alt_rtl argument. Pass to
30549 * expmed.h (extract_bit_field): Fix prototype.
30550 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30551 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30552 to extract_bit_field_calls.
30553 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30554 Pass alt_rtl to extract_bit_field calls.
30555 * calls.c (store_unaligned_arguments_into_psuedos)
30556 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30557 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30559 * optabs.h (struct expand_operand): Add target bitfield.
30561 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30563 * compare-elim.c (try_eliminate_compare): Canonicalize
30564 operation with embedded compare to
30565 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30566 (set (reg) (operation)].
30568 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30570 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30573 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30574 cost of adding a carry flag for ADC instruction.
30575 [case MINUS]: Ignore the cost of subtracting a carry flag
30576 for SBB instruction.
30578 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30580 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30582 * config/rs6000/bmiintrin.h: New file.
30583 * config/rs6000/bmi2intrin.h: New file.
30584 * config/rs6000/x86intrin.h: New file.
30586 2017-05-12 Jeff Law <law@redhat.com>
30588 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30591 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30593 PR middle-end/80707
30594 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30596 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30598 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30599 early expansion of vector divide builtins.
30600 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30601 builtins identified as having unsigned arguments.
30603 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30605 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30606 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30607 expansion of vector logical operations (and, andc, or, xor,
30610 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30612 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30613 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30615 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30617 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30618 early GIMPLE expansion of vector multiplies.
30620 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30622 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30623 TARGET_HAVE_MOVT conditional.
30624 (movt splitter): Likewise.
30626 2017-05-12 Richard Biener <rguenther@suse.de>
30628 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30629 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30630 Fold all stmts not inplace.
30632 2017-05-12 Richard Biener <rguenther@suse.de>
30634 PR tree-optimization/80713
30635 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30636 inserted_exprs bit for not removed stmts.
30638 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30640 PR middle-end/69921
30641 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30642 parallelized" attribute for parallelized OpenACC kernels.
30643 * omp-offload.c (execute_oacc_device_lower): Use it.
30645 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30646 Set "oacc kernels" attribute.
30647 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30648 parameter. Adjust all users.
30649 (oacc_fn_attrib_kernels_p): Remove function.
30650 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30651 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30652 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30653 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30654 assert "oacc kernels" attribute is set.
30656 2017-05-11 Carl Love <cel@us.ibm.com>
30658 * config/rs6000/rs6000-c: Add support for built-in functions
30659 vector unsigned char vec_popcnt (vector signed char)
30660 vector unsigned char vec_popcnt (vector unsigned char)
30661 vector unsigned short vec_popcnt (vector signed short)
30662 vector unsigned short vec_popcnt (vector unsigned short)
30663 vector unsigned int vec_popcnt (vector signed int)
30664 vector unsigned int vec_popcnt (vector unsigned int)
30665 vector unsigned long long vec_popcnt (vector signed long long)
30666 vector unsigned long long vec_popcnt (vector unsigned long long)
30667 vector signed long long vec_slo (vector signed long long,
30668 vector signed char)
30669 vector signed long long vec_slo (vector signed long long,
30670 vector unsigned char)
30671 vector unsigned long long vec_slo (vector unsigned long long,
30672 vector signed char)
30673 vector unsigned long long vec_slo (vector unsigned long long,
30674 vector unsigned char)
30675 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30676 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30677 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30678 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30679 * doc/extend.texi: Update the built-in documentation file for the
30680 new built-in functions.
30682 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30684 * attribs.h (sorted_attr_string): Move machine independent
30685 functions for target clone support from the i386 port to common
30686 code. Rename ix86_function_versions to common_function_versions.
30687 Rename make_name to make_unique_name.
30688 (common_function_versions): Likewise.
30689 (make_unique_name): Likewise.
30690 (make_dispatcher_decl): Likewise.
30691 (is_function_default_version): Likewise.
30692 * attribs.c (attr_strcmp): Likewise.
30693 (sorted_attr_string): Likewise.
30694 (common_function_versions): Likewise.
30695 (make_unique_name): Likewise.
30696 (make_dispatcher_decl): Likewise.
30697 (is_function_default_version): Likewise.
30698 * config/i386/i386.c (attr_strcmp): Likewise.
30699 (sorted_attr_string): Likewise.
30700 (ix86_function_versions): Likewise.
30701 (make_name): Likewise.
30702 (make_dispatcher_decl): Likewise.
30703 (is_function_default_version): Likewise.
30704 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30706 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30709 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30710 Account for direct move costs for vec_construct of integer
30713 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30716 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30717 (UNSPEC_STX_ATOMIC): Ditto.
30718 (loaddi_via_sse): New insn.
30719 (storedi_via_sse): Ditto.
30720 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30721 Update corresponding peephole2 patterns.
30722 (atomic_storedi_fpu): Ditto.
30724 2017-05-11 Julia Koval <julia.koval@intel.com>
30726 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30727 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30729 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30730 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30731 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30733 2017-05-11 Nathan Sidwell <nathan@acm.org>
30735 * graphite-poly.c: Include dumpfile.h.
30737 * dumpfle.h (dump_function): Declare here ...
30738 * tree-dump.h (dump_function): ... not here.
30739 * dumpfile.c: #include tree-cfg.h.
30740 (dump_function): Move here from ...
30741 * tree-dump.c (dump_function): ... here.
30742 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30743 * graphite-poly.c: Don't include tree-dump.h.
30744 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30745 * print-tree.c: Likewise.
30746 * stor-layout.c: Likewise.
30747 * tree-nested.c: Likewise.
30749 * dumpfile.c (dump_start): Use TDF_FLAGS.
30750 (dump_enable_all): Fix TDF_KIND check thinko.
30752 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30754 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30755 array entries to represent two legal parameterizations of the
30756 overloaded __builtin_cmpb function, as represented by the
30757 P6_OV_BUILTIN_CMPB constant.
30758 (altivec_resolve_overloaded_builtin): Add special case handling
30759 for the __builtin_cmpb function, as represented by the
30760 P6_OV_BUILTIN_CMPB constant.
30761 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30762 (BU_P6_64BIT_2): New macro.
30763 (BU_P6_OVERLOAD_2): New macro
30764 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30765 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30766 (CMPB): Add overload support to represent both 32-bit and 64-bit
30767 compare-bytes function.
30768 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30769 support for TARGET_CMPB.
30770 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30771 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30772 documentation of the __builtin_cmpb overloaded built-in function.
30774 2017-05-11 Richard Biener <rguenther@suse.de>
30776 PR tree-optimization/80705
30777 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30778 bases are not vectorizable.
30780 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30782 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30783 when counting register pressure.
30785 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30787 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30788 (iv_ca_more_deps): Renamed to ...
30789 (iv_ca_compare_deps): ... this.
30790 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30792 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30794 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30796 (determine_group_iv_costs): ... here.
30797 (find_inv_vars_cb): Record inv var if it's not recorded before.
30799 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30801 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30802 (get_shiftadd_cost): Ditto.
30804 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30806 * tree-ssa-address.c: Include header file.
30807 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30809 (add_to_parts): Refactor.
30810 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30811 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30814 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30816 PR tree-optimization/53090
30817 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30819 (extract_cond_operands): Detect condition with IV on both sides
30820 and return COMP_IV_EXPR_2.
30821 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30822 (rewrite_use_compare): Simplify by removing call to function
30823 extract_cond_operands.
30825 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30827 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30828 (extract_cond_operands): Detect condition comparing against non-
30829 invariant bound and return appropriate enum value.
30830 (find_interesting_uses_cond): Update use of extract_cond_operands.
30831 Handle its return value accordingly.
30832 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30834 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30836 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30837 nonlinear iv_use computation in loop invariant sensitive way.
30839 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30841 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30842 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30844 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30846 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30847 (dump_cand): Support iv_cand.inv_exprs.
30848 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30850 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30853 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30855 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30857 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30858 as local function. Include necessary header files.
30859 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30863 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30865 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30867 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30868 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30869 RSHIFT_EXPR and BIT_NOT_EXPR.
30871 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30873 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30874 (adjust_setup_cost): New parameter supporting round up adjustment.
30875 (struct address_cost_data): Delete.
30876 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30877 (split_address_cost, ptr_difference_cost): Delete.
30878 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30879 (struct ainc_cost_data): New struct.
30880 (get_address_cost_ainc): New function.
30881 (get_address_cost, get_computation_cost): Reimplement.
30882 (determine_group_iv_cost_address): Record inv_expr for all uses of
30884 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30885 (iv_ca_has_deps): Reimplemented to ...
30886 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30888 (iv_ca_extend): Call iv_ca_more_deps.
30890 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30892 * tree-ssa-address.c (struct mem_address): Move to header file.
30893 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30894 * tree-ssa-address.h (struct mem_address): Move from C file.
30895 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30897 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30899 * tree-affine.h (aff_combination_type): New interface.
30900 (aff_combination_zero_p): Remove static.
30901 (aff_combination_const_p): New interface.
30902 (aff_combination_singleton_var_p): New interfaces.
30904 2017-05-11 Richard Biener <rguenther@suse.de>
30906 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30907 Skip unreachable blocks and destinations.
30908 (eliminate): Move stmt removal and fixup ...
30909 (fini_eliminate): ... here. Skip inserted exprs.
30910 (pass_pre::execute): Move fini_pre after fini_eliminate.
30911 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30912 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30913 PRE to get rid of dead code that has invalid SSA form and
30914 split critical edges again.
30916 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30918 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30920 2017-05-11 Richard Biener <rguenther@suse.de>
30922 * passes.c (execute_function_todo): Verify loops if they are
30923 said to be up-to-date.
30924 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30925 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30927 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30930 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30931 handle calling assemble_external ourself.
30934 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30935 modes with zero size. Enhance comment.
30937 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30939 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30940 built-ins for vec_xl and vec_xst with short and char pointer
30943 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30945 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30946 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30947 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30948 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30949 (_mm_maskz_min_round_ss): New intrinsics.
30950 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30951 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30952 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30953 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30954 (__builtin_ia32_minss_mask_round): New builtins.
30955 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30956 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30957 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30959 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30960 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30962 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30965 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30967 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30968 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30969 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30970 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30971 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30972 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30973 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30974 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30975 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30976 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30977 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30978 (__builtin_ia32_mulss_mask_round): New builtins.
30979 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30980 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30981 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30983 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30984 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30986 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30989 2017-05-10 Julia Koval <julia.koval@intel.com>
30991 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30992 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30993 (_mm256_setr_m128i): New intrinsics.
30995 2017-05-10 Julia Koval <julia.koval@intel.com>
30997 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30998 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30999 (_mm_maskz_rcp14_ss): New intrinsics.
31000 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31001 (__builtin_ia32_rcp14ss_mask): New builtins.
31002 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31004 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31006 PR tree-optimization/51513
31007 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31008 (assert_unreachable_fallthru_edge_p): Use it.
31009 (group_case_labels_stmt): Likewise.
31010 * tree-cfg.h: Prototype it.
31011 * stmt.c: Include cfghooks.h and tree-cfg.h.
31012 (emit_case_dispatch_table) <gap_label>: New local variable.
31013 Use it to fill dispatch table gaps.
31014 Test for default_label before updating probabilities.
31015 (expand_case) <default_label>: Remove unneeded initialization.
31016 Test for unreachable default case statement and remove its edge.
31017 Set default_label accordingly.
31018 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31020 2017-05-10 Carl Love <cel@us.ibm.com>
31022 * config/rs6000/rs6000-c: Add support for built-in functions
31023 vector signed char vec_neg (vector signed char)
31024 vector signed short int vec_neg (vector short int)
31025 vector signed int vec_neg (vector signed int)
31026 vector signed long long vec_neg (vector signed long long)
31027 vector float vec_neg (vector float)
31028 vector double vec_neg (vector double)
31029 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31031 * config/rs6000/altivec.h: Add define for vec_neg
31032 * doc/extend.texi: Update the built-in documentation for the
31033 new built-in functions.
31035 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31037 PR tree-optimization/77644
31038 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31040 2017-05-10 Nathan Sidwell <nathan@acm.org>
31042 * dumpfile.h (TDI_lang_all): New.
31043 (TDF_KIND): New. Renumber others
31044 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31046 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31048 (get_dump_file_name): Adjust suffix generation.
31049 (dump_enable_all): Use TDF_KIND.
31050 * doc/invoke.texi (-fdump-lang-all): Document.
31052 * dumpfile.h: Tabify.
31054 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31057 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31058 Move member access before delete.
31060 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31062 * tree-inline.c (expand_call_inline): Split block at stmt
31065 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31068 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31069 are now unused after splitting mov{sf,sd}_hardfloat.
31070 (f32_lr2): Likewise.
31071 (f32_lm): Likewise.
31072 (f32_lm2): Likewise.
31073 (f32_li): Likewise.
31074 (f32_li2): Likewise.
31075 (f32_lv): Likewise.
31076 (f32_sr): Likewise.
31077 (f32_sr2): Likewise.
31078 (f32_sm): Likewise.
31079 (f32_sm2): Likewise.
31080 (f32_si): Likewise.
31081 (f32_si2): Likewise.
31082 (f32_sv): Likewise.
31083 (f32_dm): Likewise.
31084 (f32_vsx): Likewise.
31085 (f32_av): Likewise.
31086 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31087 For movsf, order stores so the VSX stores occur before the GPR
31088 store which encourages the register allocator to use a traditional
31089 FPR instead of a GPR. For movsd, order the stores so that the GPR
31090 store comes before the VSX stores to allow the power6 to work.
31091 This is due to the power6 not having a 32-bit integer store
31092 instruction from a FPR.
31093 (movsf_hardfloat): Likewise.
31094 (movsd_hardfloat): Likewise.
31096 2017-05-09 Martin Sebor <msebor@redhat.com>
31098 PR translation/80280
31099 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31102 PR translation/80280
31103 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31104 data member added in r247778.
31105 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31107 2017-05-09 Nathan Sidwell <nathan@acm.org>
31109 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31111 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31114 2017-05-09 Marek Polacek <polacek@redhat.com>
31116 * doc/invoke.texi: Fix typo.
31118 2017-05-09 Richard Biener <rguenther@suse.de>
31120 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31121 (vrp_val_is_min): Likewise.
31122 (set_value_range_to_value): Likewise.
31123 (set_value_range_to_nonnegative): Likewise.
31124 (gimple_assign_nonzero_p): Likewise.
31125 (gimple_stmt_nonzero_p): Likewise.
31126 (vrp_int_const_binop): Likewise. Remove unreachable case.
31127 (adjust_range_with_scev): Adjust comments.
31128 (compare_range_with_value): Likewise.
31129 (extract_range_from_phi_node): Likewise.
31130 (test_for_singularity): Likewise.
31132 2017-05-09 Richard Biener <rguenther@suse.de>
31134 * tree-vrp.c (get_single_symbol): Add assert that we don't
31135 get overflowed constants as invariant part.
31136 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31137 checks. Use wi::cmp instead of recursing for integer constants.
31138 (compare_values): Just ignore whether we assumed undefined
31139 overflow instead of failing the compare.
31140 (extract_range_for_var_from_comparison_expr): Add comment before the
31141 TREE_NO_WARNING sets.
31142 (test_for_singularity): Likewise.
31143 (extract_range_from_comparison): Do not disable optimization
31144 when we assumed undefined overflow.
31145 (extract_range_basic): Remove init of unused var.
31147 2017-05-09 Richard Biener <rguenther@suse.de>
31149 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31150 (extract_range_from_multiplicative_op_1): Adjust.
31151 (extract_range_from_binary_expr_1): Use int_const_binop.
31153 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31156 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31157 rs6000_store_data_bypass_p in seven define_bypass directives and
31158 in several comments.
31159 * config/rs6000/rs6000-protos.h: Add prototype for
31160 rs6000_store_data_bypass_p function.
31161 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31162 function implements slightly different (rs6000-specific) semantics
31163 than store_data_bypass_p, returning false rather than aborting
31164 with assertion error when arguments do not satisfy the
31165 requirements of store data bypass.
31166 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31167 rs6000_store_data_bypass_p.
31169 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31171 * config/xtensa/xtensa-protos.h
31172 (xtensa_initial_elimination_offset): New declaration.
31173 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31174 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31175 macro definition, add case for FRAME_POINTER_REGNUM when
31176 FRAME_GROWS_DOWNWARD.
31177 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31178 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31179 xtensa_initial_elimination_offset.
31181 2017-05-08 Nathan Sidwell <nathan@acm.org>
31183 * doc/invoke.texi: Alphabetize -fdump options.
31185 2017-05-08 Martin Sebor <msebor@redhat.com>
31187 PR translation/80280
31188 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31190 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31192 * target.def (compute_frame_layout): New optional target hook.
31193 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31194 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31195 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31197 * reload1.c (verify_initial_elim_offsets): Likewise.
31198 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31199 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31200 (arm_get_frame_offsets): Split up into this ...
31201 (arm_compute_frame_layout): ... and this function.
31203 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31205 * config/aarch64/constraints.md (Usa): New constraint.
31206 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31208 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31210 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31211 with_multilib_list after it has been checked.
31213 2017-05-08 Richard Biener <rguenther@suse.de>
31215 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31216 (bitmap_set_subtract_values): Likewise.
31218 2017-05-08 Richard Biener <rguenther@suse.de>
31220 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31221 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31223 (gimple_stmt_nonzero_warnv_p): Rename to ...
31224 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31226 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31227 (extract_range_basic): Adjust, do not disable propagation on
31228 strict overflow sensitive simplification.
31229 (vrp_visit_cond_stmt): Likewise.
31231 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31233 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31234 body size unconditionally.
31236 2017-05-07 Jeff Law <law@redhat.com>
31239 2017-05-06 Jeff Law <law@redhat.com>
31240 PR tree-optimization/78496
31241 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31244 PR tree-optimization/78496
31245 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31246 (simplify_stmt_using_ranges): Call it.
31247 (vrp_dom_walker::before_dom_children): Extract equivalences
31248 from an ASSERT_EXPR with an equality comparison against a
31251 2017-05-06 Jeff Law <law@redhat.com>
31253 PR tree-optimization/78496
31254 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31257 PR tree-optimization/78496
31258 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31259 (simplify_stmt_using_ranges): Call it.
31260 (vrp_dom_walker::before_dom_children): Extract equivalences
31261 from an ASSERT_EXPR with an equality comparison against a
31264 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31266 * lra-constraints.c (lra_copy_reg_equiv): New function.
31267 (split_reg): Use it to copy equivalence information from the
31268 original register to the spill register.
31270 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31272 PR rtl-optimization/75964
31273 * simplify-rtx.c (simplify_const_relational_operation): Remove
31274 invalid handling of comparisons of integer ABS.
31276 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31278 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31279 initialize to zero.
31280 (init_regs): Remove declaration.
31281 (function_arg_advance_32): Initialize error_p as boolean variable.
31283 2017-05-05 Nathan Sidwell <nathan@acm.org>
31285 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31286 lines. Use for (;;).
31288 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31290 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31291 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31292 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31293 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31294 VF=2 that require versioning.
31296 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31298 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31301 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31303 * diagnostic.h (diagnostic_override_option_index): Convert from
31304 macro to inline function.
31306 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31308 * diagnostic.c (last_module_changed_p): New function.
31309 (set_last_module): New function.
31310 (diagnostic_report_current_module): Convert macro usage to
31311 the above functions.
31312 * diagnostic.h (diagnostic_context::last_module): Strengthen
31313 from const line_map * to const line_map_ordinary *.
31314 (diagnostic_last_module_changed): Delete macro.
31315 (diagnostic_set_last_module): Delete macro.
31317 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31319 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31320 with diagnostic_report_diagnostic.
31321 (diagnostic_n_impl_richloc): Likewise.
31322 * diagnostic.h (report_diagnostic): Delete macro.
31323 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31324 with diagnostic_report_diagnostic.
31325 * substring-locations.c (format_warning_va): Likewise.
31327 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31329 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31330 save/restor of format_spec. Move option-printing code to...
31331 (print_option_information): ...this new function, and
31332 reimplement by simply printing to the pretty_printer,
31333 rather than appending to the format string.
31335 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31337 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31338 handling logic into...
31339 (update_effective_level_from_pragmas): ...this new function.
31341 2017-05-04 Andrew Waterman <andrew@sifive.com>
31343 * config/riscv/riscv.opt (mstrict-align): New option.
31344 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31345 (SLOW_UNALIGNED_ACCESS): Define.
31346 (riscv_slow_unaligned_access): Declare.
31347 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31349 (riscv_slow_unaligned_access): New variable.
31350 (rocket_tune_info): Set slow_unaligned_access to true.
31351 (optimize_size_tune_info): Set slow_unaligned_access to false.
31352 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31353 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31354 (riscv_option_override): Set riscv_slow_unaligned_access.
31355 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31357 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31359 * config/riscv/riscv.md: Unify indentation.
31361 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31366 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31368 (extendsi<mode>2): Add support for doing sign extension via
31369 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31370 don't have ISA 3.0 instructions.
31371 (extendsi<mode>2 splitter): Likewise.
31372 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31373 generate the normal insns since SImode can now go in vector
31374 registers. Disallow the special UNSPECs needed for previous
31375 machines to hide SImode being used. Add new insns
31376 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31377 (fix_trunc<mode>si2_stfiwx): Likewise.
31378 (fix_trunc<mode>si2_internal): Likewise.
31379 (fixuns_trunc<mode>si2): Likewise.
31380 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31381 (fctiw<u>z_<mode>_smallint): Likewise.
31382 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31383 of floating point to 32-bit integer from doing a direct move to
31384 the GPR registers to do a store.
31385 (fctiwz_<mode>): Break long line.
31387 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31389 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31390 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31391 (addr_list, addr_offset_valid_p): New.
31392 (split_address_groups): Check offset validity with above function.
31393 (gt-tree-ssa-loop-ivopts.h): Include header file.
31395 2017-05-05 Nathan Sidwell <nathan@acm.org>
31397 * config.gcc (arm*-*-*): Add missing 'fi'.
31399 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31401 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31402 included in -fopt-info does not matter.
31403 * doc/optinfo.texi (-fopt-info): Fix description of default
31404 behavour. Explicitly say order of options included in -fopt-info
31407 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31409 * config.gcc: Allow combinations of aprofile and rmprofile values for
31410 --with-multilib-list.
31411 * config/arm/t-multilib: New file.
31412 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31413 variables. Remove setting of ISA and floating-point ABI in
31414 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31415 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31416 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31417 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31419 * config/arm/t-rmprofile: Likewise except for the matches changes.
31420 * doc/install.texi (--with-multilib-list): Document the combination of
31421 aprofile and rmprofile values and warn about pitfalls in doing that.
31423 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31425 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31426 (movdi_aarch64): Likewise.
31428 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31430 PR tree-optimization/80632
31431 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31433 (build_arrays): Initialize it for virtual phis.
31434 (fix_phi_nodes): Use it for virtual phis.
31436 PR tree-optimization/80558
31437 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31438 [x, y] op z into [x op, y op z] for op & or | if conditions
31441 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31442 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31445 * config/arm/arm.md (use_literal_pool): Remove.
31446 (64-bit immediate split): No longer takes cost into consideration
31447 if arm_disable_literal_pool is enabled.
31448 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31449 used when arm_disable_literal_pool is enabled.
31450 (arm_max_const_double_inline_cost): Remove use of
31451 arm_disable_literal_pool.
31452 (push_minipool_fix): Add assert.
31453 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31454 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31455 (no_literal_pool_sf_immediate): New.
31457 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31459 PR tree-optimization/80613
31460 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31461 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31463 2017-05-05 Richard Biener <rguenther@suse.de>
31465 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31467 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31469 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31470 of this flag from insn conditions due to removal from r247495.
31472 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31474 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31476 (arm_early_store_addr_dep_ptr): Likewise.
31477 * config/arm/aarch-common-protos.h
31478 (arm_early_load_addr_dep_ptr): Add prototype.
31479 (arm_early_store_addr_dep_ptr): Likewise.
31480 * config/arm/cortex-a53.md: Add new bypasses.
31482 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31484 * tree.c (next_type_uid): Change type to unsigned.
31485 (type_hash_canon): Decrement back next_type_uid if
31486 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31487 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31490 2017-05-04 Martin Sebor <msebor@redhat.com>
31492 * builtins.c: Fix a trivial typo in a comment.
31494 PR middle-end/79234
31495 * builtins.c (check_sizes): Adjust to handle reading past the end.
31496 Avoid printing excessive upper bound of ranges. Use %E to print
31497 tree nodes instead of converting them to %wu.
31498 (expand_builtin_memchr): New function.
31499 (compute_dest_size): Rename...
31500 (compute_objsize): ...to this.
31501 (expand_builtin_memcpy): Adjust.
31502 (expand_builtin_mempcpy): Adjust.
31503 (expand_builtin_strcat): Adjust.
31504 (expand_builtin_strcpy): Adjust.
31505 (check_strncat_sizes): Adjust.
31506 (expand_builtin_strncat): Adjust.
31507 (expand_builtin_strncpy): Adjust and simplify.
31508 (expand_builtin_memset): Adjust.
31509 (expand_builtin_bzero): Adjust.
31510 (expand_builtin_memcmp): Adjust.
31511 (expand_builtin): Handle memcmp.
31512 (maybe_emit_chk_warning): Check strncat just once.
31514 2017-05-04 Martin Sebor <msebor@redhat.com>
31516 PR preprocessor/79214
31517 PR middle-end/79222
31518 PR middle-end/79223
31519 * builtins.c (check_sizes): Add inlining context and issue
31520 warnings even when -Wno-system-headers is set.
31521 (check_strncat_sizes): Same.
31522 (expand_builtin_strncat): Same.
31523 (expand_builtin_memmove): New function.
31524 (expand_builtin_stpncpy): Same.
31525 (expand_builtin): Handle memmove and stpncpy.
31527 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31529 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31530 which is not used any more.
31532 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31534 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31536 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31538 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31539 (cortexa53_tunings): Likewise.
31540 (cortexa57_tunings): Likewise.
31541 (cortexa72_tunings): Likewise.
31542 (cortexa73_tunings): Likewise.
31544 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31546 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31547 Set loop alignment to 8.
31549 2017-05-04 Martin Sebor <msebor@redhat.com>
31551 PR translation/80280
31552 * builtins.c (expand_builtin_object_size): Add missing quoting to
31553 %D and like directives.
31554 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31555 (hsa_type_for_tree_type): Same.
31556 (verify_function_arguments): Same.
31557 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31558 * varasm.c (get_section): Same.
31561 2017-05-04 Martin Sebor <msebor@redhat.com>
31563 PR translation/80280
31564 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31566 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31568 * config/aarch64/aarch64.c (generic_addrcost_table):
31569 Change HI/TI mode setting.
31571 2017-05-04 Martin Jambor <mjambor@suse.cz>
31573 PR tree-optimization/80622
31574 * tree-sra.c (comes_initialized_p): New function.
31575 (build_accesses_from_assign): Only set write lazily when
31576 comes_initialized_p is false.
31577 (analyze_access_subtree): Use comes_initialized_p.
31578 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31579 instead of testing for PARM_DECL.
31581 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31583 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31584 constraint on operand 0 to allow more general addressing modes.
31585 Adjust output template.
31586 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31588 * config/aarch64/aarch64-protos.h
31589 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31590 * config/aarch64/constraints.md (Dp): New address constraint.
31591 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31594 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31596 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31597 update use of estimate_ipcp_clone_size_and_time.
31598 (estimate_local_effects): Update use of
31599 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31600 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31601 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31602 Return nonspecialized time.
31604 2017-05-04 Richard Biener <rguenther@suse.de>
31606 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31607 for the last VUSE which def dominates the PHI. Directly call
31609 (get_continuation_for_phi_1): Remove.
31611 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31613 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31614 to explain the use of truncating division. Cap the number of
31615 iterations to the maximum given by nb_iterations_upper_bound,
31618 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31620 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31621 * configure: Regenerate.
31622 * config.in: Regenerate.
31623 * config/i386/driver-mingw32.c: new file.
31624 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31625 * config.host: Link driver-mingw32.o on MinGW host.
31626 * doc/install.texi: Document new --enable-mingw-wildcard configure
31629 2017-05-04 Marek Polacek <polacek@redhat.com>
31631 PR tree-optimization/80612
31632 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31634 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31635 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31637 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31638 (movt splitter): Likewise.
31639 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31640 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31641 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31642 block for Thumb-1 with MOVT.
31643 (thumb2_legitimate_address_p): Move code block ...
31644 (can_avoid_literal_pool_for_label_p): ... into this new function.
31645 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31647 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31648 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31649 "M-profile targets with the MOVT instruction".
31651 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31653 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31654 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31655 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31657 2017-05-04 Martin Liska <mliska@suse.cz>
31659 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31660 variable cond_code.
31662 2017-05-04 Richard Biener <rguenther@suse.de>
31664 * tree.c (array_at_struct_end_p): Handle arrays at struct
31665 end with flexarrays more conservatively. Refactor and treat
31666 arrays of arrays or aggregates more strict. Fix
31667 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31668 * tree.c (array_at_struct_end_p): Adjust prototype.
31669 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31670 * gimple-fold.c (get_range_strlen): Likewise.
31671 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31673 2017-05-04 Richard Biener <rguenther@suse.de>
31675 PR tree-optimization/31130
31676 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31678 (supports_overflow_infinity): Likewise.
31679 (is_negative_overflow_infinity): Likewise.
31680 (is_positive_overflow_infinity): Likewise.
31681 (is_overflow_infinity): Likewise.
31682 (stmt_overflow_infinity): Likewise.
31683 (overflow_infinity_range_p): Likewise.
31684 (usable_range_p): Remove as always returning true.
31685 (make_overflow_infinity): Remove.
31686 (negative_overflow_infinity): Likewise.
31687 (positive_overflow_infinity): Likewise.
31688 (avoid_overflow_infinity): Likewise.
31689 (set_value_range): Adjust accordingly.
31690 (set_value_range_to_nonnegative): Likewise, remove now unused
31691 overflow_infinity arg.
31692 (vrp_operand_equal_p): Adjust.
31693 (update_value_range): Likewise.
31694 (range_int_cst_singleton_p): Likewise.
31695 (operand_less_p): Likewise.
31696 (compare_values_warnv): Likewise.
31697 (extract_range_for_var_from_comparison_expr): Likewise.
31698 (vrp_int_const_binop): Likewise.
31699 (zero_nonzero_bits_from_vr): Likewise.
31700 (extract_range_from_multiplicative_op_1): Likewise.
31701 (extract_range_from_binary_expr_1): Likewise.
31702 (extract_range_from_unary_expr): Likewise.
31703 (extract_range_from_comparison): Likewise.
31704 (extract_range_basic): Likewise.
31705 (adjust_range_with_scev): Likewise.
31706 (compare_ranges): Likewise.
31707 (compare_range_with_value): Likewise.
31708 (dump_value_range): Likewise.
31709 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31711 (simplify_cond_using_ranges): Adjust.
31713 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31715 * brig-builtins.def: Added a builtin for class_f64.
31716 * builtin-types.def: Added a builtin type needed by class_f64.
31718 2017-05-03 Jason Merrill <jason@redhat.com>
31720 * timevar.def: Add TV_CONSTEXPR.
31722 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31724 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31726 2017-05-03 Martin Jambor <mjambor@suse.cz>
31728 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31729 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31730 * ipa-cp.c (ipcp_propagate_stage): Do not call
31731 ipa_update_after_lto_read.
31732 * ipa-inline.c (ipa_inline): Likewise.
31734 2017-05-03 Martin Jambor <mjambor@suse.cz>
31736 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31737 tag. Added a default constructor and a destructor.
31738 (ipa_edge_args_sum_t): New class;
31739 (ipa_edge_args_sum): Declare.
31740 (ipa_edge_args_vector): Remove declaration.
31741 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31742 (ipa_free_edge_args_substructures): Remove declaration.
31743 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31744 (ipa_edge_args_info_available_for_edge_p): Likewise.
31745 * ipa-prop.c (ipa_edge_args_vector): Removed.
31746 (edge_removal_hook_holder): Likewise.
31747 (edge_duplication_hook_holder): Likewise.
31748 (ipa_edge_args_sum): New variable.
31749 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31750 ipa_edge_args_vector.
31751 (ipa_free_edge_args_substructures): Likewise.
31752 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31753 ipa_edge_args_vector.
31754 (ipa_edge_removal_hook): Turned into method
31755 ipa_edge_args_sum_t::remove.
31756 (ipa_edge_duplication_hook): Turned into method
31757 ipa_edge_args_sum_t::duplicate.
31758 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31759 registering edge hooks.
31760 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31761 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31762 ipa_edge_args_sum instead of ipa_edge_args_vector.
31763 * ipa-profile.c (ipa_profile): Likewise.
31765 2017-05-03 Martin Jambor <mjambor@suse.cz>
31767 * symbol-summary.h (function_summary): New method exists.
31768 (function_summary::symtab_removal): Deallocate through release.
31769 (call_summary): New class.
31770 (gt_ggc_mx): New overload.
31771 (gt_pch_nx): Likewise.
31772 (gt_pch_nx): Likewise.
31774 2017-05-03 Jeff Law <law@redhat.com>
31776 PR tree-optimization/78496
31777 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31778 from simplify_cond_using_ranges. Split off code to walk
31779 backwards through casts into ...
31780 (simplify_cond_using_ranges_2): New function.
31781 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31782 (execute_vrp): After identifying jump threads, call
31783 simplify_cond_using_ranges_2.
31785 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31788 * ipa-inline.h (inline_summary): Add ctor.
31789 (create_ggc): Do not use ggc_cleared_alloc.
31791 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31792 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31794 * gcc.c (handle_braces): Support escaping in switch matching
31796 * doc/invoke.texi (Spec Files): Document it.
31797 Remove superfluous @code markup in items.
31799 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31801 * diagnostic-show-locus.c (struct column_range): New struct.
31802 (get_affected_columns): New function.
31803 (get_printed_columns): New function.
31804 (struct correction): New struct.
31805 (correction::ensure_capacity): New function.
31806 (correction::ensure_terminated): New function.
31807 (struct line_corrections): New struct.
31808 (line_corrections::~line_corrections): New dtor.
31809 (line_corrections::add_hint): New function.
31810 (layout::print_trailing_fixits): Reimplement in terms of the new
31812 (selftest::test_overlapped_fixit_printing): New function.
31813 (selftest::diagnostic_show_locus_c_tests): Call it.
31815 2017-05-03 Nathan Sidwell <nathan@acm.org>
31817 Canonicalize canonical type hashing
31818 * tree.h (type_hash_canon_hash): Declare.
31819 * tree.c (type_hash_list, attribute_hash_list): Move into
31820 type_hash_canon_hash.
31821 (build_type_attribute_qual_variant): Break out hash code calc into
31822 type_hash_canon_hash.
31823 (type_hash_canon_hash): New. Generic type hash computation.
31824 (build_range_type_1, build_array_type_1, build_function_type,
31825 build_method_type_directly, build_offset_type, build_complex_type,
31826 make_vector_type): Call it.
31828 2017-05-03 Richard Biener <rguenther@suse.de>
31830 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31831 When all DRs have unknown misaligned do not always peel
31832 when there is a store but apply the same costing model as if
31833 there were only loads.
31835 2017-05-03 Richard Biener <rguenther@suse.de>
31838 PR tree-optimization/80492
31839 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31840 compare_base_decls returning dont-know properly.
31842 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31844 * config/arm/iterators.md (CCSI): New mode iterator.
31845 (arch): New mode attribute.
31846 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31847 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31848 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31849 code iterator for success result mode.
31850 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31851 the corresponding new insn generators.
31853 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31856 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31857 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31859 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31861 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31862 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31863 (DDR_A): Wrap DDR argument in brackets.
31864 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31865 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31866 (DDR_REVERSED_P): Likewise.
31868 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31870 PR tree-optimization/79472
31871 * tree-switch-conversion.c (struct switch_conv_info): Add
31872 contiguous_range and default_case_nonstandard fields.
31873 (collect_switch_conv_info): Compute contiguous_range and
31874 default_case_nonstandard fields, don't clear final_bb if
31875 contiguous_range and only the default case doesn't have the required
31877 (check_all_empty_except_final): Set default_case_nonstandard instead
31878 of failing if contiguous_range and the default case doesn't have empty
31880 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31881 and only the default case doesn't have the required constants. Skip
31883 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31884 if default_case_nonstandard.
31885 (build_constructors): Build constant 1 just once. Assert that default
31886 values aren't inserted in between cases if contiguous_range. Skip
31888 (build_arrays): Skip virtual phis.
31889 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31890 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31891 Handle virtual phis.
31892 (gen_inbound_check): Handle default_case_nonstandard case.
31893 (process_switch): Adjust check_final_bb caller. Call
31894 gather_default_values with the first non-default case instead of
31895 default case if default_case_nonstandard.
31897 2017-05-02 Nathan Sidwell <nathan@acm.org>
31899 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31900 check. Fix formatting.
31902 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31904 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31905 errors when comparing specialized and unspecialized times.
31907 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31909 * diagnostic-show-locus.c
31910 (layout::should_print_annotation_line_p): Make private.
31911 (layout::print_annotation_line): Make private.
31912 (layout::annotation_line_showed_range_p): Make private.
31913 (layout::show_ruler): Make private.
31914 (layout::print_source_line): Make private. Pass in line and
31915 line_width, rather than calling location_get_source_line. Drop
31917 (layout::print_leading_fixits): New method.
31918 (layout::print_any_fixits): Rename to...
31919 (layout::print_trailing_fixits): ...this, and make private.
31920 Don't print newline fixits.
31921 (diagnostic_show_locus): Move logic for printing one row into...
31922 (layout::print_line): ...this new function. Move the
31923 location_get_source_line call and error-handling from
31924 print_source_line to here. Call print_leading_fixits, and rename
31925 print_any_fixits to print_trailing_fixits.
31926 (selftest::test_fixit_insert_containing_newline): Update now that
31927 newlines are partially supported.
31928 (selftest::test_fixit_insert_containing_newline_2): New test.
31929 (selftest::test_fixit_replace_containing_newline): Update comments.
31930 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31931 * edit-context.c (class added_line): New class.
31932 (class edited_line): Describe newline handling in comment.
31933 (edited_line::actually_edited_p): New method.
31934 (edited_line::print_content): Delete redundant decl.
31935 (edited_line::m_predecessors): New field.
31936 (edited_file::print_content): Call edited_line::print_content.
31937 (edited_file::print_diff): Update to support newlines.
31938 (edited_file::print_diff_hunk): Likewise.
31939 (edited_file::print_run_of_changed_lines): New function.
31940 (edited_file::print_diff_line): Convert to...
31941 (print_diff_line): ...this.
31942 (edited_file::get_effective_line_count): New function.
31943 (edited_line::edited_line): Initialize new field m_predecessors.
31944 (edited_line::~edited_line): Clean up m_predecessors.
31945 (edited_line::apply_fixit): Handle newlines.
31946 (edited_line::get_effective_line_count): New function.
31947 (edited_line::print_content): New function.
31948 (edited_line::print_diff_lines): New function.
31949 (selftest::test_applying_fixits_insert_containing_newline): New
31951 (selftest::test_applying_fixits_replace_containing_newline): New
31953 (selftest::insert_line): New function.
31954 (selftest::test_applying_fixits_multiple_lines): Add example of
31956 (selftest::edit_context_c_tests): Call the new tests.
31958 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31960 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31961 parameter cand. Update dump information.
31962 (get_computation_cost): Update uses.
31964 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31966 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31967 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31968 (get_computation_at, rewrite_use_address): Update use of
31969 get_computation_aff.
31971 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31973 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31974 (get_computation): Delete.
31975 (get_computation_cost): Implement like get_computation_cost_at.
31976 Use get_computation_at.
31977 (get_computation_cost_at): Delete.
31978 (rewrite_use_nonlinear_expr): Use get_computation_at.
31979 (rewrite_use_compare, remove_unused_ivs): Ditto.
31981 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31983 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31985 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31987 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31988 (ivopts_global_cost_for_size): Rename parameter and update uses.
31989 (iv_ca_recount_cost): Update uses.
31990 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31991 candidates seperately in n_invs and n_cands.
31992 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31994 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31996 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31997 (find_inv_vars_cb): New.
31998 (find_depends): Renamed to ...
31999 (find_inv_vars): ... this.
32000 (add_candidate_1, force_var_cost): Call find_inv_vars.
32001 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32003 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32005 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32006 inv_vars. Add inv_exprs.
32007 (struct iv_cand): Rename depends_on to inv_vars.
32008 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32009 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32010 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32011 (dump_cand): Dump inv_vars.
32012 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32013 (record_invariant, find_depends, add_candidate_1): Ditto.
32014 (set_group_iv_cost, force_var_cost): Ditto.
32015 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32016 (get_computation_cost_at, get_computation_cost): Ditto.
32017 (determine_group_iv_cost_generic): Ditto.
32018 (determine_group_iv_cost_address): Ditto.
32019 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32020 (determine_group_iv_costs): Ditto.
32021 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32022 (iv_ca_set_remove_invariants): Renamed to ...
32023 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32024 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32025 (iv_ca_set_add_invariants): Renamed to ...
32026 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32027 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32028 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32029 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32030 (create_new_ivs): Remove useless dump.
32032 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32034 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32036 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32037 (iv_ca_set_no_cp, create_new_iv): Ditto.
32039 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32041 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32043 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32045 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32046 function tree_check2.
32048 2017-05-02 Martin Liska <mliska@suse.cz>
32050 * doc/gcov.texi: Add missing preposition.
32051 * gcov.c (function_info::function_info): Properly fill up
32052 all member variables.
32054 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32056 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32058 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32060 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32062 2017-05-02 Martin Liska <mliska@suse.cz>
32065 * lto-streamer-in.c (lto_read_tree_1): Remove
32066 LTO_STREAMER_DEBUG.
32067 * lto-streamer.c (struct tree_hash_entry): Likewise.
32068 (struct tree_entry_hasher): Likewise.
32069 (tree_entry_hasher::hash): Likewise.
32070 (tree_entry_hasher::equal): Likewise.
32071 (lto_streamer_init): Likewise.
32072 (lto_orig_address_map): Likewise.
32073 (lto_orig_address_get): Likewise.
32074 (lto_orig_address_remove): Likewise.
32075 * lto-streamer.h: Likewise.
32076 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32077 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32079 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32081 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32082 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32083 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32084 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32085 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32086 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32087 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32088 (mm_maskz_sub_ss): New intrinsics.
32089 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32090 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32091 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32092 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32093 (__builtin_ia32_subss_mask_round): New builtins.
32094 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32095 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32096 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32098 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32099 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32101 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32104 2017-05-02 Martin Jambor <mjambor@suse.cz>
32106 PR tree-optimization/78687
32107 * tree-sra.c (access): New field parent.
32108 (process_subtree_disqualification): New function.
32109 (disqualify_candidate): Call it.
32110 (build_accesses_from_assign): Reset write flag if creating an
32112 (build_access_subtree): Fill in parent field and also prpagate
32113 down grp_write flag.
32114 (create_artificial_child_access): New parameter set_grp_write, set
32115 grp_write to its value.
32116 (propagate_subaccesses_across_link): Also propagate grp_write flag
32118 (propagate_all_subaccesses): Push the closest parent back to work
32119 queue if add_access_to_work_queue returned true.
32121 2017-05-02 Richard Biener <rguenther@suse.de>
32123 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32124 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32125 -fstrict-overflow documentation.
32126 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32127 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32128 flag_strict_overflow.
32129 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32130 * lto-opts.c (lto_write_options): Do not stream it.
32131 * lto-wrapper.c (merge_and_complain): Do not handle it.
32132 * opts.c (default_options_table): Do not set -fstrict-overflow.
32133 (finish_options): Likewise do not clear it when sanitizing.
32134 * simplify-rtx.c (simplify_const_relational_operation): Do not
32135 test flag_strict_overflow.
32137 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32139 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32140 using enabled attribute.
32141 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32142 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32143 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32144 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32145 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32146 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32147 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32148 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32149 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32150 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32152 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32154 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32156 2017-05-02 Richard Biener <rguenther@suse.de>
32158 PR tree-optimization/80591
32160 2017-04-10 Richard Biener <rguenther@suse.de>
32162 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32165 2017-05-02 Richard Biener <rguenther@suse.de>
32167 PR tree-optimization/80549
32168 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32169 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32170 headers if they do not have a preheader.
32172 2017-05-02 Martin Liska <mliska@suse.cz>
32175 * common.opt: Fix typo.
32176 * doc/invoke.texi: Likewise.
32178 2017-05-01 Jan Beulich <jbeulich@suse.com>
32180 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32181 swapping, add (x,x,m,x,n) alternative.
32183 2017-05-01 Nathan Sidwell <nathan@acm.org>
32185 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32186 unnecessary unadjusted_alignment check.
32188 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32191 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32193 * gimplify.c (gimplify_cilk_detach): New function.
32194 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32195 * tree-core.h: Document EXPR_CILK_SPAWN.
32196 * tree.h (EXPR_CILK_SPAWN): Define.
32198 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32200 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32201 to use new fixit_hint representation, using the "replace" logic.
32202 (get_line_span_for_fixit_hint): Likewise.
32203 (layout::print_any_fixits): Likewise.
32204 (selftest::test_one_liner_many_fixits): Rename to...
32205 (selftest::test_one_liner_many_fixits_1): ...this, and update
32206 comment and expected output to reflect that the multiple fix-it
32207 hints are now consolidated into one insertion.
32208 (selftest::test_one_liner_many_fixits_2): New test.
32209 (selftest::test_diagnostic_show_locus_one_liner): Update for
32211 (selftest::test_fixit_consolidation): Update for fix-it API
32213 * diagnostic.c (print_parseable_fixits): Likewise.
32214 * edit-context.c (edited_line::m_line_events): Convert from
32215 auto_vec <line_event *> to auto_vec <line_event>.
32216 (class line_event): Convert from abstract base class to a concrete
32217 class, taking over the role of replace_event.
32218 (class insert_event): Delete.
32219 (class replace_event): Rename to class line_event. Convert to
32221 (edit_context::add_fixits): Reimplement.
32222 (edit_context::apply_insert): Delete.
32223 (edit_context::apply_replace): Rename to...
32224 (edit_context::apply_fixit): ...this. Convert to half-open range.
32225 (edited_file::apply_insert): Delete.
32226 (edited_file::apply_replace): Rename to...
32227 (edited_file::apply_fixit): ...this.
32228 (edited_line::~edited_line): Drop deletion of events.
32229 (edited_line::apply_insert): Delete.
32230 (edited_line::apply_replace): Rename to...
32231 (edited_line::apply_fixit): ...this. Convert to half-open range.
32232 Update for change to type of m_line_events.
32233 * edit-context.h (edit_context::apply_insert): Delete.
32234 (edit_context::apply_replace): Rename to...
32235 (edit_context::apply_fixit): ...this.
32237 2017-05-01 Martin Sebor <msebor@redhat.com>
32239 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32242 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32245 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32246 __get_cpuid_max returns 0.
32247 (__get_cpuid_count): Ditto.
32249 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32251 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32252 replacement expression is another instance of one of its arguments.
32254 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32257 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32258 check for stack push/pop autoinc.
32259 * config/i386/i386.c (ix86_agi_dependent): Return false
32260 if the only reason why modified_in_p returned true is that
32261 addr is SP based and set_insn is a push or pop.
32263 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32265 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32268 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32271 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32272 (account_size_time): Use two predicates - exec_pred and
32274 (evaluate_conditions_for_known_args): Compute both clause and
32276 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32277 (inline_summary_t::duplicate): Update.
32278 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32280 (compute_inline_parameters): Likewise.
32281 (estimate_edge_size_and_time): Update caluclation of time.
32282 (estimate_node_size_and_time): Compute both time and nonspecialized
32284 (estimate_ipcp_clone_size_and_time): Update.
32285 (inline_merge_summary): Update.
32286 (do_estimate_edge_time): Update.
32287 (do_estimate_edge_size): Update.
32288 (do_estimate_edge_hints): Update.
32289 (inline_read_section, inline_write_summary): Stream both new predicates.
32290 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32292 (compute_inlined_call_time): Cleanup.
32293 (big_speedup_p): Update.
32294 (edge_badness): Update.
32295 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32296 (size_time_entry): Replace predicate by exec_predicate and
32297 nonconst_predicate.
32298 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32299 (estimate_edge_time): Return also nonspec_time.
32300 (reset_edge_growth_cache): Update.
32302 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32304 PR rtl-optimization/80491
32305 * ifcvt.c (noce_process_if_block): When looking for x setter
32306 with missing else_bb, don't check only the insn right before
32307 cond_earliest, but look for the last insn that x is modified in
32308 within the same bb.
32310 PR rtl-optimization/80491
32311 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32313 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32315 PR tree-optimization/80487
32316 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32318 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32320 PR tree-optimization/79697
32321 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32322 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32323 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32325 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32326 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32328 2017-04-28 Martin Sebor <msebor@redhat.com>
32330 PR tree-optimization/80523
32331 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32332 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32334 (maybe_warn, format_directive, parse_directive): Use new functions.
32335 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32337 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32339 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32341 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32343 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32344 target_header_dir): Set correctly.
32345 * configure: Regenerated.
32346 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32347 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32348 instead of SYSTEM_HEADER_DIR.
32350 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32352 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32353 (estimate_local_effects): Likewise.
32354 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32355 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32356 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32357 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32358 do_estimate_edge_time, estimate_edge_time): Likewise.
32359 * ipa-inline-analysis.c (estimate_node_size_and_time,
32360 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32361 (estimate_time_after_inlining): Remove.
32363 2017-04-28 Martin Liska <mliska@suse.cz>
32365 * doc/gcov.texi: Enhance documentation of gcov.
32367 2017-04-28 Martin Liska <mliska@suse.cz>
32369 * doc/gcov.texi: Sort options in alphabetic order.
32370 * doc/gcov-dump.texi: Likewise.
32371 * doc/gcov-tool.texi: Likewise.
32372 * gcov.c (print_usage): Likewise.
32373 * gcov-dump.c (print_usage): Likewise.
32374 * gcov-tool.c (print_merge_usage_message): Likewise.
32375 (print_rewrite_usage_message): Likewise.
32376 (print_overlap_usage_message): Likewise.
32378 2017-04-28 Martin Liska <mliska@suse.cz>
32380 PR gcov-profile/53915
32381 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32383 2017-04-28 Martin Liska <mliska@suse.cz>
32385 PR gcov-profile/79891
32386 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32387 is marked by compiler as living on a line.
32388 (get_cycles_count): Remove usage of the union.
32389 (output_intermediate_file): Likewise.
32390 (find_source): Fix GNU coding style.
32391 (accumulate_line_counts): Remove old non-all block mode.
32392 (output_lines): Remove usage of the union.
32393 * profile.c (output_location): Include all BBs, even if
32394 belonging to a same line (and file) as a previous BB.
32396 2017-04-28 Martin Liska <mliska@suse.cz>
32398 * gcov.c (process_args): Handle new argument 'w'.
32399 (read_graph_file): Assign ID to BBs.
32400 (output_branch_count): Display BB # if verbose flag is set.
32401 (output_lines): Likewise for arcs.
32402 (print_usage): Add '--verbose' option help.
32403 * doc/gcov.texi: Document --verbose (-w) option.
32405 2017-04-28 Martin Liska <mliska@suse.cz>
32407 * gcov.c (struct block_location_info): New struct.
32408 (process_file): Fill up the new structure.
32409 (read_graph_file): Replace usage of encoding by the newly added
32411 (add_line_counts): Likewise.
32412 (accumulate_line_counts): Remove usage of the union.
32413 (function_info::function_info): New function.
32414 (function_info::~function_info): Likewise.
32415 (process_file): Call delete instead of release_function.
32416 (release_function): Release the function.
32417 (release_structures): Call delete instead of release_function.
32418 (solve_flow_graph): Replace usage of num_blocks.
32419 (find_exception_blocks): Likewise.
32420 (output_lines): Fix GNU coding style.
32422 2017-04-28 Martin Liska <mliska@suse.cz>
32425 * coverage.c (coverage_remove_note_file): New function.
32426 * coverage.h: Declare the function.
32427 * toplev.c (finalize): Clean if an error has been seen.
32429 2017-04-28 Martin Liska <mliska@suse.cz>
32431 PR gcov-profile/80031
32432 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32433 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32434 * gcov.c (read_graph_file): Read just number of blocks.
32435 * profile.c (branch_prob): Do not stream 0 flags per a basic
32438 2017-04-28 Martin Liska <mliska@suse.cz>
32440 * gcov-dump.c (tag_*): Add new argument to declarations.
32441 (dump_gcov_file): Likewise.
32442 (tag_blocks): Add and use new argument depth.
32443 (tag_arcs): Likewise.
32444 (tag_lines): Likewise.
32445 (tag_counters): Likewise.
32446 (tag_summary): Likewise.
32447 (dump_working_sets): Use depth to do a proper indentation.
32449 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32452 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32453 * symtab.c (symtab_node::debug_symtab): Move definition here.
32455 2017-04-28 Richard Biener <rguenther@suse.de>
32457 * lto-streamer.h (LTO_major_version): Bump to 7.
32459 2017-04-28 Richard Biener <rguenther@suse.de>
32461 * tree-vrp.c (assert_info): New struct.
32462 (add_assert_info): New helper.
32463 (register_edge_assert_for_2): Refactor to add asserts to a vector
32465 (register_edge_assert_for_1): Likewise.
32466 (register_edge_assert_for): Likewise.
32467 (finish_register_edge_assert_for): New helper actually registering
32468 asserts where live on edge.
32469 (find_conditional_asserts): Adjust.
32470 (find_switch_asserts): Likewise.
32471 (evrp_dom_walker::try_find_new_range): Generalize.
32472 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32474 2017-04-27 Marek Polacek <polacek@redhat.com>
32477 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32478 arg10 and arg11 to itype.
32480 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32482 * doc/extend.texi (Object Size Checking): Improve grammar.
32484 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32487 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32488 that the logic for permitting reciprocal estimates matches that
32491 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32494 * tree.c (type_cache_hasher::equal): Only compare
32495 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32496 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32497 non-aggregate element types.
32498 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32499 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32501 2017-04-27 Richard Biener <rguenther@suse.de>
32503 PR middle-end/80533
32504 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32505 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32506 keeping a reference to a trailing array.
32508 2017-04-27 Richard Biener <rguenther@suse.de>
32510 PR middle-end/80539
32511 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32512 being in loop-closed SSA form conservatively.
32513 (chrec_fold_multiply_poly_poly): Likewise.
32515 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32517 PR middle-end/79665
32518 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32519 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32521 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32524 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32525 (aarch64_function_arg_alignment): Return unsigned int again, but still
32526 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32527 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32528 Don't emit -Wpsabi note.
32529 (aarch64_function_arg_boundary): Likewise.
32530 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32533 2017-04-26 Nathan Sidwell <nathan@acm.org>
32535 * tree.h (crc32_unsigned_n): Declare.
32536 (crc32_unsigned, crc32_unsigned): Make inline.
32537 * tree.c (crc32_unsigned_bits): Replace with ...
32538 (crc32_unsigned_n): ... this.
32539 (crc32_unsigned, crc32_byte): Remove.
32540 (crc32_string): Remove unnecessary braces.
32542 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32544 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32545 * ipa-inline-analysis.c (MAX_TIME): Remove.
32546 (account_size_time): Use sreal for time.
32547 (dump_inline_summary): Update.
32548 (estimate_function_body_sizes): Update.
32549 (estimate_edge_size_and_time): Update.
32550 (estimate_calls_size_and_time): Update.
32551 (estimate_node_size_and_time): Update.
32552 (inline_merge_summary): Update.
32553 (inline_update_overall_summary): Update.
32554 (estimate_time_after_inlining): Update.
32555 (inline_read_section): Update.
32556 (inline_write_summary): Update.
32557 * ipa-inline.c (compute_uninlined_call_time): Update.
32558 (compute_inlined_call_time): Update.
32559 (recursive_inlining): Update.
32560 (inline_small_functions): Update.
32561 (dump_overall_stats): Update.
32562 * ipa-inline.h: Include sreal.h.
32563 (size_time_entry): Turn time to sreal.
32564 (inline_summary): Turn self_time nad time to sreal.
32566 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32568 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32570 (sreal::stream_out, sreal::stream_in): New.
32571 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32573 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32575 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32578 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32581 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32583 (dimode_scalar_chain::compute_convert_gain): Ditto.
32584 (dimode_scalar_chain::make_vector_copies): Ditto.
32585 (dimode_scalar_chain::convert_reg): Ditto.
32586 (dimode_scalar_chain::convert_insn): Ditto.
32587 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32588 (VI248_AVX512BW_1): New mode iterator.
32589 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32590 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32593 2017-04-25 Martin Sebor <msebor@redhat.com>
32595 PR tree-optimization/80497
32596 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32597 constants are representable in HOST_WIDE_INT.
32598 (parse_directive): Ditto.
32600 2017-04-25 Martin Sebor <msebor@redhat.com>
32603 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32604 (new_zero_array): Adjust signature.
32605 (dom_info::dom_init): Used unsigned rather that size_t.
32606 (dom_info::dom_info): Same.
32608 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32609 Jakub Jelinek <jakub@redhat.com>
32612 * config/arm/arm.c: Include gimple.h.
32613 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32614 returns negative, increment ncrn only if it returned positive.
32615 (arm_needs_doubleword_align): Return int instead of bool,
32616 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32617 members, but if there is any such non-FIELD_DECL
32618 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32619 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32620 returns negative, increment nregs only if it returned positive.
32621 (arm_setup_incoming_varargs): Likewise.
32622 (arm_function_arg_boundary): Emit -Wpsabi note if
32623 arm_needs_doubleword_align returns negative, return
32624 DOUBLEWORD_ALIGNMENT only if it returned positive.
32626 2017-04-25 Marek Polacek <polacek@redhat.com>
32629 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32630 first argument to type.
32632 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32635 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32636 type checks to test for compatibility instead of equality.
32638 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32639 Jakub Jelinek <jakub@redhat.com>
32642 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32644 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32645 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32646 the alignment computation, but return their maximum in warn_alignment.
32647 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32648 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32650 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32651 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32654 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32656 * config/arc/simdext.md (dmpyh): Fix typo.
32658 2017-04-25 Richard Biener <rguenther@suse.de>
32660 PR tree-optimization/80492
32661 * alias.c (compare_base_decls): Handle registers with asm
32662 specification conservatively.
32663 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32664 compare_base_decls returning dont-know properly.
32666 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32668 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32669 (legitimate_offset_address_p): New function.
32670 (arc_legitimate_address_p): Use above function.
32672 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32674 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32676 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32678 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32679 ACCH registers whenever they are available.
32681 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32683 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32684 double regs fix when not used.
32686 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32688 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32690 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32691 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32693 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32695 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32696 when emitting short ADD instructions.
32698 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32700 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32702 (cmpsi_cc_c_insn): Likewise.
32703 (cbranchsi4_scratch): Compute proper instruction length using
32704 compact_hreg_operand.
32705 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32707 2017-04-25 Richard Biener <rguenther@suse.de>
32709 PR middle-end/80509
32710 * passes.c (pass_manager::pass_manager): Initialize
32711 m_name_to_pass_map.
32713 2017-04-25 Richard Biener <rguenther@suse.de>
32715 PR tree-optimization/79201
32716 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32718 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32721 * config/s390/vector.md: Split MEM->GPR vector moves for
32722 non-s_operand addresses.
32724 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32727 * config/s390/predicates.md (reload_const_wide_int_operand): New
32729 * config/s390/s390.md ("movti"): Remove d/P alternative.
32730 ("movti_bigconst"): New pattern definition.
32732 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32735 * s390-protos.h (s390_expand_cs_hqi): Removed.
32736 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32737 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32738 modes as well as CCZ1mode and CCZmode.
32739 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32740 signature of s390_emit_compare_and_swap.
32741 (s390_expand_cs_hqi): Likewise, make static.
32742 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32743 compare-and-swap, in some cases.
32744 (s390_expand_cs): Wrapper function.
32745 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32747 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32748 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32749 patterns for small and large integers. Forbid symref memory operands.
32750 Move expander to s390.c. Require cc register.
32751 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32752 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32753 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32754 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32755 symref memory operands. Remove CC mode and call s390_match_ccmode
32757 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32759 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32761 * config/s390/s390.md (define_peephole2): New peephole to help
32762 combining the load-and-test pattern with volatile memory.
32764 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32766 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32767 with CCZmode for TARGET_Z196.
32769 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32771 PR rtl-optimization/80501
32772 * combine.c (make_compound_operation_int): Set subreg_code to SET
32773 even for AND with mask of the sign bit of mode.
32775 PR rtl-optimization/80500
32776 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32777 sum's initial value.
32779 2017-04-25 Julian Brown <julian@codesourcery.com>
32780 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32782 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32784 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32786 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32788 2017-04-25 Julian Brown <julian@codesourcery.com>
32789 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32791 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32792 (thunderx2t99_sha): New Reservation.
32794 2017-04-25 Julian Brown <julian@codesourcery.com>
32795 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32797 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32798 type for 1-element load.
32800 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32802 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32804 2017-04-24 Martin Jambor <mjambor@suse.cz>
32806 PR tree-optimization/80293
32807 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32808 char arrays not totally scalarizable if it is false.
32809 (analyze_all_variable_accesses): Pass correct value in the new
32810 parameter. Add a statistics counter.
32812 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32814 PR middle-end/79931
32815 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32817 2017-04-24 Richard Biener <rguenther@suse.de>
32819 PR tree-optimization/80494
32820 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32821 out for complex types.
32823 2017-04-24 Richard Biener <rguenther@suse.de>
32825 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32826 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32827 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32828 (DFS): Adjust and never fail.
32829 (sccvn_dom_walker::fail): Remove.
32830 (sccvn_dom_walker::before_dom_children): Adjust.
32831 (run_scc_vn): Likewise and never fail.
32832 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32833 (pass_fre::execute): Likewise.
32835 2017-04-24 Richard Biener <rguenther@suse.de>
32837 PR tree-optimization/79725
32838 * tree-ssa-sink.c (statement_sink_location): Return whether
32839 failure reason was zero uses. Move that check later.
32840 (sink_code_in_bb): Deal with zero uses by removing the stmt
32843 2017-04-24 Richard Biener <rguenther@suse.de>
32846 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32847 pointer-based references.
32849 2017-04-24 Richard Biener <rguenther@suse.de>
32852 * pass_manager.h (pass_manager::operator new): Remove.
32853 (pass_manager::operator delete): Likewise.
32854 * passes.c (pass_manager::operator new): Remove.
32855 (pass_manager::operator delete): Likewise.
32856 (pass_manager::pass_manager): Zero individual pass members.
32858 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32861 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32862 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32863 Check "XEXP (src, 1)" operand here.
32864 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32865 Check "XEXP (src, 1)" operand here.
32866 (dimode_scalar_chain::make_vector_copies): Detect count register
32867 of a shift instruction. Zero extend count register from QImode
32868 to DImode to satisfy vector shift pattern count operand predicate.
32869 Substitute vector shift count operand with a DImode copy.
32870 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32873 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32875 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32876 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32877 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32878 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32879 (UNSPEC_NOREX_MEM): Remove definition.
32881 2017-04-21 Richard Biener <rguenther@suse.de>
32883 PR tree-optimization/79547
32884 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32885 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32886 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32887 without any constraints.
32889 2017-04-21 Richard Biener <rguenther@suse.de>
32891 PR tree-optimization/78847
32892 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32894 2017-04-21 Richard Biener <rguenther@suse.de>
32896 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32897 (build_distinct_type_copy): Likewise.
32898 (build_variant_type_copy): Likewise.
32899 * tree.c (build_qualified_type): Pass down mem-stat info.
32900 (build_distinct_type_copy): Likewise.
32901 (build_variant_type_copy): Likewise.
32903 2017-04-21 Richard Biener <rguenther@suse.de>
32905 PR tree-optimization/80237
32906 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32908 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32909 for a simplified result.
32911 2016-04-21 Richard Biener <rguenther@suse.de>
32913 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32914 sth as strict as a simple_iv but a chrec without symbols and an
32915 operand defined in the loop we are peeling (and not some subloop).
32916 (propagate_constants_for_unrolling): Propagate all constants.
32918 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32921 * config/i386/i386.c (print_reg): Remove assert for disalowed
32922 regno values, call output_operand_lossage instead.
32924 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32927 * config/i386/constraints.md (Yc): New register constraint.
32928 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32929 Use Yc constraint for alternative 2 of operand 0. Remove
32930 preferred_for_speed attribute.
32932 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32934 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32935 lastprivate clauses in SIMT case.
32937 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32939 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32941 2017-04-20 Richard Biener <rguenther@suse.de>
32943 PR tree-optimization/57796
32944 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32945 as N scalar stores.
32946 (vect_model_load_cost): Cost gathers as N scalar loads.
32948 2017-04-20 Richard Biener <rguenther@suse.de>
32950 * ggc-page.c (ggc_allocated_p): Rename to ...
32951 (safe_lookup_page_table_entry): ... this and return the lookup
32953 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32955 2017-04-20 Richard Biener <rguenther@suse.de>
32957 PR tree-optimization/80453
32958 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32959 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32960 from the conditions.
32961 (vn_phi_eq): Pass them down.
32962 (vn_phi_lookup): Record them.
32963 (vn_phi_insert): Likewise.
32965 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32967 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32968 uninitialized variable warning to avoid buffer overrun.
32970 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32973 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32974 is suppressed for '{ 0 }' in C.
32976 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32978 * BASE-VER: Set to 8.0.0.
32980 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32982 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32983 priority .init_array and .fini_array section with SECTION_NOTYPE
32986 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32988 PR middle-end/80423
32989 * tree.h (build_array_type): Add typeless_storage default argument.
32990 * tree.c (type_cache_hasher::equal): Also compare
32991 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32992 (build_array_type): Add typeless_storage argument, set
32993 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32995 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32996 (build_array_type): Likewise. Add typeless_storage argument.
32998 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32999 Jakub Jelinek <jakub@redhat.com>
33001 PR tree-optimization/80426
33002 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33003 operation on symbolic operands, also compute the overflow for the
33004 invariant part when the operation degenerates into a negation.
33006 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33009 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33010 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33013 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33015 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33018 * config/avr/avr.c (tree.h): Include it.
33019 (cgraph.h): Include it.
33020 (avr_encode_section_info): Don't warn for uninitialized progmem
33021 variable if it's just an alias.
33023 2017-04-19 Richard Biener <rguenther@suse.de>
33026 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33027 when needed by AutoPGO.
33029 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33032 * doc/lto.texi: Remove an extra 'that'.
33034 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33036 PR rtl-optimization/80429
33037 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33038 are only used in debug insns.
33040 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33041 Vladimir Makarov <vmakarov@redhat.com>
33043 * config/sparc/predicates.md (input_operand): Add comment. Return
33044 true for any memory operand when LRA is in progress.
33045 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33047 2017-04-18 Jeff Law <law@redhat.com>
33050 * mips.md ({return,simple_return}_internal): Do not overwrite
33053 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33055 PR tree-optimization/80443
33056 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33057 instead of adding 1, subtract -1 and similarly instead of subtracting
33060 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33062 PR rtl-optimization/80357
33063 * haifa-sched.c (tmp_bitmap): New variable.
33064 (model_recompute): Handle duplicate use records.
33065 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33066 (free_global_sched_pressure_data): Free it.
33068 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33071 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33072 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33073 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33074 instead of SYSTEM_HEADER_DIR.
33076 2017-04-18 Jeff Law <law@redhat.com>
33078 PR middle-end/80422
33079 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33080 predecessors after walking up the insn chain.
33082 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33085 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33086 sizetype type into debug info.
33088 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33091 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33092 unneeded test for TARGET_UPPER_REGS_SF.
33093 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33095 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33098 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33099 instead of gsi_after_labels.
33101 2017-04-18 Jeff Law <law@redhat.com>
33103 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33107 2017-04-13 Jeff Law <law@redhat.com>
33108 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33109 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33111 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33114 * config/avr/avr.c (intl.h): Include it.
33115 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33117 2017-04-18 Martin Liska <mliska@suse.cz>
33119 PR gcov-profile/78783
33120 * gcov-tool.c (gcov_output_files): Validate that destination
33121 file is either removed by the tool or by a user.
33123 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33124 Guy Benyei <guybe@mellanox.com>
33126 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33127 block, and do not negate it, the stored id is already negative.
33129 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33131 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33133 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33136 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33137 masks of options that should be turned off if the VSX vector
33138 options are turned off.
33139 (OTHER_P8_VECTOR_MASKS): Likewise.
33140 (OTHER_VSX_VECTOR_MASKS): Likewise.
33141 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33142 rs6000_disable_incompatible_switches to validate no type switches
33144 (rs6000_incompatible_switch): New function to disallow turning on
33145 other vector options if -mno-vsx, -mno-power8-vector, or
33146 -mno-power9-vector are specified.
33148 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33150 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33152 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33154 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33155 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33156 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33157 (ARG_POINTER_CFA_OFFSET): Likewise.
33159 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33161 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33162 conditions to take advantage of various optimizations.
33164 2017-04-13 Jeff Law <law@redhat.com>
33166 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33167 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33168 (zero_extendsidi2_dext): Likewise.
33170 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33173 * fold-const.c (fold_ternary_loc): Revert
33174 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33177 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33179 PR rtl-optimization/80343
33180 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33181 new scratch pseudo.
33183 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33186 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33187 to ubsan_encode_value.
33189 2017-04-13 Jeff Law <law@redhat.com>
33191 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33192 appearing in DEBUG_INSNs.
33194 2017-04-13 Martin Liska <mliska@suse.cz>
33196 PR gcov-profile/80413
33197 * gcov-io.c (gcov_write_string): Copy to buffer just when
33198 allocated size is greater than zero.
33200 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33203 * dwarf2out.c (decls_for_scope): Ignore declarations of
33204 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33206 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33209 * ipa-visibility.c (non_local_p): Fix typos.
33210 (localize_node): When localizing symbol in same comdat group,
33211 dissolve the group only when we know external symbols are going
33213 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33215 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33217 PR tree-optimization/79390
33218 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33219 order does not result in usable sequence, retry with reversed operand
33225 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33226 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33227 op0 instead of fold_convert_loc (loc, type, arg0).
33229 2017-04-12 Jeff Law <law@redhat.com>
33231 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33232 has a delay slot in the generated code.
33234 * config/cris/cris.md (cris_preferred_reload_class): Return
33235 GENNONACR_REGS rather than GENERAL_REGS.
33237 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33240 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33241 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33242 signedness of the result type.
33244 2017-04-12 Richard Biener <rguenther@suse.de>
33245 Jeff Law <law@redhat.com>
33247 PR tree-optimization/80359
33248 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33249 trim stores to TARGET_MEM_REFs.
33251 2017-04-12 Richard Biener <rguenther@suse.de>
33253 PR tree-optimization/79390
33254 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33255 threading case even more.
33257 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33260 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33261 for quad_address_p for TImode, instead of just not indexed_address.
33263 2017-04-12 Richard Biener <rguenther@suse.de>
33264 Bernd Edlinger <bernd.edlinger@hotmail.de>
33266 PR middle-end/79671
33267 * alias.c (component_uses_parent_alias_set_from): Handle
33268 TYPE_TYPELESS_STORAGE.
33269 (get_alias_set): Likewise.
33270 * tree-core.h (tree_type_common): Add typeless_storage flag.
33271 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33272 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33273 for types containing members with TYPE_TYPELESS_STORAGE.
33274 (place_field): Likewise.
33275 (layout_type): Likewise for ARRAY_TYPE.
33276 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33277 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33278 TYPE_TYPELESS_STORAGE.
33279 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33281 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33284 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33285 first argument to type.
33287 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33291 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33292 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33293 (rs6000_expand_binop_builtin): Likewise.
33294 (rs6000_expand_ternop_builtin): Likewise; also add missing
33295 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33296 vshasigma built-ins.
33297 * doc/extend.texi: Document that vec_xxpermdi's third argument
33298 must be a constant.
33300 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33302 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33303 Use shift_const cost parameter when calculating gain of STV shifts.
33305 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33307 PR rtl-optimization/70478
33308 * lra-constraints.c (process_alt_operands): Check memory for
33309 disfavoring memory insn operand.
33311 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33313 PR middle-end/80100
33314 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33315 left shift in unsigned HOST_WIDE_INT type.
33317 PR rtl-optimization/80385
33318 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33319 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33322 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33323 if they have any depend clauses.
33325 2017-04-11 Martin Liska <mliska@suse.cz>
33328 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33329 * ipa-split.c (split_function): Create a local comdat symbol
33330 if caller is in a comdat group.
33332 2017-04-11 Martin Liska <mliska@suse.cz>
33335 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33338 2017-04-11 Martin Sebor <msebor@redhat.com>
33340 PR middle-end/80364
33341 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33342 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33344 (directive::set_width, directive::set_precision, format_character):
33346 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33349 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33352 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33353 conflict, set target->arch_name instead of target->cpu_name.
33355 2017-04-11 Richard Biener <rguenther@suse.de>
33357 PR tree-optimization/80374
33358 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33359 build_zero_cst, remove fold_convertible_p check again.
33361 2017-04-11 Martin Liska <mliska@suse.cz>
33364 * ubsan.c (instrument_object_size): Do not instrument register
33367 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33370 * config/i386/i386-builtin-types.def
33371 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33372 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33373 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33374 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33375 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33376 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33377 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33378 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33379 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33380 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33381 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33382 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33383 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33384 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33385 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33386 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33387 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33388 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33389 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33390 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33391 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33392 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33393 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33394 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33395 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33396 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33397 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33398 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33399 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33400 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33401 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33402 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33403 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33404 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33405 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33406 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33407 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33408 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33409 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33410 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33411 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33412 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33413 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33414 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33415 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33416 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33418 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33419 flag to second_arg_count, handle 4 argument function type _COUNT
33420 aliases, handle second_arg_count on second argument rather than last.
33422 2017-04-10 Jeff Law <law@redhat.com>
33424 PR tree-optimization/80374
33425 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33426 record anything if we can not convert integer_zero_node to the
33429 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33432 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33433 Enhance special handling given to the TARGET_P9_MINMAX option in
33434 relation to certain other options.
33436 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33438 PR tree-optimization/80153
33439 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33440 remove POINTER_PLUS_EXPR's base part directly, rather than through
33443 2017-04-10 Richard Biener <rguenther@suse.de>
33444 Bin Cheng <bin.cheng@arm.com>
33446 PR tree-optimization/80153
33447 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33448 the first element of pointer type aff_tree. Build result expr in
33450 (add_elt_to_tree): Convert to type unconditionally. Remove other
33451 fold_convert calls.
33452 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33453 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33455 2017-04-10 Richard Biener <rguenther@suse.de>
33457 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33460 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33462 PR rtl-optimization/70478
33463 * lra-constraints.c (curr_small_class_check): New.
33464 (update_and_check_small_class_inputs): New.
33465 (process_alt_operands): Update curr_small_class_check. Disfavor
33466 alternative insn memory operands. Check available regs for small
33469 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33472 * config/mips/mips.opt (-mvirt): Update description.
33473 * doc/invoke.texi (-mvirt): Likewise.
33475 2017-04-10 Richard Biener <rguenther@suse.de>
33477 PR middle-end/80362
33478 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33479 looking for NEGATE_EXPR in -A / -B to A / B folding.
33481 2017-04-10 Martin Liska <mliska@suse.cz>
33483 PR gcov-profile/80224
33484 * gcov.c (print_usage): Fix usage string.
33485 (get_gcov_intermediate_filename): Remove.
33486 (output_gcov_file): Use both for normal and intermediate format.
33487 (generate_results): Do not initialize special file for
33488 intermediate format.
33490 2017-04-10 Richard Biener <rguenther@suse.de>
33492 PR tree-optimization/80304
33493 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33496 2017-04-10 Nathan Sidwell <nathan@acm.org>
33499 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33500 (rs6000_init_builtins): Use it.
33502 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33504 * config/arm/arm.md (<mrc>): Add mode to SET source.
33505 (<mrrc>): Likewise.
33507 2017-04-10 Richard Biener <rguenther@suse.de>
33509 PR middle-end/80344
33510 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33512 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33515 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33516 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33517 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33518 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33519 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33520 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33521 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33522 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33523 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33524 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33525 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33526 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33527 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33528 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33529 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33530 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33531 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33532 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33533 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33534 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33535 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33536 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33537 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33539 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33541 PR rtl-optimization/70478
33542 * lra-constraints.c: Reverse the last patch.
33544 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33546 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33547 Add comment for WCHAR_T.
33549 2017-04-08 Martin Liska <mliska@suse.cz>
33552 2017-04-07 Martin Liska <mliska@suse.cz>
33555 * ipa-split.c (split_function): Add function part to a same comdat
33558 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33561 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33563 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33565 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33566 (vec_pairing): ...to this.
33567 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33568 (rs6000_sched_init): Adjust for name change.
33569 (struct rs6000_sched_context): Likewise.
33570 (rs6000_init_sched_context): Likewise.
33571 (rs6000_set_sched_context): Likewise.
33573 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33579 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33581 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33582 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33583 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33585 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33587 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33589 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33591 PR rtl-optimization/70703
33592 * ira-color.c (update_conflict_hard_regno_costs): Use
33593 int64_t instead of HOST_WIDE_INT.
33595 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33597 PR rtl-optimization/70478
33598 * lra-constraints.c (process_alt_operands): Disfavor alternative
33599 insn memory operands.
33601 2017-04-07 Jeff Law <law@redhat.com>
33603 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33604 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33606 2017-04-07 Martin Liska <mliska@suse.cz>
33609 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33610 Show error message instead of an ICE.
33612 2017-04-07 Martin Liska <mliska@suse.cz>
33615 * ipa-split.c (split_function): Add function part to a same comdat
33618 2017-04-07 Richard Biener <rguenther@suse.de>
33620 PR middle-end/80341
33621 * tree.c (get_unwidened): Also handle ! for_type case for
33623 * convert.c (do_narrow): Split out from ...
33624 (convert_to_integer_1): ... here. Do not pass final truncation
33625 type to get_unwidened for TRUNC_DIV_EXPR.
33627 2017-04-07 Richard Biener <rguenther@suse.de>
33629 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33631 (aff_combination_const): Adjust.
33632 (aff_combination_scale): Likewise.
33633 (aff_combination_add_elt): Likewise.
33634 (aff_combination_add_cst): Likewise.
33635 (aff_combination_convert): Likewise.
33636 (add_elt_to_tree): Likewise. Remove unused argument.
33637 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33639 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33641 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33643 * config/arm/arm.c (arm_default_short_enums): Use
33644 ARM_DEFAULT_SHORT_ENUMS.
33645 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33647 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33650 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33651 members with redundant out-of-class redeclaration.
33653 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33656 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33657 * config/i386/i386.md (*zero_extendsidi2):
33658 Add (?*x,*x) and (?*v,*v) alternatives.
33660 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33663 * config/i386/i386.c (ix86_expand_builtin)
33664 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33665 mode from insn data. Convert operands to insn operand mode.
33666 Copy operands that don't satisfy insn predicate to a register.
33668 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33670 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33673 2017-04-06 Richard Biener <rguenther@suse.de>
33675 PR tree-optimization/80334
33676 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33677 preserve alignment of accesses.
33679 2017-04-06 Richard Biener <rguenther@suse.de>
33681 PR tree-optimization/80262
33682 * tree-sra.c (build_ref_for_offset): Preserve address-space
33684 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33685 Drop useless address-space information on MEM_REF offsets.
33687 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33689 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33691 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33693 PR rtl-optimization/70703
33694 * ira-color.c (update_conflict_hard_regno_costs): Use
33695 HOST_WIDE_INT instead of long.
33697 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33700 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33701 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33703 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33704 for x86_64 target. Handle -m3dnowa option.
33706 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33708 PR rtl-optimization/70703
33709 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33710 (update_conflict_hard_regno_costs): Use long instead of unsigned
33711 arithmetic for cost calculation.
33713 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33714 Bernd Edlinger <bernd.edlinger@hotmail.de>
33717 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33720 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33723 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33724 ptr_mode with Pmode throughout.
33725 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33726 into probe_stack_range and use DImode.
33728 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33731 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33732 call_eh_return is true.
33734 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33736 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33737 Initialize last_match_fntype_index.
33739 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33742 * tree-nvr.c: Include internal-fn.h.
33743 (pass_return_slot::execute): Ignore internal calls without
33746 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33747 Richard Biener <rguenther@suse.de>
33750 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33751 captures used multiple times, except for the last use.
33752 * generic-match-head.c: Include gimplify.h.
33754 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33756 PR tree-optimization/79390
33757 * target.h (struct noce_if_info): Declare.
33758 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33759 * target.def (noce_conversion_profitable_p): New target hook.
33760 * ifcvt.h (struct noce_if_info): New type, moved from ...
33761 * ifcvt.c (struct noce_if_info): ... here.
33762 (noce_conversion_profitable_p): Renamed to ...
33763 (default_noce_conversion_profitable_p): ... this. No longer
33765 (noce_try_store_flag_constants, noce_try_addcc,
33766 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33767 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33768 instead of noce_conversion_profitable_p.
33769 * config/i386/i386.c: Include ifcvt.h.
33770 (ix86_option_override_internal): Don't override
33771 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33772 (ix86_noce_conversion_profitable_p): New function.
33773 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33774 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33775 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33776 * doc/tm.texi: Regenerated.
33778 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33780 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33783 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33786 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33787 instructions for small multiply cores.
33789 2017-04-04 Jeff Law <law@redhat.com>
33791 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33793 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33794 that are not set by the loop over the elements.
33796 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33799 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33800 int mode, convert_modes it to mode as unsigned, otherwise use
33801 lowpart_subreg to mode rather than SImode.
33802 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33803 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33804 Use DImode instead of SImode for the shift count operand.
33805 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33808 2017-04-04 Richard Biener <rguenther@suse.de>
33810 PR middle-end/80281
33811 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33812 arithmetic done for the negate or the plus. Simplify.
33813 (A - (-B) -> A + B): Likewise.
33814 * fold-const.c (split_tree): Make sure to not negate pointers.
33816 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33818 PR rtl-optimization/60818
33819 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33820 a compare of comparisons with the thing compared if this results
33821 in a different machine mode.
33823 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33825 * alias.c (base_alias_check): Fix typo in comment.
33826 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33827 * cgraphunit.c (symbol_table::compile): Likewise.
33828 * collect2.c (maybe_run_lto_and_relink): Likewise.
33829 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33830 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33831 * config/avr/avr.c (avr_map_op_t): Likewise.
33832 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33833 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33834 * config/epiphany/epiphany.md (movcc): Likewise.
33835 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33836 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33838 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33839 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33840 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33841 * config/sh/sh.c (sh_rtx_costs): Likewise.
33842 * fold-const.c (fold_truth_andor): Likewise.
33843 * genautomata.c (collapse_flag): Likewise.
33844 * gengtype.h (struct type::u::s): Likewise.
33845 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33846 * input.c (FORMAT_AMOUNT): Likewise.
33847 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33848 (known_aggs_to_agg_replacement_list): Likewise.
33849 * ipa-inline-analysis.c: Likewise.
33850 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33851 * ipa-polymorphic-call.c
33852 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33853 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33854 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33856 * modulo-sched.c (apply_reg_moves): Likewise.
33857 * omp-expand.c (build_omp_regions_1): Likewise.
33858 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33859 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33860 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33861 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33862 * value-prof.c: Likewise.
33863 * var-tracking.c (val_reset): Likewise.
33865 2017-04-03 Richard Biener <rguenther@suse.de>
33867 PR tree-optimization/80275
33868 * fold-const.c (split_address_to_core_and_offset): Handle
33871 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33873 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33874 descriptors is at least equal to that of functions.
33876 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33878 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33880 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33883 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33884 (mov<IMOD4:mode>): New expander.
33885 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33887 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33889 PR rtl-optimization/79405
33890 * fwprop.c (propagations_left): New variable.
33891 (forward_propagate_into): Decrement it.
33892 (fwprop_init): Initialize it.
33893 (fw_prop): If the variable has reached zero, stop propagating.
33894 (fwprop_addr): Ditto.
33896 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33899 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33900 a FUNCTION_DECL, pass it as decl instead of origin to
33903 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33905 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33908 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33911 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33912 TARGET_VSX_SMALL_INTEGER.
33914 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33916 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33917 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33919 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33921 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33922 extraction from odd-numbered MSA register.
33924 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33926 PR middle-end/80173
33927 * expmed.c (store_bit_field_1): Don't attempt to create
33928 a word subreg out of hard registers wider than word if they
33929 have HARD_REGNO_NREGS of 1 for their mode.
33931 PR middle-end/80163
33932 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33933 conversions to integer types wider than word and pointer.
33936 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33937 (rtx_equal_for_cselib_p): Pass 0 to it.
33938 * cselib.c (cselib_hasher::equal): Likewise.
33939 (rtx_equal_for_cselib_1): Add depth argument. If depth
33940 is 128, don't look up VALUE locs and punt. Increment
33941 depth in recursive calls when walking VALUE locs.
33943 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33945 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33946 (make_gcov_file_name): Use the canonical path name for generating
33948 (read_line): Fix handling of files with ascii null bytes.
33950 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33952 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33953 to initialise a vector register instead
33954 of using a const_int.
33956 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33958 PR translation/80189
33959 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33960 diagnostic messages.
33962 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33965 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33966 (dfp_diex_<mode>): Update mode of operand 1.
33967 * doc/extend.texi (dxex, dxexq): Document change to return type.
33968 (diex, diexq): Document change to argument type.
33970 2017-03-30 Martin Jambor <mjambor@suse.cz>
33973 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33974 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33975 it reflects the signature changes performed at the callee side.
33976 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33977 to cgraph_build_function_type_skip_args.
33978 (build_function_decl_skip_args): Adjust call to the above function.
33980 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33983 * config/i386/sse.md
33984 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33985 register as dest whenever it is a MEM not rtx_equal_p to the
33986 corresponding dup operand, and when forcing into reg move the
33987 reg into the memory afterwards.
33988 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33989 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33990 for the force_reg mode.
33991 (avx512vl_vextractf128<mode>): Use register as dest either
33992 always when a MEM, or when it is a MEM not rtx_equal_p to the
33993 corresponding dup operand, or even not when it is a CONST_VECTOR
33994 depending on the mode and lo vs. hi.
33995 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33997 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33998 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33999 Likewise. Require that operands[2] is even.
34000 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34001 Remove extraneous parens. Require that operands[2] is a multiple
34003 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34004 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34005 disallow memory then.
34007 2017-03-30 Richard Biener <rguenther@suse.de>
34009 PR tree-optimization/77498
34010 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34011 to non-constants over backedges.
34013 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34015 PR rtl-optimization/80233
34016 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34017 as last_combined_insn. Do not test for BARRIER_P separately.
34019 2017-03-29 Andreas Schwab <schwab@suse.de>
34022 * calls.c (prepare_call_address): Convert funexp to Pmode before
34023 copying to temp reg.
34025 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34027 PR tree-optimization/80158
34028 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34029 Handle possible future case of more than one alternate
34031 (replace_rhs_if_not_dup): Likewise.
34032 (replace_one_candidate): Likewise.
34034 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34036 PR rtl-optimization/80193
34037 * ira.c (ira): Do not check allocation for LRA.
34039 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34041 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34042 (nvptx_output_simt_exit): Declare.
34043 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34044 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34045 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34046 (nvptx_declare_function_name): ...here. Emit declaration of local
34047 memory space buffer for omp_simt_enter insn.
34048 (nvptx_output_unisimt_switch): New.
34049 (nvptx_output_softstack_switch): New.
34050 (nvptx_output_simt_enter): New.
34051 (nvptx_output_simt_exit): New.
34052 * config/nvptx/nvptx.h (struct machine_function): New fields
34053 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34054 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34055 (UNSPECV_SIMT_EXIT): Ditto.
34056 (omp_simt_enter_insn): New insn.
34057 (omp_simt_enter): New expansion.
34058 (omp_simt_exit): New insn.
34059 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34061 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34062 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34063 (expand_GOMP_SIMT_EXIT): New.
34064 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34065 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34066 (GOMP_SIMT_EXIT): Ditto.
34067 * target-insns.def (omp_simt_enter): New insn.
34068 (omp_simt_exit): Ditto.
34069 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34071 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34072 (lower_rec_input_clauses): Likewise.
34073 (lower_lastprivate_clauses): Handle SIMT privatization.
34075 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34076 (ompdevlow_adjust_simt_enter): New.
34077 (find_simtpriv_var_op): New.
34078 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34079 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34081 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34082 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34083 (copy_decl_for_dup_finish): Ditto.
34085 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34087 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34090 * config/i386/i386.c (ix86_option_override_internal): Always
34091 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34093 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34095 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34097 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34099 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34100 mark new edge's irreducible flag accordign to it.
34101 (vect_do_peeling): Check loop preheader edge's irreducible flag
34102 and pass it to function slpeel_add_loop_guard.
34104 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34106 PR tree-optimization/80218
34107 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34108 Update block frequencies and counts.
34110 2017-03-28 Richard Biener <rguenther@suse.de>
34112 PR tree-optimization/78644
34113 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34114 of a simplification result we may not use it at all.
34116 2017-03-28 Richard Biener <rguenther@suse.de>
34119 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34120 without arguments, generate default definition of a SSA name.
34122 2017-03-28 Richard Biener <rguenther@suse.de>
34124 PR middle-end/80222
34125 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34126 TYPE_REF_CAN_ALIAS_ALL references.
34127 * fold-const.c (fold_indirect_ref_1): Likewise.
34129 2017-03-28 Martin Liska <mliska@suse.cz>
34132 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34133 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34135 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34136 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34138 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34139 (EXTRA_SPECS): Define.
34140 (SUBTARGET_EXTRA_SPECS): Likewise.
34141 (SUBTARGET_CPP_SPEC): Likewise.
34142 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34143 SUBTARGET_EXTRA_SPECS.
34144 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34146 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34148 * config/arc/simdext.md (vst64_insn): Update pattern.
34149 (vld32wh_insn): Likewise.
34150 (vld32wl_insn): Likewise.
34151 (vld64_insn): Likewise.
34152 (vld32_insn): Likewise.
34154 2017-03-28 Marek Polacek <polacek@redhat.com>
34157 * fold-const.c (fold_comparison): Use protected_set_expr_location
34158 instead of SET_EXPR_LOCATION.
34160 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34162 * tree.c (add_expr): Avoid name lookup warning.
34164 2017-03-27 Jeff Law <law@redhat.com>
34166 PR tree-optimization/80216
34167 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34168 function name. Limit recursion depth.
34169 (record_temporary_equivalences): Corresponding changes.
34171 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34173 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34176 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34179 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34181 * cfgcleanup.c (reg_note_cfa_p): New array.
34182 (insns_have_identical_cfa_notes): New function.
34183 (old_insns_match_p): Don't cross-jump in between /f
34184 and non-/f instructions. If both i1 and i2 are frame related,
34185 verify all CFA notes, their order and content.
34187 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34190 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34191 HImode and SImode with zero extend to DImode to one insn.
34192 (bswap<mode>2_extenddi): Likewise.
34193 (bswapsi2_extenddi): Likewise.
34194 (bswaphi2_extendsi): Likewise.
34195 (bswaphi2): Combine bswap HImode and SImode into one insn.
34196 Separate memory insns from swapping register.
34197 (bswapsi2): Likewise.
34198 (bswap<mode>2): Likewise.
34199 (bswaphi2_internal): Delete, no longer used.
34200 (bswapsi2_internal): Likewise.
34201 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34202 store, and gpr<-gpr swap insns.
34203 (bswap<mode>2_store): Likewise.
34204 (bswaphi2_reg): Register only splitter, combine with the splitter.
34205 (bswaphi2 splitter): Likewise.
34206 (bswapsi2_reg): Likewise.
34207 (bswapsi2 splitter): Likewise.
34208 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34209 the insns into load, store, and register/register insns.
34210 (bswapdi2_ldbrx): Likewise.
34211 (bswapdi2_load): Likewise.
34212 (bswapdi2_store): Likewise.
34213 (bswapdi2_reg): Likewise.
34215 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34217 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34218 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34220 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34223 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34225 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34226 special handling for target option conflicts between dform
34227 options (-mpower9-dform, -mpower9-dform-vector,
34228 -mpower9-dform-scalar) and -mno-direct-move.
34230 2017-03-27 Richard Biener <rguenther@suse.de>
34232 PR tree-optimization/80181
34233 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34235 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34237 * config/arc/predicates.md (move_double_src_operand): Replace the
34238 call to move_double_src_operand with a call to address_operand.
34240 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34242 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34243 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34244 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34246 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34248 * config/arc/predicates.md (long_immediate_loadstore_operand):
34249 Consider scaled addresses cases.
34251 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34253 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34254 restored when in interrupt.
34255 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34256 doesn't have delay slot.
34258 2017-03-27 Richard Biener <rguenther@suse.de>
34261 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34262 inlined thunk clones.
34264 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34267 * asan.c (instrument_derefs): Copy over last operand from
34268 original COMPONENT_REF to the new COMPONENT_REF with
34269 DECL_BIT_FIELD_REPRESENTATIVE.
34270 * ubsan.c (instrument_object_size): Likewise.
34272 2017-03-27 Richard Biener <rguenther@suse.de>
34274 PR tree-optimization/80170
34275 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34276 sure DR/SCEV didnt fold in constants we do not see when looking
34277 at the reference base alignment.
34279 2017-03-27 Richard Biener <rguenther@suse.de>
34281 PR middle-end/80171
34282 * gimple-fold.c (fold_ctor_reference): Properly guard against
34283 NULL return value from canonicalize_constructor_val.
34285 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34288 * config/i386/i386.c (ix86_expand_builtin)
34289 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34290 flags reg setting and flags reg using instructions.
34291 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34292 clobbering instructions to zero extend op2.
34294 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34296 * doc/install.texi (Configuration) <--with-aix-soname>:
34297 Update link to AIX ld.
34299 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34301 PR rtl-optimization/80160
34302 PR rtl-optimization/80159
34303 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34304 reg_alternate_class into account.
34306 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34309 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34310 to consider in curr_insn_transform.
34312 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34314 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34315 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34316 and emit_mode_inner.
34318 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34320 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34321 argument to the overloaded builtin variants. Use the new flag to
34322 deprecate certain builtin variants.
34323 * config/s390/s390-builtin-types.def: Add new builtin types.
34324 * config/s390/s390-builtins.h: Support new flags field for
34325 overloaded builtins.
34326 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34327 (s390_macro_to_expand): Enable vector float data type.
34328 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34329 builtins by incrementing the __VEC__ version number.
34330 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34332 (s390_resolve_overloaded_builtin): Emit error messages depending
34333 on the builtin flags.
34334 * config/s390/s390.c (s390_expand_builtin): Support additional
34335 flags argument. Change error message to match the messages
34336 emitted in s390-c.c.
34337 * config/s390/s390.md: New UNSPEC_* constants.
34338 (op_type): Add new instruction types.
34339 * config/s390/vecintrin.h: Add new builtins and test data class
34341 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34342 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34343 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34344 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34345 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34346 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34348 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34349 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34350 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34351 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34353 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34354 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34355 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34356 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34358 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34359 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34360 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34361 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34362 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34363 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34364 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34366 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34367 ("vec_scatter_element<V_HW_4:mode>_DI")
34368 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34369 ("vec_fpint<mode>", "vflls")
34370 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34371 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34372 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34373 ("*vec_cmphe<mode>_cc"): ... these.
34375 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34376 mode constant instead of magic value.
34378 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34380 * config/s390/s390.c (s390_expand_vec_compare): Support other
34381 vector floating point modes than just V2DF.
34382 (s390_expand_vcond): Likewise.
34383 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34384 (s390_cannot_change_mode_class): Prevent mode changes between TF
34385 and V1TF in vector registers.
34386 * config/s390/s390.md (DF, SF): New mode attributes.
34387 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34388 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34389 SFmode support for VRs.
34390 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34392 (VFT, VF_HW): New mode iterators.
34393 (vw, sdx): New mode attributes.
34394 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34395 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34396 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34397 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34398 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34399 also the new vector floating point modes. Renaming to ...
34401 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34402 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34403 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34404 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34405 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34406 ("vec_unordered<mode>"): ... these.
34408 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34409 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34410 ("*vec_extendv2df"): New insn definitions.
34412 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34414 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34415 ("mulditi3_2", "*muldi3_sign"): New patterns.
34416 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34417 rename the pattern definition.
34419 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34421 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34423 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34425 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34427 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34428 instruction if possible.
34429 * config/s390/vector.md (vec_halfnumelts): New mode
34431 ("*vec_vllezlf<mode>"): New pattern.
34433 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34435 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34436 ("popcountv4si2", "popcountv2di2"): Rename to ...
34437 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34438 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34440 ("popcount<mode>2_vxe"): New pattern.
34442 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34444 * common/config/s390/s390-common.c (processor_flags_table): Add
34446 * config.gcc: Add arch12.
34447 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34448 Default to arch12 for unknown CPU model numbers.
34449 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34450 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34451 PROCESSOR_max sanity check.
34452 * config/s390/s390-opts.h (enum processor_type): Add
34454 * config/s390/s390.c (processor_table): Add arch12.
34455 (s390_expand_builtin): Add check for B_VXE flag.
34456 (s390_issue_rate): Add PROCESSOR_ARCH12.
34457 (s390_get_sched_attrmask): Likewise.
34458 (s390_get_unit_mask): Likewise.
34459 (s390_sched_score): Enable z13 scheduling for arch12.
34460 (s390_sched_reorder): Likewise.
34461 (s390_sched_variable_issue): Likewise.
34462 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34464 (s390_tune_attr): Use z13 scheduling also for arch12.
34465 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34466 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34467 (TARGET_VXE_P): New macros.
34468 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34469 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34470 * config/s390/s390.opt: Add arch12 as processor_type.
34472 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34474 * config/s390/s390.md
34475 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34476 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34477 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34479 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34480 Rename expanders to ...
34482 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34483 ("fixuns_truncdddi2_emu"): ... these.
34485 ("fixuns_trunc<mode>si2_emu"): New expander.
34487 ("*fixuns_truncdfdi2_z13"): Rename to ...
34488 ("*fixuns_truncdfdi2_vx"): ... this.
34490 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34492 * config/s390/2964.md: Remove the single element vector compare
34493 instructions which are no longer used.
34494 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34496 (s390_canonicalize_comparison): Remove handling of DFmode
34498 (s390_expand_vec_compare_scalar): Remove function.
34499 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34500 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34502 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34504 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34506 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34508 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34509 will anyway by matched by mov<mode>_64dfp.
34511 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34513 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34514 vlef/vstef. Add missing operand to vleif.
34516 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34518 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34519 pair for all vector types with 64 bit elements.
34520 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34521 * config/s390/vector.md (V_HW_64): ... here.
34522 (V_128_NOSINGLE): New mode iterator.
34523 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34524 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34525 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34526 ("*vec_load_pairv2di"): Change to ...
34527 ("*vec_load_pair<mode>"): ... this one.
34529 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34531 * config/s390/constraints.md: Add comments.
34532 (jKK): Reject element sizes > 8 bytes.
34533 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34535 * config/s390/s390.md: Add the s_operand checks formerly in
34536 s390_split_ok_p to various splitters where they are still
34538 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34539 for 128 bit vectors. Plus two splitters.
34541 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34543 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34546 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34549 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34550 error if the boundary argument is not constant.
34552 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34554 PR rtl-optimization/80112
34555 * loop-doloop.c (doloop_condition_get): Don't check condition
34556 if cmp isn't SET with IF_THEN_ELSE src.
34558 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34560 PR tree-optimization/80158
34561 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34562 replacing a candidate statement, also replace it for the
34563 candidate's alternate interpretation.
34564 (replace_rhs_if_not_dup): Likewise.
34565 (replace_one_candidate): Likewise.
34567 2017-03-24 Richard Biener <rguenther@suse.de>
34569 PR tree-optimization/80167
34570 * graphite-isl-ast-to-gimple.c
34571 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34573 (translate_isl_ast_to_gimple::get_rename): Likewise.
34575 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34577 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34578 handling of certain combinations of target options, including the
34579 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34580 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34582 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34585 * config/arm/arm.md (*load_multiple): Add reload_completed to
34586 matching condition.
34588 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34589 Richard Biener <rguenth@suse.de>
34591 PR tree-optimization/79908
34592 PR tree-optimization/80136
34593 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34594 been cast away, gimplify_and_add suffices.
34596 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34598 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34600 2017-03-23 Richard Biener <rguenther@suse.de>
34602 PR tree-optimization/80032
34603 * gimplify.c (gimple_push_cleanup): Forced unconditional
34604 cleanups still have to go to the conditional_cleanups
34607 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34609 PR tree-optimization/80072
34610 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34612 (next_operand_entry_id): Change type to unsigned int.
34613 (sort_by_operand_rank): Make sure to return the right return value
34614 even if unsigned fields are bigger than INT_MAX.
34615 (struct oecount): Change cnt and id type to unsigned int.
34616 (oecount_hasher::equal): Formatting fix.
34617 (oecount_cmp): Make sure to return the right return value
34618 even if unsigned fields are bigger than INT_MAX.
34619 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34622 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34623 TREE_READONLY on result if writing it more than once.
34626 * doc/invoke.texi (-fsanitize=thread): Document that with
34627 -fnon-call-exceptions atomics are not able to throw
34631 * tsan.c: Include tree-eh.h.
34632 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34633 maybe_clean_or_replace_eh_stmt where needed.
34634 (instrument_memory_accesses): Add cfg_changed argument.
34635 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34636 if it returned true.
34637 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34639 PR rtl-optimization/63191
34640 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34641 wrapper function, moved the whole old content into ...
34642 (ix86_delegitimize_address_1): ... this. New inline function.
34643 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34644 true as last argument instead of ix86_delegitimize_address.
34646 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34648 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34649 cortexa57_branch_cost.
34651 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34653 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34655 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34658 * doc/md.texi (Constraints): Document wA constraint.
34659 * config/rs6000/constraints.md (wA): New.
34660 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34661 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34662 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34663 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34665 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34668 * gimplify.c (is_oacc_declared): New function.
34669 (oacc_default_clause): Use it to set default flags for acc declared
34670 variables inside parallel regions.
34671 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34672 declared variables.
34673 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34674 declare attribute to any decl as necessary.
34676 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34679 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34680 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34681 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34682 (arm_arch_lpae): This.
34683 * config/arm/arm.c (arm_arch7ve): Rename into ...
34684 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34685 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34688 2017-03-22 Martin Liska <mliska@suse.cz>
34691 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34692 error message instead of an ICE.
34694 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34696 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34698 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34700 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34702 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34705 2017-03-21 Martin Sebor <msebor@redhat.com>
34707 * doc/extend.texi: Use "cannot" instead of "can't."
34708 * doc/hostconfig.texi: Same.
34709 * doc/install.texi: Same.
34710 * doc/invoke.texi: Same.
34711 * doc/loop.texi: Same.
34712 * doc/md.texi: Same.
34713 * doc/objc.texi: Same.
34714 * doc/rtl.texi: Same.
34715 * doc/tm.texi: Same.
34716 * doc/tm.texi.in: Same.
34717 * doc/trouble.texi: Same.
34719 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34722 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34723 (collect_checksum_attributes): Set it.
34724 (die_checksum_ordered): Use it.
34726 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34728 PR tree-optimization/79908
34729 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34730 change: For a VA_ARG whose LHS has been cast away, use
34731 force_gimple_operand to construct the side effects.
34733 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34735 PR translation/80001
34736 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34737 more amenable to translation.
34738 (oacc_loop_auto_partitions): Likewise.
34740 2017-03-21 Marek Polacek <polacek@redhat.com>
34741 Martin Sebor <msebor@redhat.com>
34743 PR tree-optimization/80109
34744 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34745 on INTEGRAL_TYPE_P.
34747 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34748 Segher Boessenkool <segher@kernel.crashing.org>
34751 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34752 check reg_used_between_p between insn and one of succ or succ2
34753 depending on if succ is artificial insn not inserted into insn
34756 2017-03-21 Martin Liska <mliska@suse.cz>
34758 PR gcov-profile/80081
34759 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34760 * doc/gcc.texi: Include gcov-dump stuff.
34761 * doc/gcov-dump.texi: New file.
34763 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34765 PR rtl-optimization/79150
34766 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34767 conditional jump, if the jump is the last insn of the loop.
34769 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34770 Richard Biener <rguenth@suse.de>
34772 PR tree-optimization/79908
34773 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34774 been cast away, use force_gimple_operand to construct the side
34777 2017-03-21 Martin Liska <mliska@suse.cz>
34779 PR libfortran/79956
34780 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34783 2017-03-21 Brad Spengler <spender@grsecurity.net>
34786 * plugin.c (htab_hash_plugin): New function.
34787 (add_new_plugin): Use it and adjust.
34788 (parse_plugin_arg_opt): Adjust.
34789 (init_one_plugin): Likewise.
34791 2017-03-21 Richard Biener <rguenther@suse.de>
34793 PR tree-optimization/80032
34794 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34795 if set force the cleanup to happen unconditionally.
34796 (gimplify_target_expr): Push inserted clobbers with force_uncond
34797 to avoid them being removed by control-dependent DCE.
34799 2017-03-21 Richard Biener <rguenther@suse.de>
34801 PR tree-optimization/80122
34802 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34803 va_arg_pack_len when the inlined call stmt requires pack
34805 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34807 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34810 * tsan.c (instrument_builtin_call): If the memory model argument
34811 is not a constant, assume it is valid.
34814 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34817 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34819 PR rtl-optimization/79910
34820 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34821 if its dest is used by an insn before I2 (other than the combined
34822 insns themselves, which are properly handled already).
34824 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34827 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34829 * combine.c (record_used_regs): New static function.
34830 (try_combine): Handle situations where there is an additional
34831 instruction between I2 and I3 which needs to have a LOG_LINK
34835 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34837 * combine.c (try_combine): Delete redundant i1 test. Call
34838 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34840 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34843 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34844 alternatives 13/14.
34846 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34848 PR tree-optimization/80054
34849 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34850 the optimization if a PHI or any of its arguments is not dominated
34851 by the candidate's basis. Use gphi* rather than gimple* as
34853 (replace_profitable_candidates): Clean up a gimple* variable that
34854 should be a gphi* variable.
34856 2017-03-20 Martin Sebor <msebor@redhat.com>
34859 * doc/extend.texi (attribute constructor): Document present limitation.
34861 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34864 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34865 __POWER9_VECTOR__ #ifdef control, change template definition to
34866 use Power9-specific built-in function.
34867 (vec_any_eq): Likewise.
34868 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34869 to control outcomes from this test.
34870 (vector_ae_<mode>p): For VEC_F modes, likewise.
34872 2017-03-20 Ian Lance Taylor <iant@google.com>
34874 * config/i386/i386.c (ix86_function_regparm): Save an extra
34875 register for -fsplit-stack with DECL_STATIC_CHAIN.
34877 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34880 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34881 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34883 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34885 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34887 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34890 2017-03-20 Marek Polacek <polacek@redhat.com>
34893 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34895 2017-03-20 Richard Biener <rguenther@suse.de>
34897 PR tree-optimization/80113
34898 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34899 allocate extra SSA name for PHI def.
34900 (add_close_phis_to_outer_loops): Likewise.
34901 (add_close_phis_to_merge_points): Likewise.
34902 (copy_loop_close_phi_args): Likewise.
34903 (copy_cond_phi_nodes): Likewise.
34905 2017-03-20 Martin Liska <mliska@suse.cz>
34907 PR middle-end/79753
34908 * tree-chkp.c (chkp_build_returned_bound): Do not build
34909 returned bounds for a LHS that's not a BOUNDED_P type.
34911 2017-03-20 Martin Liska <mliska@suse.cz>
34915 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34916 COMPLEX_CST and VECTOR_CST.
34918 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34921 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34922 target operand. A new splitter adds the clobber statement in case
34923 the target operand is dead anyway.
34925 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34927 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34928 to age-old versions of binutils and glibc.
34930 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34932 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34934 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34936 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34938 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34940 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34941 requirement for binutils 2.13.
34943 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34945 * combine.c (try_combine): Delete redundant i1 test. Call
34946 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34948 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34950 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34951 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34953 <riscv64-*-elf>: Re-arrange section
34954 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34955 <riscv32-*-linux>: Likewise.
34956 <riscv64-*-elf>: Likewise
34957 <riscv64-*-linux>: Likewise.
34959 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34962 * aarch64.opt(verbose-cost-dump): Fix typo.
34964 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34967 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34968 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34970 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34972 * reload.c (find_reloads): When reloading a nonoffsettable address,
34973 use RELOAD_OTHER for it and its address reloads.
34975 PR rtl-optimization/79910
34976 * combine.c (record_used_regs): New static function.
34977 (try_combine): Handle situations where there is an additional
34978 instruction between I2 and I3 which needs to have a LOG_LINK
34981 2017-03-17 Jeff Law <law@redhat.com>
34983 PR tree-optimization/71437
34984 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34985 conditional in the hash table first.
34986 (vrp_dom_walker::before_dom_children): Extract condition from
34987 ASSERT_EXPR. Record condition, its inverion and any implied
34988 conditions as well.
34990 2017-03-17 Marek Polacek <polacek@redhat.com>
34991 Markus Trippelsdorf <markus@trippelsdorf.de>
34993 PR tree-optimization/80079
34994 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34997 2017-03-17 Richard Biener <rguenther@suse.de>
34999 PR middle-end/80075
35000 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35001 Properly verify the LHS before the RHS possibly claims to be
35003 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35006 2017-03-17 Martin Jambor <mjambor@suse.cz>
35008 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35009 (List of -O2 options): Likewise.
35010 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35013 2017-03-17 Tom de Vries <tom@codesourcery.com>
35015 * gcov-dump.c (print_usage): Print bug_report_url.
35017 2017-03-17 Richard Biener <rguenther@suse.de>
35019 PR middle-end/80050
35020 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35021 (parser::peek): Likewise.
35023 2017-03-17 Richard Biener <rguenther@suse.de>
35025 PR tree-optimization/80048
35026 * sese.c (free_sese_info): Properly release rename_map and
35027 copied_bb_map elements.
35029 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35031 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35032 Add linked-list forward and backlinks. Insert on
35033 construction, remove on destruction.
35034 (class pass_store_merging): Add m_stores_head field.
35035 (pass_store_merging::terminate_and_process_all_chains):
35036 Iterate over m_stores_head list.
35037 (pass_store_merging::terminate_all_aliasing_chains):
35039 (pass_store_merging::execute): Check for debug stmts first.
35040 Push new chains onto the m_stores_head stack.
35042 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35045 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35046 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35047 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35049 2017-03-16 Jeff Law <law@redhat.com>
35051 PR tree-optimization/71437
35052 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35053 member function. Implementation moved into after_dom_children
35054 member function and into the threader's thread_outgoing_edges
35056 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35057 some code into new thread_outgoing_edges.
35058 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35059 definition. Simplify marker handling (do it here). Assume we always
35060 have the available expression and the const/copies tables.
35061 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35063 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35064 * tree-vrp.c (equiv_stack): No longer file scoped.
35065 (vrp_dom_walker): New class.
35066 (vrp_dom_walker::before_dom_children): New member function.
35067 (vrp_dom_walker::after_dom_children): Likewise.
35068 (identify_jump_threads): Setup domwalker. Use it rather than
35069 walking edges in a random order by hand. Simplify setup/finalization.
35070 (finalize_jump_threads): Remove.
35071 (vrp_finalize): Do not call identify_jump_threads here.
35072 (execute_vrp): Do it here instead and call thread_through_all_blocks
35075 PR tree-optimization/71437
35076 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35078 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35080 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35081 (dom_opt_dom_walker::thread_across_edge): Remove
35082 handle_dominating_asserts argument. All callers changed.
35083 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35084 changes. Remove calls to lhs_of_dominating_assert. Other
35085 uses of handle_dominating_asserts turn into unconditional code
35086 (simplify_control_stmt_condition_1): Likewise.
35087 (simplify_control_stmt_condition): Likewise.
35088 (thread_through_normal_block, thread_across_edge): Likewise.
35089 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35090 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35091 object if it is not an SSA_NAME.
35092 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35093 before calling into the VRP specific simplifiers.
35094 (identify_jump_threads): Remove handle_dominating_asserts
35097 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35100 * tree-diagnostic.c (default_tree_printer): No longer static.
35101 * tree-diagnostic.h (default_tree_printer): New prototype.
35103 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35105 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35106 Change ins into fmov.
35108 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35110 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35111 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35112 Use h_con constraint for operand 1.
35113 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35114 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35116 2017-03-15 Jeff Law <law@redhat.com>
35118 PR tree-optimization/71437
35119 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35120 (record_temporary_equivalences): Use it.
35122 PR tree-optimization/71437
35123 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35124 tree-ssa-scopedtables.
35125 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35126 (record_conditions, record_cond, vuse_eq): Likewise.
35127 (record_edge_info): Adjust to API tweak of record_conditions.
35128 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35129 (record_temporary_equivalences, optimize_stmt): Likewise.
35130 (eliminate_redundant_computations): Likewise.
35131 (record_equivalences_from_stmt): Likewise.
35132 * tree-ssa-scopedtables.c: Include options.h and params.h.
35133 (vuse_eq): New function, moved from tree-ssa-dom.c
35134 (build_and_record_new_cond): Likewise.
35135 (record_conditions): Likewise. Accept vector of conditions rather
35136 than edge_equivalence structure for first argument.
35137 for the first argument.
35138 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35139 from tree-ssa-dom.c.
35140 (avail_exprs_stack::record_cond): Likewise.
35141 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35142 from tree-ssa-dom.c.
35143 (avail_exprs_stack): Add new member functions lookup_avail_expr
35145 (record_conditions): Declare.
35147 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35150 * lra-constraints.c (process_alt_operands): Increase reject for
35151 reloading an input/output operand.
35153 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35156 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35157 insns to convert from signed/unsigned char/short to IEEE 128-bit
35159 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35161 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35164 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35165 subreg of inner mode for values already in registers.
35167 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35169 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35170 iteration reg is used after the loop.
35172 2017-03-14 Martin Sebor <msebor@redhat.com>
35174 PR tree-optimization/79800
35175 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35176 precision in negative-positive range.
35177 (format_floating): Call non-const overload with adjusted precision.
35179 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35182 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35185 2017-03-14 Martin Sebor <msebor@redhat.com>
35187 PR middle-end/80020
35188 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35189 * builtins.def (aligned_alloc): Use it.
35192 * Makefile.in (GTFILES): Add calls.c.
35193 * calls.c: Include "gt-calls.h".
35195 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35197 PR rtl-optimization/79728
35198 * regs.h (struct target_regs): New field
35199 x_contains_allocatable_regs_of_mode.
35200 (contains_allocatable_regs_of_mode): New macro.
35201 * reginfo.c (init_reg_sets_1): Initialize it, and change
35202 contains_reg_of_mode so it includes global regs as well.
35203 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35204 rather than contains_regs_of_mode.
35206 2017-03-14 Martin Liska <mliska@suse.cz>
35208 * doc/invoke.texi: Document options that can't be combined with
35209 -fcheck-pointer-bounds.
35211 2017-03-14 Martin Liska <mliska@suse.cz>
35213 PR middle-end/79831
35214 * doc/invoke.texi (-Wchkp): Document the option.
35216 2017-03-14 Martin Liska <mliska@suse.cz>
35218 * Makefile.in: Install gcov-dump.
35220 2017-03-14 Martin Liska <mliska@suse.cz>
35222 * multiple_target.c (expand_target_clones): Bail out for
35223 an invalid attribute.
35225 2017-03-14 Richard Biener <rguenther@suse.de>
35227 * alias.c (struct alias_set_entry): Pack properly.
35228 * cfgloop.h (struct loop): Likewise.
35229 * cse.c (struct set): Likewise.
35230 * ipa-utils.c (struct searchc_env): Likewise.
35231 * loop-invariant.c (struct invariant): Likewise.
35232 * lra-remat.c (struct cand): Likewise.
35233 * recog.c (struct change_t): Likewise.
35234 * rtl.h (struct address_info): Likewise.
35235 * symbol-summary.h (function_summary): Likewise.
35236 * tree-loop-distribution.c (struct partition): Likewise.
35237 * tree-object-size.c (struct object_size_info): Likewise.
35238 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35239 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35240 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35241 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35242 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35243 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35244 (struct _stmt_vec_info): Likewise.
35246 2017-03-14 Martin Liska <mliska@suse.cz>
35249 * multiple_target.c (create_dispatcher_calls): Check that
35250 a target can create a function dispatcher.
35252 2017-03-14 Martin Liska <mliska@suse.cz>
35255 * multiple_target.c (expand_target_clones): Drop local.local
35256 flag for default implementation.
35258 2017-03-14 Richard Biener <rguenther@suse.de>
35260 PR tree-optimization/80030
35261 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35263 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35265 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35266 gcc_fallthrough() instead of __attribute__((fallthrough));
35268 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35270 * doc/gcc.texi: Remove "up" link to (DIR).
35271 * doc/gccint.texi: Ditto.
35273 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35275 * doc/install.texi (Specific) <avr>: Remove reference to
35278 2017-03-13 Jeff Law <law@redhat.com>
35280 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35281 attribute rather than comments.
35283 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35284 match_scratch operand is highest.
35286 2017-03-13 Martin Liska <mliska@suse.cz>
35288 PR middle-end/78339
35289 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35290 is a CHKP clone, use original declaration.
35292 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35294 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35295 (arc_conditional_register_usage): Use a different allocation order
35296 when optimizing for size.
35297 * common/config/arc/arc-common.c (arc_option_optimization_table):
35298 Section anchors default on when optimizing for size.
35300 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35302 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35304 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35306 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35307 * config/arc/arc.md (cpu_facility): Add cd variant.
35308 (*movqi_insn): Add code density variant.
35309 (*movhi_insn): Likewise.
35310 (*movqi_insn): Likewise.
35311 (*addsi3_mixed): Likewise.
35312 (subsi3_insn): Likewise.
35314 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35316 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35318 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35320 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35321 expressions with MINUS and UNARY ops.
35323 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35326 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35328 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35329 between vec_select and vector argument.
35330 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35331 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35332 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35333 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35334 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35335 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35337 2017-03-13 Richard Biener <rguenther@suse.de>
35340 * params.def (vect-max-peeling-for-alignment): Fix typo.
35342 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35344 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35345 issue that only occurred with binutils below 2.18.
35347 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35349 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35350 refer to binutils 2.11/2.12 minimum.
35352 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35354 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35355 ftp.kernel.org and simplify binutils requirement.
35357 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35359 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35361 (Optimize Options): Ditto. Also remove redundancy.
35363 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35365 PR translation/79848
35366 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35368 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35369 to G_ to avoid double translation.
35371 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35373 PR translation/79923
35374 * auto-profile.c (get_combined_location): Convert leading
35375 character of diagnostics to lower case and remove trailing period.
35376 (read_profile): Likewise for various diagnostics.
35377 * config/arm/arm.c (arm_option_override): Remove trailing period
35378 from various diagnostics.
35379 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35380 (msp430_expand_delay_cycles): Likewise.
35382 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35385 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35386 full command-line argument, rather than just "str".
35387 (aarch64_validate_march): Likewise.
35388 (aarch64_validate_mtune): Likewise.
35390 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35392 PR rtl-optimization/78911
35393 * lra-assigns.c (must_not_spill_p): New function.
35394 (spill_for): Use it.
35396 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35398 PR tree-optimization/79981
35399 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35400 ATOMIC_COMPARE_EXCHANGE ifn result.
35401 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35402 IFN_ATOMIC_COMPARE_EXCHANGE.
35404 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35407 * opts.c (parse_sanitizer_options): Add missing question mark to
35408 "did you mean" message.
35410 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35412 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35414 (VMULEUH_UNS): Likewise.
35415 (VMULOUB_UNS): Likewise.
35416 (VMULOUH_UNS): Likewise.
35417 * config/rs6000/rs6000.c (builtin_function_type): Remove
35418 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35420 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35423 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35424 x with result of extra_parsing_for_operand_code_0.
35425 (function_reader::extra_parsing_for_operand_code_0): Convert
35426 return type from void to rtx, returning x. When reading
35427 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35428 larger size containing struct block_symbol.
35430 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35432 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35433 -mfloat128-hardware without -m64.
35435 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35438 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35439 entries to the case statement that marks unsigned arguments to
35440 overloaded functions.
35442 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35444 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35445 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35447 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35450 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35451 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35453 2017-03-10 Martin Liska <mliska@suse.cz>
35457 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35458 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35461 2017-03-10 Olivier Hainque <hainque@adacore.com>
35463 * tree-switch-conversion (array_value_type): Start by resetting
35464 candidate type to it's main variant.
35466 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35468 PR rtl-optimization/79909
35469 * combine.c (try_combine): Use simplify_replace_rtx on individual
35470 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35471 of the whole CALL_INSN_FUNCTION_USAGE.
35473 PR tree-optimization/79972
35474 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35475 get_range_info on SSA_NAMEs. Formatting fixes.
35477 2017-03-10 Richard Biener <rguenther@suse.de>
35478 Jakub Jelinek <jakub@redhat.com>
35480 PR tree-optimization/77975
35481 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35482 edge to be constant.
35483 (get_val_for): For constant x return it. Formatting fix.
35484 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35485 would use the same bases as the current one.
35487 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35489 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35490 instead of vec_select for V1TImode.
35491 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35493 (VSX_LE_128): Add V1TI to this mode iterator.
35494 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35495 (*vsx_le_perm_store_<mode>): Likewise.
35496 (pre-reload splitter for VSX stores): Likewise.
35497 (post-reload splitter for VSX stores): Likewise.
35498 (*vsx_xxpermdi2_le_<mode>): Likewise.
35499 (*vsx_lxvd2x2_le_<mode>): Likewise.
35500 (*vsx_stxvd2x2_le_<mode>): Likewise.
35502 2017-03-09 Michael Eager <eager@eagercon.com>
35504 Correct failures with --enable-checking=yes,rtl.
35506 * config/microblaze/microblaze.c (microblaze_expand_shift):
35507 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35508 test for const0_rtx.
35509 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35510 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35512 2017-03-09 Richard Biener <rguenther@suse.de>
35514 PR tree-optimization/79977
35515 * graphite-scop-detection.c (scop_detection::merge_sese):
35516 Handle the case of extra exits to blocks dominating the entry.
35518 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35520 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35523 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35525 PR rtl-optimization/79949
35526 * lra-constraints.c (process_alt_operands): Check memory when
35527 trying to predict a cycle. Print about the overall increase.
35529 2017-03-09 Richard Biener <rguenther@suse.de>
35531 PR middle-end/79971
35532 * gimple-expr.c (useless_type_conversion_p): Preserve
35533 TYPE_SATURATING for fixed-point types.
35535 2017-03-09 Richard Biener <rguenther@suse.de>
35538 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35539 alignment of BLKmode params.
35541 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35544 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35545 (VALL_NO_V2Q): Likewise.
35547 * config/aarch64/aarch64-simd.md
35548 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35550 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35551 VALL_NO_V2Q mode iterator.
35552 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35554 2017-03-09 Martin Liska <mliska@suse.cz>
35556 PR tree-optimization/79631
35557 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35558 tree_int_cst_sign_bit just for INTEGER constants.
35560 2017-03-09 Martin Liska <mliska@suse.cz>
35564 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35567 2017-03-09 Marek Polacek <polacek@redhat.com>
35570 * tree.c (inchash::add_expr): Handle TREE_VEC.
35572 2017-03-09 Martin Liska <mliska@suse.cz>
35575 (chkp_narrow_size_and_offset): New function.
35576 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35577 (void chkp_parse_bit_field_ref): New function.
35578 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35579 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35581 2017-03-09 Martin Liska <mliska@suse.cz>
35584 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35585 (chkp_find_bounds_1): Remove gcc_unreachable.
35587 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35590 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35591 BUILT_IN_SYNC*, determine the access type from the size suffix and
35592 always build a MEM_REF with that type. Handle forgotten
35593 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35596 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35597 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35598 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35599 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35600 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35601 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35602 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35603 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35604 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35605 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35606 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35607 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35608 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35609 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35610 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35611 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35612 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35613 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35614 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35615 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35616 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35617 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35618 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35619 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35620 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35621 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35622 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35623 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35624 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35625 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35626 definitions outside of __OPTIMIZE__ guarded section.
35629 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35630 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35631 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35632 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35635 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35637 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35638 ("vfenez<mode>"): Add missing constraints.
35640 2017-03-08 Martin Sebor <msebor@redhat.com>
35643 * config/nds32/nds32.c (nds32_option_override):
35644 Fix misspelled diagnostic.
35646 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35649 * gimplify.c (gimplify_omp_for): Replace index var in outer
35650 taskloop statement with an artificial variable and add
35651 OMP_CLAUSE_PRIVATE clause for it.
35653 2017-03-08 Richard Biener <rguenther@suse.de>
35655 PR tree-optimization/79955
35656 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35657 for accesses that are completely outside of the variable.
35659 2017-03-08 Andrew Haley <aph@redhat.com>
35661 PR tree-optimization/79943
35662 * tree-ssa-loop-split.c (compute_new_first_bound): When
35663 calculating the new upper bound, (END-BEG) should be added, not
35666 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35668 * config/avr/avr.md (setmemhi): Make sure match_dup
35669 operand number comes before match_scratch.
35671 2017-03-08 Richard Biener <rguenther@suse.de>
35673 PR tree-optimization/79920
35674 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35675 with ncopies == 1 to ...
35676 (vect_transform_slp_perm_load): ... here. Properly compute
35677 all element loads by iterating VF times over the group. Do
35678 not handle ncopies (computed in a broken way) in
35679 vect_create_mask_and_perm.
35681 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35684 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35685 is a uniform vector, use uniform_vector_p return value instead of
35686 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35688 2017-03-07 Marek Polacek <polacek@redhat.com>
35690 PR middle-end/79809
35691 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35692 (alloca_call_type): Likewise.
35694 2017-03-07 Martin Liska <mliska@suse.cz>
35696 * gcov.c (process_args): Put comment to correct location.
35698 2017-03-07 Martin Liska <mliska@suse.cz>
35700 PR middle-end/68270
35701 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35702 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35703 (chkp_narrow_bounds_for_field): Likewise.
35704 (chkp_parse_array_and_component_ref): Pass one more argument to
35707 2017-03-07 Richard Biener <rguenther@suse.de>
35709 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35712 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35714 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35715 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35717 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35720 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35721 to end of description.
35722 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35724 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35726 PR rtl-optimization/79901
35727 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35729 (*avx512f_<code><mode>3<mask_name>): ... this.
35730 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35731 iterator instead of VI8_AVX2_AVX512BW.
35733 PR rtl-optimization/79901
35734 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35735 min/max expander, expand it using expand_vec_cond_expr.
35738 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35741 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35744 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35745 to void * for PCH reasons.
35746 * dwarf2out.c (output_loc_operands, output_die): Cast
35747 v.val_vec.array to unsigned char *.
35749 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35752 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35755 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35757 PR rtl-optimization/79571
35758 * lra-constraints.c (process_alt_operands): Calculate static
35759 reject and subtract it from overall when only addresses will be
35762 2017-03-06 Julia Koval <julia.koval@intel.com>
35765 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35766 incoming stack boundary to 128 for 64-bit targets.
35768 2017-03-06 Richard Biener <rguenther@suse.de>
35770 PR tree-optimization/79894
35771 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35772 to NULL after folding it.
35774 2017-03-06 Richard Biener <rguenther@suse.de>
35776 PR tree-optimization/79824
35777 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35778 check disabling peeling for gaps.
35780 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35782 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35783 attributes): Document gettimeofday.
35785 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35787 * config/s390/s390.c (s390_option_override_internal): Set
35788 PARAM_MIN_VECT_LOOP_BOUND
35790 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35792 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35793 * config/s390/s390.md: Likewise.
35795 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35798 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35799 (<avx2_avx512>_perm<mode>): Rename to ...
35800 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35802 (<avx512>_perm<mode>_mask): Rename to ...
35803 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35805 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35806 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35807 instead of VI8F_256_512.
35808 (avx512f_perm<mode>): New define_expand.
35809 (avx512f_perm<mode>_mask): Likewise.
35810 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35811 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35813 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35815 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35816 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35818 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35820 2017-03-06 Martin Liska <mliska@suse.cz>
35823 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35824 when having a SSA NAME w/o VAR_DECL assigned to it.
35826 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35828 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35829 msa_dpsub_<su>_d): Fix MODE for vec_select.
35831 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35833 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35835 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35837 2017-03-06 Richard Biener <rguenther@suse.de>
35839 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35840 * plugin.c (register_plugin_info): Likewise.
35841 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35843 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35845 * config/i386/sse.md (sse_storehps, sse_storelps,
35846 avx_<castmode><avxsizesuffix>_<castmode>,
35847 avx512f_<castmode><avxsizesuffix>_<castmode>,
35848 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35849 in condition that at least one operand is not a MEM.
35851 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35853 PR middle-end/79805
35854 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35855 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35857 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35858 gimple_call_nothrow_p flag based on whether original builtin can throw.
35859 If it can, emit following stmts on the fallthrough edge.
35860 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35861 don't create new bb if inserting just debug stmts on the edge, try to
35862 insert them on the fallthru bb or just reset debug stmts.
35864 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35867 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35868 restore recog_data (including the operand rtxes inside it) around
35869 the call to get_insn_template.
35871 2017-03-03 Martin Sebor <msebor@redhat.com>
35873 PR tree-optimization/79699
35874 * context.c (context::~context): Free MPFR caches to avoid
35875 a memory leak on program exit.
35877 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35879 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35880 Use wide_int::ulow () instead of .elt (0).
35882 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35884 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35885 (*pushxf): Limit oF constraint to 32bit targets and add oC
35886 constraint for 64bit targets.
35887 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35888 (*pushdf): Change rmF constraint to rmC.
35890 2017-03-03 Martin Liska <mliska@suse.cz>
35892 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35893 Remove unused variable.
35895 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35898 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35899 is a memory operand, increase num_memory.
35900 (ix86_expand_args_builtin): Likewise.
35902 2017-03-03 Jan Hubicka <jh@suse.cz>
35905 * ipa-devirt.c (maybe_record_node): Properly handle
35906 __cxa_pure_virtual visibility.
35908 2017-03-03 Martin Liska <mliska@suse.cz>
35910 PR tree-optimization/79803
35911 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35913 (pass_loop_prefetch::execute): Disabled optimization if an
35914 assumption about L1 cache size is not met.
35916 2017-03-03 Martin Liska <mliska@suse.cz>
35918 PR rtl-optimization/79574
35919 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35920 (hash_scan_set): Likewise.
35921 (dump_hash_table): Likewise.
35922 (hoist_code): Likewise.
35924 2017-03-03 Richard Biener <rguenther@suse.de>
35926 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35927 in place of uhigh ().
35928 (fixed_convert_from_real): Likewise.
35930 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35933 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35935 2017-03-03 Richard Biener <rguenther@suse.de>
35937 PR middle-end/79818
35938 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35939 TYPE_OVERFLOW_UNDEFINED check.
35941 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35943 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35945 (vector_ae_<mode>_p): Likewise.
35946 (vector_nez_<mode>_p): Likewise.
35947 (vector_ne_v2di_p): Likewise.
35948 (vector_ae_v2di_p): Likewise.
35949 (vector_ne_<mode>_p): Likewise.
35950 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35952 (vsx_tsqrt<mode>2_fe): Likewise.
35954 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35957 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35959 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35961 PR rtl-optimization/79780
35962 * cprop.c (one_cprop_pass): When second and further conditional trap
35963 in a single basic block is turned into an unconditional trap, turn it
35964 into a deleted note to avoid RTL verification failures.
35966 2017-03-02 Richard Biener <rguenther@suse.de>
35968 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35970 2017-03-02 Richard Biener <rguenther@suse.de>
35972 PR tree-optimization/79345
35974 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35975 param and abort the walk, returning -1 if it is hit.
35976 (walk_aliased_vdefs): Take a limit param and pass it on.
35977 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35978 defaulting to 0 and return a signed int.
35979 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35980 (check_defs): New helper.
35981 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35982 about uninitialized memory.
35983 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35984 bogus uninitialized warning.
35985 (fixed_convert_from_real): Likewise.
35987 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35989 PR tree-optimization/66768
35990 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35991 iv_use if base object can't be determined.
35993 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35995 PR tree-optimization/79345
35996 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35997 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35998 (get_pattern_stats): Initialize it.
35999 * genemit.c (gen_expand): Verify match_scratch numbers come after
36000 match_operand/match_dup numbers.
36001 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36002 match_scratch numbers.
36003 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36005 * config/s390/s390.md (trunctdsd2): Likewise.
36007 2017-03-02 Richard Biener <rguenther@suse.de>
36009 * wide-int.h (wide_int_storage::operator=): Implement in terms
36012 2017-03-02 Richard Biener <rguenther@suse.de>
36014 PR tree-optimization/79777
36015 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36016 the to insert expression to sth existing.
36018 2017-03-01 Martin Sebor <msebor@redhat.com>
36020 PR middle-end/79692
36021 * gimple-ssa-sprintf.c
36022 (directive::known_width_and_precision): New function.
36023 (format_integer): Use it.
36024 (get_mpfr_format_length): Consider the full range of precision
36025 when computing %g output with the # flag. Set the likely byte
36026 count to 3 rather than 1 when precision is indeterminate.
36027 (format_floating): Correct the lower bound of precision.
36029 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36031 * doc/invoke.texi: Document default code model for 64-bit Linux.
36033 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36036 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36037 udiv rather than div since input pattern is unsigned.
36039 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36041 * config/i386/i386.c (print_reg): Warn for values of
36042 unsupported size in integer register.
36044 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36047 * config/rs6000/predicates.md (current_file_function_operand): Do
36048 not allow self calls to be local if the function is replaceable.
36050 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36053 * config/rs6000/altivec.h (vec_ctz and others): Change the
36054 preprocessor macro that controls conditional compilation from
36055 _ARCH_PWR9 to __POWER9_VECTOR__.
36056 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36057 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36058 control (instead of _ARCH_PWR9 control) so that template
36059 definition uses power9-specific function.
36060 (vec_any_eq): Likewise.
36061 (vec_all_ne): Change macro definition to use a power9-specific
36062 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36063 _ARCH_PWR9 control).
36064 (vec_any_eq) Likewise.
36065 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36066 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36067 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36068 support for xvcmpnedp instruction.
36069 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36070 macro expansion so that Power9 implementation of vec_all_ne does
36071 not use the AltiVec predicate framework.
36072 (VCMPNEH_P): Likewise.
36073 (VCMPNEW_P): Likewise.
36074 (VCMPNED_P): Likewise.
36075 (VCMPNEFP_P): Likewise.
36076 (VCMPNEDP_P): Likewise.
36077 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36078 implementation of vec_any_eq to not use AltiVec predicate
36080 (VCMPAEH_P): Likewise.
36081 (VCMPAEW_P): Likewise.
36082 (VCMPAED_P): Likewise.
36083 (VCMPAEFP_P): Likewise.
36084 (VCMPAEDP_P): Likewise.
36085 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36086 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36087 not use the AltiVec predicate framework.
36088 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36089 of vec_any_eq to not use AltiVec predicate framework.
36090 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36091 support for predefined __POWER9_VECTOR__ macro to indicate that
36092 Power9 instruction selection is enabled.
36093 (altivec_overloaded_builtins): Remove extraneous
36094 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36095 function argument types RS6000_BTI_bool_V16QI and
36096 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36097 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36098 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36099 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36100 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36101 Power9 for implementations of vec_cmpne. Change the signature for
36102 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36103 (representing vec_all_ne) to remove the previously described first
36104 argument of type RS6000_BTI_INTSI, as this was an artifact of
36105 reliance on the AltiVec predicate framework, which is no longer
36106 used in the implementation of these functions. Add
36107 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36108 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36109 since, unlike the AltiVec predicate framework implementation, we
36110 do not share function descriptors between vec_alle and vec_anyeq.
36111 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36112 set of modes that receive special treatment even when
36113 TARGET_P9_VECTOR is true. The special treatment emits code that
36114 does not depend on Power9 instructions.
36115 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36116 define_expand to not rely on AltiVec predicate framework.
36117 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36119 (vector_ne_v2di_p): Change this define_expand to not rely on
36120 AltiVec predicate framework.
36121 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36123 (vector_ne_<mode>_p): Change this define_expand to not rely on
36124 AltiVec predicate framework.
36125 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36127 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36128 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36129 define_insn pattern.
36130 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36131 define_insn pattern because the xvcmpne<VSs>. instruction is not
36133 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36134 instruction is not supported.
36136 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36138 * config/nvptx/nvptx.c: Include intl.h.
36140 2017-03-01 Martin Jambor <mjambor@suse.cz>
36143 * ipa-prop.h (ipa_bits): Removed field known.
36144 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36145 to pointers. Adjusted their comments to warn about their sharing.
36146 (ipcp_transformation_summary): Change bits to a vector of pointers.
36147 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36148 (ipa_get_ipa_bits_for_value): Declare.
36149 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36150 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36151 (ipa_bits_hash_table): Likewise.
36152 (ipa_vr_ggc_hash_traits): Likewise.
36153 (ipa_vr_hash_table): Likewise.
36154 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36155 being pointers and vr_known being removed.
36156 (ipa_set_jf_unknown): Likewise.
36157 (ipa_get_ipa_bits_for_value): New function.
36158 (ipa_set_jfunc_bits): Likewise.
36159 (ipa_get_value_range): New overloaded functions.
36160 (ipa_set_jfunc_vr): Likewise.
36161 (ipa_compute_jump_functions_for_edge): Use the above functions to
36162 construct bits and vr parts of jump functions.
36163 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36164 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36166 (ipcp_grow_transformations_if_necessary): Also allocate
36167 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36169 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36170 them. Fix too long lines.
36171 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36172 vr_known being removed.
36173 (ipa_read_jump_function): Use new setter functions to construct bits
36174 and vr parts of jump functions or set them to NULL.
36175 (write_ipcp_transformation_info): Adjust for bits being pointers.
36176 (read_ipcp_transformation_info): Likewise.
36177 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36179 Include gt-ipa-prop.h.
36180 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36182 (ipcp_store_bits_results): Likewise.
36183 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36184 Do not write to existing jump functions but use a temporary instead.
36186 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36189 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36190 attempt to use its first operand as BIT_FIELD_REF base.
36192 2017-03-01 Richard Biener <rguenther@suse.de>
36194 PR middle-end/79721
36195 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36196 interpolating formula in wrapping arithmetic.
36197 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36199 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36201 PR tree-optimization/79734
36202 * tree-vect-generic.c (expand_vector_condition): Optimize
36203 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36204 Handle VEC_COND_EXPR where comparison has different inner width from
36205 type's inner width.
36207 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36209 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36210 markup, and similar issues. Remove @opindex entries for things
36211 that aren't options. Add missing -mmpy-option entries.
36213 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36215 PR tree-optimization/79737
36216 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36217 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36218 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36219 instead of byte_size. Formatting fix.
36220 (shift_bytes_in_array_right): Formatting fix.
36222 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36225 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36226 condition on optimize for the leaf function test.
36228 2017-02-28 Martin Liska <mliska@suse.cz>
36231 * read-rtl-function.c (function_reader::handle_unknown_directive):
36232 Bail out when one uses -flto.
36234 2017-02-28 Martin Liska <mliska@suse.cz>
36236 * common.opt: Replace space with tabular for options of <number>
36238 * config/i386/i386.opt: Show <number> value for
36239 -mlarge-data-threshold.
36240 * opts.c (print_filtered_help): Do not display number in hexadecimal
36243 2017-02-28 Martin Liska <mliska@suse.cz>
36245 * common.opt: Fix --help=option -Q for options which are of
36248 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36250 * config/i386/i386.c (print_reg): Error out for values
36251 of 8-bit size in invalid integer register.
36253 2017-02-28 Martin Sebor <msebor@redhat.com>
36255 PR tree-optimization/79691
36256 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36258 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36261 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36262 gcc_unreachable with output_operand_lossage.
36264 2017-02-28 Richard Biener <rguenther@suse.de>
36266 PR tree-optimization/79740
36267 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36269 (visit_nary_op): Insert the nary into the hashtable if we
36270 pattern-matched sth.
36271 * tree-ssa-pre.c (eliminate_insert): Robustify.
36273 2017-02-28 Richard Biener <rguenther@suse.de>
36275 PR middle-end/79731
36276 * fold-const.c (decode_field_reference): Reject out-of-bound
36279 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36281 * config/i386/i386.c: Include intl.h.
36282 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36283 instead of just cond ? "..." : "...".
36284 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36285 * coverage.c (read_counts_file): Likewise.
36286 * omp-offload.c: Include intl.h.
36287 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36288 of just cond ? "..." : "...".
36289 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36290 of just cond ? "..." : "...".
36292 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36295 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36297 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36298 'tune for' CPU name.
36299 * config/arm/arm-cpu-data.h: Regenerated.
36301 2017-02-28 Richard Biener <rguenther@suse.de>
36303 PR tree-optimization/79732
36304 * tree-inline.c (expand_call_inline): Do not shadow var.
36306 2017-02-28 Richard Biener <rguenther@suse.de>
36308 PR tree-optimization/79723
36309 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36310 address-space properly.
36312 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36314 * doc/optinfo.texi (Optimization groups): Fix option used for
36316 * doc/invoke.texi (-fopt-info): Document "omp".
36317 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36318 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36319 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36320 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36321 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36323 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36325 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36326 OPTGROUP_OMP with "omp".
36328 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36331 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36332 for arithmetic shift of unsigned V2DI.
36334 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36336 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36337 arc/linux.h headers.
36338 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36339 (LINK_SPEC): Likewise.
36340 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36341 (EXTRA_SPECS): Likewise.
36342 (STARTFILE_SPEC): Likewise.
36343 (ENDFILE_SPEC): Likewise.
36344 (LIB_SPEC): Likewise.
36345 (TARGET_SDATA_DEFAULT): Likewise.
36346 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36347 (MULTILIB_DEFAULTS): Likewise.
36348 (DWARF2_UNWIND_INFO): Likewise.
36349 * config/arc/big.h: New file.
36350 * config/arc/elf.h: Likewise.
36351 * config/arc/linux.h: Likewise.
36352 * config/arc/t-uClibc: Remove.
36354 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36356 PR tree-optimization/77536
36357 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36358 (tree_transform_and_unroll_loop): Use above function to compute the
36359 estimated niter of unrolled loop and use it when scaling profile.
36360 Also use count info rather than frequency if it's non-zero.
36361 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36362 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36363 (vect_transform_loop): Call above function.
36365 2017-02-27 Richard Biener <rguenther@suse.de>
36367 PR tree-optimization/45397
36368 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36369 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36370 (visit_nary_op): Add pattern matching for CSEing sign-changed
36371 or truncated operations with wider ones.
36373 2017-02-27 Richard Biener <rguenther@suse.de>
36375 PR tree-optimization/79690
36376 * tree-vect-stmts.c (vectorizable_store): Use vector type
36377 built from the DR with address-space.
36379 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36381 * doc/invoke.texi (Optimize Options): Refine the description
36382 of asan-use-after-return.
36384 2017-02-25 Alan Modra <amodra@gmail.com>
36386 PR rtl-optimization/79584
36387 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36388 base, not ad->base_term, the reg within base. Remove assertion
36389 that ad->base == ad->base_term. Replace gen_int_mode using
36390 bogus mode with const0_rtx.
36392 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36394 PR middle-end/79396
36395 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36396 FMA_EXPR like tcc_binary or tcc_unary.
36398 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36401 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36403 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36404 (output_loc_operands): Handle DW_OP_call_ref and
36405 DW_OP_GNU_variable_value.
36406 (struct variable_value_struct): New type.
36407 (struct variable_value_hasher): Likewise.
36408 (variable_value_hash): New variable.
36409 (string_types): Remove.
36410 (copy_loc_descr): New function.
36411 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36412 (prepend_loc_descr_to_each): New function.
36413 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36414 instead of add_loc_descr_to_each if the first argument is single
36415 location list and the second has multiple.
36416 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36417 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36418 when looking for variable value which doesn't have other location info.
36419 (loc_list_from_tree): Formatting fix.
36420 (gen_array_type_die): Simplify DW_AT_string_length handling.
36421 (adjust_string_types): Remove.
36422 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36423 string_types. Call resolve_variable_values.
36424 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36425 (resolve_addr_in_expr): Likewise. Add A argument.
36426 (copy_deref_exprloc): Remove deref argument. Adjust for the
36427 original expression being DW_OP_GNU_variable_value with optionally
36428 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36429 optionally after it.
36430 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36431 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36432 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36433 (variable_value_hasher::hash, variable_value_hasher::equal): New
36435 (resolve_variable_value_in_expr, resolve_variable_value,
36436 resolve_variable_values, note_variable_value_in_expr,
36437 note_variable_value): New functions.
36438 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36441 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36444 * opts.h (handle_generated_option): Add GENERATED_P argument.
36445 * opts-common.c (handle_option): Adjust function comment.
36446 (handle_generated_option): Add GENERATED_P argument, pass it to
36448 (control_warning_option): Pass false to handle_generated_option
36450 * opts.c (maybe_default_option): Pass true to handle_generated_option
36452 * optc-gen.awk: Likewise.
36454 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36456 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36457 a REG, look at the REG it is a SUBREG of.
36458 (splitter for cmpeqsi_t): Ditto.
36460 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36462 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36463 the special USEs with the pattern of the insn, not the insn itself.
36465 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36468 * doc/invoke.texi: Document -mload-store-pairs.
36470 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36471 Sandra Loosemore <sandra@codesourcery.com>
36473 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36474 argument isn't a CONST_INT.
36475 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36476 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36477 (nios2_validate_compare): Bypass alternate compare logic if *op2
36478 is not a CONST_INT.
36479 (ldstwm_operation_p): Return false if first_base is not a REG or
36480 if first_offset is not a CONST_INT.
36482 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36484 * config/cris/cris.md: Use correct operand in a define_peephole2.
36486 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36488 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36490 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36492 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36493 this_insn if it is an INSN or JUMP_INSN.
36494 (force_offsettable): Look at base, not at addr.
36495 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36496 on things that aren't necessarily CONST_INTs.
36498 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36500 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36501 -mfpmath=sse is the default also for x86-32 targets with SSE2
36502 instruction set when @option{-ffast-math} is enabled
36504 2017-02-24 Jeff Law <law@redhat.com>
36506 PR rtl-optimizatoin/79286
36507 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36510 2017-02-24 Richard Biener <rguenther@suse.de>
36512 PR tree-optimization/79389
36513 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36516 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36518 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36519 function comment to reflect reality.
36520 (loop_exits_before_overflow): Fix typo in function description.
36522 2017-02-24 Richard Biener <rguenther@suse.de>
36524 PR tree-optimization/79389
36525 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36526 properly that a threading opportunity exists. Detect conditional
36527 copy/constant propagation opportunities.
36529 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36531 * config/visium/visium.md (type): Add trap.
36532 (b): New mode attribute.
36533 (*btst): Rename into...
36534 (*btst<mode>): ...this and adjust.
36535 (*cbranchsi4_btst_insn): Rename into...
36536 (*cbranch<mode>4_btst_insn): ...this and adjust.
36537 (trap): New define_insn.
36539 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36541 PR tree-optimization/79389
36542 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36543 (noce_reversed_cond_code): New function.
36544 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36545 reversed_comparison_code. Formatting fix.
36546 (noce_try_store_flag): Test rev_cond != NULL in addition to
36547 reversed_comparison_code.
36548 (noce_try_store_flag_constants): Likewise.
36549 (noce_try_store_flag_mask): Likewise.
36550 (noce_try_addcc): Use rev_cond if non-NULL instead of
36551 reversed_comparison_code.
36552 (noce_try_cmove_arith): Likewise. Formatting fixes.
36553 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36554 (noce_find_if_block): Initialize rev_cond.
36555 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36556 instead of false as last argument never attempt to reverse it
36559 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36561 PR tree-optimization/79663
36562 * tree-predcom.c (combine_chains): Process refs in reverse order
36563 only for ZERO length chains, and add explaining comment.
36565 2017-02-23 Jeff Law <law@redhat.com>
36567 PR tree-optimization/79578
36568 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36569 in call to operand_equal_p.
36571 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36574 * config/i386/cpuid.h: Fix another undefined behavior.
36576 2017-02-23 Richard Biener <rguenther@suse.de>
36578 PR tree-optimization/79683
36579 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36580 vector types for data-refs.
36582 2017-02-23 Martin Liska <mliska@suse.cz>
36584 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36586 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36588 PR middle-end/79665
36589 * internal-fn.c (get_range_pos_neg): Moved to ...
36590 * tree.c (get_range_pos_neg): ... here. No longer static.
36591 * tree.h (get_range_pos_neg): New prototype.
36592 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36593 are known to be in between 0 and signed maximum inclusive, try to
36594 expand both unsigned and signed divmod and use the cheaper one from
36597 2017-02-22 Jeff Law <law@redhat.com>
36599 PR tree-optimization/79578
36600 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36601 to compare base operands.
36603 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36606 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36607 gpc_reg_operand instead of fpr_reg_operand.
36609 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36611 * config/mips/mips.c (mips_return_in_memory): Force FP
36612 vector types to be returned in memory for o32 ABI.
36614 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36616 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36617 instead of DW_TAG_member for static data member declarations and don't
36618 set no_linkage_name for static inline data members.
36619 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36622 2017-02-22 Martin Liska <mliska@suse.cz>
36624 * doc/invoke.texi: Replace inequality signs with square brackets
36627 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36630 * lra-constraints.c (simplify_operand_subreg): Handle
36631 WORD_REGISTER_OPERATIONS targets.
36633 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36636 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36637 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36638 elimination by swapping fld*.
36640 2017-02-22 Richard Biener <rguenther@suse.de>
36642 PR tree-optimization/79673
36643 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36644 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36645 irrelevant address-space qualifiers and avoiding a
36646 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36648 2017-02-22 Richard Biener <rguenther@suse.de>
36650 PR tree-optimization/79666
36651 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36652 to not symbolically negate if that may introduce undefined
36655 2017-02-22 Martin Liska <mliska@suse.cz>
36658 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36659 * data-streamer-out.c (streamer_write_gcov_count_stream):
36661 * value-prof.c (stream_out_histogram_value): Make assert more
36662 precise based on type of counter.
36664 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36667 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36668 Use nonimmediate_operand instead of memory_operand for operand 1.
36669 (float-extend standard_x87sse_constant_load splitter): Ditto.
36671 2017-02-21 Jeff Law <law@redhat.com>
36673 PR tree-optimization/79621
36674 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36675 blocks with edges to themselves.
36677 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36680 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36681 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36682 Use gimple_call_builtin_p.
36685 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36686 on temporarily removed DEBUG_INSNs.
36688 PR tree-optimization/79649
36689 * tree-loop-distribution.c (classify_partition): Give up on
36690 non-generic address space loads/stores.
36692 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36694 * doc/loop.texi (Loop manipulation): Remove nonexistent
36695 tree_ssa_loop_version from the documentation.
36696 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36698 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36701 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36702 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36703 * config/rs6000/rs6000.c: Include except.h.
36704 (rs6000_expand_split_stack_prologue): Call
36705 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36707 2017-02-21 Martin Jambor <mjambor@suse.cz>
36710 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36711 have been analyzed.
36713 2017-02-21 Martin Jambor <mjambor@suse.cz>
36715 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36716 for backward compatibility only.
36717 * doc/invoke.texi (Option Summary): Remove all references to
36718 -fipa-cp-alignment.
36720 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36724 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36726 * lra-constraints.c (curr_insn_transform): Handle
36727 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36729 2017-02-21 Martin Liska <mliska@suse.cz>
36731 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36733 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36735 PR translation/79638
36736 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36738 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36741 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36742 (arm_function_ok_for_sibcall): Return false for an indirect call by
36743 descriptor if all the argument registers are used.
36744 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36745 alignment of the function.
36747 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36749 PR tree-optimization/61441
36750 * simplify-rtx.c (simplify_const_unary_operation): For
36751 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36752 the sNaN unmodified.
36754 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36756 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36757 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36758 instead of SYSTEM_HEADER_DIR.
36760 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36761 Martin Liška <mliska@suse.cz>
36763 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36764 Fix typos and grammar, use active voice, and clarify.
36766 2017-02-20 Marek Polacek <polacek@redhat.com>
36768 PR middle-end/79537
36769 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36772 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36774 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36777 * config/i386/i386.c (ix86_expand_builtin): Handle
36778 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36779 ix86_builtins_isa[fcode].isa as a requirement of those
36780 flags and any other flag in the bitmask.
36781 (ix86_init_mmx_sse_builtins): Use 0 instead of
36782 ~OPTION_MASK_ISA_64BIT as mask.
36783 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36784 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36785 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36786 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36788 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36791 * lra-constraints.c (split_reg): Check requested split mode
36792 is supported by the register.
36794 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36796 * lra-constraints.c (simplify_operand_subreg): Remove early
36799 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36802 * lra-constraints.c (curr_insn_transform): Tighten condition
36803 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36805 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36808 * lra-constraints.c (curr_insn_transform): Handle
36809 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36811 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36814 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36816 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36818 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36821 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36824 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36826 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36829 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36831 * final.c (last_columnnum, override_columnnum): New variables.
36832 (final_start_function): Set last_columnnum, pass it to begin_prologue
36833 hook and pass 0 to dwarf2out_begin_prologue.
36834 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36835 to source_line debug hook.
36836 (notice_source_line): Compute last_columnnum and for debug_column_info
36837 return true on column changes.
36838 * debug.h (struct gcc_debug_hooks): Add column argument to
36839 source_line and begin_prologue hooks.
36840 (debug_nothing_int_charstar_int_bool): Remove prototype.
36841 (debug_nothing_int_int_charstar,
36842 debug_nothing_int_int_charstar_int_bool): New prototypes.
36843 (dwarf2out_begin_prologue): Add column argument.
36844 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36845 begin_prologue hooks.
36846 (debug_nothing_int_charstar_int_bool): Remove.
36847 (debug_nothing_int_int_charstar,
36848 debug_nothing_int_int_charstar_int_bool): New functions.
36849 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36850 through to dwarf2out_source_line.
36851 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36852 (dwarf2out_source_line): Add column argument, emit it if requested.
36853 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36855 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36856 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36857 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36858 through to dwarf2out_begin_prologue.
36859 (vmsdbgout_source_line): Add column argument, pass it through to
36860 dwarf2out_source_line.
36861 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36862 dbxout_source_line caller.
36863 (dbxout_source_line): Add column argument.
36865 * common.opt (gno-column-info, gcolumn-info): New options.
36866 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36867 (check_die): Also test for multiple DW_AT_decl_column attributes.
36868 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36869 DW_AT_decl_column if requested.
36870 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36872 (gen_variable_die): Likewise.
36873 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36874 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36877 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36878 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36879 (ix86_handle_option): Handle OPT_m3dnowa.
36880 * doc/invoke.texi (-m3dnowa): Document.
36881 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36882 -m3dnowa instead of -m3dnow -march=athlon.
36885 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36886 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36888 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36891 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36892 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36893 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36894 generator for vsx_xxpermdi_<mode>_be.
36895 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36896 force big-endian semantics.
36897 (vsx_xxpermdi_<mode>_be): New define_expand with same
36898 implementation as previous version of vsx_xxpermdi_<mode>.
36900 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36902 PR tree-optimization/79327
36903 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36904 variable, its initialization and use.
36906 2017-02-17 Julia Koval <julia.koval@intel.com>
36908 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36909 (OPTION_MASK_ISA_PKU_UNSET): New.
36910 (ix86_handle_option): Handle -mrdpid.
36911 * config/i386/cpuid.h (bit_RDPID): New.
36912 * config/i386/driver-i386.c (host_detect_local_cpu):
36913 Detect RDPID feature.
36914 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36915 * config/i386/i386-c.c (ix86_target_macros_internal):
36917 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36918 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36919 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36920 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36921 * config/i386/i386.md (define_insn "rdpid"): New.
36922 * config/i386/i386.opt Add -mrdpid.
36923 * config/i386/immintrin.h (_rdpid_u32): New.
36925 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36927 PR rtl-optimization/79541
36928 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36929 instead of transforming it into USE.
36931 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36933 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36934 If HONOR_SNANS (SFmode) force the input to a register.
36935 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36936 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36937 an frsp or similar insn.
36939 2017-02-17 Martin Liska <mliska@suse.cz>
36941 PR rtl-optimization/79577
36942 * params.def (selsched-max-sched-times): Increase minimum to 1.
36944 2017-02-17 Martin Liska <mliska@suse.cz>
36946 PR rtl-optimization/79574
36947 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36949 2017-02-17 Martin Liska <mliska@suse.cz>
36951 PR tree-optimization/79529
36952 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36953 ssa_defined_default_def_p to handle cases which are implicitly
36955 * tree-ssa.c (ssa_defined_default_def_p): New function.
36956 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36957 which are implicitly defined.
36958 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36960 2017-02-17 Richard Biener <rguenther@suse.de>
36962 PR middle-end/79576
36963 * params.def (max-ssa-name-query-depth): Limit to 10.
36965 2017-02-17 Richard Biener <rguenther@suse.de>
36967 PR tree-optimization/79552
36968 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36971 2017-02-17 Richard Biener <rguenther@suse.de>
36974 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36976 2017-02-17 Marek Polacek <polacek@redhat.com>
36978 PR middle-end/79536
36979 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36980 (fold_negate_expr): New wrapper.
36982 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36984 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36985 Correct terminology and de-emphasize pre-standard behavior.
36987 2017-02-16 Alan Modra <amodra@gmail.com>
36989 PR rtl-optimization/79286
36990 * ira.c (def_dominates_uses): New function.
36991 (update_equiv_regs): Don't create an equivalence for insns that
36992 may trap where the register def does not dominate the use.
36994 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36996 PR rtl-optimization/78127
36997 * lra.c (lra): Call lra_eliminate before finish the loop after
37000 2017-02-16 Richard Biener <rguenther@suse.de>
37002 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37004 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37005 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37006 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37007 (isl_val_int_from_wi): New function.
37008 (extract_affine_gmp): Rename to ...
37009 (extract_affine_wi): ... this, take a widest_int.
37010 (extract_affine_int): Just wrap extract_affine_wi.
37011 (add_param_constraints): Use isl_val_int_from_wi.
37012 (add_loop_constraints): Likewise, and extract_affine_wi.
37014 2017-02-15 Jeff Law <law@redhat.com>
37016 PR middle-end/79521
37017 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37018 ira_init_register_move_cost_if_necessary.
37020 2017-02-15 Martin Sebor <msebor@redhat.com>
37022 PR middle-end/32003
37023 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37024 removed in a prior commit.
37026 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37028 PR tree-optimization/79347
37029 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37030 counters during peeling.
37032 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37034 * Makefile.in (site.exp): Remove "set ISLVER".
37036 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37039 * real.c (real_from_integer): Call real_convert even for decimal.
37041 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37044 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37046 2017-02-14 Andrew Pinski <apinski@cavium.com>
37048 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37049 cores and change the partno/implementer to be correct.
37050 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37051 the 'B" as the implementer.
37052 * config/aarch64/aarch64-tune.md: Regenerate.
37054 2017-02-14 Carl Love <cel@us.ibm.com>
37056 * config/rs6000/rs6000.c: Add case statement entry to make the
37057 xvcvuxdsp built-in argument unsigned.
37058 * config/rs6000/vsx.md: Fix the source and return operand types so they
37059 match the instruction definitions from the ISA document. Fix typo
37060 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37063 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37066 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37067 member early_clobber_alts.
37068 * lra-lives.c (reg_early_clobber_p): New.
37069 (process_bb_lives): Use it.
37070 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37071 (debug_operand_data): Initialize early_clobber_alts.
37072 (setup_operand_alternative): Set up early_clobber_alts.
37073 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37074 alternatives to new_insn_reg.
37075 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37077 (lra_update_insn_regno_info): Pass the new arg.
37079 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37081 PR middle-end/79505
37082 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37083 (new_oacc_loop_raw): Don't clear already cleared fields.
37086 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37087 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37088 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37090 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37093 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37095 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37098 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37099 the extra instruction to the right place to store 128-bit constant
37102 2017-02-14 Martin Sebor <msebor@redhat.com>
37104 PR middle-end/79448
37105 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37106 warning for strings of unknown length.
37108 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37110 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37112 2017-02-14 Jeff Law <law@redhat.com>
37115 * ira-costs.c (scan_one_insn): Initialize register move costs
37116 for pseudos seen in USE/CLOBBER insns.
37118 PR tree-optimization/79095
37119 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37120 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37121 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37122 if the operands are known to be not equal, then the resulting range
37124 (intersect_ranges): If the new range is ~[0,0] and the old range is
37125 wide, then prefer ~[0,0].
37126 * tree-vrp.c (overflow_comparison_p_1): New function.
37127 (overflow_comparison_p): New function.
37128 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37129 if NAME is used in an overflow test.
37130 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37131 overflow check that can be expressed as an equality test, then adjust
37132 ops to be that equality test.
37134 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37136 * config/s390/s390-builtin-types.def: Remove flags argument.
37137 * config/s390/s390.c (s390_init_builtins): Likewise.
37139 2017-02-14 Martin Liska <mliska@suse.cz>
37141 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37142 vector. Fix trailing white spaces.
37144 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37146 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37149 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37151 PR rtl-optimization/68664
37152 * config/arm/arm.c (arm_sched_can_speculate_insn):
37153 New function. Declare prototype.
37154 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37156 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37158 PR rtl-optimization/68664
37159 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37161 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37163 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37165 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37166 max skip bytes for function, loop and jump.
37168 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37170 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37171 ABS_EXPR for gimple dump.
37173 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37176 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37178 PR tree-optimization/79408
37179 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37180 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37181 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37182 also if rhs1 is INTEGER_CST.
37184 2017-02-14 Richard Biener <rguenther@suse.de>
37186 PR middle-end/79432
37187 * tree-into-ssa.c (insert_phi_nodes): When the function can
37188 have abnormal edges rewrite SSA names with broken use-def
37189 dominance out of SSA and register them for PHI insertion.
37191 2017-02-13 Martin Sebor <msebor@redhat.com>
37193 PR middle-end/79496
37194 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37195 clearing info.nowrite flag when snprintf size argument is a range.
37197 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37199 * cprop.c (cprop_jump): Add missing space in string literal.
37200 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37201 (get_constraint_for_component_ref): Likewise.
37202 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37203 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37204 * lra-constraints.c (process_alt_operands): Likewise.
37205 * ipa-inline.c (inline_small_functions): Likewise.
37206 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37207 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37208 * trans-mem.c (diagnose_tm_1_op): Likewise.
37209 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37210 (grid_parallel_clauses_gridifiable): Likewise.
37212 * config/nvptx/mkoffload.c (process): Add space in between
37215 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37216 "MOD4_SSE_REGS" and "ALL_REGS".
37218 * spellcheck.c (test_data): Add , in between "foo" and "food".
37220 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37223 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37224 boundary crossing check and subsequent code generation agree.
37226 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37228 * config/aarch64/aarch64.c (has_memory_op): Delete.
37229 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37232 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37234 PR rtl-optimization/79388
37235 PR rtl-optimization/79450
37236 * combine.c (distribute_notes): When removing TEM_INSN for which
37237 corresponding dest has last value recorded, invalidate that last
37240 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37242 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37243 of explicit '@'. Add missing assembly comment marker on branch costs
37246 2017-02-13 Nathan Sidwell <nathan@acm.org>
37248 * gengtype-lex.l (<in_struct>): Add '/'.
37250 2017-02-13 Martin Liska <mliska@suse.cz>
37253 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37255 2017-02-13 Richard Biener <rguenther@suse.de>
37257 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37259 * configure: Re-generate.
37260 * config.in: Likewise.
37261 * graphite-dependences.c: Simplify as if
37262 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37263 * graphite-isl-ast-to-gimple.c: Likewise.
37264 * graphite-optimize-isl.c: Likewise.
37265 * graphite-poly.c: Likewise.
37266 * graphite-sese-to-poly.c: Likewise.
37267 * graphite.h: Likewise.
37268 * toplev.c: Include isl/version.h and use isl_version () for
37269 printing the ISL version.
37270 * doc/install.texi: Update ISL requirement.
37272 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37274 * doc/standards.texi (Standards): Update reference to
37277 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37279 * doc/extend.texi (Named Address Spaces): sourceware.org now
37281 * doc/install.texi (Binaries): Ditto.
37284 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37286 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37287 standard terminology "stringize"/"stringizing" throughout.
37288 * doc/cppinternals.texi: Likewise.
37290 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37292 * doc/extend.texi: Fix some spelling mistakes and typos.
37293 * doc/invoke.texi: Likewise.
37295 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37298 * params.def (inline-min-speedup) Change from 10 to 8.
37300 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37302 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37305 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37308 * ipa-inline-analysis.c (get_minimal_bb): New function.
37309 (record_modified): Use it.
37310 (remap_edge_change_prob): Handle also ancestor functions.
37312 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37314 * doc/contrib.texi (Contributors): Remove broken link into
37315 the Mauve CVS repository.
37317 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37319 PR middle-end/79454
37320 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37321 result computation whenever lhs doesn't have vector mode, not
37322 just when it has BLKmode.
37324 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37326 * doc/makefile.texi (profiledbootstrap): Refer to the
37327 installation instructions only in textual form.
37329 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37332 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37334 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37336 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37337 (Specific): Update mingw-w64 reference.
37339 (Specific): Remove broken link to Renesas RX processor.
37341 2017-02-10 Richard Biener <rguenther@suse.de>
37343 * toplev.c (process_options): Do not mention obsolete graphite
37344 options when printing sorry message about missing graphite support.
37345 Mention -floop-nest-optimize.
37347 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37349 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37350 (vtst_p16): Likewise.
37351 (vtstq_p8): Likewise.
37352 (vtstq_p16): Likewise.
37354 (vtstq_p64): Likewise.
37355 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37356 (vset_lane_p64): New.
37357 (vsetq_lane_p64): New.
37359 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37361 PR tree-optimization/79411
37362 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37363 stmt operands are SSA_NAMEs used in abnormal phis.
37364 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37367 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37370 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37373 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37375 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37377 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37379 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37383 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37384 not arbitrary TREE_CONSTANT.
37387 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37388 "omp declare target link" attribute unless is_global_var.
37389 * omp-offload.c (find_link_var_op): Likewise.
37391 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37392 Chung-Lin Tang <cltang@codesourcery.com>
37394 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37396 (gimplify_adjust_omp_clauses): Don't delete TILE.
37397 (gimplify_omp_for): Deal with TILE.
37398 * internal-fn.c (expand_GOACC_TILE): New function.
37399 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37401 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37402 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37404 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37405 avoid DIV for outermost collapse var.
37406 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37407 Remove out of date comments, fix whitespace.
37408 * omp-general.c (omp_extract_for_data): Deal with tiling.
37409 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37410 adjust OLF_DIM_BASE value.
37411 (struct omp_for_data): Add tiling field.
37412 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37413 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37414 for auto loops. Remove default auto determining, moved to
37415 oacc_loop_fixed_partitions.
37416 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37417 stmts, add e_mask field.
37418 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37419 (oacc_thread_numbers): Use oacc_dim_call.
37420 (oacc_xform_tile): New.
37421 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37422 (finish_oacc_loop): Adjust for ifns vector.
37423 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37424 add case for GOACC_TILE fns.
37425 (oacc_loop_xform_loop): Delete.
37426 (oacc_loop_process): Iterate over call list directly, and add
37427 handling for GOACC_TILE fns.
37428 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37430 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37431 vector partitioning to outer loops. Assign 2 partitions to loops
37432 when available. Add TILE handling.
37433 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37434 (execite_oacc_device_lower): Process GOACC_TILE fns,
37435 ignore unknown specs.
37436 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37437 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37438 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37440 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37442 * configure.ac (ACX_BUGURL): Update.
37443 * configure: Regenerate.
37445 2017-02-09 Richard Biener <rguenther@suse.de>
37447 PR tree-optimization/69823
37448 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37449 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37451 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37453 * config/arc/arc-c.def: Add __NPS400__ definition.
37454 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37455 (TARGET_NPS400): Define.
37457 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37459 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37461 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37462 pointer, arch_info.
37463 (arc_cpu_types): Fill the arch_info field with a pointer into the
37464 arc_arch_types table.
37465 (arc_selected_cpu): Declare.
37466 * config/arc/arc.c (arc_selected_cpu): Make global.
37467 (arc_selected_arch): Delete.
37468 (arc_base_cpu): Delete.
37469 (arc_override_options): Remove references to deleted variables,
37470 update access to arch information.
37471 (ARC_OPT): Update access to arch information.
37472 (ARC_OPTX): Likewise.
37473 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37474 (TARGET_ARC600): Update access to arch information.
37475 (TARGET_ARC601): Likewise.
37476 (TARGET_ARC700): Likewise.
37477 (TARGET_EM): Likewise.
37478 (TARGET_HS): Likewise.
37479 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37482 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37485 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37486 condition/operands for integer GE/LE/GEU/LEU operations.
37488 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37490 PR translation/79397
37491 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37494 2017-02-08 Martin Jambor <mjambor@suse.cz>
37497 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37498 whether allocation happened.
37499 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37500 nothing was allocated.
37502 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37504 PR tree-optimization/79408
37505 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37506 constant, but SSA_NAME with a known integer range, use the minimum
37507 of that range instead of op1 to determine if modulo can be replaced
37508 with its first operand.
37510 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37512 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37514 2017-02-08 Richard Biener <rguenther@suse.de>
37516 PR tree-optimization/71824
37517 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37518 Check all loops contained in the merged region.
37520 2017-02-07 Andrew Pinski <apinski@cavium.com>
37522 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37524 2017-02-07 Andrew Pinski <apinski@cavium.com>
37526 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37527 (thunderxt88): Likewise.
37528 (thunderxt81): Disable LSE and change v8.1 to v8.
37529 (thunderxt83): Likewise.
37531 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37532 Richard Biener <rguenther@suse.de>
37534 PR middle-end/79399
37535 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37536 type from int to size_t.
37537 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37539 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37541 PR rtl-optimization/79386
37542 * cprop.c (bypass_conditional_jumps): Initialize
37543 bypass_last_basic_block already before splitting bbs after
37544 unconditional traps...
37545 (bypass_conditional_jumps): ... rather than here.
37548 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37549 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37550 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37551 fix -masm=intel patterns.
37553 2017-02-07 Richard Biener <rguenther@suse.de>
37555 PR tree-optimization/79256
37556 PR middle-end/79278
37557 * builtins.c (get_object_alignment_2): Use min_align_of_type
37558 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37559 and ADJUST_FIELD_ALIGN.
37561 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37563 * doc/tm.texi: Regenerate.
37564 * stor-layout.c (layout_decl): Adjust.
37565 (update_alignment_for_field): Likewise.
37566 (place_field): Likewise.
37567 (min_align_of_type): Likewise.
37568 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37569 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37570 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37571 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37572 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37573 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37574 * config/i386/i386.c (x86_field_alignment): Likewise.
37575 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37576 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37577 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37578 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37579 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37580 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37584 2017-01-30 Richard Biener <rguenther@suse.de>
37586 PR tree-optimization/79256
37587 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37588 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37591 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37593 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37594 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37595 builtins to SImode and emit a zero-extend, if necessary.
37597 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37599 * docs/invoke.texi (RISC-V Options): Alphabetize.
37601 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37603 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37606 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37608 * config/riscv/riscv.c: New file.
37609 * common/config/riscv/riscv-common.c: Likewise.
37610 * config.gcc: Likewise.
37611 * config/riscv/constraints.md: Likewise.
37612 * config/riscv/elf.h: Likewise.
37613 * config/riscv/generic.md: Likewise.
37614 * config/riscv/linux.h: Likewise.
37615 * config/riscv/multilib-generator: Likewise.
37616 * config/riscv/peephole.md: Likewise.
37617 * config/riscv/pic.md: Likewise.
37618 * config/riscv/predicates.md: Likewise.
37619 * config/riscv/riscv-builtins.c: Likewise.
37620 * config/riscv/riscv-c.c: Likewise.
37621 * config/riscv/riscv-ftypes.def: Likewise.
37622 * config/riscv/riscv-modes.def: Likewise.
37623 * config/riscv/riscv-opts.h: Likewise.
37624 * config/riscv/riscv-protos.h: Likewise.
37625 * config/riscv/riscv.h: Likewise.
37626 * config/riscv/riscv.md: Likewise.
37627 * config/riscv/riscv.opt: Likewise.
37628 * config/riscv/sync.md: Likewise.
37629 * config/riscv/t-elf-multilib: Likewise.
37630 * config/riscv/t-linux: Likewise.
37631 * config/riscv/t-linux-multilib: Likewise.
37632 * config/riscv/t-riscv: Likewise.
37633 * configure.ac: Likewise.
37634 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37635 Waterman as RISC-V maintainers.
37636 * doc/install.texi: Add RISC-V entries.
37637 * doc/invoke.texi: Add RISC-V options section.
37638 * doc/md.texi: Add RISC-V constraints section.
37639 * configure: Regenerated.
37641 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37644 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37645 false values to be constant vectors with all 0 or all 1 bits set.
37646 (vcondu<mode><mode>): Likewise.
37647 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37649 (fpmask_comparison_operator): Update comment.
37650 (vecint_comparison_operator): New predicate.
37651 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37652 vector conditionals when the true and false values are constant
37653 vectors with all 0 bits or all 1 bits set.
37655 2017-02-06 Martin Sebor <msebor@redhat.com>
37657 PR tree-optimization/79376
37658 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37660 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37662 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37663 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37664 to simplify split condition.
37666 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37668 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37669 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37672 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37674 PR rtl-optimization/68664
37675 * target.def (can_speculate_insn): New hook.
37676 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37677 * doc/tm.texi: Regenerate.
37678 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37679 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37680 (rs6000_sched_can_speculate_insn): New function.
37682 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37684 PR tree-optimization/79284
37685 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37686 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37687 vectorizable_mask_load_store, vectorizable_operation,
37688 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37689 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37690 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37691 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37692 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37693 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37694 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37695 is_gimple_assign (stmt). Replace another such test with
37696 is_gimple_assign (stmt).
37698 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37701 * config/avr/avr.c (rtl-iter.h): Include it.
37702 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37703 (avr_legitimate_combined_insn): ...and implementation.
37705 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37707 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37708 * config/s390/s390.c (s390_const_operand_ok)
37709 (s390_canonicalize_comparison, s390_extract_part)
37710 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37711 (s390_contiguous_bitmask_p, s390_rtx_costs)
37712 (legitimize_pic_address): Likewise.
37713 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37714 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37715 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37716 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37717 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37719 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37721 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37722 REGNO($0) == REGNO($1).
37724 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37726 * config/s390/linux.h(SIZE_TYPE): Add comment.
37728 2017-02-06 Julian Brown <julian@codesourcery.com>
37729 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37730 Virendra Pathak <virendra.pathak@broadcom.com>
37732 * config/aarch64/aarch64-cores.def: Change the scheduler
37734 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37735 * config/aarch64/thunderx2t99.md: New file.
37737 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37739 * doc/standards.texi (Go Language): Update link to language
37742 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37744 * tree-eh.c (lower_resx): Sanitize profile.
37745 (cleanup_empty_eh_move_lp): Likewise.
37747 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37750 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37752 * cfgloopmanip.h (loop_version): Update prototype.
37753 * modulo-sched.c (sms_schedule): Update call of loop_version.
37754 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37755 * tree-parloops.c (gen_parallel_loop): Likewise.
37756 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37757 * tree-ssa-loop-split.c (split_loop): Likewise.
37758 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37759 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37761 2017-02-05 Martin Liska <mliska@suse.cz>
37764 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37766 (print_operand_address): Initialize a struct to zero.
37768 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37770 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37771 garbage collector only in textual form.
37773 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37775 * doc/extend.texi (x86 specific memory model extensions for
37776 transactional memory): Simplify a phrase.
37778 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37781 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37782 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37783 (atomic_storedi_1): Likewise.
37785 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37787 PR tree-optimization/79338
37788 * tree-parloops.c (gather_scalar_reductions): Don't call
37789 vect_analyze_loop_form for loop->inner before destroying loop's
37792 2017-02-03 Martin Sebor <msebor@redhat.com>
37794 PR tree-optimization/79327
37795 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37796 when precision has resulted in leading zeros.
37797 (format_integer): Adjust the likely counter to assume an unknown
37798 argument that may be zero is non-zero.
37800 2017-02-03 Jason Merrill <jason@redhat.com>
37803 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37804 avoid copying non-taken branch.
37806 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37808 PR tree-optimization/79340
37809 * tree-vect-loop.c (vectorizable_reduction): Release
37810 vec_defs elements after safe_splicing them into other vectors.
37813 PR tree-optimization/79327
37814 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37815 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37817 (format_integer): Use wide_int_to_tree instead of build_int_cst
37818 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37819 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37820 of shortest and longest sequence.
37822 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37824 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37825 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37827 2017-02-03 Walter Lee <walt@tilera.com>
37830 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37831 after initial stackframe link reg save.
37832 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37834 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37837 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37838 wu for stxssp alternative.
37840 2017-02-03 Martin Sebor <msebor@redhat.com>
37842 PR tree-optimization/79352
37843 * gimple-fold.c (get_range_strlen): Add argument.
37844 (get_range_strlen): Change return type to bool.
37845 (get_maxval_strlen): Pass in a dummy argument.
37846 * gimple-fold.h (get_range_strlen): Change return type to bool.
37847 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37848 * tree.h (array_at_struct_end_p): Add argument.
37849 * tree.c (array_at_struct_end_p): Handle it.
37851 2017-02-03 Martin Liska <mliska@suse.cz>
37854 * multiple_target.c (create_dispatcher_calls): Redirect edge
37855 from a caller of a dispatcher.
37856 (expand_target_clones): Make the clones local.
37857 (ipa_target_clone): Do both target clones and resolvers.
37858 (ipa_dispatcher_calls): Remove the pass.
37859 (pass_dispatcher_calls::gate): Likewise.
37860 (make_pass_dispatcher_calls): Likewise.
37861 * passes.def (pass_target_clone): Put as very first IPA early
37864 2017-02-03 Martin Liska <mliska@suse.cz>
37866 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37867 in case of a function with ifunc attribute.
37869 2017-02-03 Martin Liska <mliska@suse.cz>
37871 * cgraph.c (cgraph_node::dump): Dump function version info.
37872 * symtab.c (symtab_node::dump_base): Add missing new line.
37874 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37876 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37877 (ifcombine_ifandif): Use it.
37879 2017-02-03 Martin Liska <mliska@suse.cz>
37881 * doc/invoke.texi: Document default value for
37882 use-after-scope-direct-emission-threshold.
37884 2017-02-03 Martin Liska <mliska@suse.cz>
37886 PR tree-optimization/79339
37887 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37888 (format_floating): Likewise.
37890 2017-02-03 Martin Liska <mliska@suse.cz>
37893 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37895 (ipa_node_params_t::remove): Likewise.
37896 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37897 initialization from removed ipa_node_params_t::insert.
37898 (ipa_node_params::~ipa_node_params): Move from removed
37899 ipa_node_params_t::release.
37900 * symbol-summary.h (symbol_summary::m_released): New member.
37901 Do not release a summary twice. Do not allow to call finalizer
37902 for types of a summary that live in GGC memory.
37904 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37906 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37909 2017-02-02 Martin Sebor <msebor@redhat.com>
37911 PR middle-end/79275
37912 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37913 (format_string): Tighten up the range of output for non-constant
37914 strings and correct the expected range for wide non-constant strings.
37916 2017-02-02 Martin Sebor <msebor@redhat.com>
37918 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37920 PR middle-end/32003
37921 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37923 (-fdump-tree-@var): Add to index and document how to come up
37924 with pass-specific option and dump file names.
37925 (-fdump-passes): Clarify where to look for output.
37927 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37929 PR middle-end/77445
37930 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37931 statistics of the analyzed path; allow threading for speed when
37932 any of BBs along the path are optimized for speed.
37934 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37936 PR middle-end/78468
37937 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37938 settings of the virtual registers.
37941 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37943 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37944 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37947 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37949 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37950 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37952 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37954 * config/s390/s390.md: Add missing comments with the expanded
37956 * config/s390/vector.md: Likewise.
37957 * config/s390/vx-builtins.md: Likewise.
37959 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37962 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37963 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37964 conditions on a single line.
37966 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37968 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37969 __S390_VX__ to __VX__.
37971 2017-02-01 Andrew Pinski <apinski@cavium.com>
37973 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37974 stmt_info to record_stmt_cost.
37975 (vect_get_known_peeling_cost): Pass stmt_info if known to
37977 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37978 cpu_vector_cost field into
37979 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37980 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37981 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37982 splitting of scalar_stmt_cost and vec_stmt_cost.
37983 (thunderx_vector_cost): Likewise.
37984 (cortexa57_vector_cost): LIkewise.
37985 (exynosm1_vector_cost): Likewise.
37986 (xgene1_vector_cost): Likewise.
37987 (thunderx2t99_vector_cost): Improve after the splitting of the two
37989 (aarch64_builtin_vectorization_cost): Update for the splitting of
37990 scalar_stmt_cost and vec_stmt_cost.
37992 2017-02-01 Torvald Riegel <triegel@redhat.com>
37993 Richard Henderson <rth@redhat.com>
37995 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37996 conditional on existance of a fast atomic load.
37997 * optabs-query.c (can_atomic_load_p): New function.
37998 * optabs-query.h (can_atomic_load_p): Declare it.
37999 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38000 no fast atomic load is available for the particular size of access.
38001 (expand_atomic_compare_and_swap): Likewise.
38002 (expand_atomic_load): Likewise.
38003 (expand_atomic_store): Likewise.
38004 (expand_atomic_fetch_op): Likewise.
38005 * testsuite/lib/target-supports.exp
38006 (check_effective_target_sync_int_128): Remove x86 because it provides
38007 no fast atomic load.
38008 (check_effective_target_sync_int_128_runtime): Likewise.
38010 2017-02-01 Richard Biener <rguenther@suse.de>
38012 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38013 (graphite_transform_loops): Provide opt-info for optimized nests.
38014 * tree-parloop.c (parallelize_loops): Provide opt-info for
38015 parallelized loops.
38017 2017-02-01 Richard Biener <rguenther@suse.de>
38019 PR middle-end/79315
38020 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38021 was not set before.
38023 2017-02-01 Richard Biener <rguenther@suse.de>
38025 PR tree-optimization/71824
38026 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38027 Verify the loops are valid in the merged SESE region.
38028 (scop_detection::can_represent_loop_1): Check analyzing the
38029 evolution of the number of iterations in the region succeeds.
38031 2017-01-31 Ian Lance Taylor <iant@golang.org>
38033 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38034 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38036 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38038 PR preprocessor/79210
38039 * input.c (get_substring_ranges_for_loc): Replace line_width
38040 assertion with error-handling.
38042 2017-01-31 Richard Biener <rguenther@suse.de>
38044 PR tree-optimization/77318
38045 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38046 (create_pw_aff_from_tree): Take loop parameter.
38047 (add_condition_to_pbb): Pass loop of the condition to
38048 create_pw_aff_from_tree.
38050 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38052 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38053 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38055 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38059 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38061 (convert_int_to_float128): Likewise.
38062 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38063 (convert_int_to_float128): Likewise.
38064 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38065 (UNSPEC_IEEE128_CONVERT): Likewise.
38066 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38067 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38068 Use local variables for IBM extended format.
38069 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38070 (fix_trunc<mode>si2_fprs): Likewise.
38071 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38072 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38073 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38074 to know that we can now have integers of all sizes in vector
38076 (fix<uns>_<mode>di2_hw): Likewise.
38077 (float<uns>_<mode>si2_hw): Likewise.
38078 (fix_<mode>si2_hw): Likewise.
38079 (fixuns_<mode>si2_hw): Likewise.
38080 (float<uns>_<mode>di2_hw): Likewise.
38081 (float_<mode>di2_hw): Likewise.
38082 (float_<mode>si2_hw): Likewise.
38083 (floatuns_<mode>di2_hw): Likewise.
38084 (floatuns_<mode>si2_hw): Likewise.
38085 (xscvqp<su>wz_<mode>): Delete, no longer used.
38086 (xscvqp<su>dz_<mode>): Likewise.
38087 (xscv<su>dqp_<mode>): Likewise.
38088 (ieee128_mfvsrd_64bit): Likewise.
38089 (ieee128_mfvsrd_32bit): Likewise.
38090 (ieee128_mfvsrwz): Likewise.
38091 (ieee128_mtvsrw): Likewise.
38092 (ieee128_mtvsrd_64bit): Likewise.
38093 (ieee128_mtvsrd_32bit): Likewise.
38095 2017-01-31 Martin Liska <mliska@suse.cz>
38098 * ipa-prop.c (ipa_free_all_node_params): Call release method
38099 instead of ~sumbol_summary to not to trigger double times
38102 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38104 PR tree-optimization/71691
38105 * bitmap.h (class auto_bitmap): New.
38106 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38107 is_maybe_undefined instead of ssa_undefined_value_p.
38109 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38111 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38112 __S390_ARCH_LEVEL__ to __ARCH__.
38114 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38116 PR tree-optimization/79267
38117 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38118 if should_remove_lhs_p is true.
38120 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38123 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38124 (add_alignment_attribute): New.
38125 (base_type_die): Add alignment attribute.
38126 (subrange_type_die): Likewise.
38127 (modified_type_die): Likewise.
38128 (gen_array_type_die): Likewise.
38129 (gen_descr_array_type_die: Likewise.
38130 (gen_enumeration_type_die): Likewise.
38131 (gen_subprogram_die): Likewise.
38132 (gen_variable_die): Likewise.
38133 (gen_field_die): Likewise.
38134 (gen_ptr_to_mbr_type_die): Likewise.
38135 (gen_struct_or_union_type_die): Likewise.
38136 (gen_subroutine_type_die): Likewise.
38137 (gen_typedef_die): Likewise.
38138 (base_type_cmp): Compare alignment attribute.
38140 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38143 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38144 (setb_unsigned) New pattern for setb with CCUNS.
38145 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38146 subfc./subfe sequence to avoid overflow problems. Generate a
38147 shorter sequence with cmpld/setb for power9.
38148 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38149 for generating subfc. instruction.
38150 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38151 now uses this instruction.
38153 2017-01-30 Ian Lance Taylor <iant@google.com>
38156 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38157 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38159 2017-01-30 Martin Sebor <msebor@redhat.com>
38161 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38162 Move constant to the right of a relational operator.
38163 (get_mpfr_format_length, format_character, format_string): Ditto.
38164 (should_warn_p, maybe_warn): Same.
38166 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38168 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38171 * asan.c (get_translation_unit_decl): Remove function.
38172 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38174 2017-01-30 Martin Liska <mliska@suse.cz>
38176 PR gcov-profile/79259
38177 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38178 -fprofile-generate.
38180 2017-01-30 Martin Liska <mliska@suse.cz>
38183 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38184 Initialize variables with NULL value.
38186 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38189 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38191 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38193 2017-01-30 Richard Biener <rguenther@suse.de>
38195 PR tree-optimization/79276
38196 * tree-vrp.c (process_assert_insertions): Properly adjust common
38197 when removing a duplicate.
38199 2017-01-30 Richard Biener <rguenther@suse.de>
38201 PR tree-optimization/79256
38202 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38203 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38205 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38207 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38210 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38211 ("*r<noxa>sbg_<mode>_sll_bitmask")
38212 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38213 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38214 Use contiguous_bitmask_nowrap_operand.
38216 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38219 * config/rs6000/altivec.h (vec_xl): Revise #define.
38220 (vec_xst): Likewise.
38222 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38224 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38226 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38228 PR rtl-optimization/79194
38229 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38230 traps before call to bypass_conditional_jumps.
38232 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38234 PR tree-optimization/71374
38235 * lra-constraints.c (check_conflict_input_operands): New.
38236 (match_reload): Use it.
38238 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38241 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38242 account to calculate conflict_set.
38244 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38246 PR rtl-optimization/78559
38247 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38248 other_insn in combine.
38250 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38252 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38253 uint16_type_node for BT_UINT16.
38255 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38257 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38258 "RTL Tests" to menu.
38259 (GIMPLE Tests): New node.
38260 (RTL Tests): New node.
38262 2017-01-27 Richard Biener <rguenther@suse.de>
38264 PR tree-optimization/79245
38265 * tree-loop-distribution.c (distribute_loop): Apply cost
38266 modeling also to detected patterns.
38268 2017-01-27 Richard Biener <rguenther@suse.de>
38270 PR tree-optimization/71433
38271 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38272 (compare_assert_loc): New function.
38273 (process_assert_insertions): Sort and optimize assert locations
38274 to remove duplicates and push down identical assertions on
38275 edges to their destination block.
38277 2017-01-27 Richard Biener <rguenther@suse.de>
38279 PR tree-optimization/79244
38280 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38281 out SSA names even if abnormal.
38283 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38285 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38286 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38287 instead of MPFR_RNDN.
38289 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38292 * arm.c (arm_option_override): Don't call build_target_option_node
38293 until after doing all option overrides.
38294 (arm_valid_target_attribute_tree): Likewise.
38296 2017-01-27 Martin Liska <mliska@suse.cz>
38298 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38299 for {cd}tors and C++ {cd}tors.
38301 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38303 * config/s390/s390.md ("*setmem_long_and")
38304 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38306 2017-01-26 Martin Sebor <msebor@redhat.com>
38308 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38311 2017-01-26 Martin Sebor <msebor@redhat.com>
38313 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38314 HAVE_DFmode before using XFmode or DFmode.
38315 (parse_directive): Avoid using the z length modifier to avoid
38316 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38318 PR middle-end/78703
38319 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38320 to accept adjustment as an array.
38321 (get_int_range): New function.
38322 (struct directive): Make width and prec arrays.
38323 (directive::set_width, directive::set_precision): Call get_int_range.
38324 (format_integer, format_floating): Handle width and precision ranges.
38325 (format_string, parse_directive): Same.
38327 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38330 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38331 comdat_type_p set, just clone them, but keep the children in the
38335 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38336 which have direct callers with -fvar-tracking-assignments enabled
38338 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38339 inside of type units.
38341 2017-01-26 Martin Sebor <msebor@redhat.com>
38343 PR middle-end/78703
38344 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38346 (struct format_result): Replace number_chars, number_chars_min,
38347 and number_chars_max with a single member of struct result_range.
38349 (format_result::operator+=): Adjust.
38350 (struct fmtresult): Remove bounded. Handle likely and unlikely
38352 (fmtresult::adjust_for_width_or_precision): New function.
38353 (fmtresult:type_max_digits): New function.
38354 (bytes_remaining): Handle likely and unlikely counters.
38355 (min_bytes_remaining): Remove.
38356 (format_percent): Simplify.
38357 (format_integer, format_floating): Set likely and unlikely counters.
38358 (get_string_length, format_character, format_string): Same.
38359 (format_plain, should_warn_p): New function.
38360 (maybe_warn): Call should_warn_p. Update diagnostic messages
38361 and handle those for all directives, including plain strings.
38362 (format_directive): Handle likely and unlikely counters.
38363 Remove unnecessary quoting from diagnostics. Add an informational
38365 (add_bytes): Remove.
38366 (pass_sprintf_length::compute_format_length): Simplify.
38367 (try_substitute_return_value): Handle likely and unlikely counters.
38369 2017-01-26 Carl Love <cel@us.ibm.com>
38371 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38372 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38374 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38377 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38378 endianess for subregs into account.
38379 * lra-constraints.c (lra_constraints): Do risky transformations
38380 always on the first iteration.
38381 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38382 last_call_used_reg_set.
38383 (process_bb_lives): Define and use last_call_used_reg_set.
38384 * lra.c (lra): Always continue after lra_constraints on the first
38387 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38389 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38391 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38393 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38395 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38396 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38397 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38398 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38399 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38400 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38401 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38402 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38403 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38405 2017-01-26 Marek Polacek <polacek@redhat.com>
38408 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38409 for the third operand.
38411 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38413 PR middle-end/79236
38414 * omp-low.c (struct omp_context): Add simt_stmt field.
38415 (scan_omp_for): Return omp_context *.
38416 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38417 context to the _simt_ SIMD stmt.
38418 (lower_omp_for): For combined SIMD with sibling _simt_
38419 SIMD, make sure to use the same decls in _looptemp_
38420 clauses as in the sibling.
38422 2017-01-26 David Sherwood <david.sherwood@arm.com>
38424 PR middle-end/79212
38425 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38428 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38431 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38432 emit fld b; fld a; if possible.
38434 * brig-builtins.def: Update copyright years.
38435 * config/arm/arm_acle_builtins.def: Update copyright years.
38437 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38440 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38441 constraint instead of o for the stxsd instruction.
38443 2017-01-25 Carl Love <cel@us.ibm.com>
38445 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38446 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38448 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38450 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38452 2017-01-25 Richard Biener <rguenther@suse.de>
38454 PR tree-optimization/69264
38455 * target.def (vector_alignment_reachable): Improve documentation.
38456 * doc/tm.texi: Regenerate.
38457 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38459 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38460 earlier changes with respect to TYPE_USER_ALIGN.
38461 (vector_alignment_reachable_p): Likewise. Improve dumping.
38463 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38466 * config/arm/arm.md (xordi3): Force constant operand into a register
38469 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38471 * doc/invoke.texi (-fstore-merging): Correct default optimization
38472 levels at which it is enabled.
38473 (-O): Move -fstore-merging from list to...
38476 2017-01-25 Richard Biener <rguenther@suse.de>
38479 * omp-expand.c: Include debug.h.
38480 (expand_omp_taskreg): Make sure to generate early debug before
38481 outlining anything from a function.
38482 (expand_omp_target): Likewise.
38483 (grid_expand_target_grid_body): Likewise.
38485 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38488 * asan.c (get_translation_unit_decl): New function.
38489 (asan_add_global): Extract modules file name from globals
38490 TRANSLATION_UNIT_DECL name.
38492 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38495 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38496 for long calls with APCS frame and VFP.
38498 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38500 * cfg.c (original_copy_tables_initialized_p): New function.
38501 * cfg.h (original_copy_tables_initialized_p): New decl.
38502 * cfgrtl.c (relink_block_chain): Guard the call to
38503 free_original_copy_tables with a call to
38504 original_copy_tables_initialized_p.
38505 * cgraph.h (symtab_node::native_rtl_p): New decl.
38506 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38507 (symtab_node::needed_p): Don't assert for early assembly output
38508 for __RTL functions.
38509 (cgraph_node::finalize_function): Set "force_output" for __RTL
38511 (cgraph_node::analyze): Bail out early for __RTL functions.
38512 (analyze_functions): Update assertion to support __RTL functions.
38513 (cgraph_node::expand): Bail out early for __RTL functions.
38514 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38516 * function.h (struct function): Update comment for field
38518 * gimple-expr.c: Include "tree-pass.h".
38519 (gimple_has_body_p): Return false for __RTL functions.
38520 * Makefile.in (OBJS): Add run-rtl-passes.o.
38521 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38523 (gcc::pass_manager::get_clean_slate): New accessor.
38524 * passes.c: Include "insn-addr.h".
38525 (should_skip_pass_p): Add logging. Update logic for running
38526 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38527 property-provider override so it is only done for gimple passes.
38529 (skip_pass): New function.
38530 (execute_one_pass): Call skip_pass when skipping passes.
38531 * read-md.c (md_reader::read_char): Support filtering
38532 the input to a subset of line numbers.
38533 (md_reader::md_reader): Initialize fields
38534 m_first_line and m_last_line.
38535 (md_reader::read_file_fragment): New function.
38536 * read-md.h (md_reader::read_file_fragment): New decl.
38537 (md_reader::m_first_line): New field.
38538 (md_reader::m_last_line): New field.
38539 * read-rtl-function.c (function_reader::create_function): Only
38540 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38541 curr_properties. Set DECL_INITIAL to a dummy block.
38542 (read_rtl_function_body_from_file_range): New function.
38543 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38545 * run-rtl-passes.c: New file.
38546 * run-rtl-passes.h: New file.
38548 2017-01-24 Jeff Law <law@redhat.com>
38550 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38553 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38555 PR tree-optimization/79159
38556 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38557 (record_nonwrapping_iv): Improve boundary using above function if no
38558 value range information.
38560 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38561 Martin Jambor <mjambor@suse.cz>
38563 * brig-builtins.def: New file.
38564 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38565 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38566 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38567 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38568 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38569 * builtin-types.def (BT_INT8): New.
38570 (BT_INT16): Likewise.
38571 (BT_UINT8): Likewise.
38572 (BT_UINT16): Likewise.
38573 (BT_FN_ULONG): Likewise.
38574 (BT_FN_UINT_INT): Likewise.
38575 (BT_FN_UINT_ULONG): Likewise.
38576 (BT_FN_UINT_LONG): Likewise.
38577 (BT_FN_UINT_PTR): Likewise.
38578 (BT_FN_ULONG_PTR): Likewise.
38579 (BT_FN_INT8_FLOAT): Likewise.
38580 (BT_FN_INT16_FLOAT): Likewise.
38581 (BT_FN_UINT32_FLOAT): Likewise.
38582 (BT_FN_UINT16_FLOAT): Likewise.
38583 (BT_FN_UINT8_FLOAT): Likewise.
38584 (BT_FN_UINT64_FLOAT): Likewise.
38585 (BT_FN_UINT16_UINT32): Likewise.
38586 (BT_FN_UINT32_UINT16): Likewise.
38587 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38588 (BT_FN_INT_PTR_INT): Likewise.
38589 (BT_FN_UINT_PTR_UINT): Likewise.
38590 (BT_FN_LONG_PTR_LONG): Likewise.
38591 (BT_FN_ULONG_PTR_ULONG): Likewise.
38592 (BT_FN_VOID_UINT64_UINT64): Likewise.
38593 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38594 (BT_FN_INT8_INT8_INT8): Likewise.
38595 (BT_FN_INT16_INT16_INT16): Likewise.
38596 (BT_FN_INT_INT_INT): Likewise.
38597 (BT_FN_UINT_FLOAT_UINT): Likewise.
38598 (BT_FN_FLOAT_UINT_UINT): Likewise.
38599 (BT_FN_ULONG_UINT_UINT): Likewise.
38600 (BT_FN_ULONG_UINT_PTR): Likewise.
38601 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38602 (BT_FN_UINT_UINT_UINT): Likewise.
38603 (BT_FN_VOID_UINT_PTR): Likewise.
38604 (BT_FN_UINT_UINT_PTR: Likewise.
38605 (BT_FN_UINT32_UINT64_PTR): Likewise.
38606 (BT_FN_INT_INT_UINT_UINT): Likewise.
38607 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38608 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38609 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38610 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38611 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38612 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38613 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38614 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38615 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38616 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38617 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38618 * doc/frontends.texi: List BRIG FE.
38619 * doc/install.texi (Testing): Add BRIG tesring requirements.
38620 * doc/invoke.texi (Overall Options): Mention BRIG.
38621 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38623 2017-01-24 Richard Biener <rguenther@suse.de>
38625 PR translation/79208
38626 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38628 2017-01-24 Martin Jambor <mjambor@suse.cz>
38631 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38632 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38633 and known_contexts.
38635 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38637 PR middle-end/79123
38638 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38639 casts from signed to unsigned really don't have a range.
38641 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38643 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38644 GMP_RNDx for compatiblity.
38646 2017-01-24 Martin Liska <mliska@suse.cz>
38649 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38650 that would prevent us to call alloca with -1 as argument.
38652 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38654 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38655 Avoid -Wformat-security warning.
38657 2017-01-23 Andrew Pinski <apinski@cavium.com>
38659 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38662 2017-01-23 Martin Sebor <msebor@redhat.com>
38664 PR middle-end/78703
38665 * gimple-ssa-sprintf.c (warn_level): New global.
38666 (format_integer): Use it here and throughout the rest of the file.
38667 Use the same switch to compute sign as base.
38668 (maybe_warn): New function.
38669 (format_directive): Factor out warnings into maybe_warn.
38670 Add debugging output. Use warn_level.
38671 (add_bytes): Use warn_level.
38672 (pass_sprintf_length::compute_format_length): Add debugging output.
38673 (try_substitute_return_value): Same.
38674 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38676 PR middle-end/78703
38677 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38678 (struct fmtresult, format_integer, format_floating): Adjust.
38679 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38680 (get_string_length, format_string,format_directive): Same.
38681 (pass_sprintf_length::compute_format_length): Same.
38682 (try_substitute_return_value): Simplify slightly.
38684 PR middle-end/78703
38685 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38686 (fmtresult::operator+=): Outlined.
38687 (struct fmtresult): Add ctors.
38688 (struct conversion_spec): Rename...
38689 (struct directive): ...to this. Add and remove data members.
38690 (directive::set_width, directive::set_precision): New functions.
38691 (format_percent): Use fmtresult ctor.
38692 (get_width_and_precision): Remove.
38693 (format_integer): Make naming changes. Avoid computing width and
38695 (format_floating): Same. Adjust indentation.
38696 (format_character, format_none): New functions.
38697 (format_string): Moved character handling to format_character.
38698 (format_directive): Remove arguments, change return type.
38699 (parse_directive): New function.
38700 (pass_sprintf_length::compute_format_length): Move directive
38701 parsing to parse_directive.
38703 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38705 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38706 (assign_assembler_name_if_needed): ... this.
38707 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38708 (assign_assembler_name_if_needed): ... this.
38709 (free_lang_data_in_cgraph): Adjust callers.
38710 * cgraphunit.c (cgraph_node::analyze): Likewise.
38711 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38714 2017-01-23 Richard Biener <rguenther@suse.de>
38716 PR tree-optimization/79088
38717 PR tree-optimization/79188
38718 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38719 resetting loop bounds after last path deletion. Reset loop
38720 bounds of the target loop, make code match the comments.
38721 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38722 Make sure loops need no fixups.
38724 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38726 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38727 exponent support with double type for first argument.
38728 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38729 type returned by __builtin_vec_extract_sig,
38730 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38731 functions from "vector int" to "vector unsigned int" or from
38732 "vector long long int" to "vector unsigned long long int".
38733 Changed type returned by __builtin_vec_extract_exp,
38734 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38735 functions from "vector int" to "vector unsigned int" or from
38736 "vector long long int" to "vector unsigned long long int".
38737 Changed return type of __builtin_vec_test_data_class,
38738 __builtin_vec_test_data_class_sp, and
38739 __builtin_vec_test_data_class_dp from "vector int" to
38740 "vector bool int" or from "vector long long int" to "vector bool
38741 long long int" and changed second argument type from "unsigned
38742 int" to "int". Added new overloaded function forms "vector float
38743 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38744 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38745 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38746 double, vector unsigned long long int)" and "vector double
38747 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38748 long int)". Changed return type of
38749 __builtin_scalar_test_data_class and
38750 __builtin_scalar_test_data_class_sp and
38751 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38752 int" and changed second argument from "unsigned int" to "int".
38753 Changed type returned by __builtin_scalar_test_neg,
38754 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38755 from "int" to "bool int". Added new overloaded function form
38756 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38757 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38758 exponent double-precision with floating point first argument.
38759 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38760 documentation of scalar_test_data_class, scalar_test_neg,
38761 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38762 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38763 vec_test_data_class built-in functions to reflect refinements in
38764 their type signatures.
38766 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38768 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38770 (aarch64_elf_asm_destructor): Likewise.
38772 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38774 PR rtl-optimization/78634
38775 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38776 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38777 * ifcvt.c (noce_try_cmove): Add missing cost check.
38779 PR rtl-optimization/71724
38780 * combine.c (if_then_else_cond): Look for situations where it is
38781 beneficial to undo the work of one of the recursive calls.
38783 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38785 PR tree-optimization/70754
38786 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38787 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38788 combined stmt before it if not NULL.
38789 (combine_chains): Process refs reversely and compute dominance point
38792 2017-01-23 Martin Liska <mliska@suse.cz>
38794 PR tree-optimization/79196
38795 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38796 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38798 (strlen_optimize_stmt): Call the renamed function.
38800 2017-01-23 Michael Matz <matz@suse.de>
38802 PR tree-optimization/78384
38803 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38805 2017-01-23 Richard Biener <rguenther@suse.de>
38807 PR tree-optimization/79186
38808 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38809 both incoming edges before moving an assert.
38811 2017-01-23 Martin Jambor <mjambor@suse.cz>
38813 * ipa-prop.c (load_from_param_1): Removed.
38814 (load_from_unmodified_param): Bits from load_from_param_1 put back
38816 (load_from_param): Removed.
38817 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38818 with stmt. Reverted back to use of load_from_unmodified_param.
38820 2017-01-23 Martin Jambor <mjambor@suse.cz>
38823 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38824 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38825 field a pointer to garbage collected vector, mark lattices and
38826 ipcp_orig_node with GTY((skip)).
38827 (ipa_get_param_count): Adjust to descriptors being a pointer.
38828 (ipa_get_param): Likewise.
38829 (ipa_get_type): Likewise.
38830 (ipa_get_param_move_cost): Likewise.
38831 (ipa_set_param_used): Likewise.
38832 (ipa_get_controlled_uses): Likewise.
38833 (ipa_set_controlled_uses): Likewise.
38834 (ipa_is_param_used): Likewise.
38835 (ipa_node_params_t): Move into garbage collector. New methods insert
38837 (ipa_node_params_sum): Annotate wth GTY(()).
38838 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38840 (ipa_load_from_parm_agg): Adjust declaration.
38841 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38842 * ipa-profile.c (ipa_profile): Likewise.
38843 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38844 (ipa_populate_param_decls): Make descriptors parameter garbage
38846 (ipa_dump_param): Adjust to descriptors being a pointer.
38847 (ipa_alloc_node_params): Likewise.
38848 (ipa_initialize_node_params): Likewise.
38849 (load_from_param_1): Make descriptors parameter garbage collected.
38850 (load_from_unmodified_param): Likewise.
38851 (load_from_param): Likewise.
38852 (ipa_load_from_parm_agg): Likewise.
38853 (ipa_node_params::~ipa_node_params): Removed.
38854 (ipa_free_all_node_params): Remove call to delete operator.
38855 (ipa_node_params_t::insert): New.
38856 (ipa_node_params_t::remove): Likewise.
38857 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38858 copy known_csts and known_contexts vectors.
38859 (ipa_read_node_info): Adjust to descriptors being a pointer.
38860 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38862 (ipcp_transform_function): Make descriptors variable garbage
38865 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38867 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38868 * config/i386/avx512dqintrin.h: Ditto.
38869 * config/i386/avx512fintrin.h: Ditto.
38870 * config/i386/i386.c: Handle new builtins.
38871 * config/i386/i386-builtin.def: Add new builtins.
38872 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38873 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38875 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38876 Martin Liska <mliska@suse.cz>
38878 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38879 * asan.c (asan_expand_poison_ifn): Support stores and use
38880 appropriate ASAN report function.
38881 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38882 * internal-fn.def (ASAN_POISON_USE): Declare.
38883 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38884 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38885 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38886 ASAN_POISON calls w/o LHS.
38887 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38888 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38889 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38890 * gimplify.c (asan_poison_variables): Add attribute
38891 use_after_scope_memory to variables that really needs to live
38893 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38894 having the attribute.
38896 2017-01-23 Martin Liska <mliska@suse.cz>
38898 * asan.c (create_asan_shadow_var): New function.
38899 (asan_expand_poison_ifn): Likewise.
38900 * asan.h (asan_expand_poison_ifn): New declaration.
38901 * internal-fn.c (expand_ASAN_POISON): Likewise.
38902 * internal-fn.def (ASAN_POISON): New builtin.
38903 * sanopt.c (pass_sanopt::execute): Expand
38904 asan_expand_poison_ifn.
38905 * tree-inline.c (copy_decl_for_dup_finish): Make function
38907 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38908 * tree-ssa.c (is_asan_mark_p): New function.
38909 (execute_update_addresses_taken): Rewrite local variables
38910 (identified just by use-after-scope as addressable) into SSA.
38912 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38914 * doc/install.texi (Specific): opensource.apple.com uses https
38915 now. Remove trailing slash.
38917 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38919 * README.Portability: Remove note on an Irix compatibility issue.
38921 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38923 * gcov.c (INCLUDE_ALGORITHM): Define.
38924 (INCLUDE_VECTOR): Define.
38925 No longer include <vector> and <algorithm> directly.
38927 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38929 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38931 * doc/invoke.texi (Code Gen Options): Ditto.
38933 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38936 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38938 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38940 rtl-optimization/79125
38941 * cprop.c (local_cprop_pass): Handle cases where we make an
38942 unconditional trap.
38944 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38948 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38949 read from, for big endian.
38951 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38953 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38954 register pauth builtins for LP64 only.
38956 2017-01-20 Marek Polacek <polacek@redhat.com>
38959 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38962 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38964 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38966 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38967 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38968 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38970 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38973 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38974 in big-endian mode when they are not a single duplicated value.
38976 2017-01-20 Richard Biener <rguenther@suse.de>
38978 * BASE-VER: Bump to 7.0.1.
38980 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38982 * omp-low.c (omplow_simd_context): New struct. Use it...
38983 (lower_rec_simd_input_clauses): ...here and...
38984 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38985 references to idx, lane, max_vf, is_simt.
38987 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38989 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38992 2017-01-20 Martin Jambor <mjambor@suse.cz>
38994 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38995 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38997 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38998 (GTFILES): Rename hsa.c to hsa-common.c.
38999 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39000 * hsa-dump.c: Likewise.
39001 * hsa-gen.c: Likewise.
39002 * hsa-regalloc.c: Likewise.
39003 * ipa-hsa.c: Likewise.
39004 * omp-expand.c: Likewise.
39005 * omp-low.c: Likewise.
39006 * toplev.c: Likewise.
39008 2017-01-20 Marek Polacek <polacek@redhat.com>
39011 * doc/invoke.texi: Document -Wduplicated-branches.
39012 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39013 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39014 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39015 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39016 return 0 only when not OEP_LEXICOGRAPHIC.
39017 (fold_build_cleanup_point_expr): Use the expression
39018 location when building CLEANUP_POINT_EXPR.
39019 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39020 * tree.c (add_expr): Handle error_mark_node.
39022 2017-01-20 Martin Liska <mliska@suse.cz>
39025 * tree-profile.c (init_ic_make_global_vars): Do not call
39027 (gimple_init_gcov_profiler): Likewise.
39029 2017-01-20 Martin Liska <mliska@suse.cz>
39032 * cgraph.h (maybe_create_reference): Remove argument and
39034 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39036 * ipa-cp.c (create_specialized_node): Likewise.
39037 * symtab.c (symtab_node::maybe_create_reference): Handle
39038 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39040 2017-01-20 Martin Liska <mliska@suse.cz>
39042 * read-rtl-function.c (function_reader::create_function): Use
39043 build_decl instread of build_decl_stat.
39045 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39047 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39048 * config/i386/avx512dqintrin.h: Ditto.
39049 * config/i386/avx512fintrin.h: Ditto.
39050 * config/i386/i386-builtin-types.def: Add new types.
39051 * config/i386/i386.c: Handle new types.
39052 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39053 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39054 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39055 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39056 (__builtin_ia32_kshiftridi): New.
39057 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39059 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39063 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39064 define to rs6000_init_stack_protect_guard.
39065 (rs6000_init_stack_protect_guard): New function.
39067 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39068 Yunqiang Su <yunqiang.su@imgtec.com>
39070 * config.gcc (supported_defaults): Add madd4.
39071 (with_madd4): Add validation.
39072 (all_defaults): Add madd4.
39073 * config/mips/mips.opt (mmadd4): New option.
39074 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39076 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39078 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39079 (ISA_HAS_FUSED_MADD4): Likewise.
39080 * doc/invoke.texi (-mmadd4): Document the new option.
39081 * doc/install.texi (--with-madd4): Document the new option.
39083 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39085 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39086 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39087 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39088 (aarch64_init_pauth_hint_builtins): New.
39089 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39090 (aarch64_expand_builtin): Expand new builtins.
39092 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39094 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39095 * combine-stack-adj.c (no_unhandled_cfa): Handle
39096 REG_CFA_TOGGLE_RA_MANGLE.
39097 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39098 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39099 info for return address signing.
39100 (aarch64_expand_epilogue): Likewise.
39102 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39104 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39105 * config/aarch64/aarch64-protos.h
39106 (aarch64_return_address_signing_enabled): New declaration.
39107 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39109 (aarch64_expand_prologue): Sign return address before it's pushed onto
39111 (aarch64_expand_epilogue): Authenticate return address fetched from
39113 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39114 (aarch64_attributes): New function attributes for "sign-return-address".
39115 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39116 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39117 ("*do_return"): Generate combined instructions according to key index.
39118 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39119 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39121 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39122 * config/aarch64/aarch64.opt (msign-return-address=): New.
39123 * doc/extend.texi (AArch64 Function Attributes): Documents
39124 "sign-return-address=".
39125 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39127 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39129 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39130 overall option summary.
39132 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39134 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39135 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39136 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39137 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39139 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39141 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39142 -mpower9-minmax by default for -mcpu=power9.
39143 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39144 128-bit floating point.
39146 2017-01-20 Alan Modra <amodra@gmail.com>
39148 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39149 optimizing for size.
39151 2017-01-20 Alan Modra <amodra@gmail.com>
39154 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39155 for strcmp and strncmp from corresponding builtin decl.
39157 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39159 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39160 instead of i386/rtems-64.h.
39161 * config/i386/rtems-64.h: Remove.
39163 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39167 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39169 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39171 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39173 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39174 Change int to HOST_WIDE_INT.
39175 * config/aarch64/aarch64-protos.h
39176 (aarch64_simd_gen_const_vector_dup): Likewise.
39177 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39179 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39181 * langhooks-def.h (lhd_type_for_size): New decl.
39182 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39183 * langhooks.c (lhd_type_for_size): New function, taken from
39186 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39188 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39189 define_bypass for CR latency.
39190 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39191 (power9-alu2): Add define_bypass for CR latency.
39193 (power9-mul): Update insn latency.
39194 (power9-mul-compare): Update insn latency, bypass latency and remove
39197 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39199 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39201 * config/aarch64/aarch64.md
39202 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39203 aarch64_nopcrelative_literal_loads.
39204 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39206 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39208 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39209 TARGET_LOONGSON_3A.
39210 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39212 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39215 * config.gcc (supported_defaults): Add lxc1-sxc1.
39216 (with_lxc1_sxc1): Add validation.
39217 (all_defaults): Add lxc1-sxc1.
39218 * config/mips/mips.opt (mlxc1-sxc1): New option.
39219 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39221 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39222 __mips_no_lxc1_sxc1.
39223 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39224 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39225 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39227 2017-01-19 Richard Biener <rguenther@suse.de>
39229 PR tree-optimization/72488
39230 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39231 sure to restore SSA info.
39232 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39234 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39236 PR rtl-optimization/79121
39237 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39238 of the inner type when shifting an extended value.
39240 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39243 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39244 interposable aliases.
39246 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39249 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39250 Use the evmergelohi instruction.
39251 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39252 (mov_sitf_e500_subreg8_2_be): Likewise.
39253 (mov_sitf_e500_subreg12_2_le): Likewise.
39254 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39255 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39256 (mov_sitf_e500_subreg8_2_le): Likewise.
39257 (mov_sitf_e500_subreg12_2_be): Likewise.
39259 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39261 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39262 attribute from vecsimple to vecperm.
39263 (altivec_vbpermq2): Likewise.
39265 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39268 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39270 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39271 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39272 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39273 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39274 case where N arg is SIZE_MAX.
39275 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39276 (cmpstrsi): Add pattern.
39278 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39280 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39281 __builtin_vec_revb builtins.
39282 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39283 built-in functions to support generation of the ISA 3.0 XXBR<x>
39284 vector byte reverse instructions.
39285 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39286 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39287 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39288 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39289 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39290 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39291 (P9V_BUILTIN_VEC_REVB): Likewise.
39292 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39293 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39294 (p9_xxbrq_v16qi): Likewise.
39295 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39296 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39297 (p9_xxbrh_v8hi): Likewise.
39298 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39299 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39300 vec_revb built-in functions.
39302 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39304 PR rtl-optimization/78952
39305 * config/i386/i386.md (any_extract): New code iterator.
39306 (*insvqi_2): Use any_extract for source operand.
39307 (*insvqi_3): Use any_shiftrt for source operand.
39309 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39311 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39313 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39315 2017-01-18 Matthias Klose <doko@ubuntu.com>
39317 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39319 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39321 * config/rs6000/altivec.h (vec_bperm): Change #define.
39322 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39323 (altivec_vbpermq2): New define_insn.
39324 (altivec_vbpermd): Likewise.
39325 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39326 function interface.
39327 (VBPERMD): Likewise.
39328 (VBPERM): New polymorphic function interface.
39329 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39330 Add entries for P9V_BUILTIN_VEC_VBPERM.
39331 * doc/extend.texi: Add interfaces for vec_bperm.
39333 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39335 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39336 first letter of error messages.
39337 (s390_resolve_overloaded_builtin): Likewise.
39338 * config/s390/s390.c (s390_expand_builtin): Likewise.
39339 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39340 (s390_valid_target_attribute_inner_p): Likewise.
39341 * config/s390/s390.md ("tabort"): Likewise.
39343 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39345 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39346 (ISA_AVOID_DIV_HILO): New macro.
39347 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39348 (ISA_HAS_DDIV): Likewise.
39350 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39352 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39354 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39356 * doc/invoke.texi (fabi-version): Spelling fix.
39358 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39361 * doc/invoke.texi (fabi-version): Mention mangling fix for
39364 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39367 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39369 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39372 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39373 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39375 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39377 (stack_protect_test): Ditto.
39378 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39379 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39381 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39382 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39383 -mstack-protector-guard-offset=.
39384 (RS/6000 and PowerPC Options): Ditto.
39386 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39388 * config/i386/i386.h (MASK_CLASS_P): New define.
39389 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39390 there are no registers from different register sets also when
39391 mask registers are used. Update function comment.
39392 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39393 to (*k/*r) and (*k/*km) alternatives.
39395 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39397 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39398 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39399 (EH_RETURN_HANDLER_RTX): New define.
39400 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39401 Force frame pointer in EH return functions.
39402 (aarch64_expand_epilogue): Add barrier for eh_return.
39403 (aarch64_final_eh_return_addr): Remove.
39404 (aarch64_eh_return_handler_rtx): New function.
39405 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39407 (aarch64_eh_return_handler_rtx): New prototype.
39409 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39411 * config/rs6000/altivec.h (vec_rlmi): New #define.
39412 (vec_vrlnm): Likewise.
39413 (vec_rlnm): Likewise.
39414 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39415 (UNSPEC_VRLNM): Likewise.
39416 (VIlong): New mode iterator.
39417 (altivec_vrl<VI_char>mi): New define_insn.
39418 (altivec_vrl<VI_char>nm): Likewise.
39419 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39421 (VRLDNM): Likewise.
39422 (RLNM): New polymorphic function entry.
39423 (VRLWMI): New monomorphic function entry.
39424 (VRLDMI): Likewise.
39425 (RLMI): New polymorphic function entry.
39426 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39427 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39428 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39431 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39434 * dwarf2out.c (field_byte_offset): Restore the
39435 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39436 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39437 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39440 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39443 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39445 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39448 * dwarf2out.c (add_data_member_location_attribute): For constant
39449 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39450 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39451 DW_AT_byte_size attributes.
39453 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39455 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39456 after forcing to constant memory when the code model is medium.
39458 2017-01-17 Julia Koval <julia.koval@intel.com>
39461 * config/i386/avx512fintrin.h
39462 (_mm512_i32gather_ps): Change __addr type to void const*.
39463 (_mm512_mask_i32gather_ps): Ditto.
39464 (_mm512_i32gather_pd): Ditto.
39465 (_mm512_mask_i32gather_pd): Ditto.
39466 (_mm512_i64gather_ps): Ditto.
39467 (_mm512_mask_i64gather_ps): Ditto.
39468 (_mm512_i64gather_pd): Ditto.
39469 (_mm512_mask_i64gather_pd): Ditto.
39470 (_mm512_i32gather_epi32): Ditto.
39471 (_mm512_mask_i32gather_epi32): Ditto.
39472 (_mm512_i32gather_epi64): Ditto.
39473 (_mm512_mask_i32gather_epi64): Ditto.
39474 (_mm512_i64gather_epi32): Ditto.
39475 (_mm512_mask_i64gather_epi32): Ditto.
39476 (_mm512_i64gather_epi64): Ditto.
39477 (_mm512_mask_i64gather_epi64): Ditto.
39478 (_mm512_i32scatter_ps): Change __addr type to void*.
39479 (_mm512_mask_i32scatter_ps): Ditto.
39480 (_mm512_i32scatter_pd): Ditto.
39481 (_mm512_mask_i32scatter_pd): Ditto.
39482 (_mm512_i64scatter_ps): Ditto.
39483 (_mm512_mask_i64scatter_ps): Ditto.
39484 (_mm512_i64scatter_pd): Ditto.
39485 (_mm512_mask_i64scatter_pd): Ditto.
39486 (_mm512_i32scatter_epi32): Ditto.
39487 (_mm512_mask_i32scatter_epi32): Ditto.
39488 (_mm512_i32scatter_epi64): Ditto.
39489 (_mm512_mask_i32scatter_epi64): Ditto.
39490 (_mm512_i64scatter_epi32): Ditto.
39491 (_mm512_mask_i64scatter_epi32): Ditto.
39492 (_mm512_i64scatter_epi64): Ditto.
39493 (_mm512_mask_i64scatter_epi64): Ditto.
39494 * config/i386/avx512pfintrin.h
39495 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39496 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39497 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39498 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39499 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39500 (_mm512_prefetch_i32scatter_ps): Ditto.
39501 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39502 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39503 (_mm512_prefetch_i64scatter_pd): Ditto.
39504 (_mm512_prefetch_i64scatter_ps): Ditto.
39505 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39506 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39507 * config/i386/avx512vlintrin.h
39508 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39509 (_mm_mmask_i32gather_ps): Ditto.
39510 (_mm256_mmask_i32gather_pd): Ditto.
39511 (_mm_mmask_i32gather_pd): Ditto.
39512 (_mm256_mmask_i64gather_ps): Ditto.
39513 (_mm_mmask_i64gather_ps): Ditto.
39514 (_mm256_mmask_i64gather_pd): Ditto.
39515 (_mm_mmask_i64gather_pd): Ditto.
39516 (_mm256_mmask_i32gather_epi32): Ditto.
39517 (_mm_mmask_i32gather_epi32): Ditto.
39518 (_mm256_mmask_i32gather_epi64): Ditto.
39519 (_mm_mmask_i32gather_epi64): Ditto.
39520 (_mm256_mmask_i64gather_epi32): Ditto.
39521 (_mm_mmask_i64gather_epi32): Ditto.
39522 (_mm256_mmask_i64gather_epi64): Ditto.
39523 (_mm_mmask_i64gather_epi64): Ditto.
39524 (_mm256_i32scatter_ps): Change __addr type to void*.
39525 (_mm256_mask_i32scatter_ps): Ditto.
39526 (_mm_i32scatter_ps): Ditto.
39527 (_mm_mask_i32scatter_ps): Ditto.
39528 (_mm256_i32scatter_pd): Ditto.
39529 (_mm256_mask_i32scatter_pd): Ditto.
39530 (_mm_i32scatter_pd): Ditto.
39531 (_mm_mask_i32scatter_pd): Ditto.
39532 (_mm256_i64scatter_ps): Ditto.
39533 (_mm256_mask_i64scatter_ps): Ditto.
39534 (_mm_i64scatter_ps): Ditto.
39535 (_mm_mask_i64scatter_ps): Ditto.
39536 (_mm256_i64scatter_pd): Ditto.
39537 (_mm256_mask_i64scatter_pd): Ditto.
39538 (_mm_i64scatter_pd): Ditto.
39539 (_mm_mask_i64scatter_pd): Ditto.
39540 (_mm256_i32scatter_epi32): Ditto.
39541 (_mm256_mask_i32scatter_epi32): Ditto.
39542 (_mm_i32scatter_epi32): Ditto.
39543 (_mm_mask_i32scatter_epi32): Ditto.
39544 (_mm256_i32scatter_epi64): Ditto.
39545 (_mm256_mask_i32scatter_epi64): Ditto.
39546 (_mm_i32scatter_epi64): Ditto.
39547 (_mm_mask_i32scatter_epi64): Ditto.
39548 (_mm256_i64scatter_epi32): Ditto.
39549 (_mm256_mask_i64scatter_epi32): Ditto.
39550 (_mm_i64scatter_epi32): Ditto.
39551 (_mm_mask_i64scatter_epi32): Ditto.
39552 (_mm256_i64scatter_epi64): Ditto.
39553 (_mm256_mask_i64scatter_epi64): Ditto.
39554 (_mm_i64scatter_epi64): Ditto.
39555 (_mm_mask_i64scatter_epi64): Ditto.
39556 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39557 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39558 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39559 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39560 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39561 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39562 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39563 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39564 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39565 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39566 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39567 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39568 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39569 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39570 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39571 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39572 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39573 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39574 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39575 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39576 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39577 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39578 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39579 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39580 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39581 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39582 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39583 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39584 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39585 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39586 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39587 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39588 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39589 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39590 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39591 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39592 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39593 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39594 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39595 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39596 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39597 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39598 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39599 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39600 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39601 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39602 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39603 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39604 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39605 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39606 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39607 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39608 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39609 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39610 definitions accordingly.
39612 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39613 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39616 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39619 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39622 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39623 hard regs for inner regno.
39625 2017-01-17 Martin Liska <mliska@suse.cz>
39628 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39629 assumption and add comment.
39631 2017-01-17 Nathan Sidwell <nathan@acm.org>
39633 * ipa-visibility.c (localize_node): New function, broken out of ...
39634 (function_and_variable_visibility): ... here. Call it.
39636 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39638 PR middle-end/77445
39639 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39640 correctly set frequency of oudgoing edge.
39641 (duplicate_thread_path): Fix profile updating.
39643 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39646 * configure.ac: Add GCC_BASE_VER.
39647 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39648 version from BASE-VER file.
39649 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39650 (gcc.o): Depend on $(BASEVER).
39651 * common.opt (dumpfullversion): New option.
39652 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39653 * doc/invoke.texi: Document -dumpfullversion.
39654 * doc/install.texi: Document --with-gcc-major-version-only.
39655 * configure: Regenerated.
39657 2017-01-17 Richard Biener <rguenther@suse.de>
39659 PR tree-optimization/71433
39660 * tree-vrp.c (register_new_assert_for): Merge same asserts
39661 on all incoming edges.
39662 (process_assert_insertions_for): Handle insertions at the
39665 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39667 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39668 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39670 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39673 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39676 2017-01-17 Alan Modra <amodra@gmail.com>
39679 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39680 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39681 symbolic stack limit when pic.
39683 2017-01-16 Martin Sebor <msebor@redhat.com>
39685 PR tree-optimization/78608
39686 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39688 2017-01-16 Jeff Law <law@redhat.com>
39691 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39692 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39693 for several include directories that may be relative to sysroot.
39694 * config/i386/x-mingw32 (gplus_includedir): Define.
39695 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39696 (native_system_includedir): Likewise.
39697 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39698 override if TARGET_SYSTEM_ROOT is defined.
39699 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39701 PR tree-optimization/79090
39702 PR tree-optimization/33562
39703 PR tree-optimization/61912
39704 PR tree-optimization/77485
39705 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39706 and computed trims into the dump file.
39708 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39710 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39712 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39715 * gimplify.c (gimplify_init_constructor): If want_value and
39716 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39720 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39721 sequence. Formatting fixes.
39722 (doloop_optimize): Formatting fixes.
39725 * gcc.c (debug_level_greater_than_spec_func): New function.
39726 (static_spec_functions): Add debug-level-gt spec function.
39727 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39729 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39730 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39731 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39732 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39733 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39734 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39736 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39738 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39739 QImode fixups to general and mask registers only.
39741 2017-01-16 Carl Love <cel@us.ibm.com>
39743 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39744 for built-in functions
39745 vector signed char vec_nabs (vector signed char)
39746 vector signed short vec_nabs (vector signed short)
39747 vector signed int vec_nabs (vector signed int)
39748 vector signed long long vec_nabs (vector signed long long)
39749 vector float vec_nabs (vector float)
39750 vector double vec_nabs (vector double)
39751 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39753 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39754 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39755 * doc/extend.texi: Update the documentation file for the new built-in
39758 2017-01-16 Martin Sebor <msebor@redhat.com>
39760 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39763 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39765 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39766 UNSPEC_VSX__XXSPLTD to require special splat handling.
39768 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39771 * system.h: Poison strndup.
39773 2017-01-16 Alan Modra <amodra@gmail.com>
39776 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39779 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39781 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39783 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39785 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39786 call recog here. Assert that INSN_CODE (insn) is non-negative.
39788 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39791 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39793 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39794 in the currently scheduled RTL fragment.
39796 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39798 PR rtl-optimization/78751
39799 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39802 2017-01-14 Jeff Law <law@redhat.com>
39804 PR tree-optimization/79090
39805 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39806 variable length stores.
39807 (compute_trims): Delete dead assignment to *trim_tail.
39808 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39811 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39813 PR rtl-optimization/78626
39814 PR rtl-optimization/78727
39815 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39816 of a block, and split such blocks after everything else is finished.
39818 2017-01-14 Alan Modra <amodra@gmail.com>
39821 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39822 target legitimate_combined_insn.
39823 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39824 (rs6000_legitimate_combined_insn): New function.
39825 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39827 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39828 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39829 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39831 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39833 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39835 2017-01-13 Jeff Law <law@redhat.com>
39837 PR tree-optimization/33562
39838 PR tree-optimization/61912
39839 PR tree-optimization/77485
39840 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39842 (delete_dead_assignment): Likewise.
39843 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39844 statement to delete_dead_call and delete_dead_assignment.
39846 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39849 * substring-locations.c (format_warning_va): Strengthen case 1 so
39850 that both endpoints of the substring must be within the format
39851 range for just the substring to be printed.
39853 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39855 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39856 * config/i386/i386.c (ix86_target_string): Add missing options
39857 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39858 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39859 flags_other and ix86_target_other to flags2_other. Display unknown
39861 (ix86_valid_target_attribute_inner_p): Add missing options and
39862 reorder options by implied ISAs, as in ix86_target_string.
39864 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39866 * hash-table.h (hash_table::too_empty_p): New function.
39867 (hash_table::expand): Use it.
39868 (hash_table::traverse): Likewise.
39869 (hash_table::empty_slot): Use sizeof (value_type) instead of
39870 sizeof (PTR) to convert bytes to elements. Shrink the table
39871 if the current size is excessive for the current number of
39874 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39876 * ira-costs.c (record_reg_classes): Break from the inner loop
39877 early once alt_fail is known to be true. Update outer loop
39878 handling accordingly.
39880 2017-01-13 Jeff Law <law@redhat.com>
39882 * tree-ssa-dse.c (decrement_count): New function.
39883 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39884 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39885 when we know the partially dead statement is a mem* function.
39887 PR tree-optimization/61912
39888 PR tree-optimization/77485
39889 * tree-ssa-dse.c: Include expr.h.
39890 (maybe_trim_constructor_store): New function.
39891 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39893 PR tree-optimization/33562
39894 PR tree-optimization/61912
39895 PR tree-optimization/77485
39896 * doc/invoke.texi: Document new dse-max-object-size param.
39897 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39898 * tree-ssa-dse.c: Include params.h.
39899 (dse_store_status): New enum.
39900 (initialize_ao_ref_for_dse): New, partially extracted from
39902 (valid_ao_ref_for_dse, normalize_ref): New.
39903 (setup_live_bytes_from_ref, compute_trims): Likewise.
39904 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39905 (maybe_trim_partially_dead_store): Likewise.
39906 (maybe_trim_complex_store): Likewise.
39907 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39908 Track what bytes live from the original store. Return tri-state
39909 for dead, partially dead or live.
39910 (dse_dom_walker): Add constructor, destructor and new private members.
39911 (delete_dead_call, delete_dead_assignment): New extracted from
39913 (dse_optimize_stmt): Make a member of dse_dom_walker.
39914 Use initialize_ao_ref_for_dse.
39916 PR tree-optimization/33562
39917 PR tree-optimization/61912
39918 PR tree-optimization/77485
39919 * sbitmap.h (bitmap_count_bits): Prototype.
39920 (bitmap_clear_range, bitmap_set_range): Likewise.
39921 * sbitmap.c (bitmap_clear_range): New function.
39922 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39924 2017-01-13 Martin Liska <mliska@suse.cz>
39927 * function.c (set_cfun): Add new argument force.
39928 * function.h (set_cfun): Likewise.
39929 * ipa-inline-transform.c (inline_call): Use the function when
39930 strict alising from is dropped for function we inline to.
39932 2017-01-13 Richard Biener <rguenther@suse.de>
39934 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39935 for dumping GIMPLE INTEGER_CSTs.
39937 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39939 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39940 to 201112L since C++17.
39942 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39945 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39946 if -fsanitize=kernel-address is present.
39948 2017-01-13 Richard Biener <rguenther@suse.de>
39950 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39951 as _Literal ( type ) number in case usual suffixes do not
39952 preserve all information.
39954 2017-01-13 Richard Biener <rguenther@suse.de>
39956 PR tree-optimization/77283
39957 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39958 and ssa-iterators.h.
39959 (is_feasible_trace): Implement a cost model based on joiner
39962 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39965 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39966 char or short to __float128/_Float128 directly.
39968 2017-01-12 Martin Sebor <msebor@redhat.com>
39970 to -Wformat-overflow.
39971 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39972 (min_bytes_remaining): Same.
39973 (get_string_length): Same.
39974 (format_string): Same.
39975 (format_directive): Same.
39977 (pass_sprintf_length::handle_gimple_call): Same.
39979 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39981 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39982 info.nowrite calls with no lhs that can't throw. Return bool
39983 whether gsi_remove has been called or not.
39984 (pass_sprintf_length::handle_gimple_call): Return bool whether
39985 try_substitute_return_value called gsi_remove. Formatting fix.
39986 (pass_sprintf_length::execute): Don't use gsi_remove if
39987 handle_gimple_call returned true.
39990 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39991 be removed due to side-effects, don't remove following barrier nor
39992 turn the successor edge into fallthru edge.
39994 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39997 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39998 element-reversing loads and stores as not swappable.
40000 2017-01-12 Nathan Sidwell <nathan@acm.org>
40001 Nicolai Stange <nicstange@gmail.com>
40003 * combine.c (try_combine): Don't ignore result of overlap checking
40004 loop. Combine overlap & asm check into single loop.
40006 2017-01-12 Richard Biener <rguenther@suse.de>
40008 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40009 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40011 2017-01-12 Richard Biener <rguenther@suse.de>
40013 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40014 and TS_TARGET_OPTION directly derive from TS_BASE.
40015 * tree-core.h (tree_optimization_option): Derive from tree_base.
40016 (tree_target_option): Likewise.
40018 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40020 * config/i386/i386.c (memory_address_length): Increase len
40021 only when rip_relative_addr_p returns false.
40023 2017-01-11 Julia Koval <julia.koval@intel.com>
40025 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40026 (OPTION_MASK_ISA_SGX_SET): New.
40027 (ix86_handle_option): Handle OPT_msgx.
40028 * config.gcc: Added sgxintrin.h.
40029 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40030 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40031 * config/i386/i386.c (ix86_target_string): Add -msgx.
40033 (ix86_option_override_internal): Handle new options.
40034 (ix86_valid_target_attribute_inner_p): Add sgx.
40035 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40036 * config/i386/i386.opt: Add msgx.
40037 * config/i386/sgxintrin.h: New file.
40038 * config/i386/x86intrin.h: Add sgxintrin.h.
40040 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40043 * fold-const.c (maybe_nonzero_address): Return 1 for function
40045 (tree_single_nonzero_warnv_p): Don't handle function local objects
40049 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40052 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40055 * opts.c: Include "spellcheck.h"
40056 (struct string_fragment): New struct.
40057 (struct edit_distance_traits<const string_fragment &>): New
40059 (get_closest_sanitizer_option): New function.
40060 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40062 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40064 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40066 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40067 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40068 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40069 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40070 for initial die_offset if dwarf_split_debug_info.
40071 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40072 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40074 (output_skeleton_debug_sections): Formatting fix. Use
40075 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40076 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40078 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40080 * config/arm/cortex-a53.md: Add bypasses for
40081 cortex_a53_r2f_cvt.
40082 (cortex_a53_r2f): Only use for transfers.
40083 (cortex_a53_f2r): Likewise.
40084 (cortex_a53_r2f_cvt): Add reservation for conversions.
40085 (cortex_a53_f2r_cvt): Likewise.
40087 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40089 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40090 to all inlined functions, change static to extern.
40092 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40095 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40097 (arm_assemble_integer): Likewise.
40099 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40101 * config.gcc: Use new awk script to check CPU, FPU and architecture
40102 parameters for --with-... options.
40103 * config/arm/parsecpu.awk: New file
40104 * config/arm/arm-cpus.in: New file.
40105 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40107 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40109 * config/arm/t-arm: Update dependency rules.
40110 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40111 of processing .def files.
40112 * config/arm/genopt.sh: Deleted.
40113 * config/arm/gentune.sh: Deleted.
40114 * config/arm/arm-cores.def: Deleted.
40115 * config/arm/arm-arches.def: Deleted.
40116 * config/arm/arm-fpus.def: Deleted.
40117 * config/arm/arm-tune.md: Regenerated.
40118 * config/arm/arm-tables.opt: Regenerated.
40119 * config/arm/arm-cpu.h: New generated file.
40120 * config/arm/arm-cpu-data.h: New generated file.
40121 * config/arm/arm-cpu-cdata.h: New generated file.
40123 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40126 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40128 (input_varpool_node): Unpack dynamically_initialized bit.
40130 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40132 PR rtl-optimization/79032
40133 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40134 the alignment of the adjusted memory reference against that of MODE,
40135 instead of the alignment of the original memory reference.
40137 2017-01-11 Martin Jambor <mjambor@suse.cz>
40139 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40141 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40142 decorated functions.
40144 2017-01-11 Richard Biener <rguenther@suse.de>
40146 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40147 set range/nonnull info for PHI results. Do not set it on
40148 stmts marked for removal.
40150 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40152 * expr.c (store_field): In the bitfield case, fetch the return value
40153 from the registers before applying a single big-endian adjustment.
40154 Always do a final load for a BLKmode value not larger than a word.
40156 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40159 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40160 that we correctly handle column numbers greater than
40161 LINE_MAP_MAX_COLUMN_NUMBER.
40163 2017-01-10 Martin Sebor <msebor@redhat.com>
40165 PR middle-end/78245
40166 * gimple-ssa-sprintf.c (get_destination_size): Call
40167 {init,fini}object_sizes.
40168 * tree-object-size.c (addr_object_size): Adjust.
40169 (pass_through_call): Adjust.
40170 (pass_object_sizes::execute): Adjust.
40171 * tree-object-size.h (fini_object_sizes): Declare.
40173 2017-01-10 Martin Sebor <msebor@redhat.com>
40175 PR tree-optimization/78775
40176 * builtins.c (get_size_range): Move...
40177 * calls.c: ...to here.
40178 (alloc_max_size): Accept zero argument.
40179 (operand_signed_p): Remove.
40180 (maybe_warn_alloc_args_overflow): Call get_size_range.
40181 * calls.h (get_size_range): Declare.
40183 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40185 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40186 from TI's devices.csv file as of September 2016.
40187 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40189 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40191 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40192 * doc/invoke.texi: Likewise.
40193 * doc/md.texi: Likewise.
40194 * doc/objc.texi: Likewise.
40196 2017-01-10 Joshua Conner <joshconner@google.com>
40198 * config/arm/fuchsia-elf.h: New file.
40199 * config/fuchsia.h: New file.
40200 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40201 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40203 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40205 2016-01-10 Richard Biener <rguenther@suse.de>
40207 PR tree-optimization/79034
40208 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40209 Propagate out degenerate PHIs in the joiner.
40211 2017-01-10 Martin Liska <mliska@suse.cz>
40213 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40214 (sort_congruence_classes_by_decl_uid): Likewise.
40215 (sort_congruence_class_groups_by_decl_uid): Likewise.
40216 (sem_item_optimizer::merge_classes): Sort class, groups in these
40217 classes and members in the groups by DECL_UID of declarations.
40218 This would make merge operations stable.
40220 2017-01-10 Martin Liska <mliska@suse.cz>
40222 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40223 usage of m_classes_vec.
40224 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40225 (sem_item_optimizer::get_group_by_hash): Likewise.
40226 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40227 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40228 (sem_item_optimizer::verify_classes): Likewise.
40229 (sem_item_optimizer::process_cong_reduction): Likewise.
40230 (sem_item_optimizer::dump_cong_classes): Likewise.
40231 (sem_item_optimizer::merge_classes): Likewise.
40232 * ipa-icf.h (congruence_class_hash): Rename from
40233 congruence_class_group_hash. Remove declaration of m_classes_vec.
40235 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40237 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40238 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40239 * config.gcc: Add avx512vpopcntdqintrin.h.
40240 * config/i386/avx512vpopcntdqintrin.h: New.
40241 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40242 * config/i386/i386-builtin-types.def: Add new types.
40243 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40244 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40245 __builtin_ia32_vpopcountq_v8di_mask): New.
40246 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40247 __AVX512VPOPCNTDQ__.
40248 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40249 (PTA_AVX512VPOPCNTDQ): Define.
40250 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40251 TARGET_AVX512VPOPCNTDQ_P): Define.
40252 * config/i386/i386.opt: Add mavx512vpopcntdq.
40253 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40254 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40256 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40258 PR middle-end/77484
40259 * predict.def (PRED_CALL): Set to 67.
40261 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40263 * expr.c (store_field): In the bitfield case, if the value comes from
40264 a function call and is of an aggregate type returned in registers, do
40265 not modify the field mode; extract the value in all cases if the mode
40266 is BLKmode and the size is not larger than a word.
40268 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40271 * config/i386/cpuid.h: Fix undefined behavior.
40273 2017-01-04 Jeff Law <law@redhat.com>
40275 PR tree-optimization/79007
40276 PR tree-optimization/67955
40277 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40278 conservative for pt.null when flag_non_call_exceptions is on.
40280 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40282 PR translation/79019
40283 PR translation/79020
40284 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40285 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40286 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40288 * config/avr/avr.opt (maccumulate-args): Likewise.
40289 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40290 * common.opt (freport-bug): Likewise.
40291 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40292 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40293 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40294 translatable string.
40295 * config/i386/i386.c (function_value_32): Likewise.
40296 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40297 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40299 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40300 * common/config/msp430/msp430-common.c (msp430_handle_option):
40302 * symtab.c (symtab_node::verify_base): Likewise.
40303 * opts.c (set_debug_level): Likewise.
40304 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40305 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40306 missing whitespace to translatable strings.
40307 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40308 * config/sh/superh.h: Likewise.
40309 * config/i386/xopintrin.h: Likewise.
40310 * config/i386/znver1.md: Likewise.
40311 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40312 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40313 * double-int.h (struct double_int): Likewise.
40314 * double-int.c (div_and_round_double): Likewise.
40315 * wide-int.cc: Likewise.
40316 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40317 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40318 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40319 (crossjumps_occurred): ... this.
40320 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40323 PR tree-optimization/78899
40324 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40325 returning bool return struct loop *, NULL for failure and the new
40327 (versionable_outer_loop_p): Don't version outer loop if it has
40328 dont_vectorized bit set.
40329 (tree_if_conversion): When versioning outer loop, ensure
40330 tree_if_conversion is performed also on the inner loop of the
40331 non-vectorizable outer loop copy.
40332 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40333 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40334 prevent vectorization of it.
40335 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40336 the outer loop vectorization of the non-scalar version is attempted
40337 before vectorization of the inner loop in scalar version. If
40338 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40339 vectorization of its inner loop.
40340 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40341 has 2 inner loops, rename also on edges from bb whose single pred
40342 is outer_loop->header. Fix typo in function comment.
40344 2017-01-09 Martin Sebor <msebor@redhat.com>
40347 * asan.c (asan_emit_stack_protection): Increase local buffer size
40348 to avoid snprintf truncation warning.
40350 2017-01-09 Andrew Pinski <apinski@cavium.com>
40352 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40353 to reference thunderx2t99 for the tuning structure
40354 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40356 (thunderx2t99_extra_costs): This.
40357 * config/aarch64/aarch64-tune.md: Regenerate.
40358 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40359 (vulcan_addrcost_table): This.
40360 (vulcan_regmove_cost): Rename to ...
40361 (thunderx2t99_regmove_cost): This.
40362 (vulcan_vector_cost): Rename to ...
40363 (thunderx2t99_vector_cost): this.
40364 (vulcan_branch_cost): Rename to ...
40365 (thunderx2t99_branch_cost): This.
40366 (vulcan_tunings): Rename to ...
40367 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40368 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40370 2017-01-09 Martin Jambor <mjambor@suse.cz>
40374 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40375 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40376 (propagate_vr_accross_jump_function): Use the above function for all
40377 value range computations for pass-through jump functions and type
40378 converasion from explicit value range values.
40379 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40380 parameters from TYPE_ARG_TYPES.
40381 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40382 (ipa_write_node_info): Stream type of the actual argument.
40383 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40385 2017-01-09 Martin Liska <mliska@suse.cz>
40388 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40389 (lookup_compiler): Do not show error message with have_E.
40391 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40393 PR tree-optimization/78938
40394 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40395 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40396 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40397 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40400 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40402 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40405 2017-01-09 Richard Biener <rguenther@suse.de>
40407 PR tree-optimization/78997
40408 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40409 name condition properly.
40411 2017-01-09 Richard Biener <rguenther@suse.de>
40414 * dwarf2out.c (is_cxx): New overload with context.
40415 (is_naming_typedef_decl): Use it.
40417 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40419 * invoke.texi (Option Summary): Correct spacing in option lists
40420 and add line breaks to fix over-long lines.
40422 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40424 PR middle-end/17660
40426 * extend.texi (Common Variable Attributes): Add xref to GCC
40427 Internals manual to explain mode attribute keywords.
40429 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40432 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40433 and Preprocessor Options.
40434 (Options for Linking): Document -pthread here....
40435 (RS/6000 and PowerPC Options): ...not here.
40436 (Solaris 2 Options): ...or here.
40437 * doc/cppopts.texi: Document -pthread.
40439 2017-01-08 Martin Sebor <msebor@redhat.com>
40441 PR middle-end/77708
40442 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40443 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40444 New member functions.
40445 (format_directive): Used them.
40447 (pass_sprintf_length::handle_gimple_call): Same.
40448 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40449 to avoid truncation for any argument.
40450 (extract_affine_mul): Same.
40451 * tree.c (get_file_function_name): Same.
40453 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40455 PR middle-end/77484
40456 * predict.def (PRED_INDIR_CALL): Set to 86.
40458 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40460 PR preprocessor/54124
40461 * doc/cppopts.texi: Reformat -d subtable to list the full name
40462 of the options. Add cross-reference to the docs for the general
40463 compiler -d options.
40464 * doc/invoke.texi (Developer Options): Add cross-reference to the
40465 preprocessor-specific -d option documentation.
40467 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40469 PR preprocessor/13498
40470 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40471 redudant material, and reflect new command-line options.
40472 (System Headers): Likewise.
40474 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40476 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40477 -isystem, and -idirafter. Copy-edit.
40478 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40479 default for -ftrack-macro-expansion. Delete obsolete and
40480 badly-formatted implementation details about -fdebug-cpp output.
40481 * doc/cppwarnopts.texi: Copy-edit.
40483 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40486 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40487 that the transition from a max line width >= 1<<10 to narrower
40488 lines works correctly.
40490 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40492 * doc/options.texi (PerFunction): New.
40493 * opt-functions.awk (switch_flags): Map both Optimization and
40494 PerFunction to CL_OPTIMIZATION.
40495 * opth-gen.awk: Test for PerFunction flag along with
40497 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40498 it only when the latter is present. Skip those that don't in
40499 the hash function generator.
40500 * common.opt (fvar-tracking): Mark as PerFunction instead of
40502 (fvar-tracking-assignments): Likewise.
40503 (fvar-tracking-assignments-toggle): Likewise.
40504 (fvar-tracking-uninit): Likewise.
40506 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40508 PR translation/79018
40509 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40512 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40515 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40516 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40517 TARGET_LONG_JUMP_TABLE_OFFSETS.
40518 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40519 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40520 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40521 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40522 * config/m68k/m68k.md (tablejump expander): Likewise.
40523 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40524 TARGET_LONG_JUMP_TABLE_OFFSETS.
40525 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40526 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40528 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40529 David Holsgrove <david.holsgrove@xilinx.com>
40531 * common/config/microblaze/microblaze-common.c
40532 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40533 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40535 * config/microblaze/microblaze.c (microblaze_must_save_register)
40536 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40538 (microblaze_eh_return): New function.
40539 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40540 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40541 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40542 * config/microblaze/microblaze.md (eh_return): New pattern.
40544 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40546 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40547 GCC_DIAGNOSTIC_STRINGIFY): Define.
40549 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40551 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40553 * config/arm/arm.md (<mcrr>): New.
40555 * config/arm/arm.c (arm_arch5te): New.
40556 (arm_option_override): Set arm_arch5te.
40557 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40559 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40560 (arm_mcrr_qualifiers): ... this. New.
40561 (MRRC_QUALIFIERS): Define to...
40562 (arm_mrrc_qualifiers): ... this. New.
40563 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40565 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40566 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40567 (MRRCI, mrrc, MRRC): New.
40568 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40569 VUNSPEC_MRRC2): New.
40571 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40573 * config/arm/arm.md (<mcr>): New.
40575 * config/arm/arm.c (arm_coproc_builtin_available): Add
40576 support for mcr, mrc, mcr2 and mrc2.
40577 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40578 (arm_mcr_qualifiers): ... this. New.
40579 (MRC_QUALIFIERS): Define to ...
40580 (arm_mrc_qualifiers): ... this. New.
40581 (MCR_QUALIFIERS): Define to ...
40582 (arm_mcr_qualifiers): ... this. New.
40583 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40585 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40586 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40587 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40588 VUNSPEC_MRC2): New.
40590 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40592 * config/arm/arm.md (*ldc): New.
40596 * config/arm/arm.c (arm_coproc_builtin_available): Add
40597 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40598 (arm_coproc_ldc_stc_legitimate_address): New.
40599 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40600 'qualifier_const_pointer'.
40601 (LDC_QUALIFIERS): Define to...
40602 (arm_ldc_qualifiers): ... this. New.
40603 (STC_QUALIFIERS): Define to...
40604 (arm_stc_qualifiers): ... this. New.
40605 * config/arm/arm-protos.h
40606 (arm_coproc_ldc_stc_legitimate_address): New.
40607 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40608 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40609 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40610 stc2, stcl, stc2l): New.
40611 * config/arm/constraints.md (Uz): New.
40612 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40613 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40614 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40615 VUNSPEC_STC2L): New.
40617 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40619 * config/arm/arm.md (<cdp>): New.
40620 * config/arm/arm.c (neon_const_bounds): Rename this ...
40621 (arm_const_bounds): ... this.
40622 (arm_coproc_builtin_available): New.
40623 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40624 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40625 (CDP_QUALIFIERS): Define to...
40626 (arm_cdp_qualifiers): ... this. New.
40628 (arm_expand_builtin_args): Add case for 6 arguments.
40629 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40630 (arm_const_bounds): ... this.
40631 (arm_coproc_builtin_available): New.
40632 * config/arm/arm_acle.h (__arm_cdp): New.
40634 * config/arm/arm_acle_builtins.def (cdp): New.
40636 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40637 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40638 'arm_const_bounds'.
40639 * config/arm/types.md (coproc): New.
40640 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40641 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40642 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40643 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40645 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40647 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40648 (UBINOP_QUALIFIERS): New.
40650 (acle_builtin_data): New. Change comment.
40651 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40652 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40653 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40654 arm_acle_builtins.def.
40655 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40656 (arm_init_acle_builtins): New.
40657 (CRC32_BUILTIN): Remove.
40658 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40659 crc32cb, crc32ch and crc32cw.
40660 (arm_init_crc32_builtins): Remove.
40661 (arm_init_builtins): Use arm_init_acle_builtins rather
40662 than arm_init_crc32_builtins.
40663 (arm_expand_acle_builtin): New.
40664 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40665 * config/arm/arm_acle_builtins.def: New.
40667 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40669 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40670 (arm_builtin_datum): ... this.
40671 (arm_init_neon_builtin): Rename to ...
40672 (arm_init_builtin): ... this. Add a new parameters PREFIX
40673 and USE_SIG_IN_NAME.
40674 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40675 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40676 'arm_builtin_datum'.
40677 (arm_init_vfp_builtins): Likewise.
40678 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40679 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40680 (arm_expand_neon_args): Rename to ...
40681 (arm_expand_builtin_args): ... this. Rename builtin_arg
40682 enum values and differentiate between ARG_BUILTIN_MEMORY
40683 and ARG_BUILTIN_NEON_MEMORY.
40684 (arm_expand_neon_builtin_1): Rename to ...
40685 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40686 values, arm_expand_builtin_args and add bool parameter NEON.
40687 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40688 (arm_expand_vfp_builtin): Likewise.
40689 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40691 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40693 PR middle-end/77484
40694 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40695 * predict.c (tree_estimate_probability_bb): Reverse direction of
40696 polymorphic call predictor.
40698 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40700 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40701 (determine_pass_name_match): ...this new function and...
40702 (should_skip_pass_p): ...this new function.
40704 2017-01-06 Nathan Sidwell <nathan@acm.org>
40706 * ipa-visibility.c (function_and_variable_visibility): Reformat
40707 comments and long lines. Remove extrneous if.
40708 * symtab.c (symtab_node::make_decl_local): Fix code format.
40709 (symtab_node::set_section_for_node): Fix comment typo.
40711 2017-01-06 Martin Liska <mliska@suse.cz>
40714 * lra-constraints.c: Rename invariant to lra_invariant.
40715 * predict.c (set_even_probabilities): Initialize e to NULL.
40717 2017-01-05 Martin Sebor <msebor@redhat.com>
40719 PR tree-optimization/78910
40720 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40721 (format_integer): Correct off-by-one error in the handling
40722 of precision with negative numbers in signed conversions..
40724 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40726 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40728 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40730 PR tree-optimization/71016
40731 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40732 factor_out_conditional_conversion. Formatting fix.
40733 (factor_out_conditional_conversion): Add cond_stmt argument.
40734 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40735 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40738 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40740 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40741 read-rtl-function.o, and selftest-rtl.o.
40742 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40743 (selftest::aarch64_test_loading_full_dump): New function.
40744 (selftest::aarch64_run_selftests): New function.
40745 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40746 selftest::aarch64_run_selftests.
40747 * config/i386/i386.c
40748 (selftest::ix86_test_loading_dump_fragment_1): New function.
40749 (selftest::ix86_test_loading_call_insn): New function.
40750 (selftest::ix86_test_loading_full_dump): New function.
40751 (selftest::ix86_test_loading_unspec): New function.
40752 (selftest::ix86_run_selftests): Call the new functions.
40753 * emit-rtl.c (maybe_set_max_label_num): New function.
40754 * emit-rtl.h (maybe_set_max_label_num): New decl.
40755 * function.c (instantiate_decls): Guard call to
40756 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40757 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40759 * gensupport.c (gen_reader::gen_reader): Pass "false"
40760 for new "compact" param of rtx_reader.
40761 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40762 rather than an empty string for NULL strings.
40763 * read-md.c: Potentially include config.h rather than bconfig.h.
40764 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40765 (have_error): New global, copied from errors.c.
40766 (md_reader::read_name): Rename to...
40767 (md_reader::read_name_1): ...this, adding "out_loc" param,
40768 and converting "missing name or number" to returning false, rather
40770 (md_reader::read_name): Reimplement in terms of read_name_1.
40771 (md_reader::read_name_or_nil): New function.
40772 (md_reader::read_string): Handle "(nil)" by returning NULL.
40773 (md_reader::md_reader): Add new param "compact".
40774 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40775 (md_reader::read_file): New method.
40776 * read-md.h (md_reader::md_reader): Add new param "compact".
40777 (md_reader::read_file): New method.
40778 (md_reader::is_compact): New accessor.
40779 (md_reader::read_name): Convert return type from void to file_location.
40780 (md_reader::read_name_or_nil): New decl.
40781 (md_reader::read_name_1): New decl.
40782 (md_reader::m_compact): New field.
40783 (noop_reader::noop_reader): Pass "false" for new "compact" param
40785 (rtx_reader::rtx_reader): Add new "compact" param.
40786 (rtx_reader::read_rtx_operand): Make virtual and convert return
40787 type from void to rtx.
40788 (rtx_reader::read_until): New decl.
40789 (rtx_reader::handle_any_trailing_information): New virtual function.
40790 (rtx_reader::postprocess): New virtual function.
40791 (rtx_reader::finalize_string): New virtual function.
40792 (rtx_reader::m_in_call_function_usage): New field.
40793 (rtx_reader::m_reuse_rtx_by_id): New field.
40794 * read-rtl-function.c: New file.
40795 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40796 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40797 (selftest::verify_three_block_rtl_cfg): New decl.
40798 * read-rtl-function.h: New file.
40799 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40800 For host, include function.h, memmodel.h, and emit-rtl.h.
40801 (one_time_initialization): New function.
40802 (struct compact_insn_name): New struct.
40803 (compact_insn_names): New array.
40804 (find_code): Handle insn codes in compact dumps.
40805 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40806 (bind_subst_iter_and_attr): Likewise.
40807 (add_condition_to_string): Likewise.
40808 (add_condition_to_rtx): Likewise.
40809 (apply_attribute_uses): Likewise.
40810 (add_current_iterators): Likewise.
40811 (apply_iterators): Likewise.
40812 (initialize_iterators): Guard usage of apply_subst_iterator with
40813 #ifdef GENERATOR_FILE.
40814 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40815 (md_reader::read_mapping): Likewise.
40816 (add_define_attr_for_define_subst): Likewise.
40817 (add_define_subst_attr): Likewise.
40818 (read_subst_mapping): Likewise.
40819 (check_code_iterator): Likewise.
40820 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40822 (one_time_initialization): New function.
40823 (rtx_reader::read_until): New method.
40824 (read_flags): New function.
40825 (parse_reg_note_name): New function.
40826 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40827 Handle reuse_rtx ids.
40828 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40829 Add parsing support for RTL dumps, mirroring the special-cases in
40830 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40831 values, and calling handle_any_trailing_information.
40832 (rtx_reader::read_rtx_operand): Convert return type from void
40833 to rtx, returning return_rtx. Handle case 'e'. Call
40834 finalize_string on XSTR and XTMPL fields.
40835 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40836 "(nil)" values were omitted. Call the postprocess vfunc on the
40838 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40839 class ctor. Initialize m_in_call_function_usage. Call
40840 one_time_initialization.
40841 * rtl-tests.c (selftest::test_uncond_jump): Call
40842 set_new_first_and_last_insn.
40843 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40844 * selftest-rtl.c: New file.
40845 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40846 (selftest::get_insn_by_uid): New decl.
40847 * selftest-run-tests.c (selftest::run_tests): Call
40848 read_rtl_function_c_tests.
40849 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40850 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40853 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40855 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40856 operands in a special way. Assert that pos+len <= mode precision.
40858 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40860 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40861 3 argument Alias with unlimited for the negative form.
40862 (fno-vect-cost-model): Removed.
40864 2017-01-05 Martin Liska <mliska@suse.cz>
40866 * hsa-gen.c (gen_hsa_divmod): New function.
40867 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40869 2017-01-05 Martin Liska <mliska@suse.cz>
40872 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40875 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40877 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40878 small constant length operands.
40880 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40882 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40883 between loop iterations.
40885 2017-01-05 Martin Liska <mliska@suse.cz>
40888 * gimplify.c (gimplify_decl_expr): Compare to
40889 asan_poisoned_variables instread of checking flags.
40890 (gimplify_target_expr): Likewise.
40891 (gimplify_expr): Likewise.
40892 (gimplify_function_tree): Conditionally initialize
40893 asan_poisoned_variables.
40895 2017-01-04 Jeff Law <law@redhat.com>
40897 PR tree-optimizatin/78812
40898 * rtl.h (contains_mem_rtx_p): Prototype.
40899 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40900 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40901 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40902 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40903 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40905 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40907 * input.c (assert_char_at_range): Default-initialize actual_range.
40909 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40911 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40912 to match the caller.
40914 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40916 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40917 insns after final jump in test to emit dummy move.
40919 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40921 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40922 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40924 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40926 * multiple_target.c (create_dispatcher_calls): Init e_next.
40927 * tree-ssa-loop-split.c (split_loop): Init border.
40928 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40931 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40936 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40937 (altivec_register_operand): Do not return true if the operand
40938 contains a SUBREG mixing SImode and SFmode.
40939 (vsx_register_operand): Likewise.
40940 (vsx_reg_sfsubreg_ok): New predicate.
40941 (vfloat_operand): Do not return true if the operand contains a
40942 SUBREG mixing SImode and SFmode.
40943 (vint_operand): Likewise.
40944 (vlogical_operand): Likewise.
40945 (gpc_reg_operand): Likewise.
40946 (int_reg_operand): Likewise.
40947 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40948 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40949 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40951 (rs6000_emit_move_si_sf_subreg): New helper function.
40952 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40953 fixup SUBREGs involving SImode and SFmode.
40954 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40955 numbers for the new peephole2 optimization.
40956 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40957 the GLIBC math library that do AND/IOR/XOR operations on single
40958 precision floating point.
40959 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40960 target macros to say whether we need to avoid SUBREGs mixing
40962 (TARGET_ALLOW_SF_SUBREG): Likewise.
40963 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40964 (UNSPEC_SI_FROM_SF): Likewise.
40965 (iorxor): Change spacing.
40966 (and_ior_xor): New iterator for AND, IOR, and XOR.
40967 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40968 (movdi_from_sf_zero_ext): Likewise.
40969 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40970 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40971 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40972 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40973 (fms<mode>4): Likewise.
40974 (fnma<mode>4): Likewise.
40975 (fnms<mode>4): Likewise.
40976 (nfma<mode>4): Likewise.
40977 (nfms<mode>4): Likewise.
40979 2017-01-04 Marek Polacek <polacek@redhat.com>
40982 * doc/invoke.texi: Document -Wpointer-compare.
40984 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40986 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40989 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40990 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40993 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40996 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40997 documentation of the powerpc_popcntb_ok attribute.
40998 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40999 code to issue warning messages if a requested CPU configuration is
41000 not supported by the binary (assembler and loader) toolchain.
41001 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41002 made to define a built-in function that has been disabled.
41003 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41004 made to define a built-in function that has been disabled.
41005 (altivec_init_builtins): Add comment explaining why definition
41006 of the DST built-in functions is not preceded by an assertion
41007 check. Add assertions to prevent ICE if attempts are made to
41008 define an altivec predicate or an abs* built-in function that has
41010 (htm_init_builtins): Add comment explaining why definition of the
41011 htm built-in functions is not preceded by an assertion check.
41013 2017-01-04 Jeff Law <law@redhat.com>
41015 PR tree-optimizatin/67955
41016 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41017 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41018 the points-to solution does not include pt_null. Use DECL_PT_UID
41021 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41023 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41024 Use gen_int_mode instead of gen_lopwart for const_int operands.
41026 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41028 PR tree-optimization/71563
41029 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41030 out of range value - has low bits known to be zero.
41032 2017-01-04 Alan Modra <amodra@gmail.com>
41034 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41035 * configure: Regenerate.
41036 * config.in: Regenerate.
41038 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41041 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41042 a substring of the message, but strcmp with the whole message. Ifdef
41043 ENABLE_NLS, translate the message first using dgettext.
41045 2017-01-03 Jeff Law <law@redhat.com>
41047 PR tree-optimizatin/78856
41048 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41049 (mark_threaded_blocks): Remove code to truncate thread paths that
41050 cross multiple loop headers. Instead invalidate the cached loop
41051 iteration information and handle case of a thread path walking
41052 into an irreducible region.
41054 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41057 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41058 assertions. Add support for doing the signbit if the IEEE 128-bit
41059 floating point value is in a GPR.
41060 * config/rs6000/rs6000.md (Fsignbit): Delete.
41061 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41062 Update the length attribute if the value is in a GPR.
41063 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41064 the sign or zero extension instruction, since the value is always 0/1.
41065 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41068 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41069 extracting SImode to a GPR register so that we can generate a
41070 store, limit the vector to be in a traditional Altivec register
41071 for the vextuwrx instruction.
41073 2017-01-03 Ian Lance Taylor <iant@google.com>
41075 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41077 2017-01-03 Martin Sebor <msebor@redhat.com>
41079 PR tree-optimization/78696
41080 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41081 precision. Use MPFR for %f for greater fidelity. Correct handling
41083 (pass_sprintf_length::compute_format_length): Set width and precision
41084 specified by asrerisk to void_node for vararg functions.
41085 (try_substitute_return_value): Adjust dump output.
41087 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41089 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41091 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41093 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41094 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41095 -mlra/-mno-lra was passed to the compiler.
41097 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41099 PR rtl-optimization/65618
41100 * emit-rtl.c (try_split): Move initialization of "before" and
41101 "after" to just before the call to emit_insn_after_setloc.
41103 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41105 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41107 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41109 * dwarf2out.c (gen_enumeration_type_die): When
41110 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41112 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41114 PR tree-optimization/78965
41115 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41116 Change first argument from const call_info & to call_info &. For %n
41117 set info.nowrite to false.
41119 PR middle-end/78901
41120 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41121 possibly throwing calls.
41123 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41124 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41125 and fns handling, rather than in a separate case SSA_NAME.
41127 2017-01-02 Jeff Law <law@redhat.com>
41129 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41130 fixes for first_period and second_period variables.
41132 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41135 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41136 (*insvqi_1): New insn pattern.
41137 (*insvqi_1_mem_rex64): Ditto.
41138 (*insvqi_2): Ditto.
41139 (*insvqi_3): Rename from *insvqi.
41141 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41143 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41145 * doc/cfg.texi (Edges): Remove reference to Java.
41146 (Maintaining the CFG): Ditto.
41148 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41150 PR middle-end/77674
41151 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41152 transparent aliases.
41154 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41156 PR middle-end/77484
41157 * predict.def (PRED_CALL): Update hitrate.
41158 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41159 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41160 into direct/indirect/polymorphic variants.
41162 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41164 Update copyright years.
41166 * gcc.c (process_command): Update copyright notice dates.
41167 * gcov-dump.c (print_version): Ditto.
41168 * gcov.c (print_version): Ditto.
41169 * gcov-tool.c (print_version): Ditto.
41170 * gengtype.c (create_file): Ditto.
41171 * doc/cpp.texi: Bump @copying's copyright year.
41172 * doc/cppinternals.texi: Ditto.
41173 * doc/gcc.texi: Ditto.
41174 * doc/gccint.texi: Ditto.
41175 * doc/gcov.texi: Ditto.
41176 * doc/install.texi: Ditto.
41177 * doc/invoke.texi: Ditto.
41179 Copyright (C) 2017 Free Software Foundation, Inc.
41181 Copying and distribution of this file, with or without modification,
41182 are permitted in any medium without royalty provided the copyright
41183 notice and this notice are preserved.