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 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
8 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
9 Alan Hayward <alan.hayward@arm.com>
10 David Sherwood <david.sherwood@arm.com>
12 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
13 as a poly_uint64 rather than an unsigned int.
14 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
16 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
17 Alan Hayward <alan.hayward@arm.com>
18 David Sherwood <david.sherwood@arm.com>
20 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
22 (decode_addr_const): Update accordingly.
24 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
25 Alan Hayward <alan.hayward@arm.com>
26 David Sherwood <david.sherwood@arm.com>
28 * tree.h (bit_field_size, bit_field_offset): New functions.
29 * hsa-gen.c (gen_hsa_addr): Use them.
30 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
31 (simplify_vector_constructor): Likewise.
32 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
33 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
34 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
35 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
38 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
39 Alan Hayward <alan.hayward@arm.com>
40 David Sherwood <david.sherwood@arm.com>
42 * expr.h (emit_group_load, emit_group_load_into_temps)
43 (emit_group_store): Take the size as a poly_int64 rather than an int.
44 * expr.c (emit_group_load_1, emit_group_load): Likewise.
45 (emit_group_load_into_temp, emit_group_store): Likewise.
47 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
48 Alan Hayward <alan.hayward@arm.com>
49 David Sherwood <david.sherwood@arm.com>
51 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
52 an unsigned int to a poly_uint64.
53 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
54 sizes as poly_uint64s rather than unsigned ints.
55 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
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 * lra-constraints.c (emit_inc): Change inc_amount from an int
65 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
66 Alan Hayward <alan.hayward@arm.com>
67 David Sherwood <david.sherwood@arm.com>
69 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
71 (add_stack_var, stack_var_cmp, partition_stack_vars)
72 (dump_stack_var_partition): Update accordingly.
73 (alloc_stack_frame_space): Take the size as a poly_int64 rather
75 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
76 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
77 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
78 rather than a HOST_WIDE_INT.
80 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
81 Alan Hayward <alan.hayward@arm.com>
82 David Sherwood <david.sherwood@arm.com>
84 * target.def (return_pops_args): Treat both the input and output
85 sizes as poly_int64s rather than HOST_WIDE_INTS.
86 * targhooks.h (default_return_pops_args): Update accordingly.
87 * targhooks.c (default_return_pops_args): Likewise.
88 * doc/tm.texi: Regenerate.
89 * emit-rtl.h (incoming_args): Change pops_args, size and
90 outgoing_args_size from int to poly_int64_pod.
91 * function.h (expr_status): Change x_pending_stack_adjust and
92 x_stack_pointer_delta from int to poly_int64.
93 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
94 (ARGS_SIZE_RTX): Update accordingly.
95 * calls.c (highest_outgoing_arg_in_use): Change from int to
97 (stack_usage_watermark, stored_args_watermark): New variables.
98 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
99 (emit_call_1): Change the stack_size and rounded_stack_size
100 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
102 (save_fixed_argument_area): Check stack_usage_watermark.
103 (initialize_argument_information): Change old_pending_adj from
104 a HOST_WIDE_INT * to a poly_int64_pod *.
105 (compute_argument_block_size): Return the size as a poly_int64
107 (finalize_must_preallocate): Track polynomial argument sizes.
108 (compute_argument_addresses): Likewise.
109 (internal_arg_pointer_based_exp): Track polynomial offsets.
110 (mem_overlaps_already_clobbered_arg_p): Rename to...
111 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
112 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
113 Check stored_args_used_watermark.
114 (load_register_parameters): Update accordingly.
115 (check_sibcall_argument_overlap_1): Likewise.
116 (combine_pending_stack_adjustment_and_call): Take the unadjusted
117 args size as a poly_int64 rather than an int. Return a bool
118 indicating whether the optimization was possible and return
119 the new adjustment by reference.
120 (check_sibcall_argument_overlap): Track polynomail argument sizes.
121 Update stored_args_watermark.
122 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
123 (expand_call): Likewise. Maintain stack_usage_watermark and
124 stored_args_watermark. Update calls to
125 combine_pending_stack_adjustment_and_call.
126 (emit_library_call_value_1): Handle polynomial argument sizes.
127 Call stack_region_maybe_used_p and mark_stack_region_used.
128 Maintain stack_usage_watermark.
129 (store_one_arg): Likewise. Update call to
130 mem_overlaps_already_clobbered_arg_p.
131 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
133 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
134 * config/microblaze/microblaze.c (microblaze_function_prologue):
136 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
137 TARGET_RETURN_POPS_ARGS interface.
138 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
140 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
141 * config/i386/i386.c (ix86_return_pops_args): Update for new
142 TARGET_RETURN_POPS_ARGS interface.
143 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
144 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
145 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
146 TARGET_RETURN_POPS_ARGS interface.
147 * config/vax/vax.c (vax_return_pops_args): Likewise.
148 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
149 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
150 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
151 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
152 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
153 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
154 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
155 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
156 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
157 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
158 and x_stack_pointer_delta from int to poly_int64.
159 * dojump.c (do_pending_stack_adjust): Update accordingly.
160 * explow.c (allocate_dynamic_stack_space): Handle polynomial
161 stack_pointer_deltas.
162 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
163 (pad_to_arg_alignment): Track polynomial offsets.
164 (assign_parm_find_stack_rtl): Likewise.
165 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
166 * toplev.c (output_stack_usage): Update reference to
167 current_function_pushed_stack_size.
169 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
170 Alan Hayward <alan.hayward@arm.com>
171 David Sherwood <david.sherwood@arm.com>
173 * function.c (in_arg_offset, var_offset, dynamic_offset)
174 (out_arg_offset, cfa_offset): Change from int to poly_int64.
175 (instantiate_new_reg): Return the new offset as a poly_int64_pod
176 rather than a HOST_WIDE_INT.
177 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
178 (instantiate_virtual_regs_in_insn): Likewise.
180 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
181 Alan Hayward <alan.hayward@arm.com>
182 David Sherwood <david.sherwood@arm.com>
184 * rtl.h (get_args_size, add_args_size_note): New functions.
185 (find_args_size_adjust): Return a poly_int64 rather than a
187 (fixup_args_size_notes): Likewise. Make the same change to the
188 end_args_size parameter.
189 * rtlanal.c (get_args_size, add_args_size_note): New functions.
190 * builtins.c (expand_builtin_trap): Use add_args_size_note.
191 * calls.c (emit_call_1): Likewise.
192 * explow.c (adjust_stack_1): Likewise.
193 * cfgcleanup.c (old_insns_match_p): Update use of
194 find_args_size_adjust.
195 * combine.c (distribute_notes): Track polynomial arg sizes.
196 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
197 end_true_args_size, beg_delay_args_size and end_delay_args_size
198 from HOST_WIDE_INT to poly_int64.
199 (add_cfi_args_size): Take the args_size as a poly_int64 rather
200 than a HOST_WIDE_INT.
201 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
202 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
203 polynomial arg sizes.
204 * emit-rtl.c (try_split): Use get_args_size.
205 * recog.c (peep2_attempt): Likewise.
206 * reload1.c (reload_as_needed): Likewise.
207 * expr.c (find_args_size_adjust): Return the adjustment as a
208 poly_int64 rather than a HOST_WIDE_INT.
209 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
210 to a poly_int64 and change the return type in the same way.
211 (emit_single_push_insn): Track polynomial arg sizes.
213 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
214 Alan Hayward <alan.hayward@arm.com>
215 David Sherwood <david.sherwood@arm.com>
217 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
218 from HOST_WIDE_INT to poly_int64.
219 * expr.c (push_block, emit_push_insn): Likewise.
221 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
222 Alan Hayward <alan.hayward@arm.com>
223 David Sherwood <david.sherwood@arm.com>
225 * function.h (frame_space): Change start and length from HOST_WIDE_INT
227 (get_frame_size): Return the size as a poly_int64 rather than a
229 (frame_offset_overflow): Take the offset as a poly_int64 rather
230 than a HOST_WIDE_INT.
231 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
232 (assign_stack_temp): Likewise for the size.
233 * function.c (get_frame_size): Return a poly_int64 rather than
235 (frame_offset_overflow): Take the offset as a poly_int64 rather
236 than a HOST_WIDE_INT.
237 (try_fit_stack_local): Take the start, length and size as poly_int64s
238 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
239 rather than a HOST_WIDE_INT.
240 (add_frame_space): Take the start and end as poly_int64s rather than
242 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
243 (assign_stack_temp): Likewise for the size.
244 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
246 (find_temp_slot_from_address): Handle polynomial offsets.
247 (combine_temp_slots): Likewise.
248 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
250 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
251 poly_int64 rather than a HOST_WIDE_INT.
252 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
253 than a HOST_WIDE_INT.
254 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
255 polynomial frame offsets.
256 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
257 as a poly_int64 rather than an int.
258 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
259 * config/v850/v850-protos.h (compute_frame_size): Likewise.
260 * config/v850/v850.c (compute_frame_size): Likewise.
261 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
262 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
263 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
264 * config/pa/pa.c (pa_compute_frame_size): Likewise.
265 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
266 rather than a HOST_WIDE_INT.
267 * explow.c (get_dynamic_stack_base): Likewise.
268 * final.c (final_start_function): Use the constant lower bound
269 of the frame size for -Wframe-larger-than.
270 * ira.c (do_reload): Adjust for new get_frame_size return type.
271 * lra.c (lra): Likewise.
272 * reload1.c (reload): Likewise.
273 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
274 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
275 * rtlanal.c (get_initial_register_offset): Return the offset as
276 a poly_int64 rather than a HOST_WIDE_INT.
278 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
279 Alan Hayward <alan.hayward@arm.com>
280 David Sherwood <david.sherwood@arm.com>
282 * reload1.c (elim_table): Change initial_offset, offset and
283 previous_offset from HOST_WIDE_INT to poly_int64_pod.
284 (offsets_at): Change the target array's element type from
285 HOST_WIDE_INT to poly_int64_pod.
286 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
287 (elimination_costs_in_insn, update_eliminable_offsets)
288 (verify_initial_elim_offsets, set_offsets_for_label)
289 (init_eliminable_invariants): Update after above changes.
291 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
292 Alan Hayward <alan.hayward@arm.com>
293 David Sherwood <david.sherwood@arm.com>
295 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
296 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
297 (decomposition): Change start and end from HOST_WIDE_INT
299 (decompose, immune_p): Update accordingly.
300 (find_inc_amount): Return a poly_int64 rather than an int.
301 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
304 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
305 Alan Hayward <alan.hayward@arm.com>
306 David Sherwood <david.sherwood@arm.com>
308 * tree.h (get_inner_reference): Return the bitsize and bitpos
309 as poly_int64_pods rather than HOST_WIDE_INT.
310 * fold-const.h (ptr_difference_const): Return the pointer difference
311 as a poly_int64_pod rather than a HOST_WIDE_INT.
312 * expr.c (get_inner_reference): Return the bitsize and bitpos
313 as poly_int64_pods rather than HOST_WIDE_INT.
314 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
316 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
317 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
318 (optimize_bit_field_compare): Update call to get_inner_reference.
319 (decode_field_reference): Likewise.
320 (fold_unary_loc): Track polynomial offsets and sizes.
321 (split_address_to_core_and_offset): Return the bitpos as a
322 poly_int64_pod rather than a HOST_WIDE_INT.
323 (ptr_difference_const): Likewise for the pointer difference.
324 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
325 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
326 * dbxout.c (dbxout_expand_expr): Likewise.
327 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
328 (loc_list_from_tree_1, fortran_common): Likewise.
329 * gimple-laddress.c (pass_laddress::execute): Likewise.
330 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
331 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
332 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
333 * tree-affine.c (tree_to_aff_combination): Likewise.
334 (get_inner_reference_aff): Likewise.
335 * tree-data-ref.c (split_constant_offset_1): Likewise.
336 (dr_analyze_innermost): Likewise.
337 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
338 * tree-sra.c (ipa_sra_check_caller): Likewise.
339 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
340 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
341 (instrument_bool_enum_load, instrument_object_size): Likewise.
342 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
343 to get_inner_reference.
344 * hsa-gen.c (gen_hsa_addr): Likewise.
345 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
346 * tsan.c (instrument_expr): Likewise.
347 * match.pd: Update call to ptr_difference_const.
349 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
350 Alan Hayward <alan.hayward@arm.com>
351 David Sherwood <david.sherwood@arm.com>
353 * fold-const.c (fold_comparison): Track sizes and offsets as
354 poly_int64s rather than HOST_WIDE_INTs when folding address
357 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
358 Alan Hayward <alan.hayward@arm.com>
359 David Sherwood <david.sherwood@arm.com>
361 * expr.h (get_bit_range): Return the bitstart and bitend as
362 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
363 as a poly_int64 rather than a HOST_WIDE_INT.
364 * expr.c (get_bit_range): Likewise.
365 (expand_assignment): Update call accordingly.
366 * fold-const.c (optimize_bit_field_compare): Likewise.
368 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
369 Alan Hayward <alan.hayward@arm.com>
370 David Sherwood <david.sherwood@arm.com>
372 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
373 (aarch64_legitimate_address_p): Use it instead of an rtx code,
374 as an optional final parameter.
375 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
376 (aarch64_legitimate_address_p): Likewise.
377 (aarch64_print_address_internal): Take an aarch64_addr_query_type
378 instead of an rtx code.
379 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
380 (aarch64_legitimate_address_hook_p): Likewise.
381 (aarch64_print_ldpstp_address): Likewise.
382 (aarch64_print_operand_address): Likewise.
383 (aarch64_address_cost): Likewise.
384 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
385 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
386 (aarch64_mem_pair_lanes_operand): Likewise.
388 2017-12-20 Richard Biener <rguenther@suse.de>
390 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
391 update_stmt_if_modified.
393 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
395 PR tree-optimization/83491
396 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
397 before walking uses. Improve coding style and comments.
399 2017-12-20 Tom de Vries <tom@codesourcery.com>
401 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
403 2017-12-20 Jakub Jelinek <jakub@redhat.com>
406 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
408 (pass_ipa_free_fn_summary): Add small_p private data member,
409 initialize to false in the ctor.
410 (pass_ipa_free_fn_summary::clone,
411 pass_ipa_free_fn_summary::set_pass_param,
412 pass_ipa_free_fn_summary::gate): New methods.
413 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
414 | TODO_dump_symtab if small_p.
415 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
416 entry and add another instance of the pass with false parm after
418 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
419 ipa_free_fn_summary here.
421 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
423 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
425 2017-12-20 Martin Sebor <msebor@redhat.com>
428 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
429 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
430 (expand_builtin_strncmp): Same.
432 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
435 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
436 allowed debug stmts before labels.
437 (expand_gimple_basic_block): Likewise.
438 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
439 * gimple-iterator.h (gsi_after_labels): Likewise.
440 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
441 rename reused variable, and simplify using gsi_move_before.
442 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
443 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
444 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
445 (gimple_verify_flow_info, gimple_block_label): Likewise.
446 (make_blocks): Move debug markers after adjacent labels.
447 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
448 allowed debug insns outside blocks.
449 * df-scan.c (df_insn_delete): Likewise.
450 * lra-constraints.c (update_ebb_live_info): Likewise.
451 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
452 (vt_initialize, delete_vta_debug_insns): Likewise.
453 (reemit_marker_as_note): Drop BB parm. Adjust callers.
455 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
456 Alan Hayward <alan.hayward@arm.com>
457 David Sherwood <david.sherwood@arm.com>
459 * poly-int-types.h (round_down_to_byte_boundary): New macro.
460 (round_up_to_byte_boundary): Likewise.
461 * expr.h (get_bit_range): Add temporary shim.
462 * gimple-ssa-store-merging.c (store_operand_info): Change the
463 bitsize, bitpos, bitregion_start and bitregion_end fields from
464 unsigned HOST_WIDE_INT to poly_uint64.
465 (merged_store_group): Likewise load_align_base.
466 (compatible_load_p, compatible_load_p): Update accordingly.
467 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
468 (split_group, imm_store_chain_info::output_merged_store): Likewise.
469 (mem_valid_for_store_merging): Return the bitsize, bitpos,
470 bitregion_start and bitregion_end as poly_uint64s rather than
471 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
472 (handled_load): Take the bitsize, bitpos,
473 bitregion_start and bitregion_end as poly_uint64s rather than
474 unsigned HOST_WIDE_INTs.
475 (pass_store_merging::process_store): Update call to
476 mem_valid_for_store_merging.
478 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
479 Alan Hayward <alan.hayward@arm.com>
480 David Sherwood <david.sherwood@arm.com>
482 * builtins.c (get_object_alignment_2): Track polynomial offsets
483 and sizes. Update the alignment handling.
485 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
486 Alan Hayward <alan.hayward@arm.com>
487 David Sherwood <david.sherwood@arm.com>
489 * tree.h (get_inner_reference): Add a version that returns the
490 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
491 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
492 the case in which bitpos is not associated with the first byte.
494 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
495 Alan Hayward <alan.hayward@arm.com>
496 David Sherwood <david.sherwood@arm.com>
498 * tree-affine.h (get_inner_reference_aff): Return the size as a
500 * tree-affine.c (get_inner_reference_aff): Likewise.
501 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
502 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
504 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
505 Alan Hayward <alan.hayward@arm.com>
506 David Sherwood <david.sherwood@arm.com>
508 * fold-const.c (pointer_may_wrap_p): Take the offset as a
509 HOST_WIDE_INT rather than a poly_int64.
511 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
512 Alan Hayward <alan.hayward@arm.com>
513 David Sherwood <david.sherwood@arm.com>
515 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
516 HOST_WIDE_INT to poly_int64_pod.
517 (perform_symbolic_merge): Update accordingly.
518 (bswap_replace): Likewise.
520 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
521 Alan Hayward <alan.hayward@arm.com>
522 David Sherwood <david.sherwood@arm.com>
524 * tree-affine.h (aff_tree::offset): Change from widest_int
526 (wide_int_ext_for_comb): Delete.
527 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
528 constants as poly_widest_int rather than widest_int.
529 (aff_combination_constant_multiple_p): Return the multiplier
530 as a poly_widest_int.
531 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
533 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
534 version static and add an overload for poly_widest_int.
535 (aff_combination_const, aff_combination_add_cst)
536 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
537 the constants as poly_widest_int rather than widest_int.
538 (tree_to_aff_combination): Generalize INTEGER_CST case to
540 (aff_combination_to_tree): Track offsets as poly_widest_ints.
541 (aff_combination_add_product, aff_combination_mult): Handle
543 (aff_combination_constant_multiple_p): Return the multiplier
544 as a poly_widest_int.
545 * tree-predcom.c (determine_offset): Return the offset as a
547 (split_data_refs_to_components, suitable_component_p): Update
549 (valid_initializer_p): Update call to
550 aff_combination_constant_multiple_p.
551 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
552 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
553 as a poly_int64 rather than a HOST_WIDE_INT.
554 (get_address_cost): Handle polynomial offsets.
555 (iv_elimination_compare_lt): Likewise.
556 (rewrite_use_nonlinear_expr): Likewise.
558 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
559 Alan Hayward <alan.hayward@arm.com>
560 David Sherwood <david.sherwood@arm.com>
562 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
563 as a poly_int64_pod rather than a HOST_WIDE_INT.
564 (get_addr_base_and_unit_offset): Likewise.
565 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
566 (get_addr_base_and_unit_offset): Likewise.
567 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
568 to poly_int64 in example.
569 * fold-const.c (fold_binary_loc): Update call to
570 get_addr_base_and_unit_offset.
571 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
572 (maybe_canonicalize_mem_ref_addr): Likewise.
573 (gimple_fold_stmt_to_constant_1): Likewise.
574 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
576 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
577 * match.pd: Likewise.
578 * omp-low.c (lower_omp_target): Likewise.
579 * tree-sra.c (build_ref_for_offset): Likewise.
580 (build_debug_ref_for_model): Likewise.
581 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
582 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
583 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
584 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
585 (constant_pointer_difference): Likewise.
586 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
587 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
588 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
589 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
590 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
591 (set_ssa_val_to): Likewise.
592 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
593 (maybe_diag_stxncpy_trunc): Likewise.
594 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
595 * tree.c (build_simple_mem_ref_loc): Likewise.
596 (array_at_struct_end_p): Likewise.
598 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
599 Alan Hayward <alan.hayward@arm.com>
600 David Sherwood <david.sherwood@arm.com>
602 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
603 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
604 (get_ref_base_and_extent_hwi): Declare.
605 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
606 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
607 (get_ref_base_and_extent_hwi): New function.
608 * cfgexpand.c (expand_debug_expr): Update call to
609 get_ref_base_and_extent.
610 * dwarf2out.c (add_var_loc_to_decl): Likewise.
611 * gimple-fold.c (get_base_constructor): Return the offset as a
612 poly_int64_pod rather than a HOST_WIDE_INT.
613 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
614 * ipa-polymorphic-call.c
615 (ipa_polymorphic_call_context::set_by_invariant)
616 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
617 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
618 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
619 rather than get_ref_base_and_extent.
620 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
621 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
622 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
624 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
625 call to get_ref_base_and_extent.
626 * tree-sra.c (create_access, get_access_for_expr): Likewise.
627 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
628 (stmt_kills_ref_p): Likewise.
629 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
630 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
632 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
633 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
634 when calling native_encode_expr.
635 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
636 call to get_ref_base_and_extent.
637 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
638 get_ref_base_and_extent.
639 * var-tracking.c (track_expr_p): Likewise.
641 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
642 Alan Hayward <alan.hayward@arm.com>
643 David Sherwood <david.sherwood@arm.com>
645 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
646 HOST_WIDE_INT to poly_int64_pod.
647 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
648 polynomail parameter offsets.
650 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
651 Alan Hayward <alan.hayward@arm.com>
652 David Sherwood <david.sherwood@arm.com>
654 * gengtype.c (main): Handle poly_int64_pod.
655 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
656 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
657 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
658 from HOST_WIDE_INT to poly_int64_pod.
659 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
660 (copy_cfa): New function.
661 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
662 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
663 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
664 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
665 (initial_return_save): Treat offsets as poly_ints.
666 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
667 offset is nonconstant.
668 (reg_save): Take the offset as a poly_int64. Fall back to
669 DW_CFA_expression for nonconstant offsets.
670 (queue_reg_save): Take the offset as a poly_int64.
671 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
673 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
674 Alan Hayward <alan.hayward@arm.com>
675 David Sherwood <david.sherwood@arm.com>
677 * rtl.h (operand_subword, operand_subword_force): Take the offset
678 as a poly_uint64 an unsigned int.
679 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
681 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
682 Alan Hayward <alan.hayward@arm.com>
683 David Sherwood <david.sherwood@arm.com>
685 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
686 'p' format code. Use INT_LIST rather than SUBREG as the example of
687 a code with an XINT and an XEXP. Remove the implication that
688 accessing an rtx field using XINT is expected to work.
689 * rtl.def (SUBREG): Change format from "ei" to "ep".
690 * rtl.h (rtunion::rt_subreg): New field.
691 (XCSUBREG): New macro.
692 (SUBREG_BYTE): Use it.
693 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
694 Update constructor accordingly.
695 (subreg_shape::operator ==): Update accordingly.
696 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
697 than an unsigned int.
698 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
699 a poly_uint64 rather than an unsigned int.
700 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
701 than an unsigned int.
702 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
703 (subreg_size_highpart_offset): Return a poly_uint64 rather than
704 an unsigned int. Take the sizes as poly_uint64s.
705 (subreg_offset_from_lsb): Return a poly_uint64 rather than
706 an unsigned int. Take the shift as a poly_uint64 rather than
708 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
709 as a poly_uint64 rather than an unsigned int.
710 (simplify_subreg_regno): Likewise.
711 (byte_lowpart_offset): Return the memory offset as a poly_int64
713 (subreg_memory_offset): Likewise. Take the subreg offset as a
714 poly_uint64 rather than an unsigned int.
715 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
716 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
717 poly_uint64 rather than an unsigned int.
718 * rtl.c (rtx_format): Describe 'p' in comment.
719 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
720 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
721 offset as a poly_uint64 rather than an unsigned int.
722 (byte_lowpart_offset): Return the memory offset as a poly_int64
724 (subreg_memory_offset): Likewise. Take the subreg offset as a
725 poly_uint64 rather than an unsigned int.
726 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
727 mode sizes as poly_uint64s rather than unsigned ints. Return a
728 poly_uint64 rather than an unsigned int.
729 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
730 (copy_insn_1): Handle 'p'.
731 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
732 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
733 an unsigned int. Return the shift in the same way.
734 (subreg_lsb): Return the shift as a poly_uint64 rather than an
736 (subreg_size_offset_from_lsb): Take the sizes and shift as
737 poly_uint64s rather than unsigned ints. Return the offset as
739 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
740 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
742 * rtlhash.c (add_rtx): Handle 'p'.
743 * genemit.c (gen_exp): Likewise.
744 * gengenrtl.c (type_from_format, gendef): Likewise.
745 * gensupport.c (subst_pattern_match, get_alternatives_number)
746 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
747 (subst_dup): Likewise.
748 * gengtype.c (adjust_field_rtx_def): Likewise.
749 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
750 (match_pattern_2): Likewise.
751 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
752 (rtx_test::subreg_field): New function.
753 (operator ==, safe_to_hoist_p, transition_parameter_type)
754 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
755 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
756 * genpeep.c (match_rtx): Likewise.
757 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
758 (rtx_writer::print_rtx_operand): Handle 'p'.
759 (print_value): Handle SUBREG.
760 * read-rtl.c (apply_int_iterator): Likewise.
761 (rtx_reader::read_rtx_operand): Handle 'p'.
762 * alias.c (rtx_equal_for_memref_p): Likewise.
763 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
764 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
766 * calls.c (expand_call): Likewise.
767 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
768 (make_extraction, gen_lowpart_for_combine): Likewise.
769 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
771 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
772 rather than an unsigned int. Treat subreg offsets as poly_ints.
773 (exp_equiv_p): Handle 'p'.
774 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
775 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
776 * dse.c (find_shift_sequence): Likewise.
777 * dwarf2out.c (rtl_for_decl_location): Likewise.
778 * expmed.c (extract_low_bits): Likewise.
779 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
780 (expand_expr_real_2): Likewise.
781 * final.c (alter_subreg): Likewise.
782 (leaf_renumber_regs_insn): Handle 'p'.
783 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
784 Treat subreg offsets as poly_ints.
785 * fwprop.c (forward_propagate_and_simplify): Likewise.
786 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
787 * ira.c (get_subreg_tracking_sizes): Likewise.
788 * ira-conflicts.c (go_through_subreg): Likewise.
789 * ira-lives.c (process_single_reg_class_operands): Likewise.
790 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
791 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
792 as a poly_uint64 rather than an unsigned int.
793 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
794 subreg offsets as poly_ints.
795 * lra-constraints.c (operands_match_p): Handle 'p'.
796 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
797 * lra-spills.c (assign_mem_slot): Likewise.
798 * postreload.c (move2add_valid_value_p): Likewise.
799 * recog.c (general_operand, indirect_operand): Likewise.
800 * regcprop.c (copy_value, maybe_mode_change): Likewise.
801 (copyprop_hardreg_forward_1): Likewise.
802 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
803 (record_subregs_of_mode): Likewise.
804 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
805 * reload.c (operands_match_p): Handle 'p'.
806 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
807 * reload1.c (alter_reg, choose_reload_regs): Likewise.
808 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
809 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
810 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
811 (simplify_const_poly_int_tests<N>::run): Likewise.
812 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
813 a poly_uint64 rather than an unsigned int.
814 * valtrack.c (debug_lowpart_subreg): Likewise.
815 * var-tracking.c (var_lowpart): Likewise.
816 (loc_cmp): Handle 'p'.
818 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
819 Alan Hayward <alan.hayward@arm.com>
820 David Sherwood <david.sherwood@arm.com>
822 * ira.c (get_subreg_tracking_sizes): New function.
823 (init_live_subregs): Take an integer size rather than a register.
824 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
825 to init_live_subregs.
827 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
828 Alan Hayward <alan.hayward@arm.com>
829 David Sherwood <david.sherwood@arm.com>
831 * expr.c (store_constructor_field): Change bitsize from a
832 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
833 HOST_WIDE_INT to a poly_int64.
834 (store_constructor): Change size from a HOST_WIDE_INT to
836 (store_field): Likewise bitsize and bitpos.
838 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
839 Alan Hayward <alan.hayward@arm.com>
840 David Sherwood <david.sherwood@arm.com>
842 * expmed.h (store_bit_field): Change bitregion_start and
843 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
844 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
845 (store_bit_field_1, store_integral_bit_field, store_bit_field)
846 (store_fixed_bit_field, store_split_bit_field): Likewise.
847 * expr.c (store_constructor_field, store_field): Likewise.
848 (optimize_bitfield_assignment_op): Likewise. Make the same change
849 to bitsize and bitpos.
850 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
851 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
852 same change in the constructor arguments.
853 (get_best_mode): Change bitregion_start and bitregion_end from
854 unsigned HOST_WIDE_INT to poly_uint64.
855 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
856 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
858 (bit_field_mode_iterator::next_mode): Update for new types
859 of m_bitregion_start and m_bitregion_end.
860 (get_best_mode): Change bitregion_start and bitregion_end from
861 unsigned HOST_WIDE_INT to poly_uint64.
863 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
864 Alan Hayward <alan.hayward@arm.com>
865 David Sherwood <david.sherwood@arm.com>
867 * rtl.h (simplify_gen_subreg): Add a temporary overload that
868 accepts poly_uint64 offsets.
869 * expmed.h (extract_bit_field): Take bitsize and bitnum as
870 poly_uint64s rather than unsigned HOST_WIDE_INTs.
871 * expmed.c (lowpart_bit_field_p): Likewise.
872 (extract_bit_field_as_subreg): New function, split out from...
873 (extract_bit_field_1): ...here. Take bitsize and bitnum as
874 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
875 extractions, check that BITSIZE matches the size of the extracted
876 value and that BITNUM is an exact multiple of that size.
877 If all else fails, try forcing the value into memory if
878 BITNUM is variable, and adjusting the address so that the
879 offset is constant. Split the part that can only handle constant
880 bitsize and bitnum out into...
881 (extract_integral_bit_field): ...this new function.
882 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
883 rather than unsigned HOST_WIDE_INTs.
885 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
886 Alan Hayward <alan.hayward@arm.com>
887 David Sherwood <david.sherwood@arm.com>
889 * expmed.h (store_bit_field): Take bitsize and bitnum as
890 poly_uint64s rather than unsigned HOST_WIDE_INTs.
891 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
892 that returns the byte size.
893 (store_bit_field_1): Take bitsize and bitnum as
894 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
895 to simple_mem_bitfield_p. Split the part that can only handle
896 constant bitsize and bitnum out into...
897 (store_integral_bit_field): ...this new function.
898 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
899 than unsigned HOST_WIDE_INTs.
900 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
902 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
903 Alan Hayward <alan.hayward@arm.com>
904 David Sherwood <david.sherwood@arm.com>
906 * lra-int.h (lra_reg): Change offset from int to poly_int64.
907 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
909 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
910 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
911 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
912 offset as a poly_int64 rather than an int.
913 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
914 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
915 * lra-constraints.c (equiv_address_substitution): Track offsets
917 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
918 (curr_insn_transform): Handle the new form of sp_offset.
919 * lra-eliminations.c (lra_elim_table): Change previous_offset
920 and offset from HOST_WIDE_INT to poly_int64.
921 (print_elim_table, update_reg_eliminate): Update accordingly.
922 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
923 (get_elimination): Update accordingly.
924 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
925 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
926 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
927 poly_int64 offsets generally.
928 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
929 (mark_not_eliminable, init_elimination): Update accordingly.
930 (remove_reg_equal_offset_note): Return a bool and pass the new
931 offset back by pointer as a poly_int64.
932 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
933 rather than a HOST_WIDE_INT.
934 (do_remat): Track offsets poly_int64s.
935 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
937 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
938 Alan Hayward <alan.hayward@arm.com>
939 David Sherwood <david.sherwood@arm.com>
941 * rtl.h (mem_attrs): Add a default constructor. Change size and
942 offset from HOST_WIDE_INT to poly_int64.
943 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
944 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
945 (widen_memory_access): Take the sizes and offsets as poly_int64s
946 rather than HOST_WIDE_INTs.
947 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
948 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
950 (adjust_offset_for_component_ref): Change the offset from a
951 HOST_WIDE_INT to a poly_int64.
952 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
953 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
954 * dce.c (find_call_stack_args): Likewise.
955 * dse.c (record_store): Likewise.
956 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
957 * print-rtl.c (rtx_writer::print_rtx): Likewise.
958 * read-rtl-function.c (test_loading_mem): Likewise.
959 * rtlanal.c (may_trap_p_1): Likewise.
960 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
961 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
962 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
963 (mem_attrs::mem_attrs): New function.
964 (set_mem_attributes_minus_bitpos): Change bitpos from a
965 HOST_WIDE_INT to poly_int64.
966 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
967 (clear_mem_offset, clear_mem_size, change_address)
968 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
969 initialize mem_attrs.
970 (set_mem_offset, set_mem_size, adjust_address_1)
971 (adjust_automodify_address_1, offset_address, widen_memory_access):
972 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
974 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
975 Alan Hayward <alan.hayward@arm.com>
976 David Sherwood <david.sherwood@arm.com>
978 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
979 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
980 rather than 0 to represent an unknown size. Assert that the size
981 is known when the mode isn't BLKmode.
982 (may_trap_p_1): Use -1 for unknown sizes.
983 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
985 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
986 Alan Hayward <alan.hayward@arm.com>
987 David Sherwood <david.sherwood@arm.com>
989 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
990 to poly_int64. Update commentary for positions_needed.large.
991 (read_info_type): Change offset and width from HOST_WIDE_INT
993 (set_usage_bits): Likewise.
994 (canon_address): Return the offset as a poly_int64 rather than
995 a HOST_WIDE_INT. Use strip_offset_and_add.
996 (set_all_positions_unneeded, any_positions_needed_p): Use
997 positions_needed.large to track stores with non-constant widths.
998 (all_positions_needed_p): Likewise. Take the offset and width
999 as poly_int64s rather than ints. Assert that rhs is nonnull.
1000 (record_store): Cope with non-constant offsets and widths.
1001 Nullify the rhs of an earlier store if we can't tell which bytes
1003 (find_shift_sequence): Take the access_size and shift as poly_int64s
1005 (get_stored_val): Take the read_offset and read_width as poly_int64s
1006 rather than HOST_WIDE_INTs.
1007 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1008 non-constant offsets and widths.
1010 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1011 Alan Hayward <alan.hayward@arm.com>
1012 David Sherwood <david.sherwood@arm.com>
1014 * inchash.h (inchash::hash::add_poly_int): New function.
1015 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1016 Use poly_int64 rather than HOST_WIDE_INT.
1017 (ao_ref::max_size_known_p): New function.
1018 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1019 rather than HOST_WIDE_INT.
1020 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1021 to temporaries until its interface is adjusted to match.
1022 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1023 (aliasing_component_refs_p, decl_refs_may_alias_p)
1024 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1025 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1026 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1028 * alias.c (ao_ref_from_mem): Likewise.
1029 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1030 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1031 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1032 (maybe_trim_complex_store, maybe_trim_constructor_store)
1033 (live_bytes_read, dse_classify_store): Likewise.
1034 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1035 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1036 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1037 (vn_reference_lookup_3): Likewise.
1038 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1040 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1041 Alan Hayward <alan.hayward@arm.com>
1042 David Sherwood <david.sherwood@arm.com>
1044 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1045 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1046 instead of ranges_overlap_p.
1048 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1049 Alan Hayward <alan.hayward@arm.com>
1050 David Sherwood <david.sherwood@arm.com>
1052 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1053 sizes as poly_int64s rather than HOST_WIDE_INTs.
1055 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1056 Alan Hayward <alan.hayward@arm.com>
1057 David Sherwood <david.sherwood@arm.com>
1059 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1060 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1061 * gimple-fold.c (fold_ctor_reference): Likewise.
1063 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1064 Alan Hayward <alan.hayward@arm.com>
1065 David Sherwood <david.sherwood@arm.com>
1067 * target.def (dwarf_poly_indeterminate_value): New hook.
1068 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1069 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1070 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1071 * doc/tm.texi: Regenerate.
1072 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1073 offset as a poly_int64.
1074 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1075 offset as a poly_int64.
1076 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1077 Take the offset as a poly_int64.
1078 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1079 (frame_pointer_fb_offset): Change to a poly_int64.
1080 (int_loc_descriptor): Take the offset as a poly_int64. Use
1081 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1082 (based_loc_descr): Take the offset as a poly_int64.
1083 Use strip_offset_and_add to handle (plus X (const)).
1084 Use new_reg_loc_descr instead of an open-coded version of the
1085 previous implementation.
1086 (mem_loc_descriptor): Handle CONST_POLY_INT.
1087 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1088 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1090 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1091 Alan Hayward <alan.hayward@arm.com>
1092 David Sherwood <david.sherwood@arm.com>
1094 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1095 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1096 * inchash.h (inchash::hash::add_poly_hwi): New function.
1097 * gengtype.c (main): Register poly_int64.
1098 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1099 offset as a poly_int.
1100 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1101 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1102 offset as a poly_int64.
1103 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1104 * print-rtl.c (print_poly_int): New function.
1105 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1107 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1109 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1110 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1111 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1112 Enforce track_offset_p for parts of a PARALLEL.
1113 (vt_add_function_parameter): Use const_offset for the final
1114 offset to track. Use get_tracked_reg_offset for the parts
1117 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1118 Alan Hayward <alan.hayward@arm.com>
1119 David Sherwood <david.sherwood@arm.com>
1121 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1122 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1123 * doc/tm.texi: Regenerate.
1124 * hooks.h (hook_bool_uint_uint_true): Delete.
1125 (hook_bool_puint64_puint64_true): Declare.
1126 * hooks.c (hook_bool_uint_uint_true): Delete.
1127 (hook_bool_puint64_puint64_true): New function.
1128 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1129 instead of unsigned ints.
1130 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1131 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1133 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1134 Alan Hayward <alan.hayward@arm.com>
1135 David Sherwood <david.sherwood@arm.com>
1137 * optabs.h (expand_operand): Add an int_value field.
1138 (create_expand_operand): Add an int_value parameter and use it
1139 to initialize the new expand_operand field.
1140 (create_integer_operand): Replace with a declaration of a function
1141 that accepts poly_int64s. Move the implementation to...
1142 * optabs.c (create_integer_operand): ...here.
1143 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1144 the mode preserves the value of int_value, instead of calling
1145 const_int_operand on the rtx. Use gen_int_mode to generate
1148 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1149 Alan Hayward <alan.hayward@arm.com>
1150 David Sherwood <david.sherwood@arm.com>
1152 * dumpfile.h (dump_dec): Declare.
1153 * dumpfile.c (dump_dec): New function.
1154 * pretty-print.h (pp_wide_integer): Turn into a function and
1155 declare a poly_int version.
1156 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1158 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1159 Alan Hayward <alan.hayward@arm.com>
1160 David Sherwood <david.sherwood@arm.com>
1162 * doc/generic.texi (POLY_INT_CST): Document.
1163 * tree.def (POLY_INT_CST): New tree code.
1164 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1165 * tree-core.h (tree_poly_int_cst): New struct.
1166 (tree_node): Add a poly_int_cst field.
1167 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1168 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1169 instead of a wide_int_ref.
1170 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1172 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1173 instead of an unsigned HOST_WIDE_INT.
1174 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1175 (ptrdiff_tree_p): Declare.
1176 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1177 extern inline implementations if the target doesn't use POLY_INT_CST.
1178 (poly_int_tree_p): New function.
1179 (wi::unextended_tree): New class.
1180 (wi::int_traits <unextended_tree>): New override.
1181 (wi::extended_tree): Add a default constructor.
1182 (wi::extended_tree::get_tree): New function.
1183 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1184 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1185 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1186 (wi::tree_to_poly_wide_ref): New typedefs.
1187 (wi::ints_for): Provide overloads for extended_tree and
1189 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1190 (wi::to_wide): New functions.
1191 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1192 * tree.c (poly_int_cst_hasher): New struct.
1193 (poly_int_cst_hash_table): New variable.
1194 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1195 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1197 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1198 (init_ttree): Initialize poly_int_cst_hash_table.
1199 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1200 a poly_int64 instead of a HOST_WIDE_INT.
1201 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1202 instead of an unsigned HOST_WIDE_INT.
1203 (wide_int_to_tree): Rename to...
1204 (wide_int_to_tree_1): ...this.
1205 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1206 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1207 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1208 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1209 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1210 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1212 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1213 (streamer_read_tree_body): Likewise.
1214 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1215 (streamer_write_tree_body): Likewise.
1216 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1217 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1218 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1219 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1220 * gimple-expr.h (is_gimple_constant): Likewise.
1221 * gimplify.c (maybe_with_size_expr): Likewise.
1222 * print-tree.c (print_node): Likewise.
1223 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1224 * tree-pretty-print.c (dump_generic_node): Likewise.
1225 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1226 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1227 * tree-vrp.c (compare_values_warnv): Likewise.
1228 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1229 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1230 (force_expr_to_var_cost): Likewise.
1231 * tree-ssa-loop.c (for_each_index): Likewise.
1232 * fold-const.h (build_invariant_address, size_int_kind): Take a
1233 poly_int64 instead of a HOST_WIDE_INT.
1234 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1235 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1237 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1238 (int_const_binop_2): New function, split out from...
1239 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1240 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1241 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1242 * gimple-ssa-strength-reduction.c (slsr_process_add)
1243 (slsr_process_mul): Check for INTEGER_CSTs before using them
1245 * stor-layout.c (bits_from_bytes): New function.
1246 (bit_from_pos): Use it.
1247 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1248 by BITS_PER_UNIT to get the TYPE_SIZE.
1249 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1250 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1252 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1253 Alan Hayward <alan.hayward@arm.com>
1254 David Sherwood <david.sherwood@arm.com>
1256 * doc/rtl.texi (const_poly_int): Document. Also document the
1257 rtl sharing behavior.
1258 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1259 * rtl.h (const_poly_int_def): New struct.
1260 (rtx_def::u): Add a cpi field.
1261 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1262 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1263 (wi::rtx_to_poly_wide_ref): New typedef
1264 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1265 (poly_int_rtx_p): New functions.
1266 (trunc_int_for_mode): Declare a poly_int64 version.
1267 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1268 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1270 (strip_offset): Declare.
1271 (strip_offset_and_add): New function.
1272 * rtl.def (CONST_POLY_INT): New rtx code.
1273 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1274 (shared_const_p): Use poly_int_rtx_p.
1275 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1277 (gen_int_shift_amount): Likewise.
1278 * emit-rtl.c (const_poly_int_hasher): New class.
1279 (const_poly_int_htab): New variable.
1280 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1281 (const_poly_int_hasher::hash): New function.
1282 (const_poly_int_hasher::equal): Likewise.
1283 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1284 (immed_wide_int_const): Rename to...
1285 (immed_wide_int_const_1): ...this and make static.
1286 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1287 instead of a wide_int_ref.
1288 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1289 (gen_lowpart_common): Handle CONST_POLY_INT.
1290 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1291 * cselib.c (cselib_hash_rtx): Likewise.
1292 * dwarf2out.c (const_ok_for_output_1): Likewise.
1293 * expr.c (convert_modes): Likewise.
1294 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1295 * rtlhash.c (add_rtx): Likewise.
1296 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1297 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1298 Handle existing CONST_POLY_INT rtxes.
1299 * expmed.h (expand_shift): Take a poly_int64 instead of a
1301 * expmed.c (expand_shift): Likewise.
1302 * rtlanal.c (strip_offset): New function.
1303 (commutative_operand_precedence): Give CONST_POLY_INT the same
1304 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1306 * rtl-tests.c (const_poly_int_tests): New struct.
1307 (rtl_tests_c_tests): Use it.
1308 * simplify-rtx.c (simplify_const_unary_operation): Handle
1310 (simplify_const_binary_operation): Likewise.
1311 (simplify_binary_operation_1): Fold additions of symbolic constants
1312 and CONST_POLY_INTs.
1313 (simplify_subreg): Handle extensions and truncations of
1315 (simplify_const_poly_int_tests): New struct.
1316 (simplify_rtx_c_tests): Use it.
1317 * wide-int.h (storage_ref): Add default constructor.
1318 (wide_int_ref_storage): Likewise.
1319 (trailing_wide_ints): Use GTY((user)).
1320 (trailing_wide_ints::operator[]): Add a const version.
1321 (trailing_wide_ints::get_precision): New function.
1322 (trailing_wide_ints::extra_size): Likewise.
1324 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1325 Alan Hayward <alan.hayward@arm.com>
1326 David Sherwood <david.sherwood@arm.com>
1328 * emit-rtl.h (gen_int_shift_amount): Declare.
1329 * emit-rtl.c (gen_int_shift_amount): New function.
1330 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1332 * calls.c (shift_return_value): Likewise.
1333 * cse.c (fold_rtx): Likewise.
1334 * dse.c (find_shift_sequence): Likewise.
1335 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1336 (expand_shift, expand_smod_pow2): Likewise.
1337 * lower-subreg.c (shift_cost): Likewise.
1338 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1339 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1340 (expand_vec_perm_var): Likewise.
1341 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1342 (simplify_binary_operation_1): Likewise.
1343 * combine.c (try_combine, find_split_point, force_int_to_mode)
1344 (simplify_shift_const_1, simplify_shift_const): Likewise.
1345 (change_zero_ext): Likewise. Use simplify_gen_binary.
1347 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1349 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1351 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1353 * doc/invoke.texi (ARM Options): Document accepted extension options
1354 for -march=armv8.3-a.
1356 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1359 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1360 or --with-float=softfp, set the default CPU to arm10e.
1362 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1364 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1365 * config/visium/predicates.md (const_shift_operand): Likewise.
1366 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1367 (visium_legitimize_reload_address): Likewise.
1369 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1372 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1374 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1377 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1379 2017-12-20 Julia Koval <julia.koval@intel.com>
1381 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1382 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1383 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1384 * config.gcc: Include vpclmulqdqintrin.h.
1385 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1386 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1388 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1389 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1390 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1391 * config/i386/i386.c (isa2_opts): Add -mcx16.
1392 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1393 (ix86_option_override_internal): Move mcx16 to flags2.
1394 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1395 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1396 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1397 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1398 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1399 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1400 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1401 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1402 * doc/invoke.texi: Add -mvpclmulqdq.
1404 2017-12-20 Tom de Vries <tom@codesourcery.com>
1407 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1408 * calls.c (rtx_for_static_chain): ... here. New function.
1409 * calls.h (rtx_for_static_chain): Declare.
1410 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1411 instead of targetm.calls.static_chain.
1412 * df-scan.c (df_get_entry_block_def_set): Same.
1414 2017-12-19 Tom de Vries <tom@codesourcery.com>
1416 PR tree-optimization/83493
1417 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1420 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1422 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1423 inform with hardcoded english plural handling.
1425 2017-12-18 Jeff Law <law@redhat.com>
1427 PR tree-optimization/83477
1428 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1429 a non-virtual PHI, always push a new range.
1431 2017-12-19 Martin Sebor <msebor@redhat.com>
1434 * builtins.c (compute_objsize): Handle non-constant offsets.
1436 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1438 PR tree-optimization/83444
1439 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1440 character load case, if get_stridx on MEM_REF's operand doesn't
1441 look usable, retry with get_addr_stridx.
1443 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1446 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1447 (variable_tracking_main_1): Keep markers even when VTA fails.
1450 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1451 even if there are markers before it.
1452 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1454 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1456 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1457 typical order conditions.
1458 * sel-sched.c (extract_new_fences_from): Likewise.
1459 * config/visium/constraints.md (J, K, L): Likewise.
1460 * config/visium/predicates.md (const_shift_operand): Likewise.
1461 * config/visium/visium.c (visium_legitimize_address,
1462 visium_legitimize_reload_address): Likewise.
1463 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1464 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1465 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1466 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1467 SET_DUMP_DETAIL): Likewise.
1468 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1469 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1470 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1471 avr_set_core_architecture, avr_set_current_function,
1472 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1473 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1474 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1475 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1476 avr_map_decompose, avr_fold_builtin): Likewise.
1477 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1478 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1479 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1480 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1481 * config/m32c/m32c.c (m32c_conditional_register_usage,
1482 m32c_address_cost): Likewise.
1483 * config/m32c/predicates.md (shiftcount_operand,
1484 longshiftcount_operand): Likewise.
1485 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1486 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1487 can_use_cdx_ldstw): Likewise.
1488 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1489 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1491 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1492 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1493 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1494 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1495 * config/vax/vax.c (adjacent_operands_p): Likewise.
1496 * config/ft32/constraints.md (L, b, KA): Likewise.
1497 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1499 * cfgexpand.c (expand_stack_alignment): Likewise.
1500 * gcse.c (insert_expr_in_table): Likewise.
1501 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1502 * cgraphunit.c (cgraph_node::expand): Likewise.
1503 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1504 * emit-rtl.c (add_insn): Likewise.
1505 * input.c (dump_location_info): Likewise.
1506 * passes.c (NEXT_PASS): Likewise.
1507 * read-rtl-function.c (parse_note_insn_name,
1508 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1510 * sched-rgn.c (sched_rgn_init): Likewise.
1511 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1512 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1513 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1514 * explow.c (eliminate_constant_term): Likewise.
1515 * final.c (leaf_renumber_regs_insn): Likewise.
1516 * cfgrtl.c (print_rtl_with_bb): Likewise.
1517 * genhooks.c (emit_init_macros): Likewise.
1518 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1519 * tree-data-ref.c (conflict_fn): Likewise.
1520 * selftest.c (assert_streq): Likewise.
1521 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1522 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1523 fold_binary_loc, multiple_of_p): Likewise.
1524 * reload.c (push_reload, find_equiv_reg): Likewise.
1525 * et-forest.c (et_nca, et_below): Likewise.
1526 * dbxout.c (dbxout_symbol_location): Likewise.
1527 * reorg.c (relax_delay_slots): Likewise.
1528 * dojump.c (do_compare_rtx_and_jump): Likewise.
1529 * gengtype-parse.c (type): Likewise.
1530 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1531 simplify_const_relational_operation): Likewise.
1532 * reload1.c (do_output_reload): Likewise.
1533 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1534 * gengtype.c (type_for_name): Likewise.
1535 * gimple-ssa-sprintf.c (format_directive): Likewise.
1537 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1540 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1541 accessing it. Adjust comment.
1543 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1546 * predict.c (zero_one_minusone): New function.
1547 (apply_return_prediction): Avoid return prediction for functions
1548 returning only -1, 0 and 1 values, unless they only return -1 and 0
1551 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1553 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1554 fall-through warning.
1555 (arc_compute_frame_size): Remove unused variables.
1556 (arc_print_operand): Fix fprintif format.
1557 (arc_can_follow_jump): Clean fall-through warning.
1559 2017-12-19 Marek Polacek <polacek@redhat.com>
1562 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1565 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1567 * config/arc/arc.c (overriderregs): New variable.
1568 (arc_override_options): Track fixed/call saved/call options.
1569 (arc_conditional_register_usage): Check against overrideregs
1570 variable whenever we change register properties.
1572 2017-12-19 Nathan Sidwell <nathan@acm.org>
1574 * opts.c (finish_options): Don't prefix dump_base_name if it
1575 already contains directories.
1577 2017-12-19 Martin Liska <mliska@suse.cz>
1579 PR rtl-optimization/82675
1580 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1581 more element in sbitmap.
1583 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1585 * gimplify.c (gimplify_expr): Use error_operand_p.
1587 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1590 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1593 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1595 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1596 the jump is to a label.
1598 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1600 PR tree-optimization/83444
1601 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1606 * ipa-inline.c (flatten_remove_node_hook): New function.
1607 (ipa_inline): Keep only nodes with flatten attribute at the end of
1608 the array in the order from ipa_reverse_postorder, only walk that
1609 portion of array for flattening, if there is more than one such
1610 node, temporarily register a removal hook and ignore removed nodes.
1612 PR tree-optimization/80631
1613 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1614 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1615 IFN_REDUC_MAX or IFN_REDUC_MIN.
1617 2017-12-18 Jeff Law <law@redhat.com>
1619 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1620 record symbolic equivalences from backedges in the CFG.
1623 2017-11-19 Jeff Law <law@redhat.com>
1625 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1626 of degenerates resulting from ignoring an edge.
1628 2017-12-18 Martin Sebor <msebor@redhat.com>
1631 PR tree-optimization/78450
1632 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1633 (handle_builtin_strlen): Call it.
1635 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1637 PR rtl-optimization/83424
1638 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1640 2017-12-18 Marek Polacek <polacek@redhat.com>
1643 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1644 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1647 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1649 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1650 (umaddsidi4, umaddsidi_split): Likewise.
1652 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1654 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1657 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1660 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1662 2017-12-18 Richard Biener <rguenther@suse.de>
1664 PR tree-optimization/81877
1665 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1666 (outermost_indep_loop): Adjust.
1667 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1668 (can_sm_ref_p): Adjust.
1670 2017-12-18 Richard Biener <rguenther@suse.de>
1673 * tree.c (array_at_struct_end_p): Return true if the underlying
1674 object has space for at least one element in excess of what
1675 the array domain specifies.
1677 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1679 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1680 example to avoid overfull hbox.
1681 * doc/invoke.texi (Option Summary): Add missing @gol.
1682 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1685 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1686 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1688 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1689 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1690 (AArch64 Options): Document -mverbose-cost-dump.
1691 (ARM Options): Likewise, plus -mflip-thumb.
1693 2017-12-17 Martin Sebor <msebor@redhat.com>
1696 * gimple-ssa-warn-restrict.c
1697 (builtin_memref::offset_out_of_bounds): Correct the handling of
1700 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1702 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1704 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1706 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1708 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1710 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1713 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1715 * doc/invoke.texi: Fix some typos.
1717 2017-12-16 Martin Sebor <msebor@redhat.com>
1719 PR tree-optimization/78918
1720 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1721 * builtins.c (check_sizes): Rename...
1722 (check_access): ...to this. Rename function arguments for clarity.
1723 (check_memop_sizes): Adjust names.
1724 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1725 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1726 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1727 (check_strncat_sizes, expand_builtin_strncat): Same.
1728 (expand_builtin_strncpy, expand_builtin_memset): Same.
1729 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1730 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1731 (maybe_emit_sprintf_chk_warning): Same.
1732 (expand_builtin_strcpy): Adjust.
1733 (expand_builtin_stpcpy): Same.
1734 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1735 in pointer-checking forms of memcpy, memmove, and mempcpy.
1736 (gcall_to_tree_minimal, max_object_size): Define new functions.
1737 * builtins.h (max_object_size): Declare.
1738 * calls.c (alloc_max_size): Call max_object_size instead of
1739 hardcoding ssizetype limit.
1740 (get_size_range): Handle new argument.
1741 * calls.h (get_size_range): Add a new argument.
1742 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1743 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1744 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1746 (gimple_fold_builtin_memory_chk): Same.
1747 (gimple_fold_builtin_stxcpy_chk): New function.
1748 * gimple-ssa-warn-restrict.c: New source.
1749 * gimple-ssa-warn-restrict.h: New header.
1750 * gimple.c (gimple_build_call_from_tree): Propagate location.
1751 * passes.def (pass_warn_restrict): Add new pass.
1752 * tree-pass.h (make_pass_warn_restrict): Declare.
1753 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1755 (handle_builtin_strcat): Same.
1756 (strlen_optimize_stmt): Rename...
1757 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1758 stpncpy, strncpy, and their checking forms.
1760 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1762 PR rtl-optimization/82849
1763 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1764 and get_max_loop_iterations_int.
1766 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1767 Alan Hayward <alan.hayward@arm.com>
1768 David Sherwood <david.sherwood@arm.com>
1770 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1771 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1772 size as a poly_uint64.
1773 (mode_for_vector, mode_for_int_vector): Take the number of vector
1774 elements as a poly_uint64.
1775 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1776 size as a poly_uint64.
1777 (mode_for_vector, mode_for_int_vector): Take the number of vector
1778 elements as a poly_uint64.
1780 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1781 Alan Hayward <alan.hayward@arm.com>
1782 David Sherwood <david.sherwood@arm.com>
1784 * machmode.h (MACRO_MODE): New macro.
1785 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1786 * targhooks.c (default_libcall_value, default_secondary_reload)
1787 (default_memory_move_cost, default_register_move_cost)
1788 (default_class_max_nregs): Likewise.
1790 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1791 Alan Hayward <alan.hayward@arm.com>
1792 David Sherwood <david.sherwood@arm.com>
1794 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1795 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1797 * genautomata.c (main): Likewise.
1798 * genconditions.c (write_header): Likewise.
1799 * genemit.c (main): Likewise.
1800 * genextract.c (print_header): Likewise.
1801 * genopinit.c (main): Likewise.
1802 * genoutput.c (output_prologue): Likewise.
1803 * genpeep.c (main): Likewise.
1804 * genpreds.c (write_insn_preds_c): Likewise.
1805 * genrecog.c (writer_header): Likewise.
1806 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1807 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1808 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1809 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1810 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1811 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1812 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1813 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1814 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1815 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1816 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1817 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1818 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1819 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1820 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1821 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1822 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1823 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1824 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1825 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1826 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1827 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1828 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1829 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1830 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1831 * config/darwin.c (IN_TARGET_CODE): Likewise.
1832 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1833 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1834 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1835 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1836 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1837 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1838 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1839 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1840 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1841 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1842 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1843 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1844 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1845 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1846 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1847 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1848 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1849 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1850 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1851 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1852 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1853 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1854 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1855 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1856 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1857 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1858 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1859 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1860 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1861 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1862 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1863 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1864 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1865 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1866 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1867 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1868 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1869 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1870 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1871 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1872 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1873 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1874 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1875 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1876 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1877 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1878 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1879 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1880 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1881 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1882 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1883 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1884 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1885 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1886 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1887 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1888 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1889 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1890 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1891 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1892 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1893 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1894 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1895 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1896 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1897 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1898 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1899 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1900 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1901 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1902 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1903 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1904 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1905 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1906 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1907 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1908 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1909 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1910 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1911 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1912 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1913 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1914 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1915 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1916 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1917 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1918 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1919 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1920 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1921 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1922 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1923 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1924 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1925 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1926 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1927 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1928 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1929 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1930 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1931 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1932 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1933 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1934 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1935 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1936 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1937 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1939 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1940 Alan Hayward <alan.hayward@arm.com>
1941 David Sherwood <david.sherwood@arm.com>
1943 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1946 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1947 Alan Hayward <alan.hayward@arm.com>
1948 David Sherwood <david.sherwood@arm.com>
1950 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1951 * doc/md.texi (vec_series@var{m}): Document.
1952 * tree.def (VEC_SERIES_EXPR): New tree code.
1953 * tree.h (build_vec_series): Declare.
1954 * tree.c (build_vec_series): New function.
1955 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1956 * tree-pretty-print.c (dump_generic_node): Likewise.
1957 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1958 * tree-inline.c (estimate_operator_cost): Likewise.
1959 * expr.c (expand_expr_real_2): Likewise.
1960 * optabs-tree.c (optab_for_tree_code): Likewise.
1961 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1962 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1963 * expmed.c (make_tree): Handle VEC_SERIES.
1964 * optabs.def (vec_series_optab): New optab.
1965 * optabs.h (expand_vec_series_expr): Declare.
1966 * optabs.c (expand_vec_series_expr): New function.
1967 * tree-vect-generic.c (expand_vector_operations_1): Check that
1968 the operands also have vector type.
1970 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1971 Alan Hayward <alan.hawyard@arm.com>
1972 David Sherwood <david.sherwood@arm.com>
1974 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1975 (VEC_COND_EXPR): Add missing @tindex.
1976 * doc/md.texi (vec_duplicate@var{m}): Document.
1977 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1978 * tree.c (build_vector_from_val): Add stubbed-out handling of
1979 variable-length vectors, using VEC_DUPLICATE_EXPR.
1980 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1981 * cfgexpand.c (expand_debug_expr): Likewise.
1982 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1983 * tree-inline.c (estimate_operator_cost): Likewise.
1984 * tree-pretty-print.c (dump_generic_node): Likewise.
1985 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1986 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1987 (test_vec_duplicate_folding): New function.
1988 (fold_const_c_tests): Call it.
1989 * optabs.def (vec_duplicate_optab): New optab.
1990 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1991 * optabs.h (expand_vector_broadcast): Declare.
1992 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1993 vec_duplicate_optab.
1994 * expr.c (store_constructor): Try using vec_duplicate_optab for
1996 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1998 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2001 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2002 div/mod latencies a bit.
2004 2017-12-15 Jeff Law <law@redhat.com>
2006 PR tree-optimization/36550
2007 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2008 (mark_threaded_blocks): Rewrite code to avoid block copying when
2009 optimizing for size. Don't pessimize blocks which will be
2010 copied, but all the statements will be dead.
2012 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2014 PR tree-optimization/81165
2015 * tree-ssa-threadupdate.c (uses_in_bb): New.
2016 (estimate_threading_killed_stmts): New.
2017 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2018 * tree-ssa-threadedge.c
2019 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2022 2017-12-15 Jeff Law <law@redhat.com>
2024 PR tree-optimization/83410
2025 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2026 threads when parallelizing loops.
2028 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2030 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2032 * config/alpha/alpha.c (vms_attribute_table): Swap
2033 affects_type_identity and handler fields, adjust comments.
2034 * config/mips/mips.c (mips_attribute_table): Likewise.
2035 * config/visium/visium.c (visium_attribute_table): Likewise.
2036 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2037 * config/microblaze/microblaze.c (microblaze_attribute_table):
2039 * config/spu/spu.c (spu_attribute_table): Likewise.
2040 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2041 * config/arc/arc.c (arc_attribute_table): Likewise.
2042 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2043 * config/v850/v850.c (v850_handle_interrupt_attribute,
2044 v850_handle_data_area_attribute): Formatting fixes.
2045 (v850_attribute_table): Swap affects_type_identity and handler
2046 fields, adjust comments.
2047 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2048 * config/arm/arm.c (arm_attribute_table): Likewise.
2049 * config/avr/avr.c (avr_attribute_table): Likewise.
2050 * config/s390/s390.c (s390_attribute_table): Likewise.
2051 * config/sh/sh.c (sh_attribute_table): Likewise.
2052 * config/i386/i386.c (ix86_handle_cconv_attribute,
2053 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2054 (ix86_attribute_table): Swap affects_type_identity and handler
2055 fields, adjust comments.
2056 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2057 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2058 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2059 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2060 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2061 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2062 * config/rx/rx.c (rx_attribute_table): Likewise.
2063 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2064 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2065 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2066 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2067 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2068 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2069 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2070 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2071 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2072 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2073 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2075 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2077 (register_scoped_attributes, decl_attributes): Formatting fixes.
2079 PR tree-optimization/83269
2080 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2081 subtraction in arg0's type if type is signed and arg0 is unsigned.
2085 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2086 convert? on @0 instead of convert. Check type of @1, not @0.
2087 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2088 convert? on @0 instead of convert. Check type of @1, not @0.
2089 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2090 only optimize if either both @1 and @2 types are narrower
2091 precision, or both are wider or equal precision, and in the former
2092 case only if both have undefined overflow.
2094 2017-12-15 Richard Biener <rguenther@suse.de>
2097 * internal-fn.def (IFN_NOP): Add.
2098 * internal-fn.c (expand_NOP): Do nothing.
2099 * lto-streamer-in.c (input_function): Instead of removing
2100 sanitizer calls replace them with IFN_NOP calls.
2102 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2103 Alan Hayward <alan.hayward@arm.com>
2104 David Sherwood <david.sherwood@arm.com>
2106 * dse.c (store_info, read_info_type): Replace begin and end with
2108 (print_range): New function.
2109 (set_all_positions_unneeded, any_positions_needed_p)
2110 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2112 (record_store): Likewise. Optimize the case in which all positions
2114 (get_stored_val): Replace read_begin and read_end with read_offset
2116 (replace_read): Update call accordingly.
2118 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2120 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2121 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2122 (loop_cand::loop_cand): Initialize above members.
2123 (loop_cand::supported_operations): Delete.
2124 (loop_cand::can_interchange_p): Inline above function.
2125 (loop_cand::classify_simple_reduction): Record number of constant
2126 initialized simple reductions.
2127 (should_interchange_loops): New parameters. Check stmt cost of loops
2129 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2130 Update call to should_interchange_loops.
2131 (should_interchange_loop_nest): Update call to
2132 should_interchange_loops.
2134 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2137 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2138 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2139 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2141 2017-12-15 Julia Koval <julia.koval@intel.com>
2143 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2144 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2145 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2146 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2147 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2149 2017-12-15 Julia Koval <julia.koval@intel.com>
2151 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2152 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2153 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2154 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2155 _mm_aesenc_epi128): New intrinsics.
2157 2017-12-15 Julia Koval <julia.koval@intel.com>
2159 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2160 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2161 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2162 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2163 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2165 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2167 * gimple-ssa-strength-reduction.c (analyze_increments):
2168 Distinguish replacement costs for constant strides from those for
2171 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2173 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2175 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2177 * doc/invoke.texi: Document -Wcast-function-type.
2178 * recog.h (stored_funcptr): Change signature.
2179 * tree-dump.c (dump_node): Avoid warning.
2180 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2182 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2185 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2188 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2190 PR tree-optimization/83312
2191 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2192 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2193 "val" param, and to cope with arbitrary basic blocks.
2194 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2195 handle NULL_TREE for "val", dropping "bb" param.
2196 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2197 drop "bb" param. Handle NULL_TREE for "val".
2198 (find_case_label_for_value): Make "switch_stmt" param const.
2199 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2201 (vrp_prop::check_all_array_refs): Reimplement as...
2202 (check_array_bounds_dom_walker::before_dom_children): ...this new
2203 vfunc. Replace linear search through BB block list, excluding
2204 those with non-executable in-edges via dominator walk.
2206 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2208 * config/arm/arm.opt (mverbose-cost-dump): New option.
2209 * config/arm/arm.c (arm_rtx_costs): Use it.
2211 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2214 * reload1.c (emit_input_reload_insns): Skip debug markers.
2216 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2218 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2222 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2223 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2224 * config/sh/sh.c: Skip debug insns besides notes.
2225 * config/sh/sh.md: Likewise.
2226 * config/sh/sh_treg_combine.cc: Likewise.
2227 * config/sh/sync.md: Likewise.
2229 2017-12-14 Tom de Vries <tom@codesourcery.com>
2231 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2232 Add item for weak_undefined.
2234 2017-12-14 Richard Biener <rguenther@suse.de>
2236 PR tree-optimization/67842
2237 * sese.h (bb_in_region): Remove #if 0'ed code.
2239 2017-12-14 Richard Biener <rguenther@suse.de>
2241 PR tree-optimization/83326
2242 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2243 may_be_zero parameter and handle it by not marking the first
2244 peeled copy as not exiting the loop.
2245 (try_peel_loop): Likewise.
2246 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2247 to handle the case of constant or zero iterations and perform
2248 loop header copying on-the-fly.
2250 2017-12-14 Richard Biener <rguenther@suse.de>
2252 PR tree-optimization/83418
2253 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2254 Instead of asserting we don't get unfolded comparisons deal with
2257 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2260 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2261 basic blocks. Assert debug bind insns don't appear outside of bbs,
2262 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2265 PR tree-optimization/83198
2266 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2267 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2268 value if arg is a REAL_CST with incompatible type.
2270 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2271 Bin Cheng <bin.cheng@arm.com>
2274 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2276 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2279 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2281 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2282 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2284 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2285 Alan Hayward <alan.hayward@arm.com>
2286 David Sherwood <david.sherwood@arm.com>
2288 * poly-int.h: New file.
2289 * poly-int-types.h: Likewise.
2290 * coretypes.h: Include them.
2291 (POLY_INT_CONVERSION): Define.
2292 * target.def (estimated_poly_value): New hook.
2293 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2294 * doc/tm.texi: Regenerate.
2295 * doc/poly-int.texi: New file.
2296 * doc/gccint.texi: Include it.
2297 * doc/rtl.texi: Describe restrictions on subreg modes.
2298 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2299 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2300 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2301 * targhooks.h (default_estimated_poly_value): Declare.
2302 * targhooks.c (default_estimated_poly_value): New function.
2303 * target.h (estimated_poly_value): Likewise.
2304 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2305 (wi::unary_traits): Delete.
2306 (wi::binary_traits::signed_shift_result_type): Define for
2307 offset_int << HOST_WIDE_INT, etc.
2308 (generic_wide_int::operator <<=): Define for all types and use
2309 wi::lshift instead of <<.
2310 (wi::hwi_with_prec): Add a default constructor.
2311 (wi::ints_for): New class.
2312 (operator <<): Define for all wide-int types.
2313 (operator /): New function.
2314 (operator %): Likewise.
2315 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2316 (ASSERT_MAYBE_NE_AT): New macros.
2318 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2319 Dominik Vogt <vogt@linux.vnet.ibm.com>
2322 * emit-rtl.c (init_emit): Remove ??? comment.
2323 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2324 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2325 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2326 alignment of 3 virtual registers to BITS_PER_WORD.
2328 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2330 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2332 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2333 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2335 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2338 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2340 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2342 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2343 with the exception of debug begin stmt markers appear before
2347 * final.c (rest_of_handle_final): Call variable_tracking_main only
2348 if !flag_var_tracking.
2350 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2351 Jakub Jelinek <jakub@redhat.com>
2355 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2356 labels debug stmts that can only appear after labels.
2358 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2360 PR rtl-optimization/82398
2361 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2362 EXPR_USEFULNESS in priority comparison.
2364 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2366 PR rtl-optimization/83393
2367 * combine.c (move_deaths): If reg_stat points to a too new insn in
2368 last_death, do not use it: find the proper insn instead.
2370 2017-12-12 Jeff Law <law@redhat.com>
2372 PR tree-optimization/83298
2373 PR tree-optimization/83362
2374 PR tree-optimization/83383
2375 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2376 push_value_range a public interface. Add new argument to
2377 record_ranges_from_stmt.
2378 * gimple-ssa-evrp-analyze.c
2379 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2380 Update comments. Handle recording temporary equivalences.
2381 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2382 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2383 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2384 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2385 gimple-ssa-evrp-analyze.h.
2386 (record_temporary_equivalences_from_phis): Add new argument. When
2387 the PHI arg is an SSA_NAME, set the result's range to the range
2389 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2390 from statements too.
2391 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2392 Pass it down to children as needed.
2393 (thread_outgoing_edges): Likewise.
2394 (thread_across_edge): Likewise. Push/pop range state as needed.
2395 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2397 2017-12-12 Julia Koval <julia.koval@intel.com>
2399 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2400 (PTA_CANNONLAKE): Remove PTA_CLWB.
2402 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2405 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2406 (vcondv2div2df): Likewise.
2407 (vconduv2dfv2di): Likewise.
2409 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2411 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2412 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2413 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2414 ATTR_MATHFN_FPROUNDING_ERRNO.
2416 2017-12-12 Richard Biener <rguenther@suse.de>
2418 PR tree-optimization/83385
2419 * graphite-scop-detection.c (get_order, order): Remove.
2420 (bb_to_rpo): New global.
2422 (build_scops): Sort pbbs in RPO order.
2424 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2426 * combine.c (simplify_set): Do not transform subregs to zero_extends
2427 if the destination mode is a vector mode.
2429 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2431 PR tree-optimization/83359
2432 * tree-cfg.h (fold_loop_internal_call): Declare.
2433 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2434 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2435 (find_loop_dist_alias): New function.
2436 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2437 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2440 PR tree-optimization/80631
2441 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2442 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2443 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2444 hardcoding zero as the value if COND_EXPR is never true. For
2445 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2446 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2447 hardcoding MAX_EXPR as the reduction operation.
2448 (is_nonwrapping_integer_induction): Allow negative step.
2449 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2450 vect_create_epilog_for_reduction, if no value is suitable, don't
2451 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2453 2017-12-12 Richard Biener <rguenther@suse.de>
2455 PR tree-optimization/81889
2456 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2457 range info from the non-wrapping IV instead of just the range
2460 2017-12-12 Julia Koval <julia.koval@intel.com>
2462 * config.gcc: Add vaesintrin.h.
2463 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2465 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2466 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2468 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2469 * config/i386/immintrin.h: Include vaesintrin.h.
2470 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2471 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2472 _mm_aesdec_epi128): New intrinsics.
2474 2017-12-12 Julia Koval <julia.koval@intel.com>
2476 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2477 OPTION_MASK_ISA_VAES_UNSET): New.
2478 (ix86_handle_option): Handle -mvaes.
2479 * config/i386/cpuid.h: Define bit_VAES.
2480 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2481 * config/i386/i386-c.c (__VAES__): New.
2482 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2483 (ix86_valid_target_attribute_inner_p): Ditto.
2484 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2485 * config/i386/i386.opt: Add -mvaes.
2486 * doc/invoke.texi: Ditto.
2488 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2490 * debug.h (gcc_debug_hooks): Add inline_entry.
2491 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2492 * debug.c (do_nothing_debug_hooks): Likewise.
2493 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2494 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2495 (dwarf2_lineno_debug_hooks): Likewise.
2497 * common.opt (gstatement-frontiers): New, setting
2498 debug_nonbind_markers_p.
2499 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2500 * toplev.c (process_options): Autodetect value for debug statement
2502 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2503 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2505 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2506 markers. Integrate source bind into debug stmt expand loop.
2507 (pass_expand::execute): Check debug marker limit. Avoid deep
2508 TER and expand debug locations for debug bind insns only.
2509 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2511 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2512 * final.c (reemit_insn_block_notes): Take current block from
2513 nonbind markers. Declare note where it's first set.
2514 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2515 begin stmt markers if enabled.
2516 (notice_source_line): Handle nonbind markers. Fail if their
2517 location is unknown or that of builtins.
2518 (rest_of_handle_final): Convert begin stmt markers to notes if
2519 var-tracking didn't run.
2520 (rest_of_clean_state): Skip begin stmt markers.
2521 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2523 * function.c (allocate_struct_function): Set begin_stmt_markers.
2524 * function.h (struct function): Add debug_marker_count counter
2525 and debug_nonbind_markers flag.
2526 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2527 * gimple-low.c (lower_function_body): Adjust
2528 debug_nonbind_markers.
2529 (lower_stmt): Drop or skip gimple debug stmts.
2530 (lower_try_catch): Skip debug stmts.
2531 * gimple.c (gimple_build_debug_begin_stmt): New.
2532 (gimple_copy): Increment debug_marker_count if copying one.
2533 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2534 * gimplify.c (rexpr_location): New.
2535 (rexpr_has_location): New.
2536 (warn_switch_unreachable_r): Handle gimple debug stmts.
2537 (shortcut_cond_r): Call expr_location.
2539 (find_goto_label): New.
2540 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2542 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2544 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2545 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2546 (LANG_HOOKS_INITIALIZER): ... this.
2547 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2548 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2550 (update_ebb_live_info): Skip debug insn markers.
2551 * lra.c (debug_insn_static_data): Rename to...
2552 (debug_bind_static_data): ... this.
2553 (debug_marker_static_data): New.
2554 (lra_set_insn_recog_data): Select one of the above depending
2556 (lra_update_isn_regno_info): Don't assume debug insns have
2558 (push_insns): Skip debug insns.
2559 * lto-streamer-in.c (input_function): Drop debug stmts
2560 depending on active options. Adjust debug_nonbind_markers.
2561 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2562 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2563 begin stmt marker notes.
2564 (print_insn): Likewise.
2565 * recog.c (extract_insn): Recognize rtl for debug markers.
2566 * rtl.def (DEBUG_MARKER): New.
2567 * tree-inline.c: Include params.h.
2568 (remap_gimple_stmt): Handle nonbind markers.
2569 (maybe_move_debug_stmts_to_successors): Likewise.
2570 (copy_debug_stmt): Likewise.
2571 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2572 markers regardless of no side effects.
2573 (tsi_link_before): Don't update container's side effects when adding
2574 a begin stmt marker.
2575 (tsi_link_after): Likewise.
2576 (expr_first): Skip begin stmt markers.
2577 (expr_last): Likewise.
2578 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2579 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2580 Disregard nonbind markers.
2581 * tree.c (make_node_stat): Don't set side effects for begin stmt
2583 (build1_stat): Likewise.
2584 * tree.def (DEBUG_BEGIN_STMT): New.
2585 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2586 * var-tracking.c (delete_debug_insns): Renamed to...
2587 (delete_vta_debug_insns): ... this.
2588 (reemit_marker_as_note): New.
2589 (vt_initialize): Reemit markers.
2590 (delete_vta_debug_insns): Likewise.
2591 (vt_debug_insns_local): Reemit or delete markers.
2592 (variable_tracking_main_1): Likewise.
2593 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2594 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2595 (gimple_debug_nonbind_marker_p): New.
2596 (gimple_build_debug_bind): Adjust.
2597 (gimple_build_debug_begin_stmt): New.
2598 * doc/invoke.texi (max-debug-marker-count): New param.
2599 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2600 (debug_parameter_ref, debug_marker): New.
2601 (NOTE_INSN_BEGIN_STMT): New.
2602 (DEBUG_INSN): Describe begin stmt markers.
2604 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2605 without debug insns, we wouldn't, but clean up debug insns
2606 after a control flow insn nevertheless.
2608 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2609 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2610 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2611 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2612 (rtl_tidy_fallthru_edge): Likewise.
2613 (rtl_verify_fallthru): Likewise.
2614 (rtl_verify_bb_layout): Likewise.
2615 (skip_insns_after_block): Likewise.
2616 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2617 * dwarf2out.c: Include print-rtl.h.
2618 (dwarf2out_next_real_insn): New.
2619 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2620 Dump debug binds in asm comments.
2621 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2622 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2623 callers to use gsi_start_nondebug_bb instead.
2624 (gsi_after_labels): Skip gimple debug stmts.
2625 (gsi_start_nondebug): New.
2626 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2627 (proper_loop_form_for_interchange): Adjust.
2628 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2629 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2630 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2631 (collect_fallthrough_labels): Likewise.
2632 (should_warn_for_implicit_fallthrough): Likewise.
2633 (warn_implicit_fallthrough_r): Likewise.
2634 (expand_FALLTHROUGH_r): Likewise.
2635 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2636 (graphite_copy_stmts_from_block): Skip nonbind markers.
2637 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2638 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2639 * jump.c (clean_barriers): Skip debug insns.
2640 * omp-expand.c (expand_parallel_call): Skip debug insns.
2641 (expand_task_call): Likewise.
2642 (remove_exit_barrier): Likewise.
2643 (expand_omp_taskreg): Likewise.
2644 (expand_omp_for_init_counts): Likewise.
2645 (expand_omp_for_generic): Likewise.
2646 (expand_omp_for_static_nochunk): Likewise.
2647 (expand_omp_for_static_chunk): Likewise.
2648 (expand_omp_simd): Likewise.
2649 (expand_omp_taskloop_for_outer): Likewise.
2650 (expand_omp_taskloop_for_inner): Likewise.
2651 (expand_oacc_for): Likewise.
2652 (expand_omp_sections): Likewise.
2653 (expand_omp_single): Likewise.
2654 (expand_omp_synch): Likewise.
2655 (expand_omp_atomic_load): Likewise.
2656 (expand_omp_atomic_store): Likewise.
2657 (expand_omp_atomic_fetch_op): Likewise.
2658 (expand_omp_atomic_pipeline): Likewise.
2659 (expand_omp_atomic_mutex): Likewise.
2660 (expand_omp_target): Likewise.
2661 (grid_expand_omp_for_loop): Likewise.
2662 (grid_expand_target_grid_body): Likewise.
2663 (build_omp_regions_1): Likewise.
2664 * omp-low.c (check_combined_parallel): Skip debug stmts.
2665 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2666 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2667 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2669 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2670 (make_edges): Likewise.
2671 (cleanup_dead_labels): Likewise.
2672 (gimple_can_merge_blocks_p): Likewise.
2673 (stmt_starts_bb_p): Likewise.
2674 (gimple_block_label): Likewise.
2675 (gimple_redirect_edge_and_branch): Likewise.
2676 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2678 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2680 * tree-pretty-print (print_declaration): Omit initializer in slim
2682 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2684 (eliminate_unnecessary_stmts): Stabilize block removal order.
2685 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2686 * var-tracking.c (get_first_insn): New.
2687 (vt_emit_notes): Call it.
2688 (vt_initialize): Walk any insns before the first BB.
2689 (delete_debug_insns): Likewise.
2691 * gimple.h (enum gimple_debug_subcode): Add
2692 GIMPLE_DEBUG_BEGIN_STMT.
2693 (gimple_debug_begin_stmt_p): New.
2694 (gimple_debug_nonbind_marker_p): New.
2695 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2696 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2697 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2698 * insn-notes.def (BEGIN_STMT): New.
2699 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2700 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2701 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2702 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2703 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2704 (INSN_DEBUG_MARKER_KIND): New.
2705 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2706 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2707 (INSN_VAR_LOCATION_PTR): New.
2708 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2710 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2711 deps only if debug bind insns are enabled.
2712 (pass_expand::execute): Avoid deep TER and expand
2713 debug locations for debug bind insns only.
2714 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2715 debug stmts special handling down to debug bind stmts.
2716 * combine.c (try_combine): Narrow debug insns special handling
2717 down to debug bind insns.
2718 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2719 Narrow debug insns preexisting special handling down to debug
2721 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2722 handling down to debug bind insns.
2723 * function.c (instantiate_virtual_regs): Skip debug markers,
2724 adjust handling of debug binds.
2725 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2726 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2727 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2728 insns to debug bind insns.
2729 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2730 special handling of debug stmts to debug bind stmts.
2731 * ipa-split.c (split_function): Likewise.
2732 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2733 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2735 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2736 * regrename.c (build_def_use): Likewise.
2737 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2738 (pass_cprop_hardreg): Narrow special casing of debug insns to
2740 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2741 * reload1.c (reload): Likewise.
2742 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2743 casing of debug stmts to debug bind stmts.
2744 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2745 * ssa-iterators.h (num_imm_uses): Likewise.
2746 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2747 debug stmts to debug bind stmts.
2748 * tree-inline.c (tree_function_versioning): Narrow special casing
2749 of debug stmts to debug bind stmts.
2750 * tree-loop-distribution.c (generate_loops_for_partition):
2751 Narrow special casing of debug stmts to debug bind stmts.
2752 * tree-sra.c (analyze_access_subtree): Narrow special casing
2753 of debug stmts to debug bind stmts.
2754 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2755 stmts to debug bind stmts.
2756 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2757 casing of debug stmts to debug bind stmts.
2758 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2759 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2760 casing of debug stmts to debug bind stmts.
2761 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2763 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2764 debug stmts to debug bind stmts.
2765 (gimple_replace_ssa_lhs): Likewise.
2766 (insert_debug_temp_for_var_def): Likewise.
2767 (insert_debug_temps_for_defs): Likewise.
2768 (reset_debug_uses): Likewise.
2769 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2770 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2771 (adjust_debug_stmts): Likewise.
2772 (adjust_phi_and_debug_stmts): Likewise.
2773 (vect_do_peeling): Likewise.
2774 * tree-vect-loop.c (vect_transform_loop): Likewise.
2775 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2776 * var-tracking.c (adjust_mems): Narrow special casing of debug
2777 insns to debug bind insns.
2778 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2779 (compute_bb_dataflow, vt_find_locations): Likewise.
2780 (vt_expand_loc, emit_notes_for_changes): Likewise.
2781 (vt_init_cfa_base): Likewise.
2782 (vt_emit_notes): Likewise.
2783 (vt_initialize): Likewise.
2784 (vt_finalize): Likewise.
2786 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2787 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2788 (next_nonnote_nondebug_insn_bb): New.
2789 (prev_nonnote_nondebug_insn_bb): New.
2790 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2791 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2792 (prev_nonnote_nondebug_insn_bb): Declare.
2793 (next_nonnote_nondebug_insn_bb): Declare.
2794 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2795 * cfgrtl.c (get_last_bb_insn): Likewise.
2796 * lra.c (push_insns): Likewise.
2798 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2801 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2803 * selftest.c (selftest::test_runner::test_runner): New ctor.
2804 (selftest::test_runner::~test_runner): New dtor.
2805 * selftest.h (class selftest::test_runner): New class.
2807 2017-12-11 Carl Love <cel@us.ibm.com>
2809 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2810 vec_extract_fp32_from_shortl]): Add #defines.
2811 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2812 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2813 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2814 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2815 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2816 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2817 * doc/extend.texi: Add documentation for the added builtins.
2819 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2821 PR rtl-optimization/80693
2822 PR rtl-optimization/81019
2823 PR rtl-optimization/81020
2824 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2825 are not mentioned in i3. Place the REG_UNUSED note on i2,
2826 possibly modified to REG_DEAD, if it did not originate in i3.
2828 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2830 * recog.c (store_data_bypass_p_1): New function.
2831 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2832 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2834 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2836 PR rtl-optimization/83361
2837 * ifcvt.c (if_convert): Call fixup_partitions.
2839 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2841 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2842 early folding of splat_u{8,16,32}.
2844 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2846 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2847 output_operand_lossage first argument with capital letter.
2848 (aarch64_override_options): Don't start error and sorry first argument
2849 with capital letter.
2851 2017-12-11 Andi Kleen <ak@linux.intel.com>
2853 PR gcov-profile/83355
2854 * auto-profile.c (string_table::get_index_by_decl): Don't
2855 recurse when abstract origin points to itself.
2857 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2859 PR tree-optimization/83320
2860 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2861 (prune_datarefs_not_in_loop): Ditto.
2863 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2865 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2867 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2869 PR tree-optimization/83337
2870 * gimple-loop-interchange.cc (compute_access_stride): Handle
2871 bitfield DRs properly.
2873 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2875 PR tree-optimization/83338
2876 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2877 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2878 vector of unsigned integers to vector of signed integers.
2880 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2882 PR rtl-optimization/83317
2883 * lra-constraints.c (process_address_1): Add insn code check.
2885 2017-12-08 Michael Matz <matz@suse.de>
2887 Fix PR tree-optimization/83323
2888 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2889 head-controlled loops and loop BBs.
2890 * common.opt (funroll-and-jam): Remove, instead ...
2891 (floop-unroll-and-jam): ... reuse this option.
2892 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2893 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2894 (-floop-unroll-and-jam): ... this option.
2896 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2898 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2899 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2900 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2902 2017-12-08 Richard Biener <rguenther@suse.de>
2905 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2906 handle accesses outside of zero-sized vars.
2908 2017-12-08 Martin Jambor <mjambor@suse.cz>
2910 PR tree-optimization/83141
2911 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2912 test for MEM_REFs implicitely changing types with padding. Remove
2914 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2916 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2918 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2920 * config/avr/avr.c (avr_attribute_table): Likewise.
2921 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2922 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2923 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2924 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2925 of struct attribute_spec.
2926 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2928 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2931 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2933 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2936 * profile-count.c (profile_count::from_gcov_type): Move from
2937 profile-count.h; handle overflow.
2938 * profile-count.h (profile_count::from_gcov_type): Move offline.
2940 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2942 PR rtl-optimization/83304
2943 * combine.c (move_deaths): If we do not know where a register died,
2946 2017-12-08 Richard Biener <rguenther@suse.de>
2948 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2949 Provide -fopt-info-loop feedback when we interchange in a nest.
2951 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2953 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2954 for armv6 ARM CPU IDs.
2956 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2958 * common/config/arm/arm-common.c: Include <algorithm>.
2959 (INCLUDE_VECTOR): Define.
2960 (compare_opt_names): New function.
2961 (arm_rewrite_selected_arch): Only strip out extensions that can be
2962 expressed through -mfpu. Sort the remaining extensions
2965 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2967 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2968 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2969 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2970 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2972 2017-06-08 Tristan Gingold <gindold@adacore.com>
2975 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2976 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2978 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2980 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2981 of struct attribute_spec.
2983 2017-12-08 Julia Koval <julia.koval@intel.com>
2985 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2986 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2987 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2988 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2989 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2990 _mm_maskz_dpwssds_epi32): Ditto.
2992 2017-12-08 Richard Biener <rguenther@suse.de>
2994 PR tree-optimization/81303
2995 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2996 conditions try to create a comparison vector type matching
2997 the data vector type.
2998 (vectorizable_condition): Adjust.
2999 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3000 Leave invariant conditions alone in case we can vectorize those.
3002 2017-12-08 Julia Koval <julia.koval@intel.com>
3004 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3005 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3006 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3007 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3008 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3010 2017-12-08 Julia Koval <julia.koval@intel.com>
3012 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3013 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3014 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3015 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3016 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3017 _mm_maskz_dpbusds_epi32): New intrinsics.
3019 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3021 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3023 (ld<bh>io_signed): Likewise.
3024 (st<bhw>io): Likewise.
3025 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3026 SMALL_INT12 constant integer operand.
3028 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3030 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3032 (nios2_large_constant_allowed): ...this. Adjust uses.
3033 (nios2_plus_symbolic_constant_p): Rename to...
3034 (nios2_plus_large_constant_p): ...this. Adjust uses.
3035 (nios2_legitimate_address_p): Correct CONST_INT handling.
3036 (nios2_symbolic_memory_operand_p): Rename to...
3037 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3038 (nios2_large_constant_p): Check for large constant integers too.
3039 (nios2_split_large_constant): Handle constant integers.
3040 (nios2_split_symbolic_memory_operand): Rename to...
3041 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3042 (nios2_legitimize_constant_address): Handle constant integers.
3043 (r0rel_constant_p): Handle small constant integers.
3044 (nios2_print_operand_address): Handle r0-relative integer addresses.
3045 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3046 * config/nios2/nios2.md: Adjust for renamed functions.
3048 2017-12-07 Andrew Waterman <andrew@sifive.com>
3050 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3051 (TARGET_HAVE_SRODATA_SECTION): New define.
3052 (riscv_select_section): New function.
3054 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3055 Alexander Monakov <amonakov@ispras.ru>
3056 Jakub Jelinek <jakub@redhat.com>
3059 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3061 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3063 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3064 a bool success value. Don't call output_operand_lossage here.
3065 (aarch64_print_ldpstp_address): Return a bool success value.
3066 (aarch64_print_operand_address): Call output_addr_const if
3067 aarch64_print_address_internal fails.
3068 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3069 'y'; call output_operand_lossage instead. Call output_operand_lossage
3070 if aarch64_print_ldpstp_address fails.
3072 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3074 * tree-vector-builder.h
3075 (tree_vector_builder::binary_encoded_nelts): Declare.
3076 * tree-vector-builder.c
3077 (tree_vector_builder::binary_encoded_nelts): New function.
3078 * fold-const.c (negate_expr_p): Likewise.
3079 (operand_equal_p, fold_checksum_tree): Likewise.
3080 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3081 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3082 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3083 (uniform_vector_p): Likewise.
3084 * varasm.c (const_hash_1, compare_constant): Likewise.
3085 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3086 (valid_lattice_transition): Operate directly on the VECTOR_CST
3088 * ipa-icf.c: Include tree-vector-builder.h.
3089 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3090 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3092 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3094 * tree.c (build_vector): Delete.
3095 * tree.h (build_vector): Make static and move into the self-testing
3098 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3100 * vector-builder.h (vector_builder::derived): New const overload.
3101 (vector_builder::elt): New function.
3102 * tree-vector-builder.h (tree_vector_builder::type): New function.
3103 (tree_vector_builder::apply_step): Declare.
3104 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3106 * gimple-fold.h (tree_vector_builder): Declare.
3107 (gimple_build_vector): Take a tree_vector_builder instead of a
3108 type and vector of elements.
3109 * gimple-fold.c (gimple_build_vector): Likewise.
3110 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3112 (get_initial_defs_for_reduction): Likewise.
3113 (vectorizable_induction): Likewise.
3115 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3117 * tree-vector-builder.h
3118 (tree_vector_builder::new_binary_operation): Declare.
3119 * tree-vector-builder.c
3120 (tree_vector_builder::new_binary_operation): New function.
3121 * fold-const.c (fold_relational_const): Use it.
3122 (const_binop): Likewise. Check that both input vectors have
3123 the same number of elements, thus excluding things like WIDEN_SUM.
3124 Check whether it is possible to operate directly on the encodings
3127 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3129 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3130 new_unary_operation, operating only on the encoded elements.
3131 (const_unop): Likewise.
3132 (exact_inverse): Likewise.
3133 (distributes_over_addition_p): New function.
3134 (const_binop): Use tree_vector_builder and new_unary_operation
3135 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3136 on the encoded elements unless the encoding is strided and the
3137 operation does not distribute over addition.
3138 (fold_convert_const): Use tree_vector_builder and
3139 new_unary_operation. Operate only on the encoded elements
3140 for truncating integer conversions, or for non-stepped encodings.
3142 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3144 * config/sparc/sparc.c: Include tree-vector-builder.h.
3145 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3146 * expmed.c: Include tree-vector-builder.h.
3147 (make_tree): Use tree_vector_builder instead of build_vector.
3148 * fold-const.c: Include tree-vector-builder.h.
3149 (const_binop): Use tree_vector_builder instead of build_vector.
3150 (const_unop): Likewise.
3151 (native_interpret_vector): Likewise.
3152 (fold_vec_perm): Likewise.
3153 (fold_ternary_loc): Likewise.
3154 * gimple-fold.c: Include tree-vector-builder.h.
3155 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3157 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3158 (simplify_vector_constructor): Use tree_vector_builder instead
3160 * tree-vect-generic.c: Include tree-vector-builder.h.
3161 (add_rshift): Use tree_vector_builder instead of build_vector.
3162 (expand_vector_divmod): Likewise.
3163 (optimize_vector_constructor): Likewise.
3164 * tree-vect-loop.c: Include tree-vector-builder.h.
3165 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3166 of build_vector. Explicitly use a stepped encoding for
3168 * tree-vect-slp.c: Include tree-vector-builder.h.
3169 (vect_get_constant_vectors): Use tree_vector_builder instead
3171 (vect_transform_slp_perm_load): Likewise.
3172 (vect_schedule_slp_instance): Likewise.
3173 * tree-vect-stmts.c: Include tree-vector-builder.h.
3174 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3175 (vect_gen_perm_mask_any): Likewise.
3176 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3177 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3179 (build_vector_from_val): Likewise. Explicitly use a duplicate
3182 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3184 * doc/generic.texi (VECTOR_CST): Describe new representation of
3186 * vector-builder.h: New file.
3187 * tree-vector-builder.h: Likewise.
3188 * tree-vector-builder.c: Likewise.
3189 * Makefile.in (OBJS): Add tree-vector-builder.o.
3190 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3191 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3192 (tree_vector): Change the number of elements to
3193 vector_cst_encoded_nelts.
3194 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3195 (VECTOR_CST_ELTS): Delete.
3196 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3197 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3198 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3199 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3200 (VECTOR_CST_ENCODED_ELT): Likewise.
3201 (vector_cst_encoded_nelts): New function.
3202 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3203 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3204 (vector_cst_int_elt, vector_cst_elt): Declare.
3205 * tree.c: Include tree-vector-builder.h.
3206 (tree_code_size): Abort if passed VECTOR_CST.
3207 (tree_size): Update for new VECTOR_CST layout.
3208 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3209 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3210 (build_vector): Use tree_vector_builder.
3211 (vector_cst_int_elt, vector_cst_elt): New functions.
3212 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3213 encoded elements and then create the vector in the canonical form.
3214 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3215 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3216 (tree_c_tests): Call test_vector_cst_patterns.
3217 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3219 (hash_tree): Likewise.
3220 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3221 (streamer_write_tree_header): Likewise.
3222 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3223 (streamer_alloc_tree): Likewise. Update call to make_vector.
3224 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3226 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3228 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3229 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3230 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3232 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3233 Richard Biener <rguenther@suse.de>
3235 PR tree-optimization/81303
3236 * Makefile.in (gimple-loop-interchange.o): New object file.
3237 * common.opt (floop-interchange): Reuse the option from graphite.
3238 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3239 -floop-interchange and mention it for -O3.
3240 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3241 * gimple-loop-interchange.cc: New file.
3242 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3243 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3244 * passes.def (pass_linterchange): New pass.
3245 * timevar.def (TV_LINTERCHANGE): New time var.
3246 * tree-pass.h (make_pass_linterchange): New declaration.
3247 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3248 interchange. Record IV before/after increment in new parameters.
3249 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3250 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3252 (check_reduction_path): ...New function here.
3253 * tree-vectorizer.h (check_reduction_path): New declaration.
3255 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3258 PR rtl-optimization/80818
3259 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3260 always early clobbered.
3261 * lra-lives.c (process_bb_lives): Check input hard regs for early
3262 clobbered non-operand hard reg.
3264 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3267 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3268 types_compatible_p, just that TYPE_MODE is the same.
3270 2017-12-07 Martin Sebor <msebor@redhat.com>
3273 * attribs.c (empty_attribute_table): Initialize new member of
3274 struct attribute_spec.
3275 (decl_attributes): Add argument. Handle mutually exclusive
3276 combinations of attributes.
3277 (selftests::test_attribute_exclusions): New function.
3278 (selftests::attribute_c_tests): Ditto.
3279 * attribs.h (decl_attributes): Add default argument.
3280 * selftest.h (attribute_c_tests): Declare.
3281 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3282 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3284 * doc/extend.texi (Common Function Attributes): Update const and pure.
3285 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3286 of struct attribute_spec.
3287 * config/arc/arc.c (arc_attribute_table): Same.
3288 * config/arm/arm.c (arm_attribute_table): Same.
3289 * config/avr/avr.c ( avr_attribute_table): Same.
3290 * config/bfin/bfin.c (bfin_attribute_table): Same.
3291 * config/cr16/cr16.c (cr16_attribute_table): Same.
3292 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3293 * config/h8300/h8300.c (h8300_attribute_table): Same.
3294 * config/i386/i386.c (ix86_attribute_table): Same.
3295 * config/ia64/ia64.c (ia64_attribute_table): Same.
3296 * config/m32c/m32c.c (m32c_attribute_table): Same.
3297 * config/m32r/m32r.c (m32r_attribute_table): Same.
3298 * config/m68k/m68k.c (m68k_attribute_table): Same.
3299 * config/mcore/mcore.c (mcore_attribute_table): Same.
3300 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3301 * config/mips/mips.c (mips_attribute_table): Same.
3302 * config/msp430/msp430.c (msp430_attribute_table): Same.
3303 * config/nds32/nds32.c (nds32_attribute_table): Same.
3304 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3305 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3306 * config/rl78/rl78.c (rl78__attribute_table): Same.
3307 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3308 * onfig/rx/rx.c (rx_attribute_table): Same.
3309 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3310 * config/sh/sh.c (sh_attribute_table): Same.
3311 * config/sparc/sparc.c (sparc_attribute_table): Same.
3312 * config/spu/spu.c (spu_attribute_table): Same.
3313 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3314 * config/v850/v850.c (v850_attribute_table): Same.
3315 * config/visium/visium.c (visium_attribute_table): Same.
3317 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3320 * config/arm/arm.c (INCLUDE_STRING): Define.
3321 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3322 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3325 2017-12-07 Michael Matz <matz@suse.de>
3327 Add unroll and jam pass
3329 * gimple-loop-jam.c: New file.
3330 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3331 * common.opt (funroll-and-jam): New option.
3332 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3333 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3334 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3335 * passes.def: Add pass_loop_jam.
3336 * timevar.def (TV_LOOP_JAM): Add.
3337 * tree-pass.h (make_pass_loop_jam): Declare.
3338 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3339 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3340 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3341 to flow_loop_tree_node_add.
3342 (duplicate_subloops, copy_loops_to): Append to sibling list.
3343 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3344 * doc/invoke.texi (-funroll-and-jam): Document new option.
3345 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3347 2017-12-07 Richard Biener <rguenther@suse.de>
3349 PR tree-optimization/83296
3350 PR tree-optimization/67769
3351 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3352 flow sensitive info in an unrelated BB.
3353 (value_replacement): Use reset_flow_sensitive_info.
3354 (minmax_replacement): Reset flow sensitive info on the def
3355 we move. Do not reset flow sensitive info in the whole BB
3356 we move the stmt to.
3357 (abs_replacement): Likewise.
3359 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3362 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3363 rs6000_cpu to the given -mcpu=, or to the default processor.
3365 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3367 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3368 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3369 instead of rs6000_cpu_attr.
3370 (rs6000_adjust_cost): Ditto.
3371 (is_microcoded_insn): Ditto.
3372 (rs6000_adjust_priority): Ditto.
3373 (rs6000_issue_rate): Ditto.
3374 (rs6000_use_sched_lookahead): Ditto.
3375 (rs6000_use_sched_lookahead_guard): Ditto.
3376 (rs6000_sched_reorder): Ditto.
3377 (force_new_group): Ditto.
3378 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3379 (group_ending_nop): Ditto.
3381 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3383 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3384 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3385 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3386 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3388 (rs6000_reassociation_width): Ditto.
3389 (rs6000_emit_epilogue): Ditto.
3390 (rs6000_adjust_cost): Ditto.
3391 (is_microcoded_insn): Ditto.
3392 (is_cracked_insn): Ditto.
3393 (rs6000_adjust_priority): Ditto.
3394 (rs6000_sched_reorder): Ditto.
3395 (rs6000_sched_reorder2): Ditto.
3396 (insn_must_be_first_in_group): Ditto.
3397 (insn_must_be_last_in_group): Ditto.
3398 (rs6000_register_move_cost): Ditto.
3399 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3402 2017-12-07 Julia Koval <julia.koval@intel.com>
3404 * config.gcc: Add vaesintrin.h.
3405 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3407 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3408 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3410 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3411 * config/i386/immintrin.h: Include vaesintrin.h.
3412 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3413 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3414 _mm_aesdec_epi128): New intrinsics.
3416 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3418 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3419 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3420 c-family/c-spellcheck.cc.
3421 (best_macro_match::best_macro_match): Likewise.
3423 (struct edit_distance_traits<cpp_hashnode *>): Move to
3424 c-family/c-spellcheck.h.
3425 (class best_macro_match): Likewise.
3427 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3429 PR tree-optimization/83293
3430 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3431 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3432 might insert into empty bb.
3435 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3436 simplify for plus with :c added, and pointer_plus without that.
3437 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3438 with undefined overflow and the conversion is not widening,
3439 perform negation in utype and only convert to type afterwards.
3440 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3441 simplify for plus with :c added, and pointer_plus without that.
3442 If type is integral with undefined overflow and the conversion is
3443 not widening, perform minus in utype and only convert to type
3444 afterwards. Move the last pointer_diff_expr simplify into the
3447 2017-12-06 Martin Sebor <msebor@redhat.com>
3449 PR tree-optimization/82646
3450 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3451 strncpy, not maxlen.
3453 2017-12-06 Martin Sebor <msebor@redhat.com>
3455 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3458 PR tree-optimization/83075
3459 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3460 strncat/strncpy don't change length of source string.
3462 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3465 2017-11-29 Martin Aberg <maberg@gaisler.com>
3467 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3468 to prevent b2bst errata sequence.
3469 (sqrtdf2_fix): Likewise.
3471 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3473 PR tree-optimization/81945
3474 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3475 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3476 to dest_cfun has orig_loop_num set, either remap it to the new
3477 loop number if the loop got moved too, or clear it.
3479 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3481 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3483 (thunderx2t99_nothing): New insn reservation.
3484 (thunderx2t99_mrs): New insn reservation.
3485 (thunderx2t99_multiple): New insn reservation.
3486 (thunderx2t99_alu_basi): Add bfx to reservation.
3487 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3489 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3492 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3494 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3496 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3497 tree-ssa-pre.c::remove_dead_inserted_code.
3498 * tree-ssa-dce.h: New file.
3499 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3500 (remove_dead_inserted_code): Move and rename to function
3501 tree-ssa-dce.c::simple_dce_from_worklist.
3502 (pass_pre::execute): Update use.
3504 2017-12-05 Richard Biener <rguenther@suse.de>
3506 PR tree-optimization/83277
3507 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3508 to code-gen liveout vars.
3510 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3512 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3513 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3514 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3517 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3519 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3521 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3522 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3525 2017-12-05 Richard Biener <rguenther@suse.de>
3527 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3528 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3529 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3530 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3532 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3534 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3536 2017-12-05 Martin Liska <mliska@suse.cz>
3537 Jakub Jelinek <jakub@redhat.com>
3539 * doc/invoke.texi: Document the options.
3540 * flag-types.h (enum sanitize_code): Add
3541 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3542 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3543 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3544 * opts.c: Define new sanitizer options.
3545 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3546 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3548 2017-12-05 Julia Koval <julia.koval@intel.com>
3550 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3551 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3552 (ix86_handle_option): Handle -mavx512vnni.
3553 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3554 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3555 * config/i386/i386-c (__AVX512VNNI__): New.
3556 * config/i386/i386.c (ix86_target_string): Handle new option.
3557 (ix86_valid_target_attribute_inner_p): Handle new option.
3558 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3559 * config/i386/i386.opt (mavx512vnni): New option.
3561 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3564 * athlon.md: Disable for generic.
3565 * haswell.md: Enable for generic.
3566 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3567 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3569 (ix86_adjust_cost): Move generic to haswell path.
3571 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3573 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3574 instead of MEM_P in a couple more places. Fix formatting issues.
3576 2017-12-04 Jim Wilson <jimw@sifive.com>
3578 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3579 instead of GP_REG_LAST-1.
3580 (riscv_adjust_libcall_cfi_prologue): Likewise.
3581 (riscv_adjust_libcall_cri_epilogue): Likewise.
3582 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3585 2017-12-04 Luis Machado <luis.machado@linaro.org>
3587 * ipa-pure-const.c (check_decl): Add missing newline.
3588 (state_from_flags): Likewise.
3590 2017-12-04 Jeff Law <law@redhat.com>
3592 PR tree-optimizatin/78496
3593 * gimple-ssa-evrp-analyze.h
3594 (evrp_range_analyzer::get_vr_values): Simplify.
3595 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3596 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3597 and gimple-ssa-evrp-analyze.h.
3598 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3599 (simplify_stmt_for_jump_threading): Copy a blob of code from
3600 tree-vrp.c to use ranges to simplify statements.
3601 (dom_opt_dom_walker::before_dom_children): Call
3602 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3603 (dom_opt_dom_walker::after_dom_children): Similarly for
3604 evrp_range_analyzer::leave.
3605 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3608 * gimple-ssa-evrp-analyze.c
3609 (evrp_range_analyzer::extract_range_from_stmt): Always use
3610 vr_values::update_value_range so preexisting range info is
3611 medged with new range info, even if the new range is VR_VARYING.
3613 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3615 * combine.c: Adjust comment.
3616 (use_crosses_set_p): Delete.
3617 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3618 (try_combine): Ditto.
3620 2017-12-04 Richard Biener <rguenther@suse.de>
3622 PR tree-optimization/83255
3623 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3624 Re-add zero-iteration check.
3626 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3628 PR rtl-optimization/83245
3629 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3630 hard registers as earlyclobber, also if not in an asm.
3632 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3636 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3639 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3640 rs6000_cpu based on cpu_index, not tune_index.
3642 2017-12-04 Richard Biener <rguenther@suse.de>
3644 PR tree-optimization/83238
3645 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3646 code match comment, rejecting invalid SESE regions.
3648 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3650 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3651 require base operand is a REG_POINTER prior to reload on targets
3652 with non-equivalent space registers.
3654 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3656 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3657 (update_specialized_profile): Fix updating of counts.
3658 (perhaps_add_new_callers): Likewise.
3660 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3663 * x86-tune.def: Remove obsolette FIXMEs.
3664 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3665 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3666 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3668 (X86_TUNE_PAD_RETURNS): Disable for generic.
3670 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3672 PR tree-optimization/83170
3673 PR tree-optimization/83241
3674 * gimple-ssa-store-merging.c
3675 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3676 gimple_vuse (ins_stmt) in case it has changed.
3677 (imm_store_chain_info::output_merged_store): Likewise.
3679 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3683 * tree-cfg.c (pass_warn_function_return::execute): Handle
3684 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3685 with BUILTINS_LOCATION.
3689 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3690 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3691 instead of DECL_MODE.
3693 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3694 last argument to rtx pointer.
3695 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3696 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3697 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3698 depending on the chosen ISAs.
3699 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3700 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3701 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3703 * config/i386/sse.md (mov<mode>_internal): Likewise.
3704 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3706 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3708 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3710 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3712 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3713 parameters from prototype.
3714 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3715 parameters. Don't print a comment.
3716 (emit_fusion_gpr_load): Adjust.
3717 (emit_fusion_load_store): Adjust.
3718 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3719 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3720 comment on the second line.
3722 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3725 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3726 rs6000_cpu based on cpu_index, not tune_index.
3728 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3730 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3731 which_alternative instead of which_alternative + 1.
3732 (output_asm_insn): Print an extra tab if the template is short.
3734 2017-12-01 Jim Wilson <jimw@sifive.com>
3736 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3738 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3739 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3741 * doc/tm.texi: Regenerate.
3743 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3746 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3747 whether we can allocate pseudos before trying to fix an address.
3748 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3749 memory address is indexed or indirect.
3750 (floatuns_<mode>si2_hw2): Likewise.
3752 2017-12-01 Jason Merrill <jason@redhat.com>
3754 * Makefile.in (TAGS): Add c-family/*.cc.
3756 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3758 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3759 (call_value_insn): Likewise.
3760 (sibcall_insn): Likewise.
3761 (sibcall_value_insn): Likewise.
3762 (movsi_aarch64): Likewise.
3763 (movdi_aarch64): Likewise.
3764 (add_losym_): Likewise.
3765 (ldr_got_small_): Likewise.
3766 (ldr_got_small_sidi): Likewise.
3767 (ldr_got_small_28k_): Likewise.
3768 (ldr_got_small_28k_sidi): Likewise.
3769 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3770 Move output_addr_const to symbolic case. Add error check.
3772 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3774 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3775 * config/s390/s390.md: Disable MVC merging peephole if it would
3776 disable operand forwarding.
3777 (new peephole2): Split MVCs if it would turn them into up to 2
3780 2017-12-01 Richard Biener <rguenther@suse.de>
3782 PR tree-optimization/83232
3783 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3784 detection of same access. Instead of breaking the group here
3785 do not consider the duplicate. Add comment explaining real fix.
3787 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3789 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3791 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3793 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3795 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3797 * function.h (struct function): Remove cilk_frame_decl,
3798 is_cilk_function and calls_cilk_spawn fields.
3799 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3801 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3802 cilk_elemental field.
3803 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3804 * target.def: Adjust comment.
3805 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3806 Don't test cilk_elemental.
3808 PR tree-optimization/83233
3809 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3810 bswap_stat name for the struct.
3813 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3814 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3815 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3816 SWITCH_BREAK_LABEL_P set on the label.
3817 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3818 added for default case if it was missing and not all cases covered.
3819 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3820 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3822 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3823 to avoid -Wimplicit-fallthrough warning.
3824 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3825 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3827 PR tree-optimization/83221
3828 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3830 (init_reassoc): Formatting fix.
3833 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3834 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3838 * tree-cfg.c: Include asan.h.
3839 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3841 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3843 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3845 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3847 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3848 vpcmpeqd instruction.
3850 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3852 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3853 registers type for 128bit mode.
3855 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3857 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3859 * gengtype-state.c (read_state_pair): Likewise.
3860 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3861 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3862 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3863 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3864 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3865 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3866 * builtins.c (fold_builtin_3): Likewise.
3867 * graphite-scop-detection.c
3868 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3869 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3872 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3875 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3876 parameter. Return true if ignore_decl_rtl_set_p is true and other
3877 conditions are satisfied.
3878 * asan.h (asan_protect_global): Add new parameter.
3879 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3880 to asan_protect_global calls.
3882 2017-11-30 Jim Wilson <jimw@sifive.com>
3884 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3885 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3886 -mstrict-align, add info on default value. Delete redundant lines for
3887 -mabi. Add missing -mexplicit-relocs docs.
3889 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3891 * config/arc/arc.md (trap): New pattern.
3893 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3895 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3896 instruction to end into a delay slot.
3897 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3898 can be placed into a delay slot against reg_note.
3900 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3902 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3903 labels number of usages.
3905 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3907 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3909 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3911 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3914 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3915 multiplication by power of 2 constant into two shifts + comparison.
3917 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3920 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3922 2017-11-30 Richard Biener <rguenther@suse.de>
3924 PR tree-optimization/83202
3925 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3926 allow_peel argument and guard peeling.
3927 (canonicalize_loop_induction_variables): Likewise.
3928 (canonicalize_induction_variables): Pass false.
3929 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3930 peeling from cunrolli.
3932 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3934 * combine.c (try_combine): Print a message to dump file whenever
3935 I0, I1, or I2 cannot be combined into I3.
3937 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3939 PR rtl-optimization/83156
3940 PR rtl-optimization/82621
3941 * combine.c (try_combine): Don't split an I2 if one of the dests is
3942 set again before I3. Allow unused dests.
3944 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3946 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3948 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3950 PR rtl-optimization/80818
3951 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3952 recursively. Use insn code for clobber.
3953 (lra_set_insn_recog_data): Pass the new arg to
3954 collect_non_operand_hard_regs.
3955 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3957 (lra_update_insn_regno_info): Pass insn to
3958 add_regs_to_insn_regno_info.
3960 2017-11-29 Jim Wilson <jimw@sifive.com>
3961 Andrew Waterman <andrew@sifive.com>
3963 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3964 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3965 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3966 use SINGLE_SHIFT_COST.
3967 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3968 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3970 2017-11-29 Julia Koval <julia.koval@intel.com>
3972 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3973 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3974 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3975 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3976 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3977 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3978 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3979 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3980 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3981 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3982 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3983 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3984 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3985 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3986 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3987 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3988 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3989 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3990 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3991 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3992 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3993 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3994 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3995 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3996 __builtin_ia32_vpshldv_v2di_mask,
3997 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3998 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3999 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4001 2017-11-29 Julia Koval <julia.koval@intel.com>
4003 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4004 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4005 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4006 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4007 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4008 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4009 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4010 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4011 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4012 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4013 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4014 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4015 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4016 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4017 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4018 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4019 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4020 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4021 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4022 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4023 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4025 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4027 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4028 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4031 2017-11-29 Martin Aberg <maberg@gaisler.com>
4033 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4034 to prevent b2bst errata sequence.
4035 (sqrtdf2_fix): Likewise.
4037 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4039 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4040 (div_sqrt_insn_p): New function.
4041 (sparc_do_work_around_errata): Insert NOP instructions to
4042 prevent sequences that could trigger the TN-0013 errata for
4043 certain LEON3 processors.
4044 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4045 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4046 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4047 (in_branch_delay): Prevent div and sqrt in delay slot if
4049 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4051 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4053 * config/sparc/sparc.c (atomic_insn_p): New function.
4054 (sparc_do_work_around_errata): Insert NOP instructions to
4055 prevent sequences that could trigger the TN-0010 errata for
4057 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4058 instruction referable in atomic_insns_p.
4060 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4062 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4063 (atomic_compare_and_swap_leon3_1): Likewise.
4066 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4068 * config/sparc/sparc.c (fpop_insn_p): New function.
4069 (sparc_do_work_around_errata): Insert NOP instructions to
4070 prevent sequences that could trigger the TN-0012 errata for
4072 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4073 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4074 (in_branch_annul_delay): Prevent floating-point instructions
4075 in delay slot of annulled integer branch.
4077 2017-11-29 Richard Biener <rguenther@suse.de>
4079 PR tree-optimization/83202
4080 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4082 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4083 nodes vectorized to the same stmts multiple times.
4084 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4085 (vect_analyze_slp_instance): Adjust.
4086 (scalar_stmts_to_slp_tree_map_t): New typedef.
4087 (vect_schedule_slp_instance): Add a map recording the SLP node
4088 representing the vectorized stmts for a set of scalar stmts.
4089 Avoid code-generating redundancies.
4090 (vect_schedule_slp): Allocate map and pass it down.
4092 2017-11-29 Nathan Sidwell <nathan@acm.org>
4095 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4098 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4100 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4101 cases for printing LDP/STP memory addresses.
4102 (aarch64_print_address_internal): Renamed from
4103 aarch64_print_operand_address, added parameter, add Pmode check.
4104 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4105 (aarch64_print_operand_address): Indirect to
4106 aarch64_print_address_internal.
4107 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4108 'y' operand output specifier.
4110 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4113 * tree.c (build_simple_mem_ref_loc): Handle
4114 get_addr_base_and_unit_offset returning a MEM_REF.
4117 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4120 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4123 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4125 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4127 (layout::move_to_column): Likewise.
4129 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4132 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4133 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4135 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4136 Martin Jambor <mjambor@suse.cz>
4139 * tree.h (expr_type_first_operand_type_p): Declare
4140 * tree.c (expr_type_first_operand_type_p): New function.
4141 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4142 (ipa_value_from_jfunc): Adjust declaration.
4143 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4144 Use it as result type for arithmetics, unless it is NULL in which case
4145 be more conservative.
4146 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4147 ipa_get_jf_pass_through_result.
4148 (propagate_vals_across_pass_through): Likewise.
4149 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4150 is to propagate_vals_across_pass_through.
4151 (propagate_constants_across_call): Pass PARM_TYPE to
4152 propagate_scalar_across_jump_function.
4153 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4154 ipa_value_from_jfunc.
4155 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4156 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4157 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4158 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4159 target_type, pass it to ipa_value_from_jfunc.
4160 (update_indirect_edges_after_inlining): Pass parameter type to
4161 try_make_edge_direct_simple_call.
4163 2017-11-28 Jeff Law <law@redhat.com>
4165 * gimple-ssa-evrp-analyze.c
4166 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4167 refine ranges if scev_initialized_p returns true.
4168 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4170 2017-11-28 Julia Koval <julia.koval@intel.com>
4172 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4173 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4174 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4175 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4176 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4177 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4178 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4179 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4180 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4181 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4182 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4183 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4184 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4185 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4186 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4187 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4188 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4189 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4190 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4191 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4192 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4193 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4195 2017-11-28 Julia Koval <julia.koval@intel.com>
4197 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4198 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4199 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4200 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4201 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4202 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4203 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4204 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4205 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4206 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4207 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4208 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4209 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4210 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4211 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4212 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4213 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4214 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4215 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4216 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4217 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4218 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4219 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4220 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4221 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4222 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4223 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4224 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4225 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4226 __builtin_ia32_vpshld_v2di_mask): New builtins.
4227 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4228 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4230 2017-11-28 Richard Biener <rguenther@suse.de>
4232 PR tree-optimization/80776
4233 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4235 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4237 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4238 If the incoming edge is an effective fallthru because the other
4239 edge only reaches a __builtin_unreachable () then record ranges
4240 derived from the controlling condition in SSA info.
4241 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4242 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4244 2017-11-28 Olivier Hainque <hainque@adacore.com>
4246 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4247 on mingw build hosts.
4249 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4252 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4254 2017-11-28 Richard Biener <rguenther@suse.de>
4257 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4258 copies generated from memcpy use a character array as reference
4261 2017-11-28 Julia Koval <julia.koval@intel.com>
4262 Sebastian Peryt <sebastian.peryt@intel.com>
4264 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4265 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4266 cilk-common.o, cilk.h, cilk-common.c): Remove.
4268 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4269 * builtins.c (is_builtin_name): Remove cilkplus condition.
4270 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4271 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4272 cilk-builtins.def, cilkplus.def): Remove.
4273 * cif-code.def (CILK_SPAWN): Remove.
4274 * cilk-builtins.def: Delete.
4275 * cilk-common.c: Ditto.
4277 * cilkplus.def: Ditto.
4278 * config/darwin.h (fcilkplus): Delete.
4279 * cppbuiltin.c: Ditto.
4280 * doc/extend.texi: Remove cilkplus doc.
4281 * doc/generic.texi: Ditto.
4282 * doc/invoke.texi: Ditto.
4283 * doc/passes.texi: Ditto.
4284 * gcc.c (fcilkplus): Remove.
4285 * gengtype.c (cilk.h): Remove.
4286 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4288 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4290 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4292 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4293 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4294 cilkplus conditions.
4295 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4296 inline_read_section): Ditto.
4297 * ipa-inline-analysis.c (cilk.h): Remove.
4298 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4299 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4300 append_linker_options): Remove condition for fcilkplus.
4301 * lto/lto-lang.c (cilk.h): Remove.
4302 (lto_init): Remove condition for fcilkplus.
4303 * omp-expand.c (expand_cilk_for_call): Delete.
4304 (expand_omp_taskreg, expand_omp_for_static_chunk,
4305 expand_omp_for): Remove cilkplus
4307 (expand_cilk_for): Delete.
4308 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4309 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4310 execute_lower_omp, diagnose_sb_0): Ditto.
4311 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4312 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4313 * tree-nested.c: Ditto.
4314 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4315 (dump_generic_node): Ditto.
4316 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4317 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4319 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4321 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4323 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4324 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4325 for vector mode and !TARGET_SIMD.
4327 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4329 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4331 * tree.h (SWITCH_LABELS): Remove.
4332 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4333 assert SWITCH_BODY is non-NULL.
4334 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4336 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4338 PR tree-optimization/80788
4339 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4340 has TREE_OVERFLOW set, call drop_tree_overflow.
4342 2017-11-28 Richard Biener <rguenther@suse.de>
4344 PR tree-optimization/83158
4345 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4347 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4350 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4351 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4353 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4355 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4356 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4358 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4361 * varasm.c (compare_constant): Take the mode of the constants into
4362 account when comparing floating point constants.
4364 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4366 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4367 from explicit instantiation of debug_helper.
4368 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4370 2017-11-27 Richard Biener <rguenther@suse.de>
4372 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4375 2017-11-27 Richard Biener <rguenther@suse.de>
4377 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4378 (type_hash_canon): Also clear int_cst_hash_table entry for
4380 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4382 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4384 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4385 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4386 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4388 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4390 * hash-map.h (gt_cleare_cache): Avoid UB.
4392 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4394 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4395 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4396 (decide_unroll_runtime_iterations): Likewise.
4397 (decide_unroll_stupid): Likewise.
4399 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4402 * config/i386/i386.md: Add a loop with incssp.
4404 2017-11-27 Martin Jambor <mjambor@suse.cz>
4406 PR tree-optimization/81248
4407 * tree-sra.c (splice_param_accesses): Remove size check.
4408 (decide_one_param_reduction): Fix size check.
4409 * gimple-pretty-print.c (dump_profile): Silence warning.
4410 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4412 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4415 * dbxout.c (lastlineno): New variable.
4416 (dbx_debug_hooks): Use dbxout_switch_text_section as
4417 switch_text_section debug hook.
4418 (dbxout_function_end): Switch to current_function_section
4419 rather than function_section. If crtl->has_bb_partition,
4420 output just one N_FUN, depending on in_cold_section_p.
4421 (dbxout_source_line): Remember last lineno in lastlineno.
4422 (dbxout_switch_text_section): New function.
4423 (dbxout_function_decl): Adjust dbxout_block caller.
4424 (dbx_block_with_cold_children): New function.
4425 (dbxout_block): Return true if any LBRAC/RBRAC have been
4426 emitted. Use dbx_block_with_cold_children at depth == 0
4427 in second partition. Add PARENT_BLOCKNUM argument, pass
4428 it optionally adjusted to children. Output LBRAC/RBRAC
4429 around recursive call only if the block is in the current
4430 partition, if not and anything was output, emit empty
4432 * final.c (final_scan_insn): Compute cold_function_name
4433 before calling switch_text_section debug hook. Call
4434 that hook even if dwarf2out_do_frame if not emitting
4438 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4439 TREE_READONLY decls.
4441 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4443 PR rtl-optimization/82488
4444 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4446 2017-11-26 Julia Koval <julia.koval@intel.com>
4448 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4450 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4453 2017-11-26 Julia Koval <julia.koval@intel.com>
4455 * config/i386/driver-i386.c (host_detect_local_cpu):
4456 Detect skylake-avx512.
4458 2017-11-26 Julia Koval <julia.koval@intel.com>
4460 * config.gcc: Add -march=cannonlake.
4461 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4462 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4463 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4464 (PTA_CANNONLAKE): New.
4465 (processor_target_table): Add cannonlake.
4466 (ix86_option_override_internal): Ditto.
4467 (fold_builtin_cpu): Ditto.
4468 (get_builtin_code_for_version): Handle cannonlake.
4469 (M_INTEL_COREI7_CANNONLAKE): New.
4470 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4471 * doc/invoke.texi: Add -march=cannonlake.
4473 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4475 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4476 (try_init_one_plugin): Alternative implementation for MinGW.
4477 * Makefile.in (plugin_implib): New.
4478 (gengtype-lex.c): Fix broken AIX workaround.
4479 * configure: Regenerate.
4480 * doc/plugins.texi: Document support for MinGW.
4482 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4484 PR rtl-optimization/81553
4485 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4486 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4487 is a shift where C1 has different mode than the whole shift, use C1's
4488 mode for MULT rather than the shift's mode.
4491 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4492 builtins not enabled in the currently selected ISA.
4494 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4496 PR tree-optimization/71026
4497 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4498 (insert_reciprocals): Change to insert reciprocals before a division
4499 by a square and to insert the square of a reciprocal.
4500 (execute_cse_reciprocals_1): Change to consider division by a square.
4501 (register_division_in): Add importance parameter.
4503 2017-11-24 Richard Biener <rguenther@suse.de>
4505 PR tree-optimization/82402
4506 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4507 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4509 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4511 * match.pd (0-ptr): New transformation.
4513 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4516 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4517 when skipping checking.
4519 2017-11-24 Richard Biener <rguenther@suse.de>
4521 PR tree-optimization/83128
4522 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4523 (vn_reference_lookup_3): Likewise.
4525 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4528 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4529 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4532 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4533 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4534 check for the latter with gimple_call_builtin_p. Do not handle
4535 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4537 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4539 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4542 2017-11-23 Julia Koval <julia.koval@intel.com>
4544 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4545 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4546 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4547 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4548 _mm512_maskz_expandloadu_epi16): New intrinsics.
4549 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4550 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4551 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4552 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4553 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4554 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4555 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4556 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4557 _mm256_maskz_expandloadu_epi8): New intrinsics.
4558 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4559 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4560 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4561 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4562 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4563 * config/i386/sse.md (VI248_VLBW): New iterator.
4564 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4566 2017-11-23 Julia Koval <julia.koval@intel.com>
4568 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4569 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4570 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4571 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4572 _mm512_mask_compressstoreu_epi16): New.
4573 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4574 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4575 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4576 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4577 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4578 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4579 _mm256_mask_compressstoreu_epi8): New.
4580 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4581 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4582 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4583 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4584 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4585 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4586 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4587 __builtin_ia32_compresshi128_mask,
4588 __builtin_ia32_compressstoreuqi512_mask,
4589 __builtin_ia32_compressstoreuhi512_mask,
4590 __builtin_ia32_compressstoreuqi256_mask,
4591 __builtin_ia32_compressstoreuqi128_mask,
4592 __builtin_ia32_compressstoreuhi256_mask,
4593 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4594 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4596 (ix86_expand_special_args_builtin): Handle new types.
4597 (s4fma_expand): Handle new builtin array.
4598 * config/i386/immintrin.h: Include new headers.
4599 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4600 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4602 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4604 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4605 fixes. Declare temp and g variables at the top in order to avoid
4606 {} in most of the cases.
4608 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4610 * match.pd (ptr-0): New transformation.
4612 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4614 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4615 (struct addr_mode_cost_table): New.
4616 (struct tune_params): Add field addr_mode_costs.
4617 * config/arm/arm.c (generic_addr_mode_costs): New.
4618 (arm_slowmul_tune): Initialise addr_mode_costs field.
4619 (arm_fastmul_tune): Likewise.
4620 (arm_strongarm_tune): Likewise.
4621 (arm_xscale_tune): Likewise.
4622 (arm_9e_tune): Likewise.
4623 (arm_marvell_pj4_tune): Likewise.
4624 (arm_v6t2_tune): Likewise.
4625 (arm_cortex_tune): Likewise.
4626 (arm_cortex_a8_tune): Likewise.
4627 (arm_cortex_a7_tune): Likewise.
4628 (arm_cortex_a15_tune): Likewise.
4629 (arm_cortex_a35_tune): Likewise.
4630 (arm_cortex_a53_tune): Likewise.
4631 (arm_cortex_a57_tune): Likewise.
4632 (arm_exynosm1_tune): Likewise.
4633 (arm_xgene1_tune): Likewise.
4634 (arm_cortex_a5_tune): Likewise.
4635 (arm_cortex_a9_tune): Likewise.
4636 (arm_cortex_a12_tune): Likewise.
4637 (arm_cortex_a73_tune): Likewise.
4638 (arm_v7m_tune): Likewise.
4639 (arm_cortex_m7_tune): Likewise.
4640 (arm_v6m_tune): Likewise.
4641 (arm_fa726te_tune): Likewise.
4642 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4644 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4646 * config/arm/arm.c (arm_mem_costs): New function.
4647 (arm_rtx_costs_internal): Use arm_mem_costs.
4649 2017-11-23 Mark Wielaard <mark@klomp.org>
4651 * dwarf2out.c (init_sections_and_labels): Use generation to create
4652 unique ranges_section_label and ranges_base_label. Return generation.
4653 (output_rnglists): Add generation argument. Use generation to create
4654 unique ranges labels.
4655 (dwarf2out_finish): Get generation from init_sections_and_labels
4656 and pass generation to output_rnglists.
4658 2017-11-23 Mike Stump <mikestump@comcast.net>
4659 Eric Botcazou <ebotcazou@adacore.com>
4661 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4662 * cfgloop.h (struct loop): Add unroll field.
4663 * function.h (struct function): Add has_unroll bitfield.
4664 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4665 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4666 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4667 (pass_rtl_unroll_loops::gate): Likewise.
4668 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4669 for which loop->unroll==1.
4670 (decide_unroll_constant_iterations): Use note for consistency and
4671 take loop->unroll into account. Return early if loop->unroll is set.
4672 Fix thinko in existing test.
4673 (decide_unroll_runtime_iterations): Use note for consistency and
4674 take loop->unroll into account.
4675 (decide_unroll_stupid): Likewise.
4676 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4677 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4678 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4680 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4681 (print_loop): Print loop->unroll if set.
4682 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4683 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4684 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4686 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4687 loop->unroll is set and smaller than the trip count. Otherwise bypass
4688 entirely the heuristics if loop->unroll is set. Remove dead note.
4689 Fix off-by-one bug in other note.
4690 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4691 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4693 (tree_unroll_loops_completely): Make static.
4694 (pass_complete_unroll::execute): Use correct type for variable.
4695 (pass_complete_unrolli::execute): Fix formatting.
4696 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4698 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4700 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4701 enable when TARGET_PREFER_AVX128 is set.
4703 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4705 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4707 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4708 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4710 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4712 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4713 frequencies. Fix estimation of aggregate parameters.
4715 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4717 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4720 2017-11-23 Tom de Vries <tom@codesourcery.com>
4722 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4724 2017-11-23 Tom de Vries <tom@codesourcery.com>
4726 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4729 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4732 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4733 sibcall_value_pcrel_fdpic): Use local variable instead of
4735 (calli_tbr_rel): Add missing operand 2.
4736 (call_valuei_tbr_rel): Add missing operand 3.
4738 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4741 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4742 bitpos/bitsize covering the whole destination, use store_expr only if
4743 the complex mode is the same. Otherwise, use expand_normal and if
4744 it returns CONCAT, subreg each part separately instead of trying to
4745 subreg the whole result.
4747 2017-11-23 Richard Biener <rguenther@suse.de>
4749 PR tree-optimization/23094
4750 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4751 come here from walking over backedges in the first iteration.
4752 (vn_reference_lookup_3): Skip clobbers that store the same value.
4754 2017-11-23 Richard Biener <rguenther@suse.de>
4756 PR tree-optimization/81403
4757 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4758 a block we need a leader relative to.
4759 (phi_translate_1): For nary processing require a leader from
4760 get_representative_for given we run expression simplification
4761 using match-and-simplify. Remove previous fix.
4763 2017-11-22 Jeff Law <law@redhat.com>
4765 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4766 Use new method allocate_value_range rather than accessing the
4767 vrp_value_range_pool data member directly.
4768 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4769 to use extract_range_from_stmt method to avoid need for
4770 extract_range_from_assignment method.
4771 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4772 method rather than setting values_propgated data member directly.
4773 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4774 and values propagated data members and extract_range_from_assignment
4775 method. Reorder private data members to conform to standards.
4776 Add new methods set_lattice_propagation_complete and
4777 allocate_value_range.
4779 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4781 PR rtl-optimization/83030
4782 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4783 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4784 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4785 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4786 CROSSING_JUMP_P flag is already set before setting it.
4787 * cfgrtl.c (fixup_partition_crossing): Likewise.
4788 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4791 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4793 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4794 Handle the case where both arguments are using gen_const_vec_series.
4796 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4799 * pretty-print.c (pp_format): Move quoting implementation to
4800 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4801 to pass address of "quote" local.
4802 (pp_begin_quote): New function.
4803 (pp_end_quote): New function.
4804 * pretty-print.h (printer_fn): Convert penultimate param from bool
4806 (pp_begin_quote): New decl.
4807 (pp_end_quote): New decl.
4808 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4809 param from bool to bool *.
4810 * tree-diagnostic.h (default_tree_printer): Likewise.
4812 2017-11-22 Jeff Law <law@redhat.com>
4814 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4815 Initialize vr_values.
4816 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4817 vr_values via vr_values class instance rather than delegators.
4818 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4819 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4820 (evrp_range_analyzer::push_value_range): Likewise.
4821 (evrp_range_analyzer::pop_value_range): Likewise.
4822 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4823 most delegators. Those remaining are exposed as public interfaces.
4824 Make vr_values a pointer and private.
4825 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4827 (evrp_range_analyzer::get_vr_value): New method.
4828 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4829 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4830 (class evrp_dom_walker): Attach evrp_folder class, initialize
4831 it in the ctor. Remove temporary delegators.
4832 (evrp_dom_walker::before_dom_children): Call methods in attached
4833 evrp_range_analyzer class via class instance pointer. Use
4834 free value_range_constant_singleton to remove need for
4835 op_with_constant_singleton_value delegator method. Do not
4836 create a vrp_prop class instance for every call! Narrow
4837 scope of a couple variables.
4838 (evrp_dom_walker::cleanup): Call methods in attached
4839 evrp_range_analyzer class via class instance pointer.
4840 * vr-values.h (class vr_values): Privatize many methods and
4843 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4845 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4846 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4847 * expr.c (expand_expr_real_2): Likewise.
4848 * fold-const.c (const_unop): Likewise.
4849 * optabs-tree.c (optab_for_tree_code): Likewise.
4850 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4851 * tree-inline.c (estimate_operator_cost): Likewise.
4852 * tree-pretty-print.c (dump_generic_node): Likewise.
4853 (op_code_prio): Likewise.
4854 (op_symbol_code): Likewise.
4855 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4856 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4857 * internal-fn.c (direct_internal_fn_optab): New function.
4858 (direct_internal_fn_array, direct_internal_fn_supported_p
4859 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4860 * fold-const-call.c (fold_const_reduction): New function.
4861 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4863 * tree-vect-loop.c: Include internal-fn.h.
4864 (reduction_code_for_scalar_code): Rename to...
4865 (reduction_fn_for_scalar_code): ...this and return an internal
4867 (vect_model_reduction_cost): Take an internal_fn rather than
4869 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4871 (vectorizable_reduction): Use internal functions rather than tree
4872 codes for the reduction operation. Update calls to the functions
4874 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4875 Use calls to internal functions rather than REDUC tree codes.
4876 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4878 2017-11-22 Olivier Hainque <hainque@adacore.com>
4880 * config/vxworks.c (vxworks_override_options): Pick default
4881 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4882 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4883 DWARF_GNAT_ENCODINGS_DEFAULT.
4884 * config/vxworksae.h: Likewise.
4886 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4888 PR tree-optimization/83104
4889 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4892 2017-11-22 Marek Polacek <polacek@redhat.com>
4893 H.J. Lu <hongjiu.lu@intel.com>
4894 Jason Merrill <jason@redhat.com>
4899 * calls.c (initialize_argument_information): Call
4900 warn_parameter_passing_abi target hook.
4901 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4902 argument onto stack.
4903 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4904 * common.opt: Update -fabi-version description.
4905 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4906 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4908 (ix86_is_empty_record): New function.
4909 (ix86_warn_parameter_passing_abi): New function.
4910 (TARGET_EMPTY_RECORD_P): Redefine.
4911 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4912 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4913 * doc/tm.texi: Regenerated.
4914 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4915 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4916 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4917 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4918 instead of int_size_in_bytes.
4919 * expr.c (copy_blkmode_to_reg): Likewise.
4920 * function.c (aggregate_value_p): Return 0 for empty types.
4921 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4923 (locate_and_pad_parm): Call arg size_in_bytes instead
4925 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4926 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4927 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4929 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4930 (std_gimplify_va_arg_expr): Skip empty records. Call
4931 arg_size_in_bytes instead size_in_bytes.
4932 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4933 * tree-core.h (tree_type_common): Add empty_flag.
4934 (tree_decl_common): Update comments.
4935 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4937 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4938 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4940 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4941 * tree.c (default_is_empty_type): New function.
4942 (default_is_empty_record): New function.
4943 (arg_int_size_in_bytes): New function.
4944 (arg_size_in_bytes): New function.
4945 (get_ultimate_context): New function.
4946 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4947 TRANSLATION_UNIT_WARN_EMPTY_P.
4948 (default_is_empty_record, arg_int_size_in_bytes,
4949 arg_size_in_bytes, get_ultimate_context): Declare.
4951 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4953 * config/arm/arm.c (cmse_clear_registers): New function.
4954 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4955 code by call to cmse_clear_registers.
4956 (cmse_nonsecure_entry_clear_before_return): Likewise.
4958 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4960 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4961 (vdot_s32, vdotq_s32): New.
4962 (vdot_lane_u32, vdotq_lane_u32): New.
4963 (vdot_lane_s32, vdotq_lane_s32): New.
4966 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4969 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4970 for unsigned values with fewer HWIs than the precision.
4971 (test_overflow): New function.
4972 (wide_int_cc_tests): Call it.
4974 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4976 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4977 has the same mode class as Pmode.
4979 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4981 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4982 padding_bits_to_clear_ptr.
4983 (cmse_nonsecure_entry_clear_before_return): Likewise.
4985 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4987 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4988 auto_sbitap instead of integer bitfield to control register needing
4991 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4993 PR tree-optimization/83044
4994 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4995 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4996 that need the upper bound. Subtract offset from
4997 get_addr_base_and_unit_offset only if positive and subtract it
4998 before division by eltsize rather than after it.
5001 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5002 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5003 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5007 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5008 expand_binop with *mul_widen_optab, make sure at least one of the
5009 operands doesn't have VOIDmode.
5012 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5014 PR rtl-optimization/82044
5015 PR tree-optimization/82042
5016 * dse.c (record_store): Check for overflow.
5017 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5018 add_wild_read instead of clear_rhs_from_active_local_stores on
5019 overflow and log it into dump_file.
5021 2017-11-22 Richard Biener <rguenther@suse.de>
5023 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5024 fake edges to exit when looking for a place to insert.
5025 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5027 (insert_into_preds_of_block): Commit edge insertion immediately,
5028 assert that doesn't require new BBs.
5029 (fini_pre): Release expressions.
5030 (pass_pre::execute): Shuffle things around a bit, if the fn
5031 is too large do not compute AVAIL either as this is really the
5034 2017-11-22 Richard Biener <rguenther@suse.de>
5036 PR tree-optimization/83089
5037 * tree-if-conv.c (pass_if_conversion::execute): If anything
5038 changed reset SCEV and free the number of iteration estimates.
5040 2017-11-21 Martin Sebor <msebor@redhat.com>
5042 PR tree-optimization/82945
5043 * calls.h (warn_nonstring_bound): Remove unused function.
5045 2017-11-21 Martin Sebor <msebor@redhat.com>
5047 PR tree-optimization/82945
5048 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5049 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5050 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5052 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5053 * calls.h (get_attr_nonstring_decl): Declare new function.
5054 * doc/extend.texi (attribute nonstring): Update.
5055 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5056 get_attr_nonstring_decl and handle it.
5057 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5058 detection of nul-termination.
5059 (strlen_to_stridx): Change to a pointer.
5060 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5061 (pass_strlen::execute): Same.
5063 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5065 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5066 for the new option -mprefer-vector-width=[none|128|256|512].
5067 * config/i386/i386.c (ix86_target_string): remove old style options
5068 -mprefer-avx256 and make -mprefer-avx128 as alias.
5069 (ix86_option_override_internal): Apply defaults for the
5070 -mprefer-vector-width=[128|256] option.
5071 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5072 Implement macros to work with -mprefer-vector-width=.
5073 * config/i386/i386.opt: Implemented option
5074 -mprefer-vector-width=[none|128|256|512].
5075 * doc/invoke.texi: Documentation for
5076 -mprefer-vector-width=[none|128|256|512].
5078 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5080 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5082 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5084 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5086 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5088 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5090 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5091 * expr.c (expand_expr_real_2): Likewise.
5092 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5093 fold_binary_loc): Likewise.
5094 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5095 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5096 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5097 MINUS_EXPR transformations.
5098 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5099 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5100 * tree-inline.c (estimate_operator_cost): Likewise.
5101 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5102 op_symbol_code): Likewise.
5103 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5104 * vr-values.c (extract_range_from_binary_expr): Likewise.
5105 * varasm.c (initializer_constant_valid_p_1): Likewise.
5106 * tree.def: New tree code POINTER_DIFF_EXPR.
5108 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5110 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5111 integer suffix to movbe mnemonic.
5112 (*bswaphi2_movbe): Ditto.
5113 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5115 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5118 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5119 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5120 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5122 2017-11-21 Martin Liska <mliska@suse.cz>
5124 * tree-inline.c (expand_call_inline): Remove not needed
5127 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5128 Jakub Jelinek <jakub@redhat.com>
5131 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5132 Remove static keyword from f variable.
5134 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5136 PR tree-optimization/83086
5137 * gimple-ssa-store-merging.c
5138 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5139 rather than n.base_addr.
5141 2017-11-21 Martin Liska <mliska@suse.cz>
5143 PR rtl-optimization/82044
5144 PR tree-optimization/82042
5145 * dse.c (check_mem_read_rtx): Check for overflow.
5147 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5149 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5152 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5154 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5155 and its corresponding call arg location note.
5157 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5158 Andrew Burgess <andrew.burgess@embecosm.com>
5160 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5162 (arc_return_slot_offset): Likewise.
5163 (arc_eh_return_address_location): New declaration.
5164 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5165 (MUST_SAVE_REGISTER): Add exception handler case.
5166 (MUST_SAVE_RETURN_ADDR): Likewise.
5167 (arc_frame_pointer_required): Likewise.
5168 (arc_frame_pointer_needed): New function.
5169 (arc_compute_frame_size): Changed.
5170 (arc_expand_prologue): Likewise.
5171 (arc_expand_epilogue): Likewise.
5172 (arc_initial_elimination_offset): Likewise.
5173 (arc_return_slot_offset): Delete.
5174 (arc_eh_return_address_location): New function.
5175 (arc_builtin_setjmp_frame_value): Likewise.
5176 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5177 (EH_RETURN_STACKADJ_RTX): Define.
5178 (EH_RETURN_HANDLER_RTX): Likewise.
5179 * config/arc/arc.md (eh_return): Delete.
5181 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5183 * print-tree.h (debug_vec_tree): Remove prototype.
5184 * gdbinit.in (pvt): Remove macro.
5186 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5188 PR tree-optimization/83047
5189 * gimple-ssa-store-merging.c
5190 (imm_store_chain_info::output_merged_store): If the loads with the
5191 same vuse are in different basic blocks, for load_gsi pick a load
5192 location that is dominated by the other loads.
5195 * config/i386/i386.c (ix86_memmodel_check): Start
5196 -Winvalid-memory-model diagnostics with lowercase letter.
5199 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5200 set in_cold_section_p to the partition containing loc_list->first.
5201 When seeing loc_list->last_before_switch node, update secname and
5202 perform range_across_switch second partition handling only after that.
5205 * run-rtl-passes.c: Include debug.h.
5206 (run_rtl_passes): Call debug_hooks->assembly_start.
5207 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5211 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5212 OPTAB_DIRECT in calls to expand_simple_binop.
5214 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5217 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5219 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5222 * spellcheck.h (best_match::blithely_get_best_candidate): New
5225 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5227 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5228 with lower case letter.
5230 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5232 * config/i386/i386.md (bswaphi2): New expander.
5233 (*bswaphi2_movbe): New insn pattern.
5234 (bswaphi -> rorhi pepehole2): New peephole pattern.
5236 2017-11-20 Jeff Law <law@redhat.com>
5238 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5239 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5240 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5241 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5242 gimple-ssa-evrp-analyze.h.
5244 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5245 set BB_VISITED here.
5246 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5248 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5250 * tree-predcom.c: Add general comment on Store-Store chains.
5251 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5253 (get_chain_last_ref_at): Rename into...
5254 (get_chain_last_write_at): ...this.
5255 (get_chain_last_write_before_load): New function.
5256 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5257 CT_STORE_STORE when write reference is added.
5258 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5259 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5260 (initialize_root_vars_store_elim_1): Ditto.
5261 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5262 definition is created.
5263 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5264 chain by replacing it with dominant stored value.
5266 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5268 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5270 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5272 * vr-values.c (extract_range_from_binary_expr): Use a full range
5275 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5277 * config/arm/arm.md (R4_REGNUM): Define constant.
5278 (nonsecure_call_internal): Remove r4 clobber.
5279 (nonsecure_call_value_internal): Likewise.
5280 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5281 clobber and resequence match_operands.
5282 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5283 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5284 (nonsecure_call_value_reg_thumb2): Likewise.
5286 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5288 PR tree-optimization/78821
5289 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5290 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5291 to the address of the base rather than the base itself.
5292 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5293 (find_bswap_or_nop_finalize): New function.
5294 (find_bswap_or_nop): Use it.
5295 (bswap_replace): Return a tree rather than bool, change first
5296 argument from gimple * to gimple_stmt_iterator, allow inserting
5297 into an empty sequence, allow ins_stmt to be NULL - then emit
5298 all stmts into gsi. Fix up MEM_REF address gimplification.
5299 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5300 (struct store_immediate_info): Add N and INS_STMT non-static
5302 (store_immediate_info::store_immediate_info): Initialize them
5303 from newly added ctor args.
5304 (merged_store_group::apply_stores): Formatting fixes. Sort by
5306 (stmts_may_clobber_ref_p): For stores call also
5307 refs_anti_dependent_p.
5308 (gather_bswap_load_refs): New function.
5309 (imm_store_chain_info::try_coalesce_bswap): New method.
5310 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5311 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5312 (imm_store_chain_info::output_merged_store): Fail if number of
5313 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5314 and NOP_EXPR rhs_code.
5315 (pass_store_merging::process_store): Compute n and ins_stmt, if
5316 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5317 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5319 (pass_store_merging::execute): Calculate dominators.
5321 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5322 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5323 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5324 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5325 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5326 class pass_optimize_bswap, bswap_replace,
5327 pass_optimize_bswap::execute): Moved to ...
5328 * gimple-ssa-store-merging.c: ... this file.
5329 Include optabs-tree.h.
5330 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5331 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5332 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5333 anonymous namespace, remove static keywords.
5334 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5335 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5337 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5340 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5342 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5344 * vec.h (debug_helper): New function.
5345 (DEFINE_DEBUG_VEC): New macro.
5346 * hash-set.h (debug_helper): New function.
5347 (DEFINE_DEBUG_HASH_SET): New macro.
5348 * cfg.c (debug_slim (edge)): New function.
5349 Call DEFINE_DEBUG_VEC for edges.
5350 Call DEFINE_DEBUG_HASH_SET for edges.
5351 * cfghooks.c (debug_slim (basic_block)): New function.
5352 Call DEFINE_DEBUG_VEC for basic blocks.
5353 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5354 * print-tree.c (debug_slim): New function to handle trees.
5355 Call DEFINE_DEBUG_VEC for trees.
5356 Call DEFINE_DEBUG_HASH_SET for trees.
5357 (debug (vec<tree, va_gc>) &): Remove.
5358 (debug (<vec<tree, va_gc>) *): Remove.
5359 * print-rtl.c (debug_slim): New function to handle const_rtx.
5360 Call DEFINE_DEBUG_VEC for rtx_def.
5361 Call DEFINE_DEBUG_VEC for rtx_insn.
5362 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5363 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5364 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5365 (debug (vec<rtx_insn *> *ptr): Remove.
5366 (debug_insn_vector): Remove.
5367 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5369 2017-11-20 Tom de Vries <tom@codesourcery.com>
5371 PR rtl-optimization/82020
5372 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5373 IF_THEN_ELSE condition.
5375 2017-11-19 Jeff Law <law@redhat.com>
5377 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5378 of degenerates resulting from ignoring an edge.
5380 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5383 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5385 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5388 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5389 for uninitialied counts.
5391 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5394 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5396 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5399 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5401 2017-11-19 Tom de Vries <tom@codesourcery.com>
5403 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5406 2017-11-19 Tom de Vries <tom@codesourcery.com>
5408 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5411 2017-11-19 Tom de Vries <tom@codesourcery.com>
5413 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5416 2017-11-19 Tom de Vries <tom@codesourcery.com>
5418 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5420 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5422 2017-11-19 Tom de Vries <tom@codesourcery.com>
5424 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5426 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5427 semicolon after MCORE_EXPORT_NAME call.
5429 2017-11-19 Tom de Vries <tom@codesourcery.com>
5432 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5433 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5435 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5437 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5438 (cgraph_edge::clone): Cleanup updating of profile.
5439 * ipa-cp.c (update_profiling_info): Likewise.
5440 * ipa-inline-transform.c (inline_transform): Likewise.
5441 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5442 * ipa-split.c (execute_split_functions): Do not split when function
5444 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5445 * profile-count.c (profile_count::dump): Add global0.
5446 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5448 (profile_count::to_sreal_scale): Likewise.
5449 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5450 (profile_count::combine_with_ipa_count): New function.
5451 * profile-count.h (profile_guessed_global0adjusted): New.
5452 (profile_count::adjusted_zero): New.
5453 (profile_count::global0adjusted): New.
5454 (profile_count::combine_with_ipa_count): New.
5455 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5456 correct profile of return block of split functions.
5457 (copy_cfg_body): Remove unused profile_count.
5458 (copy_body): Likewise.
5459 (expand_call_inline): Update.
5460 (tree_function_versioning): Update.
5462 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5464 * hash-set.h (hash_set::empty): New.
5465 * tree-ssa-threadbackward.h: Delete.
5466 * tree-ssa-threadbackward.c (class thread_jumps): New.
5467 Move max_threaded_paths into class.
5468 (fsm_find_thread_path): Remove arguments that are now in class.
5469 (profitable_jump_thread_path): Rename to...
5470 (thread_jumps::profitable_jump_thread_path): ...this.
5471 (convert_and_register_jump_thread_path): Rename to...
5472 (thread_jumps::convert_and_register_current_path): ...this.
5473 (check_subpath_and_update_thread_path): Rename to...
5474 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5475 (register_jump_thread_path_if_profitable): Rename to...
5476 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5477 (handle_phi): Rename to...
5478 (thread_jumps::handle_phi): ...this.
5479 (handle_assignment): Rename to...
5480 (thread_jumps::handle_assignment): ...this.
5481 (fsm_find_control_statement_thread_paths): Rename to...
5482 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5483 (find_jump_threads_backwards): Rename to...
5484 (thread_jumps::find_jump_threads_backwards): ...this.
5485 Initialize path local data.
5486 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5487 from within thread_jumps class.
5488 (pass_early_thread_jumps::execute): Same.
5490 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5492 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5494 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5496 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5497 to canonical location.
5499 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5502 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5504 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5506 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5508 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5509 rcrt1.o%s/grcrt1.o%s for -static-pie.
5511 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5513 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5514 ix86_shift_rotate_cost): Break out from ...
5515 (ix86_rtx_costs): ... here.
5516 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5519 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5521 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5522 when entry block was promoted unlikely.
5523 (estimate_bb_frequencies): Increase frequency scale.
5524 * profile-count.h (profile_count): Export precision info.
5526 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5528 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5529 disturbin profile of entry block.
5531 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5533 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5536 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5538 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5539 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5541 2017-11-17 Jeff Law <law@redhat.com>
5543 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5544 from evrp_dom_walker class. Various methods moved into new class.
5545 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5546 (evrp_range_analyzer::enter): New method.
5547 (evrp_range_analyzer::leave): New method.
5548 (evrp_dom_walker): Remove delegators no longer needed by this class.
5549 Replace vr_values data member with evrp_range_analyzer
5551 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5552 method extracted from evrp_dom_walker::before_dom_children.
5553 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5554 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5556 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5557 Add private copy constructor and move assignment operators.
5558 Privatize methods and class data where trivially possible.
5559 (evrp_dom_walker::cleanup): New function, extracted from
5560 execute_early_vrp. Simplify access to class data.
5562 * vr-values.h (get_output_for_vrp): Prototype.
5563 * vr-values.c (get_output_for_vrp): New function extracted from
5564 vrp_visit_assignment_or_call and extract_range_from_stmt.
5565 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5567 2017-11-17 Luis Machado <luis.machado@linaro.org>
5569 * config/aarch64/aarch64.c
5570 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5571 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5572 tune_params::AUTOPREFETCHER_WEAK.
5574 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5577 * config/arm/arm.c (arm_valid_target_attribute_rec):
5578 Parse "arch=" and "+<ext>".
5579 (arm_valid_target_attribute_tree): Re-init global options.
5580 (arm_option_override): Make non-static.
5581 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5582 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5583 (__ARM_FEATURE_CRC32): Support undef.
5584 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5585 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5587 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5589 * gdbinit.in (break-on-diagnostic): New command.
5591 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5593 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5594 return to indirect jump for EH return if control-flow
5595 protection is enabled. Change explicit 'false' argument in
5596 pro_epilogue_adjust_stack with a value of
5598 * config/i386/i386.md (simple_return_indirect_internal):
5599 Remove SImode restriction to support 64-bit.
5601 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5603 * combine.c (added_notes_insn): New.
5604 (try_combine): Handle added_notes_insn like added_links_insn.
5605 Rewrite return value code.
5606 (distribute_notes): Set added_notes_insn to the earliest insn we added
5609 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5611 PR rtl-optimization/82621
5612 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5613 dest of one of those SETs is unused.
5615 2017-11-17 Richard Biener <rguenther@suse.de>
5618 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5619 * tree-pretty-print.c (dump_generic_node): Handle
5620 annot_expr_parallel_kind.
5621 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5622 * gimplify.c (gimple_boolify): Likewise.
5624 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5626 * config.gcc (extra_headers): Add cet.h for x86 targets.
5627 * config/i386/cet.h: New file.
5628 * doc/install.texi: Add --enable-cet/--disable-cet.
5630 2017-11-17 Richard Biener <rguenther@suse.de>
5632 PR tree-optimization/83017
5633 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5634 (gen_parallel_loop): Properly count iterations.
5635 (parallelize_loops): Handle loop->can_be_parallel independent
5636 of flag_loop_parallelize_all. Make static profitability test match
5638 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5639 * invoke.texi (parloops-min-per-thread): Document.
5641 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5643 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5644 upstreaming review comments.
5646 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5648 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5649 with fast unaligned access.
5650 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5652 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5654 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5655 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5657 2017-11-17 Richard Biener <rguenther@suse.de>
5659 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5660 folding of references.
5662 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5665 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5666 of replacing call to strncmp with corresponding call to strcmp when
5669 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5671 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5672 option prefer-avx256 for skylake-avx512 configuration.
5673 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5674 (get_builtin_code_for_version): Ditto.
5676 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5677 Monk Chiang <sh.chiang04@gmail.com>
5679 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5680 (FIXED_REGISTERS): Reserve more register numbers.
5681 (CALL_USED_REGISTERS): Likewise.
5682 (REG_ALLOC_ORDER): Likewise.
5683 (REG_CLASS_CONTENTS): Likewise.
5684 (REGISTER_NAMES): Likewise.
5686 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5687 Kito Cheng <kito.cheng@gmail.com>
5689 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5691 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5693 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5695 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5696 parameter list for vec_splats.
5698 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5700 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5702 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5704 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5707 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5708 with the required version of automake 1.11.6.
5710 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5712 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5714 (DU_C2_3_power9): Correct reservation combinations.
5715 (FP_DIV_power9, VEC_DIV_power9): New.
5716 (power9-alu): Split out rotate/shift...
5717 (power9-rot): ...to here, correct dispatch resource.
5718 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5720 (power9-fp): Correct latency.
5721 (power9-sdiv): Add div/sqrt resource.
5722 (power9-ddiv): Correct latency, add div/sqrt resource.
5723 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5724 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5726 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5728 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5730 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5731 switch statement mapping KF built-ins to TF built-ins if we don't
5732 have the proper ISA 3.0 assembler support.
5734 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5736 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5737 (lower_emutls_function_body): Do not compute it.
5739 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5741 * ipa-split.c (split_bb_info): Turn time to sreal.
5742 (split_point): Likewise.
5743 (dump_split_point): Likewise.
5744 (fine_split_points): Likewise.
5745 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5747 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5749 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5752 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5754 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5756 (expensive_function_p): Remove useless assert.
5757 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5759 2017-11-16 Martin Sebor <msebor@redhat.com>
5761 PR tree-optimization/82588
5762 PR tree-optimization/82583
5763 * tree-vrp.c (check_array_ref): Handle flexible array members,
5764 string literals, and inner indices.
5765 (search_for_addr_array): Add detail to diagnostics.
5767 2017-11-16 Nathan Sidwell <nathan@acm.org>
5771 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5772 (SET_DECL_ASSEMBLER_NAME): Forward to
5773 overwrite_decl_assembler_name.
5774 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5775 (overwrite_decl_assembler_name): Declare.
5776 * tree.c (overwrite_decl_assembler_name): New.
5777 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5778 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5779 (LANG_HOOKS_INITIALIZER): Add it.
5780 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5781 * langhooks.c (lhd_set_decl_assembler_name): Use
5782 SET_DECL_ASSEMBLER_NAME.
5783 (lhd_overwrite_decl_assembler_name): Default implementation.
5785 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5786 Jackson Woodruff <jackson.woodruff@arm.com>
5788 PR tree-optimization/71026
5789 * match.pd: Canonicalize constant multiplies in division.
5791 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5793 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5794 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5795 Remove OPT_fomit_frame_pointer entry.
5796 * common/config/arc/arc-common.c: Likewise.
5797 * common/config/arm/arm-common.c: Likewise.
5798 * common/config/avr/avr-common.c: Likewise.
5799 * common/config/c6x/c6x-common.c: Likewise.
5800 * common/config/cr16/cr16-common.c: Likewise.
5801 * common/config/cris/cris-common.c: Likewise.
5802 * common/config/epiphany/epiphany-common.c: Likewise.
5803 * common/config/fr30/fr30-common.c: Likewise.
5804 * common/config/frv/frv-common.c: Likewise.
5805 * common/config/ia64/ia64-common.c: Likewise.
5806 * common/config/iq2000/iq2000-common.c: Likewise.
5807 * common/config/lm32/lm32-common.c: Likewise.
5808 * common/config/m32r/m32r-common.c: Likewise.
5809 * common/config/mcore/mcore-common.c: Likewise.
5810 * common/config/microblaze/microblaze-common.c: Likewise.
5811 * common/config/mips/mips-common.c: Likewise.
5812 * common/config/mmix/mmix-common.c: Likewise.
5813 * common/config/mn10300/mn10300-common.c: Likewise.
5814 * common/config/nios2/nios2-common.c: Likewise.
5815 * common/config/pa/pa-common.c: Likewise.
5816 * common/config/pdp11/pdp11-common.c: Likewise.
5817 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5818 * common/config/riscv/riscv-common.c: Likewise.
5819 * common/config/rs6000/rs6000-common.c: Likewise.
5820 * common/config/rx/rx-common.c: Likewise.
5821 * common/config/s390/s390-common.c: Likewise.
5822 * common/config/sh/sh-common.c: Likewise.
5823 * common/config/sparc/sparc-common.c: Likewise.
5824 * common/config/tilegx/tilegx-common.c: Likewise.
5825 * common/config/tilepro/tilepro-common.c: Likewise.
5826 * common/config/v850/v850-common.c: Likewise.
5827 * common/config/visium/visium-common.c: Likewise.
5828 * common/config/xstormy16/xstormy16-common.c: Likewise.
5829 * common/config/xtensa/xtensa-common.c: Likewise.
5830 * invoke.texi (-fomit-frame-pointer): Update documentation.
5832 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5834 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5836 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5838 * cfg.c (scale_bbs_frequencies_int,
5839 cale_bbs_frequencies_gcov_type): Remove.
5840 * cfg.h (scale_bbs_frequencies_int,
5841 cale_bbs_frequencies_gcov_type): Remove.
5843 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5845 * tree-ssa-loop-manip.c
5846 (scale_dominated_blocks_in_loop): Update to profile counts.
5847 (tree_transform_and_unroll_loop): Likewise.
5849 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5851 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5852 scale_bbs_frequencies_int.
5854 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5856 * final.c (compute_alignments): Use counts rather than frequencies.
5858 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5860 * cfgloopanal.c: Include sreal.h
5861 (average_num_loop_insns): Use counts and sreal for accounting.
5863 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5865 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5868 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5869 Kito Cheng <kito.cheng@gmail.com>
5871 * config/nds32/constraints.md: Provide more constraints.
5872 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5873 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5874 support constraints usage.
5876 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5878 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5880 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5881 Kito Cheng <kito.cheng@gmail.com>
5883 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5884 * config/nds32/nds32.opt: Refine the layout.
5885 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5886 TARGET_EXT_STRING): Support new options.
5887 * config/nds32/nds32.h: Likewise.
5888 * config/nds32/nds32.md: Likewise.
5889 * config/nds32/nds32-predicates.c: Likewise.
5890 * config/nds32/constraints.md: Likewise.
5891 * common/config/nds32/nds32-common.c: Likewise.
5893 2017-11-16 Julia Koval <julia.koval@intel.com>
5896 * config/i386/gfniintrin.h: Add sse check.
5897 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5899 2017-11-16 Julia Koval <julia.koval@intel.com>
5901 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5902 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5903 (ix86_handle_option): Handle -mavx512vbmi2.
5904 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5905 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5906 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5907 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5908 (ix86_valid_target_attribute_inner_p): Ditto.
5909 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5910 * config/i386/i386.opt (mavx512vbmi2): New option.
5911 * doc/invoke.texi: Add new option.
5913 2017-11-16 Julia Koval <julia.koval@intel.com>
5915 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5916 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5917 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5918 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5919 _mm512_gf2p8mul_epi8): New intrinsics.
5920 * config/i386/i386-builtin-types.def
5921 (V64QI_FTYPE_V64QI_V64QI): New type.
5922 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5923 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5924 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5925 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5926 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5927 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5929 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5931 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5932 explicitly as a stream of bytes.
5934 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5936 * config/rs6000/altivec.h (vec_xst_be): New #define.
5937 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5938 and externalize from *altivec_vperm_<mode>_internal.
5939 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5941 (XL_BE_V8HI): Likewise.
5942 (XL_BE_V4SI): Likewise.
5943 (XL_BE_V4SI): Likewise.
5944 (XL_BE_V2DI): Likewise.
5945 (XL_BE_V4SF): Likewise.
5946 (XL_BE_V2DF): Likewise.
5947 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5948 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5949 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5950 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5951 VSX_BUILTIN_VEC_XST_BE.
5952 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5953 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5955 (altivec_init_builtins): Replace conditional calls to def_builtin
5956 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5957 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5958 with unconditional calls. Remove calls to def_builtin for
5959 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5960 __builtin_vec_xst_be.
5961 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5962 to define_expand, and add alternate RTL generation for P8.
5963 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5964 vsx_ld_elemrev_v8hi.
5965 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5966 add alternate RTL generation for P8.
5967 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5968 vsx_ld_elemrev_v16qi.
5969 (vsx_st_elemrev_v8hi): Convert define_insn
5970 to define_expand, and add alternate RTL generation for P8.
5971 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5972 vsx_st_elemrev_v8hi.
5973 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5974 add alternate RTL generation for P8.
5975 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5976 vsx_st_elemrev_v16qi.
5978 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5981 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5982 TARGET_AVX512ER check.
5983 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5984 neither -mzeroupper nor -mno-zeroupper is used and
5985 TARGET_EMIT_VZEROUPPER is set.
5986 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5987 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5989 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5991 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5992 folding of vector compares.
5993 (fold_build_vec_cmp): New helper function.
5994 (fold_compare_helper): New helper function.
5995 (builtin_function_type): Add compare builtins to the list of functions
5996 having unsigned arguments. Cosmetic updates to comment indentation.
5997 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5998 the not+eq combination.
6000 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6002 PR tree-optimization/82726
6003 PR tree-optimization/70754
6004 * tree-predcom.c (order_drefs_by_pos): New function.
6005 (combine_chains): Move code setting has_max_use_after to...
6006 (try_combine_chains): ...here. New parameter. Sort combined chains
6007 according to position information.
6008 (tree_predictive_commoning_loop): Update call to above function.
6009 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6011 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6013 PR tree-optimization/82726
6015 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6017 PR tree-optimization/70754
6018 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6019 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6020 combined stmt before it if not NULL.
6021 (combine_chains): Process refs reversely and compute dominance point
6025 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6027 PR tree-optimization/79663
6028 * tree-predcom.c (combine_chains): Process refs in reverse order
6029 only for ZERO length chains, and add explaining comment.
6031 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6033 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6034 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6035 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6037 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6039 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6041 2017-11-15 Martin Liska <mliska@suse.cz>
6043 * tree-cfg.c (pass_warn_function_return::execute):
6044 Compare warn_return_type for greater than zero.
6046 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6050 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6051 to return true on Xeon and not on Xeon Phi.
6052 (ix86_check_avx256_register): Changed to ...
6053 (ix86_check_avx_upper_register): ... this. Add extra check for
6054 VALID_AVX512F_REG_OR_XI_MODE.
6055 (ix86_avx_u128_mode_needed): Changed
6056 ix86_check_avx256_register to ix86_check_avx_upper_register.
6057 (ix86_check_avx256_stores): Changed to ...
6058 (ix86_check_avx_upper_stores): ... this. Changed
6059 ix86_check_avx256_register to ix86_check_avx_upper_register.
6060 (ix86_avx_u128_mode_after): Changed
6061 avx_reg256_found to avx_upper_reg_found. Changed
6062 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6063 (ix86_avx_u128_mode_entry): Changed
6064 ix86_check_avx256_register to ix86_check_avx_upper_register.
6065 (ix86_avx_u128_mode_exit): Ditto.
6066 * config/i386/i386.h: (host_detect_local_cpu): New define.
6068 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6070 * config/arm/xgene1.md (xgene1): Split into automatons
6071 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6072 (xgene1_f_load): Adjust reservations and/or types.
6073 (xgene1_f_store): Likewise.
6074 (xgene1_load_pair): Likewise.
6075 (xgene1_store_pair): Likewise.
6076 (xgene1_fp_load1): Likewise.
6077 (xgene1_load1): Likewise.
6078 (xgene1_store1): Likewise.
6079 (xgene1_move): Likewise.
6080 (xgene1_alu): Likewise.
6081 (xgene1_simd): Likewise.
6082 (xgene1_bfm): Likewise.
6083 (xgene1_neon_load1): Likewise.
6084 (xgene1_neon_store1): Likewise.
6085 (xgene1_neon_logic): Likewise.
6086 (xgene1_neon_st1): Likewise.
6087 (xgene1_neon_ld1r): Likewise.
6088 (xgene1_alu_cond): Added.
6089 (xgene1_shift_reg): Likwise.
6090 (xgene1_bfx): Likewise.
6091 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6093 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6096 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6098 (can_widen_mult_without_libcall): New function.
6099 (expand_mul_overflow): If only checking unsigned mul overflow,
6100 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6101 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6102 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6103 (expand_DIVMOD): Formatting fix.
6104 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6105 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6106 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6108 PR tree-optimization/82977
6109 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6110 constructed temporary to strlen_to_stridx.put.
6112 2017-11-15 Martin Liska <mliska@suse.cz>
6114 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6115 * configure: Regenerate.
6117 2017-11-15 Martin Liska <mliska@suse.cz>
6120 * config/sh/sh-mem.cc: Use proper probability for
6123 2017-11-14 Jeff Law <law@redhat.com>
6125 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6126 the red zone for stack_clash_protection_final_dynamic_probe targets
6127 when the total dynamic stack size is zero bytes.
6129 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6130 blocks is post order.
6132 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6134 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6135 * final.c (rest_of_clean_state): Set it for the
6136 -fcompare-debug dump.
6137 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6138 class when TDF_COMPARE_DEBUG is set.
6140 * dwarf2out.c (gen_producer_string): Discard
6143 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6146 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6147 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6148 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6149 (__TGMATH_REAL_2_3): Remove macros.
6150 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6151 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6153 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6155 (remquo): Define using __TGMATH_REAL_3.
6157 2017-11-14 Jeff Law <law@redhat.com>
6159 * vr-values.c: New file with contents extracted from tree-vrp.c.
6160 * Makefile.in (OBJS): Add vr-values.o
6161 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6162 (set_value_range, set_and_canonicalize_value_range): Likewise.
6163 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6164 (value_range_constant_singleton, symbolic_range_p): Likewise.
6165 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6166 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6167 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6168 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6169 (find_case_label_range, find_case_label_index): Likewise.
6170 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6171 (range_int_cst_singleton_p, value_inside_range): Likewise.
6172 (get_single_symbol): Likewise.
6173 (switch_update): Move structure definition here.
6174 (to_remove_edges, to_update_switch_stmts): Provide externs.
6175 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6176 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6177 (vrp_val_is_min, set_value_range): Likewise.
6178 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6179 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6180 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6181 (range_is_nonnull, range_int_cst_p): Likewwise.
6182 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6183 (get_single_symbol, operand_less_p): Likewise
6184 (compare_values_warnv, compare_values): Likewise.
6185 (value_inside_range, value_range_constant_singleton): Likewise.
6186 (zero_nonzero_bitgs_from_vr): Likewise.
6187 (extract_range_from_binary_expr_1): Likewise.
6188 (overflow_comparison_p): Likewise.
6189 (to_remove_edges, to_update_switch_stmts): Likewise.
6190 (find_case_label-index, find_case_label_range): Likewise.
6191 (switch_update, set_value_range_to_nonnegative): Remove.
6192 (set_value_range_to_truthvalue): Likewise.
6193 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6194 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6195 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6196 (find_case_label_ranges, test_for_singularity): Likewise.
6197 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6198 (x_vr_values): Move to its remaining use site.
6200 2017-11-10 Jeff Law <law@redhat.com>
6202 * vr-values.h (VR_INITIALIZER): Move #define here.
6203 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6204 * Makefile.in (OBJS): Add tree-evrp.o
6205 * tree-vrp.h (assert_info): Move structure definition here.
6206 (set_value_range_to_varying): Prototype.
6207 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6208 (infer_value_range, register_edge_assert_for): Likewise.
6209 (stmt_interesting_for_vrp): Likewise.
6210 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6211 (set_value_range_to_varying): No longer static.
6212 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6213 (infer_value_range, register_edge_assert_for): Likewise.
6215 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6217 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6218 generate the XXBRD instruction.
6220 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6221 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6222 as being compatible if -mabi=ieeelongdouble.
6223 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6224 to setup float128 built-ins with hardware support.
6225 (BU_FLOAT128_HW_2): Likewise.
6226 (BU_FLOAT128_HW_3): Likewise.
6227 (BU_FLOAT128_HW_VSX_1): Likewise.
6228 (BU_FLOAT128_HW_VSX_2): Likewise.
6229 (scalar_extract_expq): Change float128 built-in functions to
6230 accommodate having both KFmode and TFmode functions. Use the
6231 KFmode variant as the default.
6232 (scalar_extract_sigq): Likewise.
6233 (scalar_test_neg_qp): Likewise.
6234 (scalar_insert_exp_q): Likewise.
6235 (scalar_insert_exp_qp): Likewise.
6236 (scalar_test_data_class_qp): Likewise.
6237 (sqrtf128_round_to_odd): Delete processing the round to odd
6238 built-in functions as special built-in functions, and define them
6239 as float128 built-ins. Use the KFmode variant as the default.
6240 (truncf128_round_to_odd): Likewise.
6241 (addf128_round_to_odd): Likewise.
6242 (subf128_round_to_odd): Likewise.
6243 (mulf128_round_to_odd): Likewise.
6244 (divf128_round_to_odd): Likewise.
6245 (fmaf128_round_to_odd): Likewise.
6246 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6247 support for KFmode and TFmode xststdcqp calls.
6248 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6249 point, switch the built-in handlers for the get/set float128
6250 exponent, get float128 mantissa, float128 test built-ins, and the
6251 float128 round to odd built-in functions. Eliminate creating the
6252 float128 round to odd built-in functions as special built-ins.
6253 (rs6000_init_builtins): Eliminate special creation of the float128
6254 round to odd built-in functions.
6255 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6256 function insns to support both TFmode and KFmode variants.
6257 (xsxsigqp_<mode>): Likewise.
6258 (xsiexpqpf_<mode>): Likewise.
6259 (xsiexpqp_<mode>): Likewise.
6260 (xststdcqp_<mode>): Likewise.
6261 (xststdcnegqp_<mode>): Likewise.
6262 (xststdcqp_<mode>): Likewise.
6264 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6266 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6267 unused path_in_freq_ptr parameter.
6268 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6270 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6272 * ipa-inline.c (edge_badness): Dump sreal frequency.
6273 (compute_inlined_call_time): Match natural implementaiton ...
6274 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6275 forgotten division by CGRAPH_FREQ_BASE.
6277 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6279 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6280 Solaris 11. Update comment.
6281 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6283 * config/sol2.h (STARTFILE_SPEC): Likewise.
6284 * configure: Regenerate.
6286 2017-11-14 Carl Love <cel@us.ibm.com>
6288 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6289 le_ and be_ prefixes to swap* variables. Remove
6290 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6293 2017-11-14 Jason Merrill <jason@redhat.com>
6295 Support GTY((cache)) on hash_map.
6296 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6297 (ggc_cache_remove): Override it instead of ggc_mx.
6298 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6299 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6300 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6301 (simple_cache_map_traits): Override maybe_mx.
6302 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6303 (hash_map): Friend gt_cleare_cache.
6304 (gt_cleare_cache): New.
6305 * tree.h (tree_cache_traits): New hash_map traits class.
6306 (tree_cache_map): New typedef.
6308 2017-11-14 Richard Biener <rguenther@suse.de>
6310 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6311 paramter and handling.
6312 (cleanup_control_flow_bb): Likewise.
6313 (cleanup_control_flow_pre): New helper performing a DFS walk
6314 to call cleanup_control_flow_bb in PRE order.
6315 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6316 via cleanup_control_flow_pre.
6318 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6320 * config/aarch64/aarch64-simd.md
6321 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6322 (*aarch64_simd_bsl<mode>_alt): Likewise.
6323 (aarch64_simd_bsldi_internal): New.
6324 (aarch64_simd_bsldi_alt): Likewise.
6326 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6328 * tracer.c (better_p): Do not compare frequencies.
6329 * reg-stack.c (better_edge): Likewise.
6330 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6333 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6335 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6336 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6337 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6338 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6339 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6340 * predict.c (maybe_hot_frequency_p): Inline to ...
6341 (maybe_hot_count_p): ... here; rewrite to counts.
6342 (counts_to_freqs): Rename to ...
6343 (update_max_bb_count): ... this one.
6344 (expensive_function_p): Use counts.
6345 (estimate_bb_frequencies): Update.
6346 (rebuild_frequencies): Update.
6347 * predict.h (counts_to_freqs): Rename to ...
6348 (update_max_bb_count): ... this one.
6349 * profile.c (compute_branch_probabilities): Add debug info
6350 * tree-inline.c (expand_call_inline): Update debug info.
6351 (optimize_inline_calls): Use update_max_bb_count..
6352 (tree_function_versioning): Use update_max_bb_count..
6353 * value-prof.c (gimple_value_profile_transformations):
6354 Do not use update_max_bb_count.
6356 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6358 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6359 always use frequencies.
6361 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6363 * bb-reorder.c: Remove frequencies from comments.
6364 (better_edge_p): Use profile counts.
6365 (find_traces): Dump profile counts.
6366 (rotate_loop): Use profile counts.
6367 (find_traces_1_round): Likewise.
6368 (connect_better_edge_p): Use counts instead of probabilities for
6370 (copy_bb_p): Drop early check for non-0 frequency.
6371 (sanitize_hot_paths): Update comments.
6373 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6375 * ipa-split.c (struct split_point): Add count.
6376 (consider_split): Do not compute incoming frequency; compute incoming
6377 count and store it to split_point.
6378 (split_function): Set count of the call to split part correctly.
6380 2017-11-13 Carl Love <cel@us.ibm.com>
6382 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6384 2017-11-13 Tom Tromey <tom@tromey.com>
6386 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6388 2017-11-13 Carl Love <cel@us.ibm.com>
6390 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6391 Add support for builtins:
6392 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6393 vector {un,}signed {char,int,short},
6394 vector {un,}signed {char,int,short}) arguments.
6395 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6396 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6397 Add BU_P9V_AV_2 expansions for the builtins.
6398 * config/rs6000/altivec.h (vec_first_match_index,
6399 vec_first_mismatch_index, vec_first_match_or_eos_index,
6400 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6401 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6402 new extern declaration.
6403 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6404 * config/rs6000/vsx.md (first_match_index_<mode>,
6405 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6406 first_mismatch_or_eos_index_<mode>): Add define expand.
6407 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6408 * doc/extend.texi: Update the built-in documenation file for the new
6411 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6413 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6414 into the min/max operations for _Float<N> and _Float<N>X types.
6416 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6419 * dwarf2out.c (do_eh_frame): New static variable.
6420 (dwarf2out_begin_prologue): Set it.
6421 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6423 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6425 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6427 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6429 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6432 * debug.h (dwarf2out_do_eh_frame): Declare.
6433 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6434 (dwarf2out_do_frame): Use it.
6435 (dwarf2out_do_cfi_asm): Likewise.
6436 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6437 (dwarf2out_assembly_start): Likewise.
6438 (dwarf2out_begin_prologue): Fix comment.
6439 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6440 if the target needs either debug or unwind DWARF2 info.
6441 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6442 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6443 -ftrapping-math, -ftrapv and -fwrapv.
6445 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6447 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6448 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6449 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6450 (estimate_edge_size_and_time): Likewise.
6451 (ipa_merge_fn_summary_after_inlining): Likewise.
6452 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6453 (compute_uninlined_call_time): Use sreal_frequency.
6454 (compute_inlined_call_time): Likewise.
6455 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6456 * profile-count.c: Include sreal.h.
6457 (profile_count::to_sreal_scale): New.
6458 * profile-count.h: Forward declare sreal.
6459 (profile_count::to_sreal_scale): Declare.
6461 2017-11-13 Nathan Sidwell <nathan@acm.org>
6463 * diagnostic.c (maybe_line_and_column): New.
6464 (diagnostic_get_location_text): Use it.
6465 (diagnostic_report_current_module): Likewise.
6466 (test_diagnostic_get_location_text): Add tests.
6468 2017-11-13 Luis Machado <luis.machado@linaro.org>
6470 * doc/md.texi (Specifying processor pipeline description): Fix
6471 incorrect latency for the div instruction example.
6473 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6475 PR tree-optimization/78821
6476 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6477 that bit_not_p is the same.
6478 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6479 (split_group): Count precisely bit_not_p bits in each statement.
6480 (invert_op): New function.
6481 (imm_store_chain_info::output_merged_store): Use invert_op to
6482 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6483 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6485 2017-11-13 Martin Liska <mliska@suse.cz>
6487 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6488 (struct source_info): Likewise.
6489 (add_branch_counts): Likewise.
6490 (add_line_counts): Likewise.
6491 (function_summary): Likewise.
6492 (output_intermediate_line): Likewise.
6493 (generate_results): Likewise.
6495 2017-11-13 Martin Liska <mliska@suse.cz>
6497 * gcov.c (struct block_info): Remove typedef for block_t.
6498 (struct line_info): Likewise.
6499 (line_info::has_block): Likewise.
6500 (EXIT_BLOCK): Likewise.
6501 (unblock): Likewise.
6502 (circuit): Likewise.
6503 (get_cycles_count): Likewise.
6504 (process_file): Likewise.
6505 (read_graph_file): Likewise.
6506 (solve_flow_graph): Likewise.
6507 (find_exception_blocks): Likewise.
6508 (add_line_counts): Likewise.
6509 (accumulate_line_info): Likewise.
6510 (output_line_details): Likewise.
6512 2017-11-13 Martin Liska <mliska@suse.cz>
6514 * gcov.c (struct arc_info): Remove typedef for arc_t.
6515 (struct line_info): Likewise.
6516 (add_branch_counts): Likewise.
6517 (output_branch_count): Likewise.
6518 (function_info::~function_info): Likewise.
6519 (circuit): Likewise.
6520 (output_intermediate_line): Likewise.
6521 (read_graph_file): Likewise.
6522 (solve_flow_graph): Likewise.
6523 (find_exception_blocks): Likewise.
6524 (add_line_counts): Likewise.
6525 (accumulate_line_info): Likewise.
6526 (output_line_details): Likewise.
6527 (output_function_details): Likewise.
6529 2017-11-13 Martin Liska <mliska@suse.cz>
6531 * gcov.c (struct function_info): Remove typedef for function_t.
6532 (struct source_info): Likewise.
6533 (source_info::get_functions_at_location): Likewise.
6534 (solve_flow_graph): Likewise.
6535 (find_exception_blocks): Likewise.
6536 (add_line_counts): Likewise.
6537 (output_intermediate_file): Likewise.
6538 (process_file): Likewise.
6539 (generate_results): Likewise.
6540 (release_structures): Likewise.
6541 (read_graph_file): Likewise.
6542 (read_count_file): Likewise.
6543 (accumulate_line_counts): Likewise.
6544 (output_lines): Likewise.
6546 2017-11-13 Martin Liska <mliska@suse.cz>
6548 * gcov.c (function_info::function_info): Remove num_counts
6549 and add vector<gcov_type>.
6550 (function_info::~function_info): Use the vector.
6551 (process_file): Likewise.
6552 (read_graph_file): Likewise.
6553 (read_count_file): Likewise.
6554 (solve_flow_graph): Likewise.
6556 2017-11-13 Martin Liska <mliska@suse.cz>
6558 * gcov.c (function_info::is_artificial): New function.
6559 (process_file): Erase all artificial early.
6560 (generate_results): Skip as all artificial are already
6563 2017-11-13 Martin Liska <mliska@suse.cz>
6565 * gcov.c (read_graph_file): Store to global vector of functions.
6566 (read_count_file): Iterate the vector.
6567 (process_file): Likewise.
6568 (generate_results): Likewise.
6569 (release_structures): Likewise.
6571 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6573 PR tree-optimization/82954
6574 * gimple-ssa-store-merging.c
6575 (imm_store_chain_info::coalesce_immediate_stores): If
6576 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6578 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6580 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6581 Upddate call to ENDIAN_LANE_N.
6582 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6583 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6584 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6585 and use aarch64_endian_lane_rtx.
6586 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6588 2017-11-12 Tom de Vries <tom@codesourcery.com>
6590 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6592 2017-11-12 Tom de Vries <tom@codesourcery.com>
6594 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6596 2017-11-12 Tom de Vries <tom@codesourcery.com>
6598 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6600 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6601 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6602 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6604 2017-11-11 Martin Sebor <msebor@redhat.com>
6607 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6610 * prefic.c (translate_name): Replace strncpy with memcpy to
6611 avoid -Wstringop-truncation.
6613 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6615 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6617 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6619 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6621 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6623 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6626 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6628 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6629 enable generating XXBRH if the value is in a vector register.
6630 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6631 value is in a vector register.
6632 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6633 register to register bswap64's instead of doing the GPR sequence
6634 used on previous machines.
6635 (bswapdi2_xxbrd): New insn.
6636 (bswapdi2_reg): Disallow on ISA 3.0.
6637 (register to register bswap64 splitter): Do not split the insn on
6638 ISA 3.0 systems that use XXBRD.
6640 2017-11-10 Martin Sebor <msebor@redhat.com>
6643 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6645 (find_subframework_file): Same.
6647 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6649 * auto-profile.c (afdo_indirect_call): Drop frequency.
6650 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6651 (cgraph_node::create_edge): Drop frequency argument.
6652 (cgraph_node::create_indirect_edge): Drop frequency argument.
6653 (cgraph_edge::make_speculative): Drop frequency arguments.
6654 (cgraph_edge::resolve_speculation): Do not update frequencies
6655 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6656 (cgraph_node::dump): Check consistency in IPA mode.
6657 (cgraph_edge::maybe_hot_p): Use IPA counter.
6658 (cgraph_edge::verify_count_and_frequency): Rename to ...
6659 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6660 (cgraph_node::verify_node): Update.
6661 * cgraph.h (struct cgraph_edge): Drop frequency.
6662 (cgraph_edge::frequency): New function.
6663 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6665 (cgraph_edge::rebuild_edges): Likewise.
6666 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6667 (duplicate_thunk_for_node): Do not pass frequency.
6668 (cgraph_node::create_clone): Scale only counts.
6669 (cgraph_node::create_virtual_clone): Do not pass frequency.
6670 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6671 (cgraph_node::create_version_clone): Do not pass frequency.
6672 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6673 (cgraph_node::expand_thunk): Do not pass frequency.
6674 (cgraph_node::create_wrapper): Do not pass frequency.
6675 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6677 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6678 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6679 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6680 (gather_caller_stats): Use frequency function.
6681 (ipcp_cloning_candidate_p): Use frequency function.
6682 (ipcp_propagate_stage): Use frequency function.
6683 (get_info_about_necessary_edges): Use frequency function.
6684 (update_profiling_info): Update only IPA profile.
6685 (update_specialized_profile): Use frequency functoin.
6686 (perhaps_add_new_callers): Update only IPA profile.
6687 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6688 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6689 (dump_ipa_call_summary): Use frequency function.
6690 (estimate_edge_size_and_time): Use frequency function.
6691 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6692 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6693 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6694 (update_noncloned_counts): ... ths one; scale counts only.
6695 (clone_inlined_nodes): Do not scale frequency.
6696 (inline_call): Do not pass frequency.
6697 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6698 (compute_inlined_call_time): Use IPA profile.
6699 (want_inline_small_function_p): Use IPA profile.
6700 (want_inline_self_recursive_call_p): Use IPA profile.
6701 (edge_badness): Use IPA profile.
6702 (lookup_recursive_calls): Use IPA profile.
6703 (recursive_inlining): Do not pass frequency.
6704 (resolve_noninline_speculation): Do not update frequency.
6705 (inline_small_functions): Collect max of IPA profile.
6706 (dump_overall_stats): Dump IPA porfile.
6707 (dump_inline_stats): Dump IPA porfile.
6708 (ipa_inline): Collect IPA stats.
6709 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6710 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6711 (ipa_propagate_frequency): Use frequency function.
6712 (ipa_profile): Cleanup.
6713 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6714 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6715 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6716 (input_node): Do not stream frequency.
6717 (input_edge): Do not stream frequency.
6718 (merge_profile_summaries): Scale only IPA profiles.
6719 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6720 * predict.c (drop_profile): Do not recompute frequency.
6721 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6722 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6723 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6724 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6725 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6726 * tree-inline.c (copy_bb): Do not scale frequency.
6727 (expand_call_inline): Do not scale frequency.
6728 (tree_function_versioning): Do not scale frequency.
6729 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6731 2017-11-10 Julia Koval <julia.koval@intel.com>
6733 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6734 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6735 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6736 (_mm256_maskz_gf2p8affine_epi64_epi8)
6737 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6738 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6739 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6740 (__builtin_ia32_vgf2p8affineqb_v32qi)
6741 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6742 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6744 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6748 (arm_option_override): Refactor.
6749 (arm_option_reconfigure_globals): New.
6750 (arm_options_perform_arch_sanity_checks): New.
6751 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6753 (arm_options_perform_arch_sanity_checks): Likewise
6755 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6757 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6758 (power9-qpmul): New.
6759 * rs6000/rs6000.md ("type" attr): Add qmul.
6760 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6761 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6762 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6764 2017-11-10 Martin Sebor <msebor@redhat.com>
6767 * builtins.c (compute_objsize): Handle arrays that
6768 compute_builtin_object_size likes to fail for. Make extern.
6769 * builtins.h (compute_objsize): Declare.
6770 (check_strncpy_sizes): New function.
6771 (expand_builtin_strncpy): Call check_strncpy_sizes.
6772 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6773 -Wstringop-truncation.
6774 (gimple_fold_builtin_strncat): Same.
6775 * gimple.c (gimple_build_call_from_tree): Set call location.
6776 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6777 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6778 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6779 (handle_builtin_strlen): Use strlen_to_stridx.
6780 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6782 Use strlen_to_stridx.
6783 (pass_strlen::execute): Release strlen_to_stridx.
6784 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6785 (-Wstringop-truncation): Document new option.
6787 2017-11-10 Martin Liska <mliska@suse.cz>
6789 PR gcov-profile/82702
6790 * gcov.c (main): Handle intermediate files in a different
6792 (get_gcov_intermediate_filename): New function.
6793 (output_gcov_file): Remove support of intermediate files.
6794 (generate_results): Allocate intermediate file.
6795 (release_structures): Clean-up properly fn_end.
6796 (output_intermediate_file): Start iterating with line 1.
6798 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6800 PR tree-optimization/82929
6801 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6802 ops_swapped_p non-static data member.
6803 (store_immediate_info::store_immediate_info): Clear it.
6804 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6805 ops set ops_swapped_p.
6806 (count_multiple_uses): Handle ops_swapped_p.
6808 2017-11-10 Martin Liska <mliska@suse.cz>
6810 * coverage.c (coverage_init): Stream information about
6811 support of has_unexecuted_blocks.
6812 * doc/gcov.texi: Document that.
6813 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6814 * gcov.c (read_graph_file): Likewise.
6815 (output_line_beginning): Fix a small issue with
6818 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6820 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6821 reference of trivial component.
6823 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6826 * gimple-ssa-store-merging.c
6827 (pass_store_merging::terminate_all_aliasing_chains): For
6828 gimple_store_p stmts also call refs_output_dependent_p.
6830 PR rtl-optimization/82913
6831 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6834 2017-11-09 Jeff Law <law@redhat.com>
6836 * vr-values.h: New file with vr_values class.
6837 * tree-vrp.c: Include vr-values.h
6838 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6839 data objects into the vr_values class.
6840 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6841 (get_value_range): Make it a member function within vr_values class.
6842 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6843 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6844 (op_with_constant_singleton_value_range): Likewise.
6845 (extract_range_for_var_from_comparison_expr): Likewise.
6846 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6847 (extract_range_from_binary_expr): Likewise.
6848 (extract_range_from_unary_expr): Likewise.
6849 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6850 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6851 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6852 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6853 (compare_name_with_value, compare_names): Likewise.
6854 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6855 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6856 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6857 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6858 (extract_range_from_phi_node): Likewise.
6859 (simplify_truth_ops_using_ranges): Likewise.
6860 (simplify_div_or_mod_using_ranges): Likewise.
6861 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6862 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6863 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6864 (simplify_float_conversion_using_ranges): Likewise.
6865 (simplify_internal_call_using_ranges): Likewise.
6866 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6867 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6868 poitner via x_vr_values for calls into gimple folder.
6869 (vrp_initialize_lattice): Make this the vr_values ctor.
6870 (vrp_free_lattice): Make this the vr_values dtor.
6871 (set_vr_value): New function.
6872 (class vrp_prop): Add vr_values data member. Add various member
6873 functions as well as member functions that delegate to vr_values.
6874 (check_array_ref): Make a member function within vrp_prop class.
6875 (search_for_addr_array, vrp_initialize): Likewise.
6876 (vrp_finalize): Likewise. Revamp to avoid direct access to
6877 vr_value, values_propagated, etc.
6878 (check_array_bounds): Extract vrp_prop class instance pointer from
6879 walk info structure. Use it to call member functions.
6880 (check_all_array_refs): Make a member function within vrp_prop class.
6881 Smuggle class instance pointer via walk info structure.
6882 (x_vr_values): New local static.
6883 (vrp_valueize): Use x_vr_values to get class instance.
6884 (vr_valueize_1): Likewise.
6885 (class vrp_folder): Add vr_values data member. Add various member
6886 functions as well as member functions that delegate to vr_values.
6887 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6888 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6889 class instance from vr_values. Use it to call member functions.
6890 (vrp_dom_walker): Add vr_values data member.
6891 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6892 instance via x_vr_values.
6893 (identify_jump_threads): Accept vr_values as argument. Store
6894 it into the walker structure.
6895 (evrp_dom_walker): Add vr_values class data member. Add various
6897 (evrp_dom_walker::try_find_new_range): Use vr_values data
6898 member to access the memory allocator.
6899 (evrp_dom_walker::before_dom_children): Store vr_values class
6900 instance into the vrp_folder class.
6901 (evrp_dom_walker::push_value_range): Rework to avoid direct
6902 access to num_vr_values and vr_value.
6903 (evrp_dom_walker::pop_value_range): Likewise.
6904 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6905 Use vr_values to get to dump_all_value_ranges member function.
6906 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6907 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6908 Pass vr_values class instance down to identify_jump_threads.
6909 Remove call to vrp_free_lattice.
6910 (debug_all_value_ranges): Remove.
6912 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6913 (vrp_folder): Likewise.
6915 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6916 Get it from the existing bitmap instead.
6917 (vrp_intersect_ranges_1): Likewise.
6919 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6921 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6923 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6924 set bit_not_p to it.
6925 (imm_store_chain_info::coalesce_immediate_stores): Break group
6926 if bit_not_p is different.
6927 (count_multiple_uses, split_group,
6928 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6929 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6930 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6931 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6932 to store_immediate_info ctor.
6934 2017-11-09 Jim Wilson <jimw@sifive.com>
6936 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6937 (scan_prog_file): Likewise.
6939 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6941 * bb-reorder.c (max_entry_frequency): Remove.
6942 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6943 connect_traces, push_to_next_round_p): Remove prototypes.
6944 (find_traces_1_round): Use counts only.
6945 (push_to_next_round_p): Likewise.
6946 (find_traces): Likewise.
6947 (rotate_loop): Likewise.
6948 (find_traces_1_round): Likewise.
6949 (connect_traces): Likewise.
6950 (edge_order): Likewise.
6952 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6954 * config/arm/arm.c (output_return_instruction): Add comments to
6955 indicate requirement for cmse_nonsecure_entry return to account
6956 for the size of clearing instruction output here.
6957 (thumb_exit): Likewise.
6958 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6959 return in hardfloat mode.
6961 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6963 * config/rs6000/rs6000.c (machine_function): Add a bool,
6964 "toc_is_wrapped_separately".
6965 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6966 if it wasn't explicitly set or unset, we are optimizing for speed, and
6967 doing separate shrink-wrapping.
6968 (rs6000_get_separate_components): Enable the TOC component if
6969 saving the TOC register in the prologue.
6970 (rs6000_components_for_bb): Handle the TOC component.
6971 (rs6000_emit_prologue_components): Store the TOC register where needed.
6972 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6973 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6975 2017-11-09 Martin Jambor <mjambor@suse.cz>
6977 * ipa-param-manipulation.c: New file.
6978 * ipa-param-manipulation.h: Likewise.
6979 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6980 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6981 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6982 (ipa_parm_adjustment): Likewise.
6983 (ipa_parm_adjustment_vec): Likewise.
6984 (ipa_get_vector_of_formal_parms): Moved declaration to
6985 ipa-param-manipulation.h.
6986 (ipa_get_vector_of_formal_parm_types): Likewise.
6987 (ipa_modify_formal_parameters): Likewise.
6988 (ipa_modify_call_arguments): Likewise.
6989 (ipa_combine_adjustments): Likewise.
6990 (ipa_dump_param_adjustments): Likewise.
6991 (ipa_modify_expr): Likewise.
6992 (ipa_get_adjustment_candidate): Likewise.
6993 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6994 ipa-param-manipulation.c.
6995 (ipa_get_vector_of_formal_parm_types): Likewise.
6996 (ipa_modify_formal_parameters): Likewise.
6997 (ipa_modify_call_arguments): Likewise.
6998 (ipa_modify_expr): Likewise.
6999 (get_ssa_base_param): Likewise.
7000 (ipa_get_adjustment_candidate): Likewise.
7001 (index_in_adjustments_multiple_times_p): Likewise.
7002 (ipa_combine_adjustments): Likewise.
7003 (ipa_dump_param_adjustments): Likewise.
7004 * tree-sra.c: Also include ipa-param-manipulation.h
7005 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7008 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7009 Alan Hayward <alan.hayward@arm.com>
7010 David Sherwood <david.sherwood@arm.com>
7012 * doc/sourcebuild.texi (vect_masked_store): Document.
7014 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7015 Alan Hayward <alan.hayward@arm.com>
7016 David Sherwood <david.sherwood@arm.com>
7018 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7020 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7021 Alan Hayward <alan.hayward@arm.com>
7022 David Sherwood <david.sherwood@arm.com>
7024 * doc/sourcebuild.texi (vect_variable_length): Document.
7026 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7027 Alan Hayward <alan.hayward@arm.com>
7028 David Sherwood <david.sherwood@arm.com>
7030 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7032 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7033 Alan Hayward <alan.hayward@arm.com>
7034 David Sherwood <david.sherwood@arm.com>
7036 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7038 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7039 Alan Hayward <alan.hayward@arm.com>
7040 David Sherwood <david.sherwood@arm.com>
7042 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7043 previously undocumented selectors.
7044 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7046 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7048 * doc/rtl.texi (const_vector): Say that elements can be
7049 const_wide_ints too.
7050 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7051 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7052 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7053 * optabs.c (expand_vector_broadcast): Likewise.
7055 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7056 Alan Hayward <alan.hayward@arm.com>
7057 David Sherwood <david.sherwood@arm.com>
7059 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7060 scaled index even if the unscaled address was invalid.
7061 Don't increase the complexity of using a scale in that case.
7063 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7064 Alan Hayward <alan.hayward@arm.com>
7065 David Sherwood <david.sherwood@arm.com>
7067 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7068 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7069 UNITS_PER_WORD bytes.
7070 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7071 based on REGMODE_NATURAL_SIZE of the inner mode.
7072 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7073 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7074 * expmed.c (lowpart_bit_field_p): Divide the value up into
7075 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7076 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7077 whether something is likely to occupy more than one register.
7079 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7082 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7083 profile_count::adjust_for_ipa_scaling.
7084 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7085 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7087 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7089 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7091 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7092 (split_group): Add total_orig and total_new arguments, estimate the
7093 number of statements related to the store group without store merging
7094 and with store merging.
7095 (imm_store_chain_info::output_merged_store): Adjust split_group
7096 callers, punt if estimated number of statements with store merging
7097 is not smaller than estimated number of statements without it.
7099 (handled_load): Remove has_single_use checks.
7100 (pass_store_merging::process_store): Likewise.
7102 2017-11-09 Richard Biener <rguenther@suse.de>
7104 PR tree-optimization/82902
7105 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7107 2017-11-09 Martin Liska <mliska@suse.cz>
7110 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7113 2017-11-09 Martin Liska <mliska@suse.cz>
7115 PR tree-optimization/82669
7116 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7118 2017-11-09 Martin Liska <mliska@suse.cz>
7120 PR gcov-profile/48463
7121 * coverage.c (coverage_begin_function): Output also end locus
7122 of a function and information whether the function is
7124 * gcov-dump.c (tag_function): Parse and print the information.
7125 * gcov.c (INCLUDE_MAP): Add include.
7126 (INCLUDE_SET): Likewise.
7127 (struct line_info): Move earlier in the source file because
7128 of vector<line_info> in function_info structure.
7129 (line_info::line_info): Likewise.
7130 (line_info::has_block): Likewise.
7131 (struct source_info): Add new member index.
7132 (source_info::get_functions_at_location): New function.
7133 (function_info::group_line_p): New function.
7134 (output_intermediate_line): New function.
7135 (output_intermediate_file): Use the mentioned function.
7136 (struct function_start): New.
7137 (struct function_start_pair_hash): Likewise.
7138 (process_file): Add code that identifies group functions.
7139 Assign lines either to global or function scope.
7140 (generate_results): Skip artificial functions.
7141 (find_source): Assign index for each source file.
7142 (read_graph_file): Read new flag artificial and end_line.
7143 (add_line_counts): Assign it either to global of function scope.
7144 (accumulate_line_counts): Isolate core of the function to
7145 accumulate_line_info and call it for both function and global
7147 (accumulate_line_info): New function.
7148 (output_line_beginning): Fix GNU coding style.
7149 (print_source_line): New function.
7150 (output_line_details): Likewise.
7151 (output_function_details): Likewise.
7152 (output_lines): Iterate both source (global) scope and function
7154 (struct function_line_start_cmp): New class.
7155 * doc/gcov.texi: Reflect changes in documentation.
7157 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7160 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7161 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7162 and similarly for not instead of neg.
7164 2017-11-08 Andi Kleen <ak@linux.intel.com>
7166 * config/i386/i386.opt: Add -mforce-indirect-call.
7167 * config/i386/predicates.md: Check for flag_force_indirect_call.
7168 * doc/invoke.texi: Document -mforce-indirect-call
7170 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7172 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7174 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7175 config/riscv/riscv.c (predict.h): New include.
7176 (riscv_slow_unaligned_access_p): No longer static.
7177 (riscv_block_move_straight): Add require.
7178 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7180 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7183 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7184 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7185 nonimmediate_operand predicate for operand 1 instead of
7188 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7190 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7192 * config/aarch64/constraints.md (Uml): New constraint.
7193 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7196 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7198 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7199 of two vec_duplicates into a vec_concat.
7201 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7203 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7204 Simplify vec_merge of vec_duplicate and vec_concat.
7205 * config/aarch64/constraints.md (Utq): New constraint.
7206 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7209 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7211 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7212 Simplify vec_merge of vec_duplicate and const_vector.
7213 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7215 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7216 mode iterator. Update predicate on operand 1 to
7217 handle non-const_vec constants. Delete constraints.
7218 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7220 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7222 PR tree-optimization/78821
7223 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7225 (store_operand_info::store_operand_info): Initialize it to false.
7226 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7227 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7228 store in the group, and if chain_info is non-NULL, to ignore altogether
7230 (compatible_load_p): Fail if bit_not_p does not match.
7231 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7232 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7233 (pass_store_merging::process_store): Adjust
7234 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7235 call terminate_all_aliasing_chains newly when adding a store into
7236 a chain with non-NULL chain_info.
7238 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7240 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7242 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7244 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7246 (aarch64_layout_frame): Initialise emit_frame_chain.
7247 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7248 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7250 2017-11-08 Martin Liska <mliska@suse.cz>
7252 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7253 of gimple_call_internal_p.
7255 2017-11-07 Tom de Vries <tom@codesourcery.com>
7257 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7259 2017-11-07 Tom de Vries <tom@codesourcery.com>
7261 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7264 2017-11-08 Martin Liska <mliska@suse.cz>
7267 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7269 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7271 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7273 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7276 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7277 (*cmp<mode>_ccz_1): New insn with $k alternative.
7280 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7281 RTX_COMM_COMPARE as commutative as well.
7282 (ix86_binary_operator_ok): Formatting fix.
7283 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7284 *<code><mode>3<mask_name><round_saeonly_name>,
7285 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7286 *<s>mul<mode>3_highpart<mask_name>,
7287 *vec_widen_umult_even_v16si<mask_name>,
7288 *vec_widen_umult_even_v8si<mask_name>,
7289 *vec_widen_umult_even_v4si<mask_name>,
7290 *vec_widen_smult_even_v16si<mask_name>,
7291 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7292 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7293 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7294 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7295 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7296 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7297 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7298 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7299 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7300 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7301 ix86_binary_operator_ok. Formatting fixes.
7302 (*<plusminus_insn><mode>3<mask_name><round_name>,
7303 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7306 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7308 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7309 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7310 eq and ne if TARGET_ISEL.
7311 (cmp): New code_iterator.
7312 (UNS, UNSU_, UNSIK): New code_attrs.
7313 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7314 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7316 ("eq<mode>3"): ... this.
7317 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7319 ("ne<mode>3"): ... this.
7321 2017-11-07 Julia Koval <julia.koval@intel.com>
7324 * common/config/i386/i386-common.c
7325 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7326 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7327 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7328 * config/i386/i386.opt: Ditto.
7329 * config/i386/i386.c (ix86_target_string): Ditto.
7330 (ix86_option_override_internal): Ditto.
7331 (ix86_init_mpx_builtins): Move MPX to args2.
7332 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7333 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7334 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7335 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7336 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7337 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7338 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7340 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7343 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7344 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7345 (zero-extendsidi peephole2): Remove peephole.
7347 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7350 * stor-layout.c: Include attribs.h.
7351 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7352 explicit lookup of "aligned" attribute.
7354 2017-11-07 Andrew Waterman <andrew@sifive.com>
7356 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7357 (riscv_expand_block_move): Likewise.
7358 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7360 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7361 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7362 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7363 (riscv_adjust_block_mem): Likewise.
7364 (riscv_block_move_loop): Likewise.
7365 (riscv_expand_block_move): Likewise.
7366 * config/riscv/riscv.md (movmemsi): New pattern.
7368 2017-11-07 Michael Clark <michaeljclark@mac.com>
7370 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7371 (MUSL_DYNAMIC_LINKER): Likewise.
7373 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7375 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7376 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7379 2017-11-07 Richard Biener <rguenther@suse.de>
7381 * match.pd: Fix build.
7383 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7384 Jackson Woodruff <jackson.woodruff@arm.com>
7386 PR tree-optimization/71026
7387 * match.pd: Canonicalize negate in division.
7389 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7392 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7394 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7396 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7397 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7399 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7401 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7402 non-scalar integral types.
7403 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7404 (-(A-B), -(~A)): New transformations.
7406 2017-11-07 Tom de Vries <tom@codesourcery.com>
7408 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7409 semicolon after "do {} while (0)".
7410 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7411 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7412 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7413 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7414 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7416 2017-11-07 Tom de Vries <tom@codesourcery.com>
7418 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7419 after "do {} while (0)".
7420 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7421 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7422 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7423 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7424 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7426 2017-11-07 Tom de Vries <tom@codesourcery.com>
7429 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7431 (arm_rtx_costs_internal): Add missing semicolon after
7432 HANDLE_NARROW_SHIFT_ARITH call.
7434 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7436 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7437 disable isel if it was not set explicitly.
7439 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7441 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7442 (add_type_attribute) likewise.
7444 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7446 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7447 of struct ix86_frame.
7448 (ix86_initial_elimination_offset): Likewise.
7449 (ix86_expand_split_stack_prologue): Likewise.
7451 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7453 * tree-vrp.h (enum value_range_type): Update stale comment.
7455 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7456 Alan Hayward <alan.hayward@arm.com>
7457 David Sherwood <david.sherwood@arm.com>
7459 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7460 (aarch64_expand_vec_perm_const): Take the number of units too.
7461 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7462 (aarch64_expand_vec_perm_const): Likewise.
7463 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7464 (vec_perm<mode>): Update accordingly.
7466 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7467 Alan Hayward <alan.hayward@arm.com>
7468 David Sherwood <david.sherwood@arm.com>
7470 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7471 Take the number of units too.
7472 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7473 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7474 but check for a vector mode before rather than after the call.
7475 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7476 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7477 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7478 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7479 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7480 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7481 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7482 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7483 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7484 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7485 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7486 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7487 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7488 (aarch64_sqdmull2_n<mode>): Update accordingly.
7490 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7491 Alan Hayward <alan.hayward@arm.com>
7492 David Sherwood <david.sherwood@arm.com>
7494 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7495 the number of units too.
7496 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7497 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7498 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7499 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7500 (vec_store_lanesxi<mode>): Update accordingly.
7502 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7503 Alan Hayward <alan.hayward@arm.com>
7504 David Sherwood <david.sherwood@arm.com>
7506 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7507 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7508 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7509 of units rather than the mode.
7510 * config/aarch64/iterators.md (nunits): New mode attribute.
7511 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7512 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7513 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7514 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7515 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7516 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7517 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7518 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7519 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7520 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7521 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7522 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7523 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7524 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7525 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7526 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7527 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7528 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7529 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7530 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7531 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7532 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7533 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7534 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7535 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7536 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7537 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7538 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7539 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7540 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7541 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7542 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7543 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7544 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7545 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7546 (aarch64_simd_vec_setv2di): Likewise.
7548 2017-11-06 Carl Love <cel@us.ibm.com>
7550 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7552 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7553 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7555 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7557 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7558 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7559 Add power 8 macro expansions.
7560 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7561 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7562 power 8 instructions. (VSX_XXBR): Add iterator.
7564 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7566 * config/arm/arm.md (predicable_short_it): Change default to "no",
7567 improve documentation, remove uses that are identical to the default.
7568 (enabled_for_depr_it): Rename to enabled_for_short_it.
7569 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7570 Remove default uses.
7571 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7572 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7573 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7574 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7576 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7579 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7580 float128 helper macros, which are no longer used after deleting
7581 the old 'q' built-in functions, and moving the round to odd
7582 built-in functions to being special built-in functions.
7583 (BU_FLOAT128_2): Likewise.
7584 (BU_FLOAT128_1_HW): Likewise.
7585 (BU_FLOAT128_2_HW): Likewise.
7586 (BU_FLOAT128_3_HW): Likewise.
7587 (FABSQ): Delete old 'q' built-in functions.
7588 (COPYSIGNQ): Likewise.
7589 (SQRTF128_ODD): Move round to odd built-in functions to be
7590 special built-in functions, so that we can handle
7591 -mabi=ieeelongdouble.
7592 (TRUNCF128_ODD): Likewise.
7593 (ADDF128_ODD): Likewise.
7594 (SUBF128_ODD): Likewise.
7595 (MULF128_ODD): Likewise.
7596 (DIVF128_ODD): Likewise.
7597 (FMAF128_ODD): Likewise.
7598 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7599 built-in names to 'f128'.
7600 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7601 old 'q' built-in functions, as the machine independent code for
7602 'f128' built-in functions handles this.
7603 (rs6000_expand_builtin): Add expansion for float128 round to odd
7604 functions, keying off on -mabi=ieeelongdouble of whether to use
7605 the KFmode or TFmode variant.
7606 (rs6000_init_builtins): Initialize the _Float128 round to odd
7608 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7609 _Float128 'q' built-in functions are now mapped into the new
7610 'f128' built-in functions.
7612 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7614 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7615 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7617 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7620 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7621 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7622 * toplev.c: Include "ipa-fnsummary.h".
7623 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7625 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7627 PR tree-optimization/82838
7628 * gimple-ssa-store-merging.c
7629 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7630 on a separate gimple_seq which is then appended to seq.
7632 2017-11-06 Jeff Law <law@redhat.com>
7635 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7636 (get_probe_interval): New functions.
7637 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7638 (ix86_adjust_stack_and_probe): Likewise.
7639 (output_adjust_stack_and_probe): Likewise.
7640 (ix86_emit_probe_stack_range): Likewise.
7641 (ix86_expand_prologue): Likewise.
7643 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7645 PR tree-optimization/82816
7646 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7647 if the modes of the two types are the same.
7648 (convert_plusminus_to_widen): Likewise.
7650 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7652 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7654 (usadv16qi): New define_expand.
7655 (usadv8hi): New define_expand.
7657 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7660 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7662 (inline_transform): Likewise.
7663 * predict.c (counts_to_freqs): Remove useless conditional.
7664 * profile-count.h (profile_count::apply_scale): Move sanity check.
7665 * tree-inline.c (copy_bb): Always scale.
7666 (copy_cfg_body): Likewise.
7668 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7671 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7673 (*cmp_ite0): Add enabled_for_depr_it attribute.
7674 (*cmp_ite1): Likewise.
7676 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7678 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7681 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7683 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7684 return the value by pointer.
7685 (extract_range_from_multiplicative_op_1): Update accordingly.
7686 Return as soon as an operation fails.
7688 2017-11-05 Tom de Vries <tom@codesourcery.com>
7691 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7692 (DEF_SANITIZER_BUILTIN): ... here.
7693 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7694 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7696 2017-11-05 Tom de Vries <tom@codesourcery.com>
7699 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7701 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7702 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7703 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7705 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7706 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7707 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7709 2017-11-05 Tom de Vries <tom@codesourcery.com>
7712 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7715 2017-11-04 Michael Clark <michaeljclark@mac.com>
7717 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7718 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7720 (*addsi3_extended): Likewise.
7721 (*addsi3_extended2): Likewise.
7722 (<optab>si3): Likewise.
7723 (<optab>di3): Likewise.
7724 (<optab><mode>3): Likewise.
7725 (<*optabe>si3_internal): Likewise.
7726 (zero_extendqi<SUPERQI:mode>2): Likewise.
7727 (*add<mode>hi3): Likewise.
7728 (*xor<mode>hi3): Likewise.
7729 (<optab>di3): Likewise.
7730 (*<optab>si3_extend): Likewise.
7731 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7732 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7733 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7735 2017-11-04 Andrew Waterman <andrew@sifive.com>
7737 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7738 TARGET_STRICT_ALIGN based upon -mtune argument.
7740 2017-11-04 Andrew Waterman <andrew@sifive.com>
7742 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7744 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7746 * config/i386/i386.c (choose_basereg): Use optional scratch
7747 register and add assertion.
7748 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7749 needed, and don't allocate stack.
7750 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7751 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7752 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7754 2017-11-03 Jeff Law <law@redhat.com>
7756 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7757 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7758 to probe at the start of a noreturn function.
7760 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7762 PR tree-optimization/78821
7763 * gimple-ssa-store-merging.c: Update the file comment.
7764 (MAX_STORE_ALIAS_CHECKS): Define.
7765 (struct store_operand_info): New type.
7766 (store_operand_info::store_operand_info): New constructor.
7767 (struct store_immediate_info): Add rhs_code and ops data members.
7768 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7769 and op1r arguments to the ctor, initialize corresponding data members.
7770 (struct merged_store_group): Add load_align_base and load_align
7772 (merged_store_group::merged_store_group): Initialize them.
7773 (merged_store_group::do_merge): Update them.
7774 (merged_store_group::apply_stores): Pick the constant for
7775 encode_tree_to_bitpos from one of the two operands, or skip
7776 encode_tree_to_bitpos if neither operand is a constant.
7777 (class pass_store_merging): Add process_store method decl. Remove
7778 bool argument from terminate_all_aliasing_chains method decl.
7779 (pass_store_merging::terminate_all_aliasing_chains): Remove
7780 var_offset_p argument and corresponding handling.
7781 (stmts_may_clobber_ref_p): New function.
7782 (compatible_load_p): New function.
7783 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7784 if there is overlap and rhs_code is not INTEGER_CST. For
7785 non-overlapping stores terminate group if rhs is not mergeable.
7786 (get_alias_type_for_stmts): Change first argument from
7787 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7788 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7789 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7791 (get_location_for_stmts): Change first argument from
7792 auto_vec<gimple *> & to vec<gimple *> &.
7793 (struct split_store): Remove orig_stmts data member, add orig_stores.
7794 (split_store::split_store): Create orig_stores rather than orig_stmts.
7795 (find_constituent_stmts): Renamed to ...
7796 (find_constituent_stores): ... this. Change second argument from
7797 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7798 to info structures rather than the statements.
7799 (split_group): Rename ALLOW_UNALIGNED argument to
7800 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7801 it. Adjust find_constituent_stores caller.
7802 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7803 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7804 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7805 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7806 (mem_valid_for_store_merging): New function.
7807 (handled_load): New function.
7808 (pass_store_merging::process_store): New method.
7809 (pass_store_merging::execute): Use process_store method. Adjust
7810 terminate_all_aliasing_chains caller.
7812 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7814 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7815 Return true for more constants, symbols and label references.
7816 (aarch64_valid_floating_const): Remove unused function.
7818 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7821 * config/aarch64/aarch64.c (aarch64_layout_frame):
7822 Undo forcing of LR at bottom of frame.
7824 2017-11-03 Jeff Law <law@redhat.com>
7827 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7828 for int_registers_saved.
7830 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7831 extracted from tree-ssa-dom.c.
7832 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7833 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7834 (record_equivalences_from_incoming_edge): Add additional argument
7835 to single_pred_edge_ignoring_loop_edges call.
7836 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7837 (uncprop_dom_walker::before_dom_children): Add additional argument
7838 to single_pred_edge_ignoring_loop_edges call.
7839 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7840 single_pred_edge_ignoring_loop_edges rather than open coding.
7841 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7843 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7845 * match.pd (-(-A)): Rewrite.
7847 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7849 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7850 (rs6000_emit_int_cmove): New declaration.
7851 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7852 (rs6000_emit_sISEL): Delete.
7853 (rs6000_emit_int_cmove): Make non-static.
7854 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7855 instead of rs6000_emit_sISEL.
7857 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7859 * asan.c (create_cond_insert_point): Maintain profile.
7860 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7862 * basic-block.h (struct basic_block_def): Remove frequency.
7863 (EDGE_FREQUENCY): Use to_frequency
7864 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7866 (find_traces): Update to use to_frequency.
7867 (find_traces_1_round): Likewise; use only IPA counts.
7868 (bb_to_key): Likewise.
7869 (connect_traces): Use IPA counts only.
7870 (copy_bb_p): Update to use to_frequency.
7871 (fix_up_crossing_landing_pad): Likewise.
7872 (sanitize_hot_paths): Likewise.
7873 * bt-load.c (basic_block_freq): Likewise.
7874 * cfg.c (init_flow): Set count_max to uninitialized.
7875 (check_bb_profile): Remove frequencies; check counts.
7876 (dump_bb_info): Do not dump frequencies.
7877 (update_bb_profile_for_threading): Update counts only.
7878 (scale_bbs_frequencies_int): Likewise.
7879 (MAX_SAFE_MULTIPLIER): Remove.
7880 (scale_bbs_frequencies_gcov_type): Update counts only.
7881 (scale_bbs_frequencies_profile_count): Update counts only.
7882 (scale_bbs_frequencies): Update counts only.
7883 * cfg.h (struct control_flow_graph): Add count-max.
7884 (update_bb_profile_for_threading): Update prototype.
7885 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7886 (find_many_sub_basic_blocks): Likewise.
7887 * cfgcleanup.c (try_forward_edges): Likewise.
7888 (try_crossjump_to_edge): Likewise.
7889 * cfgexpand.c (expand_gimple_cond): Likewise.
7890 (expand_gimple_tailcall): Likewise.
7891 (construct_init_block): Likewise.
7892 (construct_exit_block): Likewise.
7893 * cfghooks.c (verify_flow_info): Check consistency of counts.
7894 (dump_bb_for_graph): Do not dump frequencies.
7895 (split_block_1): Do not update frequencies.
7896 (split_edge): Do not update frequencies.
7897 (make_forwarder_block): Do not update frequencies.
7898 (duplicate_block): Do not update frequencies.
7899 (account_profile_record): Do not update frequencies.
7900 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7901 for global heuristics.
7902 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7903 (expected_loop_iterations_unbounded): Use counts only.
7904 * cfgloopmanip.c (scale_loop_profile): Simplify.
7905 (create_empty_loop_on_edge): Simplify
7907 (duplicate_loop_to_header_edge): Simplify
7908 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7909 (update_br_prob_note): Take care of removing note when profile
7911 (relink_block_chain): Do not dump frequency.
7912 (rtl_account_profile_record): Use to_frequency.
7913 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7914 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7915 (cgraph_update_edges_for_call_stmt_node): Likewise.
7916 (cgraph_edge::verify_count_and_frequency): Update.
7917 (cgraph_node::verify_node): Temporarily disable frequency verification.
7918 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7919 to_cgraph_frequency.
7920 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7921 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7923 (cgraph_node::expand_thunk): Update profile.
7924 * except.c (dw2_build_landing_pads): Do not update frequency.
7925 * final.c (compute_alignments): Use to_frequency.
7926 (dump_basic_block_info): Do not dump frequency.
7927 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7928 (dump_gimple_bb_header): Do not dump frequency.
7929 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7931 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7932 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7933 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7934 (init_before_recovery): Do not update frequency.
7935 (sched_create_recovery_edges): Do not update frequency.
7936 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7937 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7938 (ipa_cp_c_finalize): Set max_count to uninitialized.
7939 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7940 (param_change_prob): Use counts.
7941 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7943 * ipa-split.c (consider_split): Use to_frequency.
7944 (split_function): Use to_frequency.
7945 * ira-build.c (loop_compare_func): Likewise.
7946 (mark_loops_for_removal): Likewise.
7947 (mark_all_loops_for_removal): Likewise.
7948 * loop-doloop.c (doloop_modify): Do not update frequency.
7949 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7951 * lto-streamer-in.c (input_function): Update count_max.
7952 * omp-expand.c (expand_omp_taskreg): Update count_max.
7953 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7954 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7955 (maybe_hot_count_p): Use ipa counts only.
7956 (maybe_hot_bb_p): Simplify.
7957 (maybe_hot_edge_p): Simplify.
7958 (probably_never_executed): Do not take frequency argument.
7959 (probably_never_executed_bb_p): Do not pass frequency.
7960 (probably_never_executed_edge_p): Likewise.
7961 (combine_predictions_for_bb): Check that profile is nonzero.
7962 (propagate_freq): Do not set frequency.
7963 (drop_profile): Simplify.
7964 (counts_to_freqs): Simplify.
7965 (expensive_function_p): Use to_frequency.
7966 (propagate_unlikely_bbs_forward): Simplify.
7967 (determine_unlikely_bbs): Simplify.
7968 (estimate_bb_frequencies): Add hack to silence graphite issues.
7969 (compute_function_frequency): Use ipa counts.
7970 (pass_profile::execute): Update.
7971 (rebuild_frequencies): Use counts only.
7972 (force_edge_cold): Use counts only.
7973 * profile-count.c (profile_count::dump): Dump new count types.
7974 (profile_count::differs_from_p): Check compatiblity.
7975 (profile_count::to_frequency): New function.
7976 (profile_count::to_cgraph_frequency): New function.
7977 * profile-count.h (struct function): Declare.
7978 (enum profile_quality): Add profile_guessed_local and
7979 profile_guessed_global0.
7980 (class profile_proability): Decrease number of bits to 29;
7981 update from_reg_br_prob_note and to_reg_br_prob_note.
7982 (class profile_count: Update comment; decrease number of bits
7983 to 61. Check compatibility.
7984 (profile_count::compatible_p): New private member function.
7985 (profile_count::ipa_p): New member function.
7986 (profile_count::operator<): Handle global zero correctly.
7987 (profile_count::operator>): Handle global zero correctly.
7988 (profile_count::operator<=): Handle global zero correctly.
7989 (profile_count::operator>=): Handle global zero correctly.
7990 (profile_count::nonzero_p): New member function.
7991 (profile_count::force_nonzero): New member function.
7992 (profile_count::max): New member function.
7993 (profile_count::apply_scale): Handle IPA scalling.
7994 (profile_count::guessed_local): New member function.
7995 (profile_count::global0): New member function.
7996 (profile_count::ipa): New member function.
7997 (profile_count::to_frequency): Declare.
7998 (profile_count::to_cgraph_frequency): Declare.
7999 * profile.c (OVERLAP_BASE): Delete.
8000 (compute_frequency_overlap): Delete.
8001 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8002 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8003 * sched-ebb.c (rank): Use counts only.
8004 * shrink-wrap.c (handle_simple_exit): Use counts only.
8005 (try_shrink_wrapping): Use counts only.
8006 (place_prologue_for_one_component): Use counts only.
8007 * tracer.c (find_best_predecessor): Use to_frequency.
8008 (find_trace): Use to_frequency.
8009 (tail_duplicate): Use to_frequency.
8010 * trans-mem.c (expand_transaction): Do not update frequency.
8011 * tree-call-cdce.c: Do not update frequency.
8012 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8013 (gimple_merge_blocks): Likewise.
8014 (gimple_split_edge): Likewise.
8015 (gimple_duplicate_sese_region): Likewise.
8016 (gimple_duplicate_sese_tail): Likewise.
8017 (move_sese_region_to_fn): Likewise.
8018 (gimple_account_profile_record): Likewise.
8019 (insert_cond_bb): Likewise.
8020 * tree-complex.c (expand_complex_div_wide): Likewise.
8021 * tree-eh.c (lower_resx): Update profile.
8022 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8024 (initialize_cfun): Do not initialize frequencies
8025 (freqs_to_counts): Delete.
8026 (copy_cfg_body): Ignore count parameter.
8027 (copy_body): Update.
8028 (expand_call_inline): Update count_max.
8029 (optimize_inline_calls): Update count_max.
8030 (tree_function_versioning): Update count_max.
8031 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8032 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8034 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8035 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8036 (try_peel_loop): Likewise.
8037 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8039 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8040 (tree_transform_and_unroll_loop): Do not use frequencies
8041 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8042 Use reliable prediction only.
8043 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8044 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8045 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8046 probability scaling.
8047 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8048 not update frequency
8049 (any_remaining_duplicated_blocks): Likewise.
8050 (update_profile): Likewise.
8051 (estimated_freqs_path): Delete.
8052 (freqs_to_counts_path): Delete.
8053 (clear_counts_path): Delete.
8054 (ssa_fix_duplicate_block_edges): Likewise.
8055 (duplicate_thread_path): Likewise.
8056 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8057 * tree-tailcall.c (decrease_profile): Do not update frequency.
8058 (eliminate_tail_call): Likewise.
8059 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8060 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8061 (optimize_mask_stores): Likewise.
8062 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8063 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8064 (ubsan_expand_ptr_ifn): Update profile.
8065 * value-prof.c (gimple_ic): Simplify.
8066 * value-prof.h (gimple_ic): Update prototype.
8067 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8068 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8070 (want_inline_self_recursive_call_p): Likewise.
8071 (resolve_noninline_speculation): Only cummulate defined counts.
8072 (inline_small_functions): Use nonzero_p.
8073 (ipa_inline): Do not access freed node.
8075 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8077 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8078 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8080 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8082 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8083 non-legitimate address.
8085 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8087 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8088 (*lt0_<mode>di, *lt0_<mode>si): New.
8090 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8092 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8093 TARGET_PAIRED_FLOAT.
8095 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8096 Jim Wilson <jim.wilson@linaro.org>
8098 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8099 * config/aarch64/aarch64-tune.md: Regenerated.
8100 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8101 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8103 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8105 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8106 (arc_expand_prologue): Restore blink for millicode.
8107 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8109 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8112 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8113 gen_vec_duplicate after forcing the scalar into a register.
8115 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8117 * combine (try_combine): Print the insns input to try_combine to the
8120 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8123 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8124 Remove second argument from aarch64_process_target_attr call.
8125 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8127 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8129 (aarch64_handle_attr_arch): Remove second argument.
8130 (aarch64_handle_attr_cpu): Ditto.
8131 (aarch64_handle_attr_tune): Ditto.
8132 (aarch64_handle_attr_isa_flags): Ditto.
8133 (aarch64_process_one_target_attr): Ditto.
8134 (aarch64_process_target_attr): Ditto.
8135 (aarch64_option_valid_attribute_p): Remove second argument.
8136 on aarch64_process_target_attr call.
8138 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8140 * diagnostic.c: Include "selftest-diagnostic.h".
8141 (selftest::assert_location_text): New function.
8142 (selftest::test_diagnostic_get_location_text): New function.
8143 (selftest::diagnostic_c_tests): Call it.
8145 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8147 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8148 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8149 (class selftest::test_diagnostic_context): Move to...
8150 * selftest-diagnostic.c: New file.
8151 * selftest-diagnostic.h: New file.
8153 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8155 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8156 offset range for FT32B.
8157 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8158 * config/ft32/ft32.md: Add TARGET_NOPM.
8159 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8160 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8162 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8164 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8166 2017-11-02 Jeff Law <law@redhat.com>
8168 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8169 virtual keyword on FINAL OVERRIDE members.
8171 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8172 virtuals together. Add virtual destructor.
8173 (substitute_and_fold_engine): Similarly.
8175 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8177 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8179 2017-11-02 Richard Biener <rguenther@suse.de>
8181 PR tree-optimization/82795
8182 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8184 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8186 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8188 (gcc_GAS_CHECK_FEATURE): Remove.
8189 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8190 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8191 * configure: Regenerate.
8193 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8195 * config/arc/arc.c (hwloop_optimize): Account for empty
8198 2017-11-02 Richard Biener <rguenther@suse.de>
8201 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8202 Truncate ARRAY_REF index and element size.
8204 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8206 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8208 2017-11-01 Jeff Law <law@redhat.com>
8210 * tree-ssa-ccp.c (ccp_folder): New class derived from
8211 substitute_and_fold_engine.
8212 (ccp_folder::get_value): New member function.
8213 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8214 (ccp_fold_stmt): Remove prototype.
8215 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8216 * tree-ssa-copy.c (copy_folder): New class derived from
8217 substitute_and_fold_engine.
8218 (copy_folder::get_value): Renamed from get_value.
8219 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8220 * tree-vrp.c (vrp_folder): New class derived from
8221 substitute_and_fold_engine.
8222 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8223 (vrp_folder::get_value): New member function.
8224 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8225 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8226 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8227 provide a class interface to folder/substitute routines.
8228 (ssa_prop_fold_stmt_fn): Remove typedef.
8229 (ssa_prop_get_value_fn): Likewise.
8230 (subsitute_and_fold): Remove prototype.
8231 (replace_uses_in): Likewise.
8232 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8233 Renamed from replace_uses_in. Call the virtual member function
8234 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8235 (substitute_and_fold_dom_walker): Remove initialization of
8236 data member entries for calbacks. Add substitute_and_fold_engine
8237 member and initialize it.
8238 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8239 member functions for get_value, replace_phi_args_in c
8240 replace_uses_in, and fold_stmt calls.
8241 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8242 substitute_and_fold. Remove assert. Update ctor call.
8244 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8245 (ssa_prop_visit_phi_fn): Likewise.
8246 (class ssa_propagation_engine): New class to provide an interface
8248 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8250 (ssa_prop_visit_phi): Likewise.
8251 (ssa_propagation_engine::simulate_stmt): Moved into class.
8252 Call visit_phi/visit_stmt from the class rather than via
8253 file scoped static variables.
8254 (ssa_propagation_engine::simulate_block): Moved into class.
8255 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8256 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8257 set file scoped statics for the visit_stmt/visit_phi callbacks.
8258 * tree-complex.c (complex_propagate): New class derived from
8259 ssa_propagation_engine.
8260 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8261 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8262 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8264 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8265 ssa_propagation_engine.
8266 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8267 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8268 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8269 * tree-ssa-copy.c (copy_prop): New class derived from
8270 ssa_propagation_engine.
8271 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8272 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8273 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8274 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8275 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8276 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8277 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8279 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8281 PR rtl-optimization/82778
8282 PR rtl-optimization/82597
8283 * compare-elim.c (struct comparison): Add in_a_setter field.
8284 (find_comparison_dom_walker::before_dom_children): Remove killed
8285 bitmap and df_simulate_find_defs call, instead walk the defs.
8286 Compute last_setter and initialize in_a_setter. Merge definitions
8287 with first initialization for a few variables.
8288 (try_validate_parallel): Use insn_invalid_p instead of
8289 recog_memoized. Return insn rather than just the pattern.
8290 (try_merge_compare): Fix up comment. Don't uselessly test if
8291 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8293 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8296 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8297 Alan Hayward <alan.hayward@arm.com>
8298 David Sherwood <david.sherwood@arm.com>
8300 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8301 aarch64_hard_regno_nregs to get the number of registers
8304 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8305 Alan Hayward <alan.hayward@arm.com>
8306 David Sherwood <david.sherwood@arm.com>
8308 * config/aarch64/constraints.md (Upl): Rename to...
8310 * config/aarch64/aarch64.md
8311 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8314 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8315 Alan Hayward <alan.hayward@arm.com>
8316 David Sherwood <david.sherwood@arm.com>
8318 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8319 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8322 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8323 Alan Hayward <alan.hayward@arm.com>
8324 David Sherwood <david.sherwood@arm.com>
8326 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8327 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8328 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8330 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8331 of a single element.
8332 * config/aarch64/iterators.md: Add a comment above the permute
8333 unspecs to say that they are generated directly by
8334 aarch64_expand_vec_perm_const.
8335 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8337 2017-11-01 Nathan Sidwell <nathan@acm.org>
8339 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8341 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8343 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8344 and medany code models, and describe what they do.
8346 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8348 Revert accidental duplicate:
8350 * combine.c (can_change_dest_mode): Reject changes in
8351 REGMODE_NATURAL_SIZE.
8353 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8355 PR rtl-optimization/64682
8356 PR rtl-optimization/69567
8357 PR rtl-optimization/69737
8358 PR rtl-optimization/82683
8359 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8360 register mentioned in the note, drop the note, unless it came from I3,
8361 in which case it should go to I3 again.
8363 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8365 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8366 and return false if not.
8367 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8369 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8371 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8372 range is known to be empty.
8374 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8375 Alan Hayward <alan.hayward@arm.com>
8376 David Sherwood <david.sherwood@arm.com>
8378 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8379 and CONST_VECTOR_NUNITS instead of computing the number of units from
8380 the byte sizes of the vector and element.
8381 (simplify_binary_operation_1): Likewise.
8382 (simplify_const_binary_operation): Likewise.
8383 (simplify_ternary_operation): Likewise.
8385 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8386 Alan Hayward <alan.hayward@arm.com>
8387 David Sherwood <david.sherwood@arm.com>
8389 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8390 (int_mem_offset): ...this new function.
8391 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8392 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8393 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8396 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8397 Alan Hayward <alan.hayward@arm.com>
8398 David Sherwood <david.sherwood@arm.com>
8400 * lower-subreg.c (interesting_mode_p): New function.
8401 (compute_costs, find_decomposable_subregs, decompose_register)
8402 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8403 (resolve_clobber, dump_choices): Use it.
8405 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8406 Alan Hayward <alan.hayward@arm.com>
8407 David Sherwood <david.sherwood@arm.com>
8409 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8411 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8412 Alan Hayward <alan.hayward@arm.com>
8413 David Sherwood <david.sherwood@arm.com>
8415 * alias.c (find_base_value, find_base_term): Only process integer
8416 truncations. Check the precision rather than the size.
8418 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8419 Alan Hayward <alan.hayward@arm.com>
8420 David Sherwood <david.sherwood@arm.com>
8422 * machmode.h (is_narrower_int_mode): New function
8423 * optabs.c (expand_float, expand_fix): Use it.
8424 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8426 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8427 Alan Hayward <alan.hayward@arm.com>
8428 David Sherwood <david.sherwood@arm.com>
8430 * rtl.h (narrower_subreg_mode): New function.
8431 * ira-color.c (update_costs_from_allocno): Use it.
8433 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8434 Alan Hayward <alan.hayward@arm.com>
8435 David Sherwood <david.sherwood@arm.com>
8437 * optabs-query.h (convert_optab_p): New function, split out from...
8438 (convert_optab_handler): ...here.
8439 (widening_optab_handler): Delete.
8440 (find_widening_optab_handler): Remove permit_non_widening parameter.
8441 (find_widening_optab_handler_and_mode): Likewise. Provide an
8442 override that operates on mode class wrappers.
8443 * optabs-query.c (widening_optab_handler): Delete.
8444 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8445 parameter. Assert that the two modes are the same class and that
8446 the "from" mode is narrower than the "to" mode. Use
8447 convert_optab_handler instead of widening_optab_handler.
8448 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8449 instead of widening_optab_handler.
8450 * expr.c (expand_expr_real_2): Update calls to
8451 find_widening_optab_handler.
8452 * optabs.c (expand_widen_pattern_expr): Likewise.
8453 (expand_binop_directly): Take the insn_code as a parameter.
8454 (expand_binop): Only call find_widening_optab_handler for
8455 conversion optabs; use optab_handler otherwise. Update calls
8456 to find_widening_optab_handler and expand_binop_directly.
8457 Use convert_optab_handler instead of widening_optab_handler.
8458 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8459 find_widening_optab_handler and use scalar_mode rather than
8461 (convert_plusminus_to_widen): Likewise.
8463 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8464 Alan Hayward <alan.hayward@arm.com>
8465 David Sherwood <david.sherwood@arm.com>
8467 * machmode.h (fixed_size_mode): New class.
8468 * rtl.h (get_pool_mode): Return fixed_size_mode.
8469 * gengtype.c (main): Add fixed_size_mode.
8470 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8471 (get_raw_arg_mode): Likewise.
8472 * doc/tm.texi: Regenerate.
8473 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8474 * targhooks.c (default_get_reg_raw_mode): Likewise.
8475 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8476 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8477 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8478 (msp430_get_raw_result_mode): Likewise.
8479 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8480 * dbxout.c (dbxout_parms): Require fixed-size modes.
8481 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8482 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8483 * omp-low.c (lower_oacc_reductions): Likewise.
8484 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8485 (simplify_subreg): Update accordingly.
8486 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8487 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8488 that aren't fixed-size.
8489 (get_pool_mode): Return a fixed_size_mode.
8490 (output_constant_pool_2): Take a fixed_size_mode.
8492 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8493 Alan Hayward <alan.hayward@arm.com>
8494 David Sherwood <david.sherwood@arm.com>
8496 * doc/rtl.texi (vec_series): Document.
8497 (const): Say that the operand can be a vec_series.
8498 * rtl.def (VEC_SERIES): New rtx code.
8499 * rtl.h (const_vec_series_p_1): Declare.
8500 (const_vec_series_p): New function.
8501 * emit-rtl.h (gen_const_vec_series): Declare.
8502 (gen_vec_series): Likewise.
8503 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8504 (gen_vec_series): Likewise.
8505 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8506 * simplify-rtx.c (simplify_unary_operation): Handle negations
8508 (simplify_binary_operation_series): New function.
8509 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8510 (test_vector_ops_series): New function.
8511 (test_vector_ops): Call it.
8512 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8513 gen_const_vec_series.
8514 (altivec_lvsr): Likewise.
8515 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8517 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8518 Alan Hayward <alan.hayward@arm.com>
8519 David Sherwood <david.sherwood@arm.com>
8521 * doc/rtl.texi (const): Update description of address constants.
8522 Say that vector constants are allowed too.
8523 * common.md (E, F): Use CONSTANT_P instead of checking for
8525 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8526 checking for CONST_VECTOR.
8527 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8529 * expr.c (expand_expr_real_2): Check for vector modes instead
8530 of checking for CONST_VECTOR.
8531 * rtl.h (const_vec_p): New function.
8532 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8533 (unwrap_const_vec_duplicate): Handle them here too.
8535 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8536 David Malcolm <dmalcolm@redhat.com>
8537 Alan Hayward <alan.hayward@arm.com>
8538 David Sherwood <david.sherwood@arm.com>
8540 * rtl.h (vec_duplicate_p): New function.
8541 * selftest-rtl.c (assert_rtx_eq_at): New function.
8542 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8543 (assert_rtx_eq_at): Declare.
8544 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8545 * selftest-run-tests.c (selftest::run_tests): Call it.
8546 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8547 (simplify_unary_operation_1): Recursively handle vector duplicates.
8548 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8550 (simplify_subreg): Handle subregs of vector duplicates.
8551 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8552 (selftest::simplify_rtx_c_tests): New functions.
8554 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8555 Alan Hayward <alan.hayward@arm.com>
8556 David Sherwood <david.sherwood@arm.com>
8558 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8559 (gen_vec_duplicate): Likewise.
8560 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8562 (gen_const_vector): ...here.
8563 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8564 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8565 whose elements are all equal.
8566 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8567 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8568 (simplify_relational_operation): Likewise.
8569 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8571 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8572 (aarch64_expand_vector_init): Likewise.
8573 * config/arm/arm.c (neon_vdup_constant): Likewise.
8574 (neon_expand_vector_init): Likewise.
8575 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8576 (arm_block_set_unaligned_vect): Likewise.
8577 (arm_block_set_aligned_vect): Likewise.
8578 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8579 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8580 (expand_vec_perm_even_odd_pack): Likewise.
8581 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8582 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8583 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8584 gen_const_vec_duplicate.
8585 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8586 * config/mips/mips.c (mips_gen_const_int_vector): Use
8587 gen_const_vec_duplicate.
8588 (mips_expand_vector_init): Use CONST0_RTX.
8589 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8590 (define_split): Use gen_const_vec_duplicate.
8591 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8592 (define_split): Use gen_const_vec_duplicate.
8593 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8594 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8595 * config/spu/spu.c (spu_const): Likewise.
8597 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8598 Alan Hayward <alan.hayward@arm.com>
8599 David Sherwood <david.sherwood@arm.com>
8601 * combine.c (can_change_dest_mode): Reject changes in
8602 REGMODE_NATURAL_SIZE.
8604 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8606 * configure.ac (--enable-libssp): New.
8607 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8608 trying to determine target-specific default. Adjust indentation.
8609 * configure: Regenerated.
8610 * doc/install.texi (Configuration): Expand --disable-libssp
8613 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8615 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8618 2017-10-31 Martin Jambor <mjambor@suse.cz>
8621 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8623 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8625 * auto-profile.c (autofdo_source_profile::read): Use
8626 UNKNOWN_LOCATION rather than 0.
8627 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8628 (warning_at): ...this overload.
8629 (warning_at_rich_loc_n): Rename to...
8630 (warning_n): ...this overload.
8631 (error_at_rich_loc): Rename to...
8632 (error_at): ...this overload.
8633 (pedwarn_at_rich_loc): Rename to...
8634 (pedwarn): ...this overload.
8635 (permerror_at_rich_loc): Rename to...
8636 (permerror): ...this overload.
8637 (inform_at_rich_loc): Rename to...
8638 (inform): ...this overload.
8639 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8640 (diagnostic_n_impl_richloc): Rename to...
8641 (diagnostic_n_impl): ...this rich_location *-based decl.
8642 (inform_at_rich_loc): Rename to...
8643 (inform): ...this, and add an assertion.
8644 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8645 (warning_at_rich_loc): Rename to...
8646 (warning_at): ...this, and add an assertion.
8647 (warning_at_rich_loc_n): Rename to...
8648 (warning_n): ...this, and add an assertion.
8649 (warning_n): Update location_t-based implementation for removal of
8650 location_t-based diagnostic_n_impl.
8651 (pedwarn_at_rich_loc): Rename to...
8652 (pedwarn): ...this, and add an assertion.
8653 (permerror_at_rich_loc): Rename to...
8654 (permerror): ...this, and add an assertion.
8655 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8656 (error_at_rich_loc): Rename to...
8657 (error_at): ...this, and add an assertion.
8658 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8659 (driver::do_spec_on_infiles): Likewise.
8660 * substring-locations.c (format_warning_va): Update for renaming
8661 of inform_at_rich_loc.
8663 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8665 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8666 _Float<N>X built-in functions so that the variant without the
8667 "__builtin_" prefix is only enabled for the GNU C and Objective C
8668 languages when they are in non-strict ANSI/ISO mode.
8669 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8670 * target.def (floatn_builtin_p): Add a target hook to control
8671 whether _Float<N> and _Float<N>X built-in functions without the
8672 "__builtin_" prefix are enabled, and return true for C and
8673 Objective C in the default hook. Include langhooks.h in
8675 * targhooks.h (default_floatn_builtin_p): Likewise.
8676 * targhooks.c (default_floatn_builtin_p): Likewise.
8677 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8678 floatn_builtin_p target hook.
8679 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8681 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8682 Eric Botcazou <ebotcazou@adacore.com>
8684 PR rtl-optimization/81803
8685 * lra-constraints.c (curr_insn_transform): Also reload the whole
8686 register for a strict subreg no wider than a word if this is for
8687 a WORD_REGISTER_OPERATIONS target.
8689 2017-10-31 Jason Merrill <jason@redhat.com>
8691 * gdbinit.in: Skip over inlines from timevar.h.
8693 2017-10-31 Martin Liska <mliska@suse.cz>
8695 * doc/gcov.texi: Document new option.
8696 * gcov.c (print_usage): Likewise print it.
8697 (process_args): Support the argument.
8698 (format_count): New function.
8699 (format_gcov): Use the function.
8701 2017-10-31 Martin Liska <mliska@suse.cz>
8703 * gcov.c (struct name_map): do not use typedef.
8704 Define operator== and operator<.
8705 (name_search): Remove.
8706 (name_sort): Remove.
8707 (main): Do not allocate names.
8708 (process_file): Add vertical space.
8709 (generate_results): Use std::find.
8710 (release_structures): Do not release memory.
8711 (find_source): Use std::find.
8713 2017-10-31 Martin Liska <mliska@suse.cz>
8715 * gcov.c (struct line_info): Remove it's typedef.
8716 (line_info::line_info): Add proper ctor.
8717 (line_info::has_block): Do not use a typedef.
8718 (struct source_info): Do not use typedef.
8719 (circuit): Likewise.
8720 (get_cycles_count): Likewise.
8721 (output_intermediate_file): Iterate via vector iterator.
8722 (add_line_counts): Use std::vector methods.
8723 (accumulate_line_counts): Likewise.
8724 (output_lines): Likewise.
8726 2017-10-31 Martin Liska <mliska@suse.cz>
8728 * gcov.c (struct source_info): Remove typedef.
8729 (source_info::source_info): Add proper ctor.
8730 (accumulate_line_counts): Use struct, not it's typedef.
8731 (output_gcov_file): Likewise.
8732 (output_lines): Likewise.
8733 (main): Do not allocate an array.
8734 (output_intermediate_file): Use size of vector container.
8735 (process_file): Resize the vector.
8736 (generate_results): Do not preallocate, use newly added vector
8738 (release_structures): Do not release sources.
8739 (find_source): Use vector methods.
8740 (add_line_counts): Do not use typedef.
8742 2017-10-31 Martin Liska <mliska@suse.cz>
8744 * doc/gcov.texi: Document that.
8745 * gcov.c (add_line_counts): Mark lines with a non-executed
8747 (output_line_beginning): Handle such lines.
8748 (output_lines): Pass new argument.
8749 (output_intermediate_file): Print it in intermediate format.
8751 2017-10-31 Martin Liska <mliska@suse.cz>
8753 * color-macros.h: New file.
8754 * diagnostic-color.c: Factor out color related to macros to
8756 * doc/gcov.texi: Document -k option.
8757 * gcov.c (INCLUDE_STRING): Include string.h.
8758 (print_usage): Add -k option.
8759 (process_args): Parse it.
8760 (pad_count_string): New function.
8761 (output_line_beginning): Likewise.
8762 (DEFAULT_LINE_START): New macro.
8763 (output_lines): Support color output.
8765 2017-10-31 Martin Liska <mliska@suse.cz>
8767 PR gcov-profile/82633
8768 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8769 their interaction with GCOV infrastructure.
8770 * configure.ac: Add -fkeep-{inline,static}-functions to
8772 * configure: Regenerate.
8774 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8777 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8779 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8782 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8783 into a register if it does not fit into an immediate offset field.
8785 2017-10-31 Olivier Hainque <hainque@adacore.com>
8787 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8789 2017-10-31 Julia Koval <julia.koval@intel.com>
8791 * config.gcc: Add gfniintrin.h.
8792 * config/i386/gfniintrin.h: New.
8793 * config/i386/i386-builtin-types.def
8794 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8795 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8796 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8797 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8798 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8799 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8800 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8801 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8802 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8803 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8804 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8805 * config/i386/immintrin.h: Include gfniintrin.h.
8806 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8808 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8810 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8812 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8814 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8815 (arm_ashldi3_1bit): Remove pattern.
8816 (ashrdi3): Remove shift by 1 expansion.
8817 (arm_ashrdi3_1bit): Remove pattern.
8818 (lshrdi3): Remove shift by 1 expansion.
8819 (arm_lshrdi3_1bit): Remove pattern.
8820 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8821 cost of ashldi3 by 1.
8822 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8823 (<shift>di3_neon): Likewise.
8825 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8827 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8828 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8829 and (*aarch64_simd_mov<VQ:mode>).
8830 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8831 pattern alternative.
8832 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8833 attributes to match pattern alternative.
8835 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8837 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8838 * config/rs6000/emmintrin.h: New file.
8839 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8841 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8843 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8844 * (movdi_vfp_cortexa8): Remove pattern.
8846 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8848 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8850 (Specific, alpha*-dec-osf5.1): Remove.
8851 (Specific, mips-sgi-irix5): Remove.
8852 (Specific, mips-sgi-irix6): Remove.
8854 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8857 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8858 arguments to clear_bit_region_be.
8860 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8862 * gimplify.c: Include memmodel.h.
8864 2017-10-30 Martin Jambor <mjambor@suse.cz>
8866 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8867 condition whether loop should be executed at all.
8869 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8872 gimple folding of vec_madd() intrinsics.
8873 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8874 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8875 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8877 2017-10-30 Richard Biener <rguenther@suse.de>
8879 PR tree-optimization/82762
8881 2017-10-23 Richard Biener <rguenther@suse.de>
8883 PR tree-optimization/82129
8885 2017-08-01 Richard Biener <rguenther@suse.de>
8887 PR tree-optimization/81181
8888 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8889 (compute_antic): ... end of iteration here.
8891 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8893 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8894 -std=iso9899:2017 and -std=gnu17.
8895 * doc/standards.texi (C Language): Document C17 support.
8896 * doc/cpp.texi (Overview): Mention -std=c17.
8897 (Standard Predefined Macros): Document C11 and C17 values of
8898 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8899 * doc/extend.texi (Inline): Do not list individual options for
8900 standards newer than C99.
8901 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8903 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8906 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8908 * asan.c (asan_finish_file): Align asan globals array by shadow
8911 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8914 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8915 (struct store_immediate_info): Add bitregion_start and bitregion_end
8917 (store_immediate_info::store_immediate_info): Add brs and bre
8918 arguments and initialize bitregion_{start,end} from those.
8919 (struct merged_store_group): Add bitregion_start, bitregion_end,
8920 align_base and mask fields. Drop unnecessary struct keyword from
8921 struct store_immediate_info. Add do_merge method.
8922 (clear_bit_region_be): Use memset instead of loop storing zeros.
8923 (merged_store_group::do_merge): New method.
8924 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8925 stores as long as the surrounding bitregions have no gaps.
8926 (merged_store_group::merge_overlapping): Use do_merge.
8927 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8928 is byte aligned, rather than requiring that start and width are
8929 byte aligned. Drop unnecessary struct keyword from
8930 struct store_immediate_info. Allocate and populate also mask array.
8931 Make start of the arrays relative to bitregion_start rather than
8932 start and size them according to bitregion_{end,start} difference.
8933 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8934 struct store_immediate_info.
8935 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8936 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8937 struct keyword from struct store_immediate_info.
8938 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8939 between stores as long as the surrounding bitregions have no gaps.
8941 (struct split_store): Add orig non-static data member.
8942 (split_store::split_store): Initialize orig to false.
8943 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8944 if there is exactly a single original stmt. Change stmts argument
8945 to pointer from reference, if NULL, don't push anything to it. Add
8946 first argument, use it to optimize skipping over orig stmts that
8947 are known to be before bitpos already. Simplify.
8948 (split_group): Return unsigned int count how many stores are or
8949 would be needed rather than a bool. Add allow_unaligned argument.
8950 Change split_stores argument from reference to pointer, if NULL,
8951 only do a dry run computing how many stores would be produced.
8952 Rewritten algorithm to use both alignment and misalign if
8953 !allow_unaligned and handle bitfield stores with gaps.
8954 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8955 from bitregion_start instead of start. Compute allow_unaligned
8956 here, if true, do 2 split_group dry runs to compute which one
8957 produces fewer stores and prefer aligned if equal. Punt if
8958 new count is bigger or equal than original before emitting any
8959 statements, rather than during that. Remove no longer needed
8960 new_ssa_names tracking. Replace num_stmts with
8961 split_stores.length (). Use 32-bit stack allocated entries
8962 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8963 if possible. Handle bitfields with gaps.
8964 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8965 Compute bitregion_{start,end} for the stores and construct
8966 store_immediate_info with that. Formatting fixes.
8968 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8971 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8972 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8974 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8976 * gimplify.c: Include tm_p.h.
8978 * common.opt (gcoff): Re-add as ignored option.
8979 (gcoff1, gcoff2, gcoff3): Likewise.
8981 * Makefile.in (OBJS): Delete sdbout.o.
8982 (GTFILES): Delete $(srcdir)/sdbout.c.
8983 * debug.h: Delete sdb_debug_hooks.
8984 * final.c: Delete sdbout.h include.
8985 (final_scan_insn): Delete SDB_DEBUG check.
8986 (rest_of_clean_state): Likewise.
8987 * output.h: Delete sdb_begin_function_line.
8990 * toplev.c: Delete sdbout.h include.
8991 (process_options): Delete SDB_DEBUG check.
8992 * tree-core.h (tree_type_common): Delete pointer field of
8994 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8995 TYPE_SYMTAB_POINTER.
8996 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8997 (TYPE_SYMTAB_IS_POINTER): Delete.
8998 (TYPE_SYMTAB_IS_DIE): Renumber.
8999 * xcoffout.c: Refer to former sdbout.c file.
9000 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9002 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9003 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9004 (Debugging Options): Delete -gcoff.
9005 (-gstabs): Delete SDB reference.
9007 (-gcoff@var{level}): Delete.
9008 * doc/passes.texi (Debugging information output): Delete SDB and
9009 sdbout.c references.
9010 * doc/tm.texi: Regenerate.
9011 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9012 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9014 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9015 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9016 (SDB_DEBUGGING_INFO): Delete.
9017 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9018 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9019 * target.def (output_source_filename): Delete COFF reference.
9021 * common.opt (gcoff): Delete.
9022 (gxcoff+): Update Negative chain.
9023 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9025 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9026 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9027 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9029 (expand_function_start): Change sdb reference to past tense.
9030 (expand_function_end): Change sdb reference to past tense.
9031 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9032 * opts.c (debug_type_names): Delete coff entry.
9033 (common_handle_option): Delete OPT_gcoff case.
9034 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9036 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9037 * config/cris/cris.h: Delete SDB reference in comment.
9038 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9039 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9040 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9041 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9043 (ix86_expand_prologue): Likewise.
9044 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9045 * config/ia64/ia64.h: Likewise.
9046 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9047 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9049 * config/mmix/mmix.h: Likewise.
9050 * config/nds32/nds32.c: Likewise.
9051 * config/stormy/storym16.h: Likewise.
9052 * config/visium/visium.h: Likewise.
9053 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9055 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9057 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9058 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9059 FRAME_POINTER_REGNUM point at high end of local var area.
9061 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9063 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9064 Move comment around. Do not reset best_edge for a copiable
9065 destination if the copy would cause a partition change.
9066 (better_edge_p): Remove redundant check.
9068 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9070 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9072 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9074 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9075 for math functions that have _Float<N> and _Float<N>X variants.
9076 (mathfn_built_in_2): Add support for math functions that have
9077 _Float<N> and _Float<N>X variants.
9078 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9079 (expand_builtin_mathfn_ternary): Add support for fma with
9080 _Float<N> and _Float<N>X variants.
9081 (expand_builtin): Likewise.
9082 (fold_builtin_3): Likewise.
9083 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9084 create math function _Float<N> and _Float<N>X variants as external
9086 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9087 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9088 the __builtin_ prefix and if not strict ansi, without the prefix.
9089 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9090 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9091 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9092 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9093 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9094 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9095 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9096 function signatures for fma _Float<N> and _Float<N>X variants.
9097 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9098 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9099 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9100 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9101 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9102 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9103 * gencfn-macros.c (print_case_cfn): Add support for math functions
9104 that have _Float<N> and _Float<N>X variants.
9105 (print_define_operator_list): Likewise.
9106 (fltfn_suffixes): Likewise.
9108 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9109 for math functions that have _Float<N> and _Float<N>X variants.
9110 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9111 and _Float<N>X variants.
9112 (COPYSIGN): Likewise.
9115 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9116 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9118 (integer_valued_read_call_p): Likewise.
9119 * fold-const-call.c (fold_const_call_ss): Likewise.
9120 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9121 _Float<N> and _Float<N>X variants.
9122 (fold_const_call_ssss): Add support for fma _Float<N> and
9123 _Float<N>X variants.
9124 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9125 support for copysign and fma _Float<N> and _Float<N>X variants.
9126 (backprop::process_builtin_call_use): Likewise.
9127 * tree-call-cdce.c (can_test_argument_range); Add support for
9128 sqrt _Float<N> and _Float<N>X variants.
9129 (edom_only_function): Likewise.
9130 (get_no_error_domain): Likewise.
9131 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9132 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9133 copysign _Float<N> and _Float<N>X variants.
9134 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9135 handled by machine independent code.
9136 (FMAF128): Likewise.
9137 * doc/cpp.texi (Common Predefined Macros): Document defining
9138 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9139 fma _Float<N> and _Float<N>X variants.
9141 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9144 * config/i386/i386-modes.def (CCFPU): Remove definition.
9145 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9146 (ix86_cc_modes_compatible): Ditto.
9147 (ix86_expand_carry_flag_compare): Ditto.
9148 (ix86_expand_int_movcc): Ditto.
9149 (ix86_expand_int_addcc): Ditto.
9150 (ix86_reverse_condition): Ditto.
9151 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9152 Return true/false for unordered/ordered fp comparisons.
9153 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9154 (ix86_prepare_fp_compare_args): Update for rename.
9155 (ix86_expand_fp_compare): Update for rename. Generate unordered
9156 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9157 (ix86_expand_sse_compare_and_jump): Ditto.
9158 * config/i386/predicates.md (fcmov_comparison_operator):
9159 Remove CCFPU mode handling.
9160 (ix86_comparison_operator): Ditto.
9161 (ix86_carry_flag_operator): Ditto.
9162 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9163 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9164 (*cmpu<mode>_cc_i387): Ditto.
9165 (FPCMP): Remove mode iterator.
9166 (unord): Remove mode attribute.
9167 (unord_subst): New define_subst transformation
9168 (unord): New define_subst attribute.
9170 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9171 (*cmpi<unord>xf_i387): Ditto.
9172 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9173 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9174 using unord_subst transformation.
9175 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9176 (round_saeonly): Also handle CCFP mode.
9177 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9178 Remove UNSPEC_SAHF unspec handling.
9180 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9182 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9184 2017-10-27 Jeff Law <law@redhat.com>
9186 * gimple-ssa-sprintf.c: Include domwalk.h.
9187 (class sprintf_dom_walker): New class, derived from dom_walker.
9188 (sprintf_dom_walker::before_dom_children): New function.
9189 (struct call_info): Moved into sprintf_dom_walker class
9190 (compute_formath_length, handle_gimple_call): Likewise.
9191 (sprintf_length::execute): Call the dominator walker rather
9192 than walking the statements.
9194 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9195 gimple statement locations.
9196 (check_array_bounds): Corresponding changes. Get the statement's
9197 location directly from wi->stmt.
9199 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9202 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9204 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9206 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9207 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9209 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9212 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9213 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9214 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9215 maybe_get_pool_constant.
9216 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9219 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9221 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9222 2.26 caveat. Update gas and gld versions.
9223 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9226 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9228 * cgraph.h (set_malloc_flag): Declare.
9229 * cgraph.c (set_malloc_flag_1): New function.
9230 (set_malloc_flag): Likewise.
9231 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9232 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9234 (read_ipa_call_summary): Add support for reading is_return_callee.
9235 (write_ipa_call_summary): Stream is_return_callee.
9236 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9237 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9238 ipa-prop.h, ipa-fnsummary.h.
9239 (pure_const_names): Change to static.
9240 (malloc_state_e): Define.
9241 (malloc_state_names): Define.
9242 (funct_state_d): Add field malloc_state.
9243 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9244 (check_retval_uses): New function.
9245 (malloc_candidate_p): Likewise.
9246 (analyze_function): Add support for malloc attribute.
9247 (pure_const_write_summary): Stream malloc_state.
9248 (pure_const_read_summary): Add support for reading malloc_state.
9249 (dump_malloc_lattice): New function.
9250 (propagate_malloc): New function.
9251 (warn_function_malloc): New function.
9252 (ipa_pure_const::execute): Call propagate_malloc and
9253 ipa_free_fn_summary.
9254 (pass_local_pure_const::execute): Add support for malloc attribute.
9255 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9256 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9258 2017-10-27 Martin Liska <mliska@suse.cz>
9260 PR gcov-profile/82457
9261 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9262 for fork-like functions to be properly instrumented.
9264 2017-10-27 Richard Biener <rguenther@suse.de>
9267 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9268 info when we redirected EH.
9270 2017-10-26 Michael Collison <michael.collison@arm.com>
9272 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9274 (<optab>_trunchf<GPI:mode>2: New pattern.
9275 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9276 * config/aarch64/iterators.md (wv): New mode attribute.
9277 (vf, VF): New mode attributes.
9278 (vgp, VGP): New mode attributes.
9279 (s): Update attribute with SImode and DImode prefixes.
9281 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9283 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9284 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9285 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9286 (nios2_option_overide): Initialize it. Don't allow R0-relative
9287 addressing with PIC.
9288 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9289 (nios2_symbolic_constant_p): Likewise.
9290 (nios2_legitimate_address_p): Likewise.
9291 (nios2_r0rel_section_name_p): New.
9292 (nios2_symbol_ref_in_r0rel_data_p): New.
9293 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9294 (r0rel_constant_p): New.
9295 (nios2_print_operand_address): Handle r0rel_constant_p.
9296 (nios2_cdx_narrow_form_p): Likewise.
9297 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9298 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9299 (Nios II Options): Document -mr0rel-sec.
9301 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9303 * config/nios2/nios2.c: Include xregex.h.
9304 (nios2_gprel_sec_regex): New.
9305 (nios2_option_overide): Initialize it. Don't allow GP-relative
9306 addressing with PIC.
9307 (nios2_small_section_name_p): Check for regex match.
9308 * config/nios2/nios2.opt (mgprel-sec=): New option.
9309 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9310 (Nios II Options): Document -mgprel-sec.
9312 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9314 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9316 2017-10-26 Tom de Vries <tom@codesourcery.com>
9318 PR tree-optimization/82707
9319 * gimple.c (gimple_copy): Fix unsharing of
9320 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9322 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9324 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9325 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9326 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9327 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9328 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9329 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9330 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9331 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9332 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9333 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9334 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9335 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9336 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9337 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9338 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9339 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9340 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9342 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9344 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9346 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9347 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9348 warning to rs6000.c. Remove the Undocumented flag, since it has
9350 (-mabi=ibmlongdouble): Likewise.
9351 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9352 already set, set the default format for long double.
9353 (rs6000_debug_reg_global): Print whether long double is IBM or
9355 (rs6000_option_override_internal): Rework setting long double
9356 format. Only warn if the user is changing the long double default
9357 and they did not use -Wno-psabi.
9358 * doc/invoke.texi (PowerPC options): Update the documentation for
9359 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9361 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9362 Alan Hayward <alan.hayward@arm.com>
9363 David Sherwood <david.sherwood@arm.com>
9365 * rtl.h (wider_subreg_mode): New function.
9366 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9367 rather than an unsigned int *.
9368 * ira-color.c (regno_max_ref_width): Replace with...
9369 (regno_max_ref_mode): ...this new variable.
9370 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9371 Use wider_subreg_mode.
9372 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9373 rather than an unsigned int *.
9374 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9375 (process_alt_operands): Likewise.
9376 (invariant_p): Likewise.
9377 * lra-spills.c (assign_mem_slot): Likewise.
9378 (add_pseudo_to_slot): Likewise.
9379 * lra.c (collect_non_operand_hard_regs): Likewise.
9380 (add_regs_to_insn_regno_info): Likewise.
9381 * reload1.c (regno_max_ref_width): Replace with...
9382 (regno_max_ref_mode): ...this new variable.
9383 (reload): Update accordingly. Update call to
9384 ira_sort_regnos_for_alter_reg.
9385 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9386 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9387 (scan_paradoxical_subregs): Likewise.
9389 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9391 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9392 (aarch64_frame): Add emit_frame_chain boolean.
9393 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9394 Move eh_return case to aarch64_layout_frame.
9395 (aarch64_layout_frame): Initialize emit_frame_chain.
9396 (aarch64_expand_prologue): Use emit_frame_chain.
9398 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9400 * config/aarch64/aarch64.c (aarch64_layout_frame):
9401 Ensure LR is always stored at the bottom of the callee-saves.
9402 Remove rarely used frame layout which saves callee-saves at top of
9403 frame, so the store of LR can be used as a valid probe in all cases.
9405 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9407 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9408 Improve unaligned TImode/TFmode base/offset split.
9410 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9411 Alan Hayward <alan.hayward@arm.com>
9412 David Sherwood <david.sherwood@arm.com>
9414 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9415 * combine.c (find_single_use_1): Likewise.
9416 (expand_field_assignment): Likewise.
9417 (move_deaths): Likewise.
9418 * lra-constraints.c (simplify_operand_subreg): Likewise.
9419 (curr_insn_transform): Likewise.
9420 * lra.c (collect_non_operand_hard_regs): Likewise.
9421 (add_regs_to_insn_regno_info): Likewise.
9422 * rtlanal.c (reg_referenced_p): Likewise.
9423 (covers_regno_no_parallel_p): Likewise.
9425 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9427 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9428 Don't print any bits outside the precision of the value.
9429 * wide-int.cc (test_printing): Add some new tests.
9431 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9433 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9434 supports -xbrace_comment option.
9435 * configure: Regenerate.
9436 * config.in: Regenerate.
9437 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9438 (ASM_CPU_SPEC): Use it.
9440 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9442 * target.def (static_rtx_alignment): New hook.
9443 * targhooks.h (default_static_rtx_alignment): Declare.
9444 * targhooks.c (default_static_rtx_alignment): New function.
9445 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9446 * doc/tm.texi: Regenerate.
9447 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9448 instead of targetm.constant_alignment. Remove call to
9450 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9451 (cris_preferred_mininum_alignment): New function, split out from...
9452 (cris_constant_alignment): ...here.
9453 (cris_static_rtx_alignment): New function.
9454 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9456 (ix86_constant_alignment): ...here.
9457 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9458 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9459 (mmix_static_rtx_alignment): New function.
9460 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9461 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9463 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9466 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9467 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9469 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9471 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9472 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9473 mark as a sign-extending load.
9474 (local_pic_load_u): Define.
9476 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9479 * fold-const.c (operand_equal_for_comparison_p): Also return true
9480 if ARG0 is a simple variant of ARG1 with narrower precision.
9481 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9483 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9485 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9487 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9488 scatter_static, scatter_per_elt.
9489 * x86-tune-costs.h: Add new cost entries.
9491 2017-10-25 Richard Biener <rguenther@suse.de>
9493 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9494 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9495 class pass_fre): Move to ...
9496 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9497 class pass_fre): ... here and adjust for statistics.
9499 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9502 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9503 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9504 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9505 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9508 2017-10-25 Richard Biener <rguenther@suse.de>
9510 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9511 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9512 eliminate_push_avail, eliminate_insert): Move inside...
9513 (class eliminate_dom_walker): ... this class in preparation
9515 (fini_eliminate): Remove by merging with ...
9516 (eliminate): ... this function. Adjust for class changes.
9517 (pass_pre::execute): Remove fini_eliminate call.
9518 (pass_fre::execute): Likewise.
9520 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9523 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9524 (VPERMI2, VPERMI2I): New mode iterators.
9525 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9526 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9528 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9529 mode iterator. Remove 3 old define_insn patterns.
9530 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9531 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9532 VPERMI2 mode iterator, remove the other two expanders.
9533 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9534 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9535 instructions, remove the other two patterns.
9536 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9537 mode iterator, remove the other two patterns.
9538 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9539 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9540 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9541 and adjust argument order accordingly.
9542 (ix86_expand_vec_perm): Adjust caller.
9543 (expand_vec_perm_1): Likewise.
9544 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9545 (expand_vec_perm_vpermt2_vpshub2): ... this.
9546 (ix86_expand_vec_perm_const_1): Adjust caller.
9547 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9550 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9551 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9552 (vec_shl_<mode>): Remove unused expander.
9553 (avx512bw_<shift_insn><mode>3): New define_insn.
9554 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9555 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9557 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9560 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9563 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9565 PR rtl-optimization/82396
9566 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9567 (autopref_multipass_init): Simplify initialization.
9568 (autopref_rank_data): Simplify sort order.
9569 * gcc/sched-int.h (autopref_multipass_data_): Remove
9570 multi_mem_insn_p, min_offset and max_offset.
9572 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9575 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9576 Check special value of flag_omit_frame_pointer.
9577 (aarch64_can_eliminate): Likewise.
9578 (aarch64_override_options_after_change_1): Simplify handling of
9579 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9581 2017-10-24 Richard Biener <rguenther@suse.de>
9583 PR tree-optimization/82697
9584 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9585 zero for conditional load and unconditional store.
9587 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9589 * doc/install.texi: Document bootstrap-cet.
9591 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9594 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9595 ENDBR instruction at function entrance if function is only
9598 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9601 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9602 patterns to better describe from which operation the CF is computed.
9603 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9604 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9605 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9606 is 0, use _0 suffixed expanders instead of emitting a comparison
9609 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9611 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9612 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9614 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9617 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9618 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9619 * loop-iv.c (iv_get_reaching_def): Likewise.
9620 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9621 variable is promoted and the partition contains undefined values.
9623 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9625 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9627 (nios2_address_cost): Define.
9628 (nios2_legitimize_address): Recognize (exp + constant) directly.
9629 (TARGET_ADDRESS_COST): Define.
9631 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9633 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9634 (nios2_symbolic_memory_operand_p): Declare.
9635 (nios2_split_large_constant): Declare.
9636 (nios2_split_symbolic_memory_operand): Declare.
9637 * config/nios2/nios2.c: Adjust includes.
9638 (nios2_symbolic_constant_allowed): New.
9639 (nios2_symbolic_constant_p): New.
9640 (nios2_plus_symbolic_constant_p): New.
9641 (nios2_valid_addr_expr_p): Recognize addresses involving
9643 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9644 (nios2_symbolic_memory_operand_p): New.
9645 (nios2_large_constant_p): New.
9646 (nios2_split_large_constant): New.
9647 (nios2_split_plus_large_constant): New.
9648 (nios2_split_symbolic_memory_operand): New.
9649 (nios2_legitimize_address): Code refactoring. Handle addresses
9650 involving symbolic constants.
9651 (nios2_emit_move_sequence): Likewise.
9652 (nios2_print_operand): Improve error output.
9653 (nios2_print_operand_address): Handle LO_SUM.
9654 (nios2_cdx_narrow_form_p): Likewise.
9655 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9656 operands involving symbolic constants.
9657 (movhi_internal, movsi_internal): Likewise.
9658 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9659 (extendhisi2, extendqi<mode>2): Likewise.
9661 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9663 * tree-pass.h (PROP_rtl_split_insns): Define.
9664 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9666 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9668 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9670 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9673 * target.def (const_not_ok_for_debug_p): Default to
9674 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9675 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9676 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9677 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9678 which targetm.const_not_ok_for_debug_p returned true.
9679 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9681 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9683 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9684 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9685 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9687 (ix86_const_not_ok_for_debug_p): New function.
9688 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9689 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9691 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9694 * system.h: Conditionally include "unique-ptr.h" if
9695 INCLUDE_UNIQUE_PTR is defined.
9696 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9697 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9699 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9701 * config/rl78/rl78.md: New define_expand "subdi3".
9703 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9706 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9707 DF_REF_INSN if DF_REF_INSN_INFO is false.
9709 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9711 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9712 xmm_move instead of sse_move.
9713 (sse_store_index): New function.
9714 (ix86_register_move_cost): Be more sensible about mismatch stall;
9715 model AVX moves correctly; make difference between sse->integer and
9717 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9718 moves; make difference between SSE and AVX.
9719 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9720 and zmm_move. Increase size of sse load and store tables;
9721 add unaligned load and store tables; add ssemmx_to_integer.
9722 * x86-tune-costs.h: Update all entries according to real
9723 move latencies from Agner Fog's manual and chip documentation.
9725 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9728 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9729 * config/i386/constraints.md (Wf): New constraint.
9730 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9731 (cmp<dwi>_doubleword): Removed.
9732 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9733 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9734 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9735 expand with cmp<dwi>_doubleword. For LTU and GEU use
9736 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9738 * common.opt (gcolumn-info): Enable by default.
9739 * doc/invoke.texi (gcolumn-info): Document new default.
9741 2017-10-23 Richard Biener <rguenther@suse.de>
9743 PR tree-optimization/82672
9744 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9745 Fold the stmt if we propagated into it.
9747 2017-10-23 Richard Biener <rguenther@suse.de>
9749 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9750 (bitmap_remove_expr_from_set): ... this. All callers call this
9751 for non-constant values.
9752 (bitmap_set_subtract): Rename to...
9753 (bitmap_set_subtract_expressions): ... this. Adjust and
9755 (bitmap_set_contains_value): Remove superfluous check.
9756 (bitmap_set_replace_value): Inline into single caller ...
9757 (bitmap_value_replace_in_set): ... here and simplify.
9758 (dependent_clean): Merge into ...
9759 (clean): ... this using an overload. Adjust.
9760 (prune_clobbered_mems): Adjust.
9761 (compute_antic_aux): Likewise.
9762 (compute_partial_antic_aux): Likewise.
9764 2017-10-23 Richard Biener <rguenther@suse.de>
9766 PR tree-optimization/82129
9768 2017-08-01 Richard Biener <rguenther@suse.de>
9770 PR tree-optimization/81181
9771 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9772 (compute_antic): ... end of iteration here.
9774 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9776 * target.def (starting_frame_offset): New hook.
9777 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9778 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9779 * doc/tm.texi.in: Regenerate.
9780 * hooks.h (hook_hwi_void_0): Declare.
9781 * hooks.c (hook_hwi_void_0): New function.
9782 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9783 STARTING_FRAME_OFFSET.
9784 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9785 * reload1.c (reload): Likewise.
9786 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9787 instead of STARTING_FRAME_OFFSET.
9788 * function.c (try_fit_stack_local): Likewise.
9789 (assign_stack_local_1): Likewise
9790 (instantiate_virtual_regs): Likewise.
9791 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9792 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9793 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9794 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9795 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9796 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9797 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9798 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9799 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9800 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9801 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9802 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9803 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9804 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9805 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9806 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9807 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9808 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9809 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9810 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9811 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9812 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9813 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9814 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9815 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9816 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9817 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9818 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9819 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9820 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9821 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9822 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9823 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9824 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9825 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9826 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9827 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9828 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9829 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9830 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9831 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9832 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9833 return a HOST_WIDE_INT.
9834 (avr_builtin_setjmp_frame_value): Use it instead of
9835 STARTING_FRAME_OFFSET.
9836 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9837 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9838 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9840 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9841 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9842 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9843 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9844 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9845 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9846 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9847 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9848 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9849 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9850 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9851 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9853 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9854 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9855 * config/mips/mips.c (mips_compute_frame_info): Refer to
9856 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9857 (mips_starting_frame_offset): New function.
9858 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9859 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9860 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9861 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9862 and return a HOST_WIDE_INT.
9863 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9864 (mmix_initial_elimination_offset): Refer to
9865 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9866 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9867 * config/pa/pa.c (pa_starting_frame_offset): New function.
9868 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9869 (pa_expand_prologue): Likewise.
9870 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9871 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9872 !FRAME_GROWS_DOWNWARD handling to...
9873 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9874 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9875 !FRAME_GROWS_DOWNWARD handling to...
9876 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9877 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9878 !FRAME_GROWS_DOWNWARD handling to...
9879 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9880 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9882 (rs6000_starting_frame_offset): New function.
9883 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9884 !FRAME_GROWS_DOWNWARD handling to...
9885 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9886 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9887 !FRAME_GROWS_DOWNWARD handling to...
9888 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9889 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9890 !FRAME_GROWS_DOWNWARD handling to...
9891 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9892 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9893 (rs6000_starting_frame_offset): New function.
9894 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9895 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9896 * config/vax/vax.c (vax_starting_frame_offset): New function.
9897 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9898 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9899 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9900 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9901 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9902 * system.h (STARTING_FRAME_OFFSET): Poison.
9904 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9906 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9907 SCALAR_TYPE_MODE instead of TYPE_MODE.
9909 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9910 Alan Hayward <alan.hayward@arm.com>
9911 David Sherwood <david.sherwood@arm.com>
9913 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9915 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9916 Alan Hayward <alan.hayward@arm.com>
9917 David Sherwood <david.sherwood@arm.com>
9919 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9921 2017-10-23 Richard Biener <rguenther@suse.de>
9923 PR tree-optimization/82129
9924 * tree-ssa-pre.c (bitmap_set_and): Remove.
9925 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9926 canonicalizing expressions in the set to those with lowest
9927 ID rather than taking that from the first edge.
9929 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9931 * combine.c (rtx_equal_for_field_assignment_p): Use
9932 byte_lowpart_offset.
9934 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9935 Alan Hayward <alan.hayward@arm.com>
9936 David Sherwood <david.sherwood@arm.com>
9938 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9939 to a SUBREG_PROMOTED_VAR.
9941 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9942 Alan Hayward <alan.hayward@arm.com>
9943 David Sherwood <david.sherwood@arm.com>
9945 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9946 (expand_debug_source_expr): Likewise.
9947 * combine.c (combine_simplify_rtx): Likewise.
9948 * cse.c (fold_rtx): Likewise.
9949 * optabs.c (expand_float): Likewise.
9950 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9951 (simplify_binary_operation_1): Likewise.
9953 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9954 Alan Hayward <alan.hayward@arm.com>
9955 David Sherwood <david.sherwood@arm.com>
9957 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9958 (record_promoted_value): Likewise.
9959 * expr.c (expand_expr_real_2): Likewise.
9960 * ree.c (update_reg_equal_equiv_notes): Likewise.
9961 (combine_set_extension): Likewise.
9962 * rtlanal.c (low_bitmask_len): Likewise.
9963 * simplify-rtx.c (neg_const_int): Likewise.
9964 (simplify_binary_operation_1): Likewise.
9966 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9967 Alan Hayward <alan.hayward@arm.com>
9968 David Sherwood <david.sherwood@arm.com>
9970 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9971 * regcprop.c (maybe_mode_change): Likewise.
9972 * reload1.c (alter_reg): Likewise.
9974 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9976 * inchash.h (inchash::hash::add_wide_int): New function.
9977 * lto-streamer-out.c (hash_tree): Use it.
9979 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9981 * inchash.h (inchash::hash::add_wide_int): Rename to...
9982 (inchash::hash::add_hwi): ...this.
9983 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9984 (polymorphic_call_target_hasher::hash): Likewise.
9985 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9986 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9987 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9988 * lto-streamer-out.c (hash_tree): Likewise.
9989 * optc-save-gen.awk: Likewise.
9990 * tree.c (add_expr): Likewise.
9992 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9995 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9996 for ordered inequality comparisons even with TARGET_IEEE_FP.
9998 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10001 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10002 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10003 Expand with cmp<dwi>_doubleword.
10005 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10007 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10008 List CET intrinsics.
10009 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10010 specific to -fcf-protection option.
10012 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10014 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10015 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10016 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10017 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10018 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10019 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10020 (extra_objs): Add cet.o for Linux/x86 targets.
10021 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10022 * config/i386/cet.c: New file.
10023 * config/i386/cetintrin.h: Likewise.
10024 * config/i386/t-cet: Likewise.
10025 * config/i386/cpuid.h (bit_SHSTK): New.
10026 (bit_IBT): Likewise.
10027 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10028 pass IBT and SHSTK bits.
10029 * config/i386/i386-builtin-types.def
10030 (VOID_FTYPE_UNSIGNED_PVOID): New.
10031 (VOID_FTYPE_UINT64_PVOID): Likewise.
10032 * config/i386/i386-builtin.def: Add CET intrinsics.
10033 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10034 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10035 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10036 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10038 * config/i386/i386.c (rest_of_insert_endbranch): New.
10039 (pass_data_insert_endbranch): Likewise.
10040 (pass_insert_endbranch): Likewise.
10041 (make_pass_insert_endbranch): Likewise.
10042 (ix86_notrack_prefixed_insn_p): Likewise.
10043 (ix86_target_string): Add -mibt, -mshstk flags.
10044 (ix86_option_override_internal): Add flag_cf_protection
10046 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10047 (ix86_print_operand): Add 'notrack' prefix output.
10048 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10049 (ix86_expand_builtin): Expand CET intrinsics.
10050 (x86_output_mi_thunk): Add 'endbranch' instruction.
10051 * config/i386/i386.h (TARGET_IBT): New.
10052 (TARGET_IBT_P): Likewise.
10053 (TARGET_SHSTK): Likewise.
10054 (TARGET_SHSTK_P): Likewise.
10055 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10056 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10057 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10058 (builtin_setjmp_setup): New pattern.
10059 (builtin_longjmp): Likewise.
10060 (rdssp<mode>): Likewise.
10061 (incssp<mode>): Likewise.
10062 (saveprevssp): Likewise.
10063 (rstorssp): Likewise.
10064 (wrss<mode>): Likewise.
10065 (wruss<mode>): Likewise.
10066 (setssbsy): Likewise.
10067 (clrssbsy): Likewise.
10068 (nop_endbr): Likewise.
10069 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10071 * config/i386/immintrin.h: Include <cetintrin.h>.
10072 * config/i386/linux-common.h
10073 (file_end_indicate_exec_stack_and_cet): New prototype.
10074 (TARGET_ASM_FILE_END): New.
10076 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10078 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10079 latencies instead of having separate table; make difference between
10080 integer and float costs.
10081 * i386.h (processor_costs): Remove scalar_stmt_cost,
10082 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10083 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10085 * x86-tune-costs.h: Remove entries which has been removed in
10086 procesor_costs from all tables; make cond_taken_branch_cost
10087 and cond_not_taken_branch_cost COST_N_INSNS based.
10089 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10091 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10093 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10095 * config/i386/i386.md (isa): Remove fma_avx512f.
10096 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10097 <avx512>_fmadd_<mode>_mask3<round_name>,
10098 <avx512>_fmsub_<mode>_mask<round_name>,
10099 <avx512>_fmsub_<mode>_mask3<round_name>,
10100 <avx512>_fnmadd_<mode>_mask<round_name>,
10101 <avx512>_fnmadd_<mode>_mask3<round_name>,
10102 <avx512>_fnmsub_<mode>_mask<round_name>,
10103 <avx512>_fnmsub_<mode>_mask3<round_name>,
10104 <avx512>_fmaddsub_<mode>_mask<round_name>,
10105 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10106 <avx512>_fmsubadd_<mode>_mask<round_name>,
10107 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10108 (*vec_widen_umult_even_v16si<mask_name>,
10109 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10110 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10112 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10114 * extend.texi: Add 'nocf_check' documentation.
10115 * gimple.texi: Add second parameter to
10116 gimple_build_call_from_tree.
10117 * invoke.texi: Add -fcf-protection documentation.
10118 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10120 2017-10-20 Richard Biener <rguenther@suse.de>
10122 PR tree-optimization/82473
10123 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10124 the largest input type.
10126 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10128 * c-attribs.c (handle_nocf_check_attribute): New function.
10129 (c_common_attribute_table): Add 'nocf_check' handling.
10130 * gimple-parser.c: Add second argument NULL to
10131 gimple_build_call_from_tree.
10132 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10133 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10135 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10136 * common.opt: Add fcf-protection flag.
10137 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10138 * flag-types.h: Add enum cf_protection_level.
10139 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10140 Add 'nocf_check' attribute propagation to gimple call.
10141 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10142 (gimple_build_call_from_tree): Update prototype.
10143 (gimple_call_nocf_check_p): New function.
10144 (gimple_call_set_nocf_check): Likewise.
10145 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10146 * ipa-icf.c: Add nocf_check attribute in statement hash.
10147 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10148 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10149 * toplev.c (process_options): Add flag_cf_protection handling.
10151 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10153 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10155 2017-10-20 Richard Biener <rguenther@suse.de>
10157 PR tree-optimization/82603
10158 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10159 remove false predicated stores.
10161 2017-10-20 Richard Biener <rguenther@suse.de>
10163 * graphite-isl-ast-to-gimple.c
10164 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10165 Remove return value and simplify, dump copied stmt after lhs
10167 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10168 Reduce dump verbosity.
10169 (gsi_insert_earliest): Likewise.
10170 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10171 * graphite.c (print_global_statistics): Adjust dumping.
10172 (print_graphite_scop_statistics): Likewise.
10173 (print_graphite_statistics): Do not dump loops here.
10174 (graphite_transform_loops): But here.
10176 2017-10-20 Nicolas Roche <roche@adacore.com>
10178 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10179 * configure: Regenerate.
10181 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10184 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10185 functions when optimizing replace GIMPLE_RETURN stmts with
10186 calls to __builtin_unreachable ().
10189 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10190 for -fsanitize=thread link of executables.
10191 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10192 link of executables.
10195 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10196 New mode iterators.
10197 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10198 define_insns for logical vector shifts to use VI248_AVX512BW
10199 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10200 condition, useless isa and prefix attributes. Change the first
10201 2 of these define_insns to ...
10202 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10203 define_insn for avx512vl.
10204 (<shift_insn><mode>3): ... and this, new define_insn without
10205 masking for non-avx512vl.
10208 * config/i386/sse.md (*andnot<mode>3,
10209 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10210 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10211 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10212 not applied use empty suffix even for TARGET_AVX512VL.
10213 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10214 is applied, supply evex,evex or evex,evex,evex instead of just
10217 2017-10-20 Julia Koval <julia.koval@intel.com>
10219 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10220 (OPTION_MASK_ISA_GFNI_UNSET): New.
10221 (ix86_handle_option): Handle OPT_mgfni.
10222 * config/i386/cpuid.h (bit_GFNI): New.
10223 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10224 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10225 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10226 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10227 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10228 * config/i386/i386.opt: Add mgfni.
10230 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10232 * config/msp430/msp430.c (msp430_option_override): Disable
10233 -fdelete-null-pointer-checks.
10234 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10236 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10238 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10239 of x87 and SSE instructions.
10241 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10243 * asan.c (create_cond_insert_point): Do not update edge count.
10244 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10245 (afdo_propagate_circuit): Likewise.
10246 (afdo_calculate_branch_prob): Likewise.
10247 (afdo_annotate_cfg): Likewise.
10248 * basic-block.h (struct edge_def): Remove count.
10249 (edge_def::count): New accessor.
10250 * bb-reorder.c (rotate_loop): Update.
10251 (find_traces_1_round): Update.
10252 (connect_traces): Update.
10253 (sanitize_hot_paths): Update.
10254 * cfg.c (unchecked_make_edge): Update.
10255 (make_single_succ_edge): Update.
10256 (check_bb_profile): Update.
10257 (dump_edge_info): Update.
10258 (update_bb_profile_for_threading): Update.
10259 (scale_bbs_frequencies_int): Update.
10260 (scale_bbs_frequencies_gcov_type): Update.
10261 (scale_bbs_frequencies_profile_count): Update.
10262 (scale_bbs_frequencies): Update.
10263 * cfganal.c (connect_infinite_loops_to_exit): Update.
10264 * cfgbuild.c (compute_outgoing_frequencies): Update.
10265 (find_many_sub_basic_blocks): Update.
10266 * cfgcleanup.c (try_forward_edges): Update.
10267 (try_crossjump_to_edge): Update
10268 * cfgexpand.c (expand_gimple_cond): Update
10269 (expand_gimple_tailcall): Update
10270 (construct_exit_block): Update
10271 * cfghooks.c (verify_flow_info): Update
10272 (redirect_edge_succ_nodup): Update
10273 (split_edge): Update
10274 (make_forwarder_block): Update
10275 (duplicate_block): Update
10276 (account_profile_record): Update
10277 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10278 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10279 * cfgloopmanip.c (scale_loop_profile): Update.
10281 (lv_adjust_loop_entry_edge): Update.
10282 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10283 (force_nonfallthru_and_redirect): Update.
10284 (purge_dead_edges): Update.
10285 (rtl_flow_call_edges_add): Update.
10286 * cgraphunit.c (init_lowered_empty_function): Update.
10287 (cgraph_node::expand_thunk): Update.
10288 * gimple-pretty-print.c (dump_probability): Update.
10289 (dump_edge_probability): Update.
10290 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10291 * haifa-sched.c (sched_create_recovery_edges): Update.
10292 * hsa-gen.c (convert_switch_statements): Update.
10293 * ifcvt.c (dead_or_predicable): Update.
10294 * ipa-inline-transform.c (inline_transform): Update.
10295 * ipa-split.c (split_function): Update.
10296 * ipa-utils.c (ipa_merge_profiles): Update.
10297 * loop-doloop.c (add_test): Update.
10298 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10299 * lto-streamer-in.c (input_cfg): Update.
10300 (input_function): Update.
10301 * lto-streamer-out.c (output_cfg): Update.
10302 * modulo-sched.c (sms_schedule): Update.
10303 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10304 * predict.c (maybe_hot_edge_p): Update.
10305 (unlikely_executed_edge_p): Update.
10306 (probably_never_executed_edge_p): Update.
10307 (dump_prediction): Update.
10308 (drop_profile): Update.
10309 (propagate_unlikely_bbs_forward): Update.
10310 (determine_unlikely_bbs): Update.
10311 (force_edge_cold): Update.
10312 * profile.c (compute_branch_probabilities): Update.
10313 * reg-stack.c (better_edge): Update.
10314 * shrink-wrap.c (handle_simple_exit): Update.
10315 * tracer.c (better_p): Update.
10316 * trans-mem.c (expand_transaction): Update.
10317 (split_bb_make_tm_edge): Update.
10318 * tree-call-cdce.c: Update.
10319 * tree-cfg.c (gimple_find_sub_bbs): Update.
10320 (gimple_split_edge): Update.
10321 (gimple_duplicate_sese_region): Update.
10322 (gimple_duplicate_sese_tail): Update.
10323 (gimple_flow_call_edges_add): Update.
10324 (insert_cond_bb): Update.
10325 (execute_fixup_cfg): Update.
10326 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10327 * tree-complex.c (expand_complex_div_wide): Update.
10328 * tree-eh.c (lower_resx): Update.
10329 (unsplit_eh): Update.
10330 (cleanup_empty_eh_move_lp): Update.
10331 * tree-inline.c (copy_edges_for_bb): Update.
10332 (freqs_to_counts): Update.
10333 (copy_cfg_body): Update.
10334 * tree-ssa-dce.c (remove_dead_stmt): Update.
10335 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10336 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10337 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10338 (unloop_loops): Update.
10339 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10340 * tree-ssa-loop-split.c (connect_loops): Update.
10341 (split_loop): Update.
10342 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10343 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10344 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10345 * tree-ssa-reassoc.c (branch_fixup): Update.
10346 * tree-ssa-tail-merge.c (replace_block_by): Update.
10347 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10348 (compute_path_counts): Update.
10349 (update_profile): Update.
10350 (recompute_probabilities): Update.
10351 (update_joiner_offpath_counts): Update.
10352 (estimated_freqs_path): Update.
10353 (freqs_to_counts_path): Update.
10354 (clear_counts_path): Update.
10355 (ssa_fix_duplicate_block_edges): Update.
10356 (duplicate_thread_path): Update.
10357 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10358 (case_bit_test_cmp): Update.
10359 (collect_switch_conv_info): Update.
10360 (gen_inbound_check): Update.
10361 (do_jump_if_equal): Update.
10362 (emit_cmp_and_jump_insns): Update.
10363 * tree-tailcall.c (decrease_profile): Update.
10364 (eliminate_tail_call): Update.
10365 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10366 (vect_do_peeling): Update.
10367 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10368 * ubsan.c (ubsan_expand_null_ifn): Update.
10369 (ubsan_expand_ptr_ifn): Update.
10370 * value-prof.c (gimple_divmod_fixed_value): Update.
10371 (gimple_mod_pow2): Update.
10372 (gimple_mod_subtract): Update.
10373 (gimple_ic): Update.
10374 (gimple_stringop_fixed_value): Update.
10376 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10379 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10381 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10383 PR rtl-optimization/82395
10384 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10385 based on non_spilled_static_chain_regno_p.
10387 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10389 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10390 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10391 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10393 2017-10-19 Martin Sebor <msebor@redhat.com>
10395 PR tree-optimization/82596
10396 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10398 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10400 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10401 (get_mem_refs_of_builtin_call): Likewise.
10402 * builtins.c (expand_builtin_apply): Adjust call to
10403 allocate_dynamic_stack_space.
10404 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10405 the third argument to allocate_dynamic_stack_space, otherwise -1.
10406 (expand_builtin): Deal with all alloca variants.
10407 (is_inexpensive_builtin): Likewise.
10408 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10409 * calls.c (special_function_p): Deal with all alloca variants.
10410 (initialize_argument_information): Adjust call to
10411 allocate_dynamic_stack_space.
10412 (expand_call): Likewise.
10413 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10414 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10415 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10416 use it for the stack usage computation.
10417 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10418 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10419 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10420 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10421 (in_loop_p): Remove first argument and useless check.
10422 (pass_walloca::execute): Remove useless test and adjust call to above.
10423 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10424 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10425 (gimplify_call_expr): Deal with all alloca variants.
10426 * hsa-gen.c (gen_hsa_alloca): Likewise.
10427 (gen_hsa_insns_for_call): Likewise.
10428 * ipa-pure-const.c (special_builtin_state): Likewise.
10429 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10430 * tree-object-size.c (alloc_object_size): Likewise.
10431 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10432 (call_may_clobber_ref_p_1): Likewise.
10433 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10434 (ccp_fold_stmt): Likewise.
10435 (optimize_stack_restore): Likewise.
10436 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10437 (mark_all_reaching_defs_necessary_1): Likewise.
10438 (propagate_necessity): Likewise.
10439 (eliminate_unnecessary_stmts): Likewise.
10440 * tree.c (build_common_builtin_nodes): Build
10441 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10442 (build_alloca_call_expr): New function.
10443 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10444 (CASE_BUILT_IN_ALLOCA): Likewise.
10445 (build_alloca_call_expr): Declare.
10446 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10448 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10451 * dwarf2out.c (new_die_raw): New static inline function.
10452 (new_die): Use it to create the DIE.
10453 (add_AT_external_die_ref): Likewise.
10454 (clone_die): Likewise.
10455 (clone_as_declaration): Likewise.
10456 (dwarf2out_vms_debug_main_pointer): Likewise.
10457 (base_type_die): Likewise. Remove early return for corner cases.
10458 Do not call add_pubtype on the DIE here.
10459 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10460 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10461 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10462 native order exists for base types, attach the DIE manually and call
10463 add_pubtype on it. Do not equate a reverse order DIE to the type.
10465 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10467 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10468 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10470 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10472 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10474 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10475 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10476 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10477 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10478 * opts.c (sanitizer_opts): Add builtin.
10479 * ubsan.c (instrument_builtin): New function.
10480 (pass_ubsan::execute): Call it.
10481 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10482 * doc/invoke.texi: Document -fsanitize=builtin.
10484 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10485 builtins, store max (log2 (align), 0) into uchar field instead of
10486 align into uptr field.
10487 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10488 store uchar 0 field instead of uptr 0 field.
10489 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10490 instead of passing one address of struct with 2 locations pass
10491 two addresses of structs with 1 location each.
10492 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10493 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10494 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10495 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10496 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10497 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10498 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10499 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10501 2017-10-19 Martin Liska <mliska@suse.cz>
10504 * file-find.c (remove_prefix): Remove.
10505 * file-find.h (remove_prefix): Likewise.
10506 * gcc-ar.c: Remove smartness of lookup.
10508 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10510 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10511 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10512 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10514 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10517 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10518 (setcc + and to xor + setcc): New peephole2.
10520 2017-10-19 Tom de Vries <tom@codesourcery.com>
10522 * doc/sourcebuild.texi (Test Directives, Variants of
10523 dg-require-support): Add dg-require-stack-size.
10525 2017-10-19 Martin Liska <mliska@suse.cz>
10528 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10529 that have a large alignment.
10530 (gimplify_target_expr): Likewise.
10532 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10534 PR rtl-optimization/82602
10535 * ira.c (rtx_moveable_p): Return false for volatile asm.
10537 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10540 * config/i386/i386-modes.def (CCGZ): New CC mode.
10541 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10542 * config/i386/predicates.md (ix86_comparison_operator):
10544 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10545 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10546 with double-word subtraction.
10547 (put_condition_code): Handle CCGZmode.
10549 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10551 * wide-int.cc (debug (const wide_int &)): New.
10552 (debug (const wide_int *)): New.
10553 (debug (const widest_int &)): New.
10554 (debug (const widest_int *)): New.
10556 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10558 PR middle-end/82556
10559 * lra-constraints.c (curr_insn_transform): Use non-input operand
10560 instead of output one for matched reload.
10562 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10564 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10565 (tree-ssa-loop-ivopts.h): New header file.
10566 (struct builtin_info): New fields.
10567 (classify_builtin_1): Compute and record base and offset parts for
10568 memset builtin partition by calling strip_offset.
10569 (offset_cmp, fuse_memset_builtins): New functions.
10570 (finalize_partitions): Fuse adjacent memset partitions by calling
10572 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10573 Expose the interface.
10574 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10576 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10578 PR tree-optimization/82574
10579 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10580 that data reference must be executed exactly once per iteration
10581 against the outermost loop in nest.
10582 (classify_partition): Update call to above function.
10584 2017-10-18 Richard Biener <rguenther@suse.de>
10586 PR tree-optimization/82591
10587 * graphite.c (graphite_transform_loops): Move code gen message
10589 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10590 Here. Handle scop_to_isl_ast failing.
10591 (scop_to_isl_ast): Limit the number of ISL operations.
10593 2017-10-18 Richard Biener <rguenther@suse.de>
10595 * graphite-isl-ast-to-gimple.c
10596 (translate_isl_ast_to_gimple::set_rename): Simplify.
10597 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10598 (graphite_copy_stmts_from_block): ... here.
10599 (copy_bb_and_scalar_dependences): Simplify.
10600 (add_parameters_to_ivs_params): Canonicalize.
10601 (generate_entry_out_of_ssa_copies): Simplify.
10602 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10603 by passing in ISL dimension.
10604 (parameter_index_in_region_1): Rename to ...
10605 (parameter_index_in_region): ... this.
10606 (extract_affine): Adjust assert, pass down parameter index.
10607 (add_param_constraints): Use range-info when available.
10608 (build_scop_context): Adjust.
10609 * sese.c (new_sese_info): Adjust.
10610 (free_sese_info): Likewise.
10611 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10612 Remove unused typedefs.
10613 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10615 2017-10-18 Martin Liska <mliska@suse.cz>
10617 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10619 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10621 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10622 (s390_sched_init): Do not reset s390_sched_state if we entered the
10623 current basic block via a fallthru edge and all others are unlikely.
10625 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10627 * config/s390/s390.c (NUM_SIDES): New variable.
10628 (LONGRUNNING_THRESHOLD): New variable.
10629 (LATENCY_FACTOR): New variable.
10630 (s390_sched_score): Decrease score for long-running instructions on
10632 (s390_sched_variable_issue): Perform bookkeeping for long-running
10635 2017-10-18 Richard Biener <rguenther@suse.de>
10637 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10638 Simplify with removal of the parameter rename map.
10639 (set_rename): Likewise.
10640 (should_copy_to_new_region): Likewise.
10641 (graphite_copy_stmts_from_block): Likewise.
10642 (copy_bb_and_scalar_dependences): Remove initialization of
10643 unused copied_bb_map.
10644 (copy_def): Remove.
10645 (copy_internal_parameters): Likewise.
10646 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10647 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10648 Use INTEGRAL_TYPE_P.
10649 (parameter_index_in_region_1): Rename to ...
10650 (assign_parameter_index_in_region): ... this. Assert we have
10651 a parameter we handle.
10652 (scan_tree_for_params): Adjust.
10653 * sese.h (parameter_rename_map_t): Remove.
10654 (struct sese_info_t): Remove unused parameter_rename_map and
10655 copied_bb_map members.
10656 * sese.c (new_sese_info): Adjust.
10657 (free_sese_info): Likewise.
10659 2017-10-18 Martin Liska <mliska@suse.cz>
10662 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10663 on an abnormal edge.
10665 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10667 * doc/invoke.texi (ffunction-sections and fdata-sections):
10670 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10672 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10673 the use statement can throw internally.
10675 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10677 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10678 any RTX present on the RHS of a SET.
10679 * compare-elim.c (try_eliminate_compare): Restore comment.
10681 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10683 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10684 may be also called on tcc_type nodes.
10685 * langhooks.c (lhd_tree_size): Likewise.
10687 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10689 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10690 format_warning_at_substring.
10691 (maybe_warn): Convert source_range * param to a location_t. Pass
10692 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10693 (format_directive): Remove code to extract source_ranges and
10694 source_range * in favor of just a location_t.
10695 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10697 * substring-locations.c (format_warning_va): Convert
10698 source_range * param to a location_t.
10699 (format_warning_at_substring): Likewise.
10700 * substring-locations.h (format_warning_va): Likewise.
10701 (format_warning_at_substring): Likewise.
10703 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10705 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10707 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10708 and scatter/gather ops.
10710 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10711 vec_gather_load and vec_scatter_store.
10712 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10713 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10714 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10715 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10716 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10717 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10719 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10721 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10722 Detect FTST instruction and handle its register pops. Only pop
10723 second operand if can_pop_second_op is true.
10724 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10725 set can_pop_second_op to false in the compare_for_stack_reg call.
10727 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10728 output_fp_compare for stack register operands.
10729 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10730 instructions here. Do not emit stack register pops here. Assert
10731 that FCOMPP pops next to top stack register. Rewrite function.
10733 2017-10-17 Nathan Sidwell <nathan@acm.org>
10735 PR middle-end/82577
10736 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10737 use DECL_ASSEMBLER_NAME_RAW.
10739 PR middle-end/82546
10740 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10743 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10744 Wilco Dijkstra <wilco.dijkstra@arm.com>
10746 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10747 converstion to Pmode from the buf_addr.
10749 2017-10-17 Richard Biener <rguenther@suse.de>
10751 * graphite-dependences.c (scop_get_reads_and_writes): Change
10752 output parameters to references.
10754 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10756 PR 71026/tree-optimization
10757 * fold-const.c (distribute_real_division): Removed.
10758 (fold_binary_loc): Remove calls to distribute_real_divison.
10760 2017-10-17 Richard Biener <rguenther@suse.de>
10762 * graphite-scop-detection.c
10763 (scop_detection::stmt_has_simple_data_refs_p): Always use
10764 the full nest as region.
10765 (try_generate_gimple_bb): Likewise.
10766 * sese.c (scalar_evolution_in_region): Simplify now that
10767 SCEV can handle instantiation in regions.
10768 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10769 in the non-loop part of a function if requested.
10771 2017-10-17 Richard Biener <rguenther@suse.de>
10773 PR tree-optimization/82563
10774 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10776 (graphite_regenerate_ast_isl): Call it.
10777 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10779 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10781 PR tree-optimization/82549
10782 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10783 Formatting fixes. Instead of calling make_bit_field_ref with negative
10786 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10788 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10789 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10790 _mm_maskz_reduce_ss): New.
10791 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10792 __builtin_ia32_reducess_mask): Ditto..
10793 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10794 * config/i386/sse.md (reduces<mode>): Renamed to ...
10795 (reduces<mode><mask_scalar_name>): ... this.
10796 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10798 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10799 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10801 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10803 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10804 * selftest-run-tests.c (selftest::run_tests): Call
10805 selftest::unique_ptr_tests_cc_tests.
10806 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10807 * unique-ptr-tests.cc: New file.
10809 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10812 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10814 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10815 (make_hard_regno_born, make_hard_regno_dead): Update
10816 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10818 2017-10-16 Jeff Law <law@redhat.com>
10820 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10822 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10824 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10826 2017-10-16 Olivier Hainque <hainque@adacore.com>
10828 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10829 with_cpu if we were configured for an e500v2 target cpu name.
10831 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10833 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10834 * doc/invoke.texi: Document +nodsp as a valid extension for
10837 2017-10-16 Martin Liska <mliska@suse.cz>
10839 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10840 (test_set_range): Likewise.
10841 (test_range_functions): Rename to ...
10842 (test_bit_in_range): ... this.
10843 (sbitmap_c_tests): Add new test.
10845 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10847 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10849 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10850 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10852 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10854 * config/aarch64/aarch64-builtins.c
10855 (aarch64_types_quadopu_lane_qualifiers): New.
10856 (TYPES_QUADOPU_LANE): New.
10857 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10858 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10859 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10860 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10861 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10862 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10863 (Vdottype, DOTPROD): New.
10864 (sur): Add SDOT and UDOT.
10866 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10868 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10869 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10870 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10871 Add TARGET_DOTPROD.
10872 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10873 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10874 Enable TARGET_DOTPROD.
10875 (cortex-a75.cortex-a55): Likewise.
10876 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10878 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10880 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10881 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10883 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10885 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10886 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10887 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10888 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10889 * config/arm/types.md (neon_dot, neon_dot_q): New.
10890 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10892 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10894 * config/arm/arm.h (TARGET_DOTPROD): New.
10895 * config/arm/arm.c (arm_arch_dotprod): New.
10896 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10897 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10898 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10899 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10900 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10901 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10902 * doc/invoke.texi (armv8.2-a): Document dotprod
10904 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10906 * i386.c (ix86_vec_cost): New function.
10907 (ix86_rtx_costs): Handle vector operations better.
10908 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10909 * x86-tune-costs.h: Add new costs to all tables.
10911 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10913 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10915 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10916 divsd, sqrtss and sqrtsd
10917 * x86-tune-costs.h: Add new entries to all costs.
10918 (znver1_cost): Fix to match real instruction latencies.
10920 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10921 Michael Collison <michael.collison@arm.com>
10923 * compare-elim.c: Include emit-rtl.h.
10924 (can_merge_compare_into_arith): New function.
10925 (try_validate_parallel): Likewise.
10926 (try_merge_compare): Likewise.
10927 (try_eliminate_compare): Call the above when no previous clobber
10929 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10932 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10934 PR middle-end/62263
10935 PR middle-end/82498
10936 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10937 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10939 PR middle-end/62263
10940 PR middle-end/82498
10941 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10942 to be any operand_equal_p operands. For & (B - 1) require
10943 B to be power of 2. Recognize
10944 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10946 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10949 * optabs.c (expand_memory_blockage): Fix call of
10950 targetm.have_memory_blockage.
10952 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10955 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10956 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10957 objects to extra_objs instead of overwriting it.
10959 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10961 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10962 Use any_fp_register_operand as operand[3] predicate. Simplify
10963 equality test for operands[2] and operands[4] memory location.
10964 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10965 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10966 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10967 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10968 any_fp_register_operand as operand[1] predicate. Simplify
10969 equality test for operands[0] and operands[3] memory location.
10970 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10971 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10972 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10974 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10976 * target-insns.def: Add memory_blockage.
10977 * optabs.c (expand_memory_blockage): New function.
10978 (expand_asm_memory_barrier): Rename ...
10979 (expand_asm_memory_blockage): ... to this.
10980 (expand_mem_thread_fence): Call expand_memory_blockage
10981 instead of expand_asm_memory_barrier.
10982 (expand_mem_singnal_fence): Ditto.
10983 (expand_atomic_load): Ditto.
10984 (expand_atomic_store): Ditto.
10985 * doc/md.texi (Standard Pattern Names For Generation):
10986 Document memory_blockage instruction pattern.
10988 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10990 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10991 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10992 * config/rl78/rl78.md: New define_expand "adddi3".
10994 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10996 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10999 2017-10-13 Jeff Law <law@redhat.com>
11001 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11003 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11006 * internal-fn.c (expand_mul_overflow): If both operands have
11007 the same highpart of -1 or 0 and the topmost bit of lowpart
11008 is different, overflow is if res <= 0 rather than res < 0.
11010 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11012 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11013 TARGET_P9_VECTOR code for unaligned_load case.
11015 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11017 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11019 2017-10-13 Nathan Sidwell <nathan@acm.org>
11021 * tree-core.h (tree_contains_struct): Make bool.
11022 * tree.c (tree_contains_struct): Likewise.
11023 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11024 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11025 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11026 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11027 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11029 2017-10-13 Richard Biener <rguenther@suse.de>
11031 * graphite-isl-ast-to-gimple.c
11032 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11033 parameters and dominance check.
11034 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11035 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11036 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11037 Do not update SSA form here or do intermediate IL verification.
11038 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11039 (graphite_initialize): Remove check on the number of loops in
11040 the function and inline into graphite_transform_loops.
11041 (graphite_finalize): Inline into graphite_transform_loops.
11042 (graphite_transform_loops): Perform SSA update and IL verification
11044 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11046 2017-10-13 Richard Biener <rguenther@suse.de>
11048 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11049 graphite_expression_type_precision): Avoid global constructor
11051 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11052 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11053 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11054 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11055 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11056 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11058 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11061 * config/i386/i386.h (ix86_red_zone_size): New.
11062 * config/i386/i386.md (push peephole2s): Replace
11063 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11065 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11066 Alan Hayward <alan.hayward@arm.com>
11067 David Sherwood <david.sherwood@arm.com>
11069 * combine.c (can_change_dest_mode): Reject changes in
11070 REGMODE_NATURAL_SIZE.
11072 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11073 Alan Hayward <alan.hayward@arm.com>
11074 David Sherwood <david.sherwood@arm.com>
11076 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11077 (expand_debug_source_expr): Likewise.
11078 * combine.c (combine_simplify_rtx): Likewise.
11079 * cse.c (fold_rtx): Likewise.
11080 * fwprop.c (canonicalize_address): Likewise.
11081 * targhooks.c (default_shift_truncation_mask): Likewise.
11083 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11084 Alan Hayward <alan.hayward@arm.com>
11085 David Sherwood <david.sherwood@arm.com>
11087 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11088 (widened_mode): Likewise.
11089 (expand_unop): Likewise.
11090 * ree.c (transform_ifelse): Likewise.
11091 (merge_def_and_ext): Likewise.
11092 (combine_reaching_defs): Likewise.
11093 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11095 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11096 Alan Hayward <alan.hayward@arm.com>
11097 David Sherwood <david.sherwood@arm.com>
11099 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11100 * combine.c (gen_lowpart_for_combine): Likewise.
11101 * dwarf2out.c (rtl_for_decl_location): Likewise.
11102 * final.c (alter_subreg): Likewise.
11103 * rtlhooks.c (gen_lowpart_general): Likewise.
11104 (gen_lowpart_if_possible): Likewise.
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 * calls.c (expand_call): Use subreg_lowpart_offset.
11111 * cse.c (cse_insn): Likewise.
11112 * regcprop.c (copy_value): Likewise.
11113 (copyprop_hardreg_forward_1): Likewise.
11115 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11118 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11119 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11120 =Q constraints to +Q and into insn condition add check
11121 that operands[0] and operands[1] are equal.
11122 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11123 =Q constraints to +Q and into insn condition add check
11124 that operands[0] is equal to either operands[1] or operands[2].
11127 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11128 instead of handling MINUS_EXPR twice (once for each argument),
11129 canonicalize operand order and handle just once, use rtype where
11130 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11133 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11134 any values of __C while still being pattern recognizable as a simple
11135 rotate instruction.
11137 2017-10-13 Richard Biener <rguenther@suse.de>
11139 PR tree-optimization/82451
11141 2017-10-02 Richard Biener <rguenther@suse.de>
11143 PR tree-optimization/82355
11144 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11145 a mapping for the enclosing loop but avoid generating one for
11146 the loop tree root.
11147 (copy_bb_and_scalar_dependences): Remove premature codegen
11148 error on PHIs in blocks duplicated into multiple places.
11149 * graphite-scop-detection.c
11150 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11151 in the region use it as loop and nest to analyze the DR in.
11152 (try_generate_gimple_bb): Likewise.
11153 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11154 (add_loop_constraints): For blocks in a loop not in the region
11155 create a dimension with a single iteration.
11156 * sese.h (gbb_loop_at_index): Remove assert.
11158 * cfgloop.c (loop_preheader_edge): For the loop tree root
11159 return the single successor of the entry block.
11160 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11161 Reset the SCEV hashtable and niters.
11162 * graphite-scop-detection.c
11163 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11164 assert that we only have POLYNOMIAL_CHREC that vary in loops
11165 contained in the region.
11166 (scop_detection::graphite_can_represent_expr): Adjust.
11167 (scop_detection::stmt_has_simple_data_refs_p): For loops
11168 not in the region set loop to NULL. The nest is now the
11169 entry edge to the region.
11170 (try_generate_gimple_bb): Likewise.
11171 * sese.c (scalar_evolution_in_region): Adjust for
11172 instantiate_scev change.
11173 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11174 Make nest parameter the edge into the region.
11175 (create_data_ref): Likewise.
11176 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11177 entry edge into a region and adjust instantiate_scev calls.
11178 (create_data_ref): Likewise.
11179 (graphite_find_data_references_in_stmt): Likewise.
11180 (find_data_references_in_stmt): Pass the loop preheader edge
11181 from the nest argument.
11182 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11183 parameter the edge into the region.
11184 (instantiate_parameters): Use the loop preheader edge as entry.
11185 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11187 (get_instantiated_value_entry): Make instantiate_below parameter
11188 the edge into the region.
11189 (instantiate_scev_name): Likewise. Adjust dominance checks,
11190 when we cannot use loop-based instantiation instantiate by
11191 walking use-def chains.
11192 (instantiate_scev_poly): Adjust.
11193 (instantiate_scev_binary): Likewise.
11194 (instantiate_scev_convert): Likewise.
11195 (instantiate_scev_not): Likewise.
11196 (instantiate_array_ref): Remove.
11197 (instantiate_scev_3): Likewise.
11198 (instantiate_scev_2): Likewise.
11199 (instantiate_scev_1): Likewise.
11200 (instantiate_scev_r): Do not blindly handle N-operand trees.
11201 Do not instantiate array-refs. Handle all constants and invariants.
11202 (instantiate_scev): Make instantiate_below parameter
11203 the edge into the region.
11204 (resolve_mixers): Use the loop preheader edge for the region
11205 parameter to instantiate_scev_r.
11206 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11208 2017-10-13 Richard Biener <rguenther@suse.de>
11210 PR tree-optimization/82525
11211 * graphite-isl-ast-to-gimple.c
11212 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11214 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11215 Fail code generation when we cannot represent the isl integer.
11216 (binary_op_to_tree): Elide modulo operations that are no-ops
11217 in the type we code generate. Remove now superfluous code
11220 2017-10-13 Richard Biener <rguenther@suse.de>
11222 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11223 (scop_detection::harmful_loop_in_region): Remove premature
11224 IV type restriction.
11225 (scop_detection::graphite_can_represent_scev): We can handle
11226 pointer IVs just fine.
11228 2017-10-13 Alan Modra <amodra@gmail.com>
11230 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11231 "Clobbers and Scratch Registers". Add paragraph on
11232 alternative to clobbers for scratch registers and OpenBLAS
11235 2017-10-13 Alan Modra <amodra@gmail.com>
11237 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11238 example of a memory input for a string of known length. Move
11239 commentary out of table. Add a number of new examples
11240 covering array memory inputs.
11242 2017-10-12 Martin Liska <mliska@suse.cz>
11244 PR tree-optimization/82493
11245 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11246 (test_range_functions): New function.
11247 (sbitmap_c_tests): Likewise.
11248 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11249 * selftest.h (sbitmap_c_tests): New function.
11251 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11253 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11255 * config/rs6000/amo.h: Fix spacing issue.
11257 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11260 * config/i386/i386.md (*ashl<mode>3_mask_1,
11261 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11262 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11265 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11267 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11268 (profile_probability): Set max_probability
11269 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11271 * profile-count.c (profile_probability::differs_from_p): Do not
11272 rely on max_probaiblity == 10000
11274 2017-10-12 Jeff Law <law@redhat.com>
11276 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11279 2017-10-12 Martin Sebor <msebor@redhat.com>
11283 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11284 (handle_alias_pairs): Call it.
11285 * common.opt (-Wattribute-alias): New option.
11286 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11287 * doc/invoke.texi (-Wattribute-alias): Document.
11289 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11292 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11294 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11296 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11297 (make_hard_regno_born, make_hard_regno_dead): Update
11298 bb_killed_pseudos and bb_gen_pseudos.
11300 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11302 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11304 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11306 * config/alpha/alpha.c (alpha_split_conditional_move):
11307 Use std::swap instead of manually swapping.
11308 (alpha_stdarg_optimize_hook): Ditto.
11309 (alpha_canonicalize_comparison): Ditto.
11311 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11313 * tree-loop-distribution.c (struct builtin_info): New struct.
11314 (struct partition): Refactor fields into struct builtin_info.
11315 (partition_free): Free struct builtin_info.
11316 (build_size_arg_loc, build_addr_arg_loc): Delete.
11317 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11318 information from struct builtin_info.
11319 (find_single_drs): New function refactored from classify_partition.
11320 Also moved builtin validity checks to this function.
11321 (compute_access_range, alloc_builtin): New functions.
11322 (classify_builtin_st, classify_builtin_ldst): New functions.
11323 (classify_partition): Refactor code into functions find_single_drs,
11324 classify_builtin_st and classify_builtin_ldst.
11325 (distribute_loop): Don't do runtime alias check when distributing
11327 (find_seed_stmts_for_distribution): New function.
11328 (pass_loop_distribution::execute): Refactor code finding seed
11329 stmts into above function. Support distribution for the innermost
11330 two-level loop nest. Adjust dump information.
11332 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11334 * tree-loop-distribution.c: Adjust the general comment.
11335 (NUM_PARTITION_THRESHOLD): New macro.
11336 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11337 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11338 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11339 in call to build_partition_graph.
11340 (finalize_partitions): New parameter. Make loop distribution more
11341 conservative by fusing more partitions.
11342 (distribute_loop): Don't do runtime alias check in case of loop nest
11344 (find_seed_stmts_for_distribution): New function.
11345 (prepare_perfect_loop_nest): New function.
11346 (pass_loop_distribution::execute): Refactor code finding seed stmts
11347 and loop nest into above functions. Support loop nest distribution.
11348 Adjust dump information accordingly.
11350 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11352 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11353 and set PTYPE_SEQUENTIAL for merged partition.
11355 2017-10-12 Richard Biener <rguenther@suse.de>
11357 PR tree-optimization/69728
11359 2017-09-19 Richard Biener <rguenther@suse.de>
11361 PR tree-optimization/69728
11362 * graphite-sese-to-poly.c (schedule_error): New global.
11363 (add_loop_schedule): Handle empty domain by failing the
11365 (build_original_schedule): Handle schedule_error.
11367 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11368 domain by returning an unchanged schedule.
11370 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11372 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11373 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11375 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11377 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11380 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11383 * expr.c (store_field): Don't optimize away bitsize == 0 store
11384 from CALL_EXPR with addressable return type.
11386 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11388 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11389 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11390 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11391 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11392 TARGET_ISEL instead of TARGET_ISEL<sel>.
11394 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11396 * config/rs6000/rs6000.c
11397 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11399 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11401 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11403 (reg_or_cint_operand): Fix comment.
11404 (reg_or_zero_operand): New predicate.
11405 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11406 * config/rs6000/rs6000.c (output_isel): Delete.
11407 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11408 instead of reg_or_cint_operand. Output instruction directly (not via
11410 (isel_unsigned_<mode>): Ditto.
11411 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11412 gpc_reg_operand. Add an instruction alternative for this. Output
11413 instruction directly.
11414 (*isel_reversed_unsigned_<mode>): Ditto.
11416 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11418 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11419 (TARGET_CANONICALIZE_COMPARISON): Define.
11421 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11424 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11425 Check whether the dest is REG before adding REG_EQUIV note.
11427 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11430 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11432 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11433 (make_hard_regno_born, make_hard_regno_dead): Update
11434 bb_killed_pseudos and bb_gen_pseudos.
11436 2017-10-11 Nathan Sidwell <nathan@acm.org>
11438 * incpath.h (enum incpath_kind): Name enum, prefix values.
11439 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11440 * incpath.c (heads, tails): Use INC_MAX.
11441 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11442 (merge_include_chains, split_quote_chain,
11443 register_include_chains): Update incpath_kind names.
11444 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11445 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11447 (add_framework_path, darwin_register_objc_includes): Likewise.
11448 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11450 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11452 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11453 Do not use float_operator operator predicate.
11454 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11455 * config/i386/predicates.md (float_operator): Remove predicate.
11457 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11459 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11460 (*jccxf_i387): Ditto.
11461 (*jcc<mode>_i387): Ditto.
11462 (*jccu<mode>_i387): Ditto.
11463 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11464 (*jcc_*_i387 splitters): Remove.
11465 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11466 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11467 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11470 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11472 * profile-count.h (slow_safe_scale_64bit): New function.
11473 (safe_scale_64bit): New inline.
11474 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11475 * profile-count.c: Include wide-int.h
11476 (slow_safe_scale_64bit): New.
11478 2017-10-11 Nathan Sidwell <nathan@acm.org>
11480 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11481 HAS_DECL_ASSEMBLER_NAME_P.
11482 * gimple-expr.c (gimple_decl_printable_name: Check
11483 HAS_DECL_ASSEMBLER_NAME_P too.
11484 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11485 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11486 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11487 * passes.c (rest_of_decl_compilation): Check
11488 HAS_DECL_ASSEMBLER_NAME_P too.
11489 * recog.c (verify_changes): Likewise.
11490 * tree-pretty-print.c (dump_decl_name): Likewise.
11491 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11493 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11494 (SET_DECL_ASSEMBLER_NAME): Use it.
11495 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11496 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11497 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11499 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11501 * config.gcc (i386, x86_64): Add extra objects.
11502 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11503 (ix86_min_insn_size): Declare.
11504 (ix86_issue_rate): Declare.
11505 (ix86_adjust_cost): Declare.
11506 (ia32_multipass_dfa_lookahead): Declare.
11507 (ix86_macro_fusion_p): Declare.
11508 (ix86_macro_fusion_pair_p): Declare.
11509 (ix86_bd_has_dispatch): Declare.
11510 (ix86_bd_do_dispatch): Declare.
11511 (ix86_core2i7_init_hooks): Declare.
11512 (ix86_atom_sched_reorder): Declare.
11513 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11514 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11515 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11516 (rip_relative_addr_p): Rename to ...
11517 (ix86_rip_relative_addr_p): ... this one; export.
11518 (memory_address_length): Update.
11519 (ix86_issue_rate): Move to x86-tune-sched.c.
11520 (ix86_flags_dependent): Move to x86-tune-sched.c.
11521 (ix86_agi_dependent): Move to x86-tune-sched.c.
11522 (exact_dependency_1): Move to x86-tune-sched.c.
11523 (exact_store_load_dependency): Move to x86-tune-sched.c.
11524 (ix86_adjust_cost): Move to x86-tune-sched.c.
11525 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11526 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11527 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11528 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11529 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11530 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11531 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11532 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11533 (min_insn_size): Rename to ...
11534 (ix86_min_insn_size): ... this one; export.
11535 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11536 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11537 (core2i7_first_cycle_multipass_backtrack): Move to
11538 x86-tune-sched-core.c.
11539 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11540 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11541 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11542 (ix86_avoid_jump_mispredicts): Update.
11543 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11544 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11545 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11546 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11547 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11548 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11549 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11550 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11551 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11552 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11553 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11554 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11555 (BIG): Move to ix86-tune-sched-bd.c.
11556 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11557 (enum insn_path): Move to ix86-tune-sched-bd.c.
11558 (get_mem_group): Move to ix86-tune-sched-bd.c.
11559 (is_cmp): Move to ix86-tune-sched-bd.c.
11560 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11561 (is_branch): Move to ix86-tune-sched-bd.c.
11562 (is_prefetch): Move to ix86-tune-sched-bd.c.
11563 (init_window): Move to ix86-tune-sched-bd.c.
11564 (allocate_window): Move to ix86-tune-sched-bd.c.
11565 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11566 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11567 (process_end_window): Move to ix86-tune-sched-bd.c.
11568 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11569 (find_constant): Move to ix86-tune-sched-bd.c.
11570 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11571 (has_immediate): Move to ix86-tune-sched-bd.c.
11572 (get_insn_path): Move to ix86-tune-sched-bd.c.
11573 (get_insn_group): Move to ix86-tune-sched-bd.c.
11574 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11575 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11576 (add_insn_window): Move to ix86-tune-sched-bd.c.
11577 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11578 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11579 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11580 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11581 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11582 (do_dispatch): Move to ix86-tune-sched-bd.c.
11583 (has_dispatch): Move to ix86-tune-sched-bd.c.
11584 * config/i386/t-i386: Add new object files.
11585 * config/i386/x86-tune-costs.h: New file.
11586 * config/i386/x86-tune-sched-atom.c: New file.
11587 * config/i386/x86-tune-sched-bd.c: New file.
11588 * config/i386/x86-tune-sched-core.c: New file.
11589 * config/i386/x86-tune-sched.c: New file.
11591 2017-10-11 Liu Hao <lh_mouse@126.com>
11593 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11594 the generic version below instead.
11595 (should_colorize): Recognize Windows consoles as terminals
11597 * pretty-print.c [__MINGW32__] (write_all): New function.
11598 [__MINGW32__] (find_esc_head): Likewise.
11599 [__MINGW32__] (find_esc_terminator): Likewise.
11600 [__MINGW32__] (eat_esc_sequence): Likewise.
11601 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11603 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11606 2017-10-11 Richard Biener <rguenther@suse.de>
11608 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11609 Properly call analyze_scalar_evolution with the loop of the stmt.
11611 2017-10-11 Richard Biener <rguenther@suse.de>
11613 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11614 * tree-core.h (tree_base): Add chrec_var union member.
11615 * tree.h (CHREC_VAR): Remove.
11616 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11617 * tree-chrec.h (build_polynomial_chrec): Adjust.
11618 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11619 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11621 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11623 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11624 * match.pd: ... here.
11625 ((T) X == (T) Y): Relax condition.
11627 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11629 PR tree-optimization/82472
11630 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11632 (break_alias_scc_partitions): Update postorder number.
11634 2017-10-11 Martin Liska <mliska@suse.cz>
11637 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11639 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11641 2017-10-11 Martin Liska <mliska@suse.cz>
11643 * print-rtl.c (print_insn): Move declaration of idbuf
11644 to same scope as name.
11646 2017-10-11 Martin Liska <mliska@suse.cz>
11651 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11652 volatile arguments.
11654 2017-10-11 Martin Liska <mliska@suse.cz>
11657 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11658 volatile arguments.
11660 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11662 * config.gcc (default_gnu_indirect_function): Default to yes for
11663 arm*-*-linux* with glibc.
11665 2017-10-11 Richard Biener <rguenther@suse.de>
11667 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11668 default-defs and types we do not want to analyze.
11669 (interpret_loop_phi): Replace unreachable code with an assert.
11670 (compute_scalar_evolution_in_loop): Remove and inline ...
11671 (analyze_scalar_evolution_1): ... here, replacing condition with
11672 what makes the intent clearer. Remove handling of cases
11673 get_scalar_evolution now handles.
11675 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11677 PR rtl-optimization/81434
11678 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11679 comment for main loop. In sched_group_found if, also add checks for
11680 pass and min_cost_group.
11682 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11684 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11685 (rs6000_insn_cost): New function.
11686 * config/rs6000/rs6000.md (cost): New attribute.
11688 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11689 H.J. Lu <hongjiu.lu@intel.com>
11693 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11694 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11695 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11696 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11697 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11698 OPTION_MASK_ISA_MMX and some other bit set require both
11699 mmx and the ISAs without the mmx bit.
11700 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11701 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11702 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11703 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11704 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11705 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11706 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11707 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11708 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11709 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11710 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11711 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11712 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11713 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11714 Add OPTION_MASK_ISA_MMX.
11716 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11718 * config.gcc (armv7*-*-freebsd*): New target.
11719 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11721 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11723 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11724 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11727 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11729 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11730 a template parameter.
11731 (WIDE_INT_REF_FOR): Update accordingly.
11732 * tree.h (wi::int_traits <const_tree>): Delete.
11733 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11734 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11735 (wi::tree_to_wide_ref): New typedef.
11736 (wi::to_wide): New function.
11737 * calls.c (get_size_range): Use wi::to_wide when operating on
11738 trees as wide_ints.
11739 * cgraph.c (cgraph_node::create_thunk): Likewise.
11740 * config/i386/i386.c (ix86_data_alignment): Likewise.
11741 (ix86_local_alignment): Likewise.
11742 * dbxout.c (stabstr_O): Likewise.
11743 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11744 * expr.c (const_vector_from_tree): Likewise.
11745 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11746 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11747 (fold_negate_expr_1, int_const_binop_1, const_binop)
11748 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11749 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11750 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11751 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11752 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11753 (fold_not_const, round_up_loc): Likewise.
11754 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11755 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11756 (alloca_call_type): Likewise.
11757 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11758 * godump.c (go_output_typedef): Likewise.
11759 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11760 * internal-fn.c (get_min_precision): Likewise.
11761 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11762 * ipa-polymorphic-call.c
11763 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11764 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11765 (ipa_modify_call_arguments): Likewise.
11766 * match.pd: Likewise.
11767 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11768 * print-tree.c (print_node_brief, print_node): Likewise.
11769 * stmt.c (expand_case): Likewise.
11770 * stor-layout.c (layout_type): Likewise.
11771 * tree-affine.c (tree_to_aff_combination): Likewise.
11772 * tree-cfg.c (group_case_labels_stmt): Likewise.
11773 * tree-data-ref.c (dr_analyze_indices): Likewise.
11774 (prune_runtime_alias_test_list): Likewise.
11775 * tree-dump.c (dequeue_and_dump): Likewise.
11776 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11777 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11778 * tree-pretty-print.c (dump_generic_node): Likewise.
11779 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11780 (simple_iv_with_niters): Likewise.
11781 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11782 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11783 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11784 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11785 (refine_value_range_using_guard, number_of_iterations_ne_max)
11786 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11787 (get_cst_init_from_scev, record_nonwrapping_iv)
11788 (scev_var_range_cant_overflow): Likewise.
11789 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11790 * tree-ssa-pre.c (compute_avail): Likewise.
11791 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11792 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11793 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11794 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11795 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11796 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11797 (dump_case_nodes, try_switch_expansion): Likewise.
11798 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11799 (vect_do_peeling): Likewise.
11800 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11801 * tree-vect-stmts.c (vectorizable_load): Likewise.
11802 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11803 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11804 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11805 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11806 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11807 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11808 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11809 (evrp_dom_walker::before_dom_children): Likewise.
11810 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11811 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11812 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11813 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11814 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11815 (get_range_pos_neg): Likewise.
11816 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11817 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11818 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11819 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11820 * config/avr/avr.c (avr_fold_builtin): Likewise.
11821 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11822 * config/msp430/msp430.c (msp430_attr): Likewise.
11823 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11824 * config/powerpcspe/powerpcspe-c.c
11825 (altivec_resolve_overloaded_builtin): Likewise.
11826 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11827 (rs6000_expand_ternop_builtin): Likewise.
11828 * config/rs6000/rs6000-c.c
11829 (altivec_resolve_overloaded_builtin): Likewise.
11830 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11831 (rs6000_expand_ternop_builtin): Likewise.
11832 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11834 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11836 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11837 when copying loop nest with only one inner loop.
11839 2017-10-10 Richard Biener <rguenther@suse.de>
11841 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11842 blocks if SCEV is active.
11843 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11845 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11846 (scev_initialize): Assert we are not yet initialized.
11848 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11850 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11851 inner loop's exit stmt by making it always exit the loop, otherwise
11852 we would generate an infinite empty loop.
11854 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11856 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11857 renaming variables in new preheader if it's deleted.
11859 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11861 * tree-loop-distribution.c (struct partition): Remove unused field
11862 loops of the structure.
11863 (partition_alloc, partition_free): Ditto.
11864 (build_rdg_partition_for_vertex): Ditto.
11866 2017-10-09 Jeff Law <law@redhat.com>
11868 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11869 return type to match prototype and documentation.
11871 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11873 * config/rs6000/rs6000.c (processor_costs): Move to ...
11874 * config/rs6000/rs6000.h: ... here.
11875 (rs6000_cost): Declare.
11877 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11879 * except.c (setjmp_fn): New global variable.
11880 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11881 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11882 if DONT_USE_BUILTIN_SETJMP is defined.
11884 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11886 * target.def (insn_cost): New hook.
11887 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11888 * doc/tm.texi: Regenerate.
11889 * rtlanal.c (insn_cost): Use the new hook.
11891 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11893 * combine.c (combine_validate_cost): Compute the new insn_cost,
11894 not just pattern_cost.
11895 (try_combine): Adjust comment.
11897 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11899 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11901 * combine.c (uid_insn_cost): Adjust comment.
11902 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11904 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11905 * dse.c (find_shift_sequence): Ditto.
11906 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11907 (bb_valid_for_noce_process_p): Use pattern_cost.
11908 * rtl.h (insn_rtx_cost): Delete.
11909 (pattern_cost): New prototype.
11910 (insn_cost): New prototype.
11911 * rtlanal.c (insn_rtx_cost): Rename to...
11912 (pattern_cost): ... this.
11915 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11917 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11918 (*jcc<mode>_0_r_i387): Ditto.
11919 (*jccxf_r_i387): Ditto.
11920 (*jcc<mode>_r_i387): Ditto.
11921 (*jccu<mode>_r_i387): Ditto.
11922 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11923 (*jcc): Rename from *jcc_1.
11925 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11927 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11928 deferred rescans after the lvx/stvx recombination pre-pass.
11930 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11932 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11933 memory operation instruction support.
11934 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11935 (rs6000-ibm-aix[789]*): Likewise.
11936 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11937 Document new functions.
11939 2017-10-09 Richard Biener <rguenther@suse.de>
11941 PR tree-optimization/82397
11942 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11943 equality only for semantically equal trees.
11945 2017-10-09 Richard Biener <rguenther@suse.de>
11947 PR tree-optimization/82449
11948 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11949 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11950 allow constant addresses.
11951 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11954 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11956 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11959 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11962 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11965 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11968 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11970 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11973 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11974 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11976 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11978 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11979 (WI_BINARY_PREDICATE_RESULT): Likewise.
11980 (wi::binary_traits::operator_result): New type.
11981 (wi::binary_traits::predicate_result): Likewise.
11982 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11983 (generic_wide_int::operator==, generic_wide_int::operator!=)
11984 (generic_wide_int::operator&, generic_wide_int::and_not)
11985 (generic_wide_int::operator|, generic_wide_int::or_not)
11986 (generic_wide_int::operator^, generic_wide_int::operator+
11987 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11989 (operator~, unary operator-, operator==, operator!=, operator&)
11990 (operator|, operator^, operator+, binary operator-, operator*): New
11992 * expr.c (get_inner_reference): Use wi::bit_and_not.
11993 * fold-const.c (fold_binary_loc): Likewise.
11994 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11995 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11996 (bit_value_binop): Likewise.
11997 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11998 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11999 (extract_range_from_binary_expr_1): Likewise.
12000 (masked_increment): Likewise.
12001 (simplify_bit_ops_using_ranges): Likewise.
12003 2017-10-09 Martin Jambor <mjambor@suse.cz>
12006 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12007 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12008 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12010 (hsa_fixup_mov_insn_type): New function.
12011 (hsa_op_with_type::get_in_type): Use it.
12012 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12013 immediates in an assert.
12014 (hsa_op_with_type::extend_int_to_32bit): New method.
12015 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12016 types. Convert to dest type if necessary.
12017 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12018 (reg_for_gimple_ssa): Pass false as min32int to
12019 hsa_type_for_scalar_tree_type.
12020 (gen_hsa_addr): Fixup type when creating addresable temporary.
12021 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12022 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12023 necessary. Call hsa_fixup_mov_insn_type.
12024 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12025 extend operands and convert to dest type if necessary.
12026 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12027 to dest type if necessary.
12028 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12029 if conversion nt necessary and size matches.
12030 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12031 to dest type if necessary.
12032 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12033 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12035 (gen_hsa_clrsb): Likewise.
12036 (gen_hsa_ffs): Likewise.
12037 (gen_hsa_divmod): Extend operands and convert to dest type if
12039 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12041 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12043 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12044 Remove empty default arguments. Use a brace block as output
12046 (conditional return): Ditto.
12048 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12049 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12051 (group_ending_nop): Ditto.
12052 (doloop_end): Ditto.
12053 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12054 (splitters for those): Ditto.
12056 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12058 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12059 a conditional jump (and the compare for it) so that pc_rtx is the
12061 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12062 for the deleted and renamed ctr<mode>_internal[234] patterns.
12063 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12064 Delete second conditional return pattern.
12065 (ctr<mode>_internal2): Delete this second bdnz pattern.
12066 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12067 (ctr<mode>_internal4): Delete this second bdz pattern.
12069 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12071 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12072 (always_initialized_rtx_for_ssa_name_p): New predicate.
12073 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12074 (finish_out_of_ssa): Free new field of SA.
12075 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12076 * tree-ssa-coalesce.c: Include tree-ssa.h.
12077 (get_parm_default_def_partitions): Remove extern keyword.
12078 (get_undefined_value_partitions): New function.
12079 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12080 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12081 uninitialized bits.
12082 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12084 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12086 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12088 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12090 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12091 for targets that preffer 128bit.
12093 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12095 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12097 2017-10-08 Olivier Hainque <hainque@adacore.com>
12099 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12100 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12102 (thumb_set_return_address): Likewise.
12104 2017-10-08 Olivier Hainque <hainque@adacore.com>
12106 * common/config/arm/arm-common.c (arm_except_unwind_info):
12107 Handle DWARF2_UNWIND_INFO.
12109 2017-10-07 Michael Collison <michael.collison@arm.com>
12111 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12114 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12116 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12117 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12118 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12119 defined, force the creation of a new block for a dispatch label.
12121 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12123 * invoke.texi (Wsuggest-attribute=cold): Document.
12124 * common.opt (Wsuggest-attribute=cold): New
12125 * ipa-pure-const.c (warn_function_cold): New function.
12126 * predict.c (compute_function_frequency): Use it.
12127 * predict.h (warn_function_cold): Declare.
12129 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12131 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12134 2017-10-06 Martin Liska <mliska@suse.cz>
12136 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12137 keyword for member functions.
12138 (struct sanopt_tree_couple): New struct.
12139 (struct sanopt_tree_couple_hash): New function.
12140 (struct sanopt_ctx): Add new hash_map.
12141 (has_dominating_ubsan_ptr_check): New function.
12142 (record_ubsan_ptr_check_stmt): Likewise.
12143 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12144 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12145 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12147 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12150 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12151 aarch64_simd_valid_immediate on CONST_VECTORs.
12152 (aarch64_reg_or_bic_imm): Likewise.
12154 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12156 PR rtl-optimization/82396
12157 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12159 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12161 * graphite-dependences.c (scop_get_reads): Move code to...
12162 (scop_get_must_writes): Move code to...
12163 (scop_get_may_writes): Move code to...
12164 (scop_get_reads_and_writes): ... here.
12165 (scop_get_dependences): Call scop_get_reads_and_writes.
12167 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12169 PR tree-optimization/82434
12170 * fold-const.h (can_native_encode_type_p,
12171 can_native_encode_string_p): Remove.
12172 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12173 don't encode anything, just return what would be otherwise returned.
12174 (native_encode_fixed, native_encode_complex, native_encode_vector):
12176 (native_encode_string): Likewise. Inline by hand
12177 can_native_encode_string_p.
12178 (can_native_encode_type_p): Remove.
12179 (can_native_encode_string_p): Remove.
12180 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12181 STRING_CSTs using can_native_encode_string_p, test all
12182 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12183 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12184 argument from native_encode_expr.
12185 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12186 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12189 2017-10-06 Richard Biener <rguenther@suse.de>
12191 PR tree-optimization/82397
12192 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12193 operand_equal_p but rely on data_ref_compare_tree for detecting
12195 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12196 to match up with dr_group_sort_cmp.
12198 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12201 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12203 * config/s390/s390-builtin-types.def: Regenerate.
12205 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12208 * config/s390/s390-builtin-types.def: Regenerate.
12209 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12210 Change flag from B_VXE to B_VX.
12211 (s390_vec_min_dbl): Remove B_VXE flag.
12213 2017-10-06 Richard Biener <rguenther@suse.de>
12215 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12216 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12217 translate_isl_ast_to_gimple::is_valid_rename,
12218 translate_isl_ast_to_gimple::get_rename,
12219 translate_isl_ast_to_gimple::get_def_bb_for_const,
12220 translate_isl_ast_to_gimple::get_new_name,
12221 translate_isl_ast_to_gimple::collect_all_ssa_names,
12222 translate_isl_ast_to_gimple::copy_loop_phi_args,
12223 translate_isl_ast_to_gimple::collect_all_ssa_names,
12224 translate_isl_ast_to_gimple::copy_loop_phi_args,
12225 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12226 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12227 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12228 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12229 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12230 translate_isl_ast_to_gimple::copy_cond_phi_args,
12231 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12232 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12233 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12234 translate_isl_ast_to_gimple::rename_uses,
12235 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12236 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12237 (set_rename_for_each_def): Likewise.
12238 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12239 here. Handle rewriting SCEV analyzable uses here.
12240 (copy_bb_and_scalar_dependences): Generate code for PHI
12242 (graphite_regenerate_ast_isl): Adjust.
12243 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12244 (add_write, add_read): New functions.
12245 (build_cross_bb_scalars_def): Use it and simplify.
12246 (build_cross_bb_scalars_use): Likewise.
12247 (graphite_find_cross_bb_scalar_vars): Inline into...
12248 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12249 simulating out-of-SSA. Compute liveout and add dependencies.
12250 (build_scops): Force an empty entry block.
12251 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12253 (sese_build_liveouts): Declare.
12254 (sese_trivially_empty_bb_p): Likewise.
12255 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12256 compute liveout and debug_liveout.
12257 (sese_bad_liveouts_use): Remove.
12258 (sese_reset_debug_liveouts_bb): Likewise.
12259 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12260 (sese_build_liveouts): Build liveout and debug_liveout and store
12262 (new_sese_info): Adjust.
12263 (free_sese_info): Likewise.
12264 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12265 do not build liveout here.
12266 (move_sese_in_condition): Adjust region entry.
12267 (scev_analyzable_p): Match up with chrec_apply requirements.
12268 (sese_trivially_empty_bb_p): New.
12269 * tree-into-ssa.c (get_reaching_def): Properly support generating
12270 default-defs for incremental rewrite of anonymous names.
12272 2017-10-06 Richard Biener <rguenther@suse.de>
12274 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12275 precision do not perform modulo reduction.
12277 2017-10-06 Richard Biener <rguenther@suse.de>
12279 PR tree-optimization/82436
12280 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12281 conservatively choose the vectorization factor when checking
12282 whether we can perform the required load permutation.
12283 (vect_transform_slp_perm_load): Assert when we may not fail.
12285 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12287 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12288 message for incompatible -msdata=* and -mcall-* options.
12290 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12292 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12293 rate for post-reload scheduling.
12295 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12297 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12299 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12301 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12302 to improve monte carlo in scimark.
12304 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12306 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12307 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12308 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12309 pentium4_cost, nocona_cost): Set reassociation width to 1.
12310 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12311 width to 2 for fp operations and 1 otherwise.
12312 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12314 (atom_cost): Set reassociation width to 2.
12315 (slm_cost, generic_cost): Set fp reassociation width
12316 to 2 and 1 otherwise.
12317 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12318 (core_cost): Set fp reassociation width to 4 and vector to 2.
12319 (ix86_reassociation_width): Rewrite using cost table; special case
12320 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12321 and TARGET_AVX128_OPTIMAL.
12322 * config/i386/i386.h (processor_costs): Add
12323 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12324 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12325 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12326 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12327 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12328 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12330 2017-10-05 Nathan Sidwell <nathan@acm.org>
12332 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12334 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12336 * config/arm/arm.c (arm_test_fpu_data): New.
12337 (arm_run_selftests): Call arm_test_fpu_data.
12339 2017-10-04 Nathan Sidwell <nathan@acm.org>
12341 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12343 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12345 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12347 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12348 check type for aarch64_simd_valid_immediate.
12349 (aarch64_output_simd_mov_immediate): Update prototype.
12350 (aarch64_simd_valid_immediate): Update prototype.
12351 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12352 support for ORR-immediate.
12353 (and<mode>3): modified pattern to add support for BIC-immediate.
12354 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12355 now checks for valid immediate for BIC and ORR based on new enum
12357 (aarch64_output_simd_mov_immediate): Function now used to output
12358 BIC/ORR imm as well based on new enum argument.
12359 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12361 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12362 (aarch64_reg_or_bic_imm): Likewise.
12364 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12366 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12367 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12369 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12373 PR rtl-optimization/82396
12374 * haifa-sched.c (autopref_multipass_init): Simplify
12376 (autopref_rank_data): Simplify sort order.
12377 * sched-int.h (autopref_multipass_data_): Remove
12378 multi_mem_insn_p, min_offset and max_offset.
12380 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12382 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12384 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12386 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12387 vect_doubleint_cvt.
12389 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12391 * doc/sourcebuild.texi: Document vect_long_mult.
12393 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12395 PR tree-optimization/82413
12396 * fold-const.c (build_range_check): Use widest_int when comparing
12397 the maximum ETYPE value with HIGH.
12399 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12401 PR rtl-optimization/82396
12402 * haifa-sched.c (autopref_multipass_init): Simplify
12404 (autopref_rank_data): Simplify sort order.
12405 * sched-int.h (autopref_multipass_data_): Remove
12406 multi_mem_insn_p, min_offset and max_offset.
12408 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12410 PR tree-optimization/82381
12411 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12412 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12415 PR tree-optimization/82374
12416 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12417 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12418 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12419 current_function_decl to the new decl.
12421 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12423 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12424 helper macro for IEEE float128 hardware built-in functions.
12425 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12427 (TRUNCF128_ODD): Likewise.
12428 (ADDF128_ODD): Likewise.
12429 (SUBF128_ODD): Likewise.
12430 (MULF128_ODD): Likewise.
12431 (DIVF128_ODD): Likewise.
12432 (FMAF128_ODD): Likewise.
12433 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12434 UNSPEC_TRUNC_ROUND_TO_ODD.
12435 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12436 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12437 floating point round to odd instructions.
12438 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12439 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12440 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12441 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12442 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12443 (trunc<mode>sf2_hw): Change the truncate with round to odd
12444 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12445 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12446 to odd hardware instructions.
12447 (sub<mode>3_odd): Likewise.
12448 (mul<mode>3_odd): Likewise.
12449 (div<mode>3_odd): Likewise.
12450 (sqrt<mode>2_odd): Likewise.
12451 (fma<mode>4_odd): Likewise.
12452 (fms<mode>4_odd): Likewise.
12453 (nfma<mode>4_odd): Likewise.
12454 (nfms<mode>4_odd): Likewise.
12455 (trunc<mode>df2_odd): Change the truncate with round to odd
12456 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12458 * doc/extend.texi (PowerPC built-in functions): Update documentation
12459 for existing IEEE float128-bit built-in functions. Add built-in
12460 functions that generate the IEEE 128-bit floating point round to
12463 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12465 PR rtl-optimization/77729
12466 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12467 to (X&(C1&~C2))|C2 transformations.
12469 2017-10-03 Martin Jambor <mjambor@suse.cz>
12471 PR tree-optimization/82363
12472 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12473 mismatch, mark lacc written regardless of racc.
12475 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12477 PR tree-optimization/82381
12478 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12479 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12480 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12481 fallthrough into reassoc_stmt_dominates_stmt_p.
12484 * combine.c (combine_instructions): Don't combine in unreachable
12487 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12490 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12491 function to not use the have_cpu variable. Do not set cpu_index,
12492 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12493 or the default cpu.
12494 (rs6000_valid_attribute_p): Remove duplicate initializations of
12495 old_optimize and func_optimize.
12496 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12497 (rs6000_activate_target_options): Make global.
12498 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12501 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12503 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12504 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12505 Set *poffset to 0 and *psize and *pmax_size to -1 if
12506 *poffset + *psize overflows in HOST_WIDE_INT.
12508 PR tree-optimization/82387
12509 PR tree-optimization/82388
12510 PR tree-optimization/82389
12511 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12512 instead of live_bytes non-NULL.
12514 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12517 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12520 2017-10-02 Richard Biener <rguenther@suse.de>
12522 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12523 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12524 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12526 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12528 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12529 requested precision matches the type's.
12530 * calls.c (alloc_max_size): Calculate the new candidate size as
12531 a widest_int and use wi::to_widest when comparing it with the
12532 current candidate size.
12533 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12534 zero rather than integer_zero_node.
12535 * match.pd: Check for a no-op conversion before using wi::add
12536 rather than after. Use tree_to_uhwi when summing small shift
12537 counts into an unsigned int.
12539 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12540 Alan Hayward <alan.hayward@arm.com>
12541 David Sherwood <david.sherwood@arm.com>
12544 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12545 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12546 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12547 POINTER_AND_FP_REGS.
12549 2017-10-02 Richard Biener <rguenther@suse.de>
12551 PR tree-optimization/82355
12552 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12553 a mapping for the enclosing loop but avoid generating one for
12554 the loop tree root.
12555 (copy_bb_and_scalar_dependences): Remove premature codegen
12556 error on PHIs in blocks duplicated into multiple places.
12557 * graphite-scop-detection.c
12558 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12559 in the region use it as loop and nest to analyze the DR in.
12560 (try_generate_gimple_bb): Likewise.
12561 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12562 (add_loop_constraints): For blocks in a loop not in the region
12563 create a dimension with a single iteration.
12564 * sese.h (gbb_loop_at_index): Remove assert.
12566 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12568 * omp-expand.c (adjust_context_scope): New function.
12569 (expand_parallel_call): Call adjust_context_scope.
12571 2017-10-01 Jeff Law <law@redhat.com>
12573 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12574 dom_opt_dom_walker class with direct access to private members.
12575 Add comments. Call test_for_singularity.
12576 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12577 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12578 m_dummy_cond anymore.
12579 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12581 (pass_dominator:execute): Build the dummy_cond here and pass it
12582 to the dom_opt_dom_walker ctor.
12583 (test_for_singularity): New function.
12585 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12586 Maya Rashish <coypu@sdf.org>
12588 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12589 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12590 (alpha*-*-netbsd*) Use nbsd_tm_file.
12591 (arm*-*-netbsdelf*) Likewise.
12592 (i[34567]86-*-netbsdelf*) Likewise.
12593 (x86_64-*-netbsd*) Likewise.
12594 (mips*-*-netbsd*) Likewise.
12595 (powerpc-*-netbsd*) Likewise.
12596 (sh*-*-netbsd*) Likewise.
12597 (sparc-*-netbsdelf*) Likewise.
12598 (sparc64-*-netbsd*) Likewise.
12599 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12601 (vax-*-netbsdelf*) Likewise.
12602 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12603 (UINT_FAST8_TYPE) Likewise.
12604 (INT_FAST16_TYPE) Check CHAR_FAST16.
12605 (UINT_FAST16_TYPE) Likewise.
12607 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12610 * config/i386/i386.md
12611 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12612 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12613 *divmodsi4_zext_2): New define_insn_and_split.
12614 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12615 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12616 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12617 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12618 New define_insn_and_split.
12619 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12620 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12621 operands[1] having DImode when mode is SImode.
12623 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12624 always SImode for DIV and MOD in REG_EQUAL notes.
12626 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12628 PR middle-end/82319
12629 * match.pd: Fix handling of NaNs in pattern.
12631 2017-09-29 Jeff Law <law@redhat.com>
12633 * sbitmap.c (bitmap_bit_in_range_p): New function.
12634 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12635 * tree-ssa-dse.c (live_bytes_read): New function.
12636 (dse_classify_store): Ignore reads of dead bytes.
12638 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12639 typos and whitespace errors.
12640 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12641 * config/s390/s390.c (s390_emit_prologue): Likewise.
12643 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12646 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12647 with a symbol for LRA.
12649 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12651 PR rtl-optimization/82338
12652 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12654 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12656 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12657 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12658 (qsort_chk): Declare.
12659 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12660 (qsort_chk): New function.
12662 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12664 PR tree-optimization/82337
12665 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12666 phi definition if the PHI result appears in an abnormal PHI.
12667 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12668 basis appears in an abnormal PHI.
12670 2017-09-29 Richard Biener <rguenther@suse.de>
12672 * graphite-isl-ast-to-gimple.c
12673 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12674 (binary_op_to_tree): Use it.
12675 (get_rename_from_scev): Likewise.
12676 (copy_loop_phi_nodes): Likewise.
12677 (copy_bb_and_scalar_dependences): Likewise.
12678 (translate_pending_phi_nodes): Likewise.
12680 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12683 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12684 for movabsq $(i32 << shift), r64.
12686 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12688 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12689 index when encoding %esp as %rsp to avoid 0x67 prefix.
12691 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12693 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12694 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12696 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12698 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12699 Extensions with more than 16 double VFP registers.
12700 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12701 to_clear_mask and all code related to it. Replace the remaining
12702 entry by a sbitmap and adapt code accordingly.
12704 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12706 * brig-builtins.def: Change pure attributes to const.
12708 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12710 * config.gcc (default_gnu_indirect_function): Default to yes for
12711 sparc*-*-linux* with glibc.
12713 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12715 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12716 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12717 when creating .init_array and .fini_array sections with priority
12720 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12723 * config/aarch64/aarch64.c
12724 (aarch64_builtin_support_vector_misalignment): Always return false
12725 when misalignment is unknown.
12727 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12729 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12730 this function to return false if the definition used by the swap
12731 instruction is artificial, or if the memory address from which the
12732 constant value is loaded is not represented by a base address held
12733 in a register or if the base address register is a frame or stack
12734 pointer. Additionally, return false if the base address of the
12735 loaded constant is a SYMBOL_REF but is not considered to be a
12737 (replace_swapped_load_constant): New function.
12738 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12739 loaded constant vector with a load of a swapped constant vector.
12741 2017-09-27 Carl Love <cel@us.ibm.com>
12743 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12744 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12745 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12748 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12750 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12751 first, always call autopref_rank_data otherwise.
12753 2017-09-27 Richard Biener <rguenther@suse.de>
12755 * graphite-scop-detection.c (find_scop_parameters): Move
12756 loop bound handling ...
12757 (gather_bbs::before_dom_children): ... here, avoiding the need
12758 to build scop_info->loop_nest.
12759 (record_loop_in_sese): Remove.
12760 * sese.h (sese_info_t::loop_nest): Remove.
12761 * sese.c (new_sese_info): Do not allocate loop_nest.
12762 (free_sese_info): Do not free loop_nest.
12764 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12767 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12768 lhs from calls if the lhs has addressable type.
12770 2017-09-27 Richard Biener <rguenther@suse.de>
12772 * graphite.h (scop::max_alias_set): New member.
12773 * graphite-scop-detection.c: Remove references to non-existing
12774 --param in comments.
12775 (build_alias_sets): Record the maximum alias set used for drs.
12776 (build_scops): Support zero as unlimited for
12777 --param graphite-max-arrays-per-scop.
12778 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12779 and inline into ...
12780 (build_poly_sr_1): ... here. Compute alias set based on the
12781 maximum alias set used for drs rather than
12782 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12784 2017-09-27 Richard Biener <rguenther@suse.de>
12786 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12787 --param loop-block-tile-size=0 to disable tiling.
12789 2017-09-27 Richard Biener <rguenther@suse.de>
12791 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12792 (graphite-max-nb-scop-params): Document special value zero.
12793 * domwalk.h (dom_walker::STOP): New symbolical constant.
12794 (dom_walker::dom_walker): Add optional parameter for bb to
12796 (dom_walker::~dom_walker): Declare.
12797 (dom_walker::before_dom_children): Document STOP return value.
12798 (dom_walker::m_user_bb_to_rpo): New member.
12799 (dom_walker::m_bb_to_rpo): Likewise.
12800 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12801 mapping here if not provided by the user.
12802 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12803 provided by the user.
12804 (dom_walker::STOP): Define.
12805 (dom_walker::walk): Do not compute bb to RPO mapping here.
12806 Support STOP return value from before_dom_children to stop
12808 * graphite-optimize-isl.c (optimize_isl): If the schedule
12809 is the same still generate code if -fgraphite-identity
12810 or -floop-parallelize-all are given.
12811 * graphite-scop-detection.c: Include cfganal.h.
12812 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12814 (gather_bbs::before_dom_children): Return STOP for BBs
12816 (build_scops): Compute bb to RPO mapping and pass it to
12817 the domwalk. Treat --param graphite-max-nb-scop-params=0
12818 as not limiting the number of params.
12819 * graphite.c (graphite_initialize): Remove limit on the
12820 number of basic-blocks in a function.
12821 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12822 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12823 default value of 10.
12825 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12827 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12828 Adjust code to eliminate needing to do the shift right 32-bits
12829 operation after XSCVDPSPN.
12831 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12833 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12834 ((X / Y) != 0 -> X >= Y): Likewise.
12836 2017-09-26 Carl Love <cel@us.ibm.com>
12838 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12839 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12840 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12841 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12842 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12843 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12844 definitions and overloading.
12845 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12846 statement for P9V_BUILTIN_XST_LEN_R.
12847 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12848 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12849 define_expand and define_insn for the instructions and builtins.
12850 * doc/extend.texi: Update the built-in documentation file for the new
12851 built-in functions.
12852 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12853 define_insn for the instructions
12855 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12858 * gcc/config/netbsd-protos.h: New file.
12859 * gcc/config/netbsd.c: New file.
12860 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12861 * gcc/config/t-netbsd: New file.
12862 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12863 (tmake_file) Add t-netbsd.
12864 (extra_objs) Add netbsd.o.
12866 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12870 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12872 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12874 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12875 sign extension from a vector register to a GPR by doing a 32-bit
12876 direct move and then an EXTSW.
12877 (extendsi<mode>2 splitter): Likewise.
12878 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12879 right or vector extract after doing XSCVDPSPN. Use
12880 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12882 (movdi_from_sf_zero_ext): Likewise.
12883 (reload_gpr_from_vsxsf): Likewise.
12884 (p8_mfvsrd_4_disf): Delete, no longer used.
12885 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12886 then needing to move the SFmode to a GPR to use the XSCVDPSP
12887 instruction instead of FRSP and XSCVDPSPN.
12888 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12889 it is adjacent to the other XSCVSPDP insns.
12890 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12891 SFmode to be in traditional Altivec registers.
12892 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12893 (vsx_xscvspdpn): Likewise.
12894 (vsx_xscvspdpn_scalar): Likewise.
12896 2017-09-26 Martin Jambor <mjambor@suse.cz>
12898 * tree-sra.c (compare_access_positions): Put integral types first,
12899 stabilize sorting of integral types, remove conditions putting
12900 non-full-precision integers last.
12901 (sort_and_splice_var_accesses): Disable scalarization if a
12902 non-integert would be represented by a non-full-precision integer.
12904 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12906 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12907 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12908 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12909 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12910 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12911 conditionals inside the function instead of around it. Call
12912 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12913 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12915 2017-09-26 Richard Biener <rguenther@suse.de>
12917 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12919 (scop_detection::build_scop_breadth): ... this. Removed.
12920 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12921 (scop_detection::harmful_stmt_in_bb): Likewise.
12922 (scop_detection::graphite_can_represent_stmt): Likewise.
12923 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12924 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12925 (scop_detection::can_represent_loop_1): ... this. Removed.
12926 (scop_detection::harmful_loop_in_region): Simplify after inlining
12927 the above and remove more quadraticness.
12928 (build_scops): Adjust.
12929 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12932 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12935 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12936 REGNO (base) == SP_REG if base is a REG.
12938 PR middle-end/35691
12939 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12940 if it is different SSA_NAME.
12941 (optimize_range_tests_cmp_bitwise): New function.
12942 (optimize_range_tests): Call it.
12944 2017-09-26 Richard Biener <rguenther@suse.de>
12946 PR tree-optimization/82321
12947 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12948 for the def being inside the loop.
12950 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12952 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12954 * config/s390/s390-builtins.def: Fix constraint on op4.
12956 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12958 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12959 independent expanders.
12960 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12961 ("vec_ordered", "vec_unordered"): New expanders.
12963 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12965 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12968 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12970 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12971 vec_unpacks_lo_v16qi.
12972 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12974 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12976 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12977 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12978 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12980 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12982 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12984 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12986 ("*vec_slb<mode>"): New insn pattern.
12987 ("vec_shr_<mode>"): New expander.
12988 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12989 and force the shift count operand to V16QImode.
12990 ("vec_srb<mode>"): Set shift count mode to V16QI.
12992 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12994 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12995 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12996 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12998 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13001 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13003 2017-09-26 Richard Biener <rguenther@suse.de>
13005 PR tree-optimization/82320
13006 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13009 2017-09-25 Jeff Law <law@redhat.com>
13011 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13012 prototype for new argument.
13013 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13014 mostly extracted from rs6000_emit_allocate_stack.
13015 (rs6000_emit_probe_stack_range_stack_clash): New function.
13016 (rs6000_emit_allocate_stack): Call
13017 rs6000_emit_probe_stack_range_stack_clash as needed.
13018 (rs6000_emit_probe_stack_range): Add additional argument
13019 to call to gen_probe_stack_range{si,di}.
13020 (output_probe_stack_range): New.
13021 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13022 (output_probe_stack_range_stack_clash): New.
13023 (rs6000_emit_prologue): Emit notes into dump file as requested.
13024 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13025 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13026 Add additional operand and pass it to output_probe_stack_range.
13028 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13030 PR tree-optimization/82163
13031 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13032 (checking_verify_loop_closed_ssa): New parameter.
13033 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13034 (check_loop_closed_ssa_stmt): Delete.
13035 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13036 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13037 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13040 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13042 * brig-builtins.def: Treat HSAIL barrier builtins as
13043 setjmp/longjump style functions.
13045 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13047 * target.def (constant_alignment): New hook.
13048 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13049 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13050 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13051 * doc/tm.texi: Regenerate.
13052 * targhooks.h (default_constant_alignment): Declare.
13053 (constant_alignment_word_strings): Likewise.
13054 * targhooks.c (default_constant_alignment): New function.
13055 (constant_alignment_word_strings): Likewise.
13056 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13057 instead of CONSTANT_ALIGNMENT.
13058 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13059 (force_const_mem): Likewise.
13060 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13061 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13062 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13063 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13064 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13066 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13067 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13068 constant_alignment_word_strings.
13069 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13070 (CONSTANT_ALIGNMENT): Likewise.
13071 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13072 (arm_constant_alignment): New function.
13073 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13074 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13075 constant_alignment_word_strings.
13076 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13077 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13078 constant_alignment_word_strings.
13079 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13080 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13081 (cris_constant_alignment): New function.
13082 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13083 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13084 (epiphany_constant_alignment): New function.
13085 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13086 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13087 constant_alignment_word_strings.
13088 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13089 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13090 constant_alignment_word_strings.
13091 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13092 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13093 constant_alignment_word_strings.
13094 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13095 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13096 * config/i386/i386.c (ix86_constant_alignment): Make static.
13097 Use the same interface as the target hook.
13098 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13099 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13100 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13101 constant_alignment_word_strings.
13102 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13103 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13104 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13105 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13106 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13107 constant_alignment_word_strings.
13108 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13109 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13110 constant_alignment_word_strings.
13111 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13112 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13113 constant_alignment_word_strings.
13114 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13115 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13117 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13118 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13119 * config/mips/mips.c (mips_constant_alignment): New function.
13120 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13121 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13122 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13123 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13124 (mmix_constant_alignment): Make static. Use the same interface
13125 as the target hook.
13126 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13127 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13128 constant_alignment_word_strings.
13129 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13130 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13131 constant_alignment_word_strings.
13132 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13133 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13134 constant_alignment_word_strings.
13135 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13136 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13137 (rs6000_constant_alignment): New function.
13138 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13139 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13140 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13141 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13142 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13143 (rs6000_constant_alignment): New function.
13144 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13145 * config/s390/s390.c (s390_constant_alignment): New function.
13146 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13147 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13148 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13149 constant_alignment_word_strings.
13150 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13151 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13152 (sparc_constant_alignment): New function.
13153 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13154 * config/spu/spu.c (spu_constant_alignment): New function.
13155 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13156 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13157 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13158 constant_alignment_word_strings.
13159 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13160 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13161 constant_alignment_word_strings.
13162 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13163 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13164 constant_alignment_word_strings.
13165 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13166 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13167 (visium_constant_alignment): New function.
13168 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13169 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13170 (xtensa_constant_alignment): New function.
13171 * system.h (CONSTANT_ALIGNMENT): Poison.
13173 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13175 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13176 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13177 (rs6000_builtin_valid_without_lhs): New helper function.
13178 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13179 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13181 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13183 * target.h (vec_perm_indices): Use unsigned short rather than
13185 (auto_vec_perm_indices): Likewise.
13186 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13187 Use unsigned int rather than unsigned char.
13188 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13190 2017-09-25 Richard Biener <rguenther@suse.de>
13192 * cfgloop.h (sort_sibling_loops): Declare.
13193 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13194 (sort_sibling_loops): New function sorting the sibling loop list
13196 * graphite.c (graphite_transform_loops): Sort sibling loops.
13198 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13200 * target.def (vec_perm_const_ok): Change sel parameter to
13202 * optabs-query.c (can_vec_perm_p): Update accordingly.
13203 * doc/tm.texi: Regenerate.
13204 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13205 auto_vec_perm_indices and remove separate nelt field.
13206 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13207 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13208 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13209 (aarch64_expand_vec_perm_const): Update accordingly.
13210 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13211 to vec_perm_indices.
13212 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13213 auto_vec_perm_indices and remove separate nelt field.
13214 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13215 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13216 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13218 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13219 to vec_perm_indices.
13220 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13221 sel to vec_perm_indices.
13222 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13223 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13224 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13226 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13229 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13232 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13233 on the FUNCTION_DECL function context if it has a DIE that is a
13236 2017-09-25 Richard Biener <rguenther@suse.de>
13238 PR tree-optimization/82285
13239 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13242 2017-09-25 Tom de Vries <tom@codesourcery.com>
13246 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13249 2017-09-25 Richard Biener <rguenther@suse.de>
13251 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13252 ISL errors other than isl_error_quota happen. Dump if the
13253 schedule is the same.
13254 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13255 errors instead of aborting inside ISL.
13257 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13260 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13261 of libgcc_eh for m64.
13262 * config/i386/darwin64.h: Likewise.
13264 2017-09-25 Richard Biener <rguenther@suse.de>
13266 PR middle-end/82144
13267 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13268 attribute for incomplete types nor twice for complete ones.
13270 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13273 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13274 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13277 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13280 * config/i386/i386.opt (mprefer-avx256): Use
13281 ix86_target_flags variable.
13282 * config/i386/i386.c (ix86_target_string): Move
13283 -mprefer-avx256 to flag2_opts.
13285 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13287 PR middle-end/35691
13288 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13289 and x != -1 | y != -1 into (x & y) != -1.
13291 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13293 * config.gcc: Add new case statement to set
13294 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13295 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13296 s390x-*-linux* case statements. Added aarch64 to the list of
13297 supported architectures.
13299 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13301 PR tree-optimization/82289
13302 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13303 STMT_VINFO_RELEVANT_P.
13305 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13306 Alan Hayward <alan.hayward@arm.com>
13307 David Sherwood <david.sherwood@arm.com>
13309 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13310 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13311 (extract_range_from_binary_expr_1): Don't call
13312 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13314 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13315 Alan Hayward <alan.hayward@arm.com>
13316 David Sherwood <david.sherwood@arm.com>
13318 * target.def (preferred_vector_alignment): New hook.
13319 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13321 * doc/tm.texi: Regenerate.
13322 * targhooks.h (default_preferred_vector_alignment): Declare.
13323 * targhooks.c (default_preferred_vector_alignment): New function.
13324 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13326 (DR_TARGET_ALIGNMENT): New macro.
13327 (aligned_access_p): Update commentary.
13328 (vect_known_alignment_in_bytes): New function.
13329 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13331 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13332 Calculate the misalignment based on the target alignment rather than
13334 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13335 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13336 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13337 the target alignment, rather than masking the element misalignment
13338 with the number of elements in a vector. Also use the target
13339 alignment when calculating the maximum number of peels.
13340 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13341 instead of TYPE_ALIGN_UNIT.
13342 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13343 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13344 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13345 (vect_create_data_ref_ptr): Likewise.
13346 (vect_setup_realignment): Realign by ANDing with
13347 -DR_TARGET_MISALIGNMENT.
13348 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13349 the number of peels based on DR_TARGET_ALIGNMENT.
13350 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13351 with the guaranteed alignment boundary when deciding whether
13353 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13354 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13355 (ensure_base_align): Remove stmt_info parameter. Get the
13356 target base alignment from DR_TARGET_ALIGNMENT.
13357 (vectorizable_store): Update call accordingly. Interpret
13358 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13360 (vectorizable_load): Likewise.
13362 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13363 Alan Hayward <alan.hayward@arm.com>
13364 David Sherwood <david.sherwood@arm.com>
13366 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13367 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13368 (vect_enhance_data_refs_alignment): Likewise.
13370 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13372 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13373 error. Only quit immediately if parsing is complete.
13374 (BEGIN): Initialize fatal_err and parse_done.
13375 (begin fpu, end fpu): Check number of arguments.
13376 (begin arch, end arch): Likewise.
13377 (begin cpu, end cpu): Likewise.
13378 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13379 (optalias): Likewise.
13381 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13383 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13384 * config/arm/arm-isa.h: Delete. Move definitions to ...
13385 * arm-cpus.in: ... here. Use new feature and fgroup values.
13386 * config/arm/arm.c (arm_option_override): Use lower case for feature
13388 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13389 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13390 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13392 (print_isa_bits_for): New function.
13393 (gen_isa): New function.
13394 (gen_comm_data): Use print_isa_bits_for.
13395 (define feature): New keyword.
13396 (define fgroup): New keyword.
13397 * config/arm/t-arm (TM_H): Remove.
13398 (GTM_H): Add arm-isa.h.
13399 (arm-isa.h): Add rule to generate file.
13400 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13401 case for feature bit names.
13403 2017-09-22 Richard Biener <rguenther@suse.de>
13405 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13407 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13408 print of no dependency loops ...
13409 * graphite.c (graphite_transform_loops): ... here.
13410 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13412 (same_close_phi_node, remove_duplicate_close_phi,
13413 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13414 (canonicalize_loop_closed_ssa): ... here and simplify.
13415 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13416 (optimize_isl): Use dump_printf_loc to tell when we stopped
13417 optimizing because of an ISL timeout.
13419 2017-09-22 Richard Biener <rguenther@suse.de>
13421 PR tree-optimization/82291
13422 * tree-if-conv.c (predicate_mem_writes): Make sure to
13423 remove writes in blocks predicated with false.
13425 2017-09-22 Richard Biener <rguenther@suse.de>
13427 * sese.c: Include cfganal.h.
13428 (if_region_set_false_region): Remove.
13429 (create_if_region_on_edge): Likewise.
13430 (move_sese_in_condition): Re-implement without destroying
13433 2017-09-22 Richard Biener <rguenther@suse.de>
13435 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13436 Verify both BBs contain loop PHI nodes before dispatching to
13437 copy_loop_phi_args.
13438 (graphite_regenerate_ast_isl): Do not recompute dominators,
13439 do not verify three times. Restructure for clarity.
13440 * graphite-scop-detection.c (same_close_phi_node,
13441 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13442 defined_in_loop_p, canonicalize_loop_closed_ssa,
13443 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13444 checking and SSA rewrite, move to ...
13445 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13446 (graphite_initialize): Do not pass in ctx, do not reset the
13447 SCEV cache, compute only dominators.
13448 (graphite_transform_loops): Allocate ISL ctx after
13449 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13450 Maintain post-dominators only around build_scops.
13451 * sese.c (if_region_set_false_region): Make static. Free
13452 and recompute dominators.
13453 (move_sese_in_condition): Assert we don't get called with
13454 post-dominators computed.
13455 * sese.h (if_region_set_false_region): Remove.
13457 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13459 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13460 mode attribute for TARGET_AVX512VL.
13462 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13464 * config/i386/i386.opt (mprefer-avx256): New option.
13465 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13467 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13468 for TARGET_PREFER_AVX256.
13469 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13471 2017-09-21 Jeff Law <law@redhat.com>
13473 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13474 Fix dump output if the only stack space is for pushed registers.
13476 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13478 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13481 2017-09-21 Martin Sebor <msebor@redhat.com>
13484 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13485 code (in C++) or code that triggers warnings.
13487 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13489 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13491 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13493 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13494 * sched-rgn.c: Ditto.
13495 * sel-sched-ir.c: Ditto.
13497 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13499 * toplev.h (set_random_seed): Adjust return type.
13500 * toplev.c (init_local_tick): Move eager initialization of random_seed
13501 to get_random_seed. Adjust comment.
13502 (init_random_seed): Inline to get_random_seed, delete.
13503 (get_random_seed): Initialize random_seed lazily.
13504 (set_random_seed): Do not return previous value.
13505 (print_switch_value): Do not call get_random_seed.
13507 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13509 * cgraph.c (delete_function_version): New, broken out from...
13510 (cgraph_node::delete_function_version): ...here. Rename to
13511 cgraph_node::delete_function_version_by_decl. Update all uses.
13512 (cgraph_node::remove): Call delete_function_version.
13514 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13517 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13518 VAR_DECLs to which addressable non-volatile parameters are mapped
13519 and for id->retvar after the return value assignment. Clear
13520 id->retval and id->retbnd after inlining.
13522 2017-09-21 Richard Biener <rguenther@suse.de>
13524 PR tree-optimization/82276
13525 PR tree-optimization/82244
13526 * tree-vrp.c (build_assert_expr_for): Set
13527 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13529 (remove_range_assertions): Revert earlier change.
13531 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13534 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13536 2017-09-21 Richard Biener <rguenther@suse.de>
13538 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13539 Restore valid IL after code generation errors.
13540 * graphite.c (graphite_transform_loops): Diagnose code
13541 generation issues as MSG_MISSED_OPTIMIZATION and continue
13542 with processing SCOPs.
13544 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13545 Alan Hayward <alan.hayward@arm.com>
13546 David Sherwood <david.sherwood@arm.com>
13548 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13549 rather than choosing between plus_constant and gen_rtx_<CODE>.
13550 * expr.c (emit_push_insn): Likewise.
13551 (expand_expr_real_2): Likewise.
13553 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13554 Alan Hayward <alan.hayward@arm.com>
13555 David Sherwood <david.sherwood@arm.com>
13557 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13559 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13560 Alan Hayward <alan.hayward@arm.com>
13561 David Sherwood <david.sherwood@arm.com>
13563 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13564 calling tree_to_uhwi.
13566 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13567 Alan Hayward <alan.hayward@arm.com>
13568 David Sherwood <david.sherwood@arm.com>
13570 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13571 INTEGER_CST rather than a negative test for ADDR_EXPR.
13573 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13574 Alan Hayward <alan.hayward@arm.com>
13575 David Sherwood <david.sherwood@arm.com>
13577 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13578 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13580 2017-09-21 Richard Biener <rguenther@suse.de>
13582 PR tree-optimization/71351
13583 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13584 graphite_create_new_loop_guard): Remove, fold remaining parts
13586 (translate_isl_ast_node_for): ... here and simplify.
13588 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13591 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13592 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13593 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13594 alternative always use QI mode, for -Os imov (=R,R) alternative
13595 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13598 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13599 Jeff Law <law@redhat.com>
13601 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13602 (allocate_stack_space): New function, partially extracted from
13603 s390_emit_prologue.
13604 (s390_emit_prologue): Track offset to most recent stack probe.
13605 Code to allocate space moved into allocate_stack_space.
13606 Dump actions when no stack is allocated.
13607 (s390_prologue_plus_offset): New function.
13608 (s390_emit_stack_probe): Likewise.
13610 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13612 * common.opt (Wa, Wl, Wp, g, gz=): Add
13614 (gno-column-info): Remove.
13615 (gcolumn-info): Drop RejectNegative.
13616 (gno-): New prefix.
13617 (gno-record-gcc-switches): Remove.
13618 (grecord-gcc-switches): Drop RejectNegative.
13619 (gno-split-dwarf): Remove.
13620 (gsplit-dwarf): Drop RejectNegative.
13621 (gno-strict-dwarf): Remove.
13622 (gstrict-dwarf): Drop RejectNegative.
13623 * config/darwin.opt (gfull, gused): Add RejectNegative.
13624 * dwarf2out.c (gen_producer_string): Drop
13625 gno-record-gcc-switches handler.
13626 * optc-gen.awk: Add g to prefixes with negative forms.
13627 * opts-common.c (remapping_prefix_p): New.
13628 (find_opt): Check it.
13629 (generate_canonical_option): Test g prefix.
13630 (option_map): Add -gno- mapping.
13631 (add_misspelling_candidates): Check remapping_prefix_p.
13633 2017-09-20 Jeff Law <law@redhat.com>
13635 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13636 thinko in stack clash protection support.
13638 * explow.c (compute_stack_clash_protection_loop_data): Use
13639 CONST_INT_P instead of explicit test. Verify object is a
13640 CONST_INT_P before looking at INTVAL.
13641 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13642 instead of explicit test.
13644 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13647 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13648 address instead of to r1 and r11.
13650 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13652 * config.gcc: Support "knm".
13653 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13654 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13656 * config/i386/i386.c (m_KNM): Define.
13657 (processor_target_table): Add "knm".
13659 (ix86_option_override_internal): Add "knm".
13660 (ix86_issue_rate): Add PROCESSOR_KNM.
13661 (ix86_adjust_cost): Ditto.
13662 (ia32_multipass_dfa_lookahead): Ditto.
13663 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13664 (fold_builtin_cpu): Add M_INTEL_KNM.
13665 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13666 (processor_type): Add PROCESSOR_KNM.
13667 * config/i386/x86-tune.def: Add m_KNM.
13668 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13670 2017-09-20 Richard Biener <rguenther@suse.de>
13672 PR tree-optimization/80213
13673 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13674 are allowed in empty BBs as well.
13675 (canonicalize_loop_closed_ssa): Also look for other complex
13677 (scop_detection::get_sese): Include the loop-closed PHI block
13679 (scop_detection::merge_sese): Remove code adding extra blocks.
13680 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13681 (build_scops): Assert the final returned scop is invalid.
13683 2017-09-20 Richard Biener <rguenther@suse.de>
13685 PR tree-optimization/82264
13686 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13688 (vn_phi_lookup): Likewise.
13689 (vn_phi_insert): Likewise.
13691 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13693 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13694 that fits into uhwi or shwi, add DW_AT_const_value regardless
13695 of early_dwarf without going through RTL, using add_AT_unsigned
13698 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13699 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13700 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13701 (DEBUG_MACRO_SECTION): Likewise.
13702 (DEBUG_LTO_MACRO_SECTION): Likewise.
13703 (DEBUG_STR_DWO_SECTION): Likewise.
13704 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13705 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13706 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13707 (DEBUG_LTO_LINE_STR_SECTION): Define.
13708 (init_sections_and_labels): Initialize debug_line_str_section
13709 variable. Initialize debug_loc_section for -gdwarf-5 to
13710 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13712 2017-09-20 Richard Biener <rguenther@suse.de>
13714 * graphite-sese-to-poly.c (extract_affine): Properly handle
13715 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13717 2017-09-20 Richard Biener <rguenther@suse.de>
13719 PR tree-optimization/81373
13720 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13721 Force SESE live-out defs to be handled even if they are
13724 2017-09-19 Jeff Law <law@redhat.com>
13726 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13727 nothing for stack adjustments with REG_STACK_CHECK.
13728 * sched-deps.c (parse_add_or_inc): Reject insns with
13729 REG_STACK_CHECK from dependency breaking.
13730 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13731 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13732 * reg-notes.def (STACK_CHECK): New note.
13734 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13735 (ix86_expand_prologue): Dump stack clash info as needed.
13736 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13738 * function.c (dump_stack_clash_frame_info): New function.
13739 * function.h (dump_stack_clash_frame_info): Prototype.
13740 (enum stack_clash_probes): New enum.
13742 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13743 flag_stack_clash_protection.
13744 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13745 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13746 (arm_frame_pointer_required): Likewise.
13747 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13748 (ia64_expand_prologue): Likewise.
13749 * config/mips/mips.c (mips_expand_prologue): Likewise.
13750 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13751 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13752 (sparc_flat_expand_prologue): Likewise.
13753 * config/spu/spu.c (spu_expand_prologue): Likewise.
13755 * explow.c: Include "params.h".
13756 (anti_adjust_stack_and_probe_stack_clash): New function.
13757 (get_stack_check_protect): Likewise.
13758 (compute_stack_clash_protection_loop_data): Likewise.
13759 (emit_stack_clash_protection_loop_start): Likewise.
13760 (emit_stack_clash_protection_loop_end): Likewise.
13761 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13762 Use anti_adjust_stack_and_probe_stack_clash.
13763 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13764 (emit_stack_clash_protection_loop_start): Likewise.
13765 (emit_stack_clash_protection_loop_end): Likewise.
13766 * rtl.h (get_stack_check_protect): Prototype.
13767 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13768 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13769 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13771 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13773 * doc/tm.texi: Rebuilt.
13774 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13775 get_stack_check_protect.
13776 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13777 * config/arm/arm.c (arm_expand_prologue): Likewise.
13778 (arm_frame_pointer_required): Likewise.
13779 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13780 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13781 * config/mips/mips.c (mips_expand_prologue): Likewise.
13782 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13783 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13784 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13785 (sparc_flat_expand_prologue): Likewise.
13787 * common.opt (-fstack-clash-protection): New option.
13788 * flag-types.h (enum stack_check_type): Note difference between
13789 -fstack-check= and -fstack-clash-protection.
13790 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13791 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13792 * toplev.c (process_options): Issue warnings/errors for cases
13793 not handled with -fstack-clash-protection.
13794 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13795 (-fstack-check): Note additional problem with -fstack-check=generic.
13796 Note that -fstack-check is primarily for Ada and refer users
13797 to -fstack-clash-protection for stack-clash-protection.
13798 Document new params for stack clash protection.
13800 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13802 * config/i386/i386.c (ix86_split_long_move): Do not handle
13803 address used for LEA in a special way.
13805 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13807 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13809 2017-09-19 Martin Sebor <msebor@redhat.com>
13812 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13813 of incompatible types.
13815 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13817 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13818 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13819 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13820 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13822 2017-09-19 Richard Biener <rguenther@suse.de>
13824 PR tree-optimization/82244
13825 * tree-vrp.c (remove_range_assertions): Do not propagate
13826 a constant to abnormals but replace the assert with a copy.
13828 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13830 PR rtl-optimization/57878
13831 PR rtl-optimization/68988
13832 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13833 avoidance test involving non_reload_pseudos. Move frequency test
13834 below the general fragmentation avoidance test.
13836 2017-09-19 Richard Biener <rguenther@suse.de>
13838 PR tree-optimization/69728
13839 * graphite-sese-to-poly.c (schedule_error): New global.
13840 (add_loop_schedule): Handle empty domain by failing the
13842 (build_original_schedule): Handle schedule_error.
13844 2017-09-19 Richard Biener <rguenther@suse.de>
13846 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13847 Do not iterate to sibling loops but only to siblings of inner
13850 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13853 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13854 that the registers are different.
13856 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13858 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13859 to processor_model and "amdfam17h" to arch_names_table.
13860 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13862 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13865 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13867 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13868 Alan Hayward <alan.hayward@arm.com>
13869 David Sherwood <david.sherwood@arm.com>
13871 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13873 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13875 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13876 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13877 vect_schedule_slp_instance.
13878 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13879 Update call to vect_slp_analyze_node_operations. Simplify return
13881 (vect_slp_analyze_bb_1): Update call accordingly.
13882 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13883 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13884 (vect_schedule_slp): Update call accordingly.
13886 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13887 Alan Hayward <alan.hayward@arm.com>
13888 David Sherwood <david.sherwood@arm.com>
13890 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13891 with types that aren't in fact scalar.
13893 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13895 * tree-vect-slp.c (vect_record_max_nunits): New function,
13897 (vect_build_slp_tree_1): ...here.
13898 (vect_build_slp_tree_2): Call it for phis too.
13900 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13902 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13903 to vect_get_vec_def_for_operand when getting the mask operand.
13905 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13906 Alan Hayward <alan.hayward@arm.com>
13907 David Sherwood <david.sherwood@arm.com>
13909 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13912 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13913 Alan Hayward <alan.hayward@arm.com>
13914 David Sherwood <david.sherwood@arm.com>
13916 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13917 calculation for vector booleans.
13919 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13920 Alan Hayward <alan.hayward@arm.com>
13921 David Sherwood <david.sherwood@arm.com>
13923 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13925 (vect_transform_stmt): ...here.
13926 (vect_analyze_stmt): Use it instead of calling
13927 vectorizable_live_operation directly.
13929 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13931 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13932 non-SIMT targets in acc vector loops.
13934 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13936 * configure.ac: Add arc and check if assembler supports gdwarf2.
13937 * configure: Regenerate.
13939 2017-09-18 Richard Biener <rguenther@suse.de>
13941 PR tree-optimization/82220
13942 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13943 epilogue niters from the min_profitable_iters compute.
13945 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13948 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13949 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13950 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13952 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13955 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13956 switching to a new text section.
13958 2017-09-18 Richard Biener <rguenther@suse.de>
13960 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13962 (build_alias_set): Reject aliases with no access function.
13964 2017-09-18 Richard Biener <rguenther@suse.de>
13966 PR tree-optimization/79622
13967 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13969 (build_cross_bb_scalars_use): Likewise.
13971 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13973 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13975 2017-09-18 Alan Modra <amodra@gmail.com>
13978 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13979 stack_pointer_rtx for count 0. Update comments. Break up
13980 large rtl expression.
13982 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13984 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13985 Increase to 20 bytes.
13986 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13987 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13988 or avx version of the stub.
13990 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13993 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13994 compute the minimum stack alignment. Also update preferred stack
13995 boundary for leaf functions.
13997 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13999 PR tree-optimization/82228
14000 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14003 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14005 * common/config/nds32/nds32-common.c
14006 (nds32_option_optimization_table): Refine formatting.
14007 (nds32_option_optimization_table): Use -fsched-pressure and
14008 -fomit-frame-pointer for specific optimization level.
14010 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14012 * config/nds32/nds32.c: Refine formatting and comments.
14013 * config/nds32/nds32.h: Likewise.
14014 * config/nds32/nds32.md: Likewise.
14015 * config/nds32/nds32-cost.c: Likewise.
14016 * config/nds32/nds32-isr.c: Likewise.
14017 * config/nds32/nds32-md-auxiliary.c: Likewise.
14018 * config/nds32/nds32-multiple.md: Likewise.
14019 * config/nds32/nds32-predicates.c: Likewise.
14021 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14022 Jakub Jelinek <jakub@redhat.com>
14024 Add support for -std=c++2a.
14025 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14027 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14029 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14032 * doc/extend.texi (Common Function Attributes): Add
14033 references to ARM, AArch64, and S/390 specific attributes.
14034 (Function Specific Option Pragmas): Add AArch64 and S/390
14035 to list of back ends that support the target pragma.
14037 2017-09-15 Nathan Sidwell <nathan@acm.org>
14039 * doc/standards.texi: Fix C++17 description. Update URLs for
14042 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14044 * common.opt (Wcast-align=strict): New warning option.
14045 * doc/invoke.texi: Document -Wcast-align=strict.
14047 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14049 * cgraph.h (cgraph_thunk_info): Add comments.
14050 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14051 assert for VIRTUAL_* arguments stricter.
14053 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14055 PR tree-optimization/71026
14056 * match.pd: Move RDIV patterns from fold-const.c
14057 * fold-const.c (distribute_real_division): Removed.
14058 (fold_binary_loc): Remove calls to distribute_real_divison.
14060 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14062 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14063 c++1z and gnu++1z as deprecated. Change other references to
14064 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14065 Change -Wc++1z-compat to -Wc++17-compat.
14066 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14067 * dwarf2out.c (highest_c_language): Handle C++17.
14068 (gen_compile_unit_die): Likewise.
14070 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14072 PR rtl-optimization/82192
14073 * combine.c (make_extraction): Don't look through non-paradoxical
14074 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14077 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14078 Alan Hayward <alan.hayward@arm.com>
14079 David Sherwood <david.sherwood@arm.com>
14081 * target.def (function_arg_offset): New hook.
14082 * targhooks.h (default_function_arg_offset): Declare.
14083 * targhooks.c (default_function_arg_offset): New function.
14084 * function.c (locate_and_pad_parm): Use
14085 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14086 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14087 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14088 * doc/tm.texi: Regenerate.
14089 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14090 * config/spu/spu.c (spu_function_arg_offset): New function.
14091 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14092 * system.h (FUNCTION_ARG_OFFSET): Poison.
14094 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14095 Alan Hayard <alan.hayward@arm.com>
14096 David Sherwood <david.sherwood@arm.com>
14098 * target.def (truly_noop_truncation): New hook.
14099 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14100 than TRULY_NOOP_TRUNCATION.
14101 * hooks.h (hook_bool_uint_uint_true): Declare.
14102 * hooks.c (hook_bool_uint_uint_true): New function.
14103 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14104 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14105 * doc/tm.texi: Regenerate.
14106 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14107 rather than TRULY_NOOP_TRUNCATION in comments.
14108 (simplify_comparison): Likewise.
14109 (record_truncated_value): Likewise.
14110 * expmed.c (extract_bit_field_1): Likewise.
14111 (extract_split_bit_field): Likewise.
14112 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14113 instead of TRULY_NOOP_TRUNCATION.
14114 * function.c (assign_parm_setup_block): Likewise.
14115 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14116 * rtlhooks.c: Include target.h.
14117 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14118 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14119 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14120 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14121 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14122 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14123 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14124 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14125 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14126 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14127 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14128 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14129 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14130 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14131 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14132 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14133 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14134 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14135 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14136 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14137 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14138 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14139 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14140 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14141 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14142 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14143 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14144 rather than TRULY_NOOP_TRUNCATION in comments.
14145 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14146 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14147 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14148 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14149 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14150 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14151 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14152 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14153 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14154 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14155 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14156 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14157 rather than TRULY_NOOP_TRUNCATION in comments.
14158 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14159 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14160 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14161 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14162 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14163 TRULY_NOOP_TRUNCATION condition.
14164 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14165 (TRULY_NOOP_TRUNCATION): Delete.
14166 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14167 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14168 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14169 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14170 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14171 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14172 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14173 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14174 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14175 rather than TRULY_NOOP_TRUNCATION in comments.
14176 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14177 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14178 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14179 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14180 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14181 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14183 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14186 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14187 (*cmp_ior): Likewise.
14188 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14189 (*ior_scc_scc_cmp): Likewise.
14190 (*and_scc_scc): Likewise.
14191 (*and_scc_scc_cmp): Likewise.
14193 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14194 Alan Hayard <alan.hayward@arm.com>
14195 David Sherwood <david.sherwood@arm.com>
14197 * target.def (can_change_mode_class): New hook.
14198 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14199 (hard_regno_nregs): Likewise.
14200 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14201 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14202 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14203 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14204 (LOAD_EXTEND_OP): Update accordingly.
14205 * doc/tm.texi: Regenerate.
14206 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14207 CANNOT_CHANGE_MODE_CLASS.
14208 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14209 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14210 * combine.c (simplify_set): Update accordingly.
14211 * emit-rtl.c (validate_subreg): Likewise.
14212 * recog.c (general_operand): Likewise.
14213 * regcprop.c (mode_change_ok): Likewise.
14214 * reload1.c (choose_reload_regs): Likewise.
14215 (inherit_piecemeal_p): Likewise.
14216 * rtlanal.c (simplify_subreg_regno): Likewise.
14217 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14218 instead of CANNOT_CHANGE_MODE_CLASS.
14219 (reload_cse_simplify_operands): Likewise.
14220 * reload.c (push_reload): Use targetm.can_change_mode_class
14221 instead of CANNOT_CHANGE_MODE_CLASS.
14222 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14223 REG_CANNOT_CHANGE_MODE_P.
14224 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14225 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14226 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14227 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14228 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14229 (arm_can_change_mode_class): New function.
14230 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14231 than CANNOT_CHANGE_MODE_CLASS in comments.
14232 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14233 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14234 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14235 (ix86_can_change_mode_class): ...this new function, inverting the
14236 sense of the return value.
14237 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14238 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14239 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14240 (ia64_can_change_mode_class): New function.
14241 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14242 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14243 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14244 (m32c_can_change_mode_class): ...this new function, inverting the
14245 sense of the return value.
14246 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14247 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14248 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14249 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14250 (mips_can_change_mode_class): ...this new function, inverting the
14251 sense of the return value.
14252 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14253 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14254 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14255 (msp430_can_change_mode_class): New function.
14256 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14257 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14258 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14259 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14260 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14261 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14262 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14263 (pa_cannot_change_mode_class): Replace with...
14264 (pa_can_change_mode_class): ...this new function, inverting the
14265 sense of the return value.
14266 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14267 than CANNOT_CHANGE_MODE_CLASS in comments.
14268 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14269 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14270 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14271 (pdp11_cannot_change_mode_class): Replace with...
14272 (pdp11_can_change_mode_class): ...this new function, inverting the
14273 sense of the return value.
14274 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14275 * config/powerpcspe/powerpcspe-protos.h
14276 (rs6000_cannot_change_mode_class_ptr): Delete.
14277 * config/powerpcspe/powerpcspe.c
14278 (rs6000_cannot_change_mode_class_ptr): Delete.
14279 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14280 (rs6000_option_override_internal): Assign to
14281 targetm.can_change_mode_class instead of
14282 rs6000_cannot_change_mode_class_ptr.
14283 (rs6000_cannot_change_mode_class): Replace with...
14284 (rs6000_can_change_mode_class): ...this new function, inverting the
14285 sense of the return value.
14286 (rs6000_debug_cannot_change_mode_class): Replace with...
14287 (rs6000_debug_can_change_mode_class): ...this new function.
14288 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14289 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14290 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14291 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14292 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14294 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14295 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14296 (rs6000_option_override_internal): Assign to
14297 targetm.can_change_mode_class instead of
14298 rs6000_cannot_change_mode_class_ptr.
14299 (rs6000_cannot_change_mode_class): Replace with...
14300 (rs6000_can_change_mode_class): ...this new function, inverting the
14301 sense of the return value.
14302 (rs6000_debug_cannot_change_mode_class): Replace with...
14303 (rs6000_debug_can_change_mode_class): ...this new function.
14304 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14305 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14306 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14307 (s390_can_change_mode_class): ...this new function, inverting the
14308 sense of the return value.
14309 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14310 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14311 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14312 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14313 (sh_cannot_change_mode_class): Replace with...
14314 (sh_can_change_mode_class): ...this new function, inverting the
14315 sense of the return value.
14316 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14317 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14318 (sparc_can_change_mode_class): New function.
14319 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14320 * config/spu/spu.c (spu_can_change_mode_class): New function.
14321 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14322 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14323 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14324 (visium_can_change_mode_class): New function.
14325 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14327 2017-09-15 Richard Biener <rguenther@suse.de>
14329 PR tree-optimization/82217
14330 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14331 but not undefined case.
14333 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14336 * postreload.c (reload_cse_simplify_operands): Skip
14337 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14339 2017-09-15 Richard Biener <rguenther@suse.de>
14341 PR tree-optimization/68823
14342 * graphite-scop-detection.c (build_alias_set): If we have a
14343 possible dependence check whether we can handle them by just
14344 looking at the DRs DR_ACCESS_FNs.
14345 (build_scops): If build_alias_set fails, fail the SCOP.
14347 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14349 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14350 to support float128 built-in functions that require the ISA 3.0
14352 (BU_FLOAT128_3_HW): Likewise.
14353 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14354 built-in functions.
14355 (FMAF128): Likewise.
14357 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14358 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14359 floating point instructions.
14360 (rs6000_invalid_builtin): Likewise.
14361 (rs6000_builtin_mask_names): Likewise.
14362 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14363 (RS6000_BTM_FLOAT128_HW): Likewise.
14364 (RS6000_BTM_COMMON): Likewise.
14365 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14367 * doc/extend.texi (RS/6000 built-in functions): Document the
14368 IEEE 128-bit floating point square root and fused multiply-add
14369 built-in functions.
14371 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14373 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14374 reg (r2) isn't in the set of registers defined in the prologue.
14376 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14377 Alan Hayward <alan.hayward@arm.com>
14378 David Sherwood <david.sherwood@arm.com>
14380 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14381 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14382 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14383 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14384 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14386 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14387 max_vectorization_factor.
14388 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14390 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14391 Alan Hayward <alan.hayward@arm.com>
14392 David Sherwood <david.sherwood@arm.com>
14394 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14395 (vect_worthwhile_without_simd_p): Declare.
14396 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14397 (vectorizable_reduction): Use it.
14398 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14399 (vectorizable_operation): Likewise.
14401 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14402 Alan Hayward <alan.hayward@arm.com>
14403 David Sherwood <david.sherwood@arm.com>
14405 * tree-vectorizer.h (vect_get_num_copies): New function.
14406 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14407 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14408 (vectorizable_induction): Likewise.
14409 (vectorizable_live_operation): Likewise.
14410 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14411 (vectorizable_bswap): Likewise.
14412 (vectorizable_call): Likewise.
14413 (vectorizable_conversion): Likewise.
14414 (vectorizable_assignment): Likewise.
14415 (vectorizable_shift): Likewise.
14416 (vectorizable_operation): Likewise.
14417 (vectorizable_store): Likewise.
14418 (vectorizable_load): Likewise.
14419 (vectorizable_condition): Likewise.
14420 (vectorizable_comparison): Likewise.
14421 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14423 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14424 Alan Hayward <alan.hayward@arm.com>
14425 David Sherwood <david.sherwood@arm.com>
14427 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14428 of vect_init_vector.
14430 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14431 Alan Hayward <alan.hayward@arm.com>
14432 David Sherwood <david.sherwood@arm.com>
14434 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14435 an inline wrapper that provides a location.
14436 (gimple_build_vector): Likewise.
14437 * gimple-fold.c (gimple_build_vector_from_val): New function.
14438 (gimple_build_vector): Likewise.
14439 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14440 functions to build the initial value. Always return a gimple value.
14441 (get_initial_defs_for_reduction): Likewise. Only compute
14443 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14444 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14445 (vectorizable_induction): Use gimple_build_vector rather than
14448 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14449 Alan Hayward <alan.hayward@arm.com>
14450 David Sherwood <david.sherwood@arm.com>
14452 * target.h (vec_perm_indices): New typedef.
14453 (auto_vec_perm_indices): Likewise.
14454 * optabs-query.h: Include target.h
14455 (can_vec_perm_p): Take a vec_perm_indices *.
14456 * optabs-query.c (can_vec_perm_p): Likewise.
14457 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14458 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14459 * tree-vect-generic.c (lower_vec_perm): Likewise.
14460 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14461 (vect_grouped_load_supported): Likewise.
14462 (vect_shift_permute_load_chain): Likewise.
14463 (vect_permute_store_chain): Use auto_vec_perm_indices.
14464 (vect_permute_load_chain): Likewise.
14465 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14466 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14467 Update uses of can_vec_perm_p.
14468 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14469 mode with a number of elements. Take a vec_perm_indices *.
14470 (vect_create_epilog_for_reduction): Update accordingly.
14471 Use auto_vec_perm_indices.
14472 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14473 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14474 (vect_transform_slp_perm_load): Likewise.
14475 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14476 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14477 (vect_gen_perm_mask_checked): Likewise.
14478 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14479 (vect_gen_perm_mask_checked): Likewise.
14480 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14481 (vectorizable_store): Likewise.
14482 (vectorizable_load): Likewise.
14483 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14484 (vectorizable_bswap): Likewise.
14486 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14487 Alan Hayward <alan.hayward@arm.com>
14488 David Sherwood <david.sherwood@arm.com>
14490 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14491 * tree.c (build_vector): Likewise.
14492 (build_vector_from_ctor): Update accordingly.
14493 (build_vector_from_val): Likewise.
14494 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14495 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14496 * tree-vect-generic.c (add_rshift): Likewise.
14497 (expand_vector_divmod): Likewise.
14498 (optimize_vector_constructor): Likewise.
14499 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14500 (vect_transform_slp_perm_load): Likewise.
14501 (vect_schedule_slp_instance): Likewise.
14502 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14503 (vectorizable_call): Likewise.
14504 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14505 * expmed.c (make_tree): Likewise.
14506 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14507 a vector passed to build_vector.
14508 (fold_convert_const): Likewise.
14509 (exact_inverse): Likewise.
14510 (fold_ternary_loc): Likewise.
14511 (fold_relational_const): Likewise.
14512 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14513 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14514 (const_unop): Likewise. Store the reduction accumulator in a
14515 variable rather than an array.
14516 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14517 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14518 the new vector, rather than constructing it after the input arrays.
14519 (native_interpret_vector): Use auto_vec<tree> when building
14520 a vector passed to build_vector. Add elements in order.
14521 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14522 auto_vec<tree> when building a vector passed to build_vector.
14523 (vect_create_epilog_for_reduction): Likewise.
14524 (vectorizable_induction): Likewise.
14525 (get_initial_def_for_reduction): Likewise. Fix indentation of
14527 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14529 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14530 passed to build_vector.
14532 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14533 Alan Hayward <alan.hayward@arm.com>
14534 David Sherwood <david.sherwood@arm.com>
14536 * tree-core.h (tree_base::u): Add an "nelts" field.
14537 (tree_vector): Use VECTOR_CST_NELTS as the length.
14538 * tree.c (tree_size): Likewise.
14539 (make_vector): Initialize VECTOR_CST_NELTS.
14540 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14541 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14542 TYPE_VECTOR_SUBPARTS.
14543 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14544 as the number of units, setting it from VECTOR_CST_NELTS.
14545 (const_vector_from_tree): Likewise.
14546 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14547 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14548 (fold_negate_expr_1): Likewise.
14549 (fold_convert_const): Likewise.
14550 (const_binop): Likewise. Differentiate the number of output and
14552 (const_unop): Likewise.
14553 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14554 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14555 in cases that did the opposite.
14557 2017-09-14 Richard Biener <rguenther@suse.de>
14559 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14562 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14564 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14566 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14569 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14570 if and where to split a bb, except for splitting before debug insn
14571 sequences followed by non-label real insn. Delete debug insns
14572 in between basic blocks.
14574 * combine.c (make_compound_operation_int): Formatting fixes.
14576 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14577 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14578 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14579 * config/sol2.h (LINK_EH_SPEC): Likewise.
14580 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14581 * config/s390/linux.h (LINK_SPEC): Likewise.
14582 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14583 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14584 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14585 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14586 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14587 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14588 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14589 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14591 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14593 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14595 (ENDFILE_LINUX_SPEC): Likewise.
14596 (LINK_EH_SPEC): Likewise.
14597 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14598 (LINK_OS_LINUX_SPEC32): Likewise.
14599 (LINK_OS_LINUX_SPEC64): Likewise.
14600 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14601 (LINK_OS_LINUX_SPEC): Likewise.
14603 2017-09-13 Martin Liska <mliska@suse.cz>
14605 PR middle-end/82154
14606 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14607 CASE_HIGH is NULL_TREE.
14609 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14610 Alan Hayward <alan.hayward@arm.com>
14611 David Sherwood <david.sherwood@arm.com>
14613 * target.def (secondary_memory_needed): New hook.
14614 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14615 instead of SECONDARY_MEMORY_NEEDED.
14616 (secondary_memory_needed_mode): Likewise.
14617 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14618 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14619 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14620 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14621 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14622 * doc/tm.texi: Regenerate.
14623 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14624 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14625 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14626 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14627 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14628 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14629 mode argument first and change the reg_class arguments to reg_class_t.
14630 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14631 Make static. Update the call to inline_secondary_memory_needed.
14632 (ix86_register_move_cost): Update the call to
14633 inline_secondary_memory_needed.
14634 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14635 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14637 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14638 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14640 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14641 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14642 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14643 and match hook interface. Add comment from mips.h.
14644 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14645 * config/mmix/mmix.md (truncdfsf2): Refer to
14646 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14648 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14649 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14650 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14651 (pa_secondary_memory_needed): New function.
14652 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14653 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14654 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14655 (pdp11_secondary_memory_needed): Make static and match hook interface.
14656 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14657 * config/powerpcspe/powerpcspe-protos.h
14658 (rs6000_secondary_memory_needed_ptr): Delete.
14659 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14661 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14662 (rs6000_option_override_internal): Assign to
14663 targetm.secondary_memory_needed rather than
14664 rs6000_secondary_memory_needed_ptr.
14665 (rs6000_secondary_memory_needed): Match hook interface.
14666 (rs6000_debug_secondary_memory_needed): Likewise.
14667 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14668 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14669 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14670 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14671 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14672 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14674 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14675 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14676 (rs6000_option_override_internal): Assign to
14677 targetm.secondary_memory_needed rather than
14678 rs6000_secondary_memory_needed_ptr.
14679 (rs6000_secondary_memory_needed): Match hook interface.
14680 (rs6000_debug_secondary_memory_needed): Likewise.
14681 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14682 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14683 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14684 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14685 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14686 (sparc_secondary_memory_needed): New function.
14687 * lra-constraints.c (check_and_process_move): Refer to
14688 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14690 (curr_insn_transform): Likewise.
14691 (process_alt_operands): Use targetm.secondary_memory_needed
14692 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14693 (check_secondary_memory_needed_p): Likewise.
14694 (choose_split_class): Likewise.
14695 * reload.c: Unconditionally include code that was previously
14696 conditional on SECONDARY_MEMORY_NEEDED.
14697 (push_secondary_reload): Use targetm.secondary_memory_needed
14698 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14699 (push_reload): Likewise.
14700 * reload1.c: Unconditionally include code that was previously
14701 conditional on SECONDARY_MEMORY_NEEDED.
14702 (choose_reload_regs): Use targetm.secondary_memory_needed
14703 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14704 (gen_reload): Likewise.
14705 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14707 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14708 Alan Hayward <alan.hayward@arm.com>
14709 David Sherwood <david.sherwood@arm.com>
14711 * target.def (secondary_memory_needed_mode): New hook:
14712 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14713 * targhooks.h (default_secondary_memory_needed_mode): New function.
14714 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14715 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14716 * doc/tm.texi: Regenerate.
14717 * lra-constraints.c (check_and_process_move): Use
14718 targetm.secondary_memory_needed_mode instead of
14719 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14720 (curr_insn_transform): Likewise.
14721 * reload.c (get_secondary_mem): Likewise.
14722 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14723 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14725 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14726 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14727 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14728 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14729 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14731 * config/powerpcspe/powerpcspe-protos.h
14732 (rs6000_secondary_memory_needed_mode): Delete.
14733 * config/powerpcspe/powerpcspe.c
14734 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14735 (rs6000_secondary_memory_needed_mode): Make static.
14736 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14737 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14739 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14741 (rs6000_secondary_memory_needed_mode): Make static.
14742 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14743 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14744 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14745 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14746 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14748 (sparc_secondary_memory_needed_mode): New function.
14749 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14751 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14753 * config/aarch64/constraints.md (Umq): New constraint.
14754 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14756 (mov<mode>): Update condition.
14758 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14760 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14761 when bitposition is the same.
14763 2017-09-13 Richard Biener <rguenther@suse.de>
14765 * dwarf2out.c (output_die_symbol): Remove.
14766 (output_die): Do not output a DIEs symbol.
14768 2017-09-13 Richard Biener <rguenther@suse.de>
14770 PR middle-end/82128
14771 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14772 default-def to avoid breaking iterator update with the weird
14773 interaction with cgraph_update_edges_for_call_stmt_node.
14775 2017-09-13 Richard Biener <rguenther@suse.de>
14777 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14778 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14779 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14780 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14782 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14784 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14785 Disable pc relative literal load irrespective of
14786 TARGET_FIX_ERR_A53_84341 for default.
14788 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14790 * config/sparc/sparc.c (output_return): Output the source location of
14791 the insn in the delay slot, if any.
14792 (output_sibcall): Likewise.
14794 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14797 * common.opt (-static-pie): New alias.
14798 (shared): Negate static-pie.
14799 (-no-pie): Update help text.
14801 (static-pie): New option.
14802 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14803 -static-pie support.
14804 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14805 (LINK_EH_SPEC): Likewise.
14806 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14807 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14808 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14809 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14810 (init_gcc_specs): Likewise.
14811 (init_spec): Likewise.
14812 (display_help): Update help message for -pie.
14813 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14816 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14818 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14819 (movdi_aarch64): Likewise.
14820 (movti_aarch64): Likewise.
14822 2017-09-12 Simon Wright <simon@pushface.org>
14825 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14826 calculation of the minor version, always output as 0.
14828 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14831 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14832 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14833 on it early, rather than manual conversion late. For
14834 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14835 instead of performing manual conversion.
14837 2017-09-12 Carl Love <cel@us.ibm.com>
14839 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14840 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14842 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14843 VMULOSW): Add definitions.
14844 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14845 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14846 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14847 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14848 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14850 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14852 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14854 (movti_aarch64): Likewise.
14855 (movdf_aarch64): Likewise.
14856 (movtf_aarch64): Likewise.
14857 (load_pairdi): Likewise.
14858 (store_pairdi): Likewise.
14859 (load_pairdf): Likewise.
14860 (store_pairdf): Likewise.
14861 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14862 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14863 (ldr_got_small_<mode>): Likewise.
14864 (ldr_got_small_28k_<mode>): Likewise.
14865 (ldr_got_tiny): Likewise.
14866 * config/aarch64/iterators.md (ldst_sz): New.
14867 (ldpstp_sz): Likewise.
14868 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14870 (thunderx_load): Split load_8 to load_16.
14871 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14873 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14874 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14875 (xgene1_store_pair): Split store_8 to store_16.
14876 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14877 (falkor_st_0_st_sd): Split store_8 to store_16.
14879 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14881 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14882 and store1/2/3/4 to store_4/8/12/16.
14883 * config/aarch64/aarch64.md: Update for rename.
14884 * config/arm/arm.md: Likewise.: Likewise.
14885 * config/arm/arm.c: Likewise.
14886 * config/arm/thumb1.md: Likewise.
14887 * config/arm/thumb2.md: Likewise.
14888 * config/arm/vfp.md: Likewise.
14889 * config/arm/arm-generic.md: Likewise.
14890 * config/arm/arm1020e.md: Likewise.
14891 * config/arm/arm1026ejs.md: Likewise.
14892 * config/arm/arm1136jfs.md: Likewise.
14893 * config/arm/arm926ejs.md: Likewise.
14894 * config/arm/cortex-a15.md: Likewise.
14895 * config/arm/cortex-a17.md: Likewise.
14896 * config/arm/cortex-a5.md: Likewise.
14897 * config/arm/cortex-a53.md: Likewise.
14898 * config/arm/cortex-a57.md: Likewise.
14899 * config/arm/cortex-a7.md: Likewise.
14900 * config/arm/cortex-a8.md: Likewise.
14901 * config/arm/cortex-a9.md: Likewise.
14902 * config/arm/cortex-m4.md: Likewise.
14903 * config/arm/cortex-m7.md: Likewise.
14904 * config/arm/cortex-r4.md: Likewise.
14905 * config/arm/exynos-m1.md: Likewise.
14906 * config/arm/fa526.md: Likewise.
14907 * config/arm/fa606te.md: Likewise.
14908 * config/arm/fa626te.md: Likewise.
14909 * config/arm/fa726te.md: Likewise.
14910 * config/arm/fmp626.md: Likewise.
14911 * config/arm/iwmmxt.md: Likewise.
14912 * config/arm/ldmstm.md: Likewise.
14913 * config/arm/marvell-pj4.md: Likewise.
14914 * config/arm/xgene1.md: Likewise.
14915 * config/aarch64/thunderx.md: Likewise.
14916 * config/aarch64/thunderx2t99.md: Likewise.
14917 * config/aarch64/falkor.md: Likewise.
14919 2017-09-12 Martin Liska <mliska@suse.cz>
14921 * attribs.c (private_lookup_attribute): New function.
14922 * attribs.h (private_lookup_attribute): Declared here.
14923 (lookup_attribute): Called from this place.
14925 2017-09-12 Richard Biener <rguenther@suse.de>
14927 PR tree-optimization/82157
14928 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14929 stmts with side-effects.
14931 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14932 Alan Hayward <alan.hayward@arm.com>
14933 David Sherwood <david.sherwood@arm.com>
14935 * target.def (hard_regno_nregs): New hook.
14936 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14937 * targhooks.h (default_hard_regno_nregs): Declare.
14938 * targhooks.c (default_hard_regno_nregs): New function.
14939 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14940 (TARGET_HARD_REGNO_NREGS): ...this hook.
14941 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14942 (CLASS_MAX_NREGS): Likewise.
14943 * doc/tm.texi: Regenerate.
14944 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14945 instead of HARD_REGNO_NREGS.
14946 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14947 HARD_REGNO_NREGS in the comment.
14948 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14949 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14950 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14951 Return an unsigned int.
14952 (TARGET_HARD_REGNO_NREGS): Redefine.
14953 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14954 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14955 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14956 (arc_hard_regno_nregs): New function.
14957 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14958 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14959 (arm_hard_regno_nregs): New function.
14960 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14961 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14962 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14963 (TARGET_HARD_REGNO_NREGS): Redefine.
14964 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14965 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14966 (HARD_REGNO_NREGS): Delete.
14967 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14968 (cr16_hard_regno_nregs): New function.
14969 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14970 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14971 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14972 (cris_hard_regno_nregs): New function.
14973 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14974 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14975 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14976 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14977 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14978 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14979 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14980 (frv_hard_regno_nregs): Make static. Take and return an
14982 (frv_class_max_nregs): Remove outdated copy of documentation.
14983 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14984 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14985 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14986 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14987 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14988 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14989 (TARGET_HARD_REGNO_NREGS): Redefine.
14990 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14991 (CLASS_MAX_NREGS): Update comment.
14992 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14993 (ia64_hard_regno_nregs): New function.
14994 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14995 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14996 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14997 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14998 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15000 (m32c_hard_regno_nregs): Likewise. Make static.
15001 (TARGET_HARD_REGNO_NREGS): Redefine.
15002 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15003 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15004 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15005 (m68k_hard_regno_nregs): New function.
15006 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15007 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15008 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15009 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15010 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15011 Take and return an unsigned int.
15012 (TARGET_HARD_REGNO_NREGS): Redefine.
15013 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15014 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15015 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15016 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15017 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15018 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15019 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15020 (msp430_hard_regno_nregs): Make static. Take and return an
15022 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15023 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15024 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15025 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15026 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15027 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15028 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15029 (TARGET_HARD_REGNO_NREGS): Redefine.
15030 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15031 (PA_HARD_REGNO_NREGS): ...this.
15032 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15033 (PA_HARD_REGNO_NREGS): ...this.
15034 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15035 (pa_hard_regno_nregs): New function.
15036 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15037 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15038 (pdp11_hard_regno_nregs): New function.
15039 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15040 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15041 (rs6000_hard_regno_nregs_hook): New function.
15042 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15043 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15044 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15045 Take and return an unsigned int. Move earlier in file.
15046 (TARGET_HARD_REGNO_NREGS): Redefine.
15047 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15048 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15049 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15050 (rl78_hard_regno_nregs): Make static. Take and return an
15052 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15053 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15054 (rs6000_hard_regno_nregs_hook): New function.
15055 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15056 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15057 (TARGET_HARD_REGNO_NREGS): Redefine.
15058 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15059 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15060 instead of HARD_REGNO_NREGS.
15061 (s390_hard_regno_nregs): New function.
15062 (s390_hard_regno_mode_ok): Add comment from s390.h.
15063 (TARGET_HARD_REGNO_NREGS): Redefine.
15064 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15065 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15066 (sh_hard_regno_nregs): New function.
15067 (sh_pass_in_reg_p): Use it.
15068 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15069 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15070 (sparc_hard_regno_nregs): New function.
15071 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15072 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15073 (spu_function_arg_advance): Use it, supplying a valid register number.
15074 (TARGET_HARD_REGNO_NREGS): Redefine.
15075 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15076 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15077 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15078 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15079 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15080 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15081 (CLASS_MAX_NREGS): Remove copy of old documentation.
15082 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15083 (visium_hard_regno_nregs): New function.
15084 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15085 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15086 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15087 xtensa_hard_regno_nregs): New function.
15088 * system.h (HARD_REGNO_NREGS): Poison.
15090 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15092 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15093 hard_regno_nregs instead of HARD_REGNO_NREGS.
15094 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15095 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15096 (c6x_expand_epilogue): Likewise.
15097 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15098 (frv_read_iacc_argument): Likewise.
15099 * config/sh/sh.c: Include regs.h.
15100 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15101 (regs_used): Likewise.
15102 (output_stack_adjust): Likewise.
15103 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15104 * expmed.c: Include regs.h.
15105 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15106 * ree.c: Include regs.h.
15107 (combine_reaching_defs): Use hard_regno_nregs instead of
15109 (add_removable_extension): Likewise.
15111 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15113 * regs.h (hard_regno_nregs): Turn into a function.
15114 (end_hard_regno): Update accordingly.
15115 * caller-save.c (setup_save_areas): Likewise.
15116 (save_call_clobbered_regs): Likewise.
15117 (replace_reg_with_saved_mem): Likewise.
15118 (insert_restore): Likewise.
15119 (insert_save): Likewise.
15120 * combine.c (can_change_dest_mode): Likewise.
15121 (move_deaths): Likewise.
15122 (distribute_notes): Likewise.
15123 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15124 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15125 (rs6000_split_multireg_move): Likewise.
15126 (rs6000_register_move_cost): Likewise.
15127 (rs6000_memory_move_cost): Likewise.
15128 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15129 (rs6000_split_multireg_move): Likewise.
15130 (rs6000_register_move_cost): Likewise.
15131 (rs6000_memory_move_cost): Likewise.
15132 * cselib.c (cselib_reset_table): Likewise.
15133 (cselib_lookup_1): Likewise.
15134 * emit-rtl.c (set_mode_and_regno): Likewise.
15135 * function.c (aggregate_value_p): Likewise.
15136 * ira-color.c (setup_profitable_hard_regs): Likewise.
15137 (check_hard_reg_p): Likewise.
15138 (calculate_saved_nregs): Likewise.
15139 (assign_hard_reg): Likewise.
15140 (improve_allocation): Likewise.
15141 (calculate_spill_cost): Likewise.
15142 * ira-emit.c (modify_move_list): Likewise.
15143 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15144 (ira_hard_reg_in_set_p): Likewise.
15145 * ira.c (setup_reg_mode_hard_regset): Likewise.
15146 (clarify_prohibited_class_mode_regs): Likewise.
15147 (check_allocation): Likewise.
15148 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15149 (lra_setup_reg_renumber): Likewise.
15150 (setup_try_hard_regno_pseudos): Likewise.
15151 (spill_for): Likewise.
15152 (assign_hard_regno): Likewise.
15153 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15154 * lra-constraints.c (in_class_p): Likewise.
15155 (lra_constraint_offset): Likewise.
15156 (simplify_operand_subreg): Likewise.
15157 (lra_constraints): Likewise.
15158 (split_reg): Likewise.
15159 (split_if_necessary): Likewise.
15160 (invariant_p): Likewise.
15161 (inherit_in_ebb): Likewise.
15162 * lra-lives.c (process_bb_lives): Likewise.
15163 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15164 (get_hard_regs): Likewise.
15165 (do_remat): Likewise.
15166 * lra-spills.c (assign_spill_hard_regs): Likewise.
15167 * mode-switching.c (create_pre_exit): Likewise.
15168 * postreload.c (reload_combine_recognize_pattern): Likewise.
15169 * recog.c (peep2_find_free_register): Likewise.
15170 * regcprop.c (kill_value_regno): Likewise.
15171 (set_value_regno): Likewise.
15172 (copy_value): Likewise.
15173 (maybe_mode_change): Likewise.
15174 (find_oldest_value_reg): Likewise.
15175 (copyprop_hardreg_forward_1): Likewise.
15176 * regrename.c (check_new_reg_p): Likewise.
15177 (regrename_do_replace): Likewise.
15178 * reload.c (push_reload): Likewise.
15179 (combine_reloads): Likewise.
15180 (find_dummy_reload): Likewise.
15181 (operands_match_p): Likewise.
15182 (find_reloads): Likewise.
15183 (find_equiv_reg): Likewise.
15184 (reload_adjust_reg_for_mode): Likewise.
15185 * reload1.c (count_pseudo): Likewise.
15186 (count_spilled_pseudo): Likewise.
15187 (find_reg): Likewise.
15188 (clear_reload_reg_in_use): Likewise.
15189 (free_for_value_p): Likewise.
15190 (allocate_reload_reg): Likewise.
15191 (choose_reload_regs): Likewise.
15192 (reload_adjust_reg_for_temp): Likewise.
15193 (emit_reload_insns): Likewise.
15194 (delete_output_reload): Likewise.
15195 * rtlanal.c (subreg_get_info): Likewise.
15196 * sched-deps.c (sched_analyze_reg): Likewise.
15197 * sel-sched.c (init_regs_for_mode): Likewise.
15198 (mark_unavailable_hard_regs): Likewise.
15199 (choose_best_reg_1): Likewise.
15200 (verify_target_availability): Likewise.
15201 * valtrack.c (dead_debug_insert_temp): Likewise.
15202 * var-tracking.c (track_loc_p): Likewise.
15203 (emit_note_insn_var_location): Likewise.
15204 * varasm.c (make_decl_rtl): Likewise.
15205 * reginfo.c (choose_hard_reg_mode): Likewise.
15206 (init_reg_modes_target): Refer directly to
15207 this_target_regs->x_hard_regno_nregs.
15209 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15211 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15212 instead of hard_regno_nregs.
15214 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15216 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15217 end_hard_regno instead of hard_regno_nregs.
15218 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15219 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15220 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15221 * ira-color.c (improve_allocation): Likewise.
15222 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15223 * lra-lives.c (mark_regno_live): Likewise.
15224 (mark_regno_dead): Likewise.
15225 * lra-remat.c (operand_to_remat): Likewise.
15226 * lra.c (collect_non_operand_hard_regs): Likewise.
15227 * postreload.c (reload_combine_note_store): Likewise.
15228 (move2add_valid_value_p): Likewise.
15229 * reload.c (regno_clobbered_p): Likewise.
15231 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15233 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15235 * config/v850/v850.c (v850_reorg): Likewise.
15236 * reload.c (refers_to_regno_for_reload_p): Likewise.
15237 (find_equiv_reg): Likewise.
15238 * reload1.c (reload_reg_reaches_end_p): Likewise.
15240 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15242 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15244 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15245 * config/arm/arm.c (output_move_neon): Likewise.
15246 (arm_attr_length_move_neon): Likewise.
15247 (neon_split_vcombine): Likewise.
15248 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15249 (c6x_mark_reg_written): Likewise.
15250 (c6x_dwarf_register_span): Likewise.
15251 * config/i386/i386.c (ix86_save_reg): Likewise.
15252 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15253 (rws_access_reg): Likewise.
15254 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15255 * mode-switching.c (create_pre_exit): Likewise.
15256 * ree.c (combine_reaching_defs): Likewise.
15257 (add_removable_extension): Likewise.
15258 * regcprop.c (find_oldest_value_reg): Likewise.
15259 (copyprop_hardreg_forward_1): Likewise.
15260 * reload.c (reload_inner_reg_of_subreg): Likewise.
15261 (push_reload): Likewise.
15262 (combine_reloads): Likewise.
15263 (find_dummy_reload): Likewise.
15264 (reload_adjust_reg_for_mode): Likewise.
15265 * reload1.c (find_reload_regs): Likewise.
15266 (forget_old_reloads_1): Likewise.
15267 (reload_reg_free_for_value_p): Likewise.
15268 (reload_adjust_reg_for_temp): Likewise.
15269 (emit_reload_insns): Likewise.
15270 (delete_output_reload): Likewise.
15271 * sel-sched.c (choose_best_reg_1): Likewise.
15272 (choose_best_pseudo_reg): Likewise.
15274 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15275 Alan Hayward <alan.hayward@arm.com>
15276 David Sherwood <david.sherwood@arm.com>
15278 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15279 * target.def (slow_unaligned_access): New hook.
15280 * targhooks.h (default_slow_unaligned_access): Declare.
15281 * targhooks.c (default_slow_unaligned_access): New function.
15282 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15283 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15284 * doc/tm.texi: Regenerate.
15285 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15286 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15287 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15289 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15290 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15292 (rs6000_slow_unaligned_access): New function.
15293 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15294 (expand_block_compare): Likewise.
15295 (expand_strn_compare): Likewise.
15296 (rs6000_rtx_costs): Likewise.
15297 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15298 (riscv_slow_unaligned_access): Likewise.
15299 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15300 (riscv_slow_unaligned_access_p): ...this and make static.
15301 (riscv_option_override): Update accordingly.
15302 (riscv_slow_unaligned_access): New function.
15303 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15304 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15305 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15306 (rs6000_slow_unaligned_access): New function.
15307 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15308 (rs6000_rtx_costs): Likewise.
15309 * config/rs6000/rs6000-string.c (expand_block_compare)
15310 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15311 of SLOW_UNALIGNED_ACCESS.
15312 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15313 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15314 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15315 of SLOW_UNALIGNED_ACCESS.
15316 * expmed.c (simple_mem_bitfield_p): Likewise.
15317 * expr.c (alignment_for_piecewise_move): Likewise.
15318 (emit_group_load_1): Likewise.
15319 (emit_group_store): Likewise.
15320 (copy_blkmode_from_reg): Likewise.
15321 (emit_push_insn): Likewise.
15322 (expand_assignment): Likewise.
15323 (store_field): Likewise.
15324 (expand_expr_real_1): Likewise.
15325 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15326 * lra-constraints.c (simplify_operand_subreg): Likewise.
15327 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15328 * gimple-ssa-store-merging.c: Likewise in block comment at start
15330 * tree-ssa-strlen.c: Include target.h.
15331 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15332 of SLOW_UNALIGNED_ACCESS.
15333 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15335 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15337 PR rtl-optimization/82185
15338 * expmed.c (emit_store_flag_int): Only test tem if it has been
15341 2017-09-12 Richard Biener <rguenther@suse.de>
15343 PR middle-end/82149
15344 * match.pd ((FTYPE) N CMP CST): Fix typo.
15346 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15348 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15350 (mips_near_type_p): Add 'short_call' attribute as a synonym
15352 * doc/extend.texi (short_call): Document new function attribute.
15354 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15357 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15358 assertion check that in the condition.
15359 (get_atomic_generic_size): Likewise. Before testing if parameter
15360 has pointer type, if it has array type, call for C++
15361 default_conversion to perform array-to-pointer conversion.
15363 2017-09-12 Richard Biener <rguenther@suse.de>
15365 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15366 for operations we cannot scalarize.
15368 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15370 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15371 vectors heap vectors. Clean up comments.
15372 Make visited_bbs a reference.
15373 (profitable_jump_thread_path): Make GC
15374 vectors heap vectors. Clean up comments.
15376 (convert_and_register_jump_thread_path): Make GC vectors heap
15378 (check_subpath_and_update_thread_path): Same. Clean up comments.
15379 Make visited_bbs a reference.
15380 (handle_phi): Abstract common code to to
15381 register_jump_thread_path_if_profitable.
15382 Rename VAR_BB to DEF_BB.
15384 Make GC vectors heap vectors.
15385 Make visited_bbs a reference.
15386 (handle_assignment): Same.
15387 (register_jump_thread_path_if_profitable): New.
15388 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15390 Make GC vectors heap vectors. Clean up comments.
15391 Make visited_bbs a reference.
15392 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15393 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15396 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15399 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15400 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15402 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15404 Revert r251800 and r251799.
15406 2017-09-11 Martin Jambor <mjambor@suse.cz>
15409 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15410 arguments in advance.
15411 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15412 use it to find predecessor edges.
15413 (naive_outof_ssa): Collect vector of predecessors.
15415 2017-09-08 Jason Merrill <jason@redhat.com>
15417 PR c++/70029 - ICE with ref-qualifier and -flto
15418 * langhooks.h (struct lang_hooks_for_types): Add
15419 copy_lang_qualifiers.
15420 * attribs.c (build_type_attribute_qual_variant): Use it.
15421 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15423 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15424 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15426 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15429 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15430 (*mulsi3_sp64): New instruction.
15431 (mulsi3): New expander.
15433 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15435 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15437 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15439 * sancov.c: Include memmodel.h.
15441 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15444 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15447 2017-09-07 Carl Love <cel@us.ibm.com>
15449 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15450 the sldi instruction.
15452 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15454 * sancov.c: Include tm_p.h.
15456 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15459 * output.h (switch_to_other_text_partition): New declaration.
15460 * varasm.c (switch_to_other_text_partition): New function.
15461 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15462 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15463 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15464 to the other text partition before emitting LCL label and switch back
15465 after emitting the word after it.
15467 2017-09-07 Richard Biener <rguenther@suse.de>
15469 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15470 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15471 critical edges here, after loop init.
15472 (pass_data_pre): Remove PROP_no_crit_edges flags.
15473 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15474 for valueization of call args to avoid leaking VN_TOP.
15475 (visit_use): Assert we do not visit default defs.
15476 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15477 Use error_mark_node to more easily detect leaking VN_TOP.
15478 All default-defs are varying, not VN_TOP. Mark them visited.
15479 (run_scc_vn): Make code match comment.
15481 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15483 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15484 OPTION_MASK_FLOAT128_KEYWORD.
15485 (POWERPC_MASKS): Likewise.
15486 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15487 support for the -mfloat128-type option, and make -mfloat128
15488 default on PowerPC Linux systems. Define or undefine
15489 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15490 Define __float128 to be __ieee128 if IEEE 128-bit support is
15491 enabled, or undefine it.
15492 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15493 Delete defining __FLOAT128_TYPE__.
15494 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15495 -mfloat128-type option and make -mfloat128 default on PowerPC
15497 (TARGET_FLOAT128_TYPE): Likewise.
15498 (-mfloat128-type): Likewise.
15499 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15500 Delete the -mfloat128-type option and make -mfloat128 default on
15501 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15502 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15503 128-bit floating point is enabled. Change tests from using
15504 -mfloat128-type to -mfloat128.
15505 (rs6000_mangle_type): Use the correct mangling for the __float128
15506 type even if normal long double is restricted to 64-bits.
15507 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15509 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15510 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15511 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15513 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15514 documentation for -mfloat128.
15516 2017-09-06 Olivier Hainque <hainque@adacore.com>
15518 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15520 2017-09-06 Wish Wu <wishwu007@gmail.com>
15521 Jakub Jelinek <jakub@redhat.com>
15523 * asan.c (initialize_sanitizer_builtins): Add
15524 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15525 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15526 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15527 BT_FN_VOID_UINT64_PTR variables.
15528 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15529 (BT_FN_VOID_UINT16_UINT16): Likewise.
15530 (BT_FN_VOID_UINT32_UINT32): Likewise.
15531 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15532 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15533 (BT_FN_VOID_UINT64_PTR): Likewise.
15534 * common.opt (flag_sanitize_coverage): New variable.
15535 (fsanitize-coverage=trace-pc): Remove.
15536 (fsanitize-coverage=): Add.
15537 * flag-types.h (enum sanitize_coverage_code): New enum.
15538 * fold-const.c (fold_range_test): Disable non-short-circuit
15539 optimization if flag_sanitize_coverage.
15540 (fold_truth_andor): Likewise.
15541 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15542 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15543 (coverage_sanitizer_opts): New array.
15544 (get_closest_sanitizer_option): Add OPTS argument, handle also
15545 OPT_fsanitize_coverage_.
15546 (parse_sanitizer_options): Adjusted to also handle
15547 OPT_fsanitize_coverage_.
15548 (common_handle_option): Add OPT_fsanitize_coverage_.
15549 * sancov.c (instrument_comparison, instrument_switch): New function.
15550 (sancov_pass): Add trace-cmp support.
15551 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15552 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15553 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15554 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15555 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15556 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15557 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15558 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15559 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15560 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15562 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15564 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15565 error. Only quit immediately if parsing is complete.
15566 (BEGIN): Initialize fatal_err and parse_done.
15567 (begin fpu, end fpu): Check number of arguments.
15568 (begin arch, end arch): Likewise.
15569 (begin cpu, end cpu): Likewise.
15570 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15571 (optalias): Likewise.
15573 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15575 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15576 * config/arm/arm-isa.h: Delete. Move definitions to ...
15577 * arm-cpus.in: ... here. Use new feature and fgroup values.
15578 * config/arm/arm.c (arm_option_override): Use lower case for feature
15580 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15581 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15582 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15584 (print_isa_bits_for): New function.
15585 (gen_isa): New function.
15586 (gen_comm_data): Use print_isa_bits_for.
15587 (define feature): New keyword.
15588 (define fgroup): New keyword.
15589 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15590 (arm-isa.h): Add rule to generate file.
15591 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15592 case for feature bit names.
15594 2017-09-06 Richard Biener <rguenther@suse.de>
15596 * tree-ssa-pre.c (NECESSARY): Remove.
15597 (create_expression_by_pieces): Do not touch pass-local flags.
15598 (insert_into_preds_of_block): Likewise.
15599 (do_pre_regular_insertion): Likewise.
15600 (eliminate_insert): Likewise.
15601 (eliminate_dom_walker::before_dom_children): Likewise.
15602 (fini_eliminate): Do not look at inserted_exprs.
15603 (mark_operand_necessary): Remove.
15604 (remove_dead_inserted_code): Replace with simple work-list
15605 algorithm based on inserted_exprs and SSA uses.
15606 (pass_pre::execute): Re-order fini_eliminate and
15607 remove_dead_inserted_code.
15609 2017-09-06 Olivier Hainque <hainque@adacore.com>
15611 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15612 for VxWorks 7. Adjust surrounding comments.
15614 2017-09-06 Richard Biener <rguenther@suse.de>
15616 * gimple-ssa-strength-reduction.c
15617 (find_candidates_dom_walker::before_dom_children): Also allow
15620 2017-09-06 Richard Biener <rguenther@suse.de>
15622 PR tree-optimization/82108
15623 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15624 for gap in the non-permutation SLP case.
15626 2017-09-06 Martin Jambor <mjambor@suse.cz>
15628 PR tree-optimization/82078
15629 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15630 add_access_to_work_queue...
15631 (build_accesses_from_assign): ...here.
15632 (propagate_all_subaccesses): Make sure racc is the group
15633 representative, if there is one.
15635 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15637 PR middle-end/82095
15638 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15641 2017-09-06 Richard Biener <rguenther@suse.de>
15643 * gimple-ssa-strength-reduction.c
15644 (find_candidates_dom_walker::before_doom_children): Use a
15645 type and not a mode check.
15647 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15650 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15652 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15654 2017-09-05 Jeff Law <law@redhat.com>
15656 PR tree-optimization/64910
15657 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15658 cases where we have 3 or more operands.
15660 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15662 PR middle-end/81768
15663 * omp-low.c (lower_omp_for): Recompute tree invariant if
15664 gimple_omp_for_initial/final is ADDR_EXPR.
15666 PR middle-end/81768
15667 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15668 into gimple val before gimplification fo the COND_EXPR.
15670 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15672 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15673 REGION_COPY argument.
15674 (thread_through_all_blocks): Remove unused argument to
15675 duplicate_thread_path.
15677 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15678 Alan Hayward <alan.hayward@arm.com>
15679 David Sherwood <david.sherwood@arm.com>
15681 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15682 Take a scalar_mode rather than a machine_mode.
15683 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15684 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15685 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15686 (aarch64_gen_adjusted_ldpstp): Likewise.
15687 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15689 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15690 Alan Hayward <alan.hayward@arm.com>
15691 David Sherwood <david.sherwood@arm.com>
15693 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15694 Take a scalar_int_mode instead of a machine_mode.
15695 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15696 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15697 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15698 (aarch64_simd_attr_length_rglist): Delete.
15699 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15700 a scalar_int_mode instead of a machine_mode.
15701 (aarch64_add_offset): Likewise.
15702 (aarch64_internal_mov_immediate): Likewise
15703 (aarch64_add_constant_internal): Likewise.
15704 (aarch64_add_constant): Likewise.
15705 (aarch64_movw_imm): Likewise.
15706 (aarch64_rtx_arith_op_extract_p): Likewise.
15707 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15708 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15709 Remove assert that the mode isn't a vector.
15710 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15711 (aarch64_expand_mov_immediate): Update calls after above changes.
15712 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15713 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15714 (aarch64_move_imm): Likewise.
15715 (aarch64_can_const_movi_rtx_p): Likewise.
15716 (aarch64_strip_extend): Likewise.
15717 (aarch64_extr_rtx_p): Likewise.
15718 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15719 a CONST_INT when the mode parameter is VOIDmode.
15720 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15722 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15724 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15725 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15726 (bitwise_type_for_mode): Update accordingly.
15728 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15730 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15731 * stor-layout.c (mode_for_size_tree): Likewise.
15732 (mode_for_array): Update accordingly.
15733 (layout_decl): Likewise.
15734 (compute_record_mode): Likewise. Only set the mode once.
15736 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15738 * target.def (get_mask_mode): Change return type to opt_mode.
15740 * doc/tm.texi: Regenerate.
15741 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15742 * targhooks.c (default_get_mask_mode): Likewise.
15743 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15744 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15745 targetm.get_mask_mode.
15746 * tree.c (build_truth_vector_type): Likewise.
15748 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15750 * machmode.h (mode_for_vector): Return an opt_mode.
15751 * stor-layout.c (mode_for_vector): Likewise.
15752 (mode_for_int_vector): Update accordingly.
15753 (layout_type): Likewise.
15754 * config/i386/i386.c (emit_memmov): Likewise.
15755 (ix86_expand_set_or_movmem): Likewise.
15756 (ix86_expand_vector_init): Likewise.
15757 (ix86_get_mask_mode): Likewise.
15758 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15760 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15761 * expmed.c (extract_bit_field_1): Likewise.
15762 * expr.c (expand_expr_real_2): Likewise.
15763 * optabs-query.c (can_vec_perm_p): Likewise.
15764 (can_vec_mask_load_store_p): Likewise.
15765 * optabs.c (expand_vec_perm): Likewise.
15766 * targhooks.c (default_get_mask_mode): Likewise.
15767 * tree-vect-stmts.c (vectorizable_store): Likewise.
15768 (vectorizable_load): Likewise.
15769 (get_vectype_for_scalar_type_and_size): Likewise.
15771 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15773 * machmode.h (mode_for_int_vector): New function.
15774 * stor-layout.c (mode_for_int_vector): Likewise.
15775 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15776 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15777 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15778 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15779 (s390_expand_vcond): Likewise.
15781 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15783 * machmode.h (opt_machine_mode): New type.
15784 (opt_mode<T>): Allow construction from anything that can be
15786 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15787 (mode_for_size): Return an opt_machine_mode.
15788 * stor-layout.c (mode_for_size): Likewise.
15789 (mode_for_size_tree): Update call accordingly.
15790 (bitwise_mode_for_mode): Likewise.
15791 (make_fract_type): Likewise.
15792 (make_accum_type): Likewise.
15793 * caller-save.c (replace_reg_with_saved_mem): Update call
15795 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15796 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15797 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15798 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15799 * expmed.c (extract_bit_field_1): Likewise.
15800 * reload.c (get_secondary_mem): Likewise.
15801 * varasm.c (assemble_integer): Likewise.
15802 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15805 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15807 * machmode.h (decimal_float_mode_for_size): New function.
15808 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15809 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15810 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15811 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15812 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15813 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15815 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15817 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15818 (get_builtin_sync_mode): Likewise.
15819 (expand_ifn_atomic_compare_exchange): Likewise.
15820 (expand_builtin_atomic_clear): Likewise.
15821 (expand_builtin_atomic_test_and_set): Likewise.
15822 (fold_builtin_atomic_always_lock_free): Likewise.
15823 * calls.c (compute_argument_addresses): Likewise.
15824 (emit_library_call_value_1): Likewise.
15825 (store_one_arg): Likewise.
15826 * combine.c (combine_instructions): Likewise.
15827 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15828 * config/arm/arm.c (arm_function_value): Likewise.
15829 (aapcs_allocate_return_reg): Likewise.
15830 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15831 * config/i386/i386.c (construct_container): Likewise.
15832 (ix86_gimplify_va_arg): Likewise.
15833 (ix86_expand_sse_cmp): Likewise.
15834 (emit_memmov): Likewise.
15835 (emit_memset): Likewise.
15836 (expand_small_movmem_or_setmem): Likewise.
15837 (ix86_expand_pextr): Likewise.
15838 (ix86_expand_pinsr): Likewise.
15839 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15840 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15842 * config/mips/mips.c (mips_function_value_1) Likewise.
15843 (mips_block_move_straight): Likewise.
15844 (mips_expand_ins_as_unaligned_store): Likewise.
15845 * config/powerpcspe/powerpcspe.c
15846 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15847 (rs6000_darwin64_record_arg_flush): Likewise.
15848 * config/rs6000/rs6000.c
15849 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15850 (rs6000_darwin64_record_arg_flush): Likewise.
15851 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15852 (sparc_function_value_1): Likewise.
15853 * config/spu/spu.c (adjust_operand): Likewise.
15854 (spu_emit_branch_or_set): Likewise.
15855 (arith_immediate_p): Likewise.
15856 * emit-rtl.c (gen_lowpart_common): Likewise.
15857 * expr.c (expand_expr_real_1): Likewise.
15858 * function.c (assign_parm_setup_block): Likewise.
15859 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15860 * reload1.c (alter_reg): Likewise.
15861 * stor-layout.c (mode_for_vector): Likewise.
15862 (layout_type): Likewise.
15864 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15866 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15867 (spu_convert_move): Likewise.
15868 * lower-subreg.c (resolve_simple_move): Likewise.
15870 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15873 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15874 define_insn to a define_expand.
15875 (altivec_vsum2sws_direct): New define_insn.
15876 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15878 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15880 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15883 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15888 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15889 realign stack if stack alignment needed is less than incoming
15892 2017-09-05 Marek Polacek <polacek@redhat.com>
15895 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15898 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15900 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15902 2017-09-05 Richard Biener <rguenther@suse.de>
15904 PR tree-optimization/82084
15905 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15907 2017-09-05 Richard Biener <rguenther@suse.de>
15909 PR tree-optimization/82102
15910 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15912 2017-09-05 Martin Liska <mliska@suse.cz>
15914 PR tree-optimization/82032
15915 * tree-cfg.c (generate_range_test): New function.
15916 * tree-cfg.h (generate_range_test): Declared here.
15917 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15918 (cleanup_control_expr_graph): Use it.
15919 * tree-switch-conversion.c (try_switch_expansion): Remove
15921 (emit_case_nodes): Use generate_range_test.
15923 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15926 * config/i386/i386.md (*<btsc><mode>_mask): Add
15927 TARGET_USE_BT to insn constraint.
15928 (*btr<mode>_mask): Ditto.
15930 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15932 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15933 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15935 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15938 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15939 TARGET_NEON and TARGET_IWMMXT.
15940 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15941 TARGET_NEON and TARGET_IWMMXT.
15942 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15944 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15946 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15947 (ix86_rewrite_tls_address): Ditto.
15948 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15949 (ix86_rewrite_tls_address_1): Ditto.
15950 (ix86_rewrite_tls_address): Ditto.
15951 * config/i386/predicates.md (tls_address_pattern): New predicate.
15952 * config/i386/i386.md (TLS address splitter): New splitter.
15954 2017-09-04 Richard Biener <rguenther@suse.de>
15956 PR tree-optimization/82084
15957 * fold-const.h (can_native_encode_string_p): Declare.
15958 * fold-const.c (can_native_encode_string_p): Factor out from ...
15959 (native_encode_string): ... here.
15960 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15961 vectorizing stores from constants we later cannot handle.
15963 2017-09-04 Marek Polacek <polacek@redhat.com>
15966 * doc/invoke.texi: Update -Wtautological-compare documentation.
15968 2017-09-04 Jeff Law <law@redhat.com>
15970 PR tree-optimization/64910
15971 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15972 swap the first and last operand if the last is a constant.
15974 2017-09-04 Marek Polacek <polacek@redhat.com>
15977 * convert.c (do_narrow): When sanitizing signed integer overflows,
15978 bail out for signed types.
15979 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15981 2017-09-04 Richard Biener <rguenther@suse.de>
15983 PR tree-optimization/82060
15984 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15985 Move devirtualization after stmt folding and before EH/AB/noreturn
15986 cleanup to get the stmt refs canonicalized. Use a bool instead
15987 of gimple_modified_p since that doesn't work for NOPs. Schedule
15988 NOPs generated by folding for removal.
15990 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15991 Alan Hayward <alan.hayward@arm.com>
15992 David Sherwood <david.sherwood@arm.com>
15994 * coretypes.h (pad_direction): New enum.
15995 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15996 (FUNCTION_ARG_PADDING): Likewise.
15997 * target.def (function_arg_padding): New hook.
15998 * targhooks.h (default_function_arg_padding): Declare.
15999 * targhooks.c (default_function_arg_padding): New function.
16000 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16001 (TARGET_FUNCTION_ARG_PADDING): ...this.
16002 * doc/tm.texi: Regenerate.
16003 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16004 instead of direction.
16005 (compute_argument_addresses): Likewise.
16006 (load_register_parameters): Likewise.
16007 (emit_library_call_value_1): Likewise.
16008 (store_one_arg): Use targetm.calls.function_arg_padding instead
16009 of FUNCTION_ARG_PADDING.
16010 (must_pass_in_stack_var_size_or_pad): Likewise.
16011 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16012 (emit_group_store): Likewise.
16013 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16014 instead of FUNCTION_ARG_PADDING.
16015 (emit_push_insn): Likewise, and propagate enum change throughout
16017 * function.h (direction): Delete.
16018 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16020 * function.c (assign_parm_find_stack_rtl): Likewise.
16021 (assign_parm_setup_block_p): Likewise.
16022 (assign_parm_setup_block): Likewise.
16023 (gimplify_parameters): Likewise.
16024 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16025 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16027 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16028 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16029 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16030 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16031 (aarch64_function_arg_padding): ...this new function.
16032 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16033 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16034 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16035 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16036 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16037 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16038 (arm_pad_arg_upward): Replace with...
16039 (arm_function_arg_padding): ...this new function.
16040 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16042 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16043 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16044 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16045 (ia64_hpux_function_arg_padding): Replace with...
16046 (ia64_function_arg_padding): ...this new function. Use pad_direction
16047 instead of direction. Check for TARGET_HPUX.
16048 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16049 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16050 (iq2000_function_arg_padding): New function.
16051 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16052 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16053 (mips_function_arg_padding): ...this new function.
16054 (mips_pad_reg_upward): Update accordingly.
16055 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16056 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16057 targetm.calls.function_arg_padding.
16058 (FUNCTION_ARG_PADDING): Delete.
16059 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16060 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16061 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16062 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16063 (nios2_block_reg_padding): Return pad_direction instead of direction.
16064 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16065 instead of direction.
16066 (nios2_function_arg_padding): Likewise. Make static.
16067 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16068 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16069 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16070 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16071 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16072 (pa_function_arg_padding): Make static. Return pad_direction instead
16074 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16075 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16076 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16077 instead of direction. Use targetm.calls.function_arg_padding.
16078 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16079 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16080 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16081 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16082 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16084 (function_arg_padding): Rename to...
16085 (rs6000_function_arg_padding): ...this. Make static. Return
16086 pad_direction instead of direction.
16087 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16088 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16089 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16090 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16091 instead of direction. Use targetm.calls.function_arg_padding.
16092 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16093 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16094 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16095 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16096 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16097 (function_arg_padding): Rename to...
16098 (rs6000_function_arg_padding): ...this. Make static. Return
16099 pad_direction instead of direction.
16100 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16101 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16102 * config/s390/s390.c (s390_function_arg_padding): New function.
16103 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16104 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16105 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16106 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16107 (function_arg_padding): Rename to...
16108 (sparc_function_arg_padding): ...this. Make static. Return
16109 pad_direction instead of direction.
16110 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16111 * config/spu/spu.c (spu_function_arg_padding): New function.
16112 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16113 * system.h (FUNCTION_ARG_PADDING): Poison.
16115 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16116 Alan Hayward <alan.hayward@arm.com>
16117 David Sherwood <david.sherwood@arm.com>
16119 * target.def (modes_tieable_p): New hook.
16120 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16121 (TARGET_MODES_TIEABLE_P): ...this.
16122 * doc/tm.texi.in: Regenerate.
16123 * hooks.h (hook_bool_mode_mode_true): Declare.
16124 * hooks.c (hook_bool_mode_mode_true): New function.
16125 * combine.c (subst): Use targetm.modes_tieable_p instead of
16127 * dse.c (find_shift_sequence): Likewise.
16128 * expmed.c (extract_low_bits): Likewise.
16129 * lower-subreg.c: Include target.h.
16130 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16132 * rtlanal.c (rtx_cost): Likewise.
16133 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16134 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16135 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16136 (TARGET_MODES_TIEABLE_P): Redefine.
16137 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16138 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16139 (TARGET_MODES_TIEABLE_P): Redefine.
16140 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16141 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16142 (arc_modes_tieable_p): New function.
16143 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16144 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16145 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16146 (arm_modes_tieable_p): Make static.
16147 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16148 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16149 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16150 (TARGET_MODES_TIEABLE_P): Redefine.
16151 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16152 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16153 (TARGET_MODES_TIEABLE_P): Redefine.
16154 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16155 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16156 (cr16_modes_tieable_p): New function.
16157 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16158 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16159 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16160 (TRULY_NOOP_TRUNCATION): Update comment.
16161 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16162 (TRULY_NOOP_TRUNCATION): Update comment.
16163 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16164 (frv_modes_tieable_p): New function.
16165 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16166 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16167 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16168 (TARGET_MODES_TIEABLE_P): Redefine.
16169 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16170 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16171 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16172 (TARGET_MODES_TIEABLE_P): Redefine.
16173 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16174 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16175 (ia64_modes_tieable_p): New function.
16176 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16177 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16178 (iq2000_modes_tieable_p): New function.
16179 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16180 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16181 (lm32_modes_tieable_p): New function.
16182 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16183 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16184 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16185 (TARGET_MODES_TIEABLE_P): Redefine.
16186 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16187 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16188 (m32r_modes_tieable_p): New function.
16189 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16190 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16191 (m68k_modes_tieable_p): New function.
16192 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16193 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16194 (mcore_modes_tieable_p): New function.
16195 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16196 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16198 (TARGET_MODES_TIEABLE_P): Redefine.
16199 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16200 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16201 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16202 (TARGET_MODES_TIEABLE_P): Redefine.
16203 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16204 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16205 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16206 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16207 (mn10300_modes_tieable_p): ...this and make static.
16208 (TARGET_MODES_TIEABLE_P): Redefine.
16209 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16210 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16211 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16212 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16213 (msp430_modes_tieable_p): Make static.
16214 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16215 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16216 (TARGET_MODES_TIEABLE_P): Redefine.
16217 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16218 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16219 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16220 (TARGET_MODES_TIEABLE_P): Redefine.
16221 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16222 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16223 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16224 (TARGET_MODES_TIEABLE_P): Redefine.
16225 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16226 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16227 (pdp11_modes_tieable_p): New function.
16228 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16229 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16230 (rs6000_modes_tieable_p): New function.
16231 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16232 * config/powerpcspe/powerpcspe.md: Update comment.
16233 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16234 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16235 (TARGET_MODES_TIEABLE_P): Redefine.
16236 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16237 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16238 (rl78_modes_tieable_p): New function.
16239 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16240 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16241 (rs6000_modes_tieable_p): New function.
16242 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16243 * config/rs6000/rs6000.md: Update comment.
16244 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16245 * config/rx/rx.c (rx_modes_tieable_p): New function.
16246 (TARGET_MODES_TIEABLE_P): Redefine.
16247 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16248 * config/s390/s390.c (s390_modes_tieable_p): New function.
16249 (TARGET_MODES_TIEABLE_P): Redefine.
16250 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16251 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16252 (sh_modes_tieable_p): New function.
16253 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16254 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16255 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16256 (sparc_modes_tieable_p): Make static.
16257 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16258 * config/spu/spu.c (spu_modes_tieable_p): New function.
16259 (TARGET_MODES_TIEABLE_P): Redefine.
16260 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16261 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16262 (TARGET_MODES_TIEABLE_P): Redefine.
16263 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16264 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16265 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16266 * config/v850/v850.c (v850_modes_tieable_p): New function.
16267 (TARGET_MODES_TIEABLE_P): Redefine.
16268 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16269 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16270 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16271 (visium_modes_tieable_p): New function.
16272 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16273 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16274 (xtensa_modes_tieable_p): New function.
16275 * system.h (MODES_TIEABLE_P): Poison.
16277 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16278 Alan Hayward <alan.hayward@arm.com>
16279 David Sherwood <david.sherwood@arm.com>
16281 * target.def (hard_regno_mode_ok): New hook.
16282 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16283 (TARGET_HARD_REGNO_MODE_OK): ...this.
16284 * doc/tm.texi.in: Regenerate.
16285 * hooks.h (hook_bool_uint_mode_true): Declare.
16286 * hooks.c (hook_bool_uint_mode_true): New function.
16287 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16288 HARD_REGNO_MODE_OK.
16289 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16290 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16291 instead of HARD_REGNO_MODE_OK.
16292 * caller-save.c: Include target.h.
16293 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16294 HARD_REGNO_MODE_OK.
16295 * combine.c (can_combine_p): Likewise.
16296 (combinable_i3pat): Likewise.
16297 (can_change_dest_mode): Likewise.
16298 * expr.c (init_expr_target): Likewise.
16299 (convert_move): Likewise.
16300 (convert_modes): Likewise.
16301 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16302 (setup_prohibited_mode_move_regs): Likewise.
16303 * ira.h (target_ira): Likewise.
16304 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16305 * lra-constraints.c (process_alt_operands): Likewise.
16306 (split_reg): Likewise.
16307 * recog.c (peep2_find_free_register): Likewise.
16308 * ree.c (combine_reaching_defs): Likewise.
16309 * regcprop.c (maybe_mode_change): Likewise.
16310 * reginfo.c (init_reg_sets_1): Likewise.
16311 (choose_hard_reg_mode): Likewise.
16312 (simplifiable_subregs): Likewise.
16313 * regrename.c (check_new_reg_p): Likewise.
16314 * reload.c (find_valid_class): Likewise.
16315 (find_valid_class_1): Likewise.
16316 (reload_inner_reg_of_subreg): Likewise.
16317 (push_reload): Likewise.
16318 (combine_reloads): Likewise.
16319 (find_dummy_reload): Likewise.
16320 (find_reloads): Likewise.
16321 * reload1.c (find_reg): Likewise.
16322 (set_reload_reg): Likewise.
16323 (allocate_reload_reg): Likewise.
16324 (choose_reload_regs): Likewise.
16325 (reload_adjust_reg_for_temp): Likewise.
16326 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16327 (simplify_subreg_regno): Likewise.
16328 * sel-sched.c (init_regs_for_mode): Likewise.
16329 * varasm.c (make_decl_rtl): Likewise.
16330 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16331 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16332 HARD_REGNO_MODE_OK.
16333 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16334 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16335 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16336 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16337 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16338 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16339 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16340 (arc_mode_class): Delete.
16341 (HARD_REGNO_MODE_OK): Delete.
16342 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16343 (arc_hard_regno_mode_ok): Rename old array to...
16344 (arc_hard_regno_mode_ok_modes): ...this.
16345 (arc_conditional_register_usage): Update accordingly.
16346 (arc_mode_class): Make static.
16347 (arc_hard_regno_mode_ok): New function.
16348 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16349 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16350 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16351 (arm_hard_regno_mode_ok): Make static.
16352 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16353 HARD_REGNO_MODE_OK.
16354 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16355 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16356 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16358 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16359 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16360 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16361 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16362 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16363 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16364 * config/bfin/predicates.md (valid_reg_operand): Use
16365 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16366 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16367 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16368 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16369 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16370 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16371 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16372 (cr16_hard_regno_mode_ok): Make static and return a bool.
16373 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16374 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16375 (cris_hard_regno_mode_ok): New function.
16376 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16377 (epiphany_mode_class): Delete.
16378 (HARD_REGNO_MODE_OK): Delete.
16379 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16380 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16381 (hard_regno_mode_ok): Rename to...
16382 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16383 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16384 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16385 HARD_REGNO_MODE_OK.
16386 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16387 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16388 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16389 (frv_hard_regno_mode_ok): Make static and return a bool.
16390 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16391 HARD_REGNO_MODE_OK.
16392 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16393 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16394 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16395 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16397 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16398 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16399 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16400 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16402 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16403 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16404 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16405 (ia64_hard_regno_mode_ok): New function.
16406 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16407 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16408 (iq2000_hard_regno_mode_ok): New function.
16409 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16410 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16411 (lm32_hard_regno_mode_ok): New function.
16412 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16413 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16414 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16415 instead of HARD_REGNO_MODE_OK.
16416 (m32c_hard_regno_ok): Rename to...
16417 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16418 (m32c_cannot_change_mode_class): Update accordingly.
16419 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16420 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16421 (m32r_mode_class): Delete.
16422 (HARD_REGNO_MODE_OK): Delete.
16423 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16424 (m32r_hard_regno_mode_ok): Rename to...
16425 (m32r_hard_regno_modes): ...this.
16426 (m32r_mode_class): Make static.
16427 (m32r_hard_regno_mode_ok): New function.
16428 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16429 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16430 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16431 (m68k_hard_regno_mode_ok): Make static.
16432 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16433 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16434 (mcore_hard_regno_mode_ok): New function.
16435 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16436 (HARD_REGNO_MODE_OK): Delete.
16437 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16439 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16440 (microblaze_hard_regno_mode_ok): New function.
16441 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16442 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16443 (mips_hard_regno_mode_ok): Delete.
16444 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16445 (mips_hard_regno_mode_ok_p): ...this and make static.
16446 (mips_hard_regno_mode_ok_p): Rename to...
16447 (mips_hard_regno_mode_ok_uncached): ...this.
16448 (mips_hard_regno_mode_ok): New function.
16449 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16450 of HARD_REGNO_MODE_OK.
16451 (mips_option_override): Update after above name changes.
16452 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16453 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16454 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16455 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16456 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16457 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16458 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16459 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16460 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16461 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16462 (msp430_hard_regno_mode_ok): Make static and return a bool.
16463 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16464 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16465 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16467 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16468 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16469 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16470 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16471 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16472 (PA_HARD_REGNO_MODE_OK): ...this
16473 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16474 (PA_HARD_REGNO_MODE_OK): ...this.
16475 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16476 (pa_hard_regno_mode_ok): New function.
16477 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16478 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16479 (pdp11_hard_regno_mode_ok): New function.
16480 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16481 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16483 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16485 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16486 (rs6000_hard_regno_mode_ok): Rename to...
16487 (rs6000_hard_regno_mode_ok_uncached): ...this.
16488 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16489 (rs6000_hard_regno_mode_ok): New function.
16490 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16491 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16492 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16493 (riscv_hard_regno_mode_ok): ...this and make static.
16494 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16495 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16496 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16497 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16498 (rl78_hard_regno_mode_ok): Make static and return bool.
16499 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16500 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16502 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16503 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16504 (rs6000_hard_regno_mode_ok): Rename to...
16505 (rs6000_hard_regno_mode_ok_uncached): ...this.
16506 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16507 (rs6000_hard_regno_mode_ok): New function.
16508 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16509 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16510 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16511 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16512 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16513 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16514 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16515 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16516 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16517 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16518 (sh_hard_regno_mode_ok): Make static.
16519 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16520 instead of HARD_REGNO_MODE_OK.
16521 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16522 (sparc_mode_class): Delete.
16523 (HARD_REGNO_MODE_OK): Delete.
16524 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16525 (hard_regno_mode_classes): Make static.
16526 (sparc_mode_class): Likewise.
16527 (sparc_hard_regno_mode_ok): New function.
16528 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16529 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16530 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16532 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16533 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16534 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16535 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16536 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16537 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16538 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16539 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16540 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16541 (visium_hard_regno_mode_ok): New function.
16542 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16543 instead of HARD_REGNO_MODE_OK.
16544 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16545 (HARD_REGNO_MODE_OK): Delete.
16546 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16547 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16548 (xtensa_option_override): Update accordingly.
16549 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16550 (xtensa_hard_regno_mode_ok): New function.
16551 * system.h (HARD_REGNO_MODE_OK): Poison.
16553 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16554 Alan Hayward <alan.hayward@arm.com>
16555 David Sherwood <david.sherwood@arm.com>
16557 * target.def (hard_regno_call_part_clobbered): New hook.
16558 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16559 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16560 * doc/tm.texi: Regenerate.
16561 * hooks.h (hook_bool_uint_mode_false): Declare.
16562 * hooks.c (hook_bool_uint_mode_false): New function.
16563 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16564 * cselib.c (cselib_process_insn): Use
16565 targetm.hard_regno_call_part_clobbered instead of
16566 HARD_REGNO_CALL_PART_CLOBBERED.
16567 * ira-conflicts.c (ira_build_conflicts): Likewise.
16568 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16569 * lra-constraints.c (need_for_call_save_p): Likewise.
16570 * lra-lives.c: Include target.h.
16571 (check_pseudos_live_through_calls): Use
16572 targetm.hard_regno_call_part_clobbered instead of
16573 HARD_REGNO_CALL_PART_CLOBBERED.
16574 * regcprop.c: Include target.h.
16575 (copyprop_hardreg_forward_1): Use
16576 targetm.hard_regno_call_part_clobbered instead of
16577 HARD_REGNO_CALL_PART_CLOBBERED.
16578 * reginfo.c (choose_hard_reg_mode): Likewise.
16579 * regrename.c (check_new_reg_p): Likewise.
16580 * reload.c (find_equiv_reg): Likewise.
16581 * reload1.c (emit_reload_insns): Likewise.
16582 * sched-deps.c (deps_analyze_insn): Likewise.
16583 * sel-sched.c (init_regs_for_mode): Likewise.
16584 (mark_unavailable_hard_regs): Likewise.
16585 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16586 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16587 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16589 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16590 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16591 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16593 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16595 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16596 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16597 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16599 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16600 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16601 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16603 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16604 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16606 * config/powerpcspe/powerpcspe.c
16607 (rs6000_hard_regno_call_part_clobbered): New function.
16608 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16609 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16610 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16612 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16613 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16614 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16616 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16617 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16618 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16620 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16621 Alan Hayward <alan.hayward@arm.com>
16622 David Sherwood <david.sherwood@arm.com>
16624 * rtl.h (subreg_memory_offset): Declare.
16625 * emit-rtl.c (subreg_memory_offset): New function.
16626 * expmed.c (store_bit_field_1): Use it.
16627 * expr.c (undefined_operand_subword_p): Likewise.
16628 * simplify-rtx.c (simplify_subreg): Likewise.
16630 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16632 PR rtl-optimization/57448
16635 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16636 using atomic_load pattern.
16637 (expand_atomic_store): Likewise.
16639 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16642 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16643 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16644 handling. Use replace_call_with_value with NULL instead of
16645 gsi_replace, unlink_stmt_vdef and release_defs.
16647 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16650 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16652 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16655 * rtl.h (emit_library_call_value_1): Declare.
16656 (emit_library_call): Replace declaration with a series of overloads.
16657 Remove the parameter count argument.
16658 (emit_library_call_value): Likewise.
16659 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16660 with an "rtx_mode_t *".
16661 (emit_library_call_value): Delete.
16662 (emit_library_call): Likewise.
16663 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16664 (asan_emit_allocas_unpoison): Likewise.
16665 * builtins.c (expand_builtin_powi): Likewise.
16666 (expand_asan_emit_allocas_unpoison): Likewise.
16667 * cfgexpand.c (expand_main_function): Likewise.
16668 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16669 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16670 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16671 * config/arm/arm.c (arm_trampoline_init): Likewise.
16672 (arm_call_tls_get_addr): Likewise.
16673 (arm_expand_divmod_libfunc): Likewise.
16674 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16675 (smulsi3_highpart): Likewise.
16676 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16677 (c6x_expand_compare): Likewise.
16678 (c6x_expand_movmem): Likewise.
16679 * config/frv/frv.c (frv_trampoline_init): Likewise.
16680 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16681 (ix86_expand_divmod_libfunc): Likewise.
16682 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16683 (ia64_expand_compare): Likewise.
16684 (ia64_profile_hook): Likewise.
16685 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16686 (nonlocal_goto): Likewise.
16687 (restore_stack_nonlocal): Likewise.
16688 * config/m32r/m32r.c (block_move_call): Likewise.
16689 (m32r_trampoline_init): Likewise.
16690 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16691 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16692 (m68k_call_m68k_read_tp): Likewise.
16693 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16694 (microblaze_expand_divide): Likewise.
16695 * config/mips/mips.h (mips_args): Likewise.
16696 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16697 (MIPS_ICACHE_SYNC): Likewise.
16698 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16699 (nios2_trampoline_init): Likewise.
16700 * config/pa/pa.c (hppa_tls_call): Likewise.
16701 (pa_trampoline_init): Likewise.
16702 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16703 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16704 (expand_strn_compare): Likewise.
16705 (rs6000_generate_compare): Likewise.
16706 (rs6000_expand_float128_convert): Likewise.
16707 (output_profile_hook): Likewise.
16708 (rs6000_trampoline_init): Likewise.
16709 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16710 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16711 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16712 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16713 (rs6000_generate_compare): Likewise.
16714 (rs6000_expand_float128_convert): Likewise.
16715 (output_profile_hook): Likewise.
16716 (rs6000_trampoline_init): Likewise.
16717 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16718 * config/sh/sh.c (sh_trampoline_init): Likewise.
16719 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16720 (sparc_emit_float_lib_cmp): Likewise.
16721 (sparc32_initialize_trampoline): Likewise.
16722 (sparc64_initialize_trampoline): Likewise.
16723 (sparc_profile_hook): Likewise.
16724 * config/spu/spu.c (ea_load_store): Likewise.
16725 * config/spu/spu.md (floatunssidf2): Likewise.
16726 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16727 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16728 * config/visium/visium.c (expand_block_move_4): Likewise.
16729 (expand_block_move_2): Likewise.
16730 (expand_block_move_1): Likewise.
16731 (expand_block_set_4): Likewise.
16732 (expand_block_set_2): Likewise.
16733 (expand_block_set_1): Likewise.
16734 (visium_trampoline_init): Likewise.
16735 (visium_profile_hook): Likewise.
16736 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16737 (xtensa_setup_frame_addresses): Likewise.
16738 (xtensa_trampoline_init): Likewise.
16739 * except.c (sjlj_emit_function_enter): Likewise.
16740 (sjlj_emit_function_exit): Likewise.
16741 * explow.c (allocate_dynamic_stack_space): Likewise.
16742 (probe_stack_range): Likewise.
16743 * expr.c (convert_mode_scalar): Likewise.
16744 * optabs.c (expand_binop): Likewise.
16745 (expand_twoval_binop_libfunc): Likewise.
16746 (expand_unop): Likewise.
16747 (prepare_cmp_insn): Likewise.
16748 (prepare_float_lib_cmp): Likewise.
16749 (expand_float): Likewise.
16750 (expand_fix): Likewise.
16751 (expand_fixed_convert): Likewise.
16752 (maybe_emit_sync_lock_test_and_set): Likewise.
16753 (expand_atomic_compare_and_swap): Likewise.
16754 (expand_mem_thread_fence): Likewise.
16755 (expand_atomic_fetch_op): Likewise.
16757 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16759 * doc/generic.texi (OpenACC): Adjust URL.
16760 * doc/invoke.texi (C Dialect Options): Ditto.
16762 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16764 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16765 predicate for operand 1. Add (m,<S>) constraint.
16766 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16767 Prevent memory operand 1 with register operand 2.
16769 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16771 PR rtl-optimization/82024
16772 * combine.c (try_combine): If the combination result is a PARALLEL,
16773 and we only need to retain the SET in there that would be placed
16774 at I2, check that we can place that at I3 instead, before doing so.
16776 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16779 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16781 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16782 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16785 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16786 Jeff Law <law@redhat.com>
16788 * varasm.c (bss_initializer_p): Do not put constants into .bss
16789 (categorize_decl_for_section): Handle bss_initializer_p returning
16790 false when DECL_INITIAL is NULL.
16792 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16795 * config/s390/s390.c (s390_can_inline_p): New function.
16797 2017-09-01 Jeff Law <law@redhat.com>
16799 PR tree-optimization/82052
16800 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16801 Always initialize the returned slot after a hash table miss
16802 when INSERT is true.
16804 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16806 * config/s390/s390.md (mem_signal_fence): Remove.
16807 * doc/md.texi (mem_signal_fence): Remove.
16808 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16810 * target-insns.def (mem_signal_fence): Remove.
16812 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16815 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16818 * asan.c (create_odr_indicator): Strip name encoding from assembler
16819 name before appending it after __odr_asan_.
16821 2017-09-01 Martin Liska <mliska@suse.cz>
16823 PR tree-optimization/82059
16824 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16825 frequency only when an edge is redirected.
16827 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16829 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16830 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16831 (arc_conditional_register_usage): Remove ARC600 lp_count
16833 (arc_file_start): Emit Tag_ARC_CPU_variation.
16834 (arc_can_use_doloop_p): New conditions to use ZOLs.
16835 (hwloop_fail): New function.
16836 (hwloop_optimize): Likewise.
16837 (hwloop_pattern_reg): Likewise.
16838 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16839 (arc_reorg_loops): New function, calls reorg_loops.
16840 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16841 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16843 (arc_loop_hazard): Remove function, functionality moved into
16845 (arc_hazard): Remove arc_loop_hazard call.
16846 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16847 into hwloop_optimize.
16848 (arc_label_align): Remove ZOL handling.
16849 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16850 * config/arc/arc.md (doloop_begin): Remove pattern.
16851 (doloop_begin_i): Likewise.
16852 (doloop_end_i): Likewise.
16853 (doloop_fallback): Likewise.
16854 (doloop_fallback_m): Likewise.
16855 (doloop_end): Reimplement expand.
16856 (arc_lp): New pattern for LP instruction.
16857 (loop_end): New pattern.
16858 (loop_fail): Likewise.
16859 (decrement_and_branch_until_zero): Likewise.
16860 * config/arc/arc.opt (mlpc-width): New option.
16861 * doc/invoke.texi (mlpc-width): Document option.
16863 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16865 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16866 (arc_ccfsm_advance): Fix checking for delay slots.
16867 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16869 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16871 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16873 (movhi_insn): Update store instruction constraint which are saving
16874 6-bit short immediates.
16875 (movsi_insn): Consider also short scaled load operations.
16876 (zero_extendhisi2_i): Use Usd constraint instead of T.
16877 (extendhisi2_i): Add q constraint.
16878 (arc_clzsi2): Add type and length attributes.
16879 (arc_ctzsi2): Likewise.
16880 * config/arc/constraints.md (Usc): Update constraint, the
16881 assembler can parse two relocations for a single instruction.
16883 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16885 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16886 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16888 2017-08-31 Olivier Hainque <hainque@adacore.com>
16890 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16891 match as powerpc-wrs-vxworks*.
16893 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16895 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16896 register constraint for by-element operand.
16897 (aarch64_mls_elt_merge<mode>): Likewise.
16899 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16901 * config/arc/arc.c (arc_can_follow_jump): Check for short
16904 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16906 * config.gcc: Use g.opt for arc.
16907 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16908 functionality moved to ...
16909 (legitimate_scaled_address_p): New function, ...here.
16910 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16911 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16912 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16914 (arc_override_options): Handle G option.
16915 (arc_output_pic_addr_const): Correct function definition.
16916 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16917 (arc_decl_anon_ns_mem_p): Delete.
16918 (arc_in_small_data_p): Overhaul this function to take into
16919 consideration the value given via G option.
16920 (arc_rewrite_small_data_1): Renamed and corrected old
16921 arc_rewrite_small_data function.
16922 (arc_rewrite_small_data): New function.
16923 (small_data_pattern): Don't use pic_offset_table_rtx.
16924 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16925 * config/arc/simdext.md (movmisalignv2hi): Use
16926 prepare_move_operands function.
16928 (movmisalign*): Likewise.
16929 * doc/invoke.texi (ARC options): Document -G option.
16931 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16933 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16935 * config/arc/arc.c (arc_print_operand): Output scalled address for
16936 sdata whenever is possible.
16937 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16938 load/stores are available.
16939 (compact_sda_memory_operand): Check for the alignment required by
16940 code density instructions.
16941 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16943 * config/arc/constraints.md (Usd): Update constraint.
16944 (Us0): New constraint.
16945 (Usc): Update constraint.
16947 2017-08-31 Richard Biener <rguenther@suse.de>
16949 PR middle-end/82054
16950 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16951 function only once.
16953 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16955 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16956 Resize type_signature.
16958 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16959 Alan Hayward <alan.hayward@arm.com>
16960 David Sherwood <david.sherwood@arm.com>
16962 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16963 subregs whose inner modes can be stored in GPRs.
16964 (aarch64_classify_index): Likewise.
16966 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16967 Alan Hayward <alan.hayward@arm.com>
16968 David Sherwood <david.sherwood@arm.com>
16970 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16971 (V_INT_EQUIV): ...this.
16972 (v_cmp_result): Rename to...
16973 (v_int_equiv): ...this.
16974 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16975 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16976 (copysign<mode>3): Likewise.
16977 (aarch64_simd_bsl<mode>_internal): Likewise.
16978 (aarch64_simd_bsl<mode>): Likewise.
16979 (vec_cmp<mode><mode>): Likewise.
16980 (vcond<mode><mode>): Likewise.
16981 (vcond<v_cmp_mixed><mode>): Likewise.
16982 (vcondu<mode><v_cmp_mixed>): Likewise.
16983 (aarch64_cm<optab><mode>): Likewise.
16984 (aarch64_cmtst<mode>): Likewise.
16985 (aarch64_fac<optab><mode>): Likewise.
16986 (vec_perm_const<mode>): Likewise.
16987 (vcond_mask_<mode><v_cmp_result>): Rename to...
16988 (vcond_mask_<mode><v_int_equiv>): ...this.
16989 (vec_cmp<mode><v_cmp_result>): Rename to...
16990 (vec_cmp<mode><v_int_equiv>): ...this.
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-modes.def: Remove 32-, 48- and 64-byte
16998 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16999 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17000 (UNSPEC_LD4_DREG): New unspecs.
17001 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17002 (aarch64_ld2<mode>_dreg_be): Replace with...
17003 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17005 (aarch64_ld3<mode>_dreg_le)
17006 (aarch64_ld3<mode>_dreg_be): Replace with...
17007 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17009 (aarch64_ld4<mode>_dreg_le)
17010 (aarch64_ld4<mode>_dreg_be): Replace with...
17011 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17014 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17016 PR tree-optimization/81987
17017 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17018 insert an initializer in a location not dominated by the stride
17021 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17023 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17024 on the entire header of the finally block in the fallthru case.
17026 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17028 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17030 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17032 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17033 rs6000_emit_move_from_cr and call renamed function.
17034 (rs6000_emit_prologue): Call renamed functions.
17035 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17036 movesi_from_cr, remove volatile CRs.
17038 2017-08-30 Jon Beniston <jon@beniston.com>
17039 Richard Biener <rguenther@suse.de>
17041 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17042 of VECTOR_MODE_P check.
17043 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17044 element vector types.
17046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17048 * df.h (df_read_modify_subreg_p): Remove in favor of...
17049 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17050 const_rtx instead of an rtx.
17051 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17052 * df-problems.c (df_word_lr_mark_ref): Likewise.
17053 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17054 (mark_pseudo_reg_dead): Likewise.
17055 (mark_ref_dead): Likewise.
17056 * reginfo.c (init_subregs_of_mode): Likewise.
17057 * sched-deps.c (sched_analyze_1): Likewise.
17058 * df-scan.c (df_def_record_1): Likewise.
17059 (df_uses_record): Likewise.
17060 (df_read_modify_subreg_p): Remove in favor of...
17061 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17062 const_rtx instead of an rtx.
17064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17065 Alan Hayward <alan.hayward@arm.com>
17066 David Sherwood <david.sherwood@arm.com>
17068 * rtl.h (partial_subreg_p): New function.
17069 * caller-save.c (save_call_clobbered_regs): Use it.
17070 * calls.c (expand_call): Likewise.
17071 * combine.c (combinable_i3pat): Likewise.
17072 (simplify_set): Likewise.
17073 (make_extraction): Likewise.
17074 (make_compound_operation_int): Likewise.
17075 (gen_lowpart_or_truncate): Likewise.
17076 (force_to_mode): Likewise.
17077 (make_field_assignment): Likewise.
17078 (reg_truncated_to_mode): Likewise.
17079 (record_truncated_value): Likewise.
17080 (move_deaths): Likewise.
17081 * cse.c (record_jump_cond): Likewise.
17082 (cse_insn): Likewise.
17083 * cselib.c (cselib_lookup_1): Likewise.
17084 * expmed.c (extract_bit_field_using_extv): Likewise.
17085 * function.c (assign_parm_setup_reg): Likewise.
17086 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17087 * ira-build.c (create_insn_allocnos): Likewise.
17088 * lra-coalesce.c (merge_pseudos): Likewise.
17089 * lra-constraints.c (match_reload): Likewise.
17090 (simplify_operand_subreg): Likewise.
17091 (curr_insn_transform): Likewise.
17092 * lra-lives.c (process_bb_lives): Likewise.
17093 * lra.c (new_insn_reg): Likewise.
17094 (lra_substitute_pseudo): Likewise.
17095 * regcprop.c (mode_change_ok): Likewise.
17096 (maybe_mode_change): Likewise.
17097 (copyprop_hardreg_forward_1): Likewise.
17098 * reload.c (push_reload): Likewise.
17099 (find_reloads): Likewise.
17100 (find_reloads_subreg_address): Likewise.
17101 * reload1.c (alter_reg): Likewise.
17102 (eliminate_regs_1): Likewise.
17103 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17105 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17107 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17108 back to if statements, including unpack.
17110 2017-08-30 Martin Liska <mliska@suse.cz>
17112 PR inline-asm/82001
17113 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17115 (func_checker::compare_asm_inputs_outputs): ... this function.
17116 (func_checker::compare_gimple_asm): Use the function to compare
17117 also ASM constrains.
17118 * ipa-icf-gimple.h: Rename the function.
17120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17121 Alan Hayward <alan.hayward@arm.com>
17122 David Sherwood <david.sherwood@arm.com>
17124 * coretypes.h (complex_mode): New type.
17125 * gdbhooks.py (build_pretty_printer): Handle it.
17126 * machmode.h (complex_mode): New class.
17127 (complex_mode::includes_p): New function.
17128 (is_complex_int_mode): Likewise.
17129 (is_complex_float_mode): Likewise.
17130 * genmodes.c (get_mode_class): Handle complex mode classes.
17131 * function.c (expand_function_end): Use is_complex_int_mode.
17133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17134 Alan Hayward <alan.hayward@arm.com>
17135 David Sherwood <david.sherwood@arm.com>
17137 * coretypes.h (scalar_mode_pod): New typedef.
17138 * gdbhooks.py (build_pretty_printer): Handle it.
17139 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17140 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17141 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17142 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17143 as_a <scalar_mode>.
17145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17146 Alan Hayward <alan.hayward@arm.com>
17147 David Sherwood <david.sherwood@arm.com>
17149 * machmode.h (mode_for_vector): Take a scalar_mode instead
17151 * stor-layout.c (mode_for_vector): Likewise.
17152 * explow.c (promote_mode): Use as_a <scalar_mode>.
17153 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17156 Alan Hayward <alan.hayward@arm.com>
17157 David Sherwood <david.sherwood@arm.com>
17159 * target.def (preferred_simd_mode): Take a scalar_mode
17160 instead of a machine_mode.
17161 * targhooks.h (default_preferred_simd_mode): Likewise.
17162 * targhooks.c (default_preferred_simd_mode): Likewise.
17163 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17164 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17165 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17166 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17167 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17168 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17169 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17170 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17172 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17173 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17174 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17175 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17176 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17177 * doc/tm.texi: Regenerate.
17178 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17182 Alan Hayward <alan.hayward@arm.com>
17183 David Sherwood <david.sherwood@arm.com>
17185 * target.def (scalar_mode_supported_p): Take a scalar_mode
17186 instead of a machine_mode.
17187 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17188 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17189 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17190 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17191 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17192 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17193 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17194 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17195 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17196 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17197 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17198 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17199 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17200 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17202 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17203 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17204 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17205 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17206 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17208 * doc/tm.texi: Regenerate.
17210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17211 Alan Hayward <alan.hayward@arm.com>
17212 David Sherwood <david.sherwood@arm.com>
17214 * coretypes.h (opt_scalar_mode): New typedef.
17215 * gdbhooks.py (build_pretty_printers): Handle it.
17216 * machmode.h (mode_iterator::get_2xwider): Add overload for
17218 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17220 * expr.c (convert_mode_scalar): Likewise.
17221 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17222 * optabs.c (expand_float): Likewise.
17223 (expand_fix): Likewise.
17224 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17227 Alan Hayward <alan.hayward@arm.com>
17228 David Sherwood <david.sherwood@arm.com>
17230 * optabs.c (expand_float): Explicitly check for scalars before
17231 using a branching expansion.
17232 (expand_fix): Likewise.
17234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17235 Alan Hayward <alan.hayward@arm.com>
17236 David Sherwood <david.sherwood@arm.com>
17238 * expr.c (convert_mode): Split scalar handling out into...
17239 (convert_mode_scalar): ...this new function. Treat the modes
17242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17243 Alan Hayward <alan.hayward@arm.com>
17244 David Sherwood <david.sherwood@arm.com>
17246 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17248 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17251 Alan Hayward <alan.hayward@arm.com>
17252 David Sherwood <david.sherwood@arm.com>
17254 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17255 rather than a machine_mode.
17256 (fixed_from_string): Likewise.
17257 (fixed_convert): Likewise.
17258 (fixed_convert_from_int): Likewise.
17259 (fixed_convert_from_real): Likewise.
17260 (real_convert_from_fixed): Likewise.
17261 * fixed-value.c (fixed_from_double_int): Likewise.
17262 (fixed_from_string): Likewise.
17263 (fixed_convert): Likewise.
17264 (fixed_convert_from_int): Likewise.
17265 (fixed_convert_from_real): Likewise.
17266 (real_convert_from_fixed): Likewise.
17267 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17270 Alan Hayward <alan.hayward@arm.com>
17271 David Sherwood <david.sherwood@arm.com>
17273 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17274 of separate mode class checks. Do not allow vector modes here.
17275 (immed_wide_int_const): Use as_a <scalar_mode>.
17276 * explow.c (trunc_int_for_mode): Likewise.
17277 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17278 (wi::shwi): Likewise.
17279 (wi::min_value): Likewise.
17280 (wi::max_value): Likewise.
17281 * dwarf2out.c (loc_descriptor): Likewise.
17282 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17283 for CONST_WIDE_INT.
17285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17286 Alan Hayward <alan.hayward@arm.com>
17287 David Sherwood <david.sherwood@arm.com>
17289 * tree.h (SCALAR_TYPE_MODE): New macro.
17290 * expr.c (expand_expr_addr_expr_1): Use it.
17291 (expand_expr_real_2): Likewise.
17292 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17293 (fold_convert_const_fixed_from_int): Likewise.
17294 (fold_convert_const_fixed_from_real): Likewise.
17295 (native_encode_fixed): Likewise
17296 (native_encode_complex): Likewise
17297 (native_encode_vector): Likewise.
17298 (native_interpret_fixed): Likewise.
17299 (native_interpret_real): Likewise.
17300 (native_interpret_complex): Likewise.
17301 (native_interpret_vector): Likewise.
17302 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17303 (simd_clone_adjust_argument_types): Likewise.
17304 (simd_clone_init_simd_arrays): Likewise.
17305 (simd_clone_adjust): Likewise.
17306 * stor-layout.c (layout_type): Likewise.
17307 * tree.c (build_minus_one_cst): Likewise.
17308 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17309 * tree-inline.c (estimate_move_cost): Likewise.
17310 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17311 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17312 (vectorizable_reduction): Likewise.
17313 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17314 (vect_recog_mixed_size_cond_pattern): Likewise.
17315 (check_bool_pattern): Likewise.
17316 (adjust_bool_pattern): Likewise.
17317 (search_type_for_mask_1): Likewise.
17318 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17319 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17320 (vectorizable_load): Likewise.
17321 (vectorizable_store): Likewise.
17322 * ubsan.c (ubsan_encode_value): Likewise.
17323 * varasm.c (output_constant): Likewise.
17325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17326 Alan Hayward <alan.hayward@arm.com>
17327 David Sherwood <david.sherwood@arm.com>
17329 * coretypes.h (scalar_mode): New class.
17330 * machmode.h (scalar_mode): Likewise.
17331 (scalar_mode::includes_p): New function.
17332 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17333 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17334 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17335 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17336 * expmed.c (store_bit_field_1): Likewise.
17337 (extract_bit_field_1): Likewise.
17338 * expr.c (write_complex_part): Likewise.
17339 (read_complex_part): Likewise.
17340 (emit_move_complex_push): Likewise.
17341 (expand_expr_real_2): Likewise.
17342 * function.c (assign_parm_setup_reg): Likewise.
17343 (assign_parms_unsplit_complex): Likewise.
17344 * optabs.c (expand_binop): Likewise.
17345 * rtlanal.c (subreg_get_info): Likewise.
17346 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17347 * varasm.c (output_constant_pool_2): Likewise.
17349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17350 Alan Hayward <alan.hayward@arm.com>
17351 David Sherwood <david.sherwood@arm.com>
17353 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17355 (expmed_mult_highpart_optab): Likewise.
17356 (expmed_mult_highpart): Likewise.
17358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17359 Alan Hayward <alan.hayward@arm.com>
17360 David Sherwood <david.sherwood@arm.com>
17362 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17363 instead of a machine_mode.
17364 (builtin_memset_read_str): Likewise.
17365 * builtins.c (c_readstr): Likewise.
17366 (builtin_memcpy_read_str): Likewise.
17367 (builtin_strncpy_read_str): Likewise.
17368 (builtin_memset_read_str): Likewise.
17369 (builtin_memset_gen_str): Likewise.
17370 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17371 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17372 instead of a machine_mode.
17373 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17375 (make_extraction): Likewise.
17376 (try_widen_shift_mode): Take and return scalar_int_modes instead
17378 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17379 a scalar_int_mode instead of a machine_mode.
17380 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17381 (avr_addr_space_pointer_mode): Likewise.
17382 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17383 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17384 (msp430_unwind_word_mode): Likewise.
17385 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17386 (spu_addr_space_pointer_mode): Likewise.
17387 (spu_addr_space_address_mode): Likewise.
17388 (spu_libgcc_cmp_return_mode): Likewise.
17389 (spu_libgcc_shift_count_mode): Likewise.
17390 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17391 (rl78_addr_space_pointer_mode): Likewise.
17392 (fl78_unwind_word_mode): Likewise.
17393 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17395 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17396 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17397 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17398 (mips_valid_pointer_mode): Likewise.
17399 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17400 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17401 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17403 (ft32_addr_space_address_mode): Likewise.
17404 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17405 scalar_int_mode instead of a machine_mode.
17406 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17408 (m32c_addr_space_address_mode): Likewise.
17409 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17410 (rs6000_eh_return_filter_mode): Likewise.
17411 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17412 (rs6000_eh_return_filter_mode): Likewise.
17413 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17414 (s390_libgcc_shift_count_mode): Likewise.
17415 (s390_unwind_word_mode): Likewise.
17416 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17418 * target.def (mode_rep_extended): Likewise.
17419 (valid_pointer_mode): Likewise.
17420 (addr_space.valid_pointer_mode): Likewise.
17421 (eh_return_filter_mode): Return a scalar_int_mode rather than
17423 (libgcc_cmp_return_mode): Likewise.
17424 (libgcc_shift_count_mode): Likewise.
17425 (unwind_word_mode): Likewise.
17426 (addr_space.pointer_mode): Likewise.
17427 (addr_space.address_mode): Likewise.
17428 * doc/tm.texi: Regenerate.
17429 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17431 (do_jump): Use scalar_int_mode for local variables.
17432 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17433 rather than a machine_mode.
17434 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17435 (scompare_loc_descriptor_wide): Likewise.
17436 (scompare_loc_descriptor_narrow): Likewise.
17437 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17439 * except.c (sjlj_emit_dispatch_table): Likewise.
17440 (expand_builtin_eh_copy_values): Likewise.
17441 * explow.c (convert_memory_address_addr_space_1): Likewise.
17442 Take a scalar_int_mode rather than a machine_mode.
17443 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17444 than a machine_mode.
17445 (memory_address_addr_space): Use scalar_int_mode for local variables.
17446 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17447 rather than a machine_mode.
17448 * expmed.c (mask_rtx): Likewise.
17449 (init_expmed_one_conv): Likewise.
17450 (expand_mult_highpart_adjust): Likewise.
17451 (extract_high_half): Likewise.
17452 (expmed_mult_highpart_optab): Likewise.
17453 (expmed_mult_highpart): Likewise.
17454 (expand_smod_pow2): Likewise.
17455 (expand_sdiv_pow2): Likewise.
17456 (emit_store_flag_int): Likewise.
17457 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17459 (extract_low_bits): Likewise.
17460 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17462 * expr.c (pieces_addr::adjust): Likewise.
17463 (can_store_by_pieces): Likewise.
17464 (store_by_pieces): Likewise.
17465 (clear_by_pieces_1): Likewise.
17466 (expand_expr_addr_expr_1): Likewise.
17467 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17468 (expand_expr_real_1): Likewise.
17469 (try_casesi): Likewise.
17470 * final.c (shorten_branches): Likewise.
17471 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17472 type of "mode" to machine_mode.
17473 * internal-fn.c (expand_arith_overflow_result_store): Take a
17474 scalar_int_mode rather than a machine_mode.
17475 (expand_mul_overflow): Use scalar_int_mode for local variables.
17476 * loop-doloop.c (doloop_modify): Likewise.
17477 (doloop_optimize): Likewise.
17478 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17479 than a machine_mode.
17480 (expand_doubleword_shift_condmove): Likewise.
17481 (expand_doubleword_shift): Likewise.
17482 (expand_doubleword_clz): Likewise.
17483 (expand_doubleword_popcount): Likewise.
17484 (expand_doubleword_parity): Likewise.
17485 (expand_absneg_bit): Use scalar_int_mode for local variables.
17486 (prepare_float_lib_cmp): Likewise.
17487 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17488 rather than a machine_mode.
17489 (convert_memory_address_addr_space): Likewise.
17490 (get_mode_bounds): Likewise.
17491 (get_address_mode): Return a scalar_int_mode rather than a
17493 * rtlanal.c (get_address_mode): Likewise.
17494 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17495 than a machine_mode.
17496 * targhooks.c (default_mode_rep_extended): Likewise.
17497 (default_valid_pointer_mode): Likewise.
17498 (default_addr_space_valid_pointer_mode): Likewise.
17499 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17500 than a machine_mode.
17501 (default_libgcc_cmp_return_mode): Likewise.
17502 (default_libgcc_shift_count_mode): Likewise.
17503 (default_unwind_word_mode): Likewise.
17504 (default_addr_space_pointer_mode): Likewise.
17505 (default_addr_space_address_mode): Likewise.
17506 * targhooks.h (default_eh_return_filter_mode): Likewise.
17507 (default_libgcc_cmp_return_mode): Likewise.
17508 (default_libgcc_shift_count_mode): Likewise.
17509 (default_unwind_word_mode): Likewise.
17510 (default_addr_space_pointer_mode): Likewise.
17511 (default_addr_space_address_mode): Likewise.
17512 (default_mode_rep_extended): Take a scalar_int_mode rather than
17514 (default_valid_pointer_mode): Likewise.
17515 (default_addr_space_valid_pointer_mode): Likewise.
17516 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17518 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17519 rather than a machine_mode.
17520 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17521 for local variables.
17522 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17523 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17524 than a machine_mode.
17526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17527 Alan Hayward <alan.hayward@arm.com>
17528 David Sherwood <david.sherwood@arm.com>
17530 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17531 the mode argument to scalar_int_mode.
17532 (do_jump_by_parts_zero_rtx): Likewise.
17533 (do_jump_by_parts_equality_rtx): Likewise.
17534 (do_jump_by_parts_greater): Take a mode argument.
17535 (do_jump_by_parts_equality): Likewise.
17536 (do_jump_1): Update calls accordingly.
17538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17539 Alan Hayward <alan.hayward@arm.com>
17540 David Sherwood <david.sherwood@arm.com>
17542 * is-a.h (safe_dyn_cast): New function.
17543 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17544 (jump_table_for_label): Likewise.
17545 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17546 instead of an rtx_insn *.
17547 (shorten_branches): Use dyn_cast instead of LABEL_P and
17548 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17549 rtx_jump_table_data::get_data_mode.
17550 (final_scan_insn): Likewise.
17552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17553 Alan Hayward <alan.hayward@arm.com>
17554 David Sherwood <david.sherwood@arm.com>
17556 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17557 trying to combine a full-register integer set with a subreg
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 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17565 that is always either address_mode or pointer_mode.
17567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17568 Alan Hayward <alan.hayward@arm.com>
17569 David Sherwood <david.sherwood@arm.com>
17571 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17572 when the two are known to be equal.
17574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17575 Alan Hayward <alan.hayward@arm.com>
17576 David Sherwood <david.sherwood@arm.com>
17578 * simplify-rtx.c (simplify_const_unary_operation): Use
17579 is_a <scalar_int_mode> instead of checking for a nonzero
17580 precision. Forcibly convert op_mode to a scalar_int_mode
17581 in that case. More clearly differentiate the operand and
17582 result modes and use the former when deciding what the value
17583 of a count-bits operation should be. Use is_int_mode instead
17584 of checking for a MODE_INT. Remove redundant check for whether
17585 this mode has a zero precision.
17587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17588 Alan Hayward <alan.hayward@arm.com>
17589 David Sherwood <david.sherwood@arm.com>
17591 * optabs.c (widen_leading): Change the type of the mode argument
17592 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17593 (widen_bswap): Likewise.
17594 (expand_parity): Likewise.
17595 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17596 (expand_ffs): Likewise.
17597 (epand_unop): Check for scalar integer modes before calling the
17600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17601 Alan Hayward <alan.hayward@arm.com>
17602 David Sherwood <david.sherwood@arm.com>
17604 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17606 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17613 value_mode arguments. Use scalar_int_mode internally.
17614 (store_bit_field_1): Rename the new integer mode from imode
17615 to op0_mode and use it instead of GET_MODE (op0). Update calls
17616 to store_split_bit_field, store_bit_field_using_insv and
17617 store_fixed_bit_field.
17618 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17619 Use scalar_int_mode internally. Use a bit count rather than a mode
17620 when calculating the largest bit size for get_best_mode.
17621 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17622 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17623 Remove assertion that OP0 has a scalar integer mode.
17624 (store_split_bit_field): Add op0_mode and value_mode arguments.
17625 Update calls to extract_fixed_bit_field.
17626 (extract_bit_field_using_extv): Add an op0_mode argument.
17627 Use scalar_int_mode internally.
17628 (extract_bit_field_1): Rename the new integer mode from imode to
17629 op0_mode and use it instead of GET_MODE (op0). Update calls to
17630 extract_split_bit_field, extract_bit_field_using_extv and
17631 extract_fixed_bit_field.
17632 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17633 to extract_split_bit_field and extract_fixed_bit_field_1.
17634 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17635 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17636 on the target mode.
17637 (extract_split_bit_field): Add an op0_mode argument. Update call
17638 to extract_fixed_bit_field.
17640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17641 Alan Hayward <alan.hayward@arm.com>
17642 David Sherwood <david.sherwood@arm.com>
17644 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17645 * explow.c (hard_function_value): Likewise.
17646 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17647 convert_to_mode call outside the loop.
17648 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17649 for the mode iterator. Require the mode specified by max_pieces
17651 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17653 (copy_blkmode_to_reg): Likewise.
17654 (set_storage_via_setmem): Likewise.
17655 * optabs.c (prepare_cmp_insn): Likewise.
17656 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17657 * stor-layout.c (finish_bitfield_representative): Likewise.
17659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17660 Alan Hayward <alan.hayward@arm.com>
17661 David Sherwood <david.sherwood@arm.com>
17663 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17664 * expr.c (convert_move): Use them.
17665 (convert_modes): Likewise.
17666 (store_expr_with_bounds): Likewise.
17668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17669 Alan Hayward <alan.hayward@arm.com>
17670 David Sherwood <david.sherwood@arm.com>
17672 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17673 parameter for the mode of "x". Remove the "known_x", "known_mode"
17674 and "known_ret" arguments. Change the type of the mode argument
17675 to scalar_int_mode.
17676 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17677 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17678 (reg_num_sign_bit_copies_for_combine): Likewise.
17679 * rtlanal.c (nonzero_bits1): Likewise.
17680 (num_sign_bit_copies1): Likewise.
17681 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17682 (reg_num_sign_bit_copies_general): Likewise.
17683 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17684 (reg_nonzero_bits_general): Likewise.
17686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17687 Alan Hayward <alan.hayward@arm.com>
17688 David Sherwood <david.sherwood@arm.com>
17690 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17691 than in subroutines. Return 1 for non-integer modes.
17692 (cached_num_sign_bit_copies): Change the type of the mode parameter
17693 to scalar_int_mode.
17694 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17695 classes. Handle CONST_INT_P first and then check whether X also
17696 has a scalar integer mode. Check the same thing for inner registers
17697 of a SUBREG and for values that are being extended or truncated.
17699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17700 Alan Hayward <alan.hayward@arm.com>
17701 David Sherwood <david.sherwood@arm.com>
17703 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17704 in subroutines. Return the mode mask for non-integer modes.
17705 (cached_nonzero_bits): Change the type of the mode parameter
17706 to scalar_int_mode.
17707 (nonzero_bits1): Likewise. Remove early exit for other mode
17708 classes. Handle CONST_INT_P first and then check whether X
17709 also has a scalar integer mode.
17711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17712 Alan Hayward <alan.hayward@arm.com>
17713 David Sherwood <david.sherwood@arm.com>
17715 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17716 Return a scalar_int_mode and assert that the size is greater than
17718 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17719 (op_by_pieces_d::op_by_pieces_d): Likewise.
17720 (op_by_pieces_d::run): Likewise.
17721 (can_store_by_pieces): Likewise.
17723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17724 Alan Hayward <alan.hayward@arm.com>
17725 David Sherwood <david.sherwood@arm.com>
17727 * combine.c (extract_left_shift): Add a mode argument and update
17729 (make_compound_operation_int): Change the type of the mode parameter
17730 to scalar_int_mode and update the call to extract_left_shift.
17732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17733 Alan Hayward <alan.hayward@arm.com>
17734 David Sherwood <david.sherwood@arm.com>
17736 * combine.c (simplify_and_const_int): Change the type of the mode
17737 parameter to scalar_int_mode.
17738 (simplify_and_const_int_1): Likewise. Update recursive call.
17740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17741 Alan Hayward <alan.hayward@arm.com>
17742 David Sherwood <david.sherwood@arm.com>
17744 * combine.c (simplify_compare_const): Check that the mode is a
17745 scalar_int_mode (rather than VOIDmode) before testing its
17747 (simplify_comparison): Move COMPARISON_P handling out of the
17748 loop and restrict the latter part of the loop to scalar_int_modes.
17749 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17750 and when considering SUBREG_REGs. Use is_int_mode instead of
17751 checking GET_MODE_CLASS against MODE_INT.
17753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17754 Alan Hayward <alan.hayward@arm.com>
17755 David Sherwood <david.sherwood@arm.com>
17757 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17758 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17759 shift_unit_mode and for modes involved in scalar shifts.
17761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17762 Alan Hayward <alan.hayward@arm.com>
17763 David Sherwood <david.sherwood@arm.com>
17765 * combine.c (force_int_to_mode): New function, split out from...
17766 (force_to_mode): ...here. Keep xmode up-to-date and use it
17767 instead of GET_MODE (x).
17769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17770 Alan Hayward <alan.hayward@arm.com>
17771 David Sherwood <david.sherwood@arm.com>
17773 * optabs-query.h (extraction_insn::struct_mode): Change type to
17774 opt_scalar_int_mode and update comment.
17775 (extraction_insn::field_mode): Change type to scalar_int_mode.
17776 (extraction_insn::pos_mode): Likewise.
17777 * combine.c (make_extraction): Update accordingly.
17778 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17779 (get_optab_extraction_insn): Likewise.
17780 * recog.c (simplify_while_replacing): Likewise.
17781 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17782 parameter to opt_scalar_int_mode.
17784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17785 Alan Hayward <alan.hayward@arm.com>
17786 David Sherwood <david.sherwood@arm.com>
17788 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17789 to a scalar_int_mode instead of a machine_mode.
17790 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17791 (get_best_mode): Return a boolean and use a pointer argument to store
17792 the selected mode. Replace the limit mode parameter with a bit limit.
17793 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17794 for the values returned by bit_field_mode_iterator::next_mode.
17795 (store_bit_field): Update call to get_best_mode.
17796 (store_fixed_bit_field): Likewise.
17797 (extract_fixed_bit_field): Likewise.
17798 * expr.c (optimize_bitfield_assignment_op): Likewise.
17799 * fold-const.c (optimize_bit_field_compare): Likewise.
17800 (fold_truth_andor_1): Likewise.
17801 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17802 Update for new type of m_mode.
17803 (get_best_mode): As above.
17805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17806 Alan Hayward <alan.hayward@arm.com>
17807 David Sherwood <david.sherwood@arm.com>
17809 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17810 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17811 (store_bit_field): Check is_a <scalar_int_mode> before calling
17812 strict_volatile_bitfield_p.
17813 (extract_bit_field): Likewise.
17815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17816 Alan Hayward <alan.hayward@arm.com>
17817 David Sherwood <david.sherwood@arm.com>
17819 * target.def (cstore_mode): Return a scalar_int_mode.
17820 * doc/tm.texi: Regenerate.
17821 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17822 * targhooks.h (default_cstore_mode): Likewise.
17823 * targhooks.c (default_cstore_mode): Likewise, using a forced
17825 * expmed.c (emit_cstore): Expect the target of the cstore to be
17828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17829 Alan Hayward <alan.hayward@arm.com>
17830 David Sherwood <david.sherwood@arm.com>
17832 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17834 (niter_desc): Likewise mode.
17835 (iv_analyze): Add a mode parameter.
17837 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17838 and change its type to scalar_int_mode.
17839 * loop-iv.c: Update commentary at head of file.
17840 (iv_constant): Pass the mode paraeter before the rtx it describes
17841 and change its type to scalar_int_mode. Remove VOIDmode handling.
17842 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17843 (iv_extend): Likewise.
17844 (shorten_into_mode): Likewise.
17845 (iv_add): Use scalar_int_mode.
17846 (iv_mult): Likewise.
17847 (iv_shift): Likewise.
17848 (canonicalize_iv_subregs): Likewise.
17849 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17850 it describes and change its mode to scalar_int_mode. Also change
17851 the type of the returned inner_mode to scalar_int_mode.
17852 (get_biv_step): Likewise, turning outer_mode from a pointer
17853 into a direct parameter. Update call to get_biv_step_1.
17854 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17855 iv_constant and get_biv_step.
17856 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17857 and change its type to scalar_int_mode. Don't initialise iv->mode
17858 to VOIDmode and remove later checks for its still being VOIDmode.
17859 Update calls to iv_analyze_op and iv_analyze_expr. Check
17860 is_a <scalar_int_mode> when changing the mode under consideration.
17861 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17862 Update call to iv_analyze_expr.
17863 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17864 inner register is not also a scalar_int_mode. Update call to
17866 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17867 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17868 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17869 separate mode class checks. Update calls to iv_analyze. Remove
17870 fix-up of VOIDmodes after iv_analyze_biv.
17871 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17872 don't have a scalar_int_mode. Update call to biv_p.
17874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17875 Alan Hayward <alan.hayward@arm.com>
17876 David Sherwood <david.sherwood@arm.com>
17878 * cfgexpand.c (convert_debug_memory_address): Use
17879 as_a <scalar_int_mode>.
17880 * combine.c (expand_compound_operation): Likewise.
17881 (make_extraction): Likewise.
17882 (change_zero_ext): Likewise.
17883 (simplify_comparison): Likewise.
17884 * cse.c (cse_insn): Likewise.
17885 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17886 (mem_loc_descriptor): Likewise.
17887 (loc_descriptor): Likewise.
17888 * expmed.c (init_expmed_one_mode): Likewise.
17889 (synth_mult): Likewise.
17890 (emit_store_flag_1): Likewise.
17891 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17892 of a comparison with size.
17893 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17894 (reduce_to_bit_field_precision): Likewise.
17895 * function.c (expand_function_end): Likewise.
17896 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17897 * loop-doloop.c (doloop_modify): Likewise.
17898 * optabs.c (expand_binop): Likewise.
17899 (expand_unop): Likewise.
17900 (expand_copysign_absneg): Likewise.
17901 (prepare_cmp_insn): Likewise.
17902 (maybe_legitimize_operand): Likewise.
17903 * recog.c (const_scalar_int_operand): Likewise.
17904 * rtlanal.c (get_address_mode): Likewise.
17905 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17906 (simplify_cond_clz_ctz): Likewise.
17907 * tree-nested.c (get_nl_goto_field): Likewise.
17908 * tree.c (build_vector_type_for_mode): Likewise.
17909 * var-tracking.c (use_narrower_mode): Likewise.
17911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17912 Alan Hayward <alan.hayward@arm.com>
17913 David Sherwood <david.sherwood@arm.com>
17915 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17916 * builtins.c (expand_builtin_signbit): Use it.
17917 * cfgexpand.c (expand_debug_expr): Likewise.
17918 * dojump.c (do_jump): Likewise.
17919 (do_compare_and_jump): Likewise.
17920 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17921 * expmed.c (make_tree): Likewise.
17922 * expr.c (expand_expr_real_2): Likewise.
17923 (expand_expr_real_1): Likewise.
17924 (try_casesi): Likewise.
17925 * fold-const-call.c (fold_const_call_ss): Likewise.
17926 * fold-const.c (unextend): Likewise.
17927 (extract_muldiv_1): Likewise.
17928 (fold_single_bit_test): Likewise.
17929 (native_encode_int): Likewise.
17930 (native_encode_string): Likewise.
17931 (native_interpret_int): Likewise.
17932 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17933 * internal-fn.c (expand_addsub_overflow): Likewise.
17934 (expand_neg_overflow): Likewise.
17935 (expand_mul_overflow): Likewise.
17936 (expand_arith_overflow): Likewise.
17937 * match.pd: Likewise.
17938 * stor-layout.c (layout_type): Likewise.
17939 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17940 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17941 * tree-ssanames.c (get_range_info): Likewise.
17942 * tree-switch-conversion.c (array_value_type) Likewise.
17943 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17944 (vect_recog_divmod_pattern): Likewise.
17945 (vect_recog_mixed_size_cond_pattern): Likewise.
17946 * tree-vrp.c (extract_range_basic): Likewise.
17947 (simplify_float_conversion_using_ranges): Likewise.
17948 * tree.c (int_fits_type_p): Likewise.
17949 * ubsan.c (instrument_bool_enum_load): Likewise.
17950 * varasm.c (mergeable_string_section): Likewise.
17951 (narrowing_initializer_constant_valid_p): Likewise.
17952 (output_constant): Likewise.
17954 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17955 Alan Hayward <alan.hayward@arm.com>
17956 David Sherwood <david.sherwood@arm.com>
17958 * machmode.h (NARROWEST_INT_MODE): New macro.
17959 * expr.c (alignment_for_piecewise_move): Use it instead of
17960 GET_CLASS_NARROWEST_MODE (MODE_INT).
17961 (push_block): Likewise.
17962 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17964 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17967 Alan Hayward <alan.hayward@arm.com>
17968 David Sherwood <david.sherwood@arm.com>
17970 * postreload.c (move2add_valid_value_p): Change the type of the
17971 mode parameter to scalar_int_mode.
17972 (move2add_use_add2_insn): Add a mode parameter and use it instead
17974 (move2add_use_add3_insn): Likewise.
17975 (reload_cse_move2add): Update accordingly.
17977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17978 Alan Hayward <alan.hayward@arm.com>
17979 David Sherwood <david.sherwood@arm.com>
17981 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17983 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17984 * optabs.c (expand_unop): Likewise.
17986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17987 Alan Hayward <alan.hayward@arm.com>
17988 David Sherwood <david.sherwood@arm.com>
17990 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17991 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17992 (popcount_loc_descriptor): Likewise.
17993 (bswap_loc_descriptor): Likewise.
17994 (rotate_loc_descriptor): Likewise.
17995 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17996 calling the functions above.
17998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17999 Alan Hayward <alan.hayward@arm.com>
18000 David Sherwood <david.sherwood@arm.com>
18002 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18004 (try_combine): Likewise.
18005 (simplify_if_then_else): Likewise.
18006 * cse.c (cse_insn): Likewise.
18007 * dwarf2out.c (mem_loc_descriptor): Likewise.
18008 * emit-rtl.c (gen_lowpart_common): Likewise.
18009 * simplify-rtx.c (simplify_truncation): Likewise.
18010 (simplify_binary_operation_1): Likewise.
18011 (simplify_const_relational_operation): Likewise.
18012 (simplify_ternary_operation): Likewise.
18013 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18016 Alan Hayward <alan.hayward@arm.com>
18017 David Sherwood <david.sherwood@arm.com>
18019 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18020 * reload.c (push_reload): Likewise.
18021 (find_reloads): Likewise.
18023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18024 Alan Hayward <alan.hayward@arm.com>
18025 David Sherwood <david.sherwood@arm.com>
18027 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18028 (make_compound_operation_int): Likewise.
18029 (change_zero_ext): Likewise.
18030 * expr.c (convert_move): Likewise.
18031 (convert_modes): Likewise.
18032 * fwprop.c (forward_propagate_subreg): Likewise.
18033 * loop-iv.c (get_biv_step_1): Likewise.
18034 * optabs.c (widen_operand): Likewise.
18035 * postreload.c (move2add_valid_value_p): Likewise.
18036 * recog.c (simplify_while_replacing): Likewise.
18037 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18038 (simplify_binary_operation_1): Likewise. Remove redundant
18039 mode equality check.
18041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18042 Alan Hayward <alan.hayward@arm.com>
18043 David Sherwood <david.sherwood@arm.com>
18045 * combine.c (combine_simplify_rtx): Add checks for
18046 is_a <scalar_int_mode>.
18047 (simplify_if_then_else): Likewise.
18048 (make_field_assignment): Likewise.
18049 (simplify_comparison): Likewise.
18050 * ifcvt.c (noce_try_bitop): Likewise.
18051 * loop-invariant.c (canonicalize_address_mult): Likewise.
18052 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18055 Alan Hayward <alan.hayward@arm.com>
18056 David Sherwood <david.sherwood@arm.com>
18058 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18059 is_a <scalar_int_mode> instead of != BLKmode.
18061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18062 Alan Hayward <alan.hayward@arm.com>
18063 David Sherwood <david.sherwood@arm.com>
18065 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18066 instead of != VOIDmode.
18067 * combine.c (if_then_else_cond): Likewise.
18068 (change_zero_ext): Likewise.
18069 * dwarf2out.c (mem_loc_descriptor): Likewise.
18070 (loc_descriptor): Likewise.
18071 * rtlanal.c (canonicalize_condition): Likewise.
18072 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18075 Alan Hayward <alan.hayward@arm.com>
18076 David Sherwood <david.sherwood@arm.com>
18078 * simplify-rtx.c (simplify_binary_operation_1): Use
18079 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18082 Alan Hayward <alan.hayward@arm.com>
18083 David Sherwood <david.sherwood@arm.com>
18085 * wide-int.h (int_traits<unsigned char>) New class.
18086 (int_traits<unsigned short>) Likewise.
18087 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18088 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18090 * combine.c (set_nonzero_bits_and_sign_copies): Use
18091 is_a <scalar_int_mode>.
18092 (find_split_point): Likewise.
18093 (combine_simplify_rtx): Likewise.
18094 (simplify_logical): Likewise.
18095 (expand_compound_operation): Likewise.
18096 (expand_field_assignment): Likewise.
18097 (make_compound_operation): Likewise.
18098 (extended_count): Likewise.
18099 (change_zero_ext): Likewise.
18100 (simplify_comparison): Likewise.
18101 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18102 (ucompare_loc_descriptor): Likewise.
18103 (minmax_loc_descriptor): Likewise.
18104 (mem_loc_descriptor): Likewise.
18105 (loc_descriptor): Likewise.
18106 * expmed.c (init_expmed_one_mode): Likewise.
18107 * lra-constraints.c (lra_constraint_offset): Likewise.
18108 * optabs.c (prepare_libcall_arg): Likewise.
18109 * postreload.c (move2add_note_store): Likewise.
18110 * reload.c (operands_match_p): Likewise.
18111 * rtl.h (load_extend_op): Likewise.
18112 * rtlhooks.c (gen_lowpart_general): Likewise.
18113 * simplify-rtx.c (simplify_truncation): Likewise.
18114 (simplify_unary_operation_1): Likewise.
18115 (simplify_binary_operation_1): Likewise.
18116 (simplify_const_binary_operation): Likewise.
18117 (simplify_const_relational_operation): Likewise.
18118 (simplify_subreg): Likewise.
18119 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18120 * var-tracking.c (adjust_mems): Likewise.
18121 (prepare_call_arguments): Likewise.
18123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18124 Alan Hayward <alan.hayward@arm.com>
18125 David Sherwood <david.sherwood@arm.com>
18127 * machmode.h (is_int_mode): New fuction.
18128 * combine.c (find_split_point): Use it.
18129 (combine_simplify_rtx): Likewise.
18130 (simplify_if_then_else): Likewise.
18131 (simplify_set): Likewise.
18132 (simplify_shift_const_1): Likewise.
18133 (simplify_comparison): Likewise.
18134 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18135 * cse.c (notreg_cost): Likewise.
18136 (cse_insn): Likewise.
18137 * cselib.c (cselib_lookup_1): Likewise.
18138 * dojump.c (do_jump_1): Likewise.
18139 (do_compare_rtx_and_jump): Likewise.
18140 * dse.c (get_call_args): Likewise.
18141 * dwarf2out.c (rtl_for_decl_init): Likewise.
18142 (native_encode_initializer): Likewise.
18143 * expmed.c (emit_store_flag_1): Likewise.
18144 (emit_store_flag): Likewise.
18145 * expr.c (convert_modes): Likewise.
18146 (store_field): Likewise.
18147 (expand_expr_real_1): Likewise.
18148 * fold-const.c (fold_read_from_constant_string): Likewise.
18149 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18150 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18151 * optabs.c (expand_binop): Likewise.
18152 (expand_unop): Likewise.
18153 (expand_abs_nojump): Likewise.
18154 (expand_one_cmpl_abs_nojump): Likewise.
18155 * simplify-rtx.c (mode_signbit_p): Likewise.
18156 (val_signbit_p): Likewise.
18157 (val_signbit_known_set_p): Likewise.
18158 (val_signbit_known_clear_p): Likewise.
18159 (simplify_relational_operation_1): Likewise.
18160 * tree.c (vector_type_mode): Likewise.
18162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18163 Alan Hayward <alan.hayward@arm.com>
18164 David Sherwood <david.sherwood@arm.com>
18166 * machmode.h (smallest_mode_for_size): Fix formatting.
18167 (smallest_int_mode_for_size): New function.
18168 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18169 instead of smallest_mode_for_size.
18170 * combine.c (make_extraction): Likewise.
18171 * config/arc/arc.c (arc_expand_movmem): Likewise.
18172 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18173 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18174 * config/s390/s390.c (s390_expand_insv): Likewise.
18175 * config/sparc/sparc.c (assign_int_registers): Likewise.
18176 * config/spu/spu.c (spu_function_value): Likewise.
18177 (spu_function_arg): Likewise.
18178 * coverage.c (get_gcov_type): Likewise.
18179 (get_gcov_unsigned_t): Likewise.
18180 * dse.c (find_shift_sequence): Likewise.
18181 * expmed.c (store_bit_field_1): Likewise.
18182 * expr.c (convert_move): Likewise.
18183 (store_field): Likewise.
18184 * internal-fn.c (expand_arith_overflow): Likewise.
18185 * optabs-query.c (get_best_extraction_insn): Likewise.
18186 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18187 * stor-layout.c (layout_type): Likewise.
18188 (initialize_sizetypes): Likewise.
18189 * targhooks.c (default_get_mask_mode): Likewise.
18190 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18193 Alan Hayward <alan.hayward@arm.com>
18194 David Sherwood <david.sherwood@arm.com>
18196 * machmode.h (opt_mode::else_blk): New function.
18197 (int_mode_for_mode): Declare.
18198 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18199 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18201 * cfgexpand.c (expand_debug_expr): Likewise.
18202 * combine.c (gen_lowpart_or_truncate): Likewise.
18203 (gen_lowpart_for_combine): Likewise.
18204 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18205 * config/avr/avr.c (avr_to_int_mode): Likewise.
18206 (avr_out_plus_1): Likewise.
18207 (avr_out_plus): Likewise.
18208 (avr_out_round): Likewise.
18209 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18210 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18211 (s390_expand_vcond): Likewise.
18212 * config/spu/spu.c (spu_split_immediate): Likewise.
18213 (spu_expand_mov): Likewise.
18214 * dse.c (get_stored_val): Likewise.
18215 * expmed.c (store_bit_field_1): Likewise.
18216 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18218 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18219 (extract_low_bits): Likewise.
18220 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18221 handling rather than repeating the check.
18222 (emit_group_store): Likewise.
18223 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18224 * optabs.c (expand_absneg_bit): Likewise.
18225 (expand_copysign_absneg): Likewise.
18226 (expand_copysign_bit): Likewise.
18227 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18228 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18229 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18230 * var-tracking.c (prepare_call_arguments): Likewise.
18231 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18232 int_mode_for_mode instead of mode_for_size.
18233 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18236 Alan Hayward <alan.hayward@arm.com>
18237 David Sherwood <david.sherwood@arm.com>
18239 * machmode.h (int_mode_for_size): New function.
18240 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18241 instead of mode_for_size.
18242 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18244 * combine.c (expand_field_assignment): Use int_mode_for_size
18245 instead of mode_for_size.
18246 (make_extraction): Likewise.
18247 (simplify_shift_const_1): Likewise.
18248 (simplify_comparison): Likewise.
18249 * dojump.c (do_jump): Likewise.
18250 * dwarf2out.c (mem_loc_descriptor): Likewise.
18251 * emit-rtl.c (init_derived_machine_modes): Likewise.
18252 * expmed.c (flip_storage_order): Likewise.
18253 (convert_extracted_bit_field): Likewise.
18254 * expr.c (copy_blkmode_from_reg): Likewise.
18255 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18256 * internal-fn.c (expand_mul_overflow): Likewise.
18257 * lower-subreg.c (simple_move): Likewise.
18258 * optabs-libfuncs.c (init_optabs): Likewise.
18259 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18260 * tree.c (vector_type_mode): Likewise.
18261 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18262 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18263 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18264 * tree-vect-stmts.c (vectorizable_load): Likewise.
18265 (vectorizable_store): Likewise.
18267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18268 Alan Hayward <alan.hayward@arm.com>
18269 David Sherwood <david.sherwood@arm.com>
18271 * coretypes.h (pod_mode): New type.
18272 (scalar_int_mode_pod): New typedef.
18273 * machmode.h (pod_mode): New class.
18274 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18275 * genmodes.c (emit_mode_int_n): Update accordingly.
18276 * lower-subreg.h (target_lower_subreg): Change type to
18277 scalar_int_mode_pod.
18278 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18279 scalar_int_mode_pod.
18281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18282 Alan Hayward <alan.hayward@arm.com>
18283 David Sherwood <david.sherwood@arm.com>
18285 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18286 machine_mode to scalar_int_mode.
18287 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18288 (rs6000_option_override_internal): Remove cast to int.
18289 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18290 machine_mode to scalar_int_mode.
18291 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18292 (rs6000_option_override_internal): Remove cast to int.
18293 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18294 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18296 * config/s390/s390.c (s390_expand_builtin): Likewise.
18297 * coretypes.h (scalar_int_mode): New type.
18298 (opt_scalar_int_mode): New typedef.
18299 * machmode.h (scalar_int_mode): New class.
18300 (scalar_int_mode::includes_p): New function.
18301 (byte_mode): Change type to scalar_int_mode.
18302 (word_mode): Likewise.
18303 (ptr_mode): Likewise.
18304 * emit-rtl.c (byte_mode): Likewise.
18305 (word_mode): Likewise.
18306 (ptr_mode): Likewise.
18307 (init_derived_machine_modes): Update accordingly.
18308 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18309 and MODE_PARTIAL_INT.
18310 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18311 opt_scalar_int_mode.
18313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18314 Alan Hayward <alan.hayward@arm.com>
18315 David Sherwood <david.sherwood@arm.com>
18317 * target.def (libgcc_floating_mode_supported_p): Take a
18319 * doc/tm.texi: Regenerate.
18320 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18322 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18323 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18327 Alan Hayward <alan.hayward@arm.com>
18328 David Sherwood <david.sherwood@arm.com>
18330 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18331 * doc/tm.texi: Regenerate.
18332 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18333 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18334 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18335 * targhooks.h (default_floatn_mode): Likewise.
18336 * targhooks.c (default_floatn_mode): Likewise.
18337 * tree.c (build_common_tree_nodes): Update accordingly.
18339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18340 Alan Hayward <alan.hayward@arm.com>
18341 David Sherwood <david.sherwood@arm.com>
18343 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18344 (mode_iterator::iterate_p): Likewise.
18345 (mode_iterator::get_wider): Likewise.
18346 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18349 Alan Hayward <alan.hayward@arm.com>
18350 David Sherwood <david.sherwood@arm.com>
18352 * coretypes.h (opt_scalar_float_mode): New typedef.
18353 * machmode.h (float_mode_for_size): New function.
18354 * emit-rtl.c (double_mode): Delete.
18355 (init_emit_once): Use float_mode_for_size.
18356 * stor-layout.c (layout_type): Likewise.
18357 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18360 Alan Hayward <alan.hayward@arm.com>
18361 David Sherwood <david.sherwood@arm.com>
18363 * output.h (assemble_real): Take a scalar_float_mode.
18364 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18365 * config/arm/arm.md (consttable_4): Likewise.
18366 (consttable_8): Likewise.
18367 (consttable_16): Likewise.
18368 * config/mips/mips.md (consttable_float): Likewise.
18369 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18370 * varasm.c (assemble_real): Take a scalar_float_mode.
18371 (output_constant_pool_2): Update accordingly.
18372 (output_constant): Likewise.
18374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18375 Alan Hayward <alan.hayward@arm.com>
18376 David Sherwood <david.sherwood@arm.com>
18378 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18379 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18380 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18381 (native_encode_real): Likewise.
18382 (native_interpret_real): Likewise.
18383 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18384 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18387 Alan Hayward <alan.hayward@arm.com>
18388 David Sherwood <david.sherwood@arm.com>
18390 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18391 <scalar_float_mode>. Simplify.
18392 (gen_extend_conv_libfunc): Likewise.
18394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18395 Alan Hayward <alan.hayward@arm.com>
18396 David Sherwood <david.sherwood@arm.com>
18398 * coretypes.h (scalar_float_mode): New type.
18399 * machmode.h (mode_traits::from_int): Use machine_mode if
18400 USE_ENUM_MODES is defined.
18401 (is_a): New function.
18403 (dyn_cast): Likewise.
18404 (scalar_float_mode): New class.
18405 (scalar_float_mode::includes_p): New function.
18406 (is_float_mode): Likewise.
18407 * gdbhooks.py (MachineModePrinter): New class.
18408 (build_pretty_printer): Use it for scalar_float_mode.
18409 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18410 (format_helper::format_helper): Turn into a template.
18411 * genmodes.c (get_mode_class): New function.
18412 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18413 or machine_mode if none.
18414 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18415 as_a <scalar_float_mode>.
18416 * dwarf2out.c (mem_loc_descriptor): Likewise.
18417 (insert_float): Likewise.
18418 (add_const_value_attribute): Likewise.
18419 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18420 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18421 (expand_unop): Update accordingly.
18422 (expand_abs_nojump): Likewise.
18423 (expand_copysign_absneg): Take a scalar_float_mode.
18424 (expand_copysign_bit): Likewise.
18425 (expand_copysign): Update accordingly.
18427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18428 Alan Hayward <alan.hayward@arm.com>
18429 David Sherwood <david.sherwood@arm.com>
18431 * coretypes.h (opt_mode): New class.
18432 * machmode.h (opt_mode): Likewise.
18433 (opt_mode::else_void): New function.
18434 (opt_mode::require): Likewise.
18435 (opt_mode::exists): Likewise.
18436 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18437 (GET_MODE_2XWIDER_MODE): Likewise.
18438 (mode_iterator::get_wider): Update accordingly.
18439 (mode_iterator::get_2xwider): Likewise.
18440 (mode_iterator::get_known_wider): Likewise, turning into a template.
18441 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18442 forcing a wider mode to exist.
18443 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18444 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18445 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18446 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18447 * lower-subreg.c (init_lower_subreg): Likewise.
18448 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18449 on the final iteration.
18450 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18451 a wider mode exists before asking for a move pattern.
18452 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18453 forcing a wider mode to exist.
18454 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18455 returning false if no such mode exists.
18456 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18457 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18458 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18459 Avoid checking for a MODE_INT if we already know the mode is not a
18461 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18462 forcing a wider mode to exist.
18463 (expmed_mult_highpart_optab): Likewise.
18464 (expmed_mult_highpart): Likewise.
18465 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18467 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18468 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18469 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18470 * internal-fn.c (expand_mul_overflow): Update use of
18471 GET_MODE_2XWIDER_MODE.
18472 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18473 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18474 GET_MODE_WIDER_MODE.
18475 (convert_plusminus_to_widen): Likewise.
18476 * tree-switch-conversion.c (array_value_type): Likewise.
18477 * var-tracking.c (emit_note_insn_var_location): Likewise.
18478 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18479 Return false inside rather than outside the loop if no wider mode
18481 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18482 and GET_MODE_2XWIDER_MODE
18483 (can_compare_p): Use else_void.
18484 * gdbhooks.py (OptMachineModePrinter): New class.
18485 (build_pretty_printer): Use it for opt_mode.
18487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18488 Alan Hayward <alan.hayward@arm.com>
18489 David Sherwood <david.sherwood@arm.com>
18491 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18492 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18495 Alan Hayward <alan.hayward@arm.com>
18496 David Sherwood <david.sherwood@arm.com>
18498 * machmode.h (mode_traits): New structure.
18499 (get_narrowest_mode): New function.
18500 (mode_iterator::start): Likewise.
18501 (mode_iterator::iterate_p): Likewise.
18502 (mode_iterator::get_wider): Likewise.
18503 (mode_iterator::get_known_wider): Likewise.
18504 (mode_iterator::get_2xwider): Likewise.
18505 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18506 (FOR_EACH_MODE): Likewise.
18507 (FOR_EACH_MODE_FROM): Likewise.
18508 (FOR_EACH_MODE_UNTIL): Likewise.
18509 (FOR_EACH_WIDER_MODE): Likewise.
18510 (FOR_EACH_2XWIDER_MODE): Likewise.
18511 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18512 * combine.c (simplify_comparison): Likewise
18513 * config/i386/i386.c (type_natural_mode): Likewise.
18514 * cse.c (cse_insn): Likewise.
18515 * dse.c (find_shift_sequence): Likewise.
18516 * emit-rtl.c (init_derived_machine_modes): Likewise.
18517 (init_emit_once): Likewise.
18518 * explow.c (hard_function_value): Likewise.
18519 * expmed.c (extract_fixed_bit_field_1): Likewise.
18520 (extract_bit_field_1): Likewise.
18521 (expand_divmod): Likewise.
18522 (emit_store_flag_1): Likewise.
18523 * expr.c (init_expr_target): Likewise.
18524 (convert_move): Likewise.
18525 (alignment_for_piecewise_move): Likewise.
18526 (widest_int_mode_for_size): Likewise.
18527 (emit_block_move_via_movmem): Likewise.
18528 (copy_blkmode_to_reg): Likewise.
18529 (set_storage_via_setmem): Likewise.
18530 (compress_float_constant): Likewise.
18531 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18532 * optabs-query.c (get_best_extraction_insn): Likewise.
18533 * optabs.c (expand_binop): Likewise.
18534 (expand_twoval_unop): Likewise.
18535 (expand_twoval_binop): Likewise.
18536 (widen_leading): Likewise.
18537 (widen_bswap): Likewise.
18538 (expand_parity): Likewise.
18539 (expand_unop): Likewise.
18540 (prepare_cmp_insn): Likewise.
18541 (prepare_float_lib_cmp): Likewise.
18542 (expand_float): Likewise.
18543 (expand_fix): Likewise.
18544 (expand_sfix_optab): Likewise.
18545 * postreload.c (move2add_use_add2_insn): Likewise.
18546 * reg-stack.c (reg_to_stack): Likewise.
18547 * reginfo.c (choose_hard_reg_mode): Likewise.
18548 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18549 * stor-layout.c (mode_for_size): Likewise.
18550 (smallest_mode_for_size): Likewise.
18551 (mode_for_vector): Likewise.
18552 (finish_bitfield_representative): Likewise.
18553 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18554 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18555 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18556 * var-tracking.c (prepare_call_arguments): Likewise.
18558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18559 Alan Hayward <alan.hayward@arm.com>
18560 David Sherwood <david.sherwood@arm.com>
18562 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18563 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18564 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18565 * machmode.h (mode_size): Move earlier in file.
18566 (mode_precision): Likewise.
18567 (mode_inner): Likewise.
18568 (mode_nunits): Likewise.
18569 (mode_unit_size): Likewise.
18570 (unit_unit_precision): Likewise.
18571 (mode_wider): Likewise.
18572 (mode_2xwider): Likewise.
18573 (machine_mode): New class.
18574 (mode_to_bytes): New function.
18575 (mode_to_bits): Likewise.
18576 (mode_to_precision): Likewise.
18577 (mode_to_inner): Likewise.
18578 (mode_to_unit_size): Likewise.
18579 (mode_to_unit_precision): Likewise.
18580 (mode_to_nunits): Likewise.
18581 (GET_MODE_SIZE): Use mode_to_bytes.
18582 (GET_MODE_BITSIZE): Use mode_to_bits.
18583 (GET_MODE_PRECISION): Use mode_to_precision.
18584 (GET_MODE_INNER): Use mode_to_inner.
18585 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18586 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18587 (GET_MODE_NUNITS): Use mode_to_nunits.
18588 * system.h (ALWAYS_INLINE): New macro.
18589 * config/powerpcspe/powerpcspe-c.c
18590 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18591 int for arg1_mode and arg2_mode.
18593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18594 Alan Hayward <alan.hayward@arm.com>
18595 David Sherwood <david.sherwood@arm.com>
18597 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18598 Prefix mode names with E_ in case statements.
18599 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18600 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18601 (aarch64_split_simd_move): Likewise.
18602 (aarch64_gen_storewb_pair): Likewise.
18603 (aarch64_gen_loadwb_pair): Likewise.
18604 (aarch64_gen_store_pair): Likewise.
18605 (aarch64_gen_load_pair): Likewise.
18606 (aarch64_get_condition_code_1): Likewise.
18607 (aarch64_constant_pool_reload_icode): Likewise.
18608 (get_rsqrte_type): Likewise.
18609 (get_rsqrts_type): Likewise.
18610 (get_recpe_type): Likewise.
18611 (get_recps_type): Likewise.
18612 (aarch64_gimplify_va_arg_expr): Likewise.
18613 (aarch64_simd_container_mode): Likewise.
18614 (aarch64_emit_load_exclusive): Likewise.
18615 (aarch64_emit_store_exclusive): Likewise.
18616 (aarch64_expand_compare_and_swap): Likewise.
18617 (aarch64_gen_atomic_cas): Likewise.
18618 (aarch64_emit_bic): Likewise.
18619 (aarch64_emit_atomic_swap): Likewise.
18620 (aarch64_emit_atomic_load_op): Likewise.
18621 (aarch64_evpc_trn): Likewise.
18622 (aarch64_evpc_uzp): Likewise.
18623 (aarch64_evpc_zip): Likewise.
18624 (aarch64_evpc_ext): Likewise.
18625 (aarch64_evpc_rev): Likewise.
18626 (aarch64_evpc_dup): Likewise.
18627 (aarch64_gen_ccmp_first): Likewise.
18628 (aarch64_gen_ccmp_next): Likewise.
18629 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18630 (alpha_emit_xfloating_libcall): Likewise.
18631 (emit_insxl): Likewise.
18632 (alpha_arg_type): Likewise.
18633 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18634 (arc_preferred_simd_mode): Likewise.
18635 (arc_secondary_reload): Likewise.
18636 (get_arc_condition_code): Likewise.
18637 (arc_print_operand): Likewise.
18638 (arc_legitimate_constant_p): Likewise.
18639 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18640 * config/arc/arc.md (casesi_load): Likewise.
18641 (casesi_compact_jump): Likewise.
18642 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18643 (cc_use_register): Likewise.
18644 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18645 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18646 (arm_init_iwmmxt_builtins): Likewise.
18647 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18648 (neon_expand_vector_init): Likewise.
18649 (arm_attr_length_move_neon): Likewise.
18650 (maybe_get_arm_condition_code): Likewise.
18651 (arm_emit_vector_const): Likewise.
18652 (arm_preferred_simd_mode): Likewise.
18653 (arm_output_iwmmxt_tinsr): Likewise.
18654 (thumb1_output_casesi): Likewise.
18655 (thumb2_output_casesi): Likewise.
18656 (arm_emit_load_exclusive): Likewise.
18657 (arm_emit_store_exclusive): Likewise.
18658 (arm_expand_compare_and_swap): Likewise.
18659 (arm_evpc_neon_vuzp): Likewise.
18660 (arm_evpc_neon_vzip): Likewise.
18661 (arm_evpc_neon_vrev): Likewise.
18662 (arm_evpc_neon_vtrn): Likewise.
18663 (arm_evpc_neon_vext): Likewise.
18664 (arm_validize_comparison): Likewise.
18665 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18666 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18667 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18668 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18669 (c6x_preferred_simd_mode): Likewise.
18670 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18671 (epiphany_rtx_costs): Likewise.
18672 * config/epiphany/predicates.md (proper_comparison_operator):
18674 * config/frv/frv.c (condexec_memory_operand): Likewise.
18675 (frv_emit_move): Likewise.
18676 (output_move_single): Likewise.
18677 (output_condmove_single): Likewise.
18678 (frv_hard_regno_mode_ok): Likewise.
18679 (frv_matching_accg_mode): Likewise.
18680 * config/h8300/h8300.c (split_adds_subs): Likewise.
18681 (h8300_rtx_costs): Likewise.
18682 (h8300_print_operand): Likewise.
18683 (compute_mov_length): Likewise.
18684 (output_logical_op): Likewise.
18685 (compute_logical_op_length): Likewise.
18686 (compute_logical_op_cc): Likewise.
18687 (h8300_shift_needs_scratch_p): Likewise.
18688 (output_a_shift): Likewise.
18689 (compute_a_shift_length): Likewise.
18690 (compute_a_shift_cc): Likewise.
18691 (expand_a_rotate): Likewise.
18692 (output_a_rotate): Likewise.
18693 * config/i386/i386.c (classify_argument): Likewise.
18694 (function_arg_advance_32): Likewise.
18695 (function_arg_32): Likewise.
18696 (function_arg_64): Likewise.
18697 (function_value_64): Likewise.
18698 (ix86_gimplify_va_arg): Likewise.
18699 (ix86_legitimate_constant_p): Likewise.
18700 (put_condition_code): Likewise.
18701 (split_double_mode): Likewise.
18702 (ix86_avx256_split_vector_move_misalign): Likewise.
18703 (ix86_expand_vector_logical_operator): Likewise.
18704 (ix86_split_idivmod): Likewise.
18705 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18706 (ix86_build_const_vector): Likewise.
18707 (ix86_build_signbit_mask): Likewise.
18708 (ix86_match_ccmode): Likewise.
18709 (ix86_cc_modes_compatible): Likewise.
18710 (ix86_expand_branch): Likewise.
18711 (ix86_expand_sse_cmp): Likewise.
18712 (ix86_expand_sse_movcc): Likewise.
18713 (ix86_expand_int_sse_cmp): Likewise.
18714 (ix86_expand_vec_perm_vpermi2): Likewise.
18715 (ix86_expand_vec_perm): Likewise.
18716 (ix86_expand_sse_unpack): Likewise.
18717 (ix86_expand_int_addcc): Likewise.
18718 (ix86_split_to_parts): Likewise.
18719 (ix86_vectorize_builtin_gather): Likewise.
18720 (ix86_vectorize_builtin_scatter): Likewise.
18721 (avx_vpermilp_parallel): Likewise.
18722 (inline_memory_move_cost): Likewise.
18723 (ix86_tieable_integer_mode_p): Likewise.
18724 (x86_maybe_negate_const_int): Likewise.
18725 (ix86_expand_vector_init_duplicate): Likewise.
18726 (ix86_expand_vector_init_one_nonzero): Likewise.
18727 (ix86_expand_vector_init_one_var): Likewise.
18728 (ix86_expand_vector_init_concat): Likewise.
18729 (ix86_expand_vector_init_interleave): Likewise.
18730 (ix86_expand_vector_init_general): Likewise.
18731 (ix86_expand_vector_set): Likewise.
18732 (ix86_expand_vector_extract): Likewise.
18733 (emit_reduc_half): Likewise.
18734 (ix86_emit_i387_round): Likewise.
18735 (ix86_mangle_type): Likewise.
18736 (ix86_expand_round_sse4): Likewise.
18737 (expand_vec_perm_blend): Likewise.
18738 (canonicalize_vector_int_perm): Likewise.
18739 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18740 (expand_vec_perm_1): Likewise.
18741 (expand_vec_perm_interleave3): Likewise.
18742 (expand_vec_perm_even_odd_pack): Likewise.
18743 (expand_vec_perm_even_odd_1): Likewise.
18744 (expand_vec_perm_broadcast_1): Likewise.
18745 (ix86_vectorize_vec_perm_const_ok): Likewise.
18746 (ix86_expand_vecop_qihi): Likewise.
18747 (ix86_expand_mul_widen_hilo): Likewise.
18748 (ix86_expand_sse2_abs): Likewise.
18749 (ix86_expand_pextr): Likewise.
18750 (ix86_expand_pinsr): Likewise.
18751 (ix86_preferred_simd_mode): Likewise.
18752 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18753 * config/i386/sse.md (*andnot<mode>3): Likewise.
18754 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18755 (*<code><mode>3): Likewise.
18756 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18757 (ia64_expand_atomic_op): Likewise.
18758 (ia64_arg_type): Likewise.
18759 (ia64_mode_to_int): Likewise.
18760 (ia64_scalar_mode_supported_p): Likewise.
18761 (ia64_vector_mode_supported_p): Likewise.
18762 (expand_vec_perm_broadcast): Likewise.
18763 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18764 (iq2000_function_arg_advance): Likewise.
18765 (iq2000_function_arg): Likewise.
18766 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18767 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18768 (m68k_libcall_value): Likewise.
18769 (m68k_function_value): Likewise.
18770 (sched_attr_op_type): Likewise.
18771 * config/mcore/mcore.c (mcore_output_move): Likewise.
18772 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18774 (microblaze_function_arg): Likewise.
18775 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18776 (mips_print_operand): Likewise.
18777 (mips_mode_ok_for_mov_fmt_p): Likewise.
18778 (mips_vector_mode_supported_p): Likewise.
18779 (mips_preferred_simd_mode): Likewise.
18780 (mips_expand_vpc_loongson_even_odd): Likewise.
18781 (mips_expand_vec_unpack): Likewise.
18782 (mips_expand_vi_broadcast): Likewise.
18783 (mips_expand_vector_init): Likewise.
18784 (mips_expand_vec_reduc): Likewise.
18785 (mips_expand_msa_cmp): Likewise.
18786 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18787 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18788 (cc_flags_for_mode): Likewise.
18789 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18790 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18791 (nds32_output_casesi_pc_relative): Likewise.
18792 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18793 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18794 (nvptx_gen_unpack): Likewise.
18795 (nvptx_gen_pack): Likewise.
18796 (nvptx_gen_shuffle): Likewise.
18797 (nvptx_gen_wcast): Likewise.
18798 (nvptx_preferred_simd_mode): Likewise.
18799 * config/pa/pa.c (pa_secondary_reload): Likewise.
18800 * config/pa/predicates.md (base14_operand): Likewise.
18801 * config/powerpcspe/powerpcspe-c.c
18802 (altivec_resolve_overloaded_builtin): Likewise.
18803 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18805 (rs6000_preferred_simd_mode): Likewise.
18806 (output_vec_const_move): Likewise.
18807 (rs6000_expand_vector_extract): Likewise.
18808 (rs6000_split_vec_extract_var): Likewise.
18809 (reg_offset_addressing_ok_p): Likewise.
18810 (rs6000_legitimate_offset_address_p): Likewise.
18811 (rs6000_legitimize_address): Likewise.
18812 (rs6000_emit_set_const): Likewise.
18813 (rs6000_const_vec): Likewise.
18814 (rs6000_emit_move): Likewise.
18815 (spe_build_register_parallel): Likewise.
18816 (rs6000_darwin64_record_arg_recurse): Likewise.
18817 (swap_selector_for_mode): Likewise.
18818 (spe_init_builtins): Likewise.
18819 (paired_init_builtins): Likewise.
18820 (altivec_init_builtins): Likewise.
18821 (do_load_for_compare): Likewise.
18822 (rs6000_generate_compare): Likewise.
18823 (rs6000_expand_float128_convert): Likewise.
18824 (emit_load_locked): Likewise.
18825 (emit_store_conditional): Likewise.
18826 (rs6000_output_function_epilogue): Likewise.
18827 (rs6000_handle_altivec_attribute): Likewise.
18828 (rs6000_function_value): Likewise.
18829 (emit_fusion_gpr_load): Likewise.
18830 (emit_fusion_p9_load): Likewise.
18831 (emit_fusion_p9_store): Likewise.
18832 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18833 (fusion_gpr_mem_load): Likewise.
18834 (fusion_addis_mem_combo_load): Likewise.
18835 (fusion_addis_mem_combo_store): Likewise.
18836 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18837 (fusion_gpr_mem_load): Likewise.
18838 (fusion_addis_mem_combo_load): Likewise.
18839 (fusion_addis_mem_combo_store): Likewise.
18840 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18842 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18843 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18844 (rs6000_preferred_simd_mode): Likewise.
18845 (output_vec_const_move): Likewise.
18846 (rs6000_expand_vector_extract): Likewise.
18847 (rs6000_split_vec_extract_var): Likewise.
18848 (reg_offset_addressing_ok_p): Likewise.
18849 (rs6000_legitimate_offset_address_p): Likewise.
18850 (rs6000_legitimize_address): Likewise.
18851 (rs6000_emit_set_const): Likewise.
18852 (rs6000_const_vec): Likewise.
18853 (rs6000_emit_move): Likewise.
18854 (rs6000_darwin64_record_arg_recurse): Likewise.
18855 (swap_selector_for_mode): Likewise.
18856 (paired_init_builtins): Likewise.
18857 (altivec_init_builtins): Likewise.
18858 (rs6000_expand_float128_convert): Likewise.
18859 (emit_load_locked): Likewise.
18860 (emit_store_conditional): Likewise.
18861 (rs6000_output_function_epilogue): Likewise.
18862 (rs6000_handle_altivec_attribute): Likewise.
18863 (rs6000_function_value): Likewise.
18864 (emit_fusion_gpr_load): Likewise.
18865 (emit_fusion_p9_load): Likewise.
18866 (emit_fusion_p9_store): Likewise.
18867 * config/rx/rx.c (rx_gen_move_template): Likewise.
18868 (flags_from_mode): Likewise.
18869 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18870 (s390_slb_comparison): Likewise.
18871 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18872 (s390_vector_mode_supported_p): Likewise.
18873 (s390_cc_modes_compatible): Likewise.
18874 (s390_match_ccmode_set): Likewise.
18875 (s390_canonicalize_comparison): Likewise.
18876 (s390_emit_compare_and_swap): Likewise.
18877 (s390_branch_condition_mask): Likewise.
18878 (s390_rtx_costs): Likewise.
18879 (s390_secondary_reload): Likewise.
18880 (__SECONDARY_RELOAD_CASE): Likewise.
18881 (s390_expand_cs): Likewise.
18882 (s390_preferred_simd_mode): Likewise.
18883 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18884 * config/sh/sh.c (sh_print_operand): Likewise.
18885 (dump_table): Likewise.
18886 (sh_secondary_reload): Likewise.
18887 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18888 * config/sh/sh.md (casesi_worker_1): Likewise.
18889 (casesi_worker_2): Likewise.
18890 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18891 (fcc_comparison_operator): Likewise.
18892 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18893 (emit_soft_tfmode_cvt): Likewise.
18894 (sparc_preferred_simd_mode): Likewise.
18895 (output_cbranch): Likewise.
18896 (sparc_print_operand): Likewise.
18897 (sparc_expand_vec_perm_bmask): Likewise.
18898 (vector_init_bshuffle): Likewise.
18899 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18900 (spu_vector_mode_supported_p): Likewise.
18901 (spu_expand_insv): Likewise.
18902 (spu_emit_branch_or_set): Likewise.
18903 (spu_handle_vector_attribute): Likewise.
18904 (spu_builtin_splats): Likewise.
18905 (spu_builtin_extract): Likewise.
18906 (spu_builtin_promote): Likewise.
18907 (spu_expand_sign_extend): Likewise.
18908 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18909 (tilegx_simd_int): Likewise.
18910 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18911 (tilepro_simd_int): Likewise.
18912 * config/v850/v850.c (const_double_split): Likewise.
18913 (v850_print_operand): Likewise.
18914 (ep_memory_offset): Likewise.
18915 * config/vax/vax.c (vax_rtx_costs): Likewise.
18916 (vax_output_int_move): Likewise.
18917 (vax_output_int_add): Likewise.
18918 (vax_output_int_subtract): Likewise.
18919 * config/visium/predicates.md (visium_branch_operator): Likewise.
18920 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18921 (visium_print_operand_address): Likewise.
18922 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18923 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18924 (xtensa_expand_conditional_branch): Likewise.
18925 (xtensa_copy_incoming_a7): Likewise.
18926 (xtensa_output_literal): Likewise.
18927 * dfp.c (decimal_real_maxval): Likewise.
18928 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18931 Alan Hayward <alan.hayward@arm.com>
18932 David Sherwood <david.sherwood@arm.com>
18934 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18935 (mode_nunits_inline): Likewise.
18936 (mode_inner_inline): Likewise.
18937 (mode_unit_size_inline): Likewise.
18938 (mode_unit_precision_inline): Likewise.
18939 (emit_insn_modes_h): Likewise. Also emit a #define of the
18941 (emit_mode_wider): Add an E_ prefix to mode names.
18942 (emit_mode_complex): Likewise.
18943 (emit_mode_inner): Likewise.
18944 (emit_mode_adjustments): Likewise.
18945 (emit_mode_int_n): Likewise.
18946 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18947 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18948 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18949 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18950 (CRC32_BUILTIN, ENTRY): Likewise.
18951 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18952 (aarch64_pop_regs): Likewise.
18953 (aarch64_process_components): Likewise.
18954 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18955 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18956 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18957 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18958 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18959 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18961 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18962 (pcmpistr): Likewise.
18963 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18964 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18965 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18967 * config/rl78/rl78.c (mduc_regs): Likewise.
18968 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18969 (htm_expand_builtin): Likewise.
18970 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18971 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18972 * config/xtensa/xtensa.c (print_operand): Likewise.
18973 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18974 (NUM_MODE_VECTOR_INT): Likewise.
18975 * genoutput.c (null_operand): Likewise.
18976 (output_operand_data): Likewise.
18977 * genrecog.c (print_parameter_value): Likewise.
18978 * lra.c (debug_operand_data): Likewise.
18980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18982 * dwarf2out.c (scompare_loc_descriptor_wide)
18983 (scompare_loc_descriptor_narrow): New functions, split out from...
18984 (scompare_loc_descriptor): ...here.
18985 * expmed.c (emit_store_flag_int): New function, split out from...
18986 (emit_store_flag): ...here.
18988 2017-08-30 Richard Biener <rguenther@suse.de>
18990 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18991 (dwarf2out_early_finish): Move setting of AT_pubnames from
18992 early debug output to early finish.
18994 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18996 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18997 and -mdata-region to the assembler.
18999 2017-08-30 Richard Biener <rguenther@suse.de>
19001 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19003 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19004 (dwarf2out_early_global_decl): Only generate a DIE for the
19005 abstract origin if it doesn't already exist or is a declaration DIE.
19006 (resolve_addr): Do not add the linkage name twice when
19007 generating a stub DIE for the DW_TAG_GNU_call_site target.
19009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19011 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19012 Use machine_mode rather than int for arg1_mode.
19014 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19017 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19018 that the second argument of the built-in functions to unpack
19019 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19020 switch statement instead a lot of if statements.
19021 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19022 Allow 64-bit values to be in Altivec registers as well as
19023 traditional floating point registers.
19024 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19026 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19028 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19029 MAX_REGS_PER_ADDRESS == 1.
19031 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19033 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19034 * config/i386/i386.c (ix86_option_override_internal): Simplify
19035 setting of opts->x_flag_entry.
19037 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19038 Jakub Jelinek <jakub@redhat.com>
19039 Richard Biener <rguenther@suse.de>
19041 PR tree-optimization/81503
19042 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19043 folded constant fits in the target type; reorder tests for clarity.
19045 2017-08-29 Martin Liska <mliska@suse.cz>
19047 * passes.def: Include pass_lower_switch.
19048 * stmt.c (dump_case_nodes): Remove and move to
19049 tree-switch-conversion.
19050 (case_values_threshold): Likewise.
19051 (expand_switch_as_decision_tree_p): Likewise.
19052 (emit_case_decision_tree): Likewise.
19053 (expand_case): Likewise.
19054 (balance_case_nodes): Likewise.
19055 (node_has_low_bound): Likewise.
19056 (node_has_high_bound): Likewise.
19057 (node_is_bounded): Likewise.
19058 (emit_case_nodes): Likewise.
19059 (struct simple_case_node): New struct.
19060 (add_case_node): Remove.
19061 (emit_case_dispatch_table): Use vector instead of case_list.
19062 (reset_out_edges_aux): Remove.
19063 (compute_cases_per_edge): Likewise.
19064 (expand_case): Build list of simple_case_node.
19065 (expand_sjlj_dispatch_table): Use it.
19066 * tree-switch-conversion.c (struct case_node): Moved from
19067 stmt.c and adjusted.
19068 (emit_case_nodes): Likewise.
19069 (node_has_low_bound): Likewise.
19070 (node_has_high_bound): Likewise.
19071 (node_is_bounded): Likewise.
19072 (case_values_threshold): Likewise.
19073 (reset_out_edges_aux): Likewise.
19074 (compute_cases_per_edge): Likewise.
19075 (add_case_node): Likewise.
19076 (dump_case_nodes): Likewise.
19077 (balance_case_nodes): Likewise.
19078 (expand_switch_as_decision_tree_p): Likewise.
19079 (emit_jump): Likewise.
19080 (emit_case_decision_tree): Likewise.
19081 (try_switch_expansion): Likewise.
19082 (do_jump_if_equal): Likewise.
19083 (emit_cmp_and_jump_insns): Likewise.
19084 (fix_phi_operands_for_edge): New function.
19085 (record_phi_operand_mapping): Likewise.
19086 (class pass_lower_switch): New pass.
19087 (pass_lower_switch::execute): New function.
19088 (make_pass_lower_switch): Likewise.
19089 (conditional_probability):
19090 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19091 * tree-pass.h: Add make_pass_lower_switch.
19093 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19096 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19099 2017-08-29 Richard Biener <rguenther@suse.de>
19101 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19102 we do not add a DW_AT_inline attribute twice.
19103 (gen_subprogram_die): Remove code setting DW_AT_inline on
19104 DECL_ABSTRACT_P nodes.
19106 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19108 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19109 calls to internal functions.
19110 (gimplify_modify_expr): Likewise.
19111 * tree-call-cdce.c (use_internal_fn): Likewise.
19112 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19113 (convert_to_divmod): Set the nothrow flag.
19114 * tree-if-conv.c (predicate_mem_writes): Likewise.
19115 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19116 (vectorizable_call): Likewise.
19117 (vectorizable_store): Likewise.
19118 (vectorizable_load): Likewise.
19119 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19120 (vect_recog_mask_conversion_pattern): Likewise.
19122 2017-08-29 Martin Liska <mliska@suse.cz>
19125 * gcc.c (driver_handle_option): Add new argument.
19126 * opts-common.c (handle_option): Pass
19127 target_option_override_hook.
19128 * opts-global.c (lang_handle_option): Add new option.
19129 (set_default_handlers): Add new argument.
19130 (decode_options): Likewise.
19131 * opts.c (target_handle_option): Likewise.
19132 (common_handle_option): Call target_option_override_hook.
19133 * opts.h (struct cl_option_handler_func): Add hook for
19134 target option override.
19135 (struct cl_option_handlers): Likewise.
19136 (set_default_handlers): Add new argument.
19137 (decode_options): Likewise.
19138 (common_handle_option): Likewise.
19139 (target_handle_option): Likewise.
19140 * toplev.c (toplev::main): Pass targetm.target_option.override
19143 2017-08-29 Richard Biener <rguenther@suse.de>
19144 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19146 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19147 life to the active subtree.
19149 2017-08-28 Jeff Law <law@redhat.com>
19151 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19152 derive_equivalences.
19153 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19154 Code moved into....
19155 (edge_info::derive_equivalences): New private member function
19157 * tree-ssa-dom.c (class edge_info): Changed from a struct
19158 to a class. Add ctor/dtor, methods and data members.
19159 (edge_info::edge_info): Renamed from allocate_edge_info.
19160 Initialize additional members.
19161 (edge_info::~edge_info): New.
19162 (free_dom_edge_info): Delete the edge info.
19163 (record_edge_info): Use new class & associated member functions.
19164 Tighten forms for testing for edge equivalences.
19165 (record_temporary_equivalences): Iterate over the simple
19166 equivalences rather than assuming there's only one per edge.
19167 (cprop_into_successor_phis): Iterate over the simple
19168 equivalences rather than assuming there's only one per edge.
19169 (optimize_stmt): Use operand_equal_p rather than pointer
19170 equality for mini-DSE code.
19172 2017-08-28 Nathan Sidwell <nathan@acm.org>
19174 * gcc.c (execute): Fold SIGPIPE handling into switch
19175 statement. Adjust internal error message.
19177 2017-08-28 Richard Biener <rguenther@suse.de>
19180 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19181 Do nothing for removed DIEs.
19183 2017-08-28 Richard Biener <rguenther@suse.de>
19185 PR tree-optimization/81977
19186 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19189 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19192 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19193 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19194 using targetm.gen_mem_thread_fence.
19196 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19199 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19200 predicate to register_operand. Reorder operands.
19201 (*btr<mode>): Ditto.
19202 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19203 (*btr<mode>_mask): Ditto.
19205 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19207 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19208 * config/rs6000/xmmintrin.h: New file.
19209 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19211 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19214 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19215 parameter and_insn and return it.
19216 (recombine_lvx_pattern): Insert a copy to ensure availability of
19217 the base register of the copied masking operation at the point of
19218 the instruction replacement.
19219 (recombine_stvx_pattern): Likewise.
19221 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19223 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19224 undocumented switches.
19225 (-mpower9-dform-vector): Likewise.
19226 (-mpower9-dform): Likewise.
19227 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19228 comments to delete references to -mpower9-dform* switches.
19229 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19230 Delete reference to -mpower9-dform* switches, test for
19231 -mpower9-vector instead.
19232 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19233 (OTHER_P9_VECTOR_MASKS): Likewise.
19234 (POWERPC_MASKS): Likewise.
19235 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19236 tests against -mpower9-dform* to -mpower9-vector. Delete code
19237 that checked for -mpower9-dform* consistancy with other options.
19238 Add test for -mpower9-misc to enable other power9 switches.
19239 (rs6000_init_hard_regno_mode_ok): Likewise.
19240 (rs6000_option_override_internal): Likewise.
19241 (rs6000_emit_prologue): Likewise.
19242 (rs6000_emit_epilogue): Likewise.
19243 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19244 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19245 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19247 (emit_fusion_p9_store): Likewise.
19248 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19249 resetting these macros if the assembler does not support ISA 3.0
19251 (TARGET_P9_DFORM_VECTOR): Likewise.
19252 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19253 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19255 2017-08-25 Alan Modra <amodra@gmail.com>
19257 PR rtl-optimization/81747
19258 * cse.c (cse_extended_basic_block): Don't attempt to record
19259 equivalences for degenerate conditional branches that branch
19260 to their fall-through.
19262 2017-08-24 Martin Sebor <msebor@redhat.com>
19264 PR middle-end/81908
19265 * gimple-fold.c (size_must_be_zero_p): New function.
19266 (gimple_fold_builtin_memory_op): Call it.
19268 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19270 * config/rs6000/mm_malloc.h: New file.
19272 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19274 PR tree-optimization/81913
19275 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19276 analysis when either IVs in condition can wrap.
19278 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19280 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19281 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19283 2017-08-24 Richard Biener <rguenther@suse.de>
19286 * targhooks.c (default_target_can_inline_p): Properly
19287 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19288 is present and always compare.
19289 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19290 infer -mfpmath=sse from TARGET_SSE_P.
19291 (ix86_can_inline_p): Properly use target_option_default_node when
19292 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19294 2017-08-24 Richard Biener <rguenther@suse.de>
19297 * dwarf2out.c (output_die): Handle flag_generate_offload like
19299 (output_comp_unit): Likewise.
19300 (gen_array_type_die): Likewise.
19301 (dwarf2out_early_finish): Likewise.
19302 (note_variable_value_in_expr): Likewise.
19303 (dwarf2out_finish): Likewise. Adjust assert.
19304 * cgraphunit.c (symbol_table::compile): Move setting of
19305 flag_generate_offload earlier ...
19306 (symbol_table::finalize_compilation_unit): ... here, before
19307 early debug finalization.
19309 2017-08-24 Richard Biener <rguenther@suse.de>
19311 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19312 and ipa-fnsummary.h.
19313 (ix86_can_inline_p): When ix86_fpmath flags do not match
19314 check whether the callee uses FP math at all.
19316 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19318 PR middle-end/81931
19319 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19320 instead of TYPE_PRECISION.
19322 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19323 Alan Hayward <alan.hayward@arm.com>
19324 David Sherwood <david.sherwood@arm.com>
19326 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19328 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19330 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19332 (ix86_frame::stack_realign_allocate): New field.
19333 (struct machine_frame_state): Modify comments.
19334 (machine_frame_state::sp_realigned_fp_end): New field.
19335 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19336 layout calculation.
19337 (sp_valid_at): Add assertion to assure no attempt to access invalid
19338 offset of a realigned stack.
19339 (fp_valid_at): Likewise.
19340 (choose_baseaddr): Modify comments.
19341 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19342 ix86_expand_prologue.
19343 (ix86_expand_prologue): Modify stack realignment and allocation.
19344 (ix86_expand_epilogue): Modify comments.
19345 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19346 avx2_runtime, avx512f, and avx512f_runtime.
19348 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19350 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19351 (mstackrealign): Do not init to -1.
19352 * config/i386/i386.c (ix86_option_override_internal):
19353 Check opts_set, not opts when setting default value of
19354 opts->x_ix86_force_align_arg_pointer.
19356 2017-08-23 Richard Biener <rguenther@suse.de>
19358 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19359 lang_hooks.decl_printable_name.
19360 * print-rtl-function.c (print_rtx_function): Likewise.
19361 * tree-pretty-print.c (dump_function_header): Likewise.
19363 2017-08-23 Richard Biener <rguenther@suse.de>
19366 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19367 -g0 at compile-time.
19369 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19371 PR middle-end/19706
19372 * doc/sourcebuild.texi (Other hardware attributes):
19375 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19377 PR middle-end/19706
19378 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19379 Fix single-use check.
19381 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19383 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19385 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19387 * doc/install.texi: Modify to add more details on running selected
19390 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19392 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19393 is combined with -mabi=ms.
19394 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19397 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19399 PR tree-optimization/81488
19400 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19401 and cached_basis fields.
19402 (MAX_SPREAD): New constant.
19403 (alloc_cand_and_find_basis): Initialize new fields.
19404 (clear_visited): New function.
19405 (create_phi_basis_1): Rename from create_phi_basis, set visited
19406 and cached_basis fields.
19407 (create_phi_basis): New wrapper function.
19408 (phi_add_costs_1): Rename from phi_add_costs, add spread
19409 parameter, set visited field, short-circuit when limits reached.
19410 (phi_add_costs): New wrapper function.
19411 (record_phi_increments_1): Rename from record_phi_increments, set
19413 (record_phi_increments): New wrapper function.
19414 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19415 (phi_incr_cost): New wrapper function.
19416 (all_phi_incrs_profitable_1): Rename from
19417 all_phi_incrs_profitable, set visited field.
19418 (all_phi_incrs_profitable): New wrapper function.
19420 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19421 Alan Hayward <alan.hayward@arm.com>
19422 David Sherwood <david.sherwood@arm.com>
19424 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19425 that takes the outer and inner modes.
19426 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19427 comparison as the canonical test for a paradoxical subreg.
19428 * combine.c (simplify_set): Use paradoxical_subreg_p.
19429 (make_extraction): Likewise.
19430 (force_to_mode): Likewise.
19431 (rtx_equal_for_field_assignment_p): Likewise.
19432 (gen_lowpart_for_combine): Likewise.
19433 (simplify_comparison): Likewise.
19434 * cse.c (equiv_constant): Likewise.
19435 * expmed.c (store_bit_field_1): Likewise.
19436 * final.c (alter_subreg): Likewise.
19437 * fwprop.c (propagate_rtx): Likewise.
19438 (forward_propagate_subreg): Likewise.
19439 * ira-conflicts.c (ira_build_conflicts): Likewise.
19440 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19441 * lra-constraints.c (curr_insn_transform): Likewise.
19442 (split_reg): Likewise.
19443 * lra-eliminations.c (move_plus_up): Likewise.
19444 (lra_eliminate_regs_1): Likewise.
19445 * recog.c (general_operand): Likewise.
19446 * ree.c (combine_reaching_defs): Likewise.
19447 * reload.c (push_reload): Likewise.
19448 (find_reloads): Likewise.
19449 * reload1.c (elimination_effects): Likewise.
19450 (compute_reload_subreg_offset): Likewise.
19451 (choose_reload_regs): Likewise.
19452 * rtlanal.c (subreg_lsb_1): Likewise.
19453 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19454 (simplify_subreg): Likewise.
19455 * var-tracking.c (track_loc_p): Likewise.
19456 * emit-rtl.c (byte_lowpart_offset): Likewise.
19457 (paradoxical_subreg_p): Delete out-of-line definition.
19459 2017-08-22 Jeff Law <law@redhat.com>
19461 PR tree-optimization/81741
19462 PR tree-optimization/71947
19463 * tree-ssa-dom.c: Include tree-inline.h.
19464 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19465 equivalences if one is more expensive to compute than the other.
19466 * tree-ssa-scopedtables.h (class const_or_copies): Make
19467 record_const_or_copy_raw method private.
19468 (class avail_exprs_stack): New method simplify_binary_operation.
19469 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19470 avail_exprs_stack::simplify_binary_operation as needed.
19471 (avail_exprs_stack::simplify_binary_operation): New function.
19473 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19475 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19476 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19477 (DOT_SYMBOLS): Likewise.
19478 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19479 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19480 (RS6000_ABI_NAME): Likewise.
19481 (TARGET_CMODEL): Likewise.
19482 (TOC_SECTION_ASM_OP): Likewise.
19483 (SET_CMODEL): New macro.
19484 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19486 2017-08-22 Richard Biener <rguenther@suse.de>
19488 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19489 to allow for free-lang-data replacements similar to verify_type_variant.
19491 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19492 Alan Hayward <alan.hayward@arm.com>
19493 David Sherwood <david.sherwood@arm.com>
19495 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19496 VOIDmode for the LABEL_REF.
19498 2017-08-22 Richard Biener <rguenther@suse.de>
19500 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19502 2017-08-22 Richard Biener <rguenther@suse.de>
19504 * common.opt (feliminate-dwarf2-dups): Ignore.
19505 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19506 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19507 same_die_p_wrap, compute_section_prefix,
19508 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19509 (comdat_symbol_id, comdat_symbol_number): Likewise.
19510 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19512 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19513 (output_die): Mark unreachable path unreachable.
19514 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19515 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19516 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19517 (dwarf2out_early_finish): Likewise.
19519 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19521 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19523 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19526 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19527 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19528 (avr_attribute_table) <io, io_low, address>: Initialize
19529 .decl_required with true.
19531 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19533 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19534 undocumented debugging options.
19535 (-mvsx-scalar-double): Likewise.
19536 (-mallow-df-permute): Likewise.
19537 (-mvectorize-builtins): Likewise.
19538 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19539 (rs6000_builtin_vectorized_function): Likewise.
19540 (rs6000_builtin_md_vectorized_function): Likewise.
19541 (rs6000_opt_vars): Likewise.
19543 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19546 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19547 (*btrq_imm): Rename from *btrq.
19548 (*btcq_imm): Rename from *btcq.
19549 (btsc): New code attribute.
19550 (*<btsc><mode>): New insn pattern.
19551 (*btr<mode>): Ditto.
19552 (*<btsc><mode>_mask): New insn_and_split pattern.
19553 (*btr<mode>_mask): Ditto.
19555 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19556 Alan Hayward <alan.hayward@arm.com>
19557 David Sherwood <david.sherwood@arm.com>
19559 * function.c (pad_below): Simplify padding calculation.
19561 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19562 Alan Hayward <alan.hayward@arm.com>
19563 David Sherwood <david.sherwood@arm.com>
19565 * target.def (function_prologue): Remove frame size argument.
19566 (function_epilogue): Likewise.
19567 * doc/tm.texi: Regenerate.
19568 * final.c (final_start_function): Update call to function_prologue.
19569 (final_end_function): Update call to function_epilogue.
19570 (default_function_pro_epilogue): Remove frame size argument.
19571 * output.h (default_function_pro_epilogue): Likewise.
19572 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19573 (arm_output_function_prologue): Likewise.
19574 * config/frv/frv.c (frv_function_prologue): Likewise.
19575 (frv_function_epilogue): Likewise.
19576 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19577 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19578 (ia64_output_function_epilogue): Likewise.
19579 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19580 (m32r_output_function_epilogue): Likewise.
19581 * config/microblaze/microblaze.c (microblaze_function_prologue)
19582 (microblaze_function_epilogue): Likewise.
19583 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19584 (mips_output_function_epilogue): Likewise.
19585 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19586 (mmix_target_asm_function_epilogue): Likewise.
19587 * config/msp430/msp430.c (msp430_start_function): Likewise.
19588 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19589 (nds32_asm_function_epilogue): Likewise.
19590 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19591 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19592 (pa_output_function_epilogue): Likewise.
19593 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19594 (rs6000_output_function_epilogue): Likewise.
19595 * config/rl78/rl78.c (rl78_start_function): Likewise.
19596 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19597 (rs6000_output_function_epilogue): Likewise.
19598 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19599 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19600 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19601 (sparc_asm_function_epilogue): Likewise.
19603 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19605 * tree.h (type_has_mode_precision_p): New function.
19606 * convert.c (convert_to_integer_1): Use it.
19607 * expr.c (expand_expr_real_2): Likewise.
19608 (expand_expr_real_1): Likewise.
19609 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19610 * match.pd: Likewise.
19611 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19612 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19613 * tree-tailcall.c (process_assignment): Likewise.
19614 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19615 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19616 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19617 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19618 (vectorizable_assignment): Likewise.
19619 (vectorizable_shift): Likewise.
19620 (vectorizable_operation): Likewise.
19621 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19623 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19625 * match.pd: Add pow (C, x) simplification.
19627 2017-08-21 Richard Biener <rguenther@suse.de>
19629 PR tree-optimization/81900
19630 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19631 for blocks with abnormal predecessors.
19632 (compute_antic): Do not set visited flag prematurely.
19634 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19637 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19639 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19641 * stor-layout.h (vector_type_mode): Move to...
19642 * tree.h (vector_type_mode): ...here.
19643 * stor-layout.c (vector_type_mode): Move to...
19644 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19646 2017-08-21 Richard Biener <rguenther@suse.de>
19648 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19649 register_external_die hooks.
19650 (debug_false_tree_charstarstar_uhwistar): Declare.
19651 (debug_nothing_tree_charstar_uhwi): Likewise.
19652 * debug.c (do_nothing_debug_hooks): Adjust.
19653 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19654 (debug_nothing_tree_charstar_uhwi): Likewise.
19655 * dbxout.c (dbx_debug_hooks): Adjust.
19656 (xcoff_debug_hooks): Likewise.
19657 * sdbout.c (sdb_debug_hooks): Likewise.
19658 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19659 * dwarf2out.c (macinfo_label_base): New global.
19660 (dwarf2out_register_external_die): New function for the
19661 register_external_die hook.
19662 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19663 (dwarf2_debug_hooks): Use them.
19664 (dwarf2_lineno_debug_hooks): Adjust.
19665 (struct die_struct): Add with_offset flag.
19666 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19667 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19668 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19669 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19670 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19671 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19672 defining section names for the early LTO debug variants.
19673 (reset_indirect_string): New helper.
19674 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19675 (print_dw_val): Add support for offsetted symbol references.
19676 (get_ultimate_context): Split out from is_cxx.
19677 (is_cxx): Use get_ultimate_context.
19678 (is_fortran): Add decl overload.
19679 (compute_comp_unit_symbol): Split out worker from
19680 compute_section_prefix.
19681 (compute_section_prefix): Call compute_comp_unit_symbol and
19682 set comdat_type_p here.
19683 (output_die): Skip DIE symbol output for the LTO added one.
19684 Handle DIE symbol references with offset.
19685 (output_comp_unit): Guard section name mangling properly.
19686 For LTO debug sections emit a symbol at the section beginning
19687 which we use to refer to its DIEs.
19688 (add_abstract_origin_attribute): For DIEs registered via
19689 dwarf2out_register_external_die directly refer to the early
19690 DIE rather than indirectly through the shadow one we created.
19691 Remove obsolete call to dwarf2out_abstract_function for
19692 non-function/block origins.
19693 (gen_array_type_die): When generating early LTO debug do
19694 not emit DW_AT_string_length.
19695 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19696 late when in LTO. As suggested place a gcc_unreachable for
19697 the DECL_ABSTRACT_P case.
19698 (gen_subprogram_die): Avoid another specification DIE
19699 for early built declarations/definitions for the late LTO case.
19700 (gen_variable_die): Add type references for late duplicated VLA dies
19702 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19703 we have the abstract instance already.
19704 (process_scope_var): Adjust decl DIE contexts in LTO which
19705 first puts them in limbo.
19706 (gen_decl_die): Do not generate type DIEs late apart from
19707 types for VLAs or for decls we do not yet have a DIE. Do not
19708 call dwarf2out_abstract_function late.
19709 (dwarf2out_early_global_decl): Make sure to create DIEs
19710 for abstract instances of a decl first.
19711 (dwarf2out_late_global_decl): Adjust comment.
19712 (output_macinfo_op): With multiple macro sections use
19713 macinfo_label_base to distinguish labels.
19714 (output_macinfo): Likewise. Update macinfo_label_base.
19715 Pass in the line info label.
19716 (note_variable_value_in_expr): When generating LTO resolve
19717 all variable values here by generating DIEs as needed.
19718 (init_sections_and_labels): Add early LTO debug flag parameter
19719 and generate different sections and names if set. Add generation
19720 counter for the labels so we can have multiple of them.
19721 (reset_dies): Helper to allow DIEs to be output multiple times.
19722 (dwarf2out_finish): When outputting DIEs to the fat part of an
19723 LTO object first reset DIEs.
19724 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19725 (modified_type_die): Check for decl_ultimate_origin being self
19727 (gen_type_die_with_usage): Likewise.
19728 (gen_typedef_die): Allow decl_ultimate_origin being self.
19729 (set_decl_abstract_flags): Remove.
19730 (set_block_abstract_flags): Likewise.
19731 (dwarf2out_abstract_function): Treat the early generated DIEs
19732 as the abstract copy and only add DW_AT_inline and
19733 DW_AT_artificial here and call set_decl_origin_self.
19734 If the DIE has an abstract origin don't do anything.
19735 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19736 if we have none yet (Go fails to build one, PR78628).
19737 (variably_modified_type_p): Prevent endless recursion for Ada
19738 cyclic pointer types.
19739 * lto-streamer-in.c: Include debug.h.
19740 (dref_queue): New global.
19741 (lto_read_tree_1): Stream in DIE references.
19742 (lto_input_tree): Register DIE references.
19743 (input_function): Stream DECL_DEBUG_ARGS.
19744 * lto-streamer-out.c: Include debug.h.
19745 (lto_write_tree_1): Output DIE references.
19746 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19747 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19748 (output_function): Stream DECL_DEBUG_ARGS.
19749 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19750 Stream DECL_ABSTRACT_ORIGIN.
19751 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19752 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19753 DECL_CONTEXT for file-scope decls.
19754 * lto-streamer.h (struct dref_entry): Declare.
19755 (dref_queue): Likewise.
19756 * cfgexpand.c (pass_expand::execute): Do not call the
19757 outlining_inline_function hook here.
19758 * lto-wrapper.c (debug_obj): New global.
19759 (tool_cleanup): Unlink it if required.
19760 (debug_objcopy): New function.
19761 (run_gcc): Handle early debug sections in the IL files by
19762 extracting them to separate files, partially linkin them and
19763 feeding the result back as result to the linker.
19764 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19765 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19766 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19767 sections into a separate segment.
19768 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19770 (darwin_asm_dwarf_section): Likewise.
19771 (darwin_asm_output_dwarf_offset): Likewise.
19772 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19774 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19775 Alan Hayward <alan.hayward@arm.com>
19776 David Sherwood <david.sherwood@arm.com>
19778 * read-md.h (md_reader::record_potential_iterator_use): Replace
19779 pointer argument with an rtx and an index.
19780 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19781 (apply_mode_iterator): Likewise.
19782 (apply_code_iterator): Likewise.
19783 (apply_int_iterator): Likewise.
19784 (apply_subst_iterator): Likewise.
19785 (record_iterator_use): Likewise.
19786 (record_attribute_use): Likewise.
19787 (md_reader::record_potential_iterator_use): Likewise. Update calls
19788 to record_iterator_use and apply_iterator.
19789 (iterator_use): Replace ptr with x and index.
19790 (attribute_use): Likewise.
19791 (apply_attribute_uses): Update calls to apply_iterator.
19792 (apply_iterators): Likewise. Update initialization of iterator_use.
19793 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19794 and record_potential_iterator_use.
19795 (rtx_reader::read_rtx_operand): Likewise.
19797 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19798 Alan Hayward <alan.hayward@arm.com>
19799 David Sherwood <david.sherwood@arm.com>
19801 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19804 2017-08-21 Richard Biener <rguenther@suse.de>
19806 PR middle-end/81884
19807 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19808 at struct end conservatively when comparing common bases.
19810 2017-08-21 Richard Biener <rguenther@suse.de>
19812 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19813 (mem_ref_in_stmt): Remove.
19814 (determine_max_movement): Use ref index to get at the reference.
19815 (invariantness_dom_walker::before_dom_children): Deal with
19816 lim data already initialized.
19817 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19819 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19821 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19822 (TARGET_ISA_ROUND): Ditto.
19823 (TARGET_ROUND): Ditto.
19824 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19825 * config/i386/i386.md: Ditto.
19826 * config/i386/sse.md: Ditto.
19827 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19828 with OPTION_MASK_ISA_SSE4_1.
19830 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19833 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19834 __builtin_ia32_lzcnt_u16.
19836 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19839 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19840 (rs6000_set_current_function): Rewrite function to use it.
19842 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19845 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19846 and TYPE_WARN_IF_NOT_ALIGN.
19847 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19848 (handle_warn_if_not_align): New.
19849 (place_union_field): Call handle_warn_if_not_align.
19850 (place_field): Call handle_warn_if_not_align.
19851 Copy TYPE_WARN_IF_NOT_ALIGN.
19852 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19853 (layout_type): Likewise.
19854 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19856 (tree_decl_common): Add warn_if_not_align.
19857 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19858 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19859 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19860 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19861 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19862 * doc/extend.texi: Document warn_if_not_aligned attribute.
19863 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19865 2017-08-17 Martin Liska <mliska@suse.cz>
19868 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19869 (get_data_dependence): Use it as pointer type.
19870 (distribute_loop): Likewise.
19872 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19874 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19875 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19876 (p8_vmrgow_<mode>_direct): New define_insn.
19877 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19878 handle endianness for vmrgew and vmrgow permute patterns.
19880 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19882 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19883 * config/rs6000/rs6000-cpus.def: Remove comment.
19884 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19885 (POWERPC_MASKS): Likewise.
19886 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19887 use of TARGET_VSX_TIMODE.
19888 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19889 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19890 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19891 (rs6000_option_override_internal): Remove dead code.
19892 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19893 (rs6000_legitimize_reload_address): Likewise.
19894 (rs6000_legitimate_address_p): Likewise.
19895 (rs6000_opt_masks): Delete "vsx-timode".
19896 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19897 from function comment.
19898 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19899 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19900 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19902 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19903 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19904 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19905 (VSX_TI): Likewise.
19907 (define_peephole2): Likewise.
19909 2017-08-17 Martin Sebor <msebor@redhat.com>
19912 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19913 past the end of an array.
19914 (test_pp_format): Add test cases.
19916 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19918 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19919 missing ECF_NOTHROW flags.
19921 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19924 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19925 operands residing in integer registers.
19926 (*vsx_le_perm_load_<mode>): Likewise.
19927 (*vsx_le_perm_store_<mode>): Likewise.
19928 (define_peephole2): Add peepholes to optimize the above.
19930 2017-08-17 Marek Polacek <polacek@redhat.com>
19932 PR middle-end/81814
19933 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19934 to mimic what shorten_compare did. Change the return type to bool.
19935 (fold_cond_expr_with_comparison): Update call to
19936 operand_equal_for_comparison_p.
19937 (fold_ternary_loc): Likewise.
19939 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19941 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19943 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19945 2017-08-17 Richard Biener <rguenther@suse.de>
19947 * tree-ssa-structalias.c (solve_graph): When propagating
19948 to successors update the graphs succ edges and avoid duplicate work.
19950 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19953 * config/i386/i386.c (ix86_option_override_internal): Save target
19954 specific options after ix86_stack_protector_guard_reg was changed.
19956 2017-08-17 Richard Biener <rguenther@suse.de>
19958 PR tree-optimization/81827
19959 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19960 (new_var_info): Initialize it conservatively.
19961 (get_call_vi): Mark register vars.
19962 (new_scalar_tmp_constraint_exp): Likewise.
19963 (handle_rhs_call): Likewise.
19964 (handle_const_call): Likewise.
19965 (create_function_info_for): Likewise.
19966 (solve_constraints): Sort varinfos to separate register from
19967 non-register vars to pack points-to solution bitmaps during
19970 2017-08-17 Marek Polacek <polacek@redhat.com>
19972 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19974 2017-08-17 Richard Biener <rguenther@suse.de>
19976 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19977 to true when overflow is undefined and we saturated the result.
19979 2017-08-17 Alan Modra <amodra@gmail.com>
19982 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19983 Don't use store multiple if only one reg needs saving.
19984 (interesting_frame_related_regno): New function.
19985 (rs6000_frame_related): Don't emit frame info for regs that
19987 (rs6000_emit_epilogue): Likewise.
19989 2017-08-16 Nathan Sidwell <nathan@acm.org>
19991 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19992 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19993 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19994 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19995 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19996 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19997 (verify_type): Adjust for TYPE_BINFO move.
19998 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19999 process TYPE_BINFO directly.
20000 (hash_tree): Likewise.
20001 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20003 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20006 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20008 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20010 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20013 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20014 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20015 Add ix86_binary_operator_ok to insn constraint.
20016 (*iordi_1_bts): Ditto.
20017 (*xordi_1_btc): Ditto.
20018 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20019 Update corresponding peephole2 pattern.
20023 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20025 PR tree-optimization/81832
20026 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20027 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20029 2017-08-16 Marek Polacek <polacek@redhat.com>
20032 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20033 regarding size_zero_node.
20035 2017-08-16 Martin Liska <mliska@suse.cz>
20038 * config.gcc: Respect previously set extra_objs in case
20041 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20043 PR tree-optimization/81835
20044 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20045 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20046 not depend on the phi.
20048 2017-08-16 Alan Modra <amodra@gmail.com>
20050 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20053 2017-08-16 Alan Modra <amodra@gmail.com>
20055 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20056 (save_reg_p): ..into this. Update all callers.
20057 (first_reg_to_save): Simplify.
20059 2017-08-16 Alan Modra <amodra@gmail.com>
20061 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20064 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20068 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20069 constant count if that count is less than 32.
20071 2017-08-15 Nathan Sidwell <nathan@acm.org>
20073 * gcc.c (execute): Emit friendlier message if inferior is killed
20074 by an external cause.
20076 2017-08-15 Richard Biener <rguenther@suse.de>
20078 PR tree-optimization/81790
20079 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20080 CONSTRUCTORs from simplifying and VN.
20082 2017-08-14 Martin Sebor <msebor@redhat.com>
20084 * builtin-attrs.def: Add comments.
20086 2017-08-14 Martin Sebor <msebor@redhat.com>
20089 * doc/extend.texi (attribute nonstring): Document new attribute.
20091 2017-08-14 Martin Sebor <msebor@redhat.com>
20094 * tree-diagnostic.c (default_tree_printer): Handle %G.
20095 * gimple-pretty-print.h (percent_G_format): Declare new function.
20096 * gimple-pretty-print.c (percent_G_format): Define.
20097 * tree-pretty-print.c (percent_K_format): Add argument.
20099 2017-08-14 Martin Sebor <msebor@redhat.com>
20101 PR translation/79998
20102 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20103 Remove a stray space.
20105 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20108 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20109 (*iordi_1_bts): Ditto.
20110 (*xordi_1_btc): Ditto.
20112 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20115 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20116 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20118 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20119 quoted strings, and make more translator-friendly.
20120 (darwin_rs6000_override_options): Likewise.
20121 (rs6000_option_override_internal): Likewise.
20122 (rs6000_return_in_memory): Fix overlong line.
20123 (init_cmulative_args): Use quoted strings, and make more
20124 translator-friendly.
20125 (rs6000_pass_by_reference): Fix overlong line.
20126 (def_builtin): Use quoted strings.
20127 (altivec_expand_predicate_builtin): Use quoted strings, and make
20128 more translator-friendly.
20129 (htm_expand_builtin): Use quoted strings.
20130 (cpu_expand_builtin): Use quoted strings, and make more
20131 translator-friendly.
20132 (altivec_expand_builtin): Likewise.
20133 (paired_expand_predicate_builtin): Likewise.
20134 (rs6000_invalid_builtin): Likewise.
20135 (builtin_function_type): Use quoted strings.
20136 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20137 more translator-friendly.
20138 (rs6000_trampoline_init): Likewise.
20139 (rs6000_handle_altivec_attribute): Likewise.
20140 (rs6000_inner_target_options): Use quoted strings.
20141 (rs6000_disable_incompatible_switches): Likewise.
20142 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20143 strings, and make more translator-friendly.
20144 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20146 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20148 PR tree-optimization/81799
20149 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20150 cond_expr to simple gimple operand.
20152 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20154 PR middle-end/46932
20155 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20157 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20161 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20162 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20163 TARGET_GASISR_PROLOGUES.
20164 * config/avr/avr.c (avr_option_override): Same.
20165 (avr_pass_pre_proep::execute): Same.
20167 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20170 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20171 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20173 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20175 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20176 define_insn_and_split. Split to a memory load from 0 in
20177 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20178 using PTR mode iterator.
20179 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20180 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20181 (*add_tp_<mode>): Redefine as define_insn_and_split.
20182 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20183 address space. Merge with *add_tp_x32 using PTR mode iterator.
20184 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20185 Split to an add with a memory load from 0 in
20186 DEFAULT_TLS_SEG_REG address space.
20188 2017-08-12 Andrew Pinski <apinski@cavium.com>
20190 * config/aarch64/aarch64-option-extensions.def (rdma):
20191 Fix feature string to what Linux prints out in /proc/cpuinfo.
20193 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20196 * dwarf2out.c (modified_type_die): For C typedef types that have
20197 an ultimate origin, process the ultimate origin instead of the
20199 (gen_typedef_die): Assert that input DECLs have no ultimate
20201 (gen_type_die_with_usage): For typedef variants that have an
20202 ultimate origin, just call gen_decl_die on the original DECL.
20203 (process_scope_var): Avoid creating DIEs for local typedefs and
20204 concrete static variables.
20206 2017-08-12 Alan Modra <amodra@gmail.com>
20210 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20211 match gnu-user.h startfile.
20212 (ENDFILE_LINUX_SPEC): Similarly.
20214 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20217 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20219 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20221 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20222 * config/aarch64/aarch64.md (mov<mode>): Change.
20223 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20224 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20225 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20227 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20229 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20230 for storage order barriers.
20232 2017-08-11 Martin Liska <mliska@suse.cz>
20235 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20236 variables of void type.
20238 2017-08-11 Martin Liska <mliska@suse.cz>
20240 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20241 TARGET_SUPPORTS_ALIASES.
20242 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20243 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20244 (optimize_weakref): Likewise.
20245 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20246 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20247 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20249 2017-08-11 Martin Liska <mliska@suse.cz>
20252 * config/i386/i386.c (make_resolver_func): Do complete
20253 refactoring of the function.
20255 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20258 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20259 * config/i386/i386.c (ix86_stack_protect_guard): Use
20260 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20261 * doc/invoke.texi (x86 Options): Document
20262 -mstack-protector-guard-symbol= option.
20264 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20266 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20267 * config/i386/i386.c (ix86_split_stack_guard): New function.
20268 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20269 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20270 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20271 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20272 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20273 (split_stack_space_check): Call ix86_split_stack_guard.
20275 2017-08-10 Martin Sebor <msebor@redhat.com>
20277 * print-tree.c (print_node): Print location using the established
20279 Replace spaces with colons.
20280 (debug_raw, debug): Ditto.
20282 2017-08-10 Martin Sebor <msebor@redhat.com>
20285 * pretty-print.c (pp_format): Correct the handling of %s precision.
20287 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20290 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20292 (ix86_finalize_stack_frame_flags): This. Also clear
20293 frame_pointer_needed if -fno-omit-frame-pointer is used without
20295 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20296 with ix86_finalize_stack_frame_flags.
20297 (ix86_expand_epilogue): Likewise.
20298 (ix86_expand_split_stack_prologue): Likewise.
20299 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20301 2017-08-10 Martin Liska <mliska@suse.cz>
20304 * c-attribs.c (handle_target_attribute):
20305 Report warning for an empty string argument of target attribute.
20307 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20310 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20312 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20313 or DECL_NONLOCAL labels.
20314 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20315 or DECL_NONLOCAL labels here.
20317 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20319 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20320 to indicate when early gimple folding has been disabled.
20321 (rs6000_gimple_fold_builtin): Add debug content.
20322 (rs6000_invalid_builtin): Fix whitespace.
20323 (rs6000_expand_builtin): Fix whitespace.
20324 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20326 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20329 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20330 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20332 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20334 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20335 (qdf24xx): Likewise.
20336 * config/aarch64/aarch64.md: Include falkor.md.
20337 * config/aarch64/falkor.md: New.
20339 2017-08-09 Marek Polacek <polacek@redhat.com>
20342 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20343 * diagnostic.c (emit_diagnostic): Add a comment.
20344 (emit_diagnostic_valist): New function.
20346 2017-08-09 Marek Polacek <polacek@redhat.com>
20349 * input.c (make_location): New overload.
20350 * input.h (make_location): Declare.
20352 2017-08-08 Alan Modra <amodra@gmail.com>
20353 H.J. Lu <hongjiu.lu@intel.com>
20356 * gcc.c (NO_PIE_SPEC): Delete.
20357 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20359 (LINK_PIE_SPEC): ..to here.
20360 (LINK_COMMAND_SPEC): Support -no-pie.
20361 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20362 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20364 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20365 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20366 (ENDFILE_CRTEND_SPEC): Similarly.
20368 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20371 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20372 (mstack-protector-guard-offset=): Ditto.
20373 * config/i386/i386.c (ix86_option_override): Handle
20374 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20376 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20377 ix86_stack_protect_guard_offset variables.
20378 (TARGET_STACK_PROTECT_GUARD): Always define.
20379 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20380 and -mstack-protector-guard-offset= options.
20382 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20384 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20385 boundary case for the last candidate.
20387 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20389 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20390 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20392 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20394 PR middle-end/19706
20395 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20396 * config/aarch64/aarch64-builtins.c
20397 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20398 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20399 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20401 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20402 Andrew Pinski <pinskia@gmail.com>
20404 PR middle-end/19706
20405 * internal-fn.def (XORSIGN): New.
20406 * optabs.def (xorsign_optab): New.
20407 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20408 (convert_expand_mult_copysign): New.
20409 (pass_optimize_widening_mul::execute): Call
20410 convert_expand_mult_copysign.
20412 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20414 PR tree-optimization/81354
20415 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20416 Insert on edges rather than explicitly creating landing pads.
20417 (analyze_candidates_and_replace): Commit edge inserts.
20419 2017-08-08 Richard Biener <rguenther@suse.de>
20421 PR middle-end/81719
20422 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20423 (expand_simple_operations): Also look through ADDR_EXPRs with
20424 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20426 2017-08-08 Richard Biener <rguenther@suse.de>
20428 PR tree-optimization/81723
20429 * tree-vect-slp.c (struct bst_traits): New hash traits.
20430 (bst_fail): New global.
20431 (vect_build_slp_tree_2): New worker, split out from ...
20432 (vect_build_slp_tree): ... this now wrapping it with using
20433 bst_fail set to cache SLP tree build fails. Properly handle
20435 (vect_analyze_slp_instance): Allocate and free bst_fail.
20437 2017-08-08 Martin Liska <mliska@suse.cz>
20440 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20441 LABEL_DECLs that can be from a different function.
20443 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20445 PR tree-optimization/81744
20446 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20447 loop's number of iterations.
20449 2017-08-08 Martin Liska <mliska@suse.cz>
20451 * asan.c: Include header files.
20452 * attribs.c (build_decl_attribute_variant): New function moved
20454 (build_type_attribute_qual_variant): Likewise.
20455 (cmp_attrib_identifiers): Likewise.
20456 (simple_cst_list_equal): Likewise.
20457 (omp_declare_simd_clauses_equal): Likewise.
20458 (attribute_value_equal): Likewise.
20459 (comp_type_attributes): Likewise.
20460 (build_type_attribute_variant): Likewise.
20461 (lookup_ident_attribute): Likewise.
20462 (remove_attribute): Likewise.
20463 (merge_attributes): Likewise.
20464 (merge_type_attributes): Likewise.
20465 (merge_decl_attributes): Likewise.
20466 (merge_dllimport_decl_attributes): Likewise.
20467 (handle_dll_attribute): Likewise.
20468 (attribute_list_equal): Likewise.
20469 (attribute_list_contained): Likewise.
20470 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20471 (lookup_attribute_by_prefix): Likewise.
20472 * bb-reorder.c: Include header files.
20473 * builtins.c: Likewise.
20474 * calls.c: Likewise.
20475 * cfgexpand.c: Likewise.
20476 * cgraph.c: Likewise.
20477 * cgraphunit.c: Likewise.
20478 * convert.c: Likewise.
20479 * dwarf2out.c: Likewise.
20480 * final.c: Likewise.
20481 * fold-const.c: Likewise.
20482 * function.c: Likewise.
20483 * gimple-expr.c: Likewise.
20484 * gimple-fold.c: Likewise.
20485 * gimple-pretty-print.c: Likewise.
20486 * gimple.c: Likewise.
20487 * gimplify.c: Likewise.
20488 * hsa-common.c: Likewise.
20489 * hsa-gen.c: Likewise.
20490 * internal-fn.c: Likewise.
20491 * ipa-chkp.c: Likewise.
20492 * ipa-cp.c: Likewise.
20493 * ipa-devirt.c: Likewise.
20494 * ipa-fnsummary.c: Likewise.
20495 * ipa-inline.c: Likewise.
20496 * ipa-visibility.c: Likewise.
20498 * lto-cgraph.c: Likewise.
20499 * omp-expand.c: Likewise.
20500 * omp-general.c: Likewise.
20501 * omp-low.c: Likewise.
20502 * omp-offload.c: Likewise.
20503 * omp-simd-clone.c: Likewise.
20504 * opts-global.c: Likewise.
20505 * passes.c: Likewise.
20506 * predict.c: Likewise.
20507 * sancov.c: Likewise.
20508 * sanopt.c: Likewise.
20509 * symtab.c: Likewise.
20510 * toplev.c: Likewise.
20511 * trans-mem.c: Likewise.
20512 * tree-chkp.c: Likewise.
20513 * tree-eh.c: Likewise.
20514 * tree-into-ssa.c: Likewise.
20515 * tree-object-size.c: Likewise.
20516 * tree-parloops.c: Likewise.
20517 * tree-profile.c: Likewise.
20518 * tree-ssa-ccp.c: Likewise.
20519 * tree-ssa-live.c: Likewise.
20520 * tree-ssa-loop.c: Likewise.
20521 * tree-ssa-sccvn.c: Likewise.
20522 * tree-ssa-structalias.c: Likewise.
20523 * tree-ssa.c: Likewise.
20524 * tree-streamer-in.c: Likewise.
20525 * tree-vectorizer.c: Likewise.
20526 * tree-vrp.c: Likewise.
20527 * tsan.c: Likewise.
20528 * ubsan.c: Likewise.
20529 * varasm.c: Likewise.
20530 * varpool.c: Likewise.
20531 * tree.c: Remove functions moved to attribs.[ch].
20532 * tree.h: Likewise.
20533 * config/aarch64/aarch64.c: Add attrs.h header file.
20534 * config/alpha/alpha.c: Likewise.
20535 * config/arc/arc.c: Likewise.
20536 * config/arm/arm.c: Likewise.
20537 * config/avr/avr.c: Likewise.
20538 * config/bfin/bfin.c: Likewise.
20539 * config/c6x/c6x.c: Likewise.
20540 * config/cr16/cr16.c: Likewise.
20541 * config/cris/cris.c: Likewise.
20542 * config/darwin.c: Likewise.
20543 * config/epiphany/epiphany.c: Likewise.
20544 * config/fr30/fr30.c: Likewise.
20545 * config/frv/frv.c: Likewise.
20546 * config/ft32/ft32.c: Likewise.
20547 * config/h8300/h8300.c: Likewise.
20548 * config/i386/winnt.c: Likewise.
20549 * config/ia64/ia64.c: Likewise.
20550 * config/iq2000/iq2000.c: Likewise.
20551 * config/lm32/lm32.c: Likewise.
20552 * config/m32c/m32c.c: Likewise.
20553 * config/m32r/m32r.c: Likewise.
20554 * config/m68k/m68k.c: Likewise.
20555 * config/mcore/mcore.c: Likewise.
20556 * config/microblaze/microblaze.c: Likewise.
20557 * config/mips/mips.c: Likewise.
20558 * config/mmix/mmix.c: Likewise.
20559 * config/mn10300/mn10300.c: Likewise.
20560 * config/moxie/moxie.c: Likewise.
20561 * config/msp430/msp430.c: Likewise.
20562 * config/nds32/nds32-isr.c: Likewise.
20563 * config/nds32/nds32.c: Likewise.
20564 * config/nios2/nios2.c: Likewise.
20565 * config/nvptx/nvptx.c: Likewise.
20566 * config/pa/pa.c: Likewise.
20567 * config/pdp11/pdp11.c: Likewise.
20568 * config/powerpcspe/powerpcspe.c: Likewise.
20569 * config/riscv/riscv.c: Likewise.
20570 * config/rl78/rl78.c: Likewise.
20571 * config/rx/rx.c: Likewise.
20572 * config/s390/s390.c: Likewise.
20573 * config/sh/sh.c: Likewise.
20574 * config/sol2.c: Likewise.
20575 * config/sparc/sparc.c: Likewise.
20576 * config/spu/spu.c: Likewise.
20577 * config/stormy16/stormy16.c: Likewise.
20578 * config/tilegx/tilegx.c: Likewise.
20579 * config/tilepro/tilepro.c: Likewise.
20580 * config/v850/v850.c: Likewise.
20581 * config/vax/vax.c: Likewise.
20582 * config/visium/visium.c: Likewise.
20583 * config/xtensa/xtensa.c: Likewise.
20585 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20588 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20589 constraints since the -mupper-regs-* switches have been
20591 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20592 into a vector from a double word element that was extracted from
20593 another vector, and eliminate extra XXPERMDI instructions.
20594 (vsx_concat_<mode>_2): Likewise.
20595 (vsx_concat_<mode>_3): Likewise.
20596 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20597 concat to allow optimizing inserts from previous extracts.
20599 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20601 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20602 memory reference to a SSP offset in TLS address space.
20603 (ix86_print_operand) <case '@'>: Remove.
20604 (ix86_print_operand_punct_valid_p): Remove '@' code.
20605 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20606 UNSPEC_SP_TLS_TEST.
20607 (stack_tls_protect_set_<mode>): Remove.
20608 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20609 (stack_tls_protect_test_<mode>): Remove.
20610 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20612 2017-08-07 Olivier Hainque <hainque@adacore.com>
20615 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20617 2017-08-07 Douglas Rupp <rupp@adacore.com>
20619 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20620 variable was referenced as multidir in command.
20622 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20625 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20628 2017-08-07 Martin Liska <mliska@suse.cz>
20630 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20631 * config/rl78/rl78.c: Add include of attribs.h.
20632 * config/sh/sh.c: Likewise.
20633 * config/v850/v850.c: Likewise.
20635 2017-08-07 Tom de Vries <tom@codesourcery.com>
20637 PR middle-end/78266
20638 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20640 2017-08-07 Martin Liska <mliska@suse.cz>
20642 * config/mips/mips.c: Include attribs.h.
20644 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20647 * doc/invoke.texi: Document change in behvaior for -Ofast for
20650 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20652 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20654 (aarch64_pop_regs): Likewise.
20655 (aarch64_gen_load_pair): Likewise.
20656 (aarch64_save_callee_saves): Likewise.
20657 (aarch64_restore_callee_saves): Likewise.
20659 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20661 * config/i386/i386.c: Revert the last change.
20663 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20666 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20668 (ix86_finalize_stack_frame_flags): This. Also clear
20669 frame_pointer_needed if -fno-omit-frame-pointer is used without
20671 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20672 with ix86_finalize_stack_frame_flags.
20673 (ix86_expand_epilogue): Likewise.
20674 (ix86_expand_split_stack_prologue): Likewise.
20676 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20679 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20680 to P_AES for Westmere.
20682 2017-08-07 Jonathan Yong <10walls@gmail.com>
20684 * config/i386/mingw.opt (fset-stack-executable): Removed.
20685 * config/i386/cygming.opt (fset-stack-executable): Moved
20687 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20689 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20691 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20693 2017-08-07 Marek Polacek <polacek@redhat.com>
20695 PR middle-end/81737
20696 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20698 2017-08-07 Martin Liska <mliska@suse.cz>
20700 * attribs.h (canonicalize_attr_name): New function.
20701 (cmp_attribs): Move from c-format.c and adjusted.
20702 (is_attribute_p): Moved from tree.h.
20703 * tree-inline.c: Add new includes.
20704 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20705 (private_is_attribute_p): Remove.
20706 (private_lookup_attribute): Likewise.
20707 (private_lookup_attribute_by_prefix): Simplify.
20708 (remove_attribute): Use is_attribute_p.
20709 * tree.h: Remove removed declarations.
20711 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20713 PR middle-end/81698
20714 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20715 instead of computing it in the function. Formatting fix.
20716 (expand_case): Don't rely on default_edge being the first edge,
20717 clear it if removing it, pass default_edge to
20718 emit_case_dispatch_table.
20719 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20722 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20724 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20725 insn in the function, emit NOP after the insn.
20727 2017-08-06 Tom de Vries <tom@codesourcery.com>
20729 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20732 2017-08-06 Tom de Vries <tom@codesourcery.com>
20734 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20737 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20739 PR tree-optimization/57371
20740 * match.pd: New pattern.
20742 2017-08-04 Marek Polacek <polacek@redhat.com>
20744 PR middle-end/81695
20745 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20746 perform the computation in offset_int.
20748 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20750 PR tree-optimization/81136
20751 * tree-vectorizer.h: Include tree-hash-traits.h.
20752 (vec_base_alignments): New typedef.
20753 (vec_info): Add a base_alignments field.
20754 (vect_record_base_alignments): Declare.
20755 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20757 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20758 (create_data_ref): Add an is_conditional_in_stmt argument.
20759 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20760 the is_conditional_in_stmt field.
20761 (data_ref_loc): Add an is_conditional_in_stmt field.
20762 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20763 (find_data_references_in_stmt): Update call to create_data_ref.
20764 (graphite_find_data_references_in_stmt): Likewise.
20765 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20766 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20767 (vect_record_base_alignment): New function.
20768 (vect_record_base_alignments): Likewise.
20769 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20770 for nested statements even if we fail to compute a misalignment.
20771 Use pooled base alignments for unconditional references.
20772 (vect_find_same_alignment_drs): Compare base addresses instead
20774 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20775 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20777 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20779 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20780 Add an explicit name for the enum. Use auto_vec for slp_instances
20781 and grouped_stores.
20782 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20784 (_bb_vec_info): Add a constructor and destructor.
20785 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20786 (destroy_loop_vec_info): Delete.
20787 (vect_destroy_datarefs): Likewise.
20788 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20789 (vec_info::vec_info): New function.
20790 (vec_info::~vec_info): Likewise.
20791 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20792 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20793 destroy_loop_vec_info.
20794 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20795 (_loop_vec_info::_loop_vec_info): ...this.
20796 (destroy_loop_vec_info): Replace with...
20797 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20798 the stmt_vec_infos. Leave handling of vec_info information to its
20799 destructor. Remove explicit vector releases.
20800 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20801 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20802 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20803 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20804 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20805 (destroy_bb_vec_info): Replace with...
20806 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20807 information to its destructor.
20808 (vect_slp_analyze_bb_1): Use new and delete instead of
20809 new_bb_vec_info and destroy_bb_vec_info.
20810 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20813 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20815 * tree-data-ref.h (subscript): Add access_fn field.
20816 (data_dependence_relation): Add could_be_independent_p.
20817 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20818 (same_access_functions): Move to tree-data-ref.c.
20819 * tree-data-ref.c (ref_contains_union_access_p): New function.
20820 (access_fn_component_p): Likewise.
20821 (access_fn_components_comparable_p): Likewise.
20822 (dr_analyze_indices): Add a reference to access_fn_component_p.
20823 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20825 (constant_access_functions): Likewise.
20826 (add_other_self_distances): Likewise.
20827 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20828 (initialize_data_dependence_relation): Use XCNEW and remove
20829 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20830 of access functions that have the same type. Allow the
20831 subsequence to end with different bases in some circumstances.
20832 Record the chosen access functions in SUB_ACCESS_FN.
20833 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20834 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20835 (subscript_dependence_tester_1): Likewise dra and drb.
20836 (build_classic_dist_vector): Update calls accordingly.
20837 (subscript_dependence_tester): Likewise.
20838 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20839 DDR_COULD_BE_INDEPENDENT_P.
20840 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20841 comp_alias_ddrs instead of may_alias_ddrs.
20842 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20844 (vect_analyze_data_ref_dependence): Use it if
20845 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20846 distance vectors if that fails.
20847 (dependence_distance_ge_vf): New function.
20848 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20849 LOOP_VINFO_MAY_ALIAS_DDRS.
20851 2017-08-04 Richard Biener <rguenther@suse.de>
20853 PR middle-end/81705
20854 * fold-const.c (fold_binary_loc): Properly restrict
20855 minus_var0 && minus_var1 case when associating undefined overflow
20858 2017-08-04 Tom de Vries <tom@codesourcery.com>
20860 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20862 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20864 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20865 Don't start diagnostic messages with a capital letter.
20866 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20868 (rs6000_invalid_builtin): Likewise.
20869 (rs6000_trampoline_init): Likewise.
20871 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20874 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20875 after setting changeable df flags.
20877 2017-08-03 Richard Biener <rguenther@suse.de>
20879 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20880 up if the use is in USE - X.
20882 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20884 * toplev.c (dumpfile.h): New include.
20885 (internal_error_reentered): New static function. Use it...
20886 (internal_error_function): ...here to handle reentered internal_error.
20888 2017-08-03 Richard Biener <rguenther@suse.de>
20890 PR middle-end/81148
20891 * fold-const.c (split_tree): Add minus_var and minus_con
20892 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20893 here but always use minus_*.
20894 (associate_trees): Assert we never associate with MINUS_EXPR
20895 and NULL first operand. Do not recurse for PLUS_EXPR operands
20896 when associating as MINUS_EXPR either.
20897 (fold_binary_loc): Track minus_var and minus_con.
20899 2017-08-03 Tom de Vries <tom@codesourcery.com>
20902 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20903 ACCEL_COMPILER, apply finish_options on
20904 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20906 2017-08-03 Tom de Vries <tom@codesourcery.com>
20909 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20910 function_entry_patch_area_size > 0.
20912 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20915 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20916 instead of 10??LU, perform unit multiplication in wide_int,
20917 don't change alloc_object_size_limit if the limit is larger
20920 PR tree-optimization/81655
20921 PR tree-optimization/81588
20922 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20923 the case when ranges[i].low and high are 1 for unsigned type with
20926 PR middle-end/81052
20927 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20928 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20930 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20932 * tree-vrp.h: Add include guard.
20934 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20937 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20938 (ud2): New insn pattern.
20939 * config/i386/i386.c (ix86_expand_epilogue):
20940 For naked functions, generate ud2 instead of trap insn.
20942 2017-08-02 Marek Polacek <polacek@redhat.com>
20945 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20947 2017-08-02 Tom de Vries <tom@codesourcery.com>
20948 Cesar Philippidis <cesar@codesourcery.com>
20950 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20951 Add missing edge probabilities.
20953 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20955 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20956 Correct endianness.
20958 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20960 PR middle-end/79499
20961 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20962 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20963 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20964 prologue_seq sequences - if any.
20966 2017-08-02 Richard Biener <rguenther@suse.de>
20968 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20969 via vectors if supported, integer extracts via punning if supported
20970 or otherwise vector extracts.
20972 2017-08-02 Richard Biener <rguenther@suse.de>
20974 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20976 (bitmap_insert_into_set): ... this.
20978 2017-08-02 Richard Biener <rguenther@suse.de>
20980 PR tree-optimization/81633
20982 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20984 PR tree-optimization/71752
20985 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20987 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20989 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20990 (machine_function::call_ms2sysv_pad_out): Remove field.
20991 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20992 (ix86_compute_frame_layout): Likewise.
20994 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20997 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20998 attribute with interrupt attribute.
21000 2017-08-01 Andrew Pinski <apinski@cavium.com>
21002 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21003 BIT_INSERT_EXPR's operand 1
21004 to see if the types precision matches.
21006 2017-08-01 Martin Liska <mliska@suse.cz>
21008 PR middle-end/70140
21009 * builtins.c (expand_builtin_memcpy_args): Remove.
21010 (expand_builtin_memcpy): Call newly added function
21011 expand_builtin_memory_copy_args.
21012 (expand_builtin_memcpy_with_bounds): Likewise.
21013 (expand_builtin_mempcpy): Remove last argument.
21014 (expand_builtin_mempcpy_with_bounds): Likewise.
21015 (expand_builtin_memory_copy_args): New function created from
21016 expand_builtin_mempcpy_args with small modifications.
21017 (expand_builtin_mempcpy_args): Remove.
21018 (expand_builtin_stpcpy): Remove unused argument.
21019 (expand_builtin): Likewise.
21020 (expand_builtin_with_bounds): Likewise.
21022 2017-08-01 Martin Liska <mliska@suse.cz>
21025 Make mempcpy more optimal (PR middle-end/70140).
21027 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21030 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21031 __builtin_vec_cmpne verify both arguments are compatible vectors
21032 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21033 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21034 move computation of aligned to after checking the argument types.
21038 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21039 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21042 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21043 Doug Rupp <rupp@adacore.com>
21044 Olivier Hainque <hainque@adacore.com>
21046 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21047 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21049 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21050 for TARGET_OS_CPP_BUILTIN.
21051 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21052 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21054 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21055 passing required abi options to the assembler for EABI configurations.
21056 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21057 of .text.hot and .text.unlikely sections for kernel modules when
21058 using ARM style exceptions.
21059 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21060 options. Add EXTRA_CC1_SPEC.
21061 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21063 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21065 (ARM_TARGET2_DWARF_FORMAT): Define.
21066 * config/arm/t-vxworks: Adjust multilib control to removal of the
21067 Diab command line options.
21069 2017-08-01 Martin Liska <mliska@suse.cz>
21071 PR gcov-profile/81561
21072 * gcov.c (unblock): Make unblocking safe as we need to preserve
21073 index correspondence of blocks and block_lists.
21075 2017-08-01 Richard Biener <rguenther@suse.de>
21077 PR tree-optimization/81181
21078 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21079 (compute_antic): ... end of iteration here.
21081 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21083 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21084 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21085 (ftree-slp-vectorize): Likewise.
21086 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21087 can no longer be set independent of flag_tree_loop_vectorize.
21088 * omp-general.c (emp_max_vf): Likewise.
21089 * opts.c (enable_fdo_optimizations): Remove references to
21090 flag_tree_vectorize, these are now implicit.
21091 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21092 and leave it for the options machinery.
21094 2017-08-01 Martin Liska <mliska@suse.cz>
21096 PR middle-end/70140
21097 * builtins.c (expand_builtin_memcpy_args): Remove.
21098 (expand_builtin_memcpy): Call newly added function
21099 expand_builtin_memory_copy_args.
21100 (expand_builtin_memcpy_with_bounds): Likewise.
21101 (expand_builtin_mempcpy): Remove last argument.
21102 (expand_builtin_mempcpy_with_bounds): Likewise.
21103 (expand_builtin_memory_copy_args): New function created from
21104 expand_builtin_mempcpy_args with small modifications.
21105 (expand_builtin_mempcpy_args): Remove.
21106 (expand_builtin_stpcpy): Remove unused argument.
21107 (expand_builtin): Likewise.
21108 (expand_builtin_with_bounds): Likewise.
21110 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21113 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21114 print "ds:" only for immediates in generic address space.
21116 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21119 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21120 (ix86_function_ok_for_sibcall): Return false for naked functions.
21122 2017-08-01 Richard Biener <rguenther@suse.de>
21124 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21125 (compute_antic): Seed worklist with exit block predecessors.
21126 * cfganal.c (dfs_find_deadend): For a cycle return the source
21127 of the edge closing it.
21129 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21131 * config/aarch64/aarch64.c
21132 (aarch64_can_const_movi_rtx_p): Move 0 check.
21134 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21136 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21137 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21139 * match.pd: Ditto in address comparison pattern.
21141 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21143 PR tree-optimization/81627
21144 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21145 closed ssa form for store-store chain.
21147 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21149 PR tree-optimization/81620
21150 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21151 for store-store chain.
21153 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21155 PR tree-optimization/81588
21156 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21157 ranges[i].in_p, invert comparison code ccode. For >/>=,
21158 swap rhs1 and rhs2 and comparison code unconditionally,
21159 for </<= don't do that. Don't swap rhs1/rhs2 again if
21160 ranges[i].in_p, instead invert comparison code ccode if
21161 opcode or oe->rank is BIT_IOR_EXPR.
21164 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21165 a direct optab to conversion optab.
21166 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21167 with GET_MODE_INNER as last argument instead of optab_handler.
21168 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21169 vector extraction if possible and optab is available.
21170 * expr.c (store_constructor): Use convert_optab_handler instead
21171 of optab_handler. Use vector initialization from smaller
21172 vectors if possible and optab is available.
21173 * tree-vect-stmts.c (vectorizable_load): Likewise.
21174 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21175 now have two modes.
21176 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21177 of vec_init from half-sized vectors with the same element mode.
21178 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21179 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21180 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21181 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21182 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21183 after mode in gen_vec_extract* calls.
21184 (vec_extract<mode>): Renamed to ...
21185 (vec_extract<mode><ssescalarmodelower>): ... this.
21186 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21187 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21188 element mode after mode in gen_vec_init* calls.
21189 (VEC_INIT_HALF_MODE): New mode iterator.
21190 (vec_init<mode>): Renamed to ...
21191 (vec_init<mode><ssescalarmodelower>): ... this.
21192 (vec_init<mode><ssehalfvecmodelower>): New expander.
21193 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21194 (vec_extractv2sfsf): ... this.
21195 (vec_initv2sf): Renamed to ...
21196 (vec_initv2sfsf): ... this.
21197 (vec_extractv2si): Renamed to ...
21198 (vec_extractv2sisi): ... this.
21199 (vec_initv2si): Renamed to ...
21200 (vec_initv2sisi): ... this.
21201 (vec_extractv4hi): Renamed to ...
21202 (vec_extractv4hihi): ... this.
21203 (vec_initv4hi): Renamed to ...
21204 (vec_initv4hihi): ... this.
21205 (vec_extractv8qi): Renamed to ...
21206 (vec_extractv8qiqi): ... this.
21207 (vec_initv8qi): Renamed to ...
21208 (vec_initv8qiqi): ... this.
21209 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21210 (vec_init<mode>): Renamed to ...
21211 (vec_init<mode><VEC_base_l>): ... this.
21212 (vec_extract<mode>): Renamed to ...
21213 (vec_extract<mode><VEC_base_l>): ... this.
21214 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21215 (vec_initv2sfsf): ... this.
21216 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21217 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21218 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21219 element mode after mode in gen_vec_init* calls.
21220 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21221 (vec_init<mode><Vel>): ... this.
21222 (vec_extract<mode>): Renamed to ...
21223 (vec_extract<mode><Vel>): ... this.
21224 * config/aarch64/iterators.md (Vel): New mode attribute.
21225 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21226 Add element mode after mode in gen_vec_extract* calls.
21227 * config/s390/vector.md (non_vec_l): New mode attribute.
21228 (vec_extract<mode>): Renamed to ...
21229 (vec_extract<mode><non_vec_l>): ... this.
21230 (vec_init<mode>): Renamed to ...
21231 (vec_init<mode><non_vec_l>): ... this.
21232 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21233 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21235 * config/arm/iterators.md (V_elem_l): New mode attribute.
21236 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21237 (vec_extract<mode><V_elem_l>): ... this.
21238 (vec_extractv2di): Renamed to ...
21239 (vec_extractv2didi): ... this.
21240 (vec_init<mode>): Renamed to ...
21241 (vec_init<mode><V_elem_l>): ... this.
21242 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21243 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21244 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21245 Add element mode after gen_vec_extract* calls.
21246 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21247 (vec_init<mode><unitmode>): ... this.
21248 (vec_extract<mode>): Renamed to ...
21249 (vec_extract<mode><unitmode>): ... this.
21250 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21251 (vec_init<mode><unitmode>): ... this.
21252 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21253 (vec_initv2sfsf): ... this.
21254 (vec_extractv2sf): Renamed to ...
21255 (vec_extractv2sfsf): ... this.
21256 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21257 Add element mode after gen_vec_extract* calls.
21258 * config/mips/mips.md (unitmode): New mode iterator.
21259 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21260 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21261 * config/spu/spu.md (inner_l): New mode attribute.
21262 (vec_init<mode>): Renamed to ...
21263 (vec_init<mode><inner_l>): ... this.
21264 (vec_extract<mode>): Renamed to ...
21265 (vec_extract<mode><inner_l>): ... this.
21266 * config/sparc/sparc.md (veltmode): New mode iterator.
21267 (vec_init<VMALL:mode>): Renamed to ...
21268 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21269 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21270 (vec_initv2sisi): ... this.
21271 (vec_initv2sf): Renamed to ...
21272 (vec_initv2sfsf): ... this.
21273 (vec_extractv2sf): Renamed to ...
21274 (vec_extractv2sfsf): ... this.
21275 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21276 (vec_init<mode>): Renamed to ...
21277 (vec_init<mode><VEC_base_l>): ... this.
21278 (vec_extract<mode>): Renamed to ...
21279 (vec_extract<mode><VEC_base_l>): ... this.
21280 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21281 (vec_initv2sfsf): ... this.
21282 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21283 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21284 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21285 gen_vec_init* calls.
21287 2017-08-01 Richard Biener <rguenther@suse.de>
21289 PR tree-optimization/81297
21290 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21291 TREE_OVERFLOW from INTEGER_CSTs.
21293 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21295 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21297 2017-07-31 Carl Love <cel@us.ibm.com>
21299 * config/rs6000/rs6000-c: Add support for built-in functions
21300 vector signed char vec_xl_be (signed long long, signed char *);
21301 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21302 vector signed int vec_xl_be (signed long long, signed int *);
21303 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21304 vector signed long long vec_xl_be (signed long long, signed long long *);
21305 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21306 vector signed short vec_xl_be (signed long long, signed short *);
21307 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21308 vector double vec_xl_be (signed long long, double *);
21309 vector float vec_xl_be (signed long long, float *);
21310 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21311 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21312 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21314 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21315 (altivec_expand_builtin): Add switch statement to call
21316 altivec_expand_xl_be for each builtin.
21317 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21318 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21319 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21320 __builtin_vsx_le_be_v16qi.
21321 * doc/extend.texi: Update the built-in documentation file for the
21322 new built-in functions.
21324 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21327 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21329 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21331 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21333 * config.gcc: Add z14.
21334 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21335 CPU model numbers for z13s and z14.
21336 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21338 * config/s390/s390-opts.h (enum processor_type): Rename
21339 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21340 * config/s390/s390.c (processor_table): Add field for CPU name to
21341 be passed to Binutils.
21342 (s390_asm_output_machine_for_arch): Use the new field in
21343 processor_table for Binutils.
21344 (s390_expand_builtin): Replace arch12 with z14.
21345 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21346 (s390_get_sched_attrmask): Likewise.
21347 (s390_get_unit_mask): Likewise.
21348 * config/s390/s390.opt: Add z14 to processor_type enum.
21350 2017-07-31 Martin Jambor <mjambor@suse.cz>
21353 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21354 regardless of optimization level.
21356 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21357 Martin Liska <mliska@suse.cz>
21359 * predict.def: Remove old comment and adjust probability.
21360 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21361 PREDICT statements.
21363 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21366 * config/i386/i386.c (ix86_function_naked): New function.
21367 (ix86_can_use_return_insn_p): Return false for naked functions.
21368 (ix86_expand_prologue): Skip prologue for naked functions.
21369 (ix86_expand_epilogue): Skip epilogue for naked functions
21370 and emit trap instruction.
21371 (ix86_warn_func_return): New function.
21372 (ix86_attribute_table): Add "naked" attribute specification.
21373 (TARGET_WARN_FUNC_RETURN): Define.
21374 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21376 2017-07-31 Martin Liska <mliska@suse.cz>
21378 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21379 (dump_gimple_bb_header): Always dump BB info.
21380 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21382 2017-07-31 Martin Liska <mliska@suse.cz>
21385 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21386 also with current_function_decl non-null equality.
21388 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21391 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21392 change type to the element type, instead add eltype variable and
21393 use it where we are interested in the element type.
21395 PR tree-optimization/81603
21396 * ipa-polymorphic-call.c
21397 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21398 offset arithmetic in offset_int, bail out if the resulting bit offset
21399 doesn't fit into shwi.
21401 2017-07-31 Martin Liska <mliska@suse.cz>
21403 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21404 (gimplify_save_expr): Fix comment.
21406 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21409 * config/i386/i386.c (ix86_function_arg): Update arguments for
21411 (ix86_compute_frame_layout): Set the initial stack offset to
21412 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21413 INCOMING_FRAME_SP_OFFSET.
21414 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21415 stack before exception handler returns.
21416 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21417 the 'ERROR_CODE' for exception handler.
21419 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21421 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21422 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21423 (ASM_OUTPUT_REG_POP): Ditto.
21424 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21425 instead of asm_fprintf to output pure string.
21427 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21429 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21430 to imported_module_or_decl hook.
21431 (debug_nothing_tree_tree_tree_bool): Remove.
21432 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21433 * debug.c (do_nothing_debug_hooks): Use
21434 debug_nothing_tree_tree_tree_bool_bool instead of
21435 debug_nothing_tree_tree_tree_bool.
21436 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21437 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21438 * sdbout.c (sdb_debug_hooks): Likewise.
21439 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21440 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21442 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21443 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21444 attribute, don't add anything.
21446 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21448 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21449 (fold_build2_stat_loc): Likewise.
21450 (fold_build3_stat_loc): Likewise.
21451 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21452 (fold_build1_loc): Remove macro.
21453 (fold_build2_loc): Likewise.
21454 (fold_build3_loc): Likewise.
21456 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21458 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21459 (gimple_build_debug_bind_source_stat): Likewise.
21460 * gimple.h (gimple_build_debug_bind): Remove macro.
21461 (gimple_build_debug_bind_source): Likewise.
21463 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21465 * bitmap.c (bitmap_alloc): Adjust.
21466 (bitmap_gc_alloc): Likewise.
21467 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21469 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21471 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21472 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21473 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21474 (bitmap_gc_alloc_stat): Likewise.
21475 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21477 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21479 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21480 * rtl.h (shallow_copy_rtx): Remove macro.
21482 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21484 * emit-rtl.c (gen_raw_REG): Adjust.
21485 * gengenrtl.c (gendef): Likewise.
21486 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21487 * rtl.h (rtx_alloc): Remove macro.
21489 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21491 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21492 (build_tree_list_stat): Likewise.
21493 * tree.h (build_tree_list): Remove macro.
21494 (build_tree_list_vec): Likewise.
21496 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21498 * tree.c (make_vector_stat): Remove _stat from name.
21499 (build_vector_stat): Likewise.
21500 * tree.h (make_vector_stat): Remove macro.
21501 (build_vector_stat): Likewise.
21503 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21505 * tree.h (build_var_debug_value): Remove prototype.
21507 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21509 * tree.c (tree_cons_stat): Remove _stat from name.
21510 * tree.h (tree_cons): Remove macro.
21512 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21514 * tree.c (build_vl_exp_stat): Remove _stat from name.
21515 * tree.h (build_vl_exp): Remove macro.
21517 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21519 * tree.c (build_decl_stat): Remove _stat from name.
21520 * tree.h (build_decl): Remove macro.
21522 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21524 * gimple.c (gimple_build_with_ops_stat): Adjust.
21525 (gimple_alloc_stat): Remove _stat from name.
21526 * gimple.h (gimple_alloc): Remove macro.
21528 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21530 * tree.c (make_tree_vec_stat): Remove _stat from name.
21531 (grow_tree_vec_stat): Likewise.
21532 * tree.h (make_tree_vec_stat): Adjust prototype.
21533 (grow_tree_vec_stat): Likewise.
21534 (make_tree_vec): Remove macro.
21535 (grow_tree_vec): Likewise.
21537 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21539 * fold-const.c (fold_build1_stat_loc): Adjust.
21540 (fold_build2_stat_loc): Likewise.
21541 (fold_build3_stat_loc): Likewise.
21542 * tree.c (build0_stat): Remove _stat from name.
21543 (build1_stat): Likewise.
21544 (build2_stat): Likewise.
21545 (build3_stat): Likewise.
21546 (build4_stat): Likewise.
21547 (build5_stat): Likewise.
21548 * tree.h (build1_loc): Remove macro, and rename _stat function
21550 (build2_loc): Likewise.
21551 (build3_loc): Likewise.
21552 (build4_loc): Likewise.
21553 (build5_loc): Likewise.
21555 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21557 * tree.c (make_int_cst_stat): Remove _stat from name.
21558 * tree.h (make_int_cst_stat): Adjust prototype.
21559 (make_int_cst): Remove macro.
21561 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21563 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21564 * tree.h (make_tree_binfo_stat): Adjust prototype.
21565 (make_tree_binfo): Remove.
21567 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21569 * tree.c (copy_node_stat): Rename to copy_node.
21570 (build_distinct_type_copy): Adjust.
21571 * tree.h (copy_node_stat): Adjust prototype.
21572 (copy_node): Remove macro.
21574 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21576 * tree.c (make_node_stat): rename to make_node.
21577 (build_tree_list_stat): Adjust.
21578 (build0_stat): Likewise.
21579 (build2_stat): Likewise.
21580 (build3_stat): Likewise.
21581 (build4_stat): Likewise.
21582 (build5_stat): Likewise.
21583 (build_decl_stat): Likewise.
21584 * tree.h (make_node_stat): Adjust prototype.
21585 (make_node): remove macro.
21587 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21589 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21590 (PPC_FEATURE2_SCV): Likewise.
21591 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21593 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21595 * config/aarch64/aarch64.c
21596 (aarch64_internal_mov_immediate): Add new special pattern.
21597 * config/aarch64/aarch64.md (*movdi_aarch64):
21598 Add reg/32bit const mov case.
21600 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21601 Richard Sandiford <richard.sandiford@linaro.org>
21603 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21604 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21605 Add integer and movi cases.
21606 (movi-split-hf-df-sf split, fp16): New.
21607 (enabled): Added TARGET_FP_F16INST.
21608 * config/aarch64/iterators.md (GPF_HF): New.
21609 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21611 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21613 * config/aarch64/aarch64.c
21614 (aarch64_simd_container_mode): Add prototype.
21615 (aarch64_expand_mov_immediate): Add HI support.
21616 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21617 (aarch64_can_const_movi_rtx_p): New.
21618 (aarch64_preferred_reload_class):
21619 Remove restrictions of using FP registers for certain SIMD operations.
21620 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21621 (aarch64_valid_floating_const): Add integer move validation.
21622 (aarch64_simd_imm_scalar_p): Remove.
21623 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21624 (aarch64_legitimate_constant_p): Expand list of supported cases.
21625 * config/aarch64/aarch64-protos.h
21626 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21627 (aarch64_reinterpret_float_as_int): New.
21628 (aarch64_simd_imm_scalar_p): Remove.
21629 * config/aarch64/constraints.md (Uvi): New.
21630 (Dd): Split into Ds and new Dd.
21631 * config/aarch64/aarch64.md (*movsi_aarch64):
21633 (*movdi_aarch64): Add SIMD mov case.
21635 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21637 * tree-predcom.c: (struct chain): Handle store-store chain in which
21638 stores for elimination only store loop invariant values.
21639 (execute_pred_commoning_chain): Ditto.
21640 (prepare_initializers_chain_store_elim): Ditto.
21641 (prepare_finalizers): Ditto.
21642 (is_inv_store_elimination_chain): New function.
21643 (initialize_root_vars_store_elim_1): New function.
21645 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21647 * tree-predcom.c: Revise general description of the pass.
21648 (enum chain_type): New enum type for store elimination.
21649 (struct chain): New field supporting store elimination.
21650 (struct component): Ditto.
21651 (dump_chain): Dump store-stores chain.
21652 (release_chain): Release resources.
21653 (split_data_refs_to_components): Compute and create component
21654 contains only stores for elimination.
21655 (get_chain_last_ref_at): New function.
21656 (make_invariant_chain): Initialization.
21657 (make_rooted_chain): Specify chain type in parameter and record it.
21658 (add_looparound_copies): Skip for store-stores chain.
21659 (determine_roots_comp): Compute type of chain and pass it to
21661 (initialize_root_vars_store_elim_2): New function.
21662 (finalize_eliminated_stores): New function.
21663 (remove_stmt): Handle store for elimination.
21664 (execute_pred_commoning_chain): Execute predictive commoning on
21665 store-store chains.
21666 (determine_unroll_factor): Skip unroll for store-stores chain.
21667 (prepare_initializers_chain_store_elim): New function.
21668 (prepare_initializers_chain): Hanlde store-store chain.
21669 (prepare_finalizers_chain, prepare_finalizers): New function.
21670 (tree_predictive_commoning_loop): Return integer value indicating
21671 if loop is unrolled or lcssa form is corrupted.
21672 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21674 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21676 * tree-predcom.c (initialize_root): Delete.
21677 (execute_pred_commoning_chain): Initialize root vars and replace
21678 reference of non-combined chain directly, rather than call above
21681 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21683 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21684 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21686 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21688 * tree-predcom.c (struct chain): New field init_seq.
21689 (release_chain): Release init_seq.
21690 (prepare_initializers_chain): Record intialization stmts in above
21692 (insert_init_seqs): New function.
21693 (tree_predictive_commoning_loop): Call insert_init_seqs.
21695 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21697 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21699 2017-07-28 Richard Biener <rguenther@suse.de>
21701 * match.pd: Remove superfluous :c.
21702 * genmatch.c (simplify::id): Add member.
21703 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21705 (current_id): New global.
21706 (dt_node::parent): Move from ...
21707 (dt_operand::parent): ... here. Add for_id member.
21708 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21709 (decision_tree::find_node): Relax order requirement when
21710 merging DT_TRUE nodes to ones inbetween the current simplify
21711 and the one we try to merge with. Add diagnostic whenever
21712 we need to enforce pattern order by not merging.
21713 (decision_tree::insert): Set current_id.
21714 (decision_tree::print_node): Dump parent node and for_id.
21715 (parser::last_id): Add member.
21716 (parser::push_simplify): Assign unique id.
21717 (parser::parser): Initialize last_id.
21719 2017-07-28 Martin Liska <mliska@suse.cz>
21722 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21723 gimple_build_debug_bind.
21725 2017-07-28 Richard Biener <rguenther@suse.de>
21727 PR tree-optimization/81502
21728 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21730 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21732 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21733 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21734 for BIT_FIELD_REF args.
21735 * fold-const.c (make_bit_field_ref): Likewise.
21736 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21738 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21741 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21742 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21743 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21744 Or it into SANITIZER_UNDEFINED.
21745 * ubsan.c: Include gimple-fold.h and varasm.h.
21746 (ubsan_expand_ptr_ifn): New function.
21747 (instrument_pointer_overflow): New function.
21748 (maybe_instrument_pointer_overflow): New function.
21749 (instrument_object_size): Formatting fix.
21750 (pass_ubsan::execute): Call instrument_pointer_overflow
21751 and maybe_instrument_pointer_overflow.
21752 * internal-fn.c (expand_UBSAN_PTR): New function.
21753 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21754 * sanitizer.def (__ubsan_handle_pointer_overflow,
21755 __ubsan_handle_pointer_overflow_abort): New builtins.
21756 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21757 * internal-fn.def (UBSAN_PTR): New internal function.
21758 * opts.c (sanitizer_opts): Add pointer-overflow.
21759 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21760 * fold-const.c (build_range_check): Compute pointer range check in
21761 integral type if pointer arithmetics would be needed. Formatting
21764 2017-07-28 Martin Liska <mliska@suse.cz>
21767 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21768 parameters that are of a variable-length.
21770 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21772 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21774 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21775 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21776 (CRT_CALL_STATIC_FUNCTION): Likewise.
21777 (ASM_DEFAULT_SPEC): New define.
21778 (ASM_SPEC32): Likewise.
21779 (ASM_SPEC64): Likewise.
21780 (ASM_SPEC_COMMON): Likewise.
21781 (ASM_SPEC): Likewise.
21782 (INVALID_64BIT): Likewise.
21783 (LINK_OS_DEFAULT_SPEC): Likewise.
21784 (LINK_OS_SPEC32): Likewise.
21785 (LINK_OS_SPEC64): Likewise.
21786 (POWERPC_LINUX): Likewise.
21787 (PTRDIFF_TYPE): Likewise.
21788 (RESTORE_FP_PREFIX): Likewise.
21789 (RESTORE_FP_SUFFIX): Likewise.
21790 (SAVE_FP_PREFIX): Likewise.
21791 (SAVE_FP_SUFFIX): Likewise.
21792 (SIZE_TYPE): Likewise.
21793 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21794 (TARGET_64BIT): Likewise.
21795 (TARGET_64BIT): Likewise.
21796 (TARGET_AIX): Likewise.
21797 (WCHAR_TYPE_SIZE): Likewise.
21798 (WCHAR_TYPE): Undefine.
21799 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21800 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21801 (CPP_OS_RTEMS_SPEC): Delete.
21802 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21803 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21805 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21807 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21809 PR tree-optimization/81578
21810 * tree-parloops.c (build_new_reduction): Bail out if
21811 reduction_code isn't one of the standard OpenMP reductions.
21812 Move the details printing after that decision.
21814 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21816 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21817 related to reload_in_progress.
21818 (splat_input_operand): Likewise.
21819 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21821 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21823 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21824 (TARGET_INSTANTIATE_DECLS): Likewise.
21825 (legitimate_indexed_address_p): Delete reload_in_progress code.
21826 (rs6000_debug_legitimate_address_p): Likewise.
21827 (rs6000_eliminate_indexed_memrefs): Likewise.
21828 (rs6000_emit_le_vsx_store): Likewise.
21829 (rs6000_emit_move_si_sf_subreg): Likewise.
21830 (rs6000_emit_move): Likewise.
21831 (register_to_reg_type): Likewise.
21832 (rs6000_pre_atomic_barrier): Likewise.
21833 (rs6000_machopic_legitimize_pic_address): Likewise.
21834 (rs6000_allocate_stack_temp): Likewise.
21835 (rs6000_address_for_fpconvert): Likewise.
21836 (rs6000_address_for_altivec): Likewise.
21837 (rs6000_secondary_memory_needed_rtx): Delete function.
21838 (rs6000_check_sdmode): Likewise.
21839 (rs6000_alloc_sdmode_stack_slot): Likewise.
21840 (rs6000_instantiate_decls): Likewise.
21841 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21842 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21843 Delete reload_in_progress.
21844 (*vec_reload_and_plus_<mptrsize>): Likewise.
21845 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21846 (vsx_div_v2di): Likewise.
21847 (vsx_udiv_v2di): Likewise.
21849 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21851 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21852 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21853 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21854 (rs6000_debug_reg_global): Delete print of LRA status.
21855 (rs6000_option_override_internal): Delete dead LRA related code.
21856 (rs6000_lra_p): Delete function.
21857 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21859 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21861 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21862 * config/riscv/rtems.h: New file.
21864 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21865 Sudakshina Das <sudi.das@arm.com>
21867 * config/aarch64/aarch64.md
21868 (define_split for and<mode>3nr_compare): Move
21869 non aarch64_logical_operand to a register.
21870 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21871 register immediate operand to a register.
21872 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21874 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21876 PR middle-end/81564
21877 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21879 2017-07-27 Richard Biener <rguenther@suse.de>
21881 PR tree-optimization/81573
21882 PR tree-optimization/81494
21883 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21884 multi defuse cycle case.
21886 2017-07-27 Richard Biener <rguenther@suse.de>
21888 PR tree-optimization/81571
21889 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21892 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21894 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21895 earlier and only if MASK_FPU is set. Adjust formatting.
21897 2017-07-27 Martin Liska <mliska@suse.cz>
21899 * opt-functions.awk: Add validation of value of Init.
21900 * optc-gen.awk: Pass new argument.
21902 2017-07-27 Martin Liska <mliska@suse.cz>
21904 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21905 Fix wrong condition.
21907 2017-07-27 Martin Liska <mliska@suse.cz>
21909 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21910 BBs and edges seen by autoFDO.
21912 2017-07-27 Richard Biener <rguenther@suse.de>
21914 PR tree-optimization/81502
21915 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21916 with incompatible but same sized type.
21917 (execute_update_addresses_taken): Likewise.
21919 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21921 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21922 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21924 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21927 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21928 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21929 Change s_operand to memory_operand.
21931 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21933 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21934 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21935 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21936 Emit instructions rather than returning an expression. Handle TFmode
21937 and KFmode by casting to TImode.
21938 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21939 (rs6000_emit_le_vsx_store): Likewise.
21940 * config/rs6000/vsx.md (VSX_TI): New iterator.
21941 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21942 (*vsx_le_undo_permute_<mode>): Likewise.
21943 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21944 emit the split sequence.
21945 (*vsx_le_perm_store_<mode>): Likewise.
21947 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21949 PR tree-optimization/81555
21950 PR tree-optimization/81556
21951 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21952 if true, force CHANGED for the recursive invocation.
21953 (reassociate_bb): Remember original length of ops array, pass
21954 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21956 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21957 attributes for noipa attribute. For naked attribute use
21958 lookup_attribute first before lookup_attribute_spec.
21959 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21961 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21962 for functions with noipa attribute.
21963 (cgraph_externally_visible_p): Return true for functions with noipa
21965 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21966 for functions with noipa attribute.
21967 * doc/extend.texi: Document noipa function attribute.
21968 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21969 also for functions with noipa attribute.
21970 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21972 2017-07-26 Andrew Pinski <apinski@cavium.com>
21974 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21975 vec_unalign_load_cost and vec_unalign_store_cost.
21977 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21979 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21980 -mvsx-small-integer option.
21981 (ISA_3_0_MASKS_IEEE): Likewise.
21982 (OTHER_VSX_VECTOR_MASKS): Likewise.
21983 (POWERPC_MASKS): Likewise.
21984 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21985 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21986 code, only testing for DImode being allowed in non-VSX floating
21988 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21989 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21991 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21992 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21993 TARGET_P8_VECTOR test.
21994 (rs6000_secondary_reload_simple_move): Likewise.
21995 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21996 since TARGET_P9_VECTOR was already tested.
21997 (rs6000_opt_masks): Remove -mvsx-small-integer.
21998 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21999 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22001 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22002 test for TARGET_VEXTRACTUB was used, and that uses
22004 (p9 extract splitter): Likewise.
22005 (vsx_extract_<mode>_di_p9): Likewise.
22006 (vsx_extract_<mode>_store_p9): Likewise.
22007 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22008 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22009 the elimination of TARGET_VSX_SMALL_INTEGER.
22010 (vsx_extract_<mode>_p8): Likewise.
22011 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22012 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22013 (vsx_set_<mode>_p9): Likewise.
22014 (vsx_set_v4sf_p9): Likewise.
22015 (vsx_set_v4sf_p9_zero): Likewise.
22016 (vsx_insert_extract_v4sf_p9): Likewise.
22017 (vsx_insert_extract_v4sf_p9_2): Likewise.
22018 * config/rs6000/rs6000.md (sign extend splitter): Change
22019 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22020 (floatsi<mode>2_lfiwax_mem): Likewise.
22021 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22022 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22023 since a test for TARGET_P9_VECTOR was used.
22024 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22025 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22026 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22027 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22028 TARGET_P8_VECTOR test.
22029 (fix_trunc<mode>si2_stfiwx): Likewise.
22030 (fix_trunc<mode>si2_internal): Likewise.
22031 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22032 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22034 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22035 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22036 TARGET_P8_VECTOR test.
22037 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22038 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22039 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22041 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22042 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22043 since a test for TARGET_P9_VECTOR was used.
22044 (splitter for loading small constants): Likewise.
22046 2017-07-26 Andrew Pinski <apinski@cavium.com>
22048 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22051 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22054 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22055 (fp_valid_at): Likewise.
22057 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22059 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22060 (qdf24xx_addrcost_table): Likewise.
22061 (cortexa57_tunings): Update to use generic_branch_cost.
22062 (cortexa72_tunings): Likewise.
22063 (cortexa73_tunings): Likewise.
22064 (qdf24xx_tunings): Likewise.
22066 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22068 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22069 (thunderx2t99_branch_cost): Likewise.
22070 (cortexa35_tunings): Update to use generic_branch_cost.
22071 (cortexa53_tunings): Likewise.
22072 (cortexa57_tunings): Likewise.
22073 (cortexa72_tunings): Likewise.
22074 (cortexa73_tunings): Likewise.
22075 (thunderx2t99_tunings): Likewise.
22077 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22079 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22080 (sparc_option_override): Honour MASK_FSMULD.
22081 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22082 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22083 * config/sparc/sparc.opt (mfsmuld): New option.
22084 * doc/invoke.texi (mfsmuld): Document option.
22086 2017-07-26 Marek Polacek <polacek@redhat.com>
22088 PR middle-end/70992
22089 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22091 2017-07-26 Richard Biener <rguenther@suse.de>
22093 * gimple-match-head.c (do_valueize): Return OP if valueize
22095 (get_def): New helper to get at the def stmt of a SSA name
22096 if valueize allows.
22097 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22098 do_valueize to get at the def stmt.
22099 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22101 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22103 PR middle-end/46932
22104 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22106 2017-07-26 Martin Liska <mliska@suse.cz>
22109 * function.c (expand_function_start): Make expansion of
22110 nonlocal_goto_save_area after parm_birth_insn.
22112 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22114 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22115 from all CPU target flags enable members.
22117 2017-07-26 Richard Biener <rguenther@suse.de>
22119 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22120 (decision_tree::gen): Make 'type' const.
22121 (write_predicate): Likewise.
22123 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22125 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22126 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22127 (rs6000_option_override_internal): Likewise.
22128 (rs6000_expand_vector_set): Likewise.
22129 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22130 (TARGET_UPPER_REGS_SF): Likewise.
22131 (TARGET_UPPER_REGS_DI): Likewise.
22132 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22133 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22134 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22135 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22136 (Splitters for DI constants in Altivec registers): Likewise.
22137 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22138 (vsx_set_v4sf_p9): Likewise.
22139 (vsx_set_v4sf_p9_zero): Likewise.
22140 (vsx_insert_extract_v4sf_p9): Likewise.
22141 (vsx_insert_extract_v4sf_p9_2): Likewise.
22143 2017-07-25 Carl Love <cel@us.ibm.com>
22145 * doc/extend.texi: Update the built-in documentation file for the
22146 existing built-in functions
22147 vector signed char vec_cnttz (vector signed char);
22148 vector unsigned char vec_cnttz (vector unsigned char);
22149 vector signed short vec_cnttz (vector signed short);
22150 vector unsigned short vec_cnttz (vector unsigned short);
22151 vector signed int vec_cnttz (vector signed int);
22152 vector unsigned int vec_cnttz (vector unsigned int);
22153 vector signed long long vec_cnttz (vector signed long long);
22154 vector unsigned long long vec_cnttz (vector unsigned long long);
22156 2017-07-25 Andrew Pinski <apinski@cavium.com>
22158 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22159 accesses where the use is for the first operand of a BIT_INSERT.
22161 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22164 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22165 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22167 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22169 * config/i386/gstabs.h: Delete.
22170 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22172 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22174 * config/i386/i386.c (ix86_decompose_address): Do not check for
22175 register RTX when looking at index_reg or base_reg.
22176 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22178 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22180 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22181 to update EH info here.
22183 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22185 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22187 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22189 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22191 2017-07-25 Torsten Duwe <duwe@suse.de>
22193 * common.opt: Introduce -fpatchable-function-entry
22194 command line option, and its variables function_entry_patch_area_size
22195 and function_entry_patch_area_start.
22196 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22197 including a two-value parser.
22198 * target.def (print_patchable_function_entry): New target hook.
22199 * targhooks.h (default_print_patchable_function_entry): New function.
22200 * targhooks.c (default_print_patchable_function_entry): Likewise.
22201 * toplev.c (process_options): Switch off IPA-RA if
22202 patchable function entries are being generated.
22203 * varasm.c (assemble_start_function): Look at the
22204 patchable-function-entry command line switch and current
22205 function attributes and maybe generate NOP instructions by
22206 calling the print_patchable_function_entry hook.
22207 * doc/extend.texi: Document patchable_function_entry attribute.
22208 * doc/invoke.texi: Document -fpatchable_function_entry
22209 command line option.
22210 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22212 * doc/tm.texi: Re-generate.
22214 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22217 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22218 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22220 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22222 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22224 2017-07-25 Richard Biener <rguenther@suse.de>
22226 PR tree-optimization/81455
22227 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22228 not walk in cycles when looking for guards.
22230 2017-07-25 Richard Biener <rguenther@suse.de>
22232 PR tree-optimization/81529
22233 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22234 when optimizing backedge uses.
22236 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22238 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22240 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22241 to dl_section_ref. On AIX, append an expression to subtract
22242 the size of the section length to dl_section_ref.
22244 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22246 * configure.ac: If any of the config.* scripts fail, exit 1.
22247 * configure: Regenerate.
22249 2017-07-25 Richard Biener <rguenther@suse.de>
22251 PR middle-end/81546
22252 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22253 of immediate uses, be more verbose on errors.
22255 2017-07-25 Richard Biener <rguenther@suse.de>
22257 PR tree-optimization/81510
22258 * tree-vect-loop.c (vect_is_simple_reduction): When the
22259 reduction stmt is not inside the loop bail out.
22261 2017-07-25 Richard Biener <rguenther@suse.de>
22263 PR tree-optimization/81303
22264 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22265 profitability check against LOOP_VINFO_NITERSM1.
22267 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22269 * domwalk.c (cmp_bb_postorder): Simplify.
22270 (sort_bbs_postorder): New function. Use it...
22271 (dom_walker::walk): ...here to optimize common cases.
22273 2017-07-25 Martin Liska <mliska@suse.cz>
22276 * ipa-visibility.c (function_and_variable_visibility): Make the
22277 redirection just on target that supports aliasing.
22278 Fix GNU coding style.
22280 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22283 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22285 * config/arm/rtems.h: Likewise.
22286 * config/bfin/rtems.h: Likewise.
22287 * config/i386/rtemself.h: Likewise.
22288 * config/lm32/rtems.h: Likewise.
22289 * config/m32c/rtems.h: Likewise.
22290 * config/m68k/rtemself.h: Likewise.
22291 * config/microblaze/rtems.h: Likewise.
22292 * config/mips/rtems.h: Likewise.
22293 * config/moxie/rtems.h: Likewise.
22294 * config/nios2/rtems.h: Likewise.
22295 * config/powerpcspe/rtems.h: Likewise.
22296 * config/rs6000/rtems.h: Likewise.
22297 * config/rtems.h: Likewise.
22298 * config/sh/rtems.h: Likewise.
22299 * config/sh/rtemself.h: Likewise.
22300 * config/sparc/rtemself.h: Likewise.
22302 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22305 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22306 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22307 * tree-ssa-structalias.c (alias_get_name): Same.
22309 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22312 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22313 instructions if no du chain is found.
22315 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22317 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22319 2017-07-25 Richard Biener <rguenther@suse.de>
22321 PR middle-end/81505
22322 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22325 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22327 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22328 upper-regs options.
22329 (ISA_2_7_MASKS_SERVER): Likewise.
22330 (ISA_3_0_MASKS_IEEE): Likewise.
22331 (OTHER_P8_VECTOR_MASKS): Likewise.
22332 (OTHER_VSX_VECTOR_MASKS): Likewise.
22333 (POWERPC_MASKS): Likewise.
22334 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22335 duplicate list of options.
22336 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22337 explicit -mupper-regs options.
22338 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22339 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22340 alias for -mupper-regs-df.
22341 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22342 (rs6000_init_hard_regno_mode_ok): Likewise.
22343 (rs6000_option_override_internal): Likewise.
22344 (rs6000_opt_masks): Likewise.
22345 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22346 options in terms of whether -mvsx or -mpower8-vector was used.
22347 (TARGET_UPPER_REGS_DI): Likewise.
22348 (TARGET_UPPER_REGS_SF): Likewise.
22349 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22350 -mupper-regs-* options.
22352 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22354 * passes.c (emergency_dump_function): Print some empty lines and a
22355 header before the RTL dump.
22357 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22359 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22361 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22364 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22365 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22367 2017-07-24 Carl Love <cel@us.ibm.com>
22369 * config/rs6000/rs6000-c.c: Add support for built-in functions
22370 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22371 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22372 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22373 vec_extract_fp_from_shortl): Add defines for the two builtins.
22374 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22375 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22377 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22378 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22379 * doc/extend.texi: Update the built-in documentation file for the
22380 new built-in function.
22382 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22385 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22387 * doc/generic.texi: Likewise.
22388 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22389 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22391 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22393 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22394 (aarch64_mls_elt_merge<mode>): Likewise.
22396 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22398 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22399 having __cxa_atexit.
22401 2017-07-23 Michael Collison <michael.collison@arm.com>
22403 * config/arm/arm.c (arm_option_override): Deprecate
22404 use of -mstructure-size-boundary.
22405 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22406 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22408 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22411 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22412 Reduce cost estimate for direct moves.
22414 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22417 * config/i386/i386.c (ix86_option_override_internal): Disable
22418 BMI, BMI2 and TBM instructions for -m16.
22420 2017-07-21 Carl Love <cel@us.ibm.com>
22422 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22423 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22424 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22425 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22426 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22427 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22428 VMULOSW): New enum "unspec" values.
22429 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22430 altivec_vmulosw): New patterns.
22431 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22432 VMULOSW): Add definitions.
22434 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22436 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22437 (qdf24xx): Likewise.
22438 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22439 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22440 (AARCH64_FL_V8_1): Renumber.
22441 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22442 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22443 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22444 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22445 rdma to feature modifiers list.
22447 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22449 PR middle-end/56727
22450 * ipa-visibility (function_and_variable_visibility): Convert
22451 recursive PLT call to direct call if appropriate.
22453 2017-07-21 Andrew Pinski <apinski@cavium.com>
22455 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22456 operand 1 to see if the types precision matches.
22457 * fold-const.c (operand_equal_p): Likewise.
22459 2017-07-21 Richard Biener <rguenther@suse.de>
22461 PR tree-optimization/81303
22462 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22463 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22464 (vect_peeling_hash_get_lowest_cost): Adjust.
22465 (vect_enhance_data_refs_alignment): Likewise. Use
22466 vect_get_peeling_costs_all_drs to compute the penalty for no
22467 peeling to match up costs.
22469 2017-07-21 Richard Biener <rguenther@suse.de>
22471 PR tree-optimization/81500
22472 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22473 we didn't identify a reduction path.
22475 2017-07-21 Tom de Vries <tom@codesourcery.com>
22476 Cesar Philippidis <cesar@codesourcery.com>
22478 PR gcov-profile/81442
22479 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22482 2017-07-21 Tom de Vries <tom@codesourcery.com>
22485 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22487 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22488 nvptx_override_options_after_change.
22490 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22492 * dwarf2out.c (output_file_names): Avoid double testing for
22493 dwarf_version >= 5.
22495 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22497 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22499 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22501 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22503 (try_crossjump_to_edge): Do not punt on partitioned functions.
22505 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22507 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22508 Put all BBs reachable only via paths crossing cold region to cold
22510 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22512 2016-07-21 Richard Biener <rguenther@suse.de>
22514 PR tree-optimization/81303
22515 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22516 into account prologue and epilogue iterations when raising
22517 min_profitable_iters to sth at least covering one vector iteration.
22519 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22521 * config/arm/arm.c (arm_test_cpu_arch_dat):
22524 2017-07-20 Nathan Sidwell <nathan@acm.org>
22526 Remove TYPE_METHODS.
22527 * tree.h (TYPE_METHODS): Delete.
22528 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22529 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22530 (dbxout_type_methods): Scan TYPE_FIELDS.
22531 (dbxout_type): Don't check TYPE_METHODS here.
22532 * function.c (use_register_for_decl): Always ignore register for
22533 class types when not optimizing.
22534 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22535 * tree.c (free_lang_data_in_type): Stitch out member functions and
22536 templates from TYPE_FIELDS.
22537 (build_distinct_type_copy, verify_type_variant,
22538 verify_type): Member fns are on TYPE_FIELDS.
22539 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22540 * tree-pretty-print.c (dump_generic_node): Likewise.
22542 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22545 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22546 V2TImode and V4TImode.
22547 (ix86_expand_vector_extract): Likewise.
22548 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22549 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22550 (ssescalarmode): Handle V4TImode and V2TImode.
22551 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22552 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22553 (VEXTRACTI128_MODE): New mode iterator.
22554 (splitter for *vec_extractv?ti first element): New.
22555 (VEC_INIT_MODE): New mode iterator.
22556 (vec_init<mode>): Consolidate 3 expanders into one using
22557 VEC_INIT_MODE mode iterator.
22559 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22561 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22562 non_spilled_static_chain_regno_p.
22564 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22566 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22568 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22570 * bb-reorder.c (connect_traces): Allow copying of blocks within
22573 2017-07-20 Richard Biener <rguenther@suse.de>
22575 * gimple.h (gimple_phi_result): Add gphi * overload.
22576 (gimple_phi_result_ptr): Likewise.
22577 (gimple_phi_arg): Likewise. Adjust index assert to only
22578 allow actual argument accesses rather than all slots available
22580 (gimple_phi_arg_def): Add gphi * overload.
22581 * tree-phinodes.c (make_phi_node): Initialize only actual
22583 (resize_phi_node): Clear memory not covered by old node,
22584 do not initialize excess argument slots.
22585 (reserve_phi_args_for_new_edge): Initialize new argument slot
22588 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22590 PR tree-optimization/81388
22592 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22594 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22595 by removing computation of may_be_zero.
22597 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22598 Tom de Vries <tom@codesourcery.com>
22600 PR middle-end/81030
22601 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22602 when gimple level profile disagrees with what RTL expander did.
22604 2017-07-20 Richard Biener <rguenther@suse.de>
22606 PR tree-optimization/61171
22607 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22608 (vect_analyze_stmt): Add slp instance parameter.
22609 (vectorizable_reduction): Likewise.
22610 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22611 (vect_is_simple_reduction): Deal with chains not detected
22612 as SLP reduction chain, specifically not properly associated
22613 chains containing a mix of plus/minus.
22614 (get_reduction_op): Remove.
22615 (get_initial_defs_for_reduction): Simplify, pass in whether
22616 this is a reduction chain, pass in the SLP node for the PHIs.
22617 (vect_create_epilog_for_reduction): Get the SLP instance as
22619 (vectorizable_reduction): Get the SLP instance as arg.
22620 During analysis remember the SLP node with the PHIs in the
22621 instance. Simplify getting at the vectorized reduction PHIs.
22622 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22623 through SLP instance.
22624 (vect_slp_analyze_operations): Likewise.
22625 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22626 (vect_transform_stmt): Likewise.
22628 2017-07-20 Tom de Vries <tom@codesourcery.com>
22630 PR tree-optimization/81489
22631 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22632 read of phi arg location to before loop that modifies phi.
22634 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22636 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22639 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22641 PR middle-end/81331
22642 * except.c (execute): Fix ordering issue.
22644 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22646 PR rtl-optimization/81423
22647 * combine.c (make_compound_operation_int): Don't try to optimize
22648 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22650 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22652 PR rtl-optimization/81423
22653 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22654 with a constant that is -1 in the truncated to mode.
22656 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22658 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22659 (determine_unlikely_bbs): ... here.
22660 * predict.h (propagate_unlikely_bbs_forward): Declare.
22661 * cfgexpand.c (pass_expand::execute): Use it.
22662 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22664 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22665 propagate_unlikely_bbs_forward.
22667 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22669 PR middle-end/81331
22670 * except.c (maybe_add_nop_after_section_switch): New function.
22673 2017-07-19 Tom de Vries <tom@codesourcery.com>
22675 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22677 2017-07-19 Tom de Vries <tom@codesourcery.com>
22679 * gimple.h (gimple_phi_arg): Make assert more strict.
22681 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22683 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22684 * config/rs6000/mmintrin.h: New file.
22685 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22687 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22689 PR tree-optimization/81346
22690 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22692 2017-07-19 Tom de Vries <tom@codesourcery.com>
22694 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22696 2017-07-19 Tom de Vries <tom@codesourcery.com>
22698 * config/nvptx/nvptx-modes.def: Add V2DImode.
22699 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22700 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22701 (nvptx_output_mov_insn): Handle lack of mov.b128.
22702 (nvptx_print_operand): Handle 'H' and 'L' codes.
22703 (nvptx_vector_mode_supported): Allow V2DImode.
22704 (nvptx_preferred_simd_mode): New function.
22705 (nvptx_data_alignment): New function.
22706 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22707 nvptx_preferred_simd_mode.
22708 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22710 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22712 2017-07-19 Tom de Vries <tom@codesourcery.com>
22714 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22715 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22716 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22717 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22718 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22719 (mov<VECIM>_insn): New define_insn.
22720 (define_expand "mov<VECIM>): New define_expand.
22722 2017-07-19 Tom de Vries <tom@codesourcery.com>
22724 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22726 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22728 PR tree-optimization/81346
22729 * fold-const.h (fold_div_compare, range_check_type): Declare.
22730 * fold-const.c (range_check_type): New function.
22731 (build_range_check): Use range_check_type.
22732 (fold_div_compare): No longer static, rewritten into
22733 a match.pd helper function.
22734 (fold_comparison): Don't call fold_div_compare here.
22735 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22736 as helper function.
22738 2017-07-19 Nathan Sidwell <nathan@acm.org>
22740 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22741 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22742 * tree.c (find_decls_types_r, verify_type): Use
22743 TYPE_{MIN,MAX}_VALUE_RAW.
22744 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22745 (hash_tree): Likewise.
22746 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22748 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22751 2017-07-18 Tom de Vries <tom@codesourcery.com>
22753 PR middle-end/81464
22754 * omp-expand.c (expand_omp_for_static_chunk): Handle
22755 equal-argument loop exit phi.
22757 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22760 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22761 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22762 operand 2 predicate.
22763 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22764 operand 2 predicate.
22765 (ror,rol -> rorx splitters): Use const_int_operand as
22766 operand 2 predicate.
22768 2017-06-18 Richard Biener <rguenther@suse.de>
22770 PR tree-optimization/81410
22771 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22772 the gap in the ! slp_perm SLP case after each group.
22774 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22776 PR middle-end/81463
22777 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22780 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22782 PR middle-end/81462
22783 * predict.c (set_even_probabilities): Cleanup; do not affect
22784 probabilities that are already known.
22785 (combine_predictions_for_bb): Call even when count is set.
22787 2017-07-18 Nathan Sidwell <nathan@acm.org>
22789 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22792 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22795 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22796 optimization for loop niter analysis.
22798 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22801 * config/avr/avr.c (avr_optimize_casesi): Don't use
22802 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22804 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22806 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22807 body_cost_vec from _vect_peel_extended_info.
22808 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22809 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22812 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22814 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22816 2017-07-18 Richard Biener <rguenther@suse.de>
22818 PR tree-optimization/80620
22819 PR tree-optimization/81403
22820 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22821 info when re-using a VN table entry.
22823 2017-07-18 Richard Biener <rguenther@suse.de>
22825 PR tree-optimization/81418
22826 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22827 vectype_in. Verify that with lane-reducing reduction operations
22828 we have a single def-use cycle.
22830 2017-07-17 Carl Love <cel@us.ibm.com>
22832 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22834 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22835 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22836 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22837 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22838 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22839 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22840 VMULOSW): New enum "unspec" values.
22841 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22842 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22843 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22844 altivec_vmulosw): New patterns.
22845 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22846 VMULOSW): Add definitions.
22848 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22850 * config/alpha/alpha.c: Include predict.h.
22852 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22854 * tree-vrp.c (compare_assert_loc): Fix comparison function
22855 to return predictable results.
22857 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22859 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22861 (subdi3): Likewise.
22862 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22863 * doc/invoke.texi (mexpand-adddi): Update text.
22865 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22867 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22868 that also clobbers the CC register. The old expand code is moved
22870 (*arc_clzsi2): ... here.
22871 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22872 the CC register. The old expand code is moved to ...
22873 (arc_ctzsi2): ... here.
22875 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22877 * config/arc/arc.opt (mindexed-loads): Use initial value
22878 TARGET_INDEXED_LOADS_DEFAULT.
22879 (mauto-modify-reg): Use initial value
22880 TARGET_AUTO_MODIFY_REG_DEFAULT.
22881 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22882 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22883 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22884 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22886 2017-07-17 Martin Liska <mliska@suse.cz>
22889 * opts.c (finish_options): Do not allow -fgnu-tm
22890 w/ -fsanitize={kernel-,}address. Say sorry.
22892 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22895 * tree-loop-distribution.c (classify_partition): Only assert on
22896 numer of iterations.
22897 (merge_dep_scc_partitions): Delete prameter. Update function call.
22898 (distribute_loop): Remove code handling loop with unknown niters.
22899 (pass_loop_distribution::execute): Skip loop with unknown niters.
22901 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22904 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22905 function sort_partitions_by_post_order.
22907 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22909 PR tree-optimization/81374
22910 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22911 the max index of basic blocks, rather than number of basic blocks.
22913 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22915 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22917 (arc_legitimate_pic_operand_p): Likewise.
22918 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22920 (arc_needs_pcl_p): Likewise.
22921 (arc_legitimate_pc_offset_p): Likewise.
22922 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22923 function is also used in constrains.md.
22924 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22925 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22926 PLUS. Only return true/false in known cases, otherwise assert.
22927 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22928 is already called in arc_legitimate_constant_p.
22929 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22931 (LEGITIMATE_PIC_OPERAND_P): Use
22932 arc_raw_symbolic_reference_mentioned_p function.
22933 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22938 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22939 Andrew Burgess <andrew.burgess@embecosm.com>
22941 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22942 (arc_return_address_register): New function.
22943 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22944 (arc_handle_fndecl_attribute): Add naked attribute.
22945 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22946 (TARGET_WARN_FUNC_RETURN): Likewise.
22947 (arc_allocate_stack_slots_for_args): New function.
22948 (arc_warn_func_return): Likewise.
22949 (machine_function): Change type fn_type.
22950 (arc_compute_function_type): Consider new naked function type,
22951 change function return type.
22952 (arc_must_save_register): Adapt to handle new
22953 arc_compute_function_type's return type.
22954 (arc_expand_prologue): Likewise.
22955 (arc_expand_epilogue): Likewise.
22956 (arc_return_address_regs): Delete.
22957 (arc_return_address_register): New function.
22958 (arc_epilogue_uses): Use above function.
22959 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22960 (arc_function_type): Change encoding, add naked type.
22961 (ARC_INTERRUPT_P): Change to handle the new encoding.
22962 (ARC_FAST_INTERRUPT_P): Likewise.
22963 (ARC_NORMAL_P): Define.
22964 (ARC_NAKED_P): Likewise.
22965 (arc_compute_function_type): Delete prototype.
22966 * config/arc/arc.md (in_ret_delay_slot): Use
22967 arc_return_address_register function.
22968 (simple_return): Likewise.
22969 (p_return_i): Likewise.
22971 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22973 PR tree-optimization/81428
22974 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22975 can't be built for those types.
22977 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22979 Remove stuff dead since r239246.
22981 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22982 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22983 (avr_inform_devices): Remove dead stuff.
22985 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22987 * config/arm/arm_neon.h: Fix softp typo.
22989 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22991 PR tree-optimization/81365
22992 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22993 aggregate moves onto bb predecessor edges, make sure there are no
22994 loads that could alias the lhs in between the start of bb and the
22997 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23000 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23001 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23002 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23004 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23006 PR tree-optimization/81396
23007 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23008 (init_symbolic_number): Initialize it to 1.
23009 (perform_symbolic_merge): Add n_ops from both operands into the new
23011 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23012 without base_addr as useless if they need more than one operation.
23013 (bswap_replace): Handle !bswap case for NULL base_addr.
23015 2017-07-17 Tom de Vries <tom@codesourcery.com>
23018 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23021 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23023 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23024 conditional builtin define __FIX_LEON3FT_B2BST.
23026 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23028 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23029 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23032 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23034 PR rtl-optimization/81424
23035 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23036 to remove potential trapping from operands if -fnon-call-exceptions.
23038 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23040 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23041 profile_proability for scalling.
23042 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23044 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23046 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23048 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23050 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23051 fixpoint arithmetics.
23053 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23055 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23056 fixpoint arithmetics.
23058 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23060 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23061 fixpoint arithmetics.
23063 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23065 * profile-count.h (profile_probability::from_reg_br_prob_note,
23066 profile_probability::to_reg_br_prob_note): New functions.
23067 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23068 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23069 * predict.c (probability_reliable_p): Update.
23070 (edge_probability_reliable_p): Update.
23071 (br_prob_note_reliable_p): Update.
23072 (invert_br_probabilities): Update.
23073 (add_reg_br_prob_note): New function.
23074 (combine_predictions_for_insn): Update.
23075 * asan.c (asan_clear_shadow): Update.
23076 * cfgbuild.c (compute_outgoing_frequencies): Update.
23077 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23078 (update_br_prob_note): Update.
23079 (rtl_verify_edges): Update.
23080 (purge_dead_edges): Update.
23081 (fixup_reorder_chain): Update.
23082 * emit-rtl.c (try_split): Update.
23083 * ifcvt.c (cond_exec_process_insns): Update.
23084 (cond_exec_process_if_block): Update.
23085 (dead_or_predicable): Update.
23086 * internal-fn.c (expand_addsub_overflow): Update.
23087 (expand_neg_overflow): Update.
23088 (expand_mul_overflow): Update.
23089 * loop-doloop.c (doloop_modify): Update.
23090 * loop-unroll.c (compare_and_jump_seq): Update.
23091 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23092 * predict.h: Update.
23093 * reorg.c (mostly_true_jump): Update.
23095 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23096 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23097 * config/arc/arc.c: (emit_unlikely_jump): Update.
23098 * config/arm/arm.c: (emit_unlikely_jump): Update.
23099 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23100 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23101 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23102 (ix86_print_operand): Update.
23103 (ix86_split_fp_branch): Update.
23104 (predict_jump): Update.
23105 * config/ia64/ia64.c (ia64_print_operand): Update.
23106 * config/mmix/mmix.c (mmix_print_operand): Update.
23107 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23108 (rs6000_expand_split_stack_prologue): Update.
23109 * config/rs6000/rs6000.c: Update.
23110 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23111 (s390_expand_vec_movstr): Update.
23112 (s390_expand_cs_tdsi): Update.
23113 (s390_expand_split_stack_prologue): Update.
23114 * config/sh/sh.c (sh_print_operand): Update.
23115 (expand_cbranchsi4): Update.
23116 (expand_cbranchdi4): Update.
23117 * config/sparc/sparc.c (output_v9branch): Update.
23118 * config/spu/spu.c (get_branch_target): Update.
23119 (ea_load_store_inline): Update.
23120 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23121 * config/tilepro/tilepro.c: Update.
23123 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23125 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23126 (gimplify_save_expr): Likewise.
23128 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23130 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23132 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23134 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23136 * timevar.def (TV_IPA_FNSUMMARY): Define.
23138 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23140 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23141 to back store errata sensitive sequence from being generated.
23142 (sqrtdf2_fix): Likewise.
23144 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23146 * tree-ssa-threadupdate.c (compute_path_counts,
23147 update_joiner_offpath_counts): Use profile_probability.
23149 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23152 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23154 * config/arm/arm-c.c (arm_cpu_builtins): Define
23155 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23157 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23159 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23160 array entries to represent __ieee128 versions of the
23161 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23162 scalar_extract_sig, and scalar_insert_exp built-in functions.
23163 (altivec_resolve_overloaded_builtin): Add special case handling
23164 for the __builtin_scalar_insert_exp function, as represented by
23165 the P9V_BUILTIN_VEC_VSIEDP constant.
23166 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23167 exponent support for __ieee128 argument.
23168 (VSESQP): Add scalar extract signature support for __ieee128
23170 (VSTDCNQP): Add scalar test negative support for __ieee128
23172 (VSIEQP): Add scalar insert exponent support for __int128 argument
23173 with __ieee128 result.
23174 (VSIEQPF): Add scalar insert exponent support for __ieee128
23175 argument with __ieee128 result.
23176 (VSTDCQP): Add scalar test data class support for __ieee128
23178 (VSTDCNQP): Add overload support for scalar test negative with
23179 __ieee128 argument.
23180 (VSTDCQP): Add overload support for scalar test data class
23181 __ieee128 argument.
23182 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23183 UNSPEC_VSX_SXSIGDP.
23184 (UNSPEC_VSX_SIEXPQP): New constant.
23185 (xsxexpqp): New insn for VSX scalar extract exponent quad
23187 (xsxsigqp): New insn for VSX scalar extract significand quad
23189 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23190 precision with floating point argument.
23191 (xststdcqp): New expand for VSX scalar test data class quad
23193 (xststdcnegqp): New expand for VSX scalar test negative quad
23195 (xststdcqp): New insn to match expansions for VSX scalar test data
23196 class quad precision and VSX scalar test negative quad precision.
23197 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23198 special case operand checking to enforce that second operand of
23199 VSX scalar test data class with quad precision argument is a 7-bit
23201 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23202 prototypes and descriptions of __ieee128 versions of
23203 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23204 scalar_test_data_class, and scalar_test_neg built-in functions.
23206 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23208 PR tree-optimization/81162
23209 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23210 replace a negate with an add.
23212 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23214 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23216 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23218 * config/arm/arm-c.c (arm_cpu_builtins): Define
23219 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23221 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23223 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23224 (armv8-r): Set ARM Cortex-R52 as default CPU.
23225 * config/arm/arm-tables.opt: Regenerate.
23226 * config/arm/arm-tune.md: Regenerate.
23227 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23229 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23230 extension for -mcpu=cortex-r52.
23232 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23234 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23235 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23236 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23237 (fp-armv8): Define it as FP_ARMv8 only.
23238 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23239 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23241 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23242 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23243 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23244 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23245 than TARGET_FPU_ARMV8.
23246 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23247 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23248 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23250 * config/arm/neon.md (neon_vrint): Likewise.
23251 (neon_vcvt): Likewise.
23252 (neon_<fmaxmin_op><mode>): Likewise.
23253 (<fmaxmin><mode>3): Likewise.
23254 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23255 * config/arm/predicates.md (arm_cond_move_operator): Check against
23256 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23258 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23260 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23261 to top of function.
23263 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23265 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23266 loop in comment with memset.
23268 2017-07-14 Martin Liska <mliska@suse.cz>
23270 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23271 * dwarf2out.c (is_java): Remove the function.
23272 (output_pubname): Remove usage of the function.
23273 (lower_bound_default): Remove usage of DW_LANG_Java.
23274 (gen_compile_unit_die): Likewise.
23275 * gcc.c: Remove compiler defaults for .java and .zip files.
23276 * gimple-expr.c (remove_suffix): Change as there's no longer
23277 extension than 4-letter one.
23278 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23279 (gimplify_save_expr): Likewise.
23280 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23281 as it's possible even for other languages than Java.
23282 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23283 * lto-opts.c (lto_write_options): Remove reference to Java.
23284 * opts.c (strip_off_ending): Update file extension handling.
23285 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23286 * tree-eh.c (lower_resx): Likewise.
23287 * tree.c (free_lang_data_in_type): Remove dead code.
23288 (find_decls_types_r): Likewise.
23289 (build_common_builtin_nodes): Remove Java from a comment.
23290 (verify_type): Remove dead code.
23291 * varasm.c (assemble_external): Remove Java from a comment.
23293 2017-07-14 Martin Liska <mliska@suse.cz>
23295 * opts.c (finish_options): Add quotes.
23296 (common_handle_option): Likewise.
23298 2017-07-14 Martin Liska <mliska@suse.cz>
23300 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23301 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23302 Remove N_SO_PASCAL.
23303 * dwarf2out.c (lower_bound_default): Do not handle
23305 (gen_compile_unit_die): Likewise.
23306 * gcc.c: Remove default extension binding for GNU Pascal.
23307 * stmt.c: Remove Pascal language from a comment.
23308 * xcoffout.c: Likewise.
23310 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23313 * diagnostic-show-locus.c (fixit_cmp): New function.
23314 (layout::layout): Sort m_fixit_hints.
23315 (column_range::column_range): Assert that the values are valid.
23316 (struct char_span): New struct.
23317 (correction::overwrite): New method.
23318 (struct source_line): New struct.
23319 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23320 calls in terms of classes source_line and char_span, and
23321 correction::overwrite.
23322 (selftest::test_overlapped_fixit_printing_2): New function.
23323 (selftest::diagnostic_show_locus_c_tests): Call it.
23325 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23327 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23328 early if there is no lhs.
23330 2017-07-13 Martin Liska <mliska@suse.cz>
23332 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23333 (gen_reference_type_die): Likewise.
23334 * stor-layout.c: Remove Pascal-related comment.
23336 2017-07-13 Martin Liska <mliska@suse.cz>
23338 * opts.c (finish_options): Add quotes to error messages.
23339 (parse_sanitizer_options): Likewise.
23341 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23343 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23345 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23347 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23349 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23351 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23353 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23355 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23358 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23359 provides the hardware capability bits, define the macro
23360 __BUILTIN_CPU_SUPPORTS__.
23361 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23362 if GLIBC does not provide the hardware capability bits. Add a
23363 gcc_unreachable call if the built-in cpu function is neither
23364 __builtin_cpu_is nor __builtin_cpu_supports.
23365 (rs6000_get_function_versions_dispatcher): Change the warning
23366 that an old GLIBC is used which does not export the capability
23367 bits to be an error.
23368 * doc/extend.texi (target_clones attribute): Document the
23369 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23370 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23371 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23372 the macros defined by GCC if the newer GLIBC is available.
23374 2017-07-12 Jeff Law <law@redhat.com>
23376 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23377 remaining includes slightly.
23378 * config/riscv/riscv-builtins.c: Include profile-count.h.
23380 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23383 * config/avr/avr.c (avr_set_current_function): In diagnostic
23384 messages: Quote keywords and (parts of) identifiers.
23385 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23388 2017-07-12 Carl Love <cel@us.ibm.com>
23390 * config/rs6000/rs6000-c.c: Add support for built-in functions
23391 vector bool char vec_revb (vector bool char);
23392 vector bool short vec_revb (vector short char);
23393 vector bool int vec_revb (vector bool int);
23394 vector bool long long vec_revb (vector bool long long);
23395 * doc/extend.texi: Update the built-in documentation file for the
23396 new built-in functions.
23398 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23400 * config/s390/s390.md: Remove movcc splitter.
23402 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23404 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23405 load/store on condition.
23407 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23410 * config/avr/avr.c (avr_encode_section_info)
23411 [progmem && !TREE_READONLY]: Error if progmem object needs
23414 2017-07-11 Michael Collison <michael.collison@arm.com>
23416 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23419 2017-07-11 Carl Love <cel@us.ibm.com>
23421 * config/rs6000/rs6000-c.c: Add support for builtins
23422 vector unsigned int vec_parity_lsbb (vector signed int);
23423 vector unsigned int vec_parity_lsbb (vector unsigned int);
23424 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23425 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23426 vector unsigned long long vec_parity_lsbb (vector signed long long);
23427 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23428 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23429 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23430 * doc/extend.texi: Update the built-in documentation file for the
23431 new built-in functions.
23433 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23435 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23436 (layout::m_primary_loc): New field.
23437 (layout::layout): Initialize new field. Move location filtering
23438 logic from here to...
23439 (layout::maybe_add_location_range): ...this new method. Add
23440 support for filtering to just the lines already specified by other
23442 (layout::will_show_line_p): New method.
23443 (gcc_rich_location::add_location_if_nearby): New method.
23444 (selftest::test_add_location_if_nearby): New test function.
23445 (selftest::diagnostic_show_locus_c_tests): Call it.
23446 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23449 2017-07-11 Tom de Vries <tom@codesourcery.com>
23451 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23452 (bb_first_real_insn): New function.
23453 (nvptx_single): Add extra initialization of broadcasted condition
23456 2017-07-11 Nathan Sidwell <nathan@acm.org>
23458 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23460 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23462 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23463 Binutils doc as TEXI will mess them up.
23464 * doc/invoke.texi (AVR Options): Same here.
23466 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23468 * config/sparc/sparc.opt (mfix-ut700): New option.
23469 (mfix-gr712rc): Likewise.
23470 (sparc_fix_b2bst): New variable.
23471 * doc/invoke.texi (SPARC options): Document them.
23472 (ARM options): Fix warnings.
23473 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23474 instructions to prevent sequences that can trigger the store-store
23475 errata for certain LEON3FT processors.
23476 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23477 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23478 * config/sparc/sparc.md (fix_b2bst): New attribute.
23479 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23481 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23484 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23486 (*rsqrtsf2_sse): Ditto.
23488 (div<mode>3): Macroize insn from divdf3 and divsf3
23489 using MODEF mode iterator.
23491 2017-07-10 Martin Sebor <msebor@redhat.com>
23493 PR tree-optimization/80397
23494 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23495 instead of testing for equality to INTEGER_TYPE.
23497 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23499 * config.gcc: Remove uclibc from arc target spec.
23501 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23503 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23505 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23508 * lto-wrapper.c (remove_option): New function.
23509 (merge_and_complain): Merge PIC/PIE options more realistically.
23511 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23513 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23517 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23518 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23519 * config.in: Regenerate.
23520 * configure: Regenerate.
23521 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23522 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23523 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23524 (TARGET_GASISR_PROLOGUES): ...target mask.
23525 * common/config/avr/avr-common.c
23526 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23527 Set -mgas-isr-prologues.
23528 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23529 INSERT_PASS_BEFORE for it.
23530 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23531 * config/avr/avr.c (avr_option_override)
23532 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23533 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23534 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23535 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23536 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23538 (make_avr_pass_pre_proep): New function.
23539 (emit_push_sfr) <treg>: Add argument to function and use it
23540 instead of TMP_REG.
23541 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23542 and set machine->gasisr.yes.
23543 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23544 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23545 __gcc_isr.n_pushed to .L__stack_usage.
23546 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23547 (avr_asm_final_postscan_insn): ...this new static function.
23548 * config/avr/avr.h (machine_function)
23549 <is_no_gccisr, use_L__stack_usage>: New fields.
23550 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23551 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23552 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23553 (gasisr, *gasisr): New expander and insn.
23554 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23555 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23556 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23558 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23560 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23563 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23565 Move jump-tables out of .text again.
23568 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23569 (ASM_OUTPUT_ADDR_VEC): New function.
23570 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23571 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23572 INSN_ADDRESSes as asm comment.
23573 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23574 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23575 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23576 * config/avr/avr.md (*tablejump): Adjust comment.
23577 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23578 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23580 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23581 (avr_output_addr_vec): New proto.
23582 (avr_log_t) <insn_addresses>: New field.
23584 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23587 * config/i386/i386.c (ix86_function_arg_advance): Set
23588 outgoing_args_on_stack to true if there are outgoing arguments
23590 (ix86_function_arg): Likewise.
23591 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23592 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23593 * config/i386/i386.h (machine_function): Add
23594 outgoing_args_on_stack.
23596 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23598 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23599 supporting pthreds.
23600 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23602 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23604 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23605 (REAL_H): Remove $(MACHMODE_H).
23606 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23608 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23609 $(MACHMODE_H) and double-int.h.
23610 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23612 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23615 2017-07-07 Andrew Pinski <apinski@cavium.com>
23617 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23618 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23620 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23622 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23623 Add warning if GCC was not configured to link against a GLIBC that
23624 exports the hardware capability bits.
23625 (make_resolver_func): Make resolver function private and not a
23626 COMDAT function. Create the name with clone_function_name instead
23627 of make_unique_name.
23630 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23631 correct operand in doing the split.
23633 2017-07-07 Carl Love <cel@us.ibm.com>
23635 * config/rs6000/rs6000-c: Add support for built-in function
23636 vector unsigned short vec_pack_to_short_fp32 (vector float,
23638 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23639 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23640 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23641 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23642 (convert_4f32_8i16): Add define_expand.
23643 * doc/extend.texi: Update the built-in documentation file for the
23644 new built-in function.
23646 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23648 * config/sparc/m8.md: New file.
23649 * config/sparc/sparc.md: Include m8.md.
23651 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23653 * config/sparc/sparc.opt: New option -mvis4b.
23654 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23655 (sparc_option_override): Handle VIS4B.
23656 (enum sparc_builtins): Define
23657 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23658 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23659 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23660 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23661 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23662 (check_constant_argument): New function.
23663 (sparc_vis_init_builtins): Define builtins
23664 __builtin_vis_dictunpack{8,16,32},
23665 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23666 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23667 __builtin_vis_fpcmpde{8,16,32}shl and
23668 __builtin_vis_fpcmpur{8,16,32}shl.
23669 (sparc_expand_builtin): Check that the constant operands to
23670 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23671 constant and in range.
23672 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23674 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23675 (SPARC_IMM5_P): Likewise.
23676 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23677 (enabled): Handle vis4b.
23678 (UNSPEC_DICTUNPACK): New unspec.
23679 (UNSPEC_FPCMPSHL): Likewise.
23680 (UNSPEC_FPUCMPSHL): Likewise.
23681 (UNSPEC_FPCMPDESHL): Likewise.
23682 (UNSPEC_FPCMPURSHL): Likewise.
23683 (cpu_feature): New CPU feature `vis4b'.
23684 (dictunpack{8,16,32}): New insns.
23685 (FPCSMODE): New mode iterator.
23686 (fpcscond): New code iterator.
23687 (fpcsucond): Likewise.
23688 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23689 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23690 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23691 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23692 * config/sparc/constraints.md: Define constraints `q' for unsigned
23693 2-bit integer constants and `t' for unsigned 5-bit integer
23695 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23697 (imm5_operand_dictunpack16): Likewise.
23698 (imm5_operand_dictunpack32): Likewise.
23699 (imm2_operand): Likewise.
23700 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23701 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23702 ditunpack* and fpcmp*shl builtins.
23704 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23706 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23707 * config.in: Add HAVE_AS_SPARC6 define.
23708 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23710 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23712 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23713 (CPP_CPU_SPEC): Handle m8.
23714 (ASM_CPU_SPEC): Likewise.
23715 * config/sparc/sparc-opts.h (enum processor_type): Add
23717 * config/sparc/sparc.c (m8_costs): New struct.
23718 (sparc_option_override): Handle TARGET_CPU_m8.
23719 (sparc32_initialize_trampoline): Likewise.
23720 (sparc64_initialize_trampoline): Likewise.
23721 (sparc_issue_rate): Likewise.
23722 (sparc_register_move_cost): Likewise.
23723 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23724 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23725 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23726 (CPP_CPU_SPEC): Handle M8.
23727 (ASM_CPU_SPEC): Likewise.
23728 (AS_M8_FLAG): Define.
23729 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23730 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23731 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23733 * configure: Regenerate.
23734 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23737 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23739 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23741 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23742 ("*movdi_insn_sp32"): Do not set v3pipe.
23743 ("*movsi_insn"): Likewise.
23744 ("*movdi_insn_sp64"): Likewise.
23745 ("*movsf_insn"): Likewise.
23746 ("*movdf_insn_sp32"): Likewise.
23747 ("*movdf_insn_sp64"): Likewise.
23748 ("*zero_extendsidi2_insn_sp64"): Likewise.
23749 ("*sign_extendsidi2_insn"): Likewise.
23750 ("*mov<VM32:mode>_insn"): Likewise.
23751 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23752 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23753 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23754 ("<vlop:code><VL:mode>3"): Likewise.
23755 ("*not_<vlop:code><VL:mode>3"): Likewise.
23756 ("*nand<VL:mode>_vis"): Likewise.
23757 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23758 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23759 ("one_cmpl<VL:mode>2"): Likewise.
23760 ("faligndata<VM64:mode>_vis"): Likewise.
23761 ("alignaddrsi_vis"): Likewise.
23762 ("alignaddrdi_vis"): Likweise.
23763 ("alignaddrlsi_vis"): Likewise.
23764 ("alignaddrldi_vis"): Likewise.
23765 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23766 ("bmaskdi_vis"): Likewise.
23767 ("bmasksi_vis"): Likewise.
23768 ("bshuffle<VM64:mode>_vis"): Likewise.
23769 ("cmask8<P:mode>_vis"): Likewise.
23770 ("cmask16<P:mode>_vis"): Likewise.
23771 ("cmask32<P:mode>_vis"): Likewise.
23772 ("pdistn<P:mode>_vis"): Likewise.
23773 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23775 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23777 * config/sparc/sparc.md ("subtype"): New insn attribute.
23778 ("*wrgsr_sp64"): Set insn subtype.
23779 ("*rdgsr_sp64"): Likewise.
23780 ("alignaddrsi_vis"): Likewise.
23781 ("alignaddrdi_vis"): Likewise.
23782 ("alignaddrlsi_vis"): Likewise.
23783 ("alignaddrldi_vis"): Likewise.
23784 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23785 ("fexpand_vis"): Likewise.
23786 ("fpmerge_vis"): Likewise.
23787 ("faligndata<VM64:mode>_vis"): Likewise.
23788 ("bshuffle<VM64:mode>_vis"): Likewise.
23789 ("cmask8<P:mode>_vis"): Likewise.
23790 ("cmask16<P:mode>_vis"): Likewise.
23791 ("cmask32<P:mode>_vis"): Likewise.
23792 ("fchksm16_vis"): Likewise.
23793 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23794 ("fmean16_vis"): Likewise.
23795 ("fp<plusminus_insn>64_vis"): Likewise.
23796 ("<plusminus_insn>v8qi3"): Likewise.
23797 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23798 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23799 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23800 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23801 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23802 ("*movqi_insn"): Likewise.
23803 ("*movhi_insn"): Likewise.
23804 ("*movsi_insn"): Likewise.
23805 ("movsi_pic_gotdata_op"): Likewise.
23806 ("*movdi_insn_sp32"): Likewise.
23807 ("*movdi_insn_sp64"): Likewise.
23808 ("movdi_pic_gotdata_op"): Likewise.
23809 ("*movsf_insn"): Likewise.
23810 ("*movdf_insn_sp32"): Likewise.
23811 ("*movdf_insn_sp64"): Likewise.
23812 ("*zero_extendhisi2_insn"): Likewise.
23813 ("*zero_extendqihi2_insn"): Likewise.
23814 ("*zero_extendqisi2_insn"): Likewise.
23815 ("*zero_extendqidi2_insn"): Likewise.
23816 ("*zero_extendhidi2_insn"): Likewise.
23817 ("*zero_extendsidi2_insn_sp64"): Likewise.
23818 ("ldfsr"): Likewise.
23819 ("prefetch_64"): Likewise.
23820 ("prefetch_32"): Likewise.
23821 ("tie_ld32"): Likewise.
23822 ("tie_ld64"): Likewise.
23823 ("*tldo_ldub_sp32"): Likewise.
23824 ("*tldo_ldub1_sp32"): Likewise.
23825 ("*tldo_ldub2_sp32"): Likewise.
23826 ("*tldo_ldub_sp64"): Likewise.
23827 ("*tldo_ldub1_sp64"): Likewise.
23828 ("*tldo_ldub2_sp64"): Likewise.
23829 ("*tldo_ldub3_sp64"): Likewise.
23830 ("*tldo_lduh_sp32"): Likewise.
23831 ("*tldo_lduh1_sp32"): Likewise.
23832 ("*tldo_lduh_sp64"): Likewise.
23833 ("*tldo_lduh1_sp64"): Likewise.
23834 ("*tldo_lduh2_sp64"): Likewise.
23835 ("*tldo_lduw_sp32"): Likewise.
23836 ("*tldo_lduw_sp64"): Likewise.
23837 ("*tldo_lduw1_sp64"): Likewise.
23838 ("*tldo_ldx_sp64"): Likewise.
23839 ("*mov<VM32:mode>_insn"): Likewise.
23840 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23841 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23843 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23845 * config/sparc/sparc.md ("type"): New insn type viscmp.
23846 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23848 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23849 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23850 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23851 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23853 ("n7_vis_logical_11cycle"): Likewise.
23854 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23855 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23856 * config/sparc/niagara.md ("niag_vis"): Likewise.
23857 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23858 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23860 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23862 * config/sparc/sparc.md: New instruction type `bmask'.
23863 (bmaskdi_vis): Use the `bmask' type.
23864 (bmasksi_vis): Likewise.
23865 * config/sparc/ultra3.md (us3_array): Likewise.
23866 * config/sparc/niagara7.md (n7_array): Likewise.
23867 * config/sparc/niagara4.md (n4_array): Likewise.
23868 * config/sparc/niagara2.md (niag2_vis): Likewise.
23869 (niag3_vis): Likewise.
23870 * config/sparc/niagara.md (niag_vis): Likewise.
23872 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23874 * ipa-comdats.c: Remove optimize check from gate.
23875 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23876 for functions not optimized.
23877 (ipa_fn_summary_read): Skip optimize check.
23878 (ipa_fn_summary_write): Likewise.
23879 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23881 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23883 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23884 (check_callers): Give up on caller which is not optimized.
23885 (inline_small_functions): Likewise.
23886 (ipa_inline): Do not give up when not optimizing.
23887 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23888 away unoptimizes cdtors.
23889 (whole_program_function_and_variable_visibility): Do
23890 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23891 * ipa.c (process_references): Do not check optimize.
23892 (symbol_table::remove_unreachable_nodes): Update optimize check.
23893 (set_writeonly_bit): Update optimize check.
23894 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23895 (pass_ipa_single_use::gate): Remove.
23897 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23899 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23900 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23901 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23902 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23903 permute_load, permute_store, adjust_extract, adjust_splat,
23904 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23905 replace_swap_with_copy, dump_swap_insn_table,
23906 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23907 recombine_lvx_pattern, recombine_stvx_pattern,
23908 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23909 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23910 to file rs6000-p8swap.c.
23911 * config/rs6000/rs6000-p8swap.c: New file.
23912 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23913 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23914 and rs6000*-*-* targets.
23916 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23918 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23920 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23922 * lto-wrapper.c (merge_and_complain): Do not merge
23923 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23924 fsigned_zeros, ftrapping_math, fwrapv.
23925 (append_compiler_options): Do not track these options.
23926 (append_linker_options): Likewie
23928 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23930 * cgraphunit.c (cgraph_node::finalize_function): When
23931 !flag_toplevel_reorde set no_reorder flag.
23932 (varpool_node::finalize_decl): Likewise.
23933 (symbol_table::compile): Drop no toplevel reorder path.
23935 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23937 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23938 edges; zero probability is not better than uninitialized.
23940 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23942 * asan.h (asan_sanitize_allocas_p): Declare.
23943 * asan.c (asan_sanitize_allocas_p): New function.
23944 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23945 (handle_builtin_alloca): Likewise.
23946 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23947 if !asan_sanitize_allocas_p.
23948 * params.def (asan-instrument-allocas): Add new option.
23949 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23950 * opts.c (common_handle_option): Disable allocas sanitization for
23953 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23955 * asan.c: Include gimple-fold.h.
23956 (get_last_alloca_addr): New function.
23957 (handle_builtin_stackrestore): Likewise.
23958 (handle_builtin_alloca): Likewise.
23959 (asan_emit_allocas_unpoison): Likewise.
23960 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23961 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23962 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23963 (instrument_builtin_call): Pass gimple iterator to
23964 get_mem_refs_of_builtin_call.
23965 (last_alloca_addr): New global.
23966 * asan.h (asan_emit_allocas_unpoison): Declare.
23967 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23968 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23969 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23970 if function calls alloca.
23971 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23972 * gimple-fold.h (replace_call_with_value): Declare.
23973 * internal-fn.c: Include asan.h.
23974 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23975 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23977 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23979 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23980 (C_SELFTEST_FLAGS): New.
23981 (CPP_SELFTEST_FLAGS): New.
23982 (SELFTEST_DEPS): New, from deps of s-selftest.
23983 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23984 (CPP_SELFTEST_DEPS): New.
23985 (selftest): Add dependency on s-selftest-c++.
23986 (s-selftest): Rename to...
23987 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23988 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23989 than SELFTEST_FLAGS.
23990 (selftest-gdb): Rename to...
23991 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23993 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23994 (selftest-valgrind): Rename to...
23995 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23997 (selftest-valgrind): Reintroduce as an alias for
23998 selftest-c-valgrind.
23999 (s-selftest-c++): New.
24000 (selftest-c++-gdb): New.
24001 (selftest-c++-valgrind): New.
24003 2017-07-06 Olivier Hainque <hainque@adacore.com>
24005 * gcc.c (process_command): When deciding if undefined variables
24006 should be ignored when processing specs, accept "gcc -v" as well.
24008 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24010 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24011 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24013 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24015 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24016 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24017 * config/arm/arm-tables.opt: Regenerate.
24018 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24020 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24022 2017-07-06 Carl Love <cel@us.ibm.com>
24024 * ChangeLog: Clean up from mid air collision
24026 2017-07-06 Carl Love <cel@us.ibm.com>
24028 * config/rs6000/rs6000-c.c: Add support for built-in functions
24029 vector signed int vec_subc (vector signed int, vector signed int);
24030 vector signed __int128 vec_subc (vector signed __int128,
24031 vector signed __int128);
24032 vector unsigned __int128 vec_subc (vector unsigned __int128,
24033 vector unsigned __int128);
24034 vector signed int vec_sube (vector signed int, vector signed int,
24035 vector signed int);
24036 vector unsigned int vec_sube (vector unsigned int,
24037 vector unsigned int,
24038 vector unsigned int);
24039 vector signed __int128 vec_sube (vector signed __int128,
24040 vector signed __int128,
24041 vector signed__int128);
24042 vector unsigned __int128 vec_sube (vector unsigned __int128,
24043 vector unsigned __int128,
24044 vector unsigned __int128);
24045 vector signed int vec_subec (vector signed int, vector signed int,
24046 vector signed int);
24047 vector unsigned int vec_subec (vector unsigned int,
24048 vector unsigned int,
24049 vector unsigned int);
24050 vector signed __int128 vec_subec (vector signed __int128,
24051 vector signed __int128,
24052 vector signed__int128);
24053 vector unsigned __int128 vec_subec (vector unsigned __int128,
24054 vector unsigned __int128,
24055 vector unsigned __int128);
24056 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24057 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24058 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24059 BU_ALTIVEC_OVERLOAD_X definitions.
24060 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24061 * doc/extend.texi: Update the built-in documentation file for the new
24062 built-in functions.
24064 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24067 * diagnostic-show-locus.c (layout::layout): Use start and finish
24068 spelling location for the start and finish of each range.
24069 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24070 Add unused aspect param.
24071 * input.c (expand_location_1): Add "aspect" param, and use it
24072 to access the correct part of the location.
24073 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24075 (expand_location_to_spelling_point): Likewise.
24076 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24077 param, and pass it to expand_location_1.
24079 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24081 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24082 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24083 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24084 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24085 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24086 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24087 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24088 _mm_maskz_getmant_ss): New intrinsics.
24089 (__builtin_ia32_getexpss128_mask): Changed to ...
24090 __builtin_ia32_getexpss128_round ... this.
24091 (__builtin_ia32_getexpsd128_mask): Changed to ...
24092 __builtin_ia32_getexpsd128_round ... this.
24093 * config/i386/i386-builtin-types.def
24094 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24095 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24096 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24097 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24098 __builtin_ia32_getmantss_mask_round): New builtins.
24099 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24100 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24101 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24102 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24103 * config/i386/sse.md
24104 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24105 avx512f_sgetexp<mode><mask_scalar_name>
24106 <round_saeonly_scalar_name> ... this.
24107 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24108 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24109 vgetexp<ssescalarmodesuffix>
24110 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24111 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24112 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24113 avx512f_vgetmant<mode><mask_scalar_name>
24114 <round_saeonly_scalar_name> ... this.
24115 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24116 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24117 vgetmant<ssescalarmodesuffix>
24118 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24119 %0<mask_scalar_operand4>, %1, %2
24120 <round_saeonly_scalar_mask_op4>, %3} ... this.
24121 * config/i386/subst.md (mask_scalar_operand4,
24122 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24123 round_saeonly_scalar_nimm_predicate): New subst attributes.
24125 2017-07-06 Julia Koval <julia.koval@intel.com>
24127 * config/i386/i386.c (ix86_erase_embedded_rounding):
24128 Remove code for old rounding pattern.
24130 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24132 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24134 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24136 * doc/sourcebuild.texi (Test Directives, Variants of
24137 dg-require-support): Add documentation for dg-require-stack-check.
24139 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24141 * config/i386/subst.md (mask_scalar, round_scalar,
24142 round_saeonly_scalar): New meta-templates.
24143 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24144 round_scalar_mask_operand3, round_scalar_mask_op3,
24145 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24146 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24147 round_saeonly_scalar_constraint,
24148 round_saeonly_scalar_prefix): New subst attribute.
24149 * config/i386/sse.md
24150 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24151 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24152 <round_scalar_name> ... this.
24153 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24154 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24155 <round_scalar_name> ... this.
24156 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24157 <sse>_vm<code><mode>3<mask_scalar_name>
24158 <round_saeonly_scalar_name> ... this.
24159 (v<plusminus_mnemonic><ssescalarmodesuffix>
24160 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24161 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24162 v<plusminus_mnemonic><ssescalarmodesuffix>
24163 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24164 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24165 (v<multdiv_mnemonic><ssescalarmodesuffix>
24166 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24167 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24168 v<multdiv_mnemonic><ssescalarmodesuffix>
24169 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24170 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24171 (v<maxmin_float><ssescalarmodesuffix>
24172 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24173 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24174 v<maxmin_float><ssescalarmodesuffix>
24175 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24176 %0<mask_scalar_operand3>, %1, %<iptr>2
24177 <round_saeonly_scalar_mask_op3>} ... this.
24179 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24181 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24182 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24184 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24185 Alan Hayward <alan.hayward@arm.com>
24186 David Sherwood <david.sherwood@arm.com>
24188 * combine.c (simplify_if_then_else): Remove "enum" before
24190 * compare-elim.c (can_eliminate_compare): Likewise.
24191 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24193 (aarch64_lookup_simd_builtin_type): Likewise.
24194 (aarch64_simd_builtin_type): Likewise.
24195 (aarch64_init_simd_builtin_types): Likewise.
24196 (aarch64_simd_expand_args): Likewise.
24197 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24199 (aarch64_reverse_mask): Likewise.
24200 (aarch64_simd_emit_reg_reg_move): Likewise.
24201 (aarch64_gen_adjusted_ldpstp): Likewise.
24202 (aarch64_ccmp_mode_to_code): Likewise.
24203 (aarch64_operands_ok_for_ldpstp): Likewise.
24204 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24205 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24207 (aarch64_min_divisions_for_recip_mul): Likewise.
24208 (aarch64_reassociation_width): Likewise.
24209 (aarch64_get_condition_code_1): Likewise.
24210 (aarch64_simd_emit_reg_reg_move): Likewise.
24211 (aarch64_simd_attr_length_rglist): Likewise.
24212 (aarch64_reverse_mask): Likewise.
24213 (aarch64_operands_ok_for_ldpstp): Likewise.
24214 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24215 (aarch64_gen_adjusted_ldpstp): Likewise.
24216 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24218 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24219 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24220 (arm_lookup_simd_builtin_type): Likewise.
24221 (arm_simd_builtin_type): Likewise.
24222 (arm_init_simd_builtin_types): Likewise.
24223 (arm_expand_builtin_args): Likewise.
24224 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24225 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24226 (ft32_setup_incoming_varargs): Likewise.
24227 (ft32_function_arg): Likewise.
24228 (ft32_function_arg_advance): Likewise.
24229 (ft32_pass_by_reference): Likewise.
24230 (ft32_arg_partial_bytes): Likewise.
24231 (ft32_valid_pointer_mode): Likewise.
24232 (ft32_addr_space_pointer_mode): Likewise.
24233 (ft32_addr_space_legitimate_address_p): Likewise.
24234 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24236 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24237 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24238 (iamcu_alignment): Likewise.
24239 (canonicalize_vector_int_perm): Likewise.
24240 (ix86_noce_conversion_profitable_p): Likewise.
24241 (ix86_mpx_bound_mode): Likewise.
24242 (ix86_operands_ok_for_move_multiple): Likewise.
24243 * config/microblaze/microblaze-protos.h
24244 (microblaze_expand_conditional_branch_reg): Likewise.
24245 * config/microblaze/microblaze.c
24246 (microblaze_expand_conditional_branch_reg): Likewise.
24247 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24249 (rs6000_reassociation_width): Likewise.
24250 (rs6000_invalid_binary_op): Likewise.
24251 (fusion_p9_p): Likewise.
24252 (emit_fusion_p9_load): Likewise.
24253 (emit_fusion_p9_store): Likewise.
24254 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24256 (riscv_hard_regno_mode_ok_p): Likewise.
24257 (riscv_address_insns): Likewise.
24258 (riscv_split_symbol): Likewise.
24259 (riscv_legitimize_move): Likewise.
24260 (riscv_function_value): Likewise.
24261 (riscv_hard_regno_nregs): Likewise.
24262 (riscv_expand_builtin): Likewise.
24263 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24264 (riscv_build_integer): Likewise.
24265 (riscv_split_integer): Likewise.
24266 (riscv_legitimate_constant_p): Likewise.
24267 (riscv_cannot_force_const_mem): Likewise.
24268 (riscv_regno_mode_ok_for_base_p): Likewise.
24269 (riscv_valid_base_register_p): Likewise.
24270 (riscv_valid_offset_p): Likewise.
24271 (riscv_valid_lo_sum_p): Likewise.
24272 (riscv_classify_address): Likewise.
24273 (riscv_legitimate_address_p): Likewise.
24274 (riscv_address_insns): Likewise.
24275 (riscv_load_store_insns): Likewise.
24276 (riscv_force_binary): Likewise.
24277 (riscv_split_symbol): Likewise.
24278 (riscv_force_address): Likewise.
24279 (riscv_legitimize_address): Likewise.
24280 (riscv_move_integer): Likewise.
24281 (riscv_legitimize_const_move): Likewise.
24282 (riscv_legitimize_move): Likewise.
24283 (riscv_address_cost): Likewise.
24284 (riscv_subword): Likewise.
24285 (riscv_output_move): Likewise.
24286 (riscv_canonicalize_int_order_test): Likewise.
24287 (riscv_emit_int_order_test): Likewise.
24288 (riscv_function_arg_boundary): Likewise.
24289 (riscv_pass_mode_in_fpr_p): Likewise.
24290 (riscv_pass_fpr_single): Likewise.
24291 (riscv_pass_fpr_pair): Likewise.
24292 (riscv_get_arg_info): Likewise.
24293 (riscv_function_arg): Likewise.
24294 (riscv_function_arg_advance): Likewise.
24295 (riscv_arg_partial_bytes): Likewise.
24296 (riscv_function_value): Likewise.
24297 (riscv_pass_by_reference): Likewise.
24298 (riscv_setup_incoming_varargs): Likewise.
24299 (riscv_print_operand): Likewise.
24300 (riscv_elf_select_rtx_section): Likewise.
24301 (riscv_save_restore_reg): Likewise.
24302 (riscv_for_each_saved_reg): Likewise.
24303 (riscv_register_move_cost): Likewise.
24304 (riscv_hard_regno_mode_ok_p): Likewise.
24305 (riscv_hard_regno_nregs): Likewise.
24306 (riscv_class_max_nregs): Likewise.
24307 (riscv_memory_move_cost): Likewise.
24308 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24309 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24310 (rl78_addr_space_address_mode): Likewise.
24311 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24313 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24314 (rs6000_reassociation_width): Likewise.
24315 (rs6000_invalid_binary_op): Likewise.
24316 (fusion_p9_p): Likewise.
24317 (emit_fusion_p9_load): Likewise.
24318 (emit_fusion_p9_store): Likewise.
24319 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24320 (ok_for_simple_move_operands): Likewise.
24321 (ok_for_simple_move_strict_operands): Likewise.
24322 (ok_for_simple_arith_logic_operands): Likewise.
24323 (visium_legitimize_reload_address): Likewise.
24324 (visium_select_cc_mode): Likewise.
24325 (output_cbranch): Likewise.
24326 (visium_split_double_move): Likewise.
24327 (visium_expand_copysign): Likewise.
24328 (visium_expand_int_cstore): Likewise.
24329 (visium_expand_fp_cstore): Likewise.
24330 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24331 (visium_function_arg): Likewise.
24332 (visium_function_arg_advance): Likewise.
24333 (visium_libcall_value): Likewise.
24334 (visium_setup_incoming_varargs): Likewise.
24335 (visium_legitimate_constant_p): Likewise.
24336 (visium_legitimate_address_p): Likewise.
24337 (visium_legitimize_address): Likewise.
24338 (visium_secondary_reload): Likewise.
24339 (visium_register_move_cost): Likewise.
24340 (visium_memory_move_cost): Likewise.
24341 (prepare_move_operands): Likewise.
24342 (ok_for_simple_move_operands): Likewise.
24343 (ok_for_simple_move_strict_operands): Likewise.
24344 (ok_for_simple_arith_logic_operands): Likewise.
24345 (visium_function_value_1): Likewise.
24346 (rtx_ok_for_offset_p): Likewise.
24347 (visium_legitimize_reload_address): Likewise.
24348 (visium_split_double_move): Likewise.
24349 (visium_expand_copysign): Likewise.
24350 (visium_expand_int_cstore): Likewise.
24351 (visium_expand_fp_cstore): Likewise.
24352 (visium_split_cstore): Likewise.
24353 (visium_select_cc_mode): Likewise.
24354 (visium_split_cbranch): Likewise.
24355 (output_cbranch): Likewise.
24356 (visium_print_operand_address): Likewise.
24357 * expmed.c (flip_storage_order): Likewise.
24358 * expmed.h (emit_cstore): Likewise.
24359 (flip_storage_order): Likewise.
24360 * genrecog.c (validate_pattern): Likewise.
24361 * hsa-gen.c (gen_hsa_addr): Likewise.
24362 * internal-fn.c (expand_arith_overflow): Likewise.
24363 * ira-color.c (allocno_copy_cost_saving): Likewise.
24364 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24365 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24366 (process_invariant_for_inheritance): Likewise.
24367 * lra-eliminations.c (move_plus_up): Likewise.
24368 * omp-low.c (lower_oacc_reductions): Likewise.
24369 * simplify-rtx.c (simplify_subreg): Likewise.
24370 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24371 (TARGET_CHKP_BOUND_MODE): Likewise..
24372 * targhooks.c (default_chkp_bound_mode): Likewise.
24373 (default_setup_incoming_vararg_bounds): Likewise.
24374 * targhooks.h (default_chkp_bound_mode): Likewise.
24375 (default_setup_incoming_vararg_bounds): Likewise.
24376 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24377 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24378 (have_whole_vector_shift): Likewise.
24379 * tree-vect-stmts.c (vectorizable_load): Likewise.
24380 * doc/tm.texi: Regenerate.
24382 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24384 Graceful degrade if Binutils PR21472 is not available.
24387 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24388 .rodata in flash test fails.
24389 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24390 * confgure: Regenerate.
24391 * config.in: Regenerate.
24392 * config/avr/avr.c (avr_asm_named_section)
24393 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24394 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24395 (avr_asm_init_sections): Same.
24397 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24399 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24400 (fma<VH:mode>4_intrinsic): Likewise.
24401 (*fmsub<VCVTF:mode>4): Likewise.
24402 (*fmsub<VH:mode>4_intrinsic): Likewise.
24404 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24407 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24408 Don't depend on "optimize > 0".
24409 (out_movhi_r_mr, out_movqi_mr_r): Same.
24410 (out_movhi_mr_r, out_movqi_r_mr): Same.
24411 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24412 io_address_operand on "optimize > 0".
24414 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24416 * tree-loop-distribution.c: Add general explanantion on the pass.
24417 (generate_loops_for_partition): Mark distributed loop.
24418 (pg_add_dependence_edges): New parameter. Handle alias data
24419 dependence specially and record it in the parameter if asked.
24420 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24421 (init_partition_graph_vertices, add_partition_graph_edge): New.
24422 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24423 (free_partition_graph_vdata, build_partition_graph): New.
24424 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24425 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24426 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24427 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24428 (version_for_distribution_p, finalize_partitions): New.
24429 (distribute_loop): Handle alias data dependence specially. Factor
24430 out loop fusion code as functions and call these functions.
24432 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24434 * tree-loop-distribution.c (classify_partition): New parameter and
24435 better handle reduction statement.
24436 (rdg_build_partitions): Revise comment.
24437 (distribute_loop): Compute statements in all partitions and pass it
24438 to classify_partition.
24440 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24442 * tree-loop-distribution.c (enum partition_type): New.
24443 (struct partition): New field type.
24444 (partition_merge_into): Add parameter. Update partition type.
24445 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24446 (build_rdg_partition_for_vertex): Compute partition type.
24447 (rdg_build_partitions): Dump partition type.
24448 (distribute_loop): Update calls to partition_merge_into.
24450 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24452 * tree-loop-distribution.c (struct ddr_hasher): New.
24453 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24455 (classify_partition): Call get_data_dependence.
24456 (pg_add_dependence_edges): Ditto.
24457 (distribute_loop): Release data dependence hash table.
24459 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24461 * tree-loop-distribution.c (ref_base_address): Delete.
24462 (similar_memory_accesses): Rename ...
24463 (share_memory_accesses): ... to this. Check if partitions access
24464 the same memory reference.
24465 (distribute_loop): Call share_memory_accesses.
24467 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24469 * tree-loop-distribution.c (struct partition): New field recording
24470 its data reference.
24471 (partition_alloc, partition_free): Init and release data refs.
24472 (partition_merge_into): Merge data refs.
24473 (build_rdg_partition_for_vertex): Collect data refs for partition.
24474 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24476 (distribute_loop): Remve data refs from vertice data of partition
24479 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24481 * tree-loop-distribution.c (params.h): Include header file.
24482 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24483 (datarefs_vec): New global var.
24484 (create_rdg_vertices): Use datarefs_vec directly.
24485 (free_rdg): Don't free data references.
24486 (build_rdg): Update use. Don't free data references.
24487 (distribute_loop): Compute global variable for data references.
24488 Bail out if there are too many data references.
24490 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24492 * tree-loop-distribution.c (loop_nest): New global var.
24493 (build_rdg): Use loop directly, rather than loop nest.
24494 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24496 (distribute_loop): Compute global variable loop nest. Update use.
24498 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24500 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24501 (partition_merge_into): New parameter. Dump reason for fusion.
24502 (distribute_loop): Update use of partition_merge_into.
24504 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24506 * tree-loop-distribution.c (bb_top_order_index): New.
24507 (bb_top_order_index_size, bb_top_order_cmp): New.
24508 (stmts_from_loop): Use topological order.
24509 (pass_loop_distribution::execute): Compute and release topological
24510 order for basic blocks.
24512 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24514 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24517 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24519 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24520 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24521 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24522 * internal-fn.def (LOOP_DIST_ALIAS): New.
24523 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24524 (fold_loop_internal_call): ... this.
24525 (vect_loop_dist_alias_call): New function.
24526 (set_uid_loop_bbs): Call fold_loop_internal_call.
24527 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24530 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24533 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24534 Require dead FLAGS_REG at the beginning of a peephole.
24536 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24539 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24540 arguments in the call to __builtin_ia32_sbb_u32.
24541 (_subborrow_u64): Swap _X and _Y arguments in the call to
24542 __builtin_ia32_sbb_u64.
24544 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24547 * tree-vrp.c (compare_assert_loc): Turn into a function template
24548 with stable template parameter. Only test if a->e is NULL,
24549 !a->e == !b->e has been verified already. Use e == NULL or
24550 e != NULL instead of e or ! e tests. If stable is true, don't use
24551 iterative_hash_expr, on the other side allow a or b or both NULL
24552 and sort the NULLs last.
24553 (process_assert_insertions): Sort using compare_assert_loc<false>
24554 instead of compare_assert_loc, later sort using
24555 compare_assert_loc<true> before calling process_assert_insertions_for
24556 in a loop. Use break instead of continue once seen NULL pointer.
24558 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24560 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24561 Cortex-R7 and Cortex-R8 processors.
24563 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24565 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24566 uninitialized while src is not.
24568 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24570 * common/config/arm/arm-common.c: Adjust include path for
24572 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24573 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24574 (arm-cpu-data.h): Likewise.
24575 (arm-cpu-cdata.h): Likewise.
24576 * config/arm/arm-cpu.h: Delete.
24577 * config/arm/arm-cpu-cdata.h: Delete.
24578 * config/arm/arm-cpu-data.h: Delete.
24580 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24582 * config/arm/arm-cpus.in (cortex-a55): New.
24583 (cortex-a75): Likewise.
24584 (cortex-a75.cortex-a55): Likewise.
24585 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24587 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24588 * config/arm/arm-cpu-cdata.h: Regenerate.
24589 * config/arm/arm-cpu-data.h: Regenerate.
24590 * config/arm/arm-cpu.h: Regenerate.
24591 * config/arm/arm-tables.opt: Regenerate.
24592 * config/arm/arm-tune.md: Regenerate.
24594 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24596 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24598 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24600 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24603 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24605 PR tree-optimization/81292
24606 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24607 full_string_p, also call adjust_related_strinfos if the adjustment
24608 is simple, otherwise invalidate related strinfos.
24610 2017-07-04 Martin Liska <mliska@suse.cz>
24613 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24614 newly created variable as DECL_IGNORED_P.
24616 2017-07-04 Martin Liska <mliska@suse.cz>
24619 * ipa-inline.c (inline_small_functions):
24620 Use xstrdup_for_dump.
24622 2017-07-04 Tom de Vries <tom@codesourcery.com>
24624 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24626 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24629 * config/darwin.c (darwin_function_switched_text_sections):
24632 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24634 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24636 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24638 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24640 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24642 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24643 min_profitable_iters, and th as inclusive lower bounds.
24644 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24645 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24646 for min_profitable_iters and min_profitable_estimate.
24647 (vect_transform_loop): Treat th as an inclusive lower bound.
24648 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24650 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24653 * config/darwin.c (darwin_function_switched_text_sections):
24654 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24655 in two pieces, and suppress the use of buf.
24657 2017-07-03 Nathan Sidwell <nathan@acm.org>
24659 * hash-table.h (hash_table_mod1): Fix indentation.
24661 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24663 PR middle-end/81290
24664 * predict.c (force_edge_cold): Be more careful about propagation
24666 * profile-count.h (profile_probability::guessed,
24667 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24669 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24671 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24673 * doc/invoke.texi (rcpc architecture extension): Document it.
24675 2017-07-03 Richard Biener <rguenther@suse.de>
24677 PR tree-optimization/60510
24678 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24679 the scalar reduction PHI and use it.
24680 (vectorizable_reduction): Properly guard the single_defuse_cycle
24681 path for non-SLP reduction chains where we cannot use it.
24682 Rework reduc_def/index and vector type deduction. Rework
24683 vector operand gathering during reduction op code-gen.
24684 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24685 chains dissolve the chain and leave it to non-SLP reduction
24688 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24690 * tree-data-ref.h (dr_alignment): Declare.
24691 * tree-data-ref.c (dr_alignment): New function.
24692 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24693 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24695 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24697 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24699 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24700 and base_misalignment fields.
24701 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24702 * tree-data-ref.c: Include builtins.h.
24703 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24704 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24705 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24706 * tree-vect-data-refs.c: Include tree-cfg.h.
24707 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24708 fields instead of calculating an alignment here.
24709 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24710 innermost_loop_behavior fields.
24712 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24714 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24716 (DR_STEP_ALIGNMENT): New macro.
24717 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24718 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24719 (create_data_ref): Print it.
24720 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24721 to tell whether the step preserves vector (mis)alignment.
24722 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24723 Move the check for an integer step and generalise to all INTEGER_CST.
24724 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24725 Print the outer step alignment.
24727 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24729 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24730 with offset_alignment.
24731 (DR_ALIGNED_TO): Delete.
24732 (DR_OFFSET_ALIGNMENT): New macro.
24733 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24734 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24735 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24736 (create_data_ref): Likewise.
24737 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24738 (vect_analyze_data_refs): Likewise.
24739 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24740 creating dummy innermost behavior.
24742 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24744 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24745 with a "innermost_loop_behavior *" and refeence tree.
24746 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24747 (create_data_ref): Update call accordingly.
24748 * tree-predcom.c (find_looparound_phi): Likewise.
24750 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24752 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24753 fields with dr_wrt_vec_loop.
24754 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24755 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24756 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24757 (vect_dr_behavior): New function.
24758 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24759 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24760 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24761 track whether the step preserves the misalignment.
24762 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24763 Use vect_dr_behavior.
24764 (vect_setup_realignment): Update call accordingly.
24765 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24766 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24767 call to vect_create_addr_base_for_vector_ref.
24768 (vect_create_cond_for_align_checks): Likewise.
24769 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24770 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24771 (vect_recog_mask_conversion_pattern): Likewise.
24772 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24773 (new_stmt_vec_info): Remove redundant zeroing.
24775 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24777 * common/config/arm/arm-common.c (arm_be8_option): New function.
24778 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24779 (ISA_ARMv6): Add isa_bit_be8.
24780 * config/arm/arm.h (arm_be8_option): Add prototype.
24781 (BE8_SPEC_FUNCTION): New define.
24782 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24783 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24784 (mlittle-endian): Similarly.
24785 (mbe8, mbe32): New options.
24786 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24787 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24789 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24791 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24793 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24795 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24796 (cleanup_tree_cfg_bb): Use it.
24797 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24799 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24801 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24804 * loop-doloop.c (add_test): Update profile.
24806 2017-07-03 Martin Liska <mliska@suse.cz>
24809 * sanopt.c (rewrite_usage_of_param): New function.
24810 (sanitize_rewrite_addressable_params): Likewise.
24811 (pass_sanopt::execute): Call rewrite_usage_of_param.
24813 2017-07-03 Richard Biener <rguenther@suse.de>
24815 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24816 back to using VIEW_CONVERT_EXPR.
24818 2017-07-03 Martin Liska <mliska@suse.cz>
24821 * doc/extend.texi: Document when a resolver function is
24822 generated for target_clones.
24824 2017-07-03 Martin Liska <mliska@suse.cz>
24826 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24827 and shadow memory of auto variables which are subject of
24828 use-after-scope sanitization.
24829 (asan_expand_mark_ifn): Add do set only when is_poison.
24831 2016-07-03 Richard Biener <rguenther@suse.de>
24833 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24835 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24836 (vectorizable_reduction): Perform reduction PHI creation when
24837 visiting a reduction PHI and adjust and simplify code generation
24838 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24839 (vect_transform_loop): Visit reduction PHIs.
24840 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24841 defs into the SLP tree.
24842 (vect_build_slp_tree): Reduction defs terminate the recursion.
24843 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24845 (vect_get_vec_defs_for_stmt_copy): Export.
24846 (vect_get_vec_defs): Likewise.
24847 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24849 (vect_get_vec_defs_for_stmt_copy): Declare.
24850 (vect_get_vec_defs): Likewise.
24852 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24854 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24855 parameter with a "loop" parameter and use it instead of the
24856 loop containing DR_STMT. Don't check simple_iv when doing
24857 BB analysis. Describe the two analysis modes in the comment.
24859 2017-07-03 Tom de Vries <tom@codesourcery.com>
24861 PR tree-optimization/69468
24862 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24863 (find_same_succ_bb): Handle ignore_edge_flags.
24865 2017-07-03 Tom de Vries <tom@codesourcery.com>
24867 PR tree-optimization/81192
24868 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24870 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24872 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24874 2017-07-03 Tom de Vries <tom@codesourcery.com>
24876 PR tree-optimization/81192
24877 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24878 BB_SAME_SUCC (bb) == NULL.
24880 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24882 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24885 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24887 * dumpfile.c: Include profile-count.h
24888 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24890 (insert_cond_bb): Update profile.
24891 * tree-cfg.h (insert_cond_bb): Update prototype.
24892 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24893 * tree-dump.c: Do not include tree-cfg.
24895 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24897 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24899 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24901 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24904 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24906 * tree-complex.c (expand_complex_div_wide): update profile.
24908 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24909 Alan Hayward <alan.hayward@arm.com>
24910 David Sherwood <david.sherwood@arm.com>
24912 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24913 (CORETYPES_H): New define.
24914 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24915 (insn-modes-inline.h, s-modes-inline-h): New rules.
24916 (generated_files): Add insn-modes-inline.h.
24917 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24918 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24919 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24920 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24921 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24922 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24923 (build/gencodes.o, build/genconditions.o): Likewise.
24924 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24925 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24926 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24927 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24928 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24929 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24930 * coretypes.h: Include everything up to real.h for generators.
24931 Include insn-modes.h first. Include wide-int-print.h after
24932 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24933 * machmode.h: Don't include insn-modes.h here.
24934 * function-tests.c: Remove includes of signop.h, machmode.h,
24935 double-int.h and wide-int.h.
24937 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24939 * optc-save-gen.awk: Likewise.
24940 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24941 * godump.c: Remove include of wide-int-print.h.
24942 * pretty-print.h: Likewise.
24943 * wide-int-print.cc: Likewise.
24944 * wide-int.cc: Likewise.
24945 * hash-map-tests.c: Remove include of signop.h.
24946 * hash-set-tests.c: Likewise.
24947 * rtl-tests.c: Likewise.
24948 * mkconfig.sh: Remove include of machmode.h.
24949 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24951 (emit_insn_modes_inline_h): ...this new function. Emit the code
24952 into an insn-modes-inline.h header file, adding appropriate
24953 include guards and end comments.
24954 (emit_insn_modes_c_header): Remove include of machmode.h.
24955 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24957 (main): Handle -i flag and call emit_insn_modes_inline_h when
24960 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24962 * tree-ssa-strlen.c (strinfo): Rename the length field to
24963 nonzero_chars. Add a full_string_p field.
24964 (compare_nonzero_chars, zero_length_string_p): New functions.
24965 (get_addr_stridx): Add an offset_out parameter.
24966 Use compare_nonzero_chars.
24967 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24968 (new_strinfo): Update after above changes to strinfo.
24969 (set_endptr_and_length): Set full_string_p.
24970 (get_string_length): Update after above changes to strinfo.
24971 (unshare_strinfo): Update call to new_strinfo.
24972 (maybe_invalidate): Likewise.
24973 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24974 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24975 as a uhwi instead of an shwi. Update after above changes to
24976 strinfo and new_strinfo.
24977 (zero_length_string): Assert that chainsi contains full strings.
24978 Use zero_length_string_p. Update call to new_strinfo.
24979 (adjust_related_strinfos): Update after above changes to strinfo.
24980 Copy full_string_p from origsi.
24981 (adjust_last_stmt): Use zero_length_string_p.
24982 (handle_builtin_strlen): Update after above changes to strinfo and
24983 new_strinfo. Install the lhs as the string length if the previous
24984 entry didn't describe a full string.
24985 (handle_builtin_strchr): Update after above changes to strinfo
24987 (handle_builtin_strcpy): Likewise.
24988 (handle_builtin_strcat): Likewise.
24989 (handle_builtin_malloc): Likewise.
24990 (handle_pointer_plus): Likewise.
24991 (handle_builtin_memcpy): Likewise. Track nonzero characters
24992 that aren't necessarily followed by a nul terminator.
24993 (handle_char_store): Likewise.
24995 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24997 PR tree-optimization/80769
24998 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24999 for malloc and calloc. Document the new invariant that all related
25000 strinfos have delayed lengths or none do.
25001 (verify_related_strinfos): Move earlier in file.
25002 (set_endptr_and_length): New function, split out from...
25003 (get_string_length): ...here. Also set the lengths of related
25005 (zero_length_string): Assert that chainsi has known (rather than
25007 (adjust_related_strinfos): Likewise.
25009 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25011 PR tree-optimization/81136
25012 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25013 assert that two references with the same misalignment have the same
25014 compile-time misalignment if those compile-time misalignments
25017 2017-07-01 Andi Kleen <ak@linux.intel.com>
25019 * print-tree.c (print_node): Print all attributes.
25021 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25023 * cfg.c (scale_bbs_frequencies): New function.
25024 * cfg.h (scale_bbs_frequencies): Declare it.
25025 * cfgloopanal.c (single_likely_exit): Cleanup.
25026 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25028 (scale_loop_profile): Likewise.
25029 (loop_version): Likewise.
25030 (create_empty_loop_on_edge): Update.
25031 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25032 scale_loop_frequencies, scale_loop_profile, loopify,
25033 loop_version): Update prototypes.
25034 * modulo-sched.c (sms_schedule): Update.
25035 * predict.c (unlikely_executed_edge_p): Also check probability.
25036 (probably_never_executed_edge_p): Fix typo.
25037 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25038 * tree-parloops.c (gen_parallel_loop): Update.
25039 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25040 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25041 * tree-ssa-loop-split.c (split_loop): Update.
25042 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25043 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25044 (vect_loop_versioning): Update.
25045 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25047 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25049 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25051 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25053 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25054 to keep profile consistent.
25056 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25058 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25059 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25060 * profile-count.h (max_safe_multiplier): Make unsigned.
25061 (profile_count::guessed_zero): New.
25063 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25065 * bb-reorder.c (fix_up_crossing_landing_pad,
25066 fix_crossing_conditional_branches): Use make_single_succ_edge
25067 to keep profile consistent.
25069 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25071 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25074 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25077 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25078 the right scopes, make sure cond_jump isn't preserved between multiple
25079 iterations. Search for fallthru edge whenever there are 3+ edges and
25080 use find_fallthru_edge for it.
25082 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25084 Patch by Alexander Monakov <amonakov@ispras.ru>
25085 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25086 probabilities consistently.
25088 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25090 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25091 emit_cmp_and_jump_insns.
25093 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25096 * tree-inline.c (expand_call_inline): Combine profile statuses.
25098 2017-06-30 Andrew Pinski <apinski@cavium.com>
25100 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25101 fold_stmt returned true.
25103 2017-06-30 Nathan Sidwell <nathan@acm.org>
25105 * ggc.h (empty_string): Delete.
25106 * cfgexpand.c (expand_asm_stmt): Use plain "".
25107 * optabs.c (expand_asm_memory_barrier): Likewise.
25108 * stringpool.c (empty_string): Delete.
25109 (digit_vector, digit_string): Delete.
25110 (ggc_alloc_string): Use plain "", don't optimize single digit
25111 strings. Use ggc_alloc_atomic.
25113 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25115 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25116 comparison set and one other set, use the cost of the non-comparison
25119 2017-06-30 Nathan Sidwell <nathan@acm.org>
25121 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25124 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25126 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25127 loops. Remove now unneeded calls to gimple_switch_set_label() that
25128 just set removed labels to NULL_TREE.
25130 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25132 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25133 (set_range_info): ...here. Only call set_range_info_raw if domain
25135 (set_nonzero_bits): Call set_range_info_raw.
25136 * tree-ssanames.h (set_range_info_raw): New.
25138 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25141 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25142 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25143 of nonimmediate_operand and <store_mask_constraint> instead of m
25144 for the input operand. For V8FI iterator, always split if input
25145 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25146 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25147 <store_mask_predicate> instead of register_operand and
25148 <store_mask_constraint> instead of v for the input operand. Make
25149 sure both operands aren't MEMs for if not <mask_applied>.
25151 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25153 * lto-wrapper.c (copy_file) Close both file descriptors before
25156 2017-06-30 Martin Liska <mliska@suse.cz>
25159 * multiple_target.c (create_dispatcher_calls): Make ifunc
25160 also for function that don't have calls or are not referenced.
25162 2017-06-30 Richard Biener <rguenther@suse.de>
25164 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25165 analyze the first scalar stmt. Move vector type computation
25166 for the BB case here from ...
25167 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25168 live operation processing in the SLP case properly.
25170 2017-06-30 Richard Biener <rguenther@suse.de>
25172 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25174 2017-06-30 Martin Liska <mliska@suse.cz>
25177 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25178 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25180 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25182 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25185 2017-06-30 Richard Biener <rguenther@suse.de>
25187 PR tree-optimization/81249
25188 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25189 condition reduction result to original scalar type.
25191 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25193 * profile-count.h (enum profile_quality): Fix typos and whitespace
25196 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25198 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25199 type for branch probabilities.
25201 2017-06-29 Julian Brown <julian@codesourcery.com>
25202 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25204 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25205 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25206 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25207 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25209 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25211 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25212 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25213 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25214 CC usage from generic code to here.
25215 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25216 CC usage into the target macros.
25218 2017-06-29 Maya Rashish <coypu@sdf.org>
25220 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25223 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25225 * arm/arm-builtins.c: Include profile-count.h
25226 * except.c (sjlj_emit_function_enter): Use
25227 profile_probability::unlikely.
25229 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25231 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25232 and tocrel_offset be pointer args rather than implicitly using
25234 (legitimate_constant_pool_address_p, rs6000_emit_move,
25235 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25236 tocrel_offset and use in toc_relative_expr_p call.
25237 (print_operand, print_operand_address): Use static tocrel_base_oac
25238 and tocrel_offset_oac.
25239 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25242 2017-06-29 Maya Rashish <coypu@sdf.org>
25244 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25246 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25248 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25249 objects, take into account only the alignment of 'op0' and 'mode1' if
25252 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25254 * ccmp.c (ccmp_tree_comparison_p): New function.
25255 (ccmp_candidate_p): Update to use above function.
25256 (get_compare_parts): New function.
25257 (expand_ccmp_next): Update to use new functions.
25258 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25260 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25261 take mode as argument.
25262 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25263 * expr.c (expand_expr_real_1): Pass mode as argument.
25265 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25267 * combine.c (combine_instructions): Print insns to dump_file, together
25270 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25272 * asan.c (asan_emit_stack_protection): Update.
25273 (create_cond_insert_point): Update.
25274 * auto-profile.c (afdo_propagate_circuit): Update.
25275 * basic-block.h (struct edge_def): Turn probability to
25276 profile_probability.
25277 (EDGE_FREQUENCY): Update.
25278 * bb-reorder.c (find_traces_1_round): Update.
25279 (better_edge_p): Update.
25280 (sanitize_hot_paths): Update.
25281 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25282 (make_single_succ_edge): Update.
25283 (check_bb_profile): Update.
25284 (dump_edge_info): Update.
25285 (update_bb_profile_for_threading): Update.
25286 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25287 probabilitycount to 0.
25288 * cfgbuild.c (compute_outgoing_frequencies): Update.
25289 * cfgcleanup.c (try_forward_edges): Update.
25290 (outgoing_edges_match): Update.
25291 (try_crossjump_to_edge): Update.
25292 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25293 (expand_gimple_tailcall): Update.
25294 (construct_init_block): Use make_single_succ_edge.
25295 (construct_exit_block): Use make_single_succ_edge.
25296 * cfghooks.c (verify_flow_info): Update.
25297 (redirect_edge_succ_nodup): Update.
25298 (split_edge): Update.
25299 (account_profile_record): Update.
25300 * cfgloopanal.c (single_likely_exit): Update.
25301 * cfgloopmanip.c (scale_loop_profile): Update.
25302 (set_zero_probability): Remove.
25303 (duplicate_loop_to_header_edge): Update.
25304 * cfgloopmanip.h (loop_version): Update prototype.
25305 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25306 (force_nonfallthru_and_redirect): Update.
25307 (update_br_prob_note): Update.
25308 (rtl_verify_edges): Update.
25309 (purge_dead_edges): Update.
25310 (rtl_lv_add_condition_to_bb): Update.
25311 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25312 * cgraphunit.c (init_lowered_empty_function): Update.
25313 (cgraph_node::expand_thunk): Update.
25314 * cilk-common.c: Include profile-count.h
25315 * dojump.c (inv): Remove.
25316 (jumpifnot): Update.
25317 (jumpifnot_1): Update.
25318 (do_jump_1): Update.
25320 (do_jump_by_parts_greater_rtx): Update.
25321 (do_compare_rtx_and_jump): Update.
25322 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25323 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25324 * dwarf2cfi.c: Include profile-count.h
25325 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25326 (sjlj_emit_dispatch_table): Likewise.
25327 * explow.c: Include profile-count.h
25328 * expmed.c (emit_store_flag_force): Update.
25329 (do_cmp_and_jump): Update.
25330 * expr.c (compare_by_pieces_d::generate): Update.
25331 (compare_by_pieces_d::finish_mode): Update.
25332 (emit_block_move_via_loop): Update.
25333 (store_expr_with_bounds): Update.
25334 (store_constructor): Update.
25335 (expand_expr_real_2): Update.
25336 (expand_expr_real_1): Update.
25337 * expr.h (try_casesi, try_tablejump): Update prototypes.
25338 * gimple-pretty-print.c (dump_probability): Update.
25339 (dump_profile): New.
25340 (dump_gimple_label): Update.
25341 (dump_gimple_bb_header): Update.
25342 * graph.c (draw_cfg_node_succ_edges): Update.
25343 * hsa-gen.c (convert_switch_statements): Update.
25344 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25345 (find_if_case_1): Update.
25346 (find_if_case_2): Update.
25347 * internal-fn.c (expand_arith_overflow_result_store): Update.
25348 (expand_addsub_overflow): Update.
25349 (expand_neg_overflow): Update.
25350 (expand_mul_overflow): Update.
25351 (expand_vector_ubsan_overflow): Update.
25352 * ipa-cp.c (good_cloning_opportunity_p): Update.
25353 * ipa-split.c (split_function): Use make_single_succ_edge.
25354 * ipa-utils.c (ipa_merge_profiles): Update.
25355 * loop-doloop.c (add_test): Update.
25356 (doloop_modify): Update.
25357 * loop-unroll.c (compare_and_jump_seq): Update.
25358 (unroll_loop_runtime_iterations): Update.
25359 * lra-constraints.c (lra_inheritance): Update.
25360 * lto-streamer-in.c (input_cfg): Update.
25361 * lto-streamer-out.c (output_cfg): Update.
25362 * mcf.c (adjust_cfg_counts): Update.
25363 * modulo-sched.c (sms_schedule): Update.
25364 * omp-expand.c (expand_omp_for_init_counts): Update.
25365 (extract_omp_for_update_vars): Update.
25366 (expand_omp_ordered_sink): Update.
25367 (expand_omp_for_ordered_loops): Update.
25368 (expand_omp_for_generic): Update.
25369 (expand_omp_for_static_nochunk): Update.
25370 (expand_omp_for_static_chunk): Update.
25371 (expand_cilk_for): Update.
25372 (expand_omp_simd): Update.
25373 (expand_omp_taskloop_for_outer): Update.
25374 (expand_omp_taskloop_for_inner): Update.
25375 * omp-simd-clone.c (simd_clone_adjust): Update.
25376 * optabs.c (expand_doubleword_shift): Update.
25377 (expand_abs): Update.
25378 (emit_cmp_and_jump_insn_1): Update.
25379 (expand_compare_and_swap_loop): Update.
25380 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25381 * predict.c (predictable_edge_p): Update.
25382 (edge_probability_reliable_p): Update.
25383 (set_even_probabilities): Update.
25384 (combine_predictions_for_insn): Update.
25385 (combine_predictions_for_bb): Update.
25386 (propagate_freq): Update.
25387 (estimate_bb_frequencies): Update.
25388 (force_edge_cold): Update.
25389 * profile-count.c (profile_count::dump): Add missing space into dump.
25390 (profile_count::debug): Add newline.
25391 (profile_count::differs_from_p): Explicitly convert to unsigned.
25392 (profile_count::stream_in): Update.
25393 (profile_probability::dump): New member function.
25394 (profile_probability::debug): New member function.
25395 (profile_probability::differs_from_p): New member function.
25396 (profile_probability::differs_lot_from_p): New member function.
25397 (profile_probability::stream_in): New member function.
25398 (profile_probability::stream_out): New member function.
25399 * profile-count.h (profile_count_quality): Rename to ...
25400 (profile_quality): ... this one.
25401 (profile_probability): New.
25402 (profile_count): Update.
25403 * profile.c (compute_branch_probabilities): Update.
25404 * recog.c (peep2_attempt): Update.
25405 * sched-ebb.c (schedule_ebbs): Update.
25406 * sched-rgn.c (find_single_block_region): Update.
25407 (compute_dom_prob_ps): Update.
25408 (schedule_region): Update.
25409 * sel-sched-ir.c (compute_succs_info): Update.
25410 * stmt.c (struct case_node): Update.
25411 (do_jump_if_equal): Update.
25412 (get_outgoing_edge_probs): Update.
25413 (conditional_probability): Update.
25414 (emit_case_dispatch_table): Update.
25415 (expand_case): Update.
25416 (expand_sjlj_dispatch_table): Update.
25417 (emit_case_nodes): Update.
25418 * targhooks.c: Update.
25419 * tracer.c (better_p): Update.
25420 (find_best_successor): Update.
25421 * trans-mem.c (expand_transaction): Update.
25422 * tree-call-cdce.c: Update.
25423 * tree-cfg.c (gimple_split_edge): Upate.
25424 (move_sese_region_to_fn): Upate.
25425 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25426 * tree-eh.c (lower_resx): Upate.
25427 (cleanup_empty_eh_move_lp): Upate.
25428 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25429 * tree-inline.c (copy_edges_for_bb): Update.
25430 (copy_cfg_body): Update.
25431 * tree-parloops.c (gen_parallel_loop): Update.
25432 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25433 (gimple_gen_time_profiler): Update.
25434 * tree-ssa-dce.c (remove_dead_stmt): Update.
25435 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25436 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25437 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25438 (unloop_loops): Update.
25439 (try_peel_loop): Update.
25440 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25441 * tree-ssa-loop-split.c (connect_loops): Update.
25442 (split_loop): Update.
25443 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25444 (hoist_guard): Update.
25445 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25446 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25447 (value_replacement): Update.
25448 * tree-ssa-reassoc.c (branch_fixup): Update.
25449 * tree-ssa-tail-merge.c (replace_block_by): Update.
25450 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25451 (create_edge_and_update_destination_phis): Update.
25452 (compute_path_counts): Update.
25453 (recompute_probabilities): Update.
25454 (update_joiner_offpath_counts): Update.
25455 (freqs_to_counts_path): Update.
25456 (duplicate_thread_path): Update.
25457 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25458 (struct switch_conv_info): Update.
25459 (gen_inbound_check): Update.
25460 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25461 (vect_do_peeling): Update.
25462 (vect_loop_versioning): Update.
25463 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25464 (optimize_mask_stores): Update.
25465 * ubsan.c (ubsan_expand_null_ifn): Update.
25466 * value-prof.c (gimple_divmod_fixed_value): Update.
25467 (gimple_divmod_fixed_value_transform): Update.
25468 (gimple_mod_pow2): Update.
25469 (gimple_mod_pow2_value_transform): Update.
25470 (gimple_mod_subtract): Update.
25471 (gimple_mod_subtract_transform): Update.
25472 (gimple_ic): Update.
25473 (gimple_stringop_fixed_value): Update.
25474 (gimple_stringops_transform): Update.
25475 * value-prof.h: Update.
25477 2017-06-29 Carl Love <cel@us.ibm.com>
25479 * config/rs6000/rs6000-c.c: Add support for built-in functions
25480 vector signed int vec_signed (vector float);
25481 vector signed long long vec_signed (vector double);
25482 vector signed int vec_signed2 (vector double, vector double);
25483 vector signed int vec_signede (vector double);
25484 vector signed int vec_signedo (vector double);
25485 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25486 instruction generator.
25487 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25488 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25489 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25491 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25492 vunsignede_v2df): Add define_expands.
25493 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25494 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25495 VEC_UNSIGNEDO): Add definitions.
25496 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25497 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25498 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25499 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25500 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25501 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25502 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25503 vec_unsignede, vec_unsignedo): Add builtin defines.
25504 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25506 * doc/extend.texi: Update the built-in documentation file for the
25507 new built-in functions.
25509 2017-06-29 Richard Biener <rguenther@suse.de>
25511 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25512 reduction chains to LOOP_VINFO_REDUCTIONS.
25513 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25514 SLP reductions after processing reduction chains.
25516 2017-06-29 Nathan Sidwell <nathan@acm.org>
25518 * builtins.c (fold_builtin_FUNCTION): Use
25519 lang_hooks.decl_printable_name.
25521 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25523 PR middle-end/81194
25524 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25525 with only one label.
25526 * stmt.c (expand_case): Assert NCASES is greater than one.
25528 2017-06-29 Richard Biener <rguenther@suse.de>
25530 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25532 (group_case_labels): Likewise.
25533 (find_taken_edge): Push sanity checking on val to workers...
25534 (find_taken_edge_cond_expr): ... here
25535 (find_taken_edge_switch_expr): ... and here, handle cases
25536 with just a default label.
25537 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25538 (group_case_labels): Likewise.
25539 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25540 group_case_labels does anything cleanup the CFG again.
25542 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25544 PR tree-optimization/81196
25545 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25546 exit condition comparing two IVs.
25548 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25550 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25551 profile to the dummy entry at the end of the list of architectures.
25552 * config/arm/arm-cpu-cdata.h: Regenerated.
25554 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25555 Michael Collison <michael.collison@arm.com>
25558 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25560 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25561 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25562 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25563 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25564 of shift when the shift amount is masked with constant equal to
25565 the size of the mode.
25566 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25569 2017-06-29 Martin Liska <mliska@suse.cz>
25571 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25573 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25576 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25579 (edge_growth_cache_entry::edge_growth_cache_entry): New
25581 (reset_edge_growth_cache): Update to use constructor.
25583 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25585 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25586 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25587 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25589 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25591 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25592 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25594 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25596 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25597 (*-linux-uclibc*): Add t-uclibc tmake_file.
25598 * config/t-musl: New.
25599 * config/t-uclibc: New.
25601 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25603 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25605 (gen_comm_data): Emit architectural setting of arch_prof.
25606 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25608 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25609 (armv8-m.base, armv8-m.main): Likewise.
25610 * arm-protos.h (arm_build_target): Add profile field.
25611 (arch_option): Likewise.
25612 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25614 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25615 arm_active_target.profile.
25617 2017-06-28 Richard Biener <rguenther@suse.de>
25619 PR middle-end/81227
25620 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25621 TYPE_OVERFLOW_WRAPS.
25622 * match.pd (negate_expr_p): Likewise.
25623 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25624 fold_build2, not fold_binary.
25626 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25628 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25629 Convert memory address to Pmode.
25630 (aarch64_print_operand): Assert MEM operands are always Pmode.
25632 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25635 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25636 Remove redundant if.
25637 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25638 * config/arm/aarch-common-protos.h
25639 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25640 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25642 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25645 * multiple_target.c (create_dispatcher_calls): Set the default
25646 clone to be static, not public.
25648 2017-06-28 Richard Biener <rguenther@suse.de>
25650 * tree-vect-loop.c (vectorizable_reduction): Move special
25651 cond reduction IV var creation ...
25652 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25653 parameter. Use STMT_VINFO_VECTYPE.
25654 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25657 2017-06-28 Martin Liska <mliska@suse.cz>
25660 * ipa-visibility.c (non_local_p): Handle visibility.
25662 2017-06-28 Martin Liska <mliska@suse.cz>
25665 * common.opt: Add IntegerRange to various options.
25666 * opt-functions.awk (integer_range_info): New function.
25667 * optc-gen.awk: Add integer_range_info to cl_options struct.
25668 * opts-common.c (decode_cmdline_option): Handle
25669 CL_ERR_INT_RANGE_ARG.
25670 (cmdline_handle_error): Likewise.
25671 * opts.c (print_filtered_help): Show valid interval in
25672 when --help is provided.
25673 * opts.h (struct cl_option): Add range_min and range_max fields.
25674 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25676 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25678 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25679 (x * C EQ/NE y * C): New transformation.
25681 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25683 * genmultilib (combination_space): Accept '+' in option names.
25685 2017-06-28 Martin Liska <mliska@suse.cz>
25688 * asan.c (instrument_derefs): Bail out inner references
25689 that are hard register variables.
25691 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25694 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25695 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25697 2017-06-28 Richard Biener <rguenther@suse.de>
25699 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25700 (vect_get_slp_defs): Adjust.
25701 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25703 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25705 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25706 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25707 (vectorizable_reduction): Adjust.
25708 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25710 (vect_get_slp_defs): Likewise.
25711 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25712 (vectorizable_bswap): Adjust.
25713 (vectorizable_call): Likewise.
25714 (vectorizable_conversion): Likewise.
25715 (vectorizable_assignment): Likewise.
25716 (vectorizable_shift): Likewise.
25717 (vectorizable_operation): Likewise.
25718 (vectorizable_store): Likewise.
25719 (vectorizable_condition): Likewise.
25720 (vectorizable_comparison): Likewise.
25722 2017-06-28 Michael Collison <michael.collison@arm.com>
25725 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25727 (arm_gen_movmemqi): Removed unused variable 'i'.
25728 Convert 'for' loop into 'while' loop.
25729 (arm_expand_prologue): Remove last unnecessary set of insn.
25730 (thumb_pop): Remove unused variable 'pushed_words'.
25731 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25733 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25735 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25736 * config/s390/s390-protos.h: Add prototype of
25737 s390_rel_address_ok_p.
25738 * config/s390/s390.c (s390_got_symbol): New function.
25739 (s390_rel_address_ok_p): New function.
25740 (legitimize_pic_address): Use s390_rel_address_ok_p.
25741 (s390_load_got): Use s390_got_symbol.
25742 (s390_option_override): Issue error if
25743 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25744 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25746 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25748 2017-06-27 Andrew Pinski <apinski@cavium.com>
25750 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25751 (X * copysign (1.0, X)): New pattern.
25752 (X * copysign (1.0, -X)): New pattern.
25753 (copysign (-1.0, CST)): New pattern.
25755 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25757 * genmultilib (combination_space): Remove variable.
25758 Validate reuse rules against regular expression for any sequence
25759 of multilib options in any order.
25761 2017-06-27 Michael Collison <michael.collison@arm.com>
25763 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25764 call aarch64_split_simd_combine.
25765 * (aarch64_combine_internal<mode>): Delete pattern.
25766 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25767 Allow register and subreg operands.
25769 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25771 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25772 specific need, just fallback on defaults.
25773 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25775 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25776 Olivier Hainque <hainque@adacore.com>
25778 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25780 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25781 targets. Pick a default if no particular attempt applied.
25782 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25785 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25787 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25788 (x86_64-wrs-vxworks7): Likewise.
25790 2017-06-27 Marek Polacek <polacek@redhat.com>
25793 * ubsan.c (instrument_null): Check get_base_address's result for null.
25795 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25797 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25799 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25801 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25802 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25803 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25804 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25805 New function types.
25806 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25807 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25808 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25809 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25810 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25811 BUILT_IN_FEUPDATEENV): New builtins.
25812 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25813 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25814 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25815 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25817 (builtin_structptr_types): Adjust size.
25818 * tree.c (builtin_structptr_types): Add four entries.
25820 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25821 Olivier Hainque <hainque@adacore.com>
25823 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25824 (TLS_SYM): New local macro, forcing reference to __tls__ on
25825 link command lines for VxWorks 7 RTPs, triggering initialization
25827 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25828 OS features TLS support, true for RTPs on VxWorks 7.
25829 * config/vxworks.c (vxworks_override_options): Setup emutls
25832 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25834 * predict.c (test_prediction_value_range): Use -1U instead of -1
25835 to avoid narrowing conversion warning.
25836 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25837 to avoid narrowing conversion warning.
25838 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25840 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25842 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25844 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25845 64bit configurations.
25846 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25847 (SIZE_TYPE): Likewise.
25848 * config/vxworks.c (vxworks_emutls_var_fields): Use
25849 long_unsigned_type_node instead of unsigned_type_node as the offset
25850 field type, which is "pointer" mode in emutls.c.
25852 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25855 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25857 PR middle-end/81207
25858 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25859 gimple_vuse copying separately from gimple_vdef copying.
25861 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25863 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25866 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25867 Olivier Hainque <hainque@adacore.com>
25869 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25870 for all vxworks7 targets.
25871 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25872 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25873 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25874 variations for VX6/VX7 and 32/64bits later on in ...
25875 (VXWORKS_LIB_SPEC): Leverage new macros.
25876 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25877 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25879 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25881 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25882 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25884 2017-06-26 Carl Love <cel@us.ibm.com>
25886 * config/rs6000/rs6000-c.c: Add support for built-in functions
25887 vector bool char vec_reve (vector bool char);
25888 vector signed char vec_reve (vector signed char);
25889 vector unsigned char vec_reve (vector unsigned char);
25890 vector bool int vec_reve (vector bool int);
25891 vector signed int vec_reve (vector signed int);
25892 vector unsigned int vec_reve (vector unsigned int);
25893 vector bool long long vec_reve (vector bool long long);
25894 vector signed long long vec_reve (vector signed long long);
25895 vector unsigned long long vec_reve (vector unsigned long long);
25896 vector bool short vec_reve (vector bool short);
25897 vector signed short vec_reve (vector signed short);
25898 vector double vec_reve (vector double);
25899 vector float vec_reve (vector float);
25900 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25901 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25902 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25903 (altivec_vreve): New pattern.
25904 * config/rs6000/altivec.h (vec_reve): New define.
25905 * doc/extend.texi (vec_rev): Update the built-in documentation file
25906 for the new built-in functions.
25908 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25910 PR tree-optimization/71815
25911 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25913 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25915 (slsr_process_phi): Likewise.
25916 (replace_uncond_cands_and_profitable_phis): Don't replace a
25917 multiply candidate with a stride of 1 (copy or cast).
25918 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25920 (lowest_cost_path): Likewise.
25921 (total_savings): Likewise.
25923 2017-06-26 Richard Biener <rguenther@suse.de>
25926 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25927 Use def_builtin_pure for all gather builtins.
25929 2017-06-26 Richard Biener <rguenther@suse.de>
25931 PR tree-optimization/81203
25932 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25933 non-dominating BBs.
25935 2017-06-26 Marek Polacek <polacek@redhat.com>
25938 * doc/invoke.texi: Document -Wmultistatement-macros.
25940 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25942 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25943 arm_neon_ok_no_float_abi effective target.
25945 2017-06-26 Richard Biener <rguenther@suse.de>
25947 PR tree-optimization/80928
25948 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25949 (copy_bbs): Set BB_DUPLICATED flag early.
25950 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25952 (execute_on_shrinking_pred): Likewise.
25953 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25954 BB_DUPLICATED blocks.
25955 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25956 iterate over all PHIs considering removal of *gsi.
25958 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25960 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25963 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25965 * config/rs6000/rs6000-string.c: (expand_block_clear,
25966 do_load_for_compare, select_block_compare_mode,
25967 compute_current_alignment, expand_block_compare,
25968 expand_strncmp_align_check, expand_strn_compare,
25969 expand_block_move, rs6000_output_load_multiple)
25970 Move functions related to string/block move/compare
25971 to a separate file.
25972 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25973 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25974 for this function which is now used in two files.
25975 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25976 * config.gcc: Add rs6000-string.o to extra_objs for
25977 targets powerpc*-*-* and rs6000*-*-*.
25979 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25982 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25983 32-bit, since indexed is not valid for DImode.
25984 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25985 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25986 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25987 (define_peephole2 for Altivec d-form store): Likewise.
25990 * multiple_target.c (create_dispatcher_calls): Only create the
25991 dispatcher call if the function is the default clone of a
25992 versioned function.
25994 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25996 PR middle-end/80902
25997 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25998 a call, force the call to not be a tail call.
26000 2017-06-23 Jeff Law <law@redhat.com>
26002 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26005 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26007 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26008 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26009 handling for early expansion of vector shifts (sl,sr,sra,rl).
26010 (builtin_function_type): Add vector shift right instructions
26011 to the unsigned argument list.
26013 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26015 rtl-optimizatoin/79286
26016 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26017 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26018 trap. PIC register plus a const unspec without offset can never trap.
26020 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26022 * tree.h (builtin_structptr_type): New type.
26023 (builtin_structptr_types): Declare new array.
26024 * tree.c (builtin_structptr_types): New array.
26025 (free_lang_data, build_common_tree_nodes): Use it.
26027 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26030 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26033 2017-06-22 Matt Turner <mattst88@gmail.com>
26035 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26036 Lake models to skylake case. Assume skylake for unknown
26037 models with clflushopt.
26039 2017-06-22 Jeff Law <law@redhat.com>
26041 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26042 frame sizes that do not satisfy aarch64_uimm12_shift.
26044 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26046 * profile-count.h (apply_probability,
26047 apply_scale, probability_in): Fix checks for zero.
26049 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26051 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26052 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26054 2016-06-22 Richard Biener <rguenther@suse.de>
26056 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26057 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26058 REDUC_MAX_EXPR support.
26059 (vectorizable_reduction): Likewise.
26060 (vect_create_epilog_for_reduction): Likewise.
26062 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26064 * match.pd (A / (1 << B) -> A >> B): New.
26065 * generic-match-head.c: Include optabs-tree.h.
26066 * gimple-match-head.c: Likewise.
26067 * optabs-tree.h (target_supports_op_p): New.
26068 * optabs-tree.c (target_supports_op_p): New.
26070 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26072 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26073 $gcc_cv_ld --help output.
26074 (gcc_cv_ld_demangle): Likewise.
26075 (gcc_cv_ld_eh_frame_hdr): Likewise.
26076 (gcc_cv_ld_pie): Likewise.
26077 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26078 (gcc_cv_ld_buildid): Likewise.
26079 (gcc_cv_ld_sysroot): Likewise.
26080 (ld_bndplt_support): Likewise.
26081 (ld_pushpopstate_support): Likewise.
26082 * configure: Regenerate.
26083 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26085 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26088 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26089 operands indexes to avoid gap between operands and match_dups.
26091 2017-06-21 Andrew Pinski <apinski@cavium.com>
26093 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26094 Increment Arith_shift and Arith_shift_reg by 1.
26095 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26097 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26098 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26099 (aarch64_strip_extend): Add new argument and test for it.
26100 (aarch64_cheap_mult_shift_p): New function.
26101 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26102 add a cost if it is true.
26103 Update calls to aarch64_strip_extend.
26104 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26106 2017-06-21 Andrew Pinski <apinski@cavium.com>
26108 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26110 (thunderxt88): Likewise.
26111 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26112 (thunderx_prefetch_tune): New variable.
26113 (thunderx2t99_prefetch_tune): Update for the correct values.
26114 (thunderxt88_tunings): New variable.
26115 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26116 generic_prefetch_tune.
26117 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26119 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26121 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26122 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26123 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26124 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26125 (aarch64_atomic_cas<mode>, GPI): Likewise.
26127 2017-06-21 Martin Liska <mliska@suse.cz>
26129 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26130 statements on cold and hot labels.
26131 * predict.c (tree_estimate_probability_bb): Remove the
26132 prediction from this place.
26134 2017-06-21 Martin Liska <mliska@suse.cz>
26136 PR tree-optimization/79489
26137 * gimplify.c (maybe_add_early_return_predict_stmt): New
26139 (gimplify_return_expr): Call the function.
26140 * predict.c (tree_estimate_probability_bb): Remove handling
26142 * predict.def: Update comment about early return predictor.
26143 * gimple-predict.h (is_gimple_predict): New function.
26144 * predict.def: Change default value of early return to 66.
26145 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26147 * passes.def: Put pass_strip_predict_hints to the beginning of
26150 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26152 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26153 FUNCTION_DECL declarations.
26154 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26156 (dwaf2out_decl): Likewise.
26157 * godump.c (go_early_global_decl): Skip call to the real debug hook
26158 for FUNCTION_DECL declarations.
26159 * passes.c (rest_of_decl_compilation): Skip call to the
26160 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26161 -fdump-go-spec is passed.
26163 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26165 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26166 Reorder for compactness.
26167 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26168 (def_builtin_pure, def_builtin_pure2): New functions.
26169 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26171 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26173 * match.pd (nop_convert): New predicate.
26174 ((A +- CST1) +- CST2): Allow some NOP conversions.
26176 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26179 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26180 with ctors/dtors if GOVD_SHARED is set.
26182 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26184 * config/aarch64/aarch64.md (movti_aarch64):
26185 Emit mov rather than orr.
26186 (movtf_aarch64): Likewise.
26187 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26188 Emit mov rather than orr.
26190 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26192 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26193 Swap alternatives, make integer dup more expensive.
26195 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26197 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26198 Return true for non-tls symbols.
26200 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26202 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26203 (cortex-a75): Likewise.
26204 (cortex-a75.cortex-a55): Likewise.
26205 * config/aarch64/aarch64-tune.md: Regenerate.
26206 * doc/invoke.texi (-mtune): Document new values for -mtune.
26208 2017-06-21 Tom de Vries <tom@codesourcery.com>
26210 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26211 stack_size feature.
26212 (Effective-Target Keywords, Other attributes): Suggest using
26213 dg-add-options stack_size feature to get stack limit in stack_size
26214 effective target documentation.
26216 2017-06-21 Julian Brown <julian@codesourcery.com>
26217 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26219 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26220 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26221 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26223 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26224 attribute type list for neon_multiply.
26225 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26226 attribute type list for neon_multiply.
26227 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26228 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26229 attribute type list for neon_multiply.
26230 * config/arm/types.md (crypto_pmull): Add.
26231 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26232 attribute type list.
26234 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26236 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26239 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26241 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26242 to make sure not to dereference a NULL cost_classes_ptr pointer.
26244 2017-06-20 Carl Love <cel@us.ibm.com>
26246 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26247 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26248 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26249 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26250 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26251 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26252 VMULOSW): New enum "unspec" values.
26253 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26254 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26255 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26256 altivec_vmulosw): New patterns.
26257 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26258 VMULOSW): Add definitions.
26260 2017-06-20 Julia Koval <julia.koval@intel.com>
26262 * config/i386/i386.c: Fix rounding expand for new pattern.
26263 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26265 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26267 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26268 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26270 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26272 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26275 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26277 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26278 architecture, then by implementer ID.
26279 * config/aarch64/aarch64-tune.md: Regenerate.
26281 2017-06-20 Richard Biener <rguenther@suse.de>
26283 PR middle-end/81097
26284 * fold-const.c (split_tree): Fold to type before negating.
26286 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26288 * diagnostic-show-locus.c
26289 (selftest::test_fixit_deletion_affecting_newline): New function.
26290 (selftest::diagnostic_show_locus_c_tests): Call it.
26292 2017-06-20 Andreas Schwab <schwab@suse.de>
26295 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26298 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26300 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26301 __ARM_FEATURE_COPROC according to support.
26303 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26305 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26306 Rewritten to avoid overflow for > 32-bit pointers.
26309 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26310 by removing enum keyword.
26311 (ubsan_type_descriptor): Likewise. Formatting fix.
26314 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26315 splitter): Require TARGET_SSE2 in the condition.
26317 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26320 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26321 for doing vector set of SFmode on ISA 3.0.
26322 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26323 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26325 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26326 SFmode value into a V4SF variable that was extracted from another
26327 V4SF variable without converting the element to double precision
26328 and back to single precision vector format.
26329 (vsx_insert_extract_v4sf_p9_2): Likewise.
26331 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26333 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26334 in UWHI to avoid undefined overflow.
26337 * ubsan.h (enum ubsan_encode_value_phase): New.
26338 (ubsan_encode_value): Change second argument to
26339 enum ubsan_encode_value_phase with default value of
26340 UBSAN_ENCODE_VALUE_GENERIC.
26341 * ubsan.c (ubsan_encode_value): Change second argument to
26342 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26343 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26344 create_tmp_var_raw instead of create_tmp_var and use a
26346 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26347 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26348 ubsan_encode_value callers.
26351 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26352 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26353 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26355 2017-06-19 Richard Biener <rguenther@suse.de>
26357 PR middle-end/81118
26358 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26359 estimates if we changed anything.
26361 2017-06-19 Richard Biener <rguenther@suse.de>
26363 PR tree-optimization/80887
26364 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26365 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26366 simplified lookups, then reset mprts_hook.
26367 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26369 (try_to_simplify): Likewise.
26371 2017-06-19 Martin Liska <mliska@suse.cz>
26374 * gimplify.c (gimplify_switch_expr):
26375 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26377 2017-06-19 Martin Liska <mliska@suse.cz>
26379 * doc/install.texi: Document that PGO runs in 4 stages.
26381 2017-06-19 Martin Liska <mliska@suse.cz>
26384 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26385 to dispatcher function name.
26386 * multiple_target.c (replace_function_decl): New function.
26387 (create_dispatcher_calls): Redirect both edges and references.
26389 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26391 * profile-count.c (profile_count::dump): Dump quality.
26392 (profile_count::differs_from_p): Update for unsigned val.
26393 * profile-count.h (profile_count_quality): New enum.
26394 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26396 2017-06-19 Richard Biener <rguenther@suse.de>
26398 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26399 struct function as arg.
26400 (estimate_numbers_of_iterations): Export overload with loop arg.
26401 (free_numbers_of_iterations_estimates_loop): Use an overload of
26402 free_numbers_of_iterations_estimates instead.
26403 * tree-cfg.c (remove_bb): Adjust.
26404 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26405 * tree-parloops.c (gen_parallel_loop): Likewise.
26406 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26408 (tree_unroll_loops_completely): Likewise.
26409 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26410 Use an overload instead and export.
26411 (estimated_loop_iterations): Adjust.
26412 (max_loop_iterations): Likewise.
26413 (likely_max_loop_iterations): Likewise.
26414 (estimate_numbers_of_iterations): Take struct function as arg
26416 (loop_exits_before_overflow): Adjust.
26417 (free_numbers_of_iterations_estimates_loop): Use an overload.
26418 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26419 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26421 2017-06-19 Richard Biener <rguenther@suse.de>
26424 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26425 RANGE_EXPR conservatively.
26427 2017-06-16 Carl Love <cel@us.ibm.com>
26429 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26430 definitions for vec_float, vec_float2, vec_floato,
26431 vec_floate built-ins.
26432 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26433 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26435 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26436 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26437 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26438 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26439 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26440 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26441 vec_floato): Add builtin defines.
26442 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26443 Update the built-in documentation file for the new built-in
26446 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26448 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26449 (mthumb): Mark as the negative of -marm.
26451 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26453 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26455 (ARM Options, -mtune): Document that this accepts the same
26456 extension options as -mcpu.
26457 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26459 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26461 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26462 permitted extensions.
26464 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26466 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26467 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26468 (armv8-m.main): Add option +nodsp.
26469 * config/arm/arm-cpu-cdata.h: Regenerated.
26471 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26473 * config/arm/t-fuchsia: New file.
26474 * config.gcc (arm*-*-fuchsia*): Use it.
26476 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26478 * config/arm/t-symbian: Rewrite for new option infrastructure.
26480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26482 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26483 (MULTILIB_REQUIRED): Likewise.
26485 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26487 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26488 (MULTILIB_RESUE): Likewise.
26489 (MULTILIB_MATCHES): Likewise.
26490 (MULTLIB_REQUIRED): Likewise.
26492 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26494 * config/arm/t-rtems: Rewrite for new option framework.
26496 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26498 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26499 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26500 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26501 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26502 * config/arm/t-multilib: ... here.
26503 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26504 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26505 armv7-a and armv8*-a when A-profile libraries have not been built.
26506 * config/arm/t-rmprofile: Rewrite.
26508 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26510 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26511 with a backslash. Remove the backslash after substituting unescaped
26513 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26515 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26517 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26518 the driver to pass through the default setting of -mfloat-abi.
26519 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26521 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26522 (all_feat_combs): New rule.
26523 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26525 * config/arm/t-aprofile: Rewrite.
26527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26529 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26530 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26531 fpu is not specified by the user/command-line.
26532 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26533 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26534 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26535 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26536 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26537 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26539 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26541 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26542 * config/arm/t-arm-elf: Rewritten.
26544 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26546 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26547 have some floating-point instructions.
26548 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26549 (TARGET_MAYBE_HARD_FLOAT): New macro.
26550 * config/arm/arm-builtins.c (arm_init_builtins): Use
26551 TARGET_MAYBE_HARD_FLOAT.
26552 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26554 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26556 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26557 (configargs.h): Include it.
26558 (arm_print_hint_for_fpu_option): New function.
26559 (arm_parse_fpu_option): New function.
26560 (candidate_extension): New class.
26561 (arm_canon_for_multilib): New function.
26562 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26563 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26564 (ARCH_CANONICAL_SPECS): New macro.
26565 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26567 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26569 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26570 are set after handling multilib fragments. Set target_cpu_default2
26573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26575 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26577 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26578 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26580 (check_arch): Likewise.
26581 * config/arm/arm.c (arm_configure_build_target): Handle
26582 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26583 options in the default.
26585 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26587 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26588 when an option is an alias of another.
26589 * config/arm/parsecpu.awk (optalias): New parser token.
26590 (gen_comm_data): Mark non-alias options as such. Emit entries
26591 for extension aliases.
26592 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26593 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26594 (armv6kz, armv6zk, armv6t2): Likewise.
26595 (armv7): Make vfpv3-d16 an alias.
26596 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26598 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26599 Sort in canonical order.
26600 (armv8-a): Sort in canonical order.
26601 (armv8.1-a, armv8.2-a): Likewise.
26602 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26604 (cortex-a9): Sort in canonical order.
26605 * config/arm/arm.c (selftests.h): Include it.
26606 (arm_test_cpu_arch_data): New function.
26607 (arm_run_self_tests): New function.
26608 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26609 (targetm): Move declaration to the end of the file.
26610 * arm-cpu-cdata.h: Regenerated.
26612 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26614 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26615 call to target_mode_check describing the type of option passed.
26616 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26617 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26618 arm_parse_cpu_option_name to match parameters against list of
26620 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26621 arm_arch_core_flags data structure.
26622 * config/arm/arm-cpu_cdata.h: Regenerated.
26624 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26626 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26628 (arm_print_hint_for_cpu_option): Likewise.
26629 (arm_print_hint_for_arch_option): Likewise.
26630 (arm_parse_cpu_option_name): Likewise.
26631 (arm_parse_arch_option_name): Likewise.
26632 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26633 of entries in the all_fpus list.
26634 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26635 (arm_parse_cpu_option_name): Declare.
26636 (arm_parse_arch_option_name): Declare.
26637 (arm_parse_option_features): Declare.
26638 (arm_intialize_isa): Declare.
26639 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26641 (gen_comm_data): ... here. Make definitions non-static.
26642 * config/arm/arm-cpu-data.h: Regenerated.
26643 * config/arm/arm-cpu-cdata.h: Regenerated.
26645 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26647 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26648 (cpu_arch_extension): New structure.
26649 (cpu_arch_option, arch_option, cpu_option): New structures.
26650 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26651 architecture types.
26652 (gen_data): Generate new format data tables.
26653 * config/arm/arm.c (cpu_tune): New structure.
26654 (cpu_option, processors): Delete.
26655 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26656 (arm_print_hint_for_cpu_option): ... this and ...
26657 (arm_print_hint_for_arch_option): ... this.
26658 (arm_parse_arch_cpu_name): Delete. Replace with ...
26659 (arm_parse_cpu_option_name): ... this and ...
26660 (arm_parse_arch_option_name): ... this.
26661 (arm_unrecognized_feature): Change type of target parameter to
26663 (arm_parse_arch_cpu_features): Delete. Replace with ...
26664 (arm_parse_option_features): ... this.
26665 (arm_configure_build_target): Rework to use new configuration data
26667 (arm_print_tune_info): Rework for new configuration data tables.
26668 * config/arm/arm-cpu-data.h: Regenerated.
26669 * config/arm/arm-cpu.h: Regenerated.
26671 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26673 * Makefile.in (OBJS): Move sbitmap.o from here ...
26674 (OBJS-libcommon): ... to here.
26676 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26678 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26679 (ISA_ALL_CRYPTO): New macro.
26680 (ISA_ALL_SIMD): New macro
26681 (ISA_ALL_FP): New macro.
26682 * config/arm/arm.c (fpu_bitlist): Update initializer.
26683 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26685 (arm9e): Add fpu. Add option for nofp
26686 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26687 (arm926ej-s, arm1026ej-s): Likewise.
26688 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26689 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26690 neon-fp16, neon-vfpv4, nofp and nosimd.
26691 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26692 (cortex-a8): Add fpu. Add option for nofp.
26693 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26694 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26695 (cortex-r4f): Add fpu.
26696 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26697 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26699 (cortex-r8): Likewise.
26700 (cortex-m4): Add fpu. Add option for nofp.
26701 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26702 (cortex-a17.cortex-a7): Likewise.
26703 (cortex-a32): Add fpu. Add options for crypto and nofp.
26704 (cortex-a35, cortex-a53): Likewise.
26705 (cortex-a57): Add fpu. Add option for crypto.
26706 (cortex-a72, cortex-a73): Likewise.
26707 (exynos-m1): Likewise.
26708 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26709 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26710 (cortex-m33): Add fpu. Add option for nofp.
26711 * config/arm/arm-cpu-cdata.h: Regenerated
26712 * config/arm/arm-cpu-data.h: Regenerated.
26714 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26716 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26717 (armv5te, armv5tej): Likewise.
26718 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26719 (armv7): Add options fp and vfpv3-d16.
26720 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26721 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26723 (armv7ve): Likewise.
26724 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26725 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26726 (armv8-a): Add nocrypto option.
26727 (armv8.1-a, armv8.2-a): Likewise.
26728 (armv8-m.main): add options fp, fp.dp and nofp.
26730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26732 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26734 (armv8-a+crc): Delete.
26735 (armv8.1-a): Add options simd, crypto and nofp.
26736 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26737 (armv8.2-a+fp16): Delete.
26738 (armv8-m.main): Add option dsp.
26739 (armv8-m.main+dsp): Delete.
26740 (cortex-a8): Add fpu. Add nofp option.
26741 (cortex-a9): Add fpu. Add nofp and nosimd options.
26742 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26743 link to main cpu and architecture data structures.
26744 (gen_comm_data): Only put isa attributes from the main architecture
26746 (option): New statement for architecture and CPU entries.
26747 * arm.c (struct cpu_option): New structure.
26748 (struct processors): Add entry for options.
26749 (arm_unrecognized_feature): New function.
26750 (arm_parse_arch_cpu_name): Ignore any characters after the first
26752 (arm_parse_arch_cpu_feature): New function.
26753 (arm_configure_build_target): Separate out any CPU and architecture
26754 features and parse separately. Don't error out if -mfpu=auto is
26755 used with only an architecture string.
26756 (arm_print_asm_arch_directives): New function.
26757 (arm_file_start): Call it.
26758 * config/arm/arm-cpu-cdata.h: Regenerated.
26759 * config/arm/arm-cpu-data.h: Likewise.
26760 * config/arm/arm-tables.opt: Likewise.
26762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26764 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26765 assembler when it is not -mfpu=auto.
26767 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26769 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26770 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26771 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26772 (ASM_CPU_SPEC): Rewrite.
26773 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26774 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26775 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26776 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26777 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26778 copied string is NUL-terminated. Also strip any characters prefixed
26780 (arm_rewrite_selected_arch): New function.
26781 (arm_rewrite_march): New function.
26783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26785 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26786 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26787 (march, mcpu, mtune): Convert to string-based options.
26788 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26789 (arm_parse_arch_cpu_name): New function.
26790 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26791 identify selected architecture or CPU.
26792 (arm_option_save): New function.
26793 (TARGET_OPTION_SAVE): Redefine.
26794 (arm_option_restore): Restore string options.
26795 (arm_option_print): Print string options.
26797 2017-06-16 Martin Sebor <msebor@redhat.com>
26799 PR tree-optimization/80933
26800 PR tree-optimization/80934
26801 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26802 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26803 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26804 (gimple_fold_builtin): Call them.
26806 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26808 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26809 as unlikely; update profile.
26811 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26813 * predict.c (force_edge_cold): Handle declaring edges impossible
26816 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26818 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26820 (try_unroll_loop_completely): Fix reporting.
26822 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26824 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26826 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26829 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26830 if given a non-constant argument for an intrinsic which requires a
26833 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26835 * profile.c (compare_freqs): New function.
26836 (branch_prob): Sort edge list.
26837 (find_spanning_tree): Assume that the list is priority sorted.
26839 2017-06-16 Richard Biener <rguenther@suse.de>
26841 PR tree-optimization/81090
26842 * passes.def (pass_record_bounds): Remove.
26843 * tree-pass.h (make_pass_record_bounds): Likewise.
26844 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26845 make_pass_record_bounds): Likewise.
26846 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26847 not free niter estimates at the beginning but at the end.
26848 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26850 2017-06-16 Richard Biener <rguenther@suse.de>
26852 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26853 initializer to workaround ICE in host GCC 4.8.
26855 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26857 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26859 (clone_inlined_nodes): Update.
26861 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26863 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26864 prefetch settings, and enable prefetching by default at -O3.
26866 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26868 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26869 Set flag_prefetch_loop_arrays according to tuning data.
26871 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26873 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26874 New tune structure.
26875 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26876 [Unrelated to main purpose of the patch] Place the pointer field last
26877 to enable type checking errors when tune structure are wrongly merged.
26878 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26879 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26880 (thunderx2t99_prefetch_tune): New tune constants.
26881 (tune_params *_tunings): Update all tunings (no functional change).
26882 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26883 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26884 from tunings structures.
26886 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26889 * ubsan.c (instrument_null): Add T argument, use it instead
26890 of computing it based on IS_LHS.
26891 (instrument_object_size): Likewise.
26892 (pass_ubsan::execute): Adjust instrument_null and
26893 instrument_object_size callers to pass gimple_get_lhs or
26894 gimple_assign_rhs1 result to it. Use instrument_null instead of
26895 calling get_base_address and instrument_mem_ref. Handle
26896 aggregate call arguments for object-size sanitization.
26898 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26900 PR tree-optimization/81089
26901 * tree-vrp.c (is_masked_range_test): Validate operands of
26904 2017-06-15 Martin Sebor <msebor@redhat.com>
26907 * dumpfile.c (dump_register): Avoid calling memset to initialize
26908 a class with a default ctor.
26909 * gcc.c (struct compiler): Remove const qualification.
26910 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26911 * hash-table.h: Ditto.
26912 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26914 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26915 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26917 * params.h (struct param_info): Make struct members non-const.
26918 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26919 with default initialization.
26920 * vec.h (vec_copy_construct, vec_default_construct): New helper
26922 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26923 with vec_copy_construct.
26924 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26925 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26926 * doc/invoke.texi (-Wclass-memaccess): Document.
26928 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26930 * emit-rtl.h (is_leaf): Update comment about local
26931 register allocator.
26933 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26936 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26937 for a variable to have a section before checking if the section has a
26939 Set section to.persistent if persistent attribute is set.
26940 Warn if .persistent attribute is used on an automatic variable.
26942 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26944 PR rtl-optimization/80474
26945 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26947 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26949 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26952 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26954 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26955 (casesi): Emit bounds checking as RTL.
26956 (casesi_internal_mips16_<mode>): Remove bounds checking.
26958 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26960 * config/xtensa/xtensa.c (xtensa_option_override): Append
26961 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26962 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26963 xtensa_doloop_hooks): Define unconditionally.
26964 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26966 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26967 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26968 for it in xtensa_option_override.
26969 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26970 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26972 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26974 * doc/cppopts.texi: Document '-' special value to -MF.
26976 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26978 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26979 (cortex_a53_fconst): Likewise.
26980 (cortex_a53_fpmul): Likewise.
26981 (cortex_a53_f_load_64): Likewise.
26982 (cortex_a53_f_load_many): Likewise.
26983 (cortex_a53_advsimd_alu): Likewise.
26984 (cortex_a53_advsimd_alu_q): Likewise.
26985 (cortex_a53_advsimd_mul): Likewise.
26986 (cortex_a53_advsimd_mul_q): Likewise.
26987 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26988 Add missing fmul, r2f_cvt and fconst cases.
26990 2017-06-14 Richard Biener <rguenther@suse.de>
26992 PR middle-end/81088
26993 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26995 (fold_binary_loc): When associating do not treat pre-existing
26996 TREE_OVERFLOW on literal constants as a reason to allow
26997 TREE_OVERFLOW on associated literal constants.
26999 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27001 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27002 (MASK_FEATURES): New macro.
27003 * config/sparc/sparc.c (sparc_option_override): Remove the special
27004 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27006 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27008 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27009 a division of 0 if non-call exceptions are enabled.
27011 2017-06-14 Andrew Pinski <apinski@cavium.com>
27012 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27015 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27016 Improve vector initialization code gen for only variable case.
27018 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27020 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27022 2017-06-14 Richard Biener <rguenther@suse.de>
27024 PR tree-optimization/81083
27025 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27028 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27030 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27031 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27032 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27033 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27034 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27035 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27037 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27039 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27040 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27042 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27044 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27046 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27048 * config/rs6000/t-rtems: Don't handle SPE.
27050 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27052 * config/rs6000/t-linux: Don't handle SPE.
27054 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27056 * config/rs6000/eabispe.h: Delete file.
27058 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27060 * config/rs6000/t-spe: Delete file.
27062 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27064 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27065 (rs6000_legitimate_offset_address_p): Return false for anything in
27066 V2SImode or V2SFmode.
27068 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27070 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27071 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27072 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27074 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27075 (rs6000_legitimate_offset_address_p): Ditto.
27076 (rs6000_emit_move): Ditto.
27077 (rs6000_init_builtins): Remove V4HI_type_node.
27079 2017-06-13 Martin Liska <mliska@suse.cz>
27082 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27083 (gate_asan): Likewise.
27084 * asan.h (asan_no_sanitize_address_p): Remove the function.
27085 (sanitize_flags_p): New function.
27086 * builtins.def: Fix coding style.
27087 * common.opt: Use renamed enum value.
27088 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27089 * doc/extend.texi: Document no_sanitize attribute.
27090 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27091 to SANITIZE_UNDEFINED_NONDEFAULT.
27092 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27093 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27094 Use sanitize_flags_p.
27095 * gimplify.c (gimplify_function_tree): Likewise.
27096 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27097 * opts.c (parse_no_sanitize_attribute): New function.
27098 (common_handle_option): Use renamed enum value.
27099 * opts.h (parse_no_sanitize_attribute): Declare.
27100 * tree.c (sanitize_flags_p): New function.
27101 * tree.h: Declared here.
27102 * tsan.c: Use sanitize_flags_p.
27103 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27104 (instrument_mem_ref): Likewise.
27105 (instrument_bool_enum_load): Likewise.
27106 (do_ubsan_in_current_function): Remove the function.
27107 (pass_ubsan::execute): Use sanitize_flags_p.
27108 * ubsan.h: Remove do_ubsan_in_current_function
27109 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27110 (dump_function_to_file): Use it here.
27112 2017-06-13 Martin Jambor <mjambor@suse.cz>
27114 PR tree-optimization/80803
27115 PR tree-optimization/81063
27116 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27117 (propagate_subaccesses_across_link): Enqueue subtree whenever
27118 necessary instead of relying on the caller.
27120 2017-06-13 Martin Jambor <mjambor@suse.cz>
27122 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27123 that have a first_link.
27124 (sort_and_splice_var_accesses): Do not check first_link before
27126 (subtree_mark_written_and_enqueue): Likewise.
27127 (propagate_all_subaccesses): Likewise and do not stop at first
27128 parent with a first_link.
27130 2017-06-13 Martin Jambor <mjambor@suse.cz>
27132 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27135 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27137 * match.pd: New pattern.
27139 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27141 * tree-vrp.c (is_masked_range_test): New function.
27142 (register_edge_assert_for): Determine ranges for
27145 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27147 PR tree-optimization/67328
27148 * fold-const.c (maskable_range_p): New function.
27149 (build_range_check): Generate bittests if possible.
27151 2017-06-13 Martin Liska <mliska@suse.cz>
27153 * gimple-pretty-print.c (dump_probability): Add new argument.
27154 (dump_edge_probability): Dump both probability and count.
27155 (dump_gimple_label): Likewise.
27156 (dump_gimple_bb_header): Likewise.
27158 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27161 * config/avr/avr-devices.c: Fix indentation.
27162 * config/avr/gen-avr-mmcu-specs.c: Dito.
27164 2017-06-13 Richard Biener <rguenther@suse.de>
27166 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27167 instead get vector type from stmt_info.
27168 (vectorizable_reduction): Adjust. Remove dead code.
27170 2017-06-13 Richard Biener <rguenther@suse.de>
27172 PR middle-end/81065
27173 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27174 case of C * (x * C2 + C3).
27175 (fold_addr_of_array_ref_difference): Properly fold index difference.
27177 2017-06-12 David S. Miller <davem@davemloft.net>
27180 * config/sparc/sparc.md (return expander): Emit frame blockage if
27181 function uses alloca.
27183 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27185 * combine.c (make_field_assignment): Check len rather than the mode
27186 precision when calling force_to_mode.
27188 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27190 Support multilibs and devices that see flash in RAM address range.
27193 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27194 (avr_mcu_t) <flash_pm_offset>: New field.
27195 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27196 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27197 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27198 (AVR_TINY_PM_OFFSET): Remove macro.
27199 * config/avr/avr.opt (-mshort-calls): New option.
27200 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27201 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27202 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27203 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27204 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27205 instead of avr_arch->have_jmp_call.
27206 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27207 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27208 avr_arch->flash_pm_offset to define.
27209 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27210 new field flash_pm_offset. Add entry for avrxmega3.
27211 (avr_texinfo): Add entry for avrxmega3.
27212 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27213 attiny212, attiny214,
27214 attiny412, attiny414, attiny416, attiny417,
27215 attiny814, attiny816, attiny817,
27216 attiny1614, attiny1616, attiny1617,
27217 attiny3214, attiny3216, attiny3217.
27218 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27219 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27220 (avr_print_operand_address) [AVR_TINY]: Same.
27221 (avr_asm_init_sections) <readonly_data_section>: Only patch
27222 callback if avr_arch->flash_pm_offset = 0.
27223 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27224 for rodata if avr_arch->flash_pm_offset != 0.
27225 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27226 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27227 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27228 (m_options): Append opt_rcall.
27229 (m_dirnames): Append dir_rcall.
27230 * config/avr/t-multilib: Regenerate.
27232 * configure.ac [target=avr]: Check whether avrxmega3 default
27233 linker description file works as needed.
27234 * configure: Regenerate.
27235 * doc/avr-mmcu.texi: Regenerate.
27236 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27237 <__AVR_ARCH__>: Document avrxmega3 and 103.
27238 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27239 <__AVR_SHORT_CALLS__>: Document it.
27240 <__AVR_PM_BASE_ADDRESS__>: Document it.
27241 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27242 (AVR Variable Attributes) <progmem>: Document this is
27243 not needed for avrxmega3.
27244 (AVR Named Address Spaces) <__flash>: Dito.
27246 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27248 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27250 2017-06-12 Doug Rupp <rupp@adacore.com>
27252 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27253 Append vxworks-stdint.h to the tm_file list.
27254 * config/vxworks-stdint.h: New file.
27256 2017-06-12 Martin Liska <mliska@suse.cz>
27258 PR tree-optimization/81041
27259 * tree-profile.c (gimple_gen_ic_func_profiler):
27260 Create an extra BB in profile-generate
27261 (gimple_gen_time_profiler): Likewise.
27263 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27265 PR tree-optimization/81003
27266 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27267 (update_range_test): Use it instead of force_gimple_operand_gsi.
27269 2017-06-12 Richard Biener <rguenther@suse.de>
27271 PR tree-optimization/81053
27272 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27273 with backedge value not defined in loop. Simplify def stmt
27276 2017-06-11 Tom de Vries <tom@codesourcery.com>
27279 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27281 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27282 nvptx_cannot_force_const_mem.
27284 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27286 * opts.c (finish_options): Move test for flag_split_stack after
27287 it has been initialized.
27289 2017-06-11 Jason Merrill <jason@redhat.com>
27291 * tree.h (id_equal): New.
27292 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27293 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27294 instead of strcmp of IDENTIFIER_POINTER.
27296 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27298 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27299 (mark_all_inlined_calls_cdtor): Fix formating.
27300 (inline_transform): Rescale profile before inlining.
27302 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27304 * cgraph.h (cgraph_edge::clone): Update prototype.
27305 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27306 (cgraph_node::create_clone): Update.
27307 (cgraph_node::create_version_clone): Update.
27308 * tree-inline.c (copy_bb): Update.
27309 (expand_call_inline): Update.
27311 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27313 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27314 factored out from ...
27315 (rs6000_emit_prologue): ... here.
27317 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27319 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27320 factored out from ...
27321 (rs6000_emit_prologue): ... here.
27323 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27325 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27327 (handle_missing_profiles): Fix computation of tp_first_run.
27328 (counts_to_freqs): Do not touch freqs when count is 0.
27330 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27332 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27335 2017-06-10 Tom de Vries <tom@codesourcery.com>
27337 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27338 attributes): Document signal effective target.
27340 2017-06-10 Tom de Vries <tom@codesourcery.com>
27342 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27343 Document effective target stack_size.
27345 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27347 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27348 to the edit_context if they can be auto-applied.
27350 2017-06-9 Ian Lance Taylor <iant@golang.org>
27352 * opts.c (finish_options): If -fsplit-stack, disable implicit
27353 -forder-blocks-and-partition.
27354 * doc/invoke.texi (Optimize Options): Document that when using
27355 -fsplit-stack -forder-blocks-and-partition is not implicitly
27358 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27360 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27361 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27362 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27363 * builtins.def (abort, trap, unreachable): Declare cold.
27364 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27365 * tree-core.h (ECF_COLD): New.
27366 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27367 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27369 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27371 * predict.c (unlikely_executed_stmt_p): Cleanup.
27373 2017-06-09 Richard Biener <rguenther@suse.de>
27375 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27376 model if the ref is always written to.
27378 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27380 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27382 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27384 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27387 2017-06-09 Tom de Vries <tom@codesourcery.com>
27390 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27391 "target cannot support label values" when encountering LABEL_REF.
27393 2017-06-09 Martin Liska <mliska@suse.cz>
27395 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27396 (gimple_gen_ic_func_profiler): Emit direct comparison
27397 of __gcov_indirect_call_callee with NULL.
27398 (gimple_gen_time_profiler): Change probability from
27399 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27401 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27403 * profile.c (edge_gcov_counts): Turn to pointer.
27404 (compute_branch_probabilities, compute_branch_probabilities): Update.
27405 (branch_prob): Do not clear edge_gcov_count.
27406 * profile.h (edge_gcov_counts): Turn to pointer.
27407 (edge_gcov_count): Update.
27409 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27411 * gimple.h (gimple_check_failed): Mark cold.
27413 2017-06-09 Richard Biener <rguenther@suse.de>
27415 PR tree-optimization/66623
27416 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27417 refactor check_reduction into two parts, properly computing
27418 whether we have to check reduction validity for outer loop
27421 2017-06-09 Richard Biener <rguenther@suse.de>
27423 PR tree-optimization/79483
27424 * graphite-scop-detection.c (order): New global.
27425 (get_order): Compute bb to order mapping that satisfies code
27426 generation constraints.
27427 (cmp_pbbs): New helper.
27428 (build_scops): Start domwalk at entry block, sort generated
27431 2017-06-09 Richard Biener <rguenther@suse.de>
27433 PR middle-end/81007
27434 * ipa-polymorphic-call.c
27435 (ipa_polymorphic_call_context::restrict_to_inner_class):
27436 Skip FIELD_DECLs with error_mark_node type.
27437 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27440 2017-06-09 Martin Liska <mliska@suse.cz>
27442 * predict.c (struct branch_predictor): New struct.
27443 (test_prediction_value_range): New test.
27444 (predict_c_tests): New function.
27445 * selftest-run-tests.c (selftest::run_tests): Run the function.
27446 * selftest.h: Declare new tests.
27448 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27451 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27452 gen_add3_insn did not fail.
27453 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27454 r0, construct that number in a temporary reg and add that reg to r0.
27455 If asked to put the result in r0 as well, fail.
27457 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27459 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27460 for early expansion of vec_eqv.
27462 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27464 PR middle-end/81005
27465 * ubsan.c (instrument_null): Avoid pointless code temporary.
27466 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27468 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27472 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27475 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27478 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27480 * predict.c (unlikely_executed_edge_p): Move ahead.
27481 (probably_never_executed_edge_p): Use it.
27483 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27485 PR middle-end/79988
27486 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27487 gimple_call_builtin_p call.
27489 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27491 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27492 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27493 rtl_check_failed_type2, rtl_check_failed_code1,
27494 rtl_check_failed_code2, rtl_check_failed_code_mode,
27495 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27496 rtvec_check_failed_bounds, rtl_check_failed_flag,
27497 _fatal_insn_not_found, _fatal_insn): Likewise.
27498 * tree.h (tree_contains_struct_check_failed,
27499 tree_check_failed, tree_not_check_failed,
27500 tree_class_check_failed, tree_range_check_failed,
27501 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27502 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27503 tree_operand_check_failed, omp_clause_check_failed,
27504 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27507 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27509 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27510 flag_branch_probabilities.
27511 * ipa-inline.c (edge_badness): Likewise.
27512 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27513 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27514 * predict.c (maybe_hot_frequency_p): Likewise.
27515 (probably_never_executed): Likewise.
27516 * sched-ebb.c (schedule_ebbs): Likewise.
27517 * sched-rgn.c (find_single_block_region): Likewise.
27518 * tracer.c (tail_duplicate): Likewise.
27520 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27522 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27523 longer requires x_flag_profile_use.
27525 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27527 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27528 instead of flag_reorder_blocks_and_partition.
27529 * dbxout.c (dbxout_function_end): Likewise.
27530 * dwarf2out.c (gen_subprogram_die): Likewise.
27531 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27532 * hw-doloop.c (reorg_loops): Likewise.
27533 * varasm.c (assemble_start_function,
27534 assemble_end_function): Likewise.
27535 (decide_function_section): Do not check for
27536 flag_reorder_blocks_and_partition.
27538 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27540 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27542 (chkp_get_hard_register_fake_addr_expr): Ditto.
27543 (chkp_build_addr_expr): Add check for hard reg case.
27544 (chkp_parse_array_and_component_ref): Ditto.
27545 (chkp_find_bounds_1): Ditto.
27546 (chkp_process_stmt): Don't generate bounds store for
27549 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27551 * predict.c (maybe_hot_bb_p): Do not check profile status.
27552 (maybe_hot_edge_p): Likewise.
27553 (probably_never_executed): Check for zero counts even if profile
27555 (unlikely_executed_edge_p): New function.
27556 (unlikely_executed_stmt_p): New function.
27557 (unlikely_executed_bb_p): New function.
27558 (set_even_probabilities): Use unlikely predicates.
27559 (combine_predictions_for_bb): Likewise.
27560 (predict_paths_for_bb): Likewise.
27561 (predict_paths_leading_to_edge): Likewise.
27562 (determine_unlikely_bbs): New function.
27563 (estimate_bb_frequencies): Use it.
27564 (compute_function_frequency): Use zero counts even if profile is
27566 * profile-count.h: Fix typo.
27568 2017-08-08 Julia Koval <julia.koval@intel.com>
27570 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27571 _mm512_mask_cvtsepi16_storeu_epi8,
27572 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27573 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27574 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27575 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27576 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27577 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27578 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27579 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27580 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27581 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27582 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27583 __builtin_ia32_pmovuswb256mem_mask,
27584 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27585 __builtin_ia32_pmovwb512mem_mask): New builtins.
27587 2017-08-08 Julia Koval <julia.koval@intel.com>
27589 PR target/73350,80862
27590 * config/i386/subst.md (round): Fix round pattern.
27591 * config/i386/i386.c (ix86_erase_embedded_rounding):
27592 Fix erasing rounding for the fixed pattern.
27594 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27596 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27598 2017-06-08 Martin Liska <mliska@suse.cz>
27600 PR gcov-profile/80911
27601 * gcov.c (block_info::block_info): New constructor.
27603 2017-06-07 Carl Love <cel@us.ibm.com>
27605 * config/rs6000/rs6000-c: The return type of the following
27606 built-in functions was implemented as int not long long. Fix sign
27607 of return value for the unsigned version of vec_mulo and vec_mule.
27608 vector unsigned long long vec_bperm (vector unsigned long long,
27609 vector unsigned char)
27610 vector signed long long vec_mule (vector signed int,
27612 vector unsigned long long vec_mule (vector unsigned int,
27613 vector unsigned int)
27614 vector signed long long vec_mulo (vector signed int,
27616 vector unsigned long long vec_mulo (vector unsigned int,
27617 vector unsigned int)
27618 * doc/extend.texi: Fix the documentation for the built-in
27621 2017-06-07 Carl Love <cel@us.ibm.com>
27624 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27627 2017-06-07 Carl Love <cel@us.ibm.com>
27629 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27630 support, Generate doublehv for signed int/float for BE case only.
27632 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27634 * doc/invoke.texi (mcx16): Rewrite.
27636 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27638 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27639 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27640 *mov<mode>_softfloat, and an anonymous splitter): Use
27641 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27643 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27645 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27647 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27648 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27649 (rs6000_debug_reg_global): Adjust.
27650 (rs6000_init_hard_regno_mode_ok): Adjust.
27651 (rs6000_dbx_register_number): Adjust.
27652 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27653 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27654 Remove SPE_ACC and SPEFSCR.
27655 (REG_ALLOC_ORDER): Ditto.
27656 (FRAME_POINTER_REGNUM): Change to 111.
27657 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27658 (REG_CLASS_NAMES): Ditto.
27659 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27660 (REGISTER_NAMES): Ditto.
27661 (ADDITIONAL_REG_NAMES): Ditto.
27662 (rs6000_reg_names): Ditto.
27663 * config/rs6000/rs6000.md: Renumber some register number
27666 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27668 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27670 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27671 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27673 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27674 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27675 Delete the SPE high registers.
27676 (REG_ALLOC_ORDER): Ditto.
27677 (enum reg_class): Remove SPE_HIGH_REGS.
27678 (REG_CLASS_NAMES): Ditto.
27679 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27680 (REGISTER_NAMES): Ditto.
27681 (rs6000_reg_names): Ditto.
27682 * doc/tm.texi.in: Remove SPE as example.
27683 * doc/tm.texi: Regenerate.
27685 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27687 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27688 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27689 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27690 * config/rs6000/rs6000.md (type): Remove "brinc".
27692 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27694 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27695 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27696 * config/rs6000/linuxspe.h: Delete file.
27697 * config/rs6000/rs6000.md: Don't include spe.md.
27698 * config/rs6000/spe.h: Delete file.
27699 * config/rs6000/spe.md: Delete file.
27700 * config/rs6000/t-rs6000: Remove spe.md.
27702 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27704 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27705 (reg_or_none500mem_operand): Delete.
27706 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27707 instead of reg_or_none500mem_operand.
27709 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27711 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27712 handling of SPE flags.
27713 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27715 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27717 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27719 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27720 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27721 paired_divv2sf3): Similar.
27722 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27723 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27724 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27726 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27727 Rename the paired_* instruction patterns.
27728 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27730 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27731 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27732 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27733 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27734 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27735 PAIRED_VECTOR_MODE.
27736 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27737 (spe_func_has_64bit_regs_p): Delete.
27738 (spe_expand_predicate_builtin): Delete.
27739 (spe_expand_evsel_builtin): Delete.
27740 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27741 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27742 (invalid_e500_subreg): Delete.
27743 (rs6000_legitimize_address): Always force_reg op2 as well, for
27744 paired single memory accesses.
27745 (rs6000_member_type_forces_blk): Delete.
27746 (rs6000_spe_function_arg): Delete.
27747 (rs6000_expand_unop_builtin): Delete SPE handling.
27748 (rs6000_expand_binop_builtin): Ditto.
27749 (spe_expand_stv_builtin): Delete.
27750 (bdesc_2arg_spe): Delete.
27751 (spe_expand_builtin): Delete.
27752 (spe_expand_predicate_builtin): Delete.
27753 (spe_expand_evsel_builtin): Delete.
27754 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27755 (spe_init_builtins): Delete.
27756 (spe_func_has_64bit_regs_p): Delete.
27757 (savres_routine_name): Delete "info" parameter. Adjust callers.
27758 (rs6000_emit_stack_reset): Ditto.
27759 (rs6000_dwarf_register_span): Delete.
27760 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27761 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27762 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27764 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27766 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27767 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27768 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27769 mulv2sf3, divv2sf3): Delete expanders.
27771 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27773 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27775 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27777 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27778 * config/rs6000/rs6000.c: Ditto.
27780 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27782 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27783 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27784 comparison_operator.
27786 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27788 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27789 * config/rs6000/rs6000.opt: Ditto.
27790 * config/rs6000/t-rtems: Ditto.
27792 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27794 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27795 TARGET_E500_SINGLE by 0, simplify.
27796 * config/rs6000/rs6000.c: Ditto.
27797 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27798 (spe_build_register_parallel): Delete.
27799 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27800 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27801 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27802 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27803 (E500_CONVERT): Delete.
27804 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27806 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27808 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27809 * config/rs6000/dfp.md: Ditto.
27810 (negdd2, *negdd2_fpr): Merge.
27811 (absdd2, *absdd2_fpr): Merge.
27812 (negtd2, *negtd2_fpr): Merge.
27813 (abstd2, *abstd2_fpr): Merge.
27814 * config/rs6000/e500.h: Delete file.
27815 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27816 TARGET_FPRS by 1 and simplify.
27817 * config/rs6000/rs6000-c.c: Ditto.
27818 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27819 TARGET_DF_SPE by 0.
27820 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27822 * config/rs6000/rs6000.md: Ditto.
27823 (floatdidf2, *floatdidf2_fpr): Merge.
27824 (move_from_CR_gt_bit): Delete.
27825 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27826 (E500_CR_IOR_COMPARE): Delete.
27827 (All patterns that require !TARGET_FPRS): Delete.
27828 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27830 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27832 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27834 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27836 * graphds.c (add_edge): Intitialize edge's attached data.
27837 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27838 pointer parameter. Call pointed function on each edge during
27839 graph traversing. Skip traversing the edge when the function
27841 (graphds_dfs, graphds_scc): Ditto.
27842 (for_each_edge): New parameter. Pass the new parameter to callback
27844 * graphds.h (skip_edge_callback): New function pointer type.
27845 (graphds_dfs, graphds_scc): New function pointer parameter.
27846 (graphds_edge_callback, for_each_edge): New parameter.
27848 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27850 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27851 out code checking if runtime alias check is possible to below ...
27852 Call the new function.
27853 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27854 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27856 2017-06-07 Marek Polacek <polacek@redhat.com>
27859 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27860 TYPE_OVERFLOW_WRAPS check.
27862 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27864 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27865 if versioning is required.
27866 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27867 peeling with the check for versioning.
27869 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27871 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27872 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27873 Set true to new parameter if new ssa variable is defined.
27874 (vect_gen_vector_loop_niters): Refactor. Set range information
27875 for the new vector loop bound variable.
27876 (vect_do_peeling): Ditto.
27878 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27880 * tree-affine.c (ssa.h): Include header file.
27881 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27882 has wrapping overflow behavior.
27884 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27886 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27888 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27890 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27891 (tree_to_aff_combination): ... here.
27893 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27895 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27896 reg_pressure model function.
27897 (ivopts_global_cost_for_size): Delete.
27898 (determine_set_costs, iv_ca_recount_cost): Call new model function
27899 ivopts_estimate_reg_pressure.
27901 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27903 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27904 expensive than udiv. Remove floating point cases from mod.
27906 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27908 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27909 Increase idiv cost.
27911 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27913 * config/aarch64/aarch64.md
27914 (copysignsf3): Fix mask generation.
27916 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27918 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27920 (class dump_manager): Add register_dumps method.
27921 * dumpfile.c: Include langhooks.h.
27922 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27923 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27924 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27925 (dump_manager::dump_register): Start with 512 entries instead of 32.
27926 (dump_manager::register_dumps): New method.
27927 * toplev.c (general_init): Instead of invoking register_dumps
27928 langhook, invoke register_dumps method on the dump manager.
27929 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27932 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27934 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27935 Say that # requires an associated define_split to exist, and that
27936 the define_split must be suitable for use after register allocation.
27938 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27940 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27941 (compute_outgoing_frequencies): Also initialize zero counts.
27942 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27943 around loops; preserve more of profile when nothing changes.
27945 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27947 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27949 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27950 * config/arm/arm-cpu-cdata.h: Regenerate.
27951 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27952 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27953 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27954 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27955 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27957 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27959 * config/arm/t-rmprofile: Likewise.
27960 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27962 2017-06-06 David S. Miller <davem@davemloft.net>
27965 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27966 blockage if function uses alloca.
27968 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27970 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27971 New "uid" fields to hold pretty-print IDs of group and ref.
27972 Memory references are now identified as <group_id>:<ref_id>
27973 instead of using [random] addresses.
27974 (dump_mem_details): Simplify, no functional change.
27975 (dump_mem_ref): Simplify and make output more concise.
27976 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27977 (find_or_create_group): Initialize group uid.
27978 (record_ref): Initialize ref uid. Improve debug output.
27979 (prune_group_by_reuse, should_issue_prefetch_p,)
27980 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27981 (mark_nontemporal_store, determine_loop_nest_reuse):
27982 Improve debug output.
27984 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27986 * dbgcnt.def (prefetch): New debug counter.
27987 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27988 (schedule_prefetches): Stop issueing prefetches if debug counter
27991 2017-06-06 Tom de Vries <tom@codesourcery.com>
27993 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27994 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27996 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27998 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27999 Use aarch64_reg_or_zero predicate for operand 4.
28000 (aarch64_compare_and_swap<mode> define_insn_and_split):
28001 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28002 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28004 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28006 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28007 (arm_compute_save_core_reg_mask): This.
28008 (thumb1_compute_save_reg_mask): Rename into ...
28009 (thumb1_compute_save_core_reg_mask): This.
28010 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28011 (arm_compute_frame_layout): Likewise.
28013 2017-06-06 Richard Biener <rguenther@suse.de>
28015 PR tree-optimization/80974
28016 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28017 keep or clear leaders SSA info.
28019 2017-06-06 Tom de Vries <tom@codesourcery.com>
28021 * config/nvptx/nvptx.c (split_mode_p): New function.
28022 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28024 2017-06-06 Tom de Vries <tom@codesourcery.com>
28026 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28028 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28031 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28034 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28036 * shrink-wrap.c (handle_simple_exit): Update profile.
28037 (try_shrink_wrapping): Upate profile.
28039 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28041 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28042 (tree_guess_outgoing_edge_probabilities): New.
28043 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28044 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28046 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28048 * ipa-split.c (split_function): Initialize return bb profile.
28050 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28052 * profile.c (compute_branch_probabilities): Also initialize
28053 EXIT_BLOCK profile.
28055 2017-06-06 Richard Biener <rguenther@suse.de>
28057 PR tree-optimization/80928
28058 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28059 (vect_analyze_loop_operations): Properly guard analysis for
28061 (vect_transform_loop): Likewise.
28062 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28063 (vect_model_induction_cost): Do not cost for pure SLP.
28064 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28065 of induction in inner loop vectorization.
28066 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28067 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28068 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28070 (vect_analyze_slp_cost_1): Cost induction.
28071 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28072 (vect_get_slp_vect_defs): Likewise.
28073 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28074 (vect_transform_stmt): Handle SLP reductions.
28075 * tree-vectorizer.h (vectorizable_induction): Adjust.
28077 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28079 * config/rs6000/rs6000.c (make_resolver_func): Update
28080 init_lowered_empty_function call.
28082 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28084 * doc/invoke.texi: Document the -fprofile-abs-path option.
28085 * common.opt (fprofile-abs-path): New option.
28086 * gcov-io.h (gcov_write_filename): Declare.
28087 * gcov-io.c (gcov_write_filename): New function.
28088 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28089 * profile.c (output_location): Likewise.
28091 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28093 * shring-wrap.c: Revert accidental commit.
28095 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28097 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28099 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28101 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28103 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28104 profile in callgraph edge.
28105 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28106 (apply_scale): Likewise.
28107 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28108 Also scale profile when inlining function with zero profile.
28109 (initialize_cfun): Update exit block profile even when it is zero.
28110 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28111 when profile is read.
28113 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28115 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28116 (CLONE_*): New constants to define the processors we can generate
28117 code for with the target_clone attribute.
28118 (rs6000_clone_map): New array to identify which clone processors
28119 the current program is running on.
28120 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28121 target_clone attribute.
28122 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28123 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28124 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28125 (cpu_expand_builtin): Add support for target_clone attribute.
28126 (rs6000_valid_attribute_p): Allow "default" attribute.
28127 (get_decl_name): New debug function to simplify printing the
28128 current function name in debugging statements.
28129 (rs6000_clone_priority): New functions to support the target_clone
28130 attribute, and be able to generate code to switch between ISA 2.05
28131 through ISA 3.0 (power6 through power9).
28132 (rs6000_compare_version_priority): Likewise.
28133 (rs6000_get_function_versions_dispatcher): Likewise.
28134 (make_resolver_func): Likewise.
28135 (add_condition_to_bb): Likewise.
28136 (dispatch_function_versions): Likewise.
28137 (rs6000_generate_version_dispatcher_body): Likewise.
28138 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28139 (fusion_gpr_load_p): Fix a spacing issue.
28140 * doc/extend.texi (Common Function Attributes): Document that the
28141 PowerPC supports the target_clone attribute.
28143 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28145 * config/arm/arm.h: explain F symbol found in description of ARM
28146 register allocation in its legend.
28148 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28150 * config/mips/frame-header-opt.c: Include profile-count.h.
28151 * config/riscv/riscv.c: Include profile-count.h
28153 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28155 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28157 (sm_set_flag_if_changed): Add bbs field.
28158 (execute_sm_if_changed_flag_set): Pass BBS.
28159 (execute_sm): Update.
28161 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28163 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28166 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28168 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28169 (peephole2): New peephole2 to emit the above.
28170 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28172 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28174 * config/aarch64/aarch64.c (define_peephole2 above
28175 *sub_<shift>_<mode>): New peephole.
28177 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28179 * config/i386/i386.c (make_resolver_func): Update.
28180 * Makefile.in: Add profile-count.h and profile-count.o
28181 * auto-profile.c (afdo_indirect_call): Update to new API.
28182 (afdo_set_bb_count): Update.
28183 (afdo_propagate_edge): Update.
28184 (afdo_propagate_circuit): Update.
28185 (afdo_calculate_branch_prob): Update.
28186 (afdo_annotate_cfg): Update.
28187 * basic-block.h: Include profile-count.h
28188 (struct edge_def): Turn count to profile_count.
28189 (struct basic_block_def): Likewie.
28190 (REG_BR_PROB_BASE): Move to profile-count.h
28191 (RDIV): Move to profile-count.h
28192 * bb-reorder.c (max_entry_count): Turn to profile_count.
28193 (find_traces): Update.
28194 (rotate_loop):Update.
28195 (connect_traces):Update.
28196 (sanitize_hot_paths):Update.
28197 * bt-load.c (migrate_btr_defs): Update.
28198 * cfg.c (RDIV): Remove.
28199 (init_flow): Use alloc_block.
28200 (alloc_block): Uninitialize count.
28201 (unchecked_make_edge): Uninitialize count.
28202 (check_bb_profile): Update.
28203 (dump_edge_info): Update.
28204 (dump_bb_info): Update.
28205 (update_bb_profile_for_threading): Update.
28206 (scale_bbs_frequencies_int): Update.
28207 (scale_bbs_frequencies_gcov_type): Update.
28208 (scale_bbs_frequencies_profile_count): New.
28209 * cfg.h (update_bb_profile_for_threading): Update.
28210 (scale_bbs_frequencies_profile_count): Declare.
28211 * cfgbuild.c (compute_outgoing_frequencies): Update.
28212 (find_many_sub_basic_blocks): Update.
28213 * cfgcleanup.c (try_forward_edges): Update.
28214 (try_crossjump_to_edge): Update.
28215 * cfgexpand.c (expand_gimple_tailcall): Update.
28216 (construct_exit_block): Update.
28217 * cfghooks.c (verify_flow_info): Update.
28218 (dump_bb_for_graph): Update.
28219 (split_edge): Update.
28220 (make_forwarder_block): Update.
28221 (duplicate_block): Update.
28222 (account_profile_record): Update.
28223 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28224 (get_estimated_loop_iterations): Update.
28225 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28226 (single_likely_exit): Update.
28227 * cfgloopmanip.c (scale_loop_profile): Update.
28229 (set_zero_probability): Update.
28230 (lv_adjust_loop_entry_edge): Update.
28231 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28232 (purge_dead_edges): Update.
28233 (rtl_account_profile_record): Update.
28234 * cgraph.c (cgraph_node::create): Uninitialize count.
28235 (symbol_table::create_edge): Uninitialize count.
28236 (cgraph_update_edges_for_call_stmt_node): Update.
28237 (cgraph_edge::dump_edge_flags): Update.
28238 (cgraph_node::dump): Update.
28239 (cgraph_edge::maybe_hot_p): Update.
28240 * cgraph.h: Include profile-count.h
28241 (create_clone), create_edge, create_indirect_edge): Update.
28242 (cgraph_node): Turn count to profile_count.
28243 (cgraph_edge0: Likewise.
28244 (make_speculative, clone): Update.
28245 (create_edge): Update.
28246 (init_lowered_empty_function): Update.
28247 * cgraphclones.c (cgraph_edge::clone): Update.
28248 (duplicate_thunk_for_node): Update.
28249 (cgraph_node::create_clone): Update.
28250 * cgraphunit.c (cgraph_node::analyze): Update.
28251 (cgraph_node::expand_thunk): Update.
28252 * final.c (dump_basic_block_info): Update.
28253 * gimple-streamer-in.c (input_bb): Update.
28254 * gimple-streamer-out.c (output_bb): Update.
28255 * graphite.c (print_global_statistics): Update.
28256 (print_graphite_scop_statistics): Update.
28257 * hsa-brig.c: Include basic-block.h.
28258 * hsa-dump.c: Include basic-block.h.
28259 * hsa-gen.c (T sum_slice): Update.
28260 (convert_switch_statements):Update.
28261 * hsa-regalloc.c: Include basic-block.h.
28262 * ipa-chkp.c (chkp_produce_thunks): Update.
28263 * ipa-cp.c (struct caller_statistics): Update.
28264 (init_caller_stats): Update.
28265 (gather_caller_stats): Update.
28266 (ipcp_cloning_candidate_p): Update.
28267 (good_cloning_opportunity_p): Update.
28268 (get_info_about_necessary_edges): Update.
28269 (dump_profile_updates): Update.
28270 (update_profiling_info): Update.
28271 (update_specialized_profile): Update.
28272 (perhaps_add_new_callers): Update.
28273 (decide_about_value): Update.
28274 (ipa_cp_c_finalize): Update.
28275 * ipa-devirt.c (struct odr_type_warn_count): Update.
28276 (struct decl_warn_count): Update.
28277 (struct final_warning_record): Update.
28278 (possible_polymorphic_call_targets): Update.
28279 (ipa_devirt): Update.
28280 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28281 * ipa-icf.c (sem_function::merge): Update.
28282 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28283 * ipa-inline.c (compute_uninlined_call_time): Update.
28284 (compute_inlined_call_time): Update.
28285 (want_inline_small_function_p): Update.
28286 (want_inline_self_recursive_call_p): Update.
28287 (edge_badness): Update.
28288 (lookup_recursive_calls): Update.
28289 (recursive_inlining): Update.
28290 (inline_small_functions): Update.
28291 (dump_overall_stats): Update.
28292 (dump_inline_stats): Update.
28293 * ipa-profile.c (ipa_profile_generate_summary): Update.
28294 (ipa_propagate_frequency): Update.
28295 (ipa_profile): Update.
28296 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28297 * ipa-utils.c (ipa_merge_profiles): Update.
28298 * loop-doloop.c (doloop_modify): Update.
28299 * loop-unroll.c (report_unroll): Update.
28300 (unroll_loop_runtime_iterations): Update.
28301 * lto-cgraph.c (lto_output_edge): Update.
28302 (lto_output_node): Update.
28303 (input_node): Update.
28304 (input_edge): Update.
28305 (merge_profile_summaries): Update.
28306 * lto-streamer-in.c (input_cfg): Update.
28307 * lto-streamer-out.c (output_cfg): Update.
28308 * mcf.c (create_fixup_graph): Update.
28309 (adjust_cfg_counts): Update.
28310 (sum_edge_counts): Update.
28311 * modulo-sched.c (sms_schedule): Update.
28312 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28313 * predict.c (maybe_hot_count_p): Update.
28314 (probably_never_executed): Update.
28315 (dump_prediction): Update.
28316 (combine_predictions_for_bb): Update.
28317 (propagate_freq): Update.
28318 (handle_missing_profiles): Update.
28319 (counts_to_freqs): Update.
28320 (rebuild_frequencies): Update.
28321 (force_edge_cold): Update.
28322 * predict.h: Include profile-count.h
28323 (maybe_hot_count_p, counts_to_freqs): UPdate.
28324 * print-rtl-function.c: Do not include cfg.h
28325 * print-rtl.c: Include basic-block.h
28326 * profile-count.c: New file.
28327 * profile-count.h: New file.
28328 * profile.c (is_edge_inconsistent): Update.
28329 (correct_negative_edge_counts): Update.
28330 (is_inconsistent): Update.
28331 (set_bb_counts): Update.
28332 (read_profile_edge_counts): Update.
28333 (compute_frequency_overlap): Update.
28334 (compute_branch_probabilities): Update; Initialize and deinitialize
28336 (branch_prob): Update.
28337 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28338 (edge_gcov_count): New.
28339 (bb_gcov_count): New.
28340 * shrink-wrap.c (try_shrink_wrapping): Update.
28341 * tracer.c (better_p): Update.
28342 * trans-mem.c (expand_transaction): Update.
28343 (ipa_tm_insert_irr_call): Update.
28344 (ipa_tm_insert_gettmclone_call): Update.
28345 * tree-call-cdce.c: Update.
28346 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28347 (gimple_duplicate_sese_tail): Update.
28348 (gimple_account_profile_record): Update.
28349 (execute_fixup_cfg): Update.
28350 * tree-inline.c (copy_bb): Update.
28351 (copy_edges_for_bb): Update.
28352 (initialize_cfun): Update.
28353 (freqs_to_counts): Update.
28354 (copy_cfg_body): Update.
28355 (expand_call_inline): Update.
28356 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28357 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28358 (try_unroll_loop_completely): Update.
28359 (try_peel_loop): Update.
28360 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28361 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28362 * tree-ssa-loop-split.c (connect_loops): Update.
28363 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28364 * tree-ssa-reassoc.c (branch_fixup): Update.
28365 * tree-ssa-tail-merge.c (replace_block_by): Update.
28366 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28367 (compute_path_counts): Update.
28368 (update_profile): Update.
28369 (recompute_probabilities): Update.
28370 (update_joiner_offpath_counts): Update.
28371 (estimated_freqs_path): Update.
28372 (freqs_to_counts_path): Update.
28373 (clear_counts_path): Update.
28374 (ssa_fix_duplicate_block_edges): Update.
28375 (duplicate_thread_path): Update.
28376 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28377 (struct switch_conv_info): Update.
28378 * tree-tailcall.c (decrease_profile): Update.
28379 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28380 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28381 * value-prof.c (check_counter): Update.
28382 (gimple_divmod_fixed_value): Update.
28383 (gimple_mod_pow2): Update.
28384 (gimple_mod_subtract): Update.
28385 (gimple_ic_transform): Update.
28386 (gimple_stringop_fixed_value): Update.
28387 * value-prof.h (gimple_ic): Update.
28389 2017-06-02 Carl Love <cel@us.ibm.com>
28391 * config/rs6000/rs6000-c: Add support for built-in functions
28392 vector double vec_doublee (vector signed int);
28393 vector double vec_doublee (vector unsigned int);
28394 vector double vec_doublee (vector float);
28395 vector double vec_doubleh (vector signed int);
28396 vector double vec_doubleh (vector unsigned int);
28397 vector double vec_doubleh (vector float);
28398 vector double vec_doublel (vector signed int);
28399 vector double vec_doublel (vector unsigned int);
28400 vector double vec_doublel (vector float);
28401 vector double vec_doubleo (vector signed int);
28402 vector double vec_doubleo (vector unsigned int);
28403 vector double vec_doubleo (vector float);.
28404 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28405 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28407 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28408 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28409 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28411 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28412 vec_doublel, vec_doubleh.
28413 * doc/extend.texi: Update the built-in documentation file for the
28414 new built-in functions.
28416 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28419 * ipa-inline-analysis.c (free_growth_caches): Set
28420 edge_removal_hook_holder to NULL after removing it.
28422 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28424 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28425 comparision with zero.
28427 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28428 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28429 for early expansion of vec_min and vec_max builtins.
28430 (builtin_function_type): Add min/max unsigned variants to those
28431 identified as having unsigned arguments.
28433 2017-06-02 Olivier Hainque <hainque@adacore.com>
28435 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28437 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28439 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28440 Use VALL_F16 iterator rather than VALL.
28442 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28444 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28445 Emit CBNZ inside loop when doing a strong exchange and comparing
28446 against zero. Generate the CC flags after the loop.
28448 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28450 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28451 (dl_section_ref): New.
28452 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28453 On AIX, append an expression to subtract the size of the
28454 section length to dl_section_ref.
28456 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28458 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28459 for early expansion of vector absolute builtins.
28461 2017-06-02 Richard Biener <rguenther@suse.de>
28463 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28464 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28466 2017-06-02 Richard Biener <rguenther@suse.de>
28468 PR tree-optimization/80948
28469 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28470 stmt order as well.
28472 2017-06-02 Richard Biener <rguenther@suse.de>
28474 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28476 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28477 for inductions as relevant.
28479 2017-06-02 Richard Biener <rguenther@suse.de>
28481 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28482 (vectorizable_induction): ... this. Remove dead code.
28484 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28486 * builtins. (expand_builtin_alloca): Remove second parameter and
28487 infer its value from the first parameter instead.
28488 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28490 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28492 PR rtl-optimization/80903
28493 * loop-doloop.c (add_test): Unshare sequence.
28495 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28497 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28499 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28501 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28503 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28504 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28505 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28506 (xlogue_layout::get_stub_rtx): Make static.
28507 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28508 (xlogue_layout::compute_stub_managed_regs): Rename to...
28509 (xlogue_layout::count_stub_managed_regs): ...this.
28510 (xlogue_layout::is_stub_managed_reg): New function.
28511 (xlogue_layout::m_stub_names): Rename to...
28512 (xlogue_layout::s_stub_names): ...this, make static.
28513 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28514 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28515 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28516 xlogue_layout::s_stub_names): Instantiate statics.
28517 (stub_managed_regs): Remove.
28518 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28519 (disable_call_ms2sysv_xlogues): Rename to...
28520 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28521 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28523 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28524 change after reload_completed.
28525 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28527 (ix86_expand_prologue): Likewise.
28528 (ix86_expand_epilogue): Likewise.
28529 (ix86_expand_split_stack_prologue): Likewise.
28530 (ix86_compute_frame_layout): Remove frame parameter ...
28531 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28532 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28534 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28535 (ix86_frame): Move from here ...
28536 * config/i386/i386.h (ix86_frame): ... to here.
28537 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28538 complete ix86_frame data structure instead. Remove some_ld_name.
28540 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28542 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28543 symbols that hold a DECL_VALUE_EXPR.
28545 2017-06-01 Martin Jambor <mjambor@suse.cz>
28547 PR tree-optimization/80898
28548 * tree-sra.c (process_subtree_disqualification): Removed.
28549 (disqualify_candidate): Do not acll
28550 process_subtree_disqualification.
28551 (subtree_mark_written_and_enqueue): New function.
28552 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28553 RHS has been disqualified and re-queue LHS if necessary. Apart
28554 from that, ignore disqualified RHS.
28556 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28558 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28559 address fetch for z10 or later.
28561 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28563 * config/arc/arc.md (tst_movb): Add guard when splitting.
28565 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28567 * config/arc/arc.c (arc_can_eliminate): Test against
28568 arc_frame_pointer_needed.
28570 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28572 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28573 to prevent store reordering.
28574 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28575 (type): Add block type.
28576 (stack_tie): Define special instruction to be used in
28579 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28581 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28582 constraint. It is not valid for the pattern.
28583 (noncommutative_binary_comparison): Likewise.
28585 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28587 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28590 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28592 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28593 be used by the reg-alloc.
28595 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28597 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28598 reg-alloc when having mul64 or mul32x16 instructions.
28599 (mulsidi3): Likewise.
28600 (umulsidi3): Likewise.
28601 (mulsi32x16): New pattern.
28602 (mulsi64): Likewise.
28603 (mulsidi64): Likewise.
28604 (umulsidi64): Likewise.
28605 (MUL32x16_REG): Define.
28606 (mul64_600): Use MUL32x16_REG.
28607 (mac64_600): Likewise.
28608 (umul64_600): Likewise.
28609 (umac64_600): Likewise.
28611 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28613 * config/arc/arc.md (mulsi3_700): Make it commutative.
28615 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28617 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28619 (*sign_extendsidi2_insn): Likewise for movstosw.
28621 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28623 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28624 the type of the input discriminant value. Convert the
28625 discriminant value of signedness vary.
28627 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28629 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28630 Add to -Wall section.
28632 2017-06-01 Richard Biener <rguenther@suse.de>
28634 PR middle-end/66313
28635 * fold-const.c (fold_plusminus_mult_expr): If the factored
28636 factor may be zero use a wrapping type for the inner operation.
28637 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28638 and handle moved defs.
28639 (process_assignment): Properly guard the unary op case. Return a
28640 tri-state indicating that moving the stmt before the call may allow
28641 to continue. Pass through to_move.
28642 (find_tail_calls): Handle moving unrelated defs before
28645 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28648 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28649 splitter result in the canonical way.
28651 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28653 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28654 also for 32bit target. Update insn attributes.
28655 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28657 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28659 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28660 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28661 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28662 (_mm_maskz_min_ss): New intrinsics.
28664 2017-05-31 Martin Liska <mliska@suse.cz>
28666 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28667 Change comment style to one we normally use.
28668 (vectorizable_reduction): Likewise.
28669 (vectorizable_induction): Likewise.
28670 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28671 (vectorizable_call): Likewise.
28672 (vectorizable_simd_clone_call): Likewise.
28673 (vectorizable_conversion): Likewise.
28674 (vectorizable_assignment): Likewise.
28675 (vectorizable_shift): Likewise.
28676 (vectorizable_operation): Likewise.
28677 (vectorizable_store): Likewise.
28678 (vectorizable_load): Likewise.
28679 * tree-vectorizer.h: Likewise.
28681 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28683 * passes.c (emergency_dump_function): New.
28684 * tree-pass.h (emergency_dump_function): Declare.
28685 * plugin.c (plugins_internal_error_function): Remove.
28686 * plugin.h (plugins_internal_error_function): Remove declaration.
28687 * toplev.c (internal_error_function): New static function. Use it...
28688 (general_init): ...here.
28690 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28692 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28693 (arc_rtx_costs): Add costs for new patterns.
28694 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28695 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28697 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28699 * tree-ssa-strlen.c (get_next_strinfo): New function.
28700 (get_stridx_plus_constant): Use it.
28701 (zero_length_string): Likewise.
28702 (adjust_related_strinfos): Likewise.
28703 (adjust_last_stmt): Likewise.
28705 2017-05-31 Richard Biener <rguenther@suse.de>
28708 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28709 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28711 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28713 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28714 loop_vinfo argument and use of dependence distance vectors.
28715 Check instead whether the two references differ only in their
28716 initial value and assume that they have the same alignment if the
28717 difference is a multiple of the vector alignment.
28718 (vect_analyze_data_refs_alignment): Update call accordingly.
28720 2017-05-31 Martin Liska <mliska@suse.cz>
28723 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28725 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28727 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28728 (create_intersect_range_checks): Move from ...
28729 * tree-data-ref.c (create_intersect_range_checks_index)
28730 (create_intersect_range_checks): ... to here.
28731 (create_runtime_alias_checks): New function factored from ...
28732 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28733 here. Call above function.
28734 * tree-data-ref.h (create_runtime_alias_checks): New function.
28736 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28738 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28739 segment length for dr_b and compute it in wide_int.
28741 2017-05-31 Richard Biener <rguenther@suse.de>
28743 PR tree-optimization/80906
28744 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28745 and pass through iv_map.
28746 (copy_bb_and_scalar_dependences): Adjust.
28747 (translate_pending_phi_nodes): Likewise.
28748 (copy_loop_close_phi_args): Handle code-generating IVs instead
28751 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28753 * diagnostic-color.c (color_dict): Add "type-diff".
28754 (parse_gcc_colors): Update comment.
28755 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28756 -fdiagnostics-show-template-tree and -fno-elide-type.
28757 (GCC_COLORS): Add type-diff to example.
28759 (-fdiagnostics-show-template-tree): New.
28760 (-fno-elide-type): New.
28761 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28762 the pp_format_decoder callback. Call any m_format_postprocessor's
28764 (pretty_printer::pretty_printer): Initialize
28765 m_format_postprocessor.
28766 (pretty_printer::~pretty_printer): Delete any
28767 m_format_postprocessor.
28768 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28769 (class format_postprocessor): New class.
28770 (struct pretty_printer::format_decoder): Document the new parameters.
28771 (struct pretty_printer::m_format_postprocessor): New field.
28772 * tree-diagnostic.c (default_tree_printer): Update for new
28773 bool and const char ** params.
28774 * tree-diagnostic.h (default_tree_printer): Likewise.
28776 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28778 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28779 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28780 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28781 rs6000_gen_cell_microcode code.
28782 (rs6000_final_prescan_insn): Delete.
28783 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28784 "warn-cell-microcode" entries.
28785 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28786 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28787 throughout. Change cc_reg_not_micro_cr0_operand to
28788 cc_reg_not_cr0_operand throughout.
28789 (*extendhi<mode>2_noload): Delete.
28790 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28791 (mwarn-cell-microcode): Delete.
28792 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28793 -mgen-cell-microcode and -mwarn-cell-microcode.
28795 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28798 * config/i386/constraints.md (Yd): New constraint.
28800 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28801 and (?Yd, r) alternatives. Update insn attributes.
28802 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28803 and (?*Yd, r) alternatives. Update insn attributes.
28804 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28806 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28808 * gimplify.c (gimplify_modify_expr): Don't create a
28809 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28812 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28814 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28816 2017-05-30 Richard Biener <rguenther@suse.de>
28818 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28819 and reduc_def fields.
28820 (STMT_VINFO_REDUC_TYPE): New define.
28821 (STMT_VINFO_REDUC_DEF): Likewise.
28822 (vect_force_simple_reduction): Adjust prototype.
28823 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28824 (vect_is_simple_reduction): Remove check_reduction argument.
28825 (vect_force_simple_reduction): Adjust and set
28826 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28827 (vectorizable_reduction): Do not re-do reduction analysis
28828 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28829 * tree-parloops.c (gather_scalar_reductions): Adjust.
28831 2017-05-30 Richard Biener <rguenther@suse.de>
28833 PR middle-end/80901
28834 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28837 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28839 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28840 Introduce unknown_misalignment parameter and remove vf.
28841 (vect_peeling_hash_get_lowest_cost):
28842 Pass unknown_misalignment parameter.
28843 (vect_enhance_data_refs_alignment):
28844 Fix unsupportable data ref treatment.
28846 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28848 * tree-vect-data-refs.c (vect_get_data_access_cost):
28849 Workaround for SLP handling.
28850 (vect_enhance_data_refs_alignment):
28851 Compute costs for doing no peeling at all, compare to the best
28852 peeling costs so far and avoid peeling if cheaper.
28854 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28856 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28857 Return peeling info and set costs to zero for unlimited cost
28859 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28860 with unknown misalignment. Compute and costs for unknown
28861 misalignment, compare them to the costs for known misalignment
28862 and choose the cheapest for peeling.
28864 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28866 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28867 (vect_get_peeling_costs_all_drs): Create function.
28868 (vect_peeling_hash_get_lowest_cost):
28869 Use vect_get_peeling_costs_all_drs.
28870 (vect_peeling_supportable): Create function.
28871 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28873 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28875 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28876 DR_HAS_NEGATIVE_STEP.
28877 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28878 (vect_enhance_data_refs_alignment): Use.
28879 (vect_duplicate_ssa_name_ptr_info): Use.
28880 * tree-vectorizer.h (dr_misalignment): Use.
28881 (known_alignment_for_access_p): Use.
28883 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28886 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28888 (has_section_name): New function.
28890 2017-05-30 Martin Liska <mliska@suse.cz>
28893 * auto-profile.c (get_function_decl_from_block): Fix
28896 2017-05-30 Richard Biener <rguenther@suse.de>
28898 PR middle-end/80876
28899 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28901 2017-05-30 Martin Liska <mliska@suse.cz>
28903 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28904 * dumpfile.h (struct dump_file_info): Remove ctors.
28906 2017-05-30 Martin Liska <mliska@suse.cz>
28908 * predict.def: Fix GNU coding style.
28910 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28912 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28913 Mark 'to' argument with ATTRIBUTE_UNUSED.
28915 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28917 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28918 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28919 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28922 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28924 * doc/install.texi (Options specification): Restore entry of
28925 --enable-sjlj-exceptions.
28927 2017-05-27 Michael Eager <eager@eagercon.com>
28930 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28932 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28934 * config/microblaze/microblaze.h
28935 (FIXED_REGISTERS): Update in macro.
28936 (CALL_USED_REGISTERS): Update in macro.
28938 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28940 * doc/install.texi: Add links to macOS binary distributions.
28942 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28946 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28948 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28950 2017-05-26 Martin Liska <mliska@suse.cz>
28952 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28954 2017-05-26 Martin Liska <mliska@suse.cz>
28956 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28957 always leading ';; '.
28958 (dump_bb_info): Likewise.
28959 (brief_dump_cfg): Likewise.
28960 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28961 * dumpfile.c: Remove usage of TDF_VERBOSE.
28962 * dumpfile.h (enum dump_kind): Likewise.
28963 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28964 * print-tree.c (debug_verbose): Remove.
28965 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28966 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28967 * tree-diagnostic.c (default_tree_printer): Replace
28968 TDF_DIAGNOSTIC with TDF_SLIM.
28970 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28972 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28973 in parameter loop, rather than loop_vinfo.
28974 (create_intersect_range_checks): Ditto.
28975 (vect_create_cond_for_alias_checks): Update call to above functions.
28977 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28979 PR tree-optimization/80815
28980 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28981 for merging runtime alias checks. Handle negative DR_STEPs.
28983 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28985 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28987 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28988 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28989 out code pruning runtime alias checks.
28990 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28991 factored out from above.
28992 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28994 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28996 (prune_runtime_alias_test_list): New decalaration.
28998 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29000 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29001 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29002 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29003 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29004 (operator==, comp_dr_with_seg_len_pair): Ditto.
29005 (vect_prune_runtime_alias_test_list): Ditto.
29007 2017-05-26 Martin Liska <mliska@suse.cz>
29010 * params.def: Bound partial-inlining-entry-probability param.
29012 2017-05-26 Marek Polacek <polacek@redhat.com>
29015 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29018 2017-05-26 Richard Biener <rguenther@suse.de>
29020 PR tree-optimization/80842
29021 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29024 2017-05-26 Richard Biener <rguenther@suse.de>
29026 PR tree-optimization/80844
29027 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29029 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29031 * doc/md.texi (Machine Constraints): Update x86 family
29032 machine constraints section to match 'config/i386/constraints.md'.
29034 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29036 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29038 2017-05-25 Nathan Sidwell <nathan@acm.org>
29040 * doc/invoke.texi (--enable-languages): Update documentation.
29042 2017-05-25 Martin Liska <mliska@suse.cz>
29044 * dumpfile.c: Add TDF_FOLDING.
29045 * dumpfile.h (enum dump_kind): Likewise.
29046 * genmatch.c (dt_simplify::gen_1): Use it.
29048 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29050 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29052 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29054 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29055 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29057 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29059 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29060 * match.pd (X == C): Rewrite it here.
29061 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29062 with_certain_nonzero_bits2): New predicates.
29063 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29065 2017-05-24 Nathan Sidwell <nathan@acm.org>
29067 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29070 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29073 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29075 * config/powerpcspe: New port. Files are copied from the rs6000
29076 port, with "rs6000" in filenames replaced by "powerpcspe".
29078 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29080 PR rtl-optimization/80754
29081 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29083 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29085 * config/sparc/sparc.md (length): Return the correct value for -mflat
29086 sibcalls to match output_sibcall.
29088 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29092 * config/rs6000/rs6000.c (struct machine_function): Add new field
29094 (rs6000_get_separate_components): Init that field, use it.
29095 (rs6000_components_for_bb): Use the field.
29097 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29099 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29101 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29103 PR middle-end/80823
29104 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29106 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29109 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29110 address against address_operand predicate.
29111 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29113 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29115 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29116 parameters passed indirectly.
29118 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29120 * config/i386/i386.md (*movdi_internal): Remove SSE4
29121 alternative 18 (?r, *v). Update insn attributes.
29122 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29123 Update insn attributes.
29124 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29125 Update insn attributes.
29126 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29127 alternative 1 (r, v). Remove isa attribute.
29128 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29129 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29130 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29132 2017-05-23 Tom de Vries <tom@codesourcery.com>
29134 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29137 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29139 * cgraphunit.c (symbol_table::process_new_functions): Update.
29140 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29141 (inline_generate_summary): Rename to ...
29142 (ipa_fn_summary_generate): ... this one.
29143 (inline_read_summary): Rename to ...
29144 (ipa_fn_summary_read): ... this one.
29145 (inline_write_summary): Rename to ...
29146 (ipa_fn_summary_write): ... this one.
29147 (inline_free_summary): Rename to ...
29148 (ipa_free_fn_summary): ... this one.
29149 (pass_data_local_fn_summary, pass_local_fn_summary,
29150 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29151 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29152 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29153 make_pass_ipa_fn_summary): New.
29154 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29155 inline_write_summary, inline_free_summary): Remove.
29156 (ipa_free_fn_summary) : New.
29157 * ipa-inline.c (ipa_inline): Update.
29158 (pass_ipa_inline): Do not generate summaries.
29159 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29161 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29162 and add pass_ipa_fn_summary.
29163 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29165 (make_pass_inline_parameters): Remove.
29167 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29169 * omp-low.c (struct omp_context): Remove "default_kind" member.
29172 * omp-offload.c (execute_oacc_device_lower): Remove the
29173 parallelism dimensions function attributes for unparallelized
29174 OpenACC kernels constructs.
29176 2017-05-23 Martin Liska <mliska@suse.cz>
29178 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29180 (cgraph_edge::make_speculative): Likewise.
29181 (cgraph_edge::resolve_speculation): Likewise.
29182 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29183 (cgraph_node::dump): Likewise.
29184 * cgraph.h: Likewise.
29185 * cgraphunit.c (analyze_functions): Likewise.
29186 (symbol_table::compile): Likewise.
29187 * ipa-cp.c (print_all_lattices): Likewise.
29188 (determine_versionability): Likewise.
29189 (initialize_node_lattices): Likewise.
29190 (ipcp_verify_propagated_values): Likewise.
29191 (estimate_local_effects): Likewise.
29192 (update_profiling_info): Likewise.
29193 (create_specialized_node): Likewise.
29194 (perhaps_add_new_callers): Likewise.
29195 (decide_about_value): Likewise.
29196 (decide_whether_version_node): Likewise.
29197 (identify_dead_nodes): Likewise.
29198 (ipcp_store_bits_results): Likewise.
29199 * ipa-devirt.c (dump_targets): Likewise.
29200 (ipa_devirt): Likewise.
29201 * ipa-icf.c (sem_item::dump): Likewise.
29202 (sem_function::equals): Likewise.
29203 (sem_variable::equals): Likewise.
29204 (sem_item_optimizer::read_section): Likewise.
29205 (sem_item_optimizer::execute): Likewise.
29206 (congruence_class::dump): Likewise.
29207 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29208 (dump_inline_summary): Likewise.
29209 (estimate_node_size_and_time): Likewise.
29210 (inline_analyze_function): Likewise.
29211 * ipa-inline-transform.c (inline_call): Likewise.
29212 * ipa-inline.c (report_inline_failed_reason): Likewise.
29213 (want_early_inline_function_p): Likewise.
29214 (edge_badness): Likewise.
29215 (update_edge_key): Likewise.
29216 (inline_small_functions): Likewise.
29217 * ipa-profile.c (ipa_profile): Likewise.
29218 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29219 (ipa_make_edge_direct_to_target): Likewise.
29220 (remove_described_reference): Likewise.
29221 (ipa_impossible_devirt_target): Likewise.
29222 (propagate_controlled_uses): Likewise.
29223 (ipa_print_node_params): Likewise.
29224 (ipcp_transform_function): Likewise.
29225 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29226 (propagate_pure_const): Likewise.
29227 * ipa-reference.c (generate_summary): Likewise.
29228 (read_write_all_from_decl): Likewise.
29229 (propagate): Likewise.
29230 (ipa_reference_read_optimization_summary): Likewise.
29231 * ipa-utils.c (ipa_merge_profiles): Likewise.
29232 * ipa.c (walk_polymorphic_call_targets): Likewise.
29233 (symbol_table::remove_unreachable_nodes): Likewise.
29234 (ipa_single_use): Likewise.
29235 * passes.c (execute_todo): Likewise.
29236 * predict.c (drop_profile): Likewise.
29237 * symtab.c (symtab_node::get_dump_name): New function.
29238 (symtab_node::dump_name): Likewise.
29239 (symtab_node::dump_asm_name): Likewise.
29240 (symtab_node::dump_references): Likewise.
29241 (symtab_node::dump_referring): Likewise.
29242 (symtab_node::dump_base): Likewise.
29243 (symtab_node::debug_symtab): Likewise.
29244 * tree-sra.c (convert_callers_for_node): Likewise.
29245 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29246 * value-prof.c (init_node_map): Likewise.
29248 2017-05-23 Martin Liska <mliska@suse.cz>
29250 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29251 and symtab_node::debug_symtab to symbol_table::debug.
29252 * cgraphunit.c (analyze_functions): Use the renamed function.
29253 (symbol_table::compile): Likewise.
29254 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29255 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29256 * passes.c (execute_todo): Likewise.
29257 * symtab.c (symbol_table::dump): New function.
29258 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29260 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29262 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29263 that nonconst implies exec.
29265 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29267 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29268 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29269 (inline_edge_summary_vec): Turn into ...
29270 (ipa_call_summaries): ... this one.
29271 (redirect_to_unreachable, edge_set_predicate,
29272 evaluate_properties_for_edge, inline_summary_alloc,
29273 reset_ipa_call_summary, reset_inline_summary,
29274 inline_summary_t::duplicate): Update.
29275 (inline_edge_duplication_hook): Turn to ...
29276 (ipa_call_summary_t::duplicate): ... this one.
29277 (inline_edge_removal_hook): Turn to ...
29278 (ipa_call_summary_t::remove): ... this one.
29279 (dump_inline_edge_summary): Turn to ...
29280 (dump_ipa_call_summary): ... this one.
29281 (estimate_function_body_sizes): Update.
29282 (inline_update_callee_summaries): Update.
29283 (remap_edge_change_prob): Update.
29284 (remap_edge_summaries): Update.
29285 (inline_merge_summary): Update.
29286 (do_estimate_edge_time): Update.
29287 (inline_generate_summary): Update.
29288 (inline_read_section): Update.
29289 (inline_read_summary): Update.
29290 (inline_free_summary): Update.
29291 * ipa-inline.c (can_inline_edge_p): Update.
29292 (compute_inlined_call_time): Update.
29293 (want_inline_small_function_p): Update.
29294 (edge_badness): Update.
29295 (early_inliner): Update.
29296 * ipa-inline.h (inline_edge_summary): Turn to ...
29297 (ipa_call_summary): ... this one.
29298 (ipa_call_summary_t): New class.
29299 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29300 (ipa_call_summaries): New.
29301 (inline_edge_summary): Remove.
29302 (estimate_edge_growth): Update.
29303 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29304 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29305 * ipa-split.c (execute_split_functions): Update.
29306 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29308 2017-05-23 Tom de Vries <tom@codesourcery.com>
29310 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29311 attributes): Document rdrand effective target.
29313 2017-05-23 Tom de Vries <tom@codesourcery.com>
29315 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29316 attributes): Sort alphabetically.
29318 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29320 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29322 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29325 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29326 V2DF/V2DI splat into two separate patterns, one that handles
29327 registers, and the other that only handles memory. Drop support
29328 for splatting from a GPR on ISA 2.07 and then splitting the
29329 splat into direct move and splat.
29330 (vsx_splat_<mode>_reg): Likewise.
29331 (vsx_splat_<mode>_mem): Likewise.
29333 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29335 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29337 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29339 PR middle-end/80809
29340 * omp-low.c (finish_taskreg_remap): New function.
29341 (finish_taskreg_scan): If unit size of ctx->record_type
29342 is non-constant, unshare the size expression and replace
29343 decls in it with possible outer var refs.
29345 PR middle-end/80809
29346 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29347 GOVD_SHARED rather than GOVD_PRIVATE with it.
29348 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29349 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29351 PR middle-end/80853
29352 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29353 as last argument to build_outer_var_ref for pointer bases of array
29354 section reductions.
29356 2017-05-19 Martin Sebor <msebor@redhat.com>
29358 * print-tree.c (print_node): Print DECL_READ_P flag.
29360 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29362 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29363 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29364 * cgraph.c: Likewise.
29365 * cgraphunit.c: Likewise.
29366 * gengtype.c: Likewise.
29367 * ipa-cp.c: Likewise.
29368 * ipa-devirt.c: Likewise.
29369 * ipa-icf.c: Likewise.
29370 * ipa-predicate.c: Likewise.
29371 * ipa-profile.c: Likewise.
29372 * ipa-prop.c: Likewise.
29373 * ipa-split.c: Likewise.
29375 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29376 edge_predicate_pool, dump_inline_hints,
29377 inline_summary::account_size_time, redirect_to_unreachable,
29378 edge_set_predicate, set_hint_predicate,
29379 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29380 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29381 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29382 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29383 ipa_call_summary_t::remove, initialize_growth_caches,
29384 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29385 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29386 mark_modified, unmodified_parm_1, unmodified_parm,
29387 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29388 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29389 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29390 will_be_nonconstant_predicate, record_modified_bb_info,
29391 get_minimal_bb, record_modified, param_change_prob,
29392 phi_result_unknown_predicate, predicate_for_phi_result,
29393 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29394 estimate_function_body_sizes, compute_inline_parameters,
29395 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29396 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29397 inline_update_callee_summaries, remap_edge_change_prob,
29398 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29399 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29400 inline_analyze_function, inline_summary_t::insert,
29401 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29402 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29403 inline_free_summary): Move to ipa-fnsummary.h
29404 (predicate_t): Remove.
29405 * ipa-fnsummary.c: New file.
29406 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29407 (enum inline_hints_vals, inline_hints, agg_position_info,
29408 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29409 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29410 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29411 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29412 inline_read_summary, inline_write_summary, inline_free_summary,
29413 inline_analyze_function, initialize_inline_failed,
29414 inline_merge_summary, inline_update_overall_summary,
29415 compute_inline_parameters): Move to ipa-fnsummary.h
29416 * ipa-fnsummary.h: New file.
29417 * ipa-inline-transform.h: Include ipa-inline.h.
29418 * ipa-inline.c: LIkewise.
29420 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29422 * ipa-inline.c (edge_badness): Use inlined_time instead of
29423 inline_summaries->get.
29425 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29427 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29429 2017-05-22 Nathan Sidwell <nathan@acm.org>
29431 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29432 (fdump-lang): Document 'raw' option.
29433 * dumpfile.h (TDI_tu): Delete.
29434 * dumpfile.c (dump_files): Remove translation-unit.
29435 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29437 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29439 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29440 command option from $(AWK) call.
29441 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29442 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29443 [FORMAT]: Remove handling of variable.
29444 * config/avr/t-multilib: Regenerate.
29446 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29448 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29450 (dump_inline_summary): Do not print self_time.
29451 (estimate_function_body_sizes): Do not set self_time.
29452 (compute_inline_parameters): Likewise.
29453 (inline_read_section, inline_write_summary): Do not stream self_time.
29454 * ipa-inline.h (inline_summary): Drop self_time.
29456 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29458 * ipa-inline-analysis.c (account_size_time): Rename to ...
29459 (inline_summary::account_size_time): ... this one.
29460 (reset_ipa_call_summary): Turn to ...
29461 (ipa_call_summary::reset): ... this one.
29462 (reset_inline_summary): Turn to ...
29463 (inline_summary::reset): ... this one.
29464 (inline_summary_t::remove): Update.
29465 (inline_summary_t::duplicate): Update.
29466 (ipa_call_summary_t::remove): Update.
29467 (dump_inline_summary): Update.
29468 (estimate_function_body_sizes): Update.
29469 (compute_inline_parameters): Update.
29470 (estimate_node_size_and_time): Update.
29471 (inline_merge_summary): Update.
29472 (inline_update_overall_summary): Update.
29473 (inline_read_section): Update.
29474 (inline_write_summary): Update.
29475 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29476 add account_size_time and reset member functions.
29477 (ipa_call_summary): Add reset function.
29478 * ipa-predicate.h (predicate::operator &): Constify.
29480 2017-05-22 Richard Biener <rguenther@suse.de>
29482 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29484 2017-05-19 Jason Merrill <jason@redhat.com>
29486 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29488 2017-05-19 Marek Polacek <polacek@redhat.com>
29491 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29492 TYPE_OVERFLOW_WRAPS checks.
29494 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29496 * tree-core.h (enum omp_clause_default_kind): Add
29497 "OMP_CLAUSE_DEFAULT_PRESENT".
29498 * tree-pretty-print.c (dump_omp_clause): Handle it.
29499 * gimplify.c (enum gimplify_omp_var_data): Add
29500 "GOVD_MAP_FORCE_PRESENT".
29501 (gimplify_adjust_omp_clauses_1): Map it to
29502 "GOMP_MAP_FORCE_PRESENT".
29503 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29505 * gimplify.c (oacc_default_clause): Clarify.
29507 2017-05-19 Nathan Sidwell <nathan@acm.org>
29509 LANG_HOOK_REGISTER_DUMPS
29510 * toplev.c (general_init): Call register dump lang hook.
29511 * doc/invoke.texi: Document -fdump-lang option family.
29512 * dumpfile.c (dump_files): Remove class dump here.
29513 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29514 * dumpfile.h (tree_dump_index): Remove TDI_class.
29515 * langhooks-def.h (lhd_register_dumps): Declare.
29516 (LANG_HOOKS_REGISTER_DUMPS): Define.
29517 (LANG_HOOKS_INITIALIZER): Add it.
29518 * langhooks.c (lhd_register_dumps): Define.
29519 * langhooks.h (struct lang_hooks): Add register_dumps.
29521 2017-05-19 Nathan Sidwell <nathan@acm.org>
29523 * context.h (context::set_passes): New.
29524 * context.c (context::context): Do not create pass manager.
29525 * toplev.c (general_init): Create pass manager here.
29527 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29529 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29530 use this splitter if two add or or instructions would also work for
29531 the constant we want to generate.
29533 2017-05-19 Richard Biener <rguenther@suse.de>
29536 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29537 predicate evaluation.
29539 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29541 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29543 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29546 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29548 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29549 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29550 (inline_edge_summary_vec): Turn into ...
29551 (ipa_call_summaries): ... this one.
29552 (redirect_to_unreachable, edge_set_predicate,
29553 evaluate_properties_for_edge, inline_summary_alloc,
29554 reset_ipa_call_summary, reset_inline_summary,
29555 inline_summary_t::duplicate): Update.
29556 (inline_edge_duplication_hook): Turn to ...
29557 (ipa_call_summary_t::duplicate): ... this one.
29558 (inline_edge_removal_hook): Turn to ...
29559 (ipa_call_summary_t::remove): ... this one.
29560 (dump_inline_edge_summary): Turn to ...
29561 (dump_ipa_call_summary): ... this one.
29562 (estimate_function_body_sizes): Update.
29563 (inline_update_callee_summaries): Update.
29564 (remap_edge_change_prob): Update.
29565 (remap_edge_summaries): Update.
29566 (inline_merge_summary): Update.
29567 (do_estimate_edge_time): Update.
29568 (inline_generate_summary): Update.
29569 (inline_read_section): Update.
29570 (inline_read_summary): Update.
29571 (inline_free_summary): Update.
29572 * ipa-inline.c (can_inline_edge_p): Update.
29573 (compute_inlined_call_time): Update.
29574 (want_inline_small_function_p): Update.
29575 (edge_badness): Update.
29576 (early_inliner): Update.
29577 * ipa-inline.h (inline_edge_summary): Turn to ...
29578 (ipa_call_summary): ... this one.
29579 (ipa_call_summary_t): New class.
29580 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29581 (ipa_call_summaries): New.
29582 (inline_edge_summary): Remove.
29583 (estimate_edge_growth): Update.
29584 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29585 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29586 * ipa-split.c (execute_split_functions): Update.
29587 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29589 2017-05-19 Richard Biener <rguenther@suse.de>
29591 PR middle-end/80764
29592 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29594 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29596 * config/rs6000/rs6000.c (struct machine_function): Add field
29597 fpr_is_wrapped_separately.
29598 (rs6000_get_separate_components): Use 64 components. Handle the
29599 new FPR components.
29600 (rs6000_components_for_bb): Handle the FPR components.
29601 (rs6000_emit_prologue_components): Handle the FPR components.
29602 (rs6000_emit_epilogue_components): Handle the FPR components.
29603 (rs6000_set_handled_components): Handle the FPR components.
29604 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29605 that are already separately shrink-wrapped.
29606 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29607 that are already separately shrink-wrapped.
29609 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29612 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29615 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29616 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29617 cases where the register allocator uses a move and an offsettable
29618 memory operation to/from a FPR register on ISA 2.06/2.07.
29619 (define_peephole2 for Altivec d-form store): Likewise.
29621 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29624 * config/i386/mmx.md (*mov<mode>_internal): Enable
29625 alternatives 11, 12, 13 and 14 also for 32bit targets.
29626 Remove alternatives 15, 16, 17 and 18.
29627 * config/i386/sse.md (vec_concatv2di): Change
29628 alternative (!x, *y) to (x, ?!*Yn).
29630 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29632 * dumpfile.h (enum dump_kind): Remove stray comma.
29634 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29636 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29637 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29638 predicate::num_conditions
29639 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29640 (CHANGED): turn into predicate::changed.
29641 (agg_position_info): Move to ipa-predicate.h
29642 (add_condition, predicate::add_clause, predicate::operator &=,
29643 predicate::or_with, predicate::evaluate, predicate::probability,
29644 dump_condition, dump_clause, predicate::dump,
29645 predicate::remap_after_duplication, predicate::remap_after_inlining,
29646 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29647 (evaluate_conditions_for_known_args): Update.
29648 (set_cond_stmt_execution_predicate): Update.
29649 * ipa-inline.h: Include ipa-predicate.h
29650 (condition, inline_param_summary, conditions, agg_position_info,
29651 predicate): Move to ipa-predicate.h
29652 * ipa-predicate.c: New file.
29653 * ipa-predicate.h: New file.
29655 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29657 * final.c (leaf_function_p): Check we are not in a sequence.
29659 2017-05-18 Martin Liska <mliska@suse.cz>
29661 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29662 * dumpfile.c (dump_register): Use new enum dump_kind.
29663 (get_dump_file_name): Likewise.
29664 (dump_enable_all): Likewise.
29665 (dump_switch_p_1): Likewise.
29666 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29667 * dumpfile.h (enum dump_kind): New enum type.
29668 (struct dump_file_info): Create constructor and
29669 format fields and comments.
29670 * passes.c (pass_manager::register_one_dump_file):
29672 * statistics.c (statistics_early_init): Likewise.
29673 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29674 TDF_TREE with TDF_SLIM.
29675 (gather_memory_references_ref): Likewise.
29677 2017-05-18 Martin Liska <mliska@suse.cz>
29679 * vec.h (struct vnull): Use it.
29681 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29683 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29684 (true_predicate, false_predicate, true_predicate_p,
29685 false_predicate_p): Remove.
29686 (single_cond_predicate, not_inlined_predicate): Turn to member function
29688 (add_condition): Update.
29689 (add_clause): Turn to...
29690 (predicate::add_clause): ... this one; update; allow passing NULL
29692 (and_predicates): Turn to ...
29693 (predicate::operator &=): ... this one.
29694 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29695 (or_predicates): Turn to ...
29696 (predicate::or_with): ... this one.
29697 (evaluate_predicate): Turn to ...
29698 (predicate::evaluate): ... this one.
29699 (predicate_probability): Turn to ...
29700 (predicate::probability): ... this one.
29701 (dump_condition): Update.
29702 (dump_predicate): Turn to ...
29703 (predicate::dump): ... this one.
29704 (account_size_time): Update.
29705 (edge_set_predicate): Update.
29706 (set_hint_predicate): UPdate.
29707 (evaluate_conditions_for_known_args): Update.
29708 (evaluate_properties_for_edge): Update.
29709 (remap_predicate_after_duplication): Turn to...
29710 (predicate::remap_after_duplication): ... this one.
29711 (remap_hint_predicate_after_duplication): Update.
29712 (inline_summary_t::duplicate): UPdate.
29713 (dump_inline_edge_summary): Update.
29714 (dump_inline_summary): Update.
29715 (set_cond_stmt_execution_predicate): Update.
29716 (set_switch_stmt_execution_predicate): Update.
29717 (compute_bb_predicates): Update.
29718 (will_be_nonconstant_expr_predicate): Update.
29719 (will_be_nonconstant_predicate): Update.
29720 (phi_result_unknown_predicate): Update.
29721 (predicate_for_phi_result): Update.
29722 (array_index_predicate): Update.
29723 (estimate_function_body_sizes): Update.
29724 (estimate_node_size_and_time): Update.
29725 (estimate_ipcp_clone_size_and_time): Update.
29726 (remap_predicate): Rename to ...
29727 (predicate::remap_after_inlining): ... this one.
29728 (remap_hint_predicate): Update.
29729 (inline_merge_summary): Update.
29730 (inline_update_overall_summary): Update.
29731 (estimate_size_after_inlining): Update.
29732 (read_predicate): Rename to ...
29733 (predicate::stream_in): ... this one.
29734 (read_inline_edge_summary): Update.
29735 (write_predicate): Rename to ...
29736 (predicate::stream_out): ... this one.
29737 (write_inline_edge_summary): Update.
29738 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29739 (clause_t): Turn to uint32_t
29740 (predicate): Turn to class; implement constructor and operators
29742 (size_time_entry): Update.
29743 (inline_summary): Update.
29744 (inline_edge_summary): Update.
29746 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29748 * fold-const.c (fold_binary_loc): Move transformation...
29749 * match.pd (C - X CMP X): ... here.
29751 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29753 * config/sparc/sparc.c (sparc_option_override): Set function
29754 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29755 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29757 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29759 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29761 2017-05-18 Marek Polacek <polacek@redhat.com>
29764 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29765 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29766 gimple_assign_load_p.
29768 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29770 PR middle-end/80692
29771 * real.c (do_compare): Give decimal_do_compare preference over
29772 comparing just the signs.
29774 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29776 * doc/md.texi (Canonicalization of Instructions): Describe the
29777 canonical form of instructions that inherently set a condition
29780 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29782 PR middle-end/80775
29783 * tree-cfg.c: Move deletion of unreachable case statements to after
29784 the merging of consecutive case labels.
29786 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29788 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29789 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29790 restoring of callee-saved registers.
29792 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29794 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29795 * config/visium/visium.c (single_set_and_flags): Likewise.
29796 * config/visium/visium.md (Substitutions): Likewise.
29798 2017-05-17 Martin Liska <mliska@suse.cz>
29800 * cfg.c: Introduce dump_flags_t type and
29801 use it instead of int type.
29803 * cfghooks.c: Likewise.
29804 * cfghooks.h (struct cfg_hooks): Likewise.
29805 * cfgrtl.c: Likewise.
29806 * cfgrtl.h: Likewise.
29807 * cgraph.c (cgraph_node::get_body): Likewise.
29808 * coretypes.h: Likewise.
29809 * domwalk.c: Likewise.
29810 * domwalk.h: Likewise.
29811 * dumpfile.c (struct dump_option_value_info): Likewise.
29812 (dump_enable_all): Likewise.
29813 (dump_switch_p_1): Likewise.
29814 (opt_info_switch_p): Likewise.
29815 * dumpfile.h (enum tree_dump_index): Likewise.
29816 (struct dump_file_info): Likewise.
29817 * genemit.c: Likewise.
29818 * generic-match-head.c: Likewise.
29819 * gengtype.c (open_base_files): Likewise.
29820 * gimple-pretty-print.c: Likewise.
29821 * gimple-pretty-print.h: Likewise.
29822 * graph.c (print_graph_cfg): Likewise.
29823 * graphite-scop-detection.c (dot_all_sese): Likewise.
29824 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29825 * loop-unroll.c (report_unroll): Likewise.
29826 * passes.c (pass_manager::register_one_dump_file): Likewise.
29827 * print-tree.c: Likewise.
29828 * statistics.c: Likewise.
29829 * tree-cfg.c: Likewise.
29830 * tree-cfg.h: Likewise.
29831 * tree-dfa.c: Likewise.
29832 * tree-dfa.h: Likewise.
29833 * tree-dump.c (dump_function): Likewise.
29834 * tree-dump.h (struct dump_info): Likewise.
29835 * tree-pretty-print.c: Likewise.
29836 * tree-pretty-print.h: Likewise.
29837 * tree-ssa-live.c: Likewise.
29838 * tree-ssa-live.h: Likewise.
29839 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29840 * tree-vect-loop.c: Likewise.
29841 * tree-vect-slp.c: Likewise.
29843 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29844 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29846 PR tree-optimization/80457
29847 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29848 of all arguments to a statement as scalar_to_vec operations.
29849 (vectorizable_call): Adjust call to vect_model_simple_cost for
29851 (vectorizable_conversion): Likewise.
29852 (vectorizable_assignment): Likewise.
29853 (vectorizable_shift): Likewise.
29854 (vectorizable_operation): Likewise.
29855 (vectorizable_comparison): Likewise.
29856 (vect_is_simple_cond): Record the def types for operands.
29857 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29858 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29859 for statement argument count.
29861 2017-05-16 Carl Love <cel@us.ibm.com>
29863 * config/rs6000/rs6000-c: Add support for built-in functions
29864 vector unsigned long long vec_bperm (vector unsigned long long,
29865 vector unsigned char)
29866 vector signed long long vec_mule (vector signed int,
29868 vector unsigned long long vec_mule (vector unsigned int,
29869 vector unsigned int)
29870 vector signed long long vec_mulo (vector signed int,
29872 vector unsigned long long vec_mulo (vector unsigned int,
29873 vector unsigned int)
29874 vector signed char vec_sldw (vector signed char,
29875 vector signed char,
29877 vector unsigned char vec_sldw (vector unsigned char,
29878 vector unsigned char,
29880 vector signed short vec_sldw (vector signed short,
29881 vector signed short,
29883 vector unsigned short vec_sldw (vector unsigned short,
29884 vector unsigned short,
29886 vector signed int vec_sldw (vector signed int,
29889 vector unsigned int vec_sldw (vector unsigned int,
29890 vector unsigned int,
29892 vector signed long long vec_sldw (vector signed long long,
29893 vector signed long long,
29895 vector unsigned long long vec_sldw (vector unsigned long long,
29896 vector unsigned long long,
29898 * config/rs6000/rs6000-c: Add support for built-in functions
29899 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29900 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29901 * doc/extend.texi: Update the built-in documentation for the
29902 new built-in functions.
29904 2017-05-16 Marek Polacek <polacek@redhat.com>
29908 * tree.c (save_expr): Don't fold the expression.
29910 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29912 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29913 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29914 and (?*y,m). Update insn attributes.
29916 2017-05-16 Martin Liska <mliska@suse.cz>
29918 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29919 flags argument of print_gimple_stmt, print_gimple_expr,
29920 print_generic_stmt and print_generic_expr.
29921 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29922 * coretypes.h: Likewise.
29923 * except.c (dump_eh_tree): Likewise.
29924 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29925 * gimple-pretty-print.h: Likewise.
29926 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29927 (backprop::push_to_worklist): Likewise.
29928 (backprop::pop_from_worklist): Likewise.
29929 (backprop::process_use): Likewise.
29930 (backprop::intersect_uses): Likewise.
29931 (note_replacement): Likewise.
29932 * gimple-ssa-store-merging.c
29933 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29934 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29935 (pass_store_merging::execute): Likewise.
29936 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29937 (ssa_base_cand_dump_callback): Likewise.
29938 (dump_incr_vec): Likewise.
29939 (replace_refs): Likewise.
29940 (replace_mult_candidate): Likewise.
29941 (create_add_on_incoming_edge): Likewise.
29942 (create_phi_basis): Likewise.
29943 (insert_initializers): Likewise.
29944 (all_phi_incrs_profitable): Likewise.
29945 (introduce_cast_before_cand): Likewise.
29946 (replace_one_candidate): Likewise.
29947 * gimplify.c (gimplify_expr): Likewise.
29948 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29949 (set_rename): Likewise.
29950 (rename_uses): Likewise.
29951 (copy_loop_phi_nodes): Likewise.
29952 (add_close_phis_to_merge_points): Likewise.
29953 (copy_loop_close_phi_args): Likewise.
29954 (copy_cond_phi_args): Likewise.
29955 (graphite_copy_stmts_from_block): Likewise.
29956 (translate_pending_phi_nodes): Likewise.
29957 * graphite-poly.c (print_pdr): Likewise.
29958 (dump_gbb_cases): Likewise.
29959 (dump_gbb_conditions): Likewise.
29960 (print_scop_params): Likewise.
29961 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29962 (build_cross_bb_scalars_use): Likewise.
29963 (gather_bbs::before_dom_children): Likewise.
29964 * hsa-dump.c (dump_hsa_immed): Likewise.
29965 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29966 (get_replacement_map): Likewise.
29967 * ipa-inline-analysis.c (dump_condition): Likewise.
29968 (estimate_function_body_sizes): Likewise.
29969 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29970 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29971 * ipa-prop.c (ipa_dump_param): Likewise.
29972 (ipa_print_node_jump_functions_for_edge): Likewise.
29973 (ipa_modify_call_arguments): Likewise.
29974 (ipa_modify_expr): Likewise.
29975 (ipa_dump_param_adjustments): Likewise.
29976 (ipa_dump_agg_replacement_values): Likewise.
29977 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29978 * ipa-pure-const.c (check_stmt): Likewise.
29979 (pass_nothrow::execute): Likewise.
29980 * ipa-split.c (execute_split_functions): Likewise.
29981 * omp-offload.c (dump_oacc_loop_part): Likewise.
29982 (dump_oacc_loop): Likewise.
29983 * trans-mem.c (tm_log_emit): Likewise.
29984 (tm_memopt_accumulate_memops): Likewise.
29985 (dump_tm_memopt_set): Likewise.
29986 (dump_tm_memopt_transform): Likewise.
29987 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29988 (print_loop): Likewise.
29989 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29990 (chkp_gather_checks_info): Likewise.
29991 (chkp_get_check_result): Likewise.
29992 (chkp_remove_check_if_pass): Likewise.
29993 (chkp_use_outer_bounds_if_possible): Likewise.
29994 (chkp_reduce_bounds_lifetime): Likewise.
29995 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29996 (chkp_mark_completed_bounds): Likewise.
29997 (chkp_register_incomplete_bounds): Likewise.
29998 (chkp_mark_invalid_bounds): Likewise.
29999 (chkp_maybe_copy_and_register_bounds): Likewise.
30000 (chkp_build_returned_bound): Likewise.
30001 (chkp_get_bound_for_parm): Likewise.
30002 (chkp_build_bndldx): Likewise.
30003 (chkp_get_bounds_by_definition): Likewise.
30004 (chkp_generate_extern_var_bounds): Likewise.
30005 (chkp_get_bounds_for_decl_addr): Likewise.
30006 * tree-chrec.c (chrec_apply): Likewise.
30007 * tree-data-ref.c (dump_data_reference): Likewise.
30008 (dump_subscript): Likewise.
30009 (dump_data_dependence_relation): Likewise.
30010 (analyze_overlapping_iterations): Likewise.
30011 * tree-inline.c (expand_call_inline): Likewise.
30012 (tree_function_versioning): Likewise.
30013 * tree-into-ssa.c (dump_defs_stack): Likewise.
30014 (dump_currdefs): Likewise.
30015 (dump_names_replaced_by): Likewise.
30016 (dump_update_ssa): Likewise.
30017 (update_ssa): Likewise.
30018 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30019 * tree-parloops.c (build_new_reduction): Likewise.
30020 (try_create_reduction_list): Likewise.
30021 (ref_conflicts_with_region): Likewise.
30022 (oacc_entry_exit_ok_1): Likewise.
30023 (oacc_entry_exit_single_gang): Likewise.
30024 * tree-pretty-print.h: Likewise.
30025 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30026 (get_scalar_evolution): Likewise.
30027 (add_to_evolution): Likewise.
30028 (get_loop_exit_condition): Likewise.
30029 (analyze_evolution_in_loop): Likewise.
30030 (analyze_initial_condition): Likewise.
30031 (analyze_scalar_evolution): Likewise.
30032 (instantiate_scev): Likewise.
30033 (number_of_latch_executions): Likewise.
30034 (gather_chrec_stats): Likewise.
30035 (final_value_replacement_loop): Likewise.
30036 (scev_const_prop): Likewise.
30037 * tree-sra.c (dump_access): Likewise.
30038 (disqualify_candidate): Likewise.
30039 (create_access): Likewise.
30040 (reject): Likewise.
30041 (maybe_add_sra_candidate): Likewise.
30042 (create_access_replacement): Likewise.
30043 (analyze_access_subtree): Likewise.
30044 (analyze_all_variable_accesses): Likewise.
30045 (sra_modify_assign): Likewise.
30046 (initialize_constant_pool_replacements): Likewise.
30047 (find_param_candidates): Likewise.
30048 (decide_one_param_reduction): Likewise.
30049 (replace_removed_params_ssa_names): Likewise.
30050 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30051 * tree-ssa-copy.c (dump_copy_of): Likewise.
30052 (copy_prop_visit_cond_stmt): Likewise.
30053 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30054 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30055 (record_equivalences_from_stmt): Likewise.
30056 * tree-ssa-dse.c (compute_trims): Likewise.
30057 (delete_dead_call): Likewise.
30058 (delete_dead_assignment): Likewise.
30059 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30060 (forward_propagate_into_cond): Likewise.
30061 (pass_forwprop::execute): Likewise.
30062 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30063 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30065 (move_computations_worker): Likewise.
30066 (execute_sm): Likewise.
30067 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30068 (remove_exits_and_undefined_stmts): Likewise.
30069 (remove_redundant_iv_tests): Likewise.
30070 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30071 (adjust_iv_update_pos): Likewise.
30072 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30073 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30074 (value_replacement): Likewise.
30075 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30076 * tree-ssa-pre.c (print_pre_expr): Likewise.
30077 (get_representative_for): Likewise.
30078 (create_expression_by_pieces): Likewise.
30079 (insert_into_preds_of_block): Likewise.
30080 (eliminate_insert): Likewise.
30081 (eliminate_dom_walker::before_dom_children): Likewise.
30082 (eliminate): Likewise.
30083 (remove_dead_inserted_code): Likewise.
30084 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30085 * tree-ssa-reassoc.c (get_rank): Likewise.
30086 (eliminate_duplicate_pair): Likewise.
30087 (eliminate_plus_minus_pair): Likewise.
30088 (eliminate_not_pairs): Likewise.
30089 (undistribute_ops_list): Likewise.
30090 (eliminate_redundant_comparison): Likewise.
30091 (update_range_test): Likewise.
30092 (optimize_range_tests_var_bound): Likewise.
30093 (optimize_vec_cond_expr): Likewise.
30094 (rewrite_expr_tree): Likewise.
30095 (rewrite_expr_tree_parallel): Likewise.
30096 (linearize_expr): Likewise.
30097 (break_up_subtract): Likewise.
30098 (linearize_expr_tree): Likewise.
30099 (attempt_builtin_powi): Likewise.
30100 (attempt_builtin_copysign): Likewise.
30101 (transform_stmt_to_copy): Likewise.
30102 (transform_stmt_to_multiply): Likewise.
30103 (dump_ops_vector): Likewise.
30104 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30105 (print_scc): Likewise.
30106 (set_ssa_val_to): Likewise.
30107 (visit_reference_op_store): Likewise.
30108 (visit_use): Likewise.
30109 (sccvn_dom_walker::before_dom_children): Likewise.
30110 (run_scc_vn): Likewise.
30111 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30113 (expr_hash_elt::print): Likewise.
30114 (const_and_copies::pop_to_marker): Likewise.
30115 (const_and_copies::record_const_or_copy_raw): Likewise.
30116 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30117 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30118 (dump_predicates): Likewise.
30119 (find_uninit_use): Likewise.
30120 (warn_uninitialized_phi): Likewise.
30121 (pass_late_warn_uninitialized::execute): Likewise.
30122 * tree-ssa.c (verify_vssa): Likewise.
30123 (verify_ssa): Likewise.
30124 (maybe_optimize_var): Likewise.
30125 * tree-vrp.c (dump_value_range): Likewise.
30126 (dump_all_value_ranges): Likewise.
30127 (dump_asserts_for): Likewise.
30128 (register_edge_assert_for_2): Likewise.
30129 (vrp_visit_cond_stmt): Likewise.
30130 (vrp_visit_switch_stmt): Likewise.
30131 (vrp_visit_stmt): Likewise.
30132 (vrp_visit_phi_node): Likewise.
30133 (simplify_cond_using_ranges_1): Likewise.
30134 (fold_predicate_in): Likewise.
30135 (evrp_dom_walker::before_dom_children): Likewise.
30136 (evrp_dom_walker::push_value_range): Likewise.
30137 (evrp_dom_walker::pop_value_range): Likewise.
30138 (execute_early_vrp): Likewise.
30140 2017-05-16 Richard Biener <rguenther@suse.de>
30142 * dwarf2out.c (loc_list_from_tree_1): Do not create
30143 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30145 2017-05-16 Richard Biener <rguenther@suse.de>
30147 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30149 (note_variable_value_in_expr): If we resolved the decl ref
30150 do not push to the stack.
30152 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30154 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30155 operations in fast-math mode.
30156 (vaddq_f16): Likewise.
30157 (vmul_f16): Likewise.
30158 (vmulq_f16): Likewise.
30159 (vsub_f16): Likewise.
30160 (vsubq_f16): Likewise.
30161 * config/arm/neon.md (add<mode>3): New.
30163 (fma:<VH:mode>3): New. Also remove outdated comment.
30166 2017-05-16 Martin Liska <mliska@suse.cz>
30170 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30171 (odr_types_equivalent_p): Likewise.
30173 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30175 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30177 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30180 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30181 non-interunit SSE move alternatives with '?'.
30182 (zero-extendsidi peephole2): New peephole to skip intermediate
30183 general register in SSE zero-extend sequence.
30185 2017-05-15 Jeff Law <law@redhat.com>
30187 * reorg.c (relax_delay_slots): Create a new variable to hold
30188 the temporary target rather than clobbering TARGET_LABEL.
30190 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30191 missing argument to extract_bit_field call.
30192 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30194 2017-05-15 Martin Liska <mliska@suse.cz>
30197 * gcc.c (process_command): Do not allow empty argument of -o option.
30199 2017-05-15 Renlin Li <renlin.li@arm.com>
30201 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30202 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30203 * config/aarch64/constraints.md (Usf): Add long call check.
30204 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30205 (call_value): Likewise.
30206 (sibcall): Likewise.
30207 (sibcall_value): Likewise.
30209 (call_value_insn): New.
30210 (sibcall_insn): Update rtx pattern.
30211 (sibcall_value_insn): Likewise.
30212 (call_internal): Remove.
30213 (call_value_internal): Likewise.
30214 (sibcall_internal): Likewise.
30215 (sibcall_value_internal): Likewise.
30216 (call_reg): Likewise.
30217 (call_symbol): Likewise.
30218 (call_value_reg): Likewise.
30219 (call_value_symbol): Likewise.
30221 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30224 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30226 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30228 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30229 compatible with CCGOCmode and with CCZmode.
30231 2017-05-14 Martin Sebor <msebor@redhat.com>
30233 PR middle-end/77671
30234 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30235 (gimple_fold_builtin_snprintf): Same.
30236 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30237 (gimple_fold_builtin_snprintf): Same.
30238 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30239 of character types.
30240 (is_call_safe): New function.
30241 (try_substitute_return_value): Call it.
30242 (try_simplify_call): New function.
30243 (pass_sprintf_length::handle_gimple_call): Call it.
30245 2017-05-14 Martin Sebor <msebor@redhat.com>
30247 PR middle-end/80669
30248 * builtins.c (expand_builtin_stpncpy): Simplify.
30250 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30252 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30253 * config/i386/i386.h
30254 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30255 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30256 (struct machine_function): Add new members call_ms2sysv,
30257 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30258 (struct machine_frame_state): New fields sp_realigned and
30259 sp_realigned_offset.
30260 * config/i386/i386.c
30261 (enum xlogue_stub): New enum.
30262 (enum xlogue_stub_sets): New enum.
30263 (class xlogue_layout): New class.
30264 (struct ix86_frame): New fields stack_realign_allocate_offset,
30265 stack_realign_offset and outlined_save_offset. Modify comments to
30266 detail stack layout when using out-of-line stubs.
30267 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30268 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30269 -mcall-ms2sysv-xlogues.
30270 (stub_managed_regs): New static variable.
30271 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30272 registers managed by out-of-line stub.
30273 (disable_call_ms2sysv_xlogues): New function.
30274 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30275 m->call_ms2sysv when appropriate and compute frame layout for
30277 (sp_valid_at, fp_valid_at): New inline functions.
30278 (choose_basereg): New function.
30279 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30281 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30282 Use align parameter of choose_baseaddr to generated aligned SSE movs
30284 (pro_epilogue_adjust_stack): Modify to track
30285 machine_frame_state::sp_realigned.
30286 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30287 (ix86_nsaved_sseregs): Likewise.
30288 (ix86_emit_save_regs): Likewise.
30289 (ix86_emit_save_regs_using_mov): Likewise.
30290 (ix86_emit_save_sse_regs_using_mov): Likewise.
30291 (get_scratch_register_on_entry): Likewise.
30292 (gen_frame_set): New function.
30293 (gen_frame_load): Likewise.
30294 (gen_frame_store): Likewise.
30295 (emit_outlined_ms2sysv_save): Likewise.
30296 (emit_outlined_ms2sysv_restore): Likewise.
30297 (ix86_expand_prologue): Modify stack re-alignment code and call
30298 emit_outlined_ms2sysv_save when appropriate.
30299 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30300 parameter rtx_insn *insn, which allows the function to be used to only
30301 generate the notes.
30302 (ix86_expand_epilogue): Modify validity checks of frame and stack
30303 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30304 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30305 * config/i386/predicates.md
30306 (save_multiple): New predicate.
30307 (restore_multiple): Likewise.
30308 * config/i386/sse.md
30309 (save_multiple<mode>): New pattern.
30310 (save_multiple_realign<mode>): Likewise.
30311 (restore_multiple<mode>): Likewise.
30312 (restore_multiple_and_return<mode>): Likewise.
30313 (restore_multiple_leave_return<mode>): Likewise.
30314 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30316 2017-05-14 Julia Koval <julia.koval@intel.com>
30318 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30319 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30320 (__builtin_ia32_xsetbv): New builtins.
30321 * config/i386/i386.c (ix86_expand_special_args_builtin):
30323 (ix86_expand_builtin): Special expand for new intrinsics.
30324 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30325 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30326 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30328 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30330 * cfganal.c (inverted_post_order_compute): Change argument type
30332 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30333 * df-core.c (rest_of_handle_df_initialize): Adjust.
30334 (rest_of_handle_df_finish): Likewise.
30335 (df_analyze_1): Likewise.
30336 (df_analyze): Likewise.
30337 (loop_inverted_post_order_compute): Change argument to be a vec *.
30338 (df_analyze_loop): Adjust.
30339 (df_get_n_blocks): Likewise.
30340 (df_get_postorder): Likewise.
30341 * df.h (struct df_d): Change field to be a vec.
30342 * lcm.c (compute_laterin): Adjust.
30343 (compute_available): Likewise.
30344 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30345 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30346 * tree-ssa-pre.c (compute_antic): Likewise.
30348 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30350 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30351 (depth_first_search::depth_first_search): Change structure init
30352 function to this constructor.
30353 (depth_first_search::add_bb): Rename function to this member.
30354 (depth_first_search::execute): Likewise.
30355 (flow_dfs_compute_reverse_finish): Adjust.
30357 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30359 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30360 (longest_simple_path): Likewise.
30361 * shrink-wrap.c (spread_components): Likewise.
30362 (disqualify_problematic_components): Likewise.
30363 (emit_common_heads_for_components): Likewise.
30364 (emit_common_tails_for_components): Likewise.
30365 (insert_prologue_epilogue_for_components): Likewise.
30367 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30369 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30372 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30374 * df-core.c (df_set_blocks): Start using auto_bitmap.
30375 (df_compact_blocks): Likewise.
30376 * df-problems.c (df_rd_confluence_n): Likewise.
30377 * df-scan.c (df_insn_rescan_all): Likewise.
30378 (df_process_deferred_rescans): Likewise.
30379 (df_update_entry_block_defs): Likewise.
30380 (df_update_exit_block_uses): Likewise.
30381 (df_entry_block_bitmap_verify): Likewise.
30382 (df_exit_block_bitmap_verify): Likewise.
30383 (df_scan_verify): Likewise.
30384 * lra-constraints.c (lra_constraints): Likewise.
30385 (undo_optional_reloads): Likewise.
30386 (lra_undo_inheritance): Likewise.
30387 * lra-remat.c (calculate_gen_cands): Likewise.
30388 (do_remat): Likewise.
30389 * lra-spills.c (assign_spill_hard_regs): Likewise.
30390 (spill_pseudos): Likewise.
30391 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30392 (bitmap_set_subtract_values): Likewise.
30394 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30396 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30397 management with auto_bitmap.
30398 (fix_inter_tick): Likewise.
30399 (fix_recovery_deps): Likewise.
30400 * ira.c (add_store_equivs): Likewise.
30401 (find_moveable_pseudos): Likewise.
30402 (split_live_ranges_for_shrink_wrap): Likewise.
30403 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30404 (rtx_reuse_manager::seen_def_p): Likewise.
30405 (rtx_reuse_manager::set_seen_def): Likewise.
30406 * print-rtl.h (class rtx_reuse_manager): Likewise.
30408 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30410 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30412 (migrate_btr_def): Likewise.
30413 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30414 * df-core.c (loop_post_order_compute): Likewise.
30415 (loop_inverted_post_order_compute): Likewise.
30416 * hsa-common.h: Likewise.
30417 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30418 * init-regs.c (initialize_uninitialized_regs): Likewise.
30419 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30420 (inline_small_functions): Likewise.
30421 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30422 * ira.c (combine_and_move_insns): Likewise.
30423 (build_insn_chain): Likewise.
30424 * loop-invariant.c (find_invariants): Likewise.
30425 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30426 * predict.c (tree_predict_by_opcode): Likewise.
30427 (predict_paths_leading_to): Likewise.
30428 (predict_paths_leading_to_edge): Likewise.
30429 (estimate_loops_at_level): Likewise.
30430 (estimate_loops): Likewise.
30431 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30432 (spread_components): Likewise.
30433 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30434 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30435 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30436 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30437 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30438 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30439 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30440 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30441 (mark_threaded_blocks): Likewise.
30442 (thread_through_all_blocks): Likewise.
30443 * tree-ssa.c (verify_ssa): Likewise.
30444 (execute_update_addresses_taken): Likewise.
30445 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30447 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30449 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30451 (post_order_compute): Likewise.
30452 (inverted_post_order_compute): Likewise.
30453 (pre_and_rev_post_order_compute_fn): Likewise.
30455 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30457 * genrecog.c (int_set::int_set): Explicitly construct our
30458 auto_vec base class.
30459 * vec.h (auto_vec::auto_vec): New constructor.
30461 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30463 * bitmap.h (class auto_bitmap): New constructor taking
30464 bitmap_obstack * argument.
30466 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30468 * bitmap.h (class auto_bitmap): Change type of m_bits to
30469 bitmap_head, and adjust ctor / dtor and member operators.
30471 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30473 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30474 when returned register mode doesn't match original mode.
30476 2017-05-12 Jeff Law <law@redhat.com>
30477 Jakub Jelinek <jakub@redhat.com>
30479 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30480 we look for cc setter after the compare-elim changes.
30481 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30482 within the vector to match what compare-elim now expects.
30483 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30484 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30486 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30487 after the compare-elim changes.
30488 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30489 the vector to match what compare-elim now expects.
30490 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30491 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30492 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30493 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30494 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30496 * config/visium/visium.c (single_set_and_flags): Fix where
30497 we look for cc setter after the compare-elim changes.
30498 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30499 with the vector to match what compare-elim now expects.
30500 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30501 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30502 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30503 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30504 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30505 (neg<mode>2_insn_set_overflow): Likewise.
30507 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30509 PR middle-end/79794
30510 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30511 maybe_expand_insn call, set ops[0].target. If still set after call,
30512 set alt_rtl. Add extra arg to recursive calls.
30513 (extract_bit_field): Add alt_rtl argument. Pass to
30515 * expmed.h (extract_bit_field): Fix prototype.
30516 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30517 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30518 to extract_bit_field_calls.
30519 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30520 Pass alt_rtl to extract_bit_field calls.
30521 * calls.c (store_unaligned_arguments_into_psuedos)
30522 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30523 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30525 * optabs.h (struct expand_operand): Add target bitfield.
30527 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30529 * compare-elim.c (try_eliminate_compare): Canonicalize
30530 operation with embedded compare to
30531 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30532 (set (reg) (operation)].
30534 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30536 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30539 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30540 cost of adding a carry flag for ADC instruction.
30541 [case MINUS]: Ignore the cost of subtracting a carry flag
30542 for SBB instruction.
30544 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30546 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30548 * config/rs6000/bmiintrin.h: New file.
30549 * config/rs6000/bmi2intrin.h: New file.
30550 * config/rs6000/x86intrin.h: New file.
30552 2017-05-12 Jeff Law <law@redhat.com>
30554 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30557 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30559 PR middle-end/80707
30560 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30562 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30564 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30565 early expansion of vector divide builtins.
30566 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30567 builtins identified as having unsigned arguments.
30569 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30571 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30572 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30573 expansion of vector logical operations (and, andc, or, xor,
30576 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30578 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30579 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30581 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30583 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30584 early GIMPLE expansion of vector multiplies.
30586 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30588 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30589 TARGET_HAVE_MOVT conditional.
30590 (movt splitter): Likewise.
30592 2017-05-12 Richard Biener <rguenther@suse.de>
30594 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30595 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30596 Fold all stmts not inplace.
30598 2017-05-12 Richard Biener <rguenther@suse.de>
30600 PR tree-optimization/80713
30601 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30602 inserted_exprs bit for not removed stmts.
30604 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30606 PR middle-end/69921
30607 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30608 parallelized" attribute for parallelized OpenACC kernels.
30609 * omp-offload.c (execute_oacc_device_lower): Use it.
30611 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30612 Set "oacc kernels" attribute.
30613 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30614 parameter. Adjust all users.
30615 (oacc_fn_attrib_kernels_p): Remove function.
30616 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30617 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30618 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30619 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30620 assert "oacc kernels" attribute is set.
30622 2017-05-11 Carl Love <cel@us.ibm.com>
30624 * config/rs6000/rs6000-c: Add support for built-in functions
30625 vector unsigned char vec_popcnt (vector signed char)
30626 vector unsigned char vec_popcnt (vector unsigned char)
30627 vector unsigned short vec_popcnt (vector signed short)
30628 vector unsigned short vec_popcnt (vector unsigned short)
30629 vector unsigned int vec_popcnt (vector signed int)
30630 vector unsigned int vec_popcnt (vector unsigned int)
30631 vector unsigned long long vec_popcnt (vector signed long long)
30632 vector unsigned long long vec_popcnt (vector unsigned long long)
30633 vector signed long long vec_slo (vector signed long long,
30634 vector signed char)
30635 vector signed long long vec_slo (vector signed long long,
30636 vector unsigned char)
30637 vector unsigned long long vec_slo (vector unsigned long long,
30638 vector signed char)
30639 vector unsigned long long vec_slo (vector unsigned long long,
30640 vector unsigned char)
30641 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30642 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30643 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30644 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30645 * doc/extend.texi: Update the built-in documentation file for the
30646 new built-in functions.
30648 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30650 * attribs.h (sorted_attr_string): Move machine independent
30651 functions for target clone support from the i386 port to common
30652 code. Rename ix86_function_versions to common_function_versions.
30653 Rename make_name to make_unique_name.
30654 (common_function_versions): Likewise.
30655 (make_unique_name): Likewise.
30656 (make_dispatcher_decl): Likewise.
30657 (is_function_default_version): Likewise.
30658 * attribs.c (attr_strcmp): Likewise.
30659 (sorted_attr_string): Likewise.
30660 (common_function_versions): Likewise.
30661 (make_unique_name): Likewise.
30662 (make_dispatcher_decl): Likewise.
30663 (is_function_default_version): Likewise.
30664 * config/i386/i386.c (attr_strcmp): Likewise.
30665 (sorted_attr_string): Likewise.
30666 (ix86_function_versions): Likewise.
30667 (make_name): Likewise.
30668 (make_dispatcher_decl): Likewise.
30669 (is_function_default_version): Likewise.
30670 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30672 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30675 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30676 Account for direct move costs for vec_construct of integer
30679 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30682 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30683 (UNSPEC_STX_ATOMIC): Ditto.
30684 (loaddi_via_sse): New insn.
30685 (storedi_via_sse): Ditto.
30686 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30687 Update corresponding peephole2 patterns.
30688 (atomic_storedi_fpu): Ditto.
30690 2017-05-11 Julia Koval <julia.koval@intel.com>
30692 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30693 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30695 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30696 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30697 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30699 2017-05-11 Nathan Sidwell <nathan@acm.org>
30701 * graphite-poly.c: Include dumpfile.h.
30703 * dumpfle.h (dump_function): Declare here ...
30704 * tree-dump.h (dump_function): ... not here.
30705 * dumpfile.c: #include tree-cfg.h.
30706 (dump_function): Move here from ...
30707 * tree-dump.c (dump_function): ... here.
30708 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30709 * graphite-poly.c: Don't include tree-dump.h.
30710 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30711 * print-tree.c: Likewise.
30712 * stor-layout.c: Likewise.
30713 * tree-nested.c: Likewise.
30715 * dumpfile.c (dump_start): Use TDF_FLAGS.
30716 (dump_enable_all): Fix TDF_KIND check thinko.
30718 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30720 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30721 array entries to represent two legal parameterizations of the
30722 overloaded __builtin_cmpb function, as represented by the
30723 P6_OV_BUILTIN_CMPB constant.
30724 (altivec_resolve_overloaded_builtin): Add special case handling
30725 for the __builtin_cmpb function, as represented by the
30726 P6_OV_BUILTIN_CMPB constant.
30727 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30728 (BU_P6_64BIT_2): New macro.
30729 (BU_P6_OVERLOAD_2): New macro
30730 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30731 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30732 (CMPB): Add overload support to represent both 32-bit and 64-bit
30733 compare-bytes function.
30734 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30735 support for TARGET_CMPB.
30736 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30737 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30738 documentation of the __builtin_cmpb overloaded built-in function.
30740 2017-05-11 Richard Biener <rguenther@suse.de>
30742 PR tree-optimization/80705
30743 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30744 bases are not vectorizable.
30746 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30748 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30749 when counting register pressure.
30751 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30753 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30754 (iv_ca_more_deps): Renamed to ...
30755 (iv_ca_compare_deps): ... this.
30756 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30758 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30760 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30762 (determine_group_iv_costs): ... here.
30763 (find_inv_vars_cb): Record inv var if it's not recorded before.
30765 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30767 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30768 (get_shiftadd_cost): Ditto.
30770 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30772 * tree-ssa-address.c: Include header file.
30773 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30775 (add_to_parts): Refactor.
30776 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30777 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30780 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30782 PR tree-optimization/53090
30783 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30785 (extract_cond_operands): Detect condition with IV on both sides
30786 and return COMP_IV_EXPR_2.
30787 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30788 (rewrite_use_compare): Simplify by removing call to function
30789 extract_cond_operands.
30791 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30793 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30794 (extract_cond_operands): Detect condition comparing against non-
30795 invariant bound and return appropriate enum value.
30796 (find_interesting_uses_cond): Update use of extract_cond_operands.
30797 Handle its return value accordingly.
30798 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30800 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30802 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30803 nonlinear iv_use computation in loop invariant sensitive way.
30805 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30807 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30808 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30810 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30812 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30813 (dump_cand): Support iv_cand.inv_exprs.
30814 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30816 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30819 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30821 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30823 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30824 as local function. Include necessary header files.
30825 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30827 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30829 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30831 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30833 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30834 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30835 RSHIFT_EXPR and BIT_NOT_EXPR.
30837 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30839 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30840 (adjust_setup_cost): New parameter supporting round up adjustment.
30841 (struct address_cost_data): Delete.
30842 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30843 (split_address_cost, ptr_difference_cost): Delete.
30844 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30845 (struct ainc_cost_data): New struct.
30846 (get_address_cost_ainc): New function.
30847 (get_address_cost, get_computation_cost): Reimplement.
30848 (determine_group_iv_cost_address): Record inv_expr for all uses of
30850 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30851 (iv_ca_has_deps): Reimplemented to ...
30852 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30854 (iv_ca_extend): Call iv_ca_more_deps.
30856 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30858 * tree-ssa-address.c (struct mem_address): Move to header file.
30859 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30860 * tree-ssa-address.h (struct mem_address): Move from C file.
30861 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30863 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30865 * tree-affine.h (aff_combination_type): New interface.
30866 (aff_combination_zero_p): Remove static.
30867 (aff_combination_const_p): New interface.
30868 (aff_combination_singleton_var_p): New interfaces.
30870 2017-05-11 Richard Biener <rguenther@suse.de>
30872 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30873 Skip unreachable blocks and destinations.
30874 (eliminate): Move stmt removal and fixup ...
30875 (fini_eliminate): ... here. Skip inserted exprs.
30876 (pass_pre::execute): Move fini_pre after fini_eliminate.
30877 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30878 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30879 PRE to get rid of dead code that has invalid SSA form and
30880 split critical edges again.
30882 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30884 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30886 2017-05-11 Richard Biener <rguenther@suse.de>
30888 * passes.c (execute_function_todo): Verify loops if they are
30889 said to be up-to-date.
30890 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30891 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30893 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30896 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30897 handle calling assemble_external ourself.
30900 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30901 modes with zero size. Enhance comment.
30903 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30905 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30906 built-ins for vec_xl and vec_xst with short and char pointer
30909 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30911 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30912 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30913 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30914 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30915 (_mm_maskz_min_round_ss): New intrinsics.
30916 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30917 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30918 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30919 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30920 (__builtin_ia32_minss_mask_round): New builtins.
30921 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30922 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30923 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30925 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30926 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30928 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30931 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30933 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30934 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30935 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30936 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30937 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30938 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30939 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30940 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30941 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30942 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30943 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30944 (__builtin_ia32_mulss_mask_round): New builtins.
30945 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30946 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30947 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30949 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30950 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30952 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30955 2017-05-10 Julia Koval <julia.koval@intel.com>
30957 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30958 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30959 (_mm256_setr_m128i): New intrinsics.
30961 2017-05-10 Julia Koval <julia.koval@intel.com>
30963 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30964 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30965 (_mm_maskz_rcp14_ss): New intrinsics.
30966 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30967 (__builtin_ia32_rcp14ss_mask): New builtins.
30968 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30970 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30972 PR tree-optimization/51513
30973 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30974 (assert_unreachable_fallthru_edge_p): Use it.
30975 (group_case_labels_stmt): Likewise.
30976 * tree-cfg.h: Prototype it.
30977 * stmt.c: Include cfghooks.h and tree-cfg.h.
30978 (emit_case_dispatch_table) <gap_label>: New local variable.
30979 Use it to fill dispatch table gaps.
30980 Test for default_label before updating probabilities.
30981 (expand_case) <default_label>: Remove unneeded initialization.
30982 Test for unreachable default case statement and remove its edge.
30983 Set default_label accordingly.
30984 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30986 2017-05-10 Carl Love <cel@us.ibm.com>
30988 * config/rs6000/rs6000-c: Add support for built-in functions
30989 vector signed char vec_neg (vector signed char)
30990 vector signed short int vec_neg (vector short int)
30991 vector signed int vec_neg (vector signed int)
30992 vector signed long long vec_neg (vector signed long long)
30993 vector float vec_neg (vector float)
30994 vector double vec_neg (vector double)
30995 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30997 * config/rs6000/altivec.h: Add define for vec_neg
30998 * doc/extend.texi: Update the built-in documentation for the
30999 new built-in functions.
31001 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31003 PR tree-optimization/77644
31004 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31006 2017-05-10 Nathan Sidwell <nathan@acm.org>
31008 * dumpfile.h (TDI_lang_all): New.
31009 (TDF_KIND): New. Renumber others
31010 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31012 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31014 (get_dump_file_name): Adjust suffix generation.
31015 (dump_enable_all): Use TDF_KIND.
31016 * doc/invoke.texi (-fdump-lang-all): Document.
31018 * dumpfile.h: Tabify.
31020 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31023 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31024 Move member access before delete.
31026 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31028 * tree-inline.c (expand_call_inline): Split block at stmt
31031 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31034 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31035 are now unused after splitting mov{sf,sd}_hardfloat.
31036 (f32_lr2): Likewise.
31037 (f32_lm): Likewise.
31038 (f32_lm2): Likewise.
31039 (f32_li): Likewise.
31040 (f32_li2): Likewise.
31041 (f32_lv): Likewise.
31042 (f32_sr): Likewise.
31043 (f32_sr2): Likewise.
31044 (f32_sm): Likewise.
31045 (f32_sm2): Likewise.
31046 (f32_si): Likewise.
31047 (f32_si2): Likewise.
31048 (f32_sv): Likewise.
31049 (f32_dm): Likewise.
31050 (f32_vsx): Likewise.
31051 (f32_av): Likewise.
31052 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31053 For movsf, order stores so the VSX stores occur before the GPR
31054 store which encourages the register allocator to use a traditional
31055 FPR instead of a GPR. For movsd, order the stores so that the GPR
31056 store comes before the VSX stores to allow the power6 to work.
31057 This is due to the power6 not having a 32-bit integer store
31058 instruction from a FPR.
31059 (movsf_hardfloat): Likewise.
31060 (movsd_hardfloat): Likewise.
31062 2017-05-09 Martin Sebor <msebor@redhat.com>
31064 PR translation/80280
31065 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31068 PR translation/80280
31069 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31070 data member added in r247778.
31071 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31073 2017-05-09 Nathan Sidwell <nathan@acm.org>
31075 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31077 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31080 2017-05-09 Marek Polacek <polacek@redhat.com>
31082 * doc/invoke.texi: Fix typo.
31084 2017-05-09 Richard Biener <rguenther@suse.de>
31086 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31087 (vrp_val_is_min): Likewise.
31088 (set_value_range_to_value): Likewise.
31089 (set_value_range_to_nonnegative): Likewise.
31090 (gimple_assign_nonzero_p): Likewise.
31091 (gimple_stmt_nonzero_p): Likewise.
31092 (vrp_int_const_binop): Likewise. Remove unreachable case.
31093 (adjust_range_with_scev): Adjust comments.
31094 (compare_range_with_value): Likewise.
31095 (extract_range_from_phi_node): Likewise.
31096 (test_for_singularity): Likewise.
31098 2017-05-09 Richard Biener <rguenther@suse.de>
31100 * tree-vrp.c (get_single_symbol): Add assert that we don't
31101 get overflowed constants as invariant part.
31102 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31103 checks. Use wi::cmp instead of recursing for integer constants.
31104 (compare_values): Just ignore whether we assumed undefined
31105 overflow instead of failing the compare.
31106 (extract_range_for_var_from_comparison_expr): Add comment before the
31107 TREE_NO_WARNING sets.
31108 (test_for_singularity): Likewise.
31109 (extract_range_from_comparison): Do not disable optimization
31110 when we assumed undefined overflow.
31111 (extract_range_basic): Remove init of unused var.
31113 2017-05-09 Richard Biener <rguenther@suse.de>
31115 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31116 (extract_range_from_multiplicative_op_1): Adjust.
31117 (extract_range_from_binary_expr_1): Use int_const_binop.
31119 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31122 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31123 rs6000_store_data_bypass_p in seven define_bypass directives and
31124 in several comments.
31125 * config/rs6000/rs6000-protos.h: Add prototype for
31126 rs6000_store_data_bypass_p function.
31127 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31128 function implements slightly different (rs6000-specific) semantics
31129 than store_data_bypass_p, returning false rather than aborting
31130 with assertion error when arguments do not satisfy the
31131 requirements of store data bypass.
31132 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31133 rs6000_store_data_bypass_p.
31135 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31137 * config/xtensa/xtensa-protos.h
31138 (xtensa_initial_elimination_offset): New declaration.
31139 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31140 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31141 macro definition, add case for FRAME_POINTER_REGNUM when
31142 FRAME_GROWS_DOWNWARD.
31143 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31144 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31145 xtensa_initial_elimination_offset.
31147 2017-05-08 Nathan Sidwell <nathan@acm.org>
31149 * doc/invoke.texi: Alphabetize -fdump options.
31151 2017-05-08 Martin Sebor <msebor@redhat.com>
31153 PR translation/80280
31154 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31156 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31158 * target.def (compute_frame_layout): New optional target hook.
31159 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31160 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31161 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31163 * reload1.c (verify_initial_elim_offsets): Likewise.
31164 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31165 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31166 (arm_get_frame_offsets): Split up into this ...
31167 (arm_compute_frame_layout): ... and this function.
31169 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31171 * config/aarch64/constraints.md (Usa): New constraint.
31172 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31174 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31176 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31177 with_multilib_list after it has been checked.
31179 2017-05-08 Richard Biener <rguenther@suse.de>
31181 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31182 (bitmap_set_subtract_values): Likewise.
31184 2017-05-08 Richard Biener <rguenther@suse.de>
31186 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31187 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31189 (gimple_stmt_nonzero_warnv_p): Rename to ...
31190 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31192 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31193 (extract_range_basic): Adjust, do not disable propagation on
31194 strict overflow sensitive simplification.
31195 (vrp_visit_cond_stmt): Likewise.
31197 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31199 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31200 body size unconditionally.
31202 2017-05-07 Jeff Law <law@redhat.com>
31205 2017-05-06 Jeff Law <law@redhat.com>
31206 PR tree-optimization/78496
31207 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31210 PR tree-optimization/78496
31211 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31212 (simplify_stmt_using_ranges): Call it.
31213 (vrp_dom_walker::before_dom_children): Extract equivalences
31214 from an ASSERT_EXPR with an equality comparison against a
31217 2017-05-06 Jeff Law <law@redhat.com>
31219 PR tree-optimization/78496
31220 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31223 PR tree-optimization/78496
31224 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31225 (simplify_stmt_using_ranges): Call it.
31226 (vrp_dom_walker::before_dom_children): Extract equivalences
31227 from an ASSERT_EXPR with an equality comparison against a
31230 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31232 * lra-constraints.c (lra_copy_reg_equiv): New function.
31233 (split_reg): Use it to copy equivalence information from the
31234 original register to the spill register.
31236 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31238 PR rtl-optimization/75964
31239 * simplify-rtx.c (simplify_const_relational_operation): Remove
31240 invalid handling of comparisons of integer ABS.
31242 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31244 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31245 initialize to zero.
31246 (init_regs): Remove declaration.
31247 (function_arg_advance_32): Initialize error_p as boolean variable.
31249 2017-05-05 Nathan Sidwell <nathan@acm.org>
31251 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31252 lines. Use for (;;).
31254 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31256 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31257 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31258 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31259 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31260 VF=2 that require versioning.
31262 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31264 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31267 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31269 * diagnostic.h (diagnostic_override_option_index): Convert from
31270 macro to inline function.
31272 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31274 * diagnostic.c (last_module_changed_p): New function.
31275 (set_last_module): New function.
31276 (diagnostic_report_current_module): Convert macro usage to
31277 the above functions.
31278 * diagnostic.h (diagnostic_context::last_module): Strengthen
31279 from const line_map * to const line_map_ordinary *.
31280 (diagnostic_last_module_changed): Delete macro.
31281 (diagnostic_set_last_module): Delete macro.
31283 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31285 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31286 with diagnostic_report_diagnostic.
31287 (diagnostic_n_impl_richloc): Likewise.
31288 * diagnostic.h (report_diagnostic): Delete macro.
31289 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31290 with diagnostic_report_diagnostic.
31291 * substring-locations.c (format_warning_va): Likewise.
31293 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31295 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31296 save/restor of format_spec. Move option-printing code to...
31297 (print_option_information): ...this new function, and
31298 reimplement by simply printing to the pretty_printer,
31299 rather than appending to the format string.
31301 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31303 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31304 handling logic into...
31305 (update_effective_level_from_pragmas): ...this new function.
31307 2017-05-04 Andrew Waterman <andrew@sifive.com>
31309 * config/riscv/riscv.opt (mstrict-align): New option.
31310 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31311 (SLOW_UNALIGNED_ACCESS): Define.
31312 (riscv_slow_unaligned_access): Declare.
31313 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31315 (riscv_slow_unaligned_access): New variable.
31316 (rocket_tune_info): Set slow_unaligned_access to true.
31317 (optimize_size_tune_info): Set slow_unaligned_access to false.
31318 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31319 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31320 (riscv_option_override): Set riscv_slow_unaligned_access.
31321 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31323 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31325 * config/riscv/riscv.md: Unify indentation.
31327 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31332 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31334 (extendsi<mode>2): Add support for doing sign extension via
31335 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31336 don't have ISA 3.0 instructions.
31337 (extendsi<mode>2 splitter): Likewise.
31338 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31339 generate the normal insns since SImode can now go in vector
31340 registers. Disallow the special UNSPECs needed for previous
31341 machines to hide SImode being used. Add new insns
31342 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31343 (fix_trunc<mode>si2_stfiwx): Likewise.
31344 (fix_trunc<mode>si2_internal): Likewise.
31345 (fixuns_trunc<mode>si2): Likewise.
31346 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31347 (fctiw<u>z_<mode>_smallint): Likewise.
31348 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31349 of floating point to 32-bit integer from doing a direct move to
31350 the GPR registers to do a store.
31351 (fctiwz_<mode>): Break long line.
31353 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31355 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31356 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31357 (addr_list, addr_offset_valid_p): New.
31358 (split_address_groups): Check offset validity with above function.
31359 (gt-tree-ssa-loop-ivopts.h): Include header file.
31361 2017-05-05 Nathan Sidwell <nathan@acm.org>
31363 * config.gcc (arm*-*-*): Add missing 'fi'.
31365 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31367 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31368 included in -fopt-info does not matter.
31369 * doc/optinfo.texi (-fopt-info): Fix description of default
31370 behavour. Explicitly say order of options included in -fopt-info
31373 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31375 * config.gcc: Allow combinations of aprofile and rmprofile values for
31376 --with-multilib-list.
31377 * config/arm/t-multilib: New file.
31378 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31379 variables. Remove setting of ISA and floating-point ABI in
31380 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31381 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31382 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31383 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31385 * config/arm/t-rmprofile: Likewise except for the matches changes.
31386 * doc/install.texi (--with-multilib-list): Document the combination of
31387 aprofile and rmprofile values and warn about pitfalls in doing that.
31389 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31391 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31392 (movdi_aarch64): Likewise.
31394 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31396 PR tree-optimization/80632
31397 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31399 (build_arrays): Initialize it for virtual phis.
31400 (fix_phi_nodes): Use it for virtual phis.
31402 PR tree-optimization/80558
31403 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31404 [x, y] op z into [x op, y op z] for op & or | if conditions
31407 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31408 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31411 * config/arm/arm.md (use_literal_pool): Remove.
31412 (64-bit immediate split): No longer takes cost into consideration
31413 if arm_disable_literal_pool is enabled.
31414 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31415 used when arm_disable_literal_pool is enabled.
31416 (arm_max_const_double_inline_cost): Remove use of
31417 arm_disable_literal_pool.
31418 (push_minipool_fix): Add assert.
31419 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31420 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31421 (no_literal_pool_sf_immediate): New.
31423 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31425 PR tree-optimization/80613
31426 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31427 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31429 2017-05-05 Richard Biener <rguenther@suse.de>
31431 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31433 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31435 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31436 of this flag from insn conditions due to removal from r247495.
31438 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31440 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31442 (arm_early_store_addr_dep_ptr): Likewise.
31443 * config/arm/aarch-common-protos.h
31444 (arm_early_load_addr_dep_ptr): Add prototype.
31445 (arm_early_store_addr_dep_ptr): Likewise.
31446 * config/arm/cortex-a53.md: Add new bypasses.
31448 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31450 * tree.c (next_type_uid): Change type to unsigned.
31451 (type_hash_canon): Decrement back next_type_uid if
31452 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31453 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31456 2017-05-04 Martin Sebor <msebor@redhat.com>
31458 * builtins.c: Fix a trivial typo in a comment.
31460 PR middle-end/79234
31461 * builtins.c (check_sizes): Adjust to handle reading past the end.
31462 Avoid printing excessive upper bound of ranges. Use %E to print
31463 tree nodes instead of converting them to %wu.
31464 (expand_builtin_memchr): New function.
31465 (compute_dest_size): Rename...
31466 (compute_objsize): ...to this.
31467 (expand_builtin_memcpy): Adjust.
31468 (expand_builtin_mempcpy): Adjust.
31469 (expand_builtin_strcat): Adjust.
31470 (expand_builtin_strcpy): Adjust.
31471 (check_strncat_sizes): Adjust.
31472 (expand_builtin_strncat): Adjust.
31473 (expand_builtin_strncpy): Adjust and simplify.
31474 (expand_builtin_memset): Adjust.
31475 (expand_builtin_bzero): Adjust.
31476 (expand_builtin_memcmp): Adjust.
31477 (expand_builtin): Handle memcmp.
31478 (maybe_emit_chk_warning): Check strncat just once.
31480 2017-05-04 Martin Sebor <msebor@redhat.com>
31482 PR preprocessor/79214
31483 PR middle-end/79222
31484 PR middle-end/79223
31485 * builtins.c (check_sizes): Add inlining context and issue
31486 warnings even when -Wno-system-headers is set.
31487 (check_strncat_sizes): Same.
31488 (expand_builtin_strncat): Same.
31489 (expand_builtin_memmove): New function.
31490 (expand_builtin_stpncpy): Same.
31491 (expand_builtin): Handle memmove and stpncpy.
31493 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31495 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31496 which is not used any more.
31498 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31500 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31502 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31504 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31505 (cortexa53_tunings): Likewise.
31506 (cortexa57_tunings): Likewise.
31507 (cortexa72_tunings): Likewise.
31508 (cortexa73_tunings): Likewise.
31510 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31512 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31513 Set loop alignment to 8.
31515 2017-05-04 Martin Sebor <msebor@redhat.com>
31517 PR translation/80280
31518 * builtins.c (expand_builtin_object_size): Add missing quoting to
31519 %D and like directives.
31520 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31521 (hsa_type_for_tree_type): Same.
31522 (verify_function_arguments): Same.
31523 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31524 * varasm.c (get_section): Same.
31527 2017-05-04 Martin Sebor <msebor@redhat.com>
31529 PR translation/80280
31530 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31532 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31534 * config/aarch64/aarch64.c (generic_addrcost_table):
31535 Change HI/TI mode setting.
31537 2017-05-04 Martin Jambor <mjambor@suse.cz>
31539 PR tree-optimization/80622
31540 * tree-sra.c (comes_initialized_p): New function.
31541 (build_accesses_from_assign): Only set write lazily when
31542 comes_initialized_p is false.
31543 (analyze_access_subtree): Use comes_initialized_p.
31544 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31545 instead of testing for PARM_DECL.
31547 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31549 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31550 constraint on operand 0 to allow more general addressing modes.
31551 Adjust output template.
31552 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31554 * config/aarch64/aarch64-protos.h
31555 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31556 * config/aarch64/constraints.md (Dp): New address constraint.
31557 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31560 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31562 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31563 update use of estimate_ipcp_clone_size_and_time.
31564 (estimate_local_effects): Update use of
31565 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31566 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31567 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31568 Return nonspecialized time.
31570 2017-05-04 Richard Biener <rguenther@suse.de>
31572 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31573 for the last VUSE which def dominates the PHI. Directly call
31575 (get_continuation_for_phi_1): Remove.
31577 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31579 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31580 to explain the use of truncating division. Cap the number of
31581 iterations to the maximum given by nb_iterations_upper_bound,
31584 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31586 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31587 * configure: Regenerate.
31588 * config.in: Regenerate.
31589 * config/i386/driver-mingw32.c: new file.
31590 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31591 * config.host: Link driver-mingw32.o on MinGW host.
31592 * doc/install.texi: Document new --enable-mingw-wildcard configure
31595 2017-05-04 Marek Polacek <polacek@redhat.com>
31597 PR tree-optimization/80612
31598 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31600 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31601 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31603 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31604 (movt splitter): Likewise.
31605 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31606 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31607 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31608 block for Thumb-1 with MOVT.
31609 (thumb2_legitimate_address_p): Move code block ...
31610 (can_avoid_literal_pool_for_label_p): ... into this new function.
31611 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31613 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31614 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31615 "M-profile targets with the MOVT instruction".
31617 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31619 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31620 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31621 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31623 2017-05-04 Martin Liska <mliska@suse.cz>
31625 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31626 variable cond_code.
31628 2017-05-04 Richard Biener <rguenther@suse.de>
31630 * tree.c (array_at_struct_end_p): Handle arrays at struct
31631 end with flexarrays more conservatively. Refactor and treat
31632 arrays of arrays or aggregates more strict. Fix
31633 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31634 * tree.c (array_at_struct_end_p): Adjust prototype.
31635 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31636 * gimple-fold.c (get_range_strlen): Likewise.
31637 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31639 2017-05-04 Richard Biener <rguenther@suse.de>
31641 PR tree-optimization/31130
31642 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31644 (supports_overflow_infinity): Likewise.
31645 (is_negative_overflow_infinity): Likewise.
31646 (is_positive_overflow_infinity): Likewise.
31647 (is_overflow_infinity): Likewise.
31648 (stmt_overflow_infinity): Likewise.
31649 (overflow_infinity_range_p): Likewise.
31650 (usable_range_p): Remove as always returning true.
31651 (make_overflow_infinity): Remove.
31652 (negative_overflow_infinity): Likewise.
31653 (positive_overflow_infinity): Likewise.
31654 (avoid_overflow_infinity): Likewise.
31655 (set_value_range): Adjust accordingly.
31656 (set_value_range_to_nonnegative): Likewise, remove now unused
31657 overflow_infinity arg.
31658 (vrp_operand_equal_p): Adjust.
31659 (update_value_range): Likewise.
31660 (range_int_cst_singleton_p): Likewise.
31661 (operand_less_p): Likewise.
31662 (compare_values_warnv): Likewise.
31663 (extract_range_for_var_from_comparison_expr): Likewise.
31664 (vrp_int_const_binop): Likewise.
31665 (zero_nonzero_bits_from_vr): Likewise.
31666 (extract_range_from_multiplicative_op_1): Likewise.
31667 (extract_range_from_binary_expr_1): Likewise.
31668 (extract_range_from_unary_expr): Likewise.
31669 (extract_range_from_comparison): Likewise.
31670 (extract_range_basic): Likewise.
31671 (adjust_range_with_scev): Likewise.
31672 (compare_ranges): Likewise.
31673 (compare_range_with_value): Likewise.
31674 (dump_value_range): Likewise.
31675 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31677 (simplify_cond_using_ranges): Adjust.
31679 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31681 * brig-builtins.def: Added a builtin for class_f64.
31682 * builtin-types.def: Added a builtin type needed by class_f64.
31684 2017-05-03 Jason Merrill <jason@redhat.com>
31686 * timevar.def: Add TV_CONSTEXPR.
31688 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31690 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31692 2017-05-03 Martin Jambor <mjambor@suse.cz>
31694 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31695 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31696 * ipa-cp.c (ipcp_propagate_stage): Do not call
31697 ipa_update_after_lto_read.
31698 * ipa-inline.c (ipa_inline): Likewise.
31700 2017-05-03 Martin Jambor <mjambor@suse.cz>
31702 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31703 tag. Added a default constructor and a destructor.
31704 (ipa_edge_args_sum_t): New class;
31705 (ipa_edge_args_sum): Declare.
31706 (ipa_edge_args_vector): Remove declaration.
31707 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31708 (ipa_free_edge_args_substructures): Remove declaration.
31709 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31710 (ipa_edge_args_info_available_for_edge_p): Likewise.
31711 * ipa-prop.c (ipa_edge_args_vector): Removed.
31712 (edge_removal_hook_holder): Likewise.
31713 (edge_duplication_hook_holder): Likewise.
31714 (ipa_edge_args_sum): New variable.
31715 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31716 ipa_edge_args_vector.
31717 (ipa_free_edge_args_substructures): Likewise.
31718 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31719 ipa_edge_args_vector.
31720 (ipa_edge_removal_hook): Turned into method
31721 ipa_edge_args_sum_t::remove.
31722 (ipa_edge_duplication_hook): Turned into method
31723 ipa_edge_args_sum_t::duplicate.
31724 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31725 registering edge hooks.
31726 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31727 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31728 ipa_edge_args_sum instead of ipa_edge_args_vector.
31729 * ipa-profile.c (ipa_profile): Likewise.
31731 2017-05-03 Martin Jambor <mjambor@suse.cz>
31733 * symbol-summary.h (function_summary): New method exists.
31734 (function_summary::symtab_removal): Deallocate through release.
31735 (call_summary): New class.
31736 (gt_ggc_mx): New overload.
31737 (gt_pch_nx): Likewise.
31738 (gt_pch_nx): Likewise.
31740 2017-05-03 Jeff Law <law@redhat.com>
31742 PR tree-optimization/78496
31743 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31744 from simplify_cond_using_ranges. Split off code to walk
31745 backwards through casts into ...
31746 (simplify_cond_using_ranges_2): New function.
31747 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31748 (execute_vrp): After identifying jump threads, call
31749 simplify_cond_using_ranges_2.
31751 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31754 * ipa-inline.h (inline_summary): Add ctor.
31755 (create_ggc): Do not use ggc_cleared_alloc.
31757 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31758 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31760 * gcc.c (handle_braces): Support escaping in switch matching
31762 * doc/invoke.texi (Spec Files): Document it.
31763 Remove superfluous @code markup in items.
31765 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31767 * diagnostic-show-locus.c (struct column_range): New struct.
31768 (get_affected_columns): New function.
31769 (get_printed_columns): New function.
31770 (struct correction): New struct.
31771 (correction::ensure_capacity): New function.
31772 (correction::ensure_terminated): New function.
31773 (struct line_corrections): New struct.
31774 (line_corrections::~line_corrections): New dtor.
31775 (line_corrections::add_hint): New function.
31776 (layout::print_trailing_fixits): Reimplement in terms of the new
31778 (selftest::test_overlapped_fixit_printing): New function.
31779 (selftest::diagnostic_show_locus_c_tests): Call it.
31781 2017-05-03 Nathan Sidwell <nathan@acm.org>
31783 Canonicalize canonical type hashing
31784 * tree.h (type_hash_canon_hash): Declare.
31785 * tree.c (type_hash_list, attribute_hash_list): Move into
31786 type_hash_canon_hash.
31787 (build_type_attribute_qual_variant): Break out hash code calc into
31788 type_hash_canon_hash.
31789 (type_hash_canon_hash): New. Generic type hash computation.
31790 (build_range_type_1, build_array_type_1, build_function_type,
31791 build_method_type_directly, build_offset_type, build_complex_type,
31792 make_vector_type): Call it.
31794 2017-05-03 Richard Biener <rguenther@suse.de>
31796 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31797 When all DRs have unknown misaligned do not always peel
31798 when there is a store but apply the same costing model as if
31799 there were only loads.
31801 2017-05-03 Richard Biener <rguenther@suse.de>
31804 PR tree-optimization/80492
31805 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31806 compare_base_decls returning dont-know properly.
31808 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31810 * config/arm/iterators.md (CCSI): New mode iterator.
31811 (arch): New mode attribute.
31812 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31813 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31814 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31815 code iterator for success result mode.
31816 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31817 the corresponding new insn generators.
31819 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31822 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31823 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31825 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31827 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31828 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31829 (DDR_A): Wrap DDR argument in brackets.
31830 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31831 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31832 (DDR_REVERSED_P): Likewise.
31834 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31836 PR tree-optimization/79472
31837 * tree-switch-conversion.c (struct switch_conv_info): Add
31838 contiguous_range and default_case_nonstandard fields.
31839 (collect_switch_conv_info): Compute contiguous_range and
31840 default_case_nonstandard fields, don't clear final_bb if
31841 contiguous_range and only the default case doesn't have the required
31843 (check_all_empty_except_final): Set default_case_nonstandard instead
31844 of failing if contiguous_range and the default case doesn't have empty
31846 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31847 and only the default case doesn't have the required constants. Skip
31849 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31850 if default_case_nonstandard.
31851 (build_constructors): Build constant 1 just once. Assert that default
31852 values aren't inserted in between cases if contiguous_range. Skip
31854 (build_arrays): Skip virtual phis.
31855 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31856 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31857 Handle virtual phis.
31858 (gen_inbound_check): Handle default_case_nonstandard case.
31859 (process_switch): Adjust check_final_bb caller. Call
31860 gather_default_values with the first non-default case instead of
31861 default case if default_case_nonstandard.
31863 2017-05-02 Nathan Sidwell <nathan@acm.org>
31865 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31866 check. Fix formatting.
31868 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31870 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31871 errors when comparing specialized and unspecialized times.
31873 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31875 * diagnostic-show-locus.c
31876 (layout::should_print_annotation_line_p): Make private.
31877 (layout::print_annotation_line): Make private.
31878 (layout::annotation_line_showed_range_p): Make private.
31879 (layout::show_ruler): Make private.
31880 (layout::print_source_line): Make private. Pass in line and
31881 line_width, rather than calling location_get_source_line. Drop
31883 (layout::print_leading_fixits): New method.
31884 (layout::print_any_fixits): Rename to...
31885 (layout::print_trailing_fixits): ...this, and make private.
31886 Don't print newline fixits.
31887 (diagnostic_show_locus): Move logic for printing one row into...
31888 (layout::print_line): ...this new function. Move the
31889 location_get_source_line call and error-handling from
31890 print_source_line to here. Call print_leading_fixits, and rename
31891 print_any_fixits to print_trailing_fixits.
31892 (selftest::test_fixit_insert_containing_newline): Update now that
31893 newlines are partially supported.
31894 (selftest::test_fixit_insert_containing_newline_2): New test.
31895 (selftest::test_fixit_replace_containing_newline): Update comments.
31896 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31897 * edit-context.c (class added_line): New class.
31898 (class edited_line): Describe newline handling in comment.
31899 (edited_line::actually_edited_p): New method.
31900 (edited_line::print_content): Delete redundant decl.
31901 (edited_line::m_predecessors): New field.
31902 (edited_file::print_content): Call edited_line::print_content.
31903 (edited_file::print_diff): Update to support newlines.
31904 (edited_file::print_diff_hunk): Likewise.
31905 (edited_file::print_run_of_changed_lines): New function.
31906 (edited_file::print_diff_line): Convert to...
31907 (print_diff_line): ...this.
31908 (edited_file::get_effective_line_count): New function.
31909 (edited_line::edited_line): Initialize new field m_predecessors.
31910 (edited_line::~edited_line): Clean up m_predecessors.
31911 (edited_line::apply_fixit): Handle newlines.
31912 (edited_line::get_effective_line_count): New function.
31913 (edited_line::print_content): New function.
31914 (edited_line::print_diff_lines): New function.
31915 (selftest::test_applying_fixits_insert_containing_newline): New
31917 (selftest::test_applying_fixits_replace_containing_newline): New
31919 (selftest::insert_line): New function.
31920 (selftest::test_applying_fixits_multiple_lines): Add example of
31922 (selftest::edit_context_c_tests): Call the new tests.
31924 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31926 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31927 parameter cand. Update dump information.
31928 (get_computation_cost): Update uses.
31930 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31932 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31933 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31934 (get_computation_at, rewrite_use_address): Update use of
31935 get_computation_aff.
31937 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31939 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31940 (get_computation): Delete.
31941 (get_computation_cost): Implement like get_computation_cost_at.
31942 Use get_computation_at.
31943 (get_computation_cost_at): Delete.
31944 (rewrite_use_nonlinear_expr): Use get_computation_at.
31945 (rewrite_use_compare, remove_unused_ivs): Ditto.
31947 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31949 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31951 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31953 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31954 (ivopts_global_cost_for_size): Rename parameter and update uses.
31955 (iv_ca_recount_cost): Update uses.
31956 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31957 candidates seperately in n_invs and n_cands.
31958 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31960 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31962 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31963 (find_inv_vars_cb): New.
31964 (find_depends): Renamed to ...
31965 (find_inv_vars): ... this.
31966 (add_candidate_1, force_var_cost): Call find_inv_vars.
31967 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31969 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31971 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31972 inv_vars. Add inv_exprs.
31973 (struct iv_cand): Rename depends_on to inv_vars.
31974 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31975 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31976 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31977 (dump_cand): Dump inv_vars.
31978 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31979 (record_invariant, find_depends, add_candidate_1): Ditto.
31980 (set_group_iv_cost, force_var_cost): Ditto.
31981 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31982 (get_computation_cost_at, get_computation_cost): Ditto.
31983 (determine_group_iv_cost_generic): Ditto.
31984 (determine_group_iv_cost_address): Ditto.
31985 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31986 (determine_group_iv_costs): Ditto.
31987 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31988 (iv_ca_set_remove_invariants): Renamed to ...
31989 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31990 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31991 (iv_ca_set_add_invariants): Renamed to ...
31992 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31993 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31994 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31995 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31996 (create_new_ivs): Remove useless dump.
31998 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32000 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32002 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32003 (iv_ca_set_no_cp, create_new_iv): Ditto.
32005 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32007 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32009 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32011 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32012 function tree_check2.
32014 2017-05-02 Martin Liska <mliska@suse.cz>
32016 * doc/gcov.texi: Add missing preposition.
32017 * gcov.c (function_info::function_info): Properly fill up
32018 all member variables.
32020 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32022 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32024 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32026 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32028 2017-05-02 Martin Liska <mliska@suse.cz>
32031 * lto-streamer-in.c (lto_read_tree_1): Remove
32032 LTO_STREAMER_DEBUG.
32033 * lto-streamer.c (struct tree_hash_entry): Likewise.
32034 (struct tree_entry_hasher): Likewise.
32035 (tree_entry_hasher::hash): Likewise.
32036 (tree_entry_hasher::equal): Likewise.
32037 (lto_streamer_init): Likewise.
32038 (lto_orig_address_map): Likewise.
32039 (lto_orig_address_get): Likewise.
32040 (lto_orig_address_remove): Likewise.
32041 * lto-streamer.h: Likewise.
32042 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32043 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32045 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32047 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32048 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32049 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32050 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32051 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32052 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32053 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32054 (mm_maskz_sub_ss): New intrinsics.
32055 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32056 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32057 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32058 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32059 (__builtin_ia32_subss_mask_round): New builtins.
32060 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32061 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32062 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32064 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32065 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32067 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32070 2017-05-02 Martin Jambor <mjambor@suse.cz>
32072 PR tree-optimization/78687
32073 * tree-sra.c (access): New field parent.
32074 (process_subtree_disqualification): New function.
32075 (disqualify_candidate): Call it.
32076 (build_accesses_from_assign): Reset write flag if creating an
32078 (build_access_subtree): Fill in parent field and also prpagate
32079 down grp_write flag.
32080 (create_artificial_child_access): New parameter set_grp_write, set
32081 grp_write to its value.
32082 (propagate_subaccesses_across_link): Also propagate grp_write flag
32084 (propagate_all_subaccesses): Push the closest parent back to work
32085 queue if add_access_to_work_queue returned true.
32087 2017-05-02 Richard Biener <rguenther@suse.de>
32089 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32090 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32091 -fstrict-overflow documentation.
32092 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32093 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32094 flag_strict_overflow.
32095 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32096 * lto-opts.c (lto_write_options): Do not stream it.
32097 * lto-wrapper.c (merge_and_complain): Do not handle it.
32098 * opts.c (default_options_table): Do not set -fstrict-overflow.
32099 (finish_options): Likewise do not clear it when sanitizing.
32100 * simplify-rtx.c (simplify_const_relational_operation): Do not
32101 test flag_strict_overflow.
32103 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32105 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32106 using enabled attribute.
32107 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32108 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32109 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32110 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32111 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32112 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32113 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32114 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32115 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32116 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32118 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32120 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32122 2017-05-02 Richard Biener <rguenther@suse.de>
32124 PR tree-optimization/80591
32126 2017-04-10 Richard Biener <rguenther@suse.de>
32128 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32131 2017-05-02 Richard Biener <rguenther@suse.de>
32133 PR tree-optimization/80549
32134 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32135 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32136 headers if they do not have a preheader.
32138 2017-05-02 Martin Liska <mliska@suse.cz>
32141 * common.opt: Fix typo.
32142 * doc/invoke.texi: Likewise.
32144 2017-05-01 Jan Beulich <jbeulich@suse.com>
32146 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32147 swapping, add (x,x,m,x,n) alternative.
32149 2017-05-01 Nathan Sidwell <nathan@acm.org>
32151 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32152 unnecessary unadjusted_alignment check.
32154 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32157 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32159 * gimplify.c (gimplify_cilk_detach): New function.
32160 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32161 * tree-core.h: Document EXPR_CILK_SPAWN.
32162 * tree.h (EXPR_CILK_SPAWN): Define.
32164 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32166 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32167 to use new fixit_hint representation, using the "replace" logic.
32168 (get_line_span_for_fixit_hint): Likewise.
32169 (layout::print_any_fixits): Likewise.
32170 (selftest::test_one_liner_many_fixits): Rename to...
32171 (selftest::test_one_liner_many_fixits_1): ...this, and update
32172 comment and expected output to reflect that the multiple fix-it
32173 hints are now consolidated into one insertion.
32174 (selftest::test_one_liner_many_fixits_2): New test.
32175 (selftest::test_diagnostic_show_locus_one_liner): Update for
32177 (selftest::test_fixit_consolidation): Update for fix-it API
32179 * diagnostic.c (print_parseable_fixits): Likewise.
32180 * edit-context.c (edited_line::m_line_events): Convert from
32181 auto_vec <line_event *> to auto_vec <line_event>.
32182 (class line_event): Convert from abstract base class to a concrete
32183 class, taking over the role of replace_event.
32184 (class insert_event): Delete.
32185 (class replace_event): Rename to class line_event. Convert to
32187 (edit_context::add_fixits): Reimplement.
32188 (edit_context::apply_insert): Delete.
32189 (edit_context::apply_replace): Rename to...
32190 (edit_context::apply_fixit): ...this. Convert to half-open range.
32191 (edited_file::apply_insert): Delete.
32192 (edited_file::apply_replace): Rename to...
32193 (edited_file::apply_fixit): ...this.
32194 (edited_line::~edited_line): Drop deletion of events.
32195 (edited_line::apply_insert): Delete.
32196 (edited_line::apply_replace): Rename to...
32197 (edited_line::apply_fixit): ...this. Convert to half-open range.
32198 Update for change to type of m_line_events.
32199 * edit-context.h (edit_context::apply_insert): Delete.
32200 (edit_context::apply_replace): Rename to...
32201 (edit_context::apply_fixit): ...this.
32203 2017-05-01 Martin Sebor <msebor@redhat.com>
32205 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32208 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32211 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32212 __get_cpuid_max returns 0.
32213 (__get_cpuid_count): Ditto.
32215 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32217 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32218 replacement expression is another instance of one of its arguments.
32220 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32223 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32224 check for stack push/pop autoinc.
32225 * config/i386/i386.c (ix86_agi_dependent): Return false
32226 if the only reason why modified_in_p returned true is that
32227 addr is SP based and set_insn is a push or pop.
32229 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32231 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32234 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32237 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32238 (account_size_time): Use two predicates - exec_pred and
32240 (evaluate_conditions_for_known_args): Compute both clause and
32242 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32243 (inline_summary_t::duplicate): Update.
32244 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32246 (compute_inline_parameters): Likewise.
32247 (estimate_edge_size_and_time): Update caluclation of time.
32248 (estimate_node_size_and_time): Compute both time and nonspecialized
32250 (estimate_ipcp_clone_size_and_time): Update.
32251 (inline_merge_summary): Update.
32252 (do_estimate_edge_time): Update.
32253 (do_estimate_edge_size): Update.
32254 (do_estimate_edge_hints): Update.
32255 (inline_read_section, inline_write_summary): Stream both new predicates.
32256 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32258 (compute_inlined_call_time): Cleanup.
32259 (big_speedup_p): Update.
32260 (edge_badness): Update.
32261 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32262 (size_time_entry): Replace predicate by exec_predicate and
32263 nonconst_predicate.
32264 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32265 (estimate_edge_time): Return also nonspec_time.
32266 (reset_edge_growth_cache): Update.
32268 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32270 PR rtl-optimization/80491
32271 * ifcvt.c (noce_process_if_block): When looking for x setter
32272 with missing else_bb, don't check only the insn right before
32273 cond_earliest, but look for the last insn that x is modified in
32274 within the same bb.
32276 PR rtl-optimization/80491
32277 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32279 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32281 PR tree-optimization/80487
32282 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32284 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32286 PR tree-optimization/79697
32287 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32288 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32289 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32291 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32292 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32294 2017-04-28 Martin Sebor <msebor@redhat.com>
32296 PR tree-optimization/80523
32297 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32298 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32300 (maybe_warn, format_directive, parse_directive): Use new functions.
32301 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32303 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32305 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32307 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32309 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32310 target_header_dir): Set correctly.
32311 * configure: Regenerated.
32312 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32313 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32314 instead of SYSTEM_HEADER_DIR.
32316 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32318 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32319 (estimate_local_effects): Likewise.
32320 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32321 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32322 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32323 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32324 do_estimate_edge_time, estimate_edge_time): Likewise.
32325 * ipa-inline-analysis.c (estimate_node_size_and_time,
32326 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32327 (estimate_time_after_inlining): Remove.
32329 2017-04-28 Martin Liska <mliska@suse.cz>
32331 * doc/gcov.texi: Enhance documentation of gcov.
32333 2017-04-28 Martin Liska <mliska@suse.cz>
32335 * doc/gcov.texi: Sort options in alphabetic order.
32336 * doc/gcov-dump.texi: Likewise.
32337 * doc/gcov-tool.texi: Likewise.
32338 * gcov.c (print_usage): Likewise.
32339 * gcov-dump.c (print_usage): Likewise.
32340 * gcov-tool.c (print_merge_usage_message): Likewise.
32341 (print_rewrite_usage_message): Likewise.
32342 (print_overlap_usage_message): Likewise.
32344 2017-04-28 Martin Liska <mliska@suse.cz>
32346 PR gcov-profile/53915
32347 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32349 2017-04-28 Martin Liska <mliska@suse.cz>
32351 PR gcov-profile/79891
32352 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32353 is marked by compiler as living on a line.
32354 (get_cycles_count): Remove usage of the union.
32355 (output_intermediate_file): Likewise.
32356 (find_source): Fix GNU coding style.
32357 (accumulate_line_counts): Remove old non-all block mode.
32358 (output_lines): Remove usage of the union.
32359 * profile.c (output_location): Include all BBs, even if
32360 belonging to a same line (and file) as a previous BB.
32362 2017-04-28 Martin Liska <mliska@suse.cz>
32364 * gcov.c (process_args): Handle new argument 'w'.
32365 (read_graph_file): Assign ID to BBs.
32366 (output_branch_count): Display BB # if verbose flag is set.
32367 (output_lines): Likewise for arcs.
32368 (print_usage): Add '--verbose' option help.
32369 * doc/gcov.texi: Document --verbose (-w) option.
32371 2017-04-28 Martin Liska <mliska@suse.cz>
32373 * gcov.c (struct block_location_info): New struct.
32374 (process_file): Fill up the new structure.
32375 (read_graph_file): Replace usage of encoding by the newly added
32377 (add_line_counts): Likewise.
32378 (accumulate_line_counts): Remove usage of the union.
32379 (function_info::function_info): New function.
32380 (function_info::~function_info): Likewise.
32381 (process_file): Call delete instead of release_function.
32382 (release_function): Release the function.
32383 (release_structures): Call delete instead of release_function.
32384 (solve_flow_graph): Replace usage of num_blocks.
32385 (find_exception_blocks): Likewise.
32386 (output_lines): Fix GNU coding style.
32388 2017-04-28 Martin Liska <mliska@suse.cz>
32391 * coverage.c (coverage_remove_note_file): New function.
32392 * coverage.h: Declare the function.
32393 * toplev.c (finalize): Clean if an error has been seen.
32395 2017-04-28 Martin Liska <mliska@suse.cz>
32397 PR gcov-profile/80031
32398 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32399 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32400 * gcov.c (read_graph_file): Read just number of blocks.
32401 * profile.c (branch_prob): Do not stream 0 flags per a basic
32404 2017-04-28 Martin Liska <mliska@suse.cz>
32406 * gcov-dump.c (tag_*): Add new argument to declarations.
32407 (dump_gcov_file): Likewise.
32408 (tag_blocks): Add and use new argument depth.
32409 (tag_arcs): Likewise.
32410 (tag_lines): Likewise.
32411 (tag_counters): Likewise.
32412 (tag_summary): Likewise.
32413 (dump_working_sets): Use depth to do a proper indentation.
32415 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32418 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32419 * symtab.c (symtab_node::debug_symtab): Move definition here.
32421 2017-04-28 Richard Biener <rguenther@suse.de>
32423 * lto-streamer.h (LTO_major_version): Bump to 7.
32425 2017-04-28 Richard Biener <rguenther@suse.de>
32427 * tree-vrp.c (assert_info): New struct.
32428 (add_assert_info): New helper.
32429 (register_edge_assert_for_2): Refactor to add asserts to a vector
32431 (register_edge_assert_for_1): Likewise.
32432 (register_edge_assert_for): Likewise.
32433 (finish_register_edge_assert_for): New helper actually registering
32434 asserts where live on edge.
32435 (find_conditional_asserts): Adjust.
32436 (find_switch_asserts): Likewise.
32437 (evrp_dom_walker::try_find_new_range): Generalize.
32438 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32440 2017-04-27 Marek Polacek <polacek@redhat.com>
32443 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32444 arg10 and arg11 to itype.
32446 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32448 * doc/extend.texi (Object Size Checking): Improve grammar.
32450 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32453 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32454 that the logic for permitting reciprocal estimates matches that
32457 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32460 * tree.c (type_cache_hasher::equal): Only compare
32461 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32462 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32463 non-aggregate element types.
32464 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32465 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32467 2017-04-27 Richard Biener <rguenther@suse.de>
32469 PR middle-end/80533
32470 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32471 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32472 keeping a reference to a trailing array.
32474 2017-04-27 Richard Biener <rguenther@suse.de>
32476 PR middle-end/80539
32477 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32478 being in loop-closed SSA form conservatively.
32479 (chrec_fold_multiply_poly_poly): Likewise.
32481 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32483 PR middle-end/79665
32484 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32485 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32487 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32490 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32491 (aarch64_function_arg_alignment): Return unsigned int again, but still
32492 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32493 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32494 Don't emit -Wpsabi note.
32495 (aarch64_function_arg_boundary): Likewise.
32496 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32499 2017-04-26 Nathan Sidwell <nathan@acm.org>
32501 * tree.h (crc32_unsigned_n): Declare.
32502 (crc32_unsigned, crc32_unsigned): Make inline.
32503 * tree.c (crc32_unsigned_bits): Replace with ...
32504 (crc32_unsigned_n): ... this.
32505 (crc32_unsigned, crc32_byte): Remove.
32506 (crc32_string): Remove unnecessary braces.
32508 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32510 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32511 * ipa-inline-analysis.c (MAX_TIME): Remove.
32512 (account_size_time): Use sreal for time.
32513 (dump_inline_summary): Update.
32514 (estimate_function_body_sizes): Update.
32515 (estimate_edge_size_and_time): Update.
32516 (estimate_calls_size_and_time): Update.
32517 (estimate_node_size_and_time): Update.
32518 (inline_merge_summary): Update.
32519 (inline_update_overall_summary): Update.
32520 (estimate_time_after_inlining): Update.
32521 (inline_read_section): Update.
32522 (inline_write_summary): Update.
32523 * ipa-inline.c (compute_uninlined_call_time): Update.
32524 (compute_inlined_call_time): Update.
32525 (recursive_inlining): Update.
32526 (inline_small_functions): Update.
32527 (dump_overall_stats): Update.
32528 * ipa-inline.h: Include sreal.h.
32529 (size_time_entry): Turn time to sreal.
32530 (inline_summary): Turn self_time nad time to sreal.
32532 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32534 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32536 (sreal::stream_out, sreal::stream_in): New.
32537 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32539 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32541 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32544 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32547 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32549 (dimode_scalar_chain::compute_convert_gain): Ditto.
32550 (dimode_scalar_chain::make_vector_copies): Ditto.
32551 (dimode_scalar_chain::convert_reg): Ditto.
32552 (dimode_scalar_chain::convert_insn): Ditto.
32553 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32554 (VI248_AVX512BW_1): New mode iterator.
32555 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32556 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32559 2017-04-25 Martin Sebor <msebor@redhat.com>
32561 PR tree-optimization/80497
32562 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32563 constants are representable in HOST_WIDE_INT.
32564 (parse_directive): Ditto.
32566 2017-04-25 Martin Sebor <msebor@redhat.com>
32569 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32570 (new_zero_array): Adjust signature.
32571 (dom_info::dom_init): Used unsigned rather that size_t.
32572 (dom_info::dom_info): Same.
32574 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32575 Jakub Jelinek <jakub@redhat.com>
32578 * config/arm/arm.c: Include gimple.h.
32579 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32580 returns negative, increment ncrn only if it returned positive.
32581 (arm_needs_doubleword_align): Return int instead of bool,
32582 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32583 members, but if there is any such non-FIELD_DECL
32584 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32585 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32586 returns negative, increment nregs only if it returned positive.
32587 (arm_setup_incoming_varargs): Likewise.
32588 (arm_function_arg_boundary): Emit -Wpsabi note if
32589 arm_needs_doubleword_align returns negative, return
32590 DOUBLEWORD_ALIGNMENT only if it returned positive.
32592 2017-04-25 Marek Polacek <polacek@redhat.com>
32595 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32596 first argument to type.
32598 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32601 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32602 type checks to test for compatibility instead of equality.
32604 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32605 Jakub Jelinek <jakub@redhat.com>
32608 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32610 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32611 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32612 the alignment computation, but return their maximum in warn_alignment.
32613 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32614 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32616 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32617 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32620 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32622 * config/arc/simdext.md (dmpyh): Fix typo.
32624 2017-04-25 Richard Biener <rguenther@suse.de>
32626 PR tree-optimization/80492
32627 * alias.c (compare_base_decls): Handle registers with asm
32628 specification conservatively.
32629 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32630 compare_base_decls returning dont-know properly.
32632 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32634 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32635 (legitimate_offset_address_p): New function.
32636 (arc_legitimate_address_p): Use above function.
32638 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32640 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32642 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32644 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32645 ACCH registers whenever they are available.
32647 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32649 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32650 double regs fix when not used.
32652 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32654 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32656 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32657 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32659 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32661 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32662 when emitting short ADD instructions.
32664 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32666 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32668 (cmpsi_cc_c_insn): Likewise.
32669 (cbranchsi4_scratch): Compute proper instruction length using
32670 compact_hreg_operand.
32671 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32673 2017-04-25 Richard Biener <rguenther@suse.de>
32675 PR middle-end/80509
32676 * passes.c (pass_manager::pass_manager): Initialize
32677 m_name_to_pass_map.
32679 2017-04-25 Richard Biener <rguenther@suse.de>
32681 PR tree-optimization/79201
32682 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32684 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32687 * config/s390/vector.md: Split MEM->GPR vector moves for
32688 non-s_operand addresses.
32690 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32693 * config/s390/predicates.md (reload_const_wide_int_operand): New
32695 * config/s390/s390.md ("movti"): Remove d/P alternative.
32696 ("movti_bigconst"): New pattern definition.
32698 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32701 * s390-protos.h (s390_expand_cs_hqi): Removed.
32702 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32703 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32704 modes as well as CCZ1mode and CCZmode.
32705 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32706 signature of s390_emit_compare_and_swap.
32707 (s390_expand_cs_hqi): Likewise, make static.
32708 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32709 compare-and-swap, in some cases.
32710 (s390_expand_cs): Wrapper function.
32711 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32713 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32714 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32715 patterns for small and large integers. Forbid symref memory operands.
32716 Move expander to s390.c. Require cc register.
32717 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32718 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32719 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32720 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32721 symref memory operands. Remove CC mode and call s390_match_ccmode
32723 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32725 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32727 * config/s390/s390.md (define_peephole2): New peephole to help
32728 combining the load-and-test pattern with volatile memory.
32730 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32732 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32733 with CCZmode for TARGET_Z196.
32735 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32737 PR rtl-optimization/80501
32738 * combine.c (make_compound_operation_int): Set subreg_code to SET
32739 even for AND with mask of the sign bit of mode.
32741 PR rtl-optimization/80500
32742 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32743 sum's initial value.
32745 2017-04-25 Julian Brown <julian@codesourcery.com>
32746 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32748 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32750 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32752 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32754 2017-04-25 Julian Brown <julian@codesourcery.com>
32755 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32757 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32758 (thunderx2t99_sha): New Reservation.
32760 2017-04-25 Julian Brown <julian@codesourcery.com>
32761 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32763 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32764 type for 1-element load.
32766 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32768 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32770 2017-04-24 Martin Jambor <mjambor@suse.cz>
32772 PR tree-optimization/80293
32773 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32774 char arrays not totally scalarizable if it is false.
32775 (analyze_all_variable_accesses): Pass correct value in the new
32776 parameter. Add a statistics counter.
32778 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32780 PR middle-end/79931
32781 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32783 2017-04-24 Richard Biener <rguenther@suse.de>
32785 PR tree-optimization/80494
32786 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32787 out for complex types.
32789 2017-04-24 Richard Biener <rguenther@suse.de>
32791 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32792 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32793 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32794 (DFS): Adjust and never fail.
32795 (sccvn_dom_walker::fail): Remove.
32796 (sccvn_dom_walker::before_dom_children): Adjust.
32797 (run_scc_vn): Likewise and never fail.
32798 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32799 (pass_fre::execute): Likewise.
32801 2017-04-24 Richard Biener <rguenther@suse.de>
32803 PR tree-optimization/79725
32804 * tree-ssa-sink.c (statement_sink_location): Return whether
32805 failure reason was zero uses. Move that check later.
32806 (sink_code_in_bb): Deal with zero uses by removing the stmt
32809 2017-04-24 Richard Biener <rguenther@suse.de>
32812 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32813 pointer-based references.
32815 2017-04-24 Richard Biener <rguenther@suse.de>
32818 * pass_manager.h (pass_manager::operator new): Remove.
32819 (pass_manager::operator delete): Likewise.
32820 * passes.c (pass_manager::operator new): Remove.
32821 (pass_manager::operator delete): Likewise.
32822 (pass_manager::pass_manager): Zero individual pass members.
32824 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32827 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32828 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32829 Check "XEXP (src, 1)" operand here.
32830 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32831 Check "XEXP (src, 1)" operand here.
32832 (dimode_scalar_chain::make_vector_copies): Detect count register
32833 of a shift instruction. Zero extend count register from QImode
32834 to DImode to satisfy vector shift pattern count operand predicate.
32835 Substitute vector shift count operand with a DImode copy.
32836 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32839 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32841 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32842 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32843 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32844 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32845 (UNSPEC_NOREX_MEM): Remove definition.
32847 2017-04-21 Richard Biener <rguenther@suse.de>
32849 PR tree-optimization/79547
32850 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32851 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32852 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32853 without any constraints.
32855 2017-04-21 Richard Biener <rguenther@suse.de>
32857 PR tree-optimization/78847
32858 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32860 2017-04-21 Richard Biener <rguenther@suse.de>
32862 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32863 (build_distinct_type_copy): Likewise.
32864 (build_variant_type_copy): Likewise.
32865 * tree.c (build_qualified_type): Pass down mem-stat info.
32866 (build_distinct_type_copy): Likewise.
32867 (build_variant_type_copy): Likewise.
32869 2017-04-21 Richard Biener <rguenther@suse.de>
32871 PR tree-optimization/80237
32872 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32874 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32875 for a simplified result.
32877 2016-04-21 Richard Biener <rguenther@suse.de>
32879 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32880 sth as strict as a simple_iv but a chrec without symbols and an
32881 operand defined in the loop we are peeling (and not some subloop).
32882 (propagate_constants_for_unrolling): Propagate all constants.
32884 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32887 * config/i386/i386.c (print_reg): Remove assert for disalowed
32888 regno values, call output_operand_lossage instead.
32890 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32893 * config/i386/constraints.md (Yc): New register constraint.
32894 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32895 Use Yc constraint for alternative 2 of operand 0. Remove
32896 preferred_for_speed attribute.
32898 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32900 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32901 lastprivate clauses in SIMT case.
32903 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32905 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32907 2017-04-20 Richard Biener <rguenther@suse.de>
32909 PR tree-optimization/57796
32910 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32911 as N scalar stores.
32912 (vect_model_load_cost): Cost gathers as N scalar loads.
32914 2017-04-20 Richard Biener <rguenther@suse.de>
32916 * ggc-page.c (ggc_allocated_p): Rename to ...
32917 (safe_lookup_page_table_entry): ... this and return the lookup
32919 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32921 2017-04-20 Richard Biener <rguenther@suse.de>
32923 PR tree-optimization/80453
32924 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32925 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32926 from the conditions.
32927 (vn_phi_eq): Pass them down.
32928 (vn_phi_lookup): Record them.
32929 (vn_phi_insert): Likewise.
32931 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32933 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32934 uninitialized variable warning to avoid buffer overrun.
32936 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32939 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32940 is suppressed for '{ 0 }' in C.
32942 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32944 * BASE-VER: Set to 8.0.0.
32946 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32948 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32949 priority .init_array and .fini_array section with SECTION_NOTYPE
32952 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32954 PR middle-end/80423
32955 * tree.h (build_array_type): Add typeless_storage default argument.
32956 * tree.c (type_cache_hasher::equal): Also compare
32957 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32958 (build_array_type): Add typeless_storage argument, set
32959 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32961 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32962 (build_array_type): Likewise. Add typeless_storage argument.
32964 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32965 Jakub Jelinek <jakub@redhat.com>
32967 PR tree-optimization/80426
32968 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32969 operation on symbolic operands, also compute the overflow for the
32970 invariant part when the operation degenerates into a negation.
32972 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32975 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32976 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32979 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32981 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32984 * config/avr/avr.c (tree.h): Include it.
32985 (cgraph.h): Include it.
32986 (avr_encode_section_info): Don't warn for uninitialized progmem
32987 variable if it's just an alias.
32989 2017-04-19 Richard Biener <rguenther@suse.de>
32992 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32993 when needed by AutoPGO.
32995 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32998 * doc/lto.texi: Remove an extra 'that'.
33000 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33002 PR rtl-optimization/80429
33003 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33004 are only used in debug insns.
33006 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33007 Vladimir Makarov <vmakarov@redhat.com>
33009 * config/sparc/predicates.md (input_operand): Add comment. Return
33010 true for any memory operand when LRA is in progress.
33011 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33013 2017-04-18 Jeff Law <law@redhat.com>
33016 * mips.md ({return,simple_return}_internal): Do not overwrite
33019 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33021 PR tree-optimization/80443
33022 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33023 instead of adding 1, subtract -1 and similarly instead of subtracting
33026 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33028 PR rtl-optimization/80357
33029 * haifa-sched.c (tmp_bitmap): New variable.
33030 (model_recompute): Handle duplicate use records.
33031 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33032 (free_global_sched_pressure_data): Free it.
33034 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33037 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33038 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33039 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33040 instead of SYSTEM_HEADER_DIR.
33042 2017-04-18 Jeff Law <law@redhat.com>
33044 PR middle-end/80422
33045 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33046 predecessors after walking up the insn chain.
33048 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33051 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33052 sizetype type into debug info.
33054 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33057 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33058 unneeded test for TARGET_UPPER_REGS_SF.
33059 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33061 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33064 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33065 instead of gsi_after_labels.
33067 2017-04-18 Jeff Law <law@redhat.com>
33069 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33073 2017-04-13 Jeff Law <law@redhat.com>
33074 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33075 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33077 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33080 * config/avr/avr.c (intl.h): Include it.
33081 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33083 2017-04-18 Martin Liska <mliska@suse.cz>
33085 PR gcov-profile/78783
33086 * gcov-tool.c (gcov_output_files): Validate that destination
33087 file is either removed by the tool or by a user.
33089 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33090 Guy Benyei <guybe@mellanox.com>
33092 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33093 block, and do not negate it, the stored id is already negative.
33095 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33097 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33099 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33102 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33103 masks of options that should be turned off if the VSX vector
33104 options are turned off.
33105 (OTHER_P8_VECTOR_MASKS): Likewise.
33106 (OTHER_VSX_VECTOR_MASKS): Likewise.
33107 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33108 rs6000_disable_incompatible_switches to validate no type switches
33110 (rs6000_incompatible_switch): New function to disallow turning on
33111 other vector options if -mno-vsx, -mno-power8-vector, or
33112 -mno-power9-vector are specified.
33114 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33116 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33118 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33120 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33121 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33122 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33123 (ARG_POINTER_CFA_OFFSET): Likewise.
33125 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33127 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33128 conditions to take advantage of various optimizations.
33130 2017-04-13 Jeff Law <law@redhat.com>
33132 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33133 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33134 (zero_extendsidi2_dext): Likewise.
33136 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33139 * fold-const.c (fold_ternary_loc): Revert
33140 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33143 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33145 PR rtl-optimization/80343
33146 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33147 new scratch pseudo.
33149 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33152 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33153 to ubsan_encode_value.
33155 2017-04-13 Jeff Law <law@redhat.com>
33157 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33158 appearing in DEBUG_INSNs.
33160 2017-04-13 Martin Liska <mliska@suse.cz>
33162 PR gcov-profile/80413
33163 * gcov-io.c (gcov_write_string): Copy to buffer just when
33164 allocated size is greater than zero.
33166 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33169 * dwarf2out.c (decls_for_scope): Ignore declarations of
33170 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33172 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33175 * ipa-visibility.c (non_local_p): Fix typos.
33176 (localize_node): When localizing symbol in same comdat group,
33177 dissolve the group only when we know external symbols are going
33179 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33181 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33183 PR tree-optimization/79390
33184 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33185 order does not result in usable sequence, retry with reversed operand
33191 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33192 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33193 op0 instead of fold_convert_loc (loc, type, arg0).
33195 2017-04-12 Jeff Law <law@redhat.com>
33197 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33198 has a delay slot in the generated code.
33200 * config/cris/cris.md (cris_preferred_reload_class): Return
33201 GENNONACR_REGS rather than GENERAL_REGS.
33203 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33206 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33207 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33208 signedness of the result type.
33210 2017-04-12 Richard Biener <rguenther@suse.de>
33211 Jeff Law <law@redhat.com>
33213 PR tree-optimization/80359
33214 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33215 trim stores to TARGET_MEM_REFs.
33217 2017-04-12 Richard Biener <rguenther@suse.de>
33219 PR tree-optimization/79390
33220 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33221 threading case even more.
33223 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33226 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33227 for quad_address_p for TImode, instead of just not indexed_address.
33229 2017-04-12 Richard Biener <rguenther@suse.de>
33230 Bernd Edlinger <bernd.edlinger@hotmail.de>
33232 PR middle-end/79671
33233 * alias.c (component_uses_parent_alias_set_from): Handle
33234 TYPE_TYPELESS_STORAGE.
33235 (get_alias_set): Likewise.
33236 * tree-core.h (tree_type_common): Add typeless_storage flag.
33237 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33238 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33239 for types containing members with TYPE_TYPELESS_STORAGE.
33240 (place_field): Likewise.
33241 (layout_type): Likewise for ARRAY_TYPE.
33242 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33243 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33244 TYPE_TYPELESS_STORAGE.
33245 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33247 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33250 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33251 first argument to type.
33253 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33257 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33258 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33259 (rs6000_expand_binop_builtin): Likewise.
33260 (rs6000_expand_ternop_builtin): Likewise; also add missing
33261 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33262 vshasigma built-ins.
33263 * doc/extend.texi: Document that vec_xxpermdi's third argument
33264 must be a constant.
33266 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33268 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33269 Use shift_const cost parameter when calculating gain of STV shifts.
33271 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33273 PR rtl-optimization/70478
33274 * lra-constraints.c (process_alt_operands): Check memory for
33275 disfavoring memory insn operand.
33277 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33279 PR middle-end/80100
33280 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33281 left shift in unsigned HOST_WIDE_INT type.
33283 PR rtl-optimization/80385
33284 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33285 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33288 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33289 if they have any depend clauses.
33291 2017-04-11 Martin Liska <mliska@suse.cz>
33294 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33295 * ipa-split.c (split_function): Create a local comdat symbol
33296 if caller is in a comdat group.
33298 2017-04-11 Martin Liska <mliska@suse.cz>
33301 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33304 2017-04-11 Martin Sebor <msebor@redhat.com>
33306 PR middle-end/80364
33307 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33308 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33310 (directive::set_width, directive::set_precision, format_character):
33312 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33315 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33318 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33319 conflict, set target->arch_name instead of target->cpu_name.
33321 2017-04-11 Richard Biener <rguenther@suse.de>
33323 PR tree-optimization/80374
33324 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33325 build_zero_cst, remove fold_convertible_p check again.
33327 2017-04-11 Martin Liska <mliska@suse.cz>
33330 * ubsan.c (instrument_object_size): Do not instrument register
33333 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33336 * config/i386/i386-builtin-types.def
33337 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33338 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33339 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33340 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33341 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33342 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33343 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33344 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33345 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33346 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33347 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33348 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33349 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33350 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33351 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33352 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33353 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33354 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33355 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33356 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33357 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33358 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33359 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33360 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33361 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33362 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33363 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33364 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33365 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33366 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33367 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33368 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33369 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33370 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33371 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33372 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33373 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33374 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33375 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33376 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33377 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33378 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33379 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33380 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33381 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33382 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33384 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33385 flag to second_arg_count, handle 4 argument function type _COUNT
33386 aliases, handle second_arg_count on second argument rather than last.
33388 2017-04-10 Jeff Law <law@redhat.com>
33390 PR tree-optimization/80374
33391 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33392 record anything if we can not convert integer_zero_node to the
33395 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33398 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33399 Enhance special handling given to the TARGET_P9_MINMAX option in
33400 relation to certain other options.
33402 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33404 PR tree-optimization/80153
33405 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33406 remove POINTER_PLUS_EXPR's base part directly, rather than through
33409 2017-04-10 Richard Biener <rguenther@suse.de>
33410 Bin Cheng <bin.cheng@arm.com>
33412 PR tree-optimization/80153
33413 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33414 the first element of pointer type aff_tree. Build result expr in
33416 (add_elt_to_tree): Convert to type unconditionally. Remove other
33417 fold_convert calls.
33418 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33419 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33421 2017-04-10 Richard Biener <rguenther@suse.de>
33423 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33426 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33428 PR rtl-optimization/70478
33429 * lra-constraints.c (curr_small_class_check): New.
33430 (update_and_check_small_class_inputs): New.
33431 (process_alt_operands): Update curr_small_class_check. Disfavor
33432 alternative insn memory operands. Check available regs for small
33435 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33438 * config/mips/mips.opt (-mvirt): Update description.
33439 * doc/invoke.texi (-mvirt): Likewise.
33441 2017-04-10 Richard Biener <rguenther@suse.de>
33443 PR middle-end/80362
33444 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33445 looking for NEGATE_EXPR in -A / -B to A / B folding.
33447 2017-04-10 Martin Liska <mliska@suse.cz>
33449 PR gcov-profile/80224
33450 * gcov.c (print_usage): Fix usage string.
33451 (get_gcov_intermediate_filename): Remove.
33452 (output_gcov_file): Use both for normal and intermediate format.
33453 (generate_results): Do not initialize special file for
33454 intermediate format.
33456 2017-04-10 Richard Biener <rguenther@suse.de>
33458 PR tree-optimization/80304
33459 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33462 2017-04-10 Nathan Sidwell <nathan@acm.org>
33465 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33466 (rs6000_init_builtins): Use it.
33468 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33470 * config/arm/arm.md (<mrc>): Add mode to SET source.
33471 (<mrrc>): Likewise.
33473 2017-04-10 Richard Biener <rguenther@suse.de>
33475 PR middle-end/80344
33476 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33478 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33481 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33482 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33483 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33484 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33485 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33486 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33487 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33488 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33489 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33490 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33491 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33492 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33493 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33494 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33495 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33496 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33497 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33498 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33499 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33500 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33501 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33502 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33503 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33505 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33507 PR rtl-optimization/70478
33508 * lra-constraints.c: Reverse the last patch.
33510 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33512 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33513 Add comment for WCHAR_T.
33515 2017-04-08 Martin Liska <mliska@suse.cz>
33518 2017-04-07 Martin Liska <mliska@suse.cz>
33521 * ipa-split.c (split_function): Add function part to a same comdat
33524 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33527 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33529 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33531 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33532 (vec_pairing): ...to this.
33533 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33534 (rs6000_sched_init): Adjust for name change.
33535 (struct rs6000_sched_context): Likewise.
33536 (rs6000_init_sched_context): Likewise.
33537 (rs6000_set_sched_context): Likewise.
33539 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33545 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33547 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33548 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33549 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33551 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33553 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33555 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33557 PR rtl-optimization/70703
33558 * ira-color.c (update_conflict_hard_regno_costs): Use
33559 int64_t instead of HOST_WIDE_INT.
33561 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33563 PR rtl-optimization/70478
33564 * lra-constraints.c (process_alt_operands): Disfavor alternative
33565 insn memory operands.
33567 2017-04-07 Jeff Law <law@redhat.com>
33569 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33570 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33572 2017-04-07 Martin Liska <mliska@suse.cz>
33575 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33576 Show error message instead of an ICE.
33578 2017-04-07 Martin Liska <mliska@suse.cz>
33581 * ipa-split.c (split_function): Add function part to a same comdat
33584 2017-04-07 Richard Biener <rguenther@suse.de>
33586 PR middle-end/80341
33587 * tree.c (get_unwidened): Also handle ! for_type case for
33589 * convert.c (do_narrow): Split out from ...
33590 (convert_to_integer_1): ... here. Do not pass final truncation
33591 type to get_unwidened for TRUNC_DIV_EXPR.
33593 2017-04-07 Richard Biener <rguenther@suse.de>
33595 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33597 (aff_combination_const): Adjust.
33598 (aff_combination_scale): Likewise.
33599 (aff_combination_add_elt): Likewise.
33600 (aff_combination_add_cst): Likewise.
33601 (aff_combination_convert): Likewise.
33602 (add_elt_to_tree): Likewise. Remove unused argument.
33603 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33605 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33607 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33609 * config/arm/arm.c (arm_default_short_enums): Use
33610 ARM_DEFAULT_SHORT_ENUMS.
33611 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33613 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33616 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33617 members with redundant out-of-class redeclaration.
33619 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33622 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33623 * config/i386/i386.md (*zero_extendsidi2):
33624 Add (?*x,*x) and (?*v,*v) alternatives.
33626 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33629 * config/i386/i386.c (ix86_expand_builtin)
33630 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33631 mode from insn data. Convert operands to insn operand mode.
33632 Copy operands that don't satisfy insn predicate to a register.
33634 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33636 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33639 2017-04-06 Richard Biener <rguenther@suse.de>
33641 PR tree-optimization/80334
33642 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33643 preserve alignment of accesses.
33645 2017-04-06 Richard Biener <rguenther@suse.de>
33647 PR tree-optimization/80262
33648 * tree-sra.c (build_ref_for_offset): Preserve address-space
33650 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33651 Drop useless address-space information on MEM_REF offsets.
33653 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33655 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33657 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33659 PR rtl-optimization/70703
33660 * ira-color.c (update_conflict_hard_regno_costs): Use
33661 HOST_WIDE_INT instead of long.
33663 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33666 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33667 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33669 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33670 for x86_64 target. Handle -m3dnowa option.
33672 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33674 PR rtl-optimization/70703
33675 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33676 (update_conflict_hard_regno_costs): Use long instead of unsigned
33677 arithmetic for cost calculation.
33679 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33680 Bernd Edlinger <bernd.edlinger@hotmail.de>
33683 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33686 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33689 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33690 ptr_mode with Pmode throughout.
33691 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33692 into probe_stack_range and use DImode.
33694 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33697 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33698 call_eh_return is true.
33700 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33702 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33703 Initialize last_match_fntype_index.
33705 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33708 * tree-nvr.c: Include internal-fn.h.
33709 (pass_return_slot::execute): Ignore internal calls without
33712 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33713 Richard Biener <rguenther@suse.de>
33716 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33717 captures used multiple times, except for the last use.
33718 * generic-match-head.c: Include gimplify.h.
33720 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33722 PR tree-optimization/79390
33723 * target.h (struct noce_if_info): Declare.
33724 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33725 * target.def (noce_conversion_profitable_p): New target hook.
33726 * ifcvt.h (struct noce_if_info): New type, moved from ...
33727 * ifcvt.c (struct noce_if_info): ... here.
33728 (noce_conversion_profitable_p): Renamed to ...
33729 (default_noce_conversion_profitable_p): ... this. No longer
33731 (noce_try_store_flag_constants, noce_try_addcc,
33732 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33733 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33734 instead of noce_conversion_profitable_p.
33735 * config/i386/i386.c: Include ifcvt.h.
33736 (ix86_option_override_internal): Don't override
33737 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33738 (ix86_noce_conversion_profitable_p): New function.
33739 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33740 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33741 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33742 * doc/tm.texi: Regenerated.
33744 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33746 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33749 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33752 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33753 instructions for small multiply cores.
33755 2017-04-04 Jeff Law <law@redhat.com>
33757 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33759 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33760 that are not set by the loop over the elements.
33762 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33765 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33766 int mode, convert_modes it to mode as unsigned, otherwise use
33767 lowpart_subreg to mode rather than SImode.
33768 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33769 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33770 Use DImode instead of SImode for the shift count operand.
33771 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33774 2017-04-04 Richard Biener <rguenther@suse.de>
33776 PR middle-end/80281
33777 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33778 arithmetic done for the negate or the plus. Simplify.
33779 (A - (-B) -> A + B): Likewise.
33780 * fold-const.c (split_tree): Make sure to not negate pointers.
33782 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33784 PR rtl-optimization/60818
33785 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33786 a compare of comparisons with the thing compared if this results
33787 in a different machine mode.
33789 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33791 * alias.c (base_alias_check): Fix typo in comment.
33792 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33793 * cgraphunit.c (symbol_table::compile): Likewise.
33794 * collect2.c (maybe_run_lto_and_relink): Likewise.
33795 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33796 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33797 * config/avr/avr.c (avr_map_op_t): Likewise.
33798 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33799 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33800 * config/epiphany/epiphany.md (movcc): Likewise.
33801 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33802 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33804 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33805 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33806 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33807 * config/sh/sh.c (sh_rtx_costs): Likewise.
33808 * fold-const.c (fold_truth_andor): Likewise.
33809 * genautomata.c (collapse_flag): Likewise.
33810 * gengtype.h (struct type::u::s): Likewise.
33811 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33812 * input.c (FORMAT_AMOUNT): Likewise.
33813 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33814 (known_aggs_to_agg_replacement_list): Likewise.
33815 * ipa-inline-analysis.c: Likewise.
33816 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33817 * ipa-polymorphic-call.c
33818 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33819 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33820 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33822 * modulo-sched.c (apply_reg_moves): Likewise.
33823 * omp-expand.c (build_omp_regions_1): Likewise.
33824 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33825 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33826 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33827 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33828 * value-prof.c: Likewise.
33829 * var-tracking.c (val_reset): Likewise.
33831 2017-04-03 Richard Biener <rguenther@suse.de>
33833 PR tree-optimization/80275
33834 * fold-const.c (split_address_to_core_and_offset): Handle
33837 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33839 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33840 descriptors is at least equal to that of functions.
33842 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33844 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33846 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33849 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33850 (mov<IMOD4:mode>): New expander.
33851 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33853 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33855 PR rtl-optimization/79405
33856 * fwprop.c (propagations_left): New variable.
33857 (forward_propagate_into): Decrement it.
33858 (fwprop_init): Initialize it.
33859 (fw_prop): If the variable has reached zero, stop propagating.
33860 (fwprop_addr): Ditto.
33862 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33865 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33866 a FUNCTION_DECL, pass it as decl instead of origin to
33869 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33871 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33874 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33877 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33878 TARGET_VSX_SMALL_INTEGER.
33880 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33882 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33883 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33885 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33887 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33888 extraction from odd-numbered MSA register.
33890 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33892 PR middle-end/80173
33893 * expmed.c (store_bit_field_1): Don't attempt to create
33894 a word subreg out of hard registers wider than word if they
33895 have HARD_REGNO_NREGS of 1 for their mode.
33897 PR middle-end/80163
33898 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33899 conversions to integer types wider than word and pointer.
33902 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33903 (rtx_equal_for_cselib_p): Pass 0 to it.
33904 * cselib.c (cselib_hasher::equal): Likewise.
33905 (rtx_equal_for_cselib_1): Add depth argument. If depth
33906 is 128, don't look up VALUE locs and punt. Increment
33907 depth in recursive calls when walking VALUE locs.
33909 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33911 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33912 (make_gcov_file_name): Use the canonical path name for generating
33914 (read_line): Fix handling of files with ascii null bytes.
33916 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33918 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33919 to initialise a vector register instead
33920 of using a const_int.
33922 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33924 PR translation/80189
33925 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33926 diagnostic messages.
33928 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33931 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33932 (dfp_diex_<mode>): Update mode of operand 1.
33933 * doc/extend.texi (dxex, dxexq): Document change to return type.
33934 (diex, diexq): Document change to argument type.
33936 2017-03-30 Martin Jambor <mjambor@suse.cz>
33939 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33940 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33941 it reflects the signature changes performed at the callee side.
33942 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33943 to cgraph_build_function_type_skip_args.
33944 (build_function_decl_skip_args): Adjust call to the above function.
33946 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33949 * config/i386/sse.md
33950 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33951 register as dest whenever it is a MEM not rtx_equal_p to the
33952 corresponding dup operand, and when forcing into reg move the
33953 reg into the memory afterwards.
33954 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33955 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33956 for the force_reg mode.
33957 (avx512vl_vextractf128<mode>): Use register as dest either
33958 always when a MEM, or when it is a MEM not rtx_equal_p to the
33959 corresponding dup operand, or even not when it is a CONST_VECTOR
33960 depending on the mode and lo vs. hi.
33961 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33963 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33964 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33965 Likewise. Require that operands[2] is even.
33966 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33967 Remove extraneous parens. Require that operands[2] is a multiple
33969 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33970 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33971 disallow memory then.
33973 2017-03-30 Richard Biener <rguenther@suse.de>
33975 PR tree-optimization/77498
33976 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33977 to non-constants over backedges.
33979 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33981 PR rtl-optimization/80233
33982 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33983 as last_combined_insn. Do not test for BARRIER_P separately.
33985 2017-03-29 Andreas Schwab <schwab@suse.de>
33988 * calls.c (prepare_call_address): Convert funexp to Pmode before
33989 copying to temp reg.
33991 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33993 PR tree-optimization/80158
33994 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33995 Handle possible future case of more than one alternate
33997 (replace_rhs_if_not_dup): Likewise.
33998 (replace_one_candidate): Likewise.
34000 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34002 PR rtl-optimization/80193
34003 * ira.c (ira): Do not check allocation for LRA.
34005 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34007 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34008 (nvptx_output_simt_exit): Declare.
34009 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34010 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34011 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34012 (nvptx_declare_function_name): ...here. Emit declaration of local
34013 memory space buffer for omp_simt_enter insn.
34014 (nvptx_output_unisimt_switch): New.
34015 (nvptx_output_softstack_switch): New.
34016 (nvptx_output_simt_enter): New.
34017 (nvptx_output_simt_exit): New.
34018 * config/nvptx/nvptx.h (struct machine_function): New fields
34019 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34020 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34021 (UNSPECV_SIMT_EXIT): Ditto.
34022 (omp_simt_enter_insn): New insn.
34023 (omp_simt_enter): New expansion.
34024 (omp_simt_exit): New insn.
34025 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34027 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34028 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34029 (expand_GOMP_SIMT_EXIT): New.
34030 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34031 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34032 (GOMP_SIMT_EXIT): Ditto.
34033 * target-insns.def (omp_simt_enter): New insn.
34034 (omp_simt_exit): Ditto.
34035 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34037 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34038 (lower_rec_input_clauses): Likewise.
34039 (lower_lastprivate_clauses): Handle SIMT privatization.
34041 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34042 (ompdevlow_adjust_simt_enter): New.
34043 (find_simtpriv_var_op): New.
34044 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34045 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34047 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34048 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34049 (copy_decl_for_dup_finish): Ditto.
34051 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34053 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34056 * config/i386/i386.c (ix86_option_override_internal): Always
34057 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34059 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34061 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34063 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34065 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34066 mark new edge's irreducible flag accordign to it.
34067 (vect_do_peeling): Check loop preheader edge's irreducible flag
34068 and pass it to function slpeel_add_loop_guard.
34070 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34072 PR tree-optimization/80218
34073 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34074 Update block frequencies and counts.
34076 2017-03-28 Richard Biener <rguenther@suse.de>
34078 PR tree-optimization/78644
34079 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34080 of a simplification result we may not use it at all.
34082 2017-03-28 Richard Biener <rguenther@suse.de>
34085 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34086 without arguments, generate default definition of a SSA name.
34088 2017-03-28 Richard Biener <rguenther@suse.de>
34090 PR middle-end/80222
34091 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34092 TYPE_REF_CAN_ALIAS_ALL references.
34093 * fold-const.c (fold_indirect_ref_1): Likewise.
34095 2017-03-28 Martin Liska <mliska@suse.cz>
34098 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34099 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34101 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34102 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34104 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34105 (EXTRA_SPECS): Define.
34106 (SUBTARGET_EXTRA_SPECS): Likewise.
34107 (SUBTARGET_CPP_SPEC): Likewise.
34108 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34109 SUBTARGET_EXTRA_SPECS.
34110 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34112 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34114 * config/arc/simdext.md (vst64_insn): Update pattern.
34115 (vld32wh_insn): Likewise.
34116 (vld32wl_insn): Likewise.
34117 (vld64_insn): Likewise.
34118 (vld32_insn): Likewise.
34120 2017-03-28 Marek Polacek <polacek@redhat.com>
34123 * fold-const.c (fold_comparison): Use protected_set_expr_location
34124 instead of SET_EXPR_LOCATION.
34126 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34128 * tree.c (add_expr): Avoid name lookup warning.
34130 2017-03-27 Jeff Law <law@redhat.com>
34132 PR tree-optimization/80216
34133 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34134 function name. Limit recursion depth.
34135 (record_temporary_equivalences): Corresponding changes.
34137 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34139 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34142 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34145 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34147 * cfgcleanup.c (reg_note_cfa_p): New array.
34148 (insns_have_identical_cfa_notes): New function.
34149 (old_insns_match_p): Don't cross-jump in between /f
34150 and non-/f instructions. If both i1 and i2 are frame related,
34151 verify all CFA notes, their order and content.
34153 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34156 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34157 HImode and SImode with zero extend to DImode to one insn.
34158 (bswap<mode>2_extenddi): Likewise.
34159 (bswapsi2_extenddi): Likewise.
34160 (bswaphi2_extendsi): Likewise.
34161 (bswaphi2): Combine bswap HImode and SImode into one insn.
34162 Separate memory insns from swapping register.
34163 (bswapsi2): Likewise.
34164 (bswap<mode>2): Likewise.
34165 (bswaphi2_internal): Delete, no longer used.
34166 (bswapsi2_internal): Likewise.
34167 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34168 store, and gpr<-gpr swap insns.
34169 (bswap<mode>2_store): Likewise.
34170 (bswaphi2_reg): Register only splitter, combine with the splitter.
34171 (bswaphi2 splitter): Likewise.
34172 (bswapsi2_reg): Likewise.
34173 (bswapsi2 splitter): Likewise.
34174 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34175 the insns into load, store, and register/register insns.
34176 (bswapdi2_ldbrx): Likewise.
34177 (bswapdi2_load): Likewise.
34178 (bswapdi2_store): Likewise.
34179 (bswapdi2_reg): Likewise.
34181 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34183 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34184 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34186 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34189 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34191 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34192 special handling for target option conflicts between dform
34193 options (-mpower9-dform, -mpower9-dform-vector,
34194 -mpower9-dform-scalar) and -mno-direct-move.
34196 2017-03-27 Richard Biener <rguenther@suse.de>
34198 PR tree-optimization/80181
34199 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34201 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34203 * config/arc/predicates.md (move_double_src_operand): Replace the
34204 call to move_double_src_operand with a call to address_operand.
34206 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34208 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34209 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34210 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34212 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34214 * config/arc/predicates.md (long_immediate_loadstore_operand):
34215 Consider scaled addresses cases.
34217 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34219 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34220 restored when in interrupt.
34221 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34222 doesn't have delay slot.
34224 2017-03-27 Richard Biener <rguenther@suse.de>
34227 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34228 inlined thunk clones.
34230 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34233 * asan.c (instrument_derefs): Copy over last operand from
34234 original COMPONENT_REF to the new COMPONENT_REF with
34235 DECL_BIT_FIELD_REPRESENTATIVE.
34236 * ubsan.c (instrument_object_size): Likewise.
34238 2017-03-27 Richard Biener <rguenther@suse.de>
34240 PR tree-optimization/80170
34241 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34242 sure DR/SCEV didnt fold in constants we do not see when looking
34243 at the reference base alignment.
34245 2017-03-27 Richard Biener <rguenther@suse.de>
34247 PR middle-end/80171
34248 * gimple-fold.c (fold_ctor_reference): Properly guard against
34249 NULL return value from canonicalize_constructor_val.
34251 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34254 * config/i386/i386.c (ix86_expand_builtin)
34255 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34256 flags reg setting and flags reg using instructions.
34257 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34258 clobbering instructions to zero extend op2.
34260 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34262 * doc/install.texi (Configuration) <--with-aix-soname>:
34263 Update link to AIX ld.
34265 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34267 PR rtl-optimization/80160
34268 PR rtl-optimization/80159
34269 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34270 reg_alternate_class into account.
34272 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34275 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34276 to consider in curr_insn_transform.
34278 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34280 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34281 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34282 and emit_mode_inner.
34284 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34286 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34287 argument to the overloaded builtin variants. Use the new flag to
34288 deprecate certain builtin variants.
34289 * config/s390/s390-builtin-types.def: Add new builtin types.
34290 * config/s390/s390-builtins.h: Support new flags field for
34291 overloaded builtins.
34292 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34293 (s390_macro_to_expand): Enable vector float data type.
34294 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34295 builtins by incrementing the __VEC__ version number.
34296 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34298 (s390_resolve_overloaded_builtin): Emit error messages depending
34299 on the builtin flags.
34300 * config/s390/s390.c (s390_expand_builtin): Support additional
34301 flags argument. Change error message to match the messages
34302 emitted in s390-c.c.
34303 * config/s390/s390.md: New UNSPEC_* constants.
34304 (op_type): Add new instruction types.
34305 * config/s390/vecintrin.h: Add new builtins and test data class
34307 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34308 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34309 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34310 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34311 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34312 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34314 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34315 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34316 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34317 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34319 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34320 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34321 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34322 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34324 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34325 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34326 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34327 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34328 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34329 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34330 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34332 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34333 ("vec_scatter_element<V_HW_4:mode>_DI")
34334 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34335 ("vec_fpint<mode>", "vflls")
34336 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34337 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34338 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34339 ("*vec_cmphe<mode>_cc"): ... these.
34341 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34342 mode constant instead of magic value.
34344 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34346 * config/s390/s390.c (s390_expand_vec_compare): Support other
34347 vector floating point modes than just V2DF.
34348 (s390_expand_vcond): Likewise.
34349 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34350 (s390_cannot_change_mode_class): Prevent mode changes between TF
34351 and V1TF in vector registers.
34352 * config/s390/s390.md (DF, SF): New mode attributes.
34353 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34354 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34355 SFmode support for VRs.
34356 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34358 (VFT, VF_HW): New mode iterators.
34359 (vw, sdx): New mode attributes.
34360 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34361 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34362 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34363 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34364 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34365 also the new vector floating point modes. Renaming to ...
34367 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34368 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34369 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34370 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34371 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34372 ("vec_unordered<mode>"): ... these.
34374 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34375 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34376 ("*vec_extendv2df"): New insn definitions.
34378 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34380 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34381 ("mulditi3_2", "*muldi3_sign"): New patterns.
34382 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34383 rename the pattern definition.
34385 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34387 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34389 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34391 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34393 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34394 instruction if possible.
34395 * config/s390/vector.md (vec_halfnumelts): New mode
34397 ("*vec_vllezlf<mode>"): New pattern.
34399 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34401 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34402 ("popcountv4si2", "popcountv2di2"): Rename to ...
34403 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34404 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34406 ("popcount<mode>2_vxe"): New pattern.
34408 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34410 * common/config/s390/s390-common.c (processor_flags_table): Add
34412 * config.gcc: Add arch12.
34413 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34414 Default to arch12 for unknown CPU model numbers.
34415 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34416 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34417 PROCESSOR_max sanity check.
34418 * config/s390/s390-opts.h (enum processor_type): Add
34420 * config/s390/s390.c (processor_table): Add arch12.
34421 (s390_expand_builtin): Add check for B_VXE flag.
34422 (s390_issue_rate): Add PROCESSOR_ARCH12.
34423 (s390_get_sched_attrmask): Likewise.
34424 (s390_get_unit_mask): Likewise.
34425 (s390_sched_score): Enable z13 scheduling for arch12.
34426 (s390_sched_reorder): Likewise.
34427 (s390_sched_variable_issue): Likewise.
34428 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34430 (s390_tune_attr): Use z13 scheduling also for arch12.
34431 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34432 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34433 (TARGET_VXE_P): New macros.
34434 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34435 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34436 * config/s390/s390.opt: Add arch12 as processor_type.
34438 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34440 * config/s390/s390.md
34441 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34442 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34443 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34445 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34446 Rename expanders to ...
34448 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34449 ("fixuns_truncdddi2_emu"): ... these.
34451 ("fixuns_trunc<mode>si2_emu"): New expander.
34453 ("*fixuns_truncdfdi2_z13"): Rename to ...
34454 ("*fixuns_truncdfdi2_vx"): ... this.
34456 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34458 * config/s390/2964.md: Remove the single element vector compare
34459 instructions which are no longer used.
34460 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34462 (s390_canonicalize_comparison): Remove handling of DFmode
34464 (s390_expand_vec_compare_scalar): Remove function.
34465 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34466 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34468 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34470 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34472 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34474 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34475 will anyway by matched by mov<mode>_64dfp.
34477 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34479 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34480 vlef/vstef. Add missing operand to vleif.
34482 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34484 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34485 pair for all vector types with 64 bit elements.
34486 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34487 * config/s390/vector.md (V_HW_64): ... here.
34488 (V_128_NOSINGLE): New mode iterator.
34489 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34490 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34491 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34492 ("*vec_load_pairv2di"): Change to ...
34493 ("*vec_load_pair<mode>"): ... this one.
34495 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34497 * config/s390/constraints.md: Add comments.
34498 (jKK): Reject element sizes > 8 bytes.
34499 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34501 * config/s390/s390.md: Add the s_operand checks formerly in
34502 s390_split_ok_p to various splitters where they are still
34504 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34505 for 128 bit vectors. Plus two splitters.
34507 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34509 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34512 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34515 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34516 error if the boundary argument is not constant.
34518 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34520 PR rtl-optimization/80112
34521 * loop-doloop.c (doloop_condition_get): Don't check condition
34522 if cmp isn't SET with IF_THEN_ELSE src.
34524 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34526 PR tree-optimization/80158
34527 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34528 replacing a candidate statement, also replace it for the
34529 candidate's alternate interpretation.
34530 (replace_rhs_if_not_dup): Likewise.
34531 (replace_one_candidate): Likewise.
34533 2017-03-24 Richard Biener <rguenther@suse.de>
34535 PR tree-optimization/80167
34536 * graphite-isl-ast-to-gimple.c
34537 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34539 (translate_isl_ast_to_gimple::get_rename): Likewise.
34541 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34543 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34544 handling of certain combinations of target options, including the
34545 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34546 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34548 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34551 * config/arm/arm.md (*load_multiple): Add reload_completed to
34552 matching condition.
34554 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34555 Richard Biener <rguenth@suse.de>
34557 PR tree-optimization/79908
34558 PR tree-optimization/80136
34559 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34560 been cast away, gimplify_and_add suffices.
34562 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34564 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34566 2017-03-23 Richard Biener <rguenther@suse.de>
34568 PR tree-optimization/80032
34569 * gimplify.c (gimple_push_cleanup): Forced unconditional
34570 cleanups still have to go to the conditional_cleanups
34573 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34575 PR tree-optimization/80072
34576 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34578 (next_operand_entry_id): Change type to unsigned int.
34579 (sort_by_operand_rank): Make sure to return the right return value
34580 even if unsigned fields are bigger than INT_MAX.
34581 (struct oecount): Change cnt and id type to unsigned int.
34582 (oecount_hasher::equal): Formatting fix.
34583 (oecount_cmp): Make sure to return the right return value
34584 even if unsigned fields are bigger than INT_MAX.
34585 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34588 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34589 TREE_READONLY on result if writing it more than once.
34592 * doc/invoke.texi (-fsanitize=thread): Document that with
34593 -fnon-call-exceptions atomics are not able to throw
34597 * tsan.c: Include tree-eh.h.
34598 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34599 maybe_clean_or_replace_eh_stmt where needed.
34600 (instrument_memory_accesses): Add cfg_changed argument.
34601 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34602 if it returned true.
34603 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34605 PR rtl-optimization/63191
34606 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34607 wrapper function, moved the whole old content into ...
34608 (ix86_delegitimize_address_1): ... this. New inline function.
34609 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34610 true as last argument instead of ix86_delegitimize_address.
34612 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34614 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34615 cortexa57_branch_cost.
34617 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34619 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34621 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34624 * doc/md.texi (Constraints): Document wA constraint.
34625 * config/rs6000/constraints.md (wA): New.
34626 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34627 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34628 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34629 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34631 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34634 * gimplify.c (is_oacc_declared): New function.
34635 (oacc_default_clause): Use it to set default flags for acc declared
34636 variables inside parallel regions.
34637 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34638 declared variables.
34639 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34640 declare attribute to any decl as necessary.
34642 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34645 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34646 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34647 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34648 (arm_arch_lpae): This.
34649 * config/arm/arm.c (arm_arch7ve): Rename into ...
34650 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34651 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34654 2017-03-22 Martin Liska <mliska@suse.cz>
34657 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34658 error message instead of an ICE.
34660 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34662 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34664 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34666 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34668 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34671 2017-03-21 Martin Sebor <msebor@redhat.com>
34673 * doc/extend.texi: Use "cannot" instead of "can't."
34674 * doc/hostconfig.texi: Same.
34675 * doc/install.texi: Same.
34676 * doc/invoke.texi: Same.
34677 * doc/loop.texi: Same.
34678 * doc/md.texi: Same.
34679 * doc/objc.texi: Same.
34680 * doc/rtl.texi: Same.
34681 * doc/tm.texi: Same.
34682 * doc/tm.texi.in: Same.
34683 * doc/trouble.texi: Same.
34685 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34688 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34689 (collect_checksum_attributes): Set it.
34690 (die_checksum_ordered): Use it.
34692 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34694 PR tree-optimization/79908
34695 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34696 change: For a VA_ARG whose LHS has been cast away, use
34697 force_gimple_operand to construct the side effects.
34699 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34701 PR translation/80001
34702 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34703 more amenable to translation.
34704 (oacc_loop_auto_partitions): Likewise.
34706 2017-03-21 Marek Polacek <polacek@redhat.com>
34707 Martin Sebor <msebor@redhat.com>
34709 PR tree-optimization/80109
34710 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34711 on INTEGRAL_TYPE_P.
34713 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34714 Segher Boessenkool <segher@kernel.crashing.org>
34717 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34718 check reg_used_between_p between insn and one of succ or succ2
34719 depending on if succ is artificial insn not inserted into insn
34722 2017-03-21 Martin Liska <mliska@suse.cz>
34724 PR gcov-profile/80081
34725 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34726 * doc/gcc.texi: Include gcov-dump stuff.
34727 * doc/gcov-dump.texi: New file.
34729 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34731 PR rtl-optimization/79150
34732 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34733 conditional jump, if the jump is the last insn of the loop.
34735 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34736 Richard Biener <rguenth@suse.de>
34738 PR tree-optimization/79908
34739 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34740 been cast away, use force_gimple_operand to construct the side
34743 2017-03-21 Martin Liska <mliska@suse.cz>
34745 PR libfortran/79956
34746 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34749 2017-03-21 Brad Spengler <spender@grsecurity.net>
34752 * plugin.c (htab_hash_plugin): New function.
34753 (add_new_plugin): Use it and adjust.
34754 (parse_plugin_arg_opt): Adjust.
34755 (init_one_plugin): Likewise.
34757 2017-03-21 Richard Biener <rguenther@suse.de>
34759 PR tree-optimization/80032
34760 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34761 if set force the cleanup to happen unconditionally.
34762 (gimplify_target_expr): Push inserted clobbers with force_uncond
34763 to avoid them being removed by control-dependent DCE.
34765 2017-03-21 Richard Biener <rguenther@suse.de>
34767 PR tree-optimization/80122
34768 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34769 va_arg_pack_len when the inlined call stmt requires pack
34771 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34773 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34776 * tsan.c (instrument_builtin_call): If the memory model argument
34777 is not a constant, assume it is valid.
34780 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34783 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34785 PR rtl-optimization/79910
34786 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34787 if its dest is used by an insn before I2 (other than the combined
34788 insns themselves, which are properly handled already).
34790 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34793 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34795 * combine.c (record_used_regs): New static function.
34796 (try_combine): Handle situations where there is an additional
34797 instruction between I2 and I3 which needs to have a LOG_LINK
34801 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34803 * combine.c (try_combine): Delete redundant i1 test. Call
34804 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34806 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34809 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34810 alternatives 13/14.
34812 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34814 PR tree-optimization/80054
34815 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34816 the optimization if a PHI or any of its arguments is not dominated
34817 by the candidate's basis. Use gphi* rather than gimple* as
34819 (replace_profitable_candidates): Clean up a gimple* variable that
34820 should be a gphi* variable.
34822 2017-03-20 Martin Sebor <msebor@redhat.com>
34825 * doc/extend.texi (attribute constructor): Document present limitation.
34827 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34830 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34831 __POWER9_VECTOR__ #ifdef control, change template definition to
34832 use Power9-specific built-in function.
34833 (vec_any_eq): Likewise.
34834 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34835 to control outcomes from this test.
34836 (vector_ae_<mode>p): For VEC_F modes, likewise.
34838 2017-03-20 Ian Lance Taylor <iant@google.com>
34840 * config/i386/i386.c (ix86_function_regparm): Save an extra
34841 register for -fsplit-stack with DECL_STATIC_CHAIN.
34843 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34846 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34847 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34849 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34851 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34853 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34856 2017-03-20 Marek Polacek <polacek@redhat.com>
34859 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34861 2017-03-20 Richard Biener <rguenther@suse.de>
34863 PR tree-optimization/80113
34864 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34865 allocate extra SSA name for PHI def.
34866 (add_close_phis_to_outer_loops): Likewise.
34867 (add_close_phis_to_merge_points): Likewise.
34868 (copy_loop_close_phi_args): Likewise.
34869 (copy_cond_phi_nodes): Likewise.
34871 2017-03-20 Martin Liska <mliska@suse.cz>
34873 PR middle-end/79753
34874 * tree-chkp.c (chkp_build_returned_bound): Do not build
34875 returned bounds for a LHS that's not a BOUNDED_P type.
34877 2017-03-20 Martin Liska <mliska@suse.cz>
34881 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34882 COMPLEX_CST and VECTOR_CST.
34884 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34887 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34888 target operand. A new splitter adds the clobber statement in case
34889 the target operand is dead anyway.
34891 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34893 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34894 to age-old versions of binutils and glibc.
34896 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34898 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34900 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34902 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34904 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34906 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34907 requirement for binutils 2.13.
34909 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34911 * combine.c (try_combine): Delete redundant i1 test. Call
34912 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34914 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34916 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34917 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34919 <riscv64-*-elf>: Re-arrange section
34920 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34921 <riscv32-*-linux>: Likewise.
34922 <riscv64-*-elf>: Likewise
34923 <riscv64-*-linux>: Likewise.
34925 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34928 * aarch64.opt(verbose-cost-dump): Fix typo.
34930 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34933 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34934 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34936 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34938 * reload.c (find_reloads): When reloading a nonoffsettable address,
34939 use RELOAD_OTHER for it and its address reloads.
34941 PR rtl-optimization/79910
34942 * combine.c (record_used_regs): New static function.
34943 (try_combine): Handle situations where there is an additional
34944 instruction between I2 and I3 which needs to have a LOG_LINK
34947 2017-03-17 Jeff Law <law@redhat.com>
34949 PR tree-optimization/71437
34950 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34951 conditional in the hash table first.
34952 (vrp_dom_walker::before_dom_children): Extract condition from
34953 ASSERT_EXPR. Record condition, its inverion and any implied
34954 conditions as well.
34956 2017-03-17 Marek Polacek <polacek@redhat.com>
34957 Markus Trippelsdorf <markus@trippelsdorf.de>
34959 PR tree-optimization/80079
34960 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34963 2017-03-17 Richard Biener <rguenther@suse.de>
34965 PR middle-end/80075
34966 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34967 Properly verify the LHS before the RHS possibly claims to be
34969 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34972 2017-03-17 Martin Jambor <mjambor@suse.cz>
34974 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34975 (List of -O2 options): Likewise.
34976 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34979 2017-03-17 Tom de Vries <tom@codesourcery.com>
34981 * gcov-dump.c (print_usage): Print bug_report_url.
34983 2017-03-17 Richard Biener <rguenther@suse.de>
34985 PR middle-end/80050
34986 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34987 (parser::peek): Likewise.
34989 2017-03-17 Richard Biener <rguenther@suse.de>
34991 PR tree-optimization/80048
34992 * sese.c (free_sese_info): Properly release rename_map and
34993 copied_bb_map elements.
34995 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34997 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34998 Add linked-list forward and backlinks. Insert on
34999 construction, remove on destruction.
35000 (class pass_store_merging): Add m_stores_head field.
35001 (pass_store_merging::terminate_and_process_all_chains):
35002 Iterate over m_stores_head list.
35003 (pass_store_merging::terminate_all_aliasing_chains):
35005 (pass_store_merging::execute): Check for debug stmts first.
35006 Push new chains onto the m_stores_head stack.
35008 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35011 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35012 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35013 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35015 2017-03-16 Jeff Law <law@redhat.com>
35017 PR tree-optimization/71437
35018 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35019 member function. Implementation moved into after_dom_children
35020 member function and into the threader's thread_outgoing_edges
35022 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35023 some code into new thread_outgoing_edges.
35024 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35025 definition. Simplify marker handling (do it here). Assume we always
35026 have the available expression and the const/copies tables.
35027 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35029 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35030 * tree-vrp.c (equiv_stack): No longer file scoped.
35031 (vrp_dom_walker): New class.
35032 (vrp_dom_walker::before_dom_children): New member function.
35033 (vrp_dom_walker::after_dom_children): Likewise.
35034 (identify_jump_threads): Setup domwalker. Use it rather than
35035 walking edges in a random order by hand. Simplify setup/finalization.
35036 (finalize_jump_threads): Remove.
35037 (vrp_finalize): Do not call identify_jump_threads here.
35038 (execute_vrp): Do it here instead and call thread_through_all_blocks
35041 PR tree-optimization/71437
35042 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35044 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35046 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35047 (dom_opt_dom_walker::thread_across_edge): Remove
35048 handle_dominating_asserts argument. All callers changed.
35049 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35050 changes. Remove calls to lhs_of_dominating_assert. Other
35051 uses of handle_dominating_asserts turn into unconditional code
35052 (simplify_control_stmt_condition_1): Likewise.
35053 (simplify_control_stmt_condition): Likewise.
35054 (thread_through_normal_block, thread_across_edge): Likewise.
35055 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35056 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35057 object if it is not an SSA_NAME.
35058 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35059 before calling into the VRP specific simplifiers.
35060 (identify_jump_threads): Remove handle_dominating_asserts
35063 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35066 * tree-diagnostic.c (default_tree_printer): No longer static.
35067 * tree-diagnostic.h (default_tree_printer): New prototype.
35069 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35071 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35072 Change ins into fmov.
35074 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35076 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35077 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35078 Use h_con constraint for operand 1.
35079 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35080 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35082 2017-03-15 Jeff Law <law@redhat.com>
35084 PR tree-optimization/71437
35085 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35086 (record_temporary_equivalences): Use it.
35088 PR tree-optimization/71437
35089 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35090 tree-ssa-scopedtables.
35091 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35092 (record_conditions, record_cond, vuse_eq): Likewise.
35093 (record_edge_info): Adjust to API tweak of record_conditions.
35094 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35095 (record_temporary_equivalences, optimize_stmt): Likewise.
35096 (eliminate_redundant_computations): Likewise.
35097 (record_equivalences_from_stmt): Likewise.
35098 * tree-ssa-scopedtables.c: Include options.h and params.h.
35099 (vuse_eq): New function, moved from tree-ssa-dom.c
35100 (build_and_record_new_cond): Likewise.
35101 (record_conditions): Likewise. Accept vector of conditions rather
35102 than edge_equivalence structure for first argument.
35103 for the first argument.
35104 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35105 from tree-ssa-dom.c.
35106 (avail_exprs_stack::record_cond): Likewise.
35107 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35108 from tree-ssa-dom.c.
35109 (avail_exprs_stack): Add new member functions lookup_avail_expr
35111 (record_conditions): Declare.
35113 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35116 * lra-constraints.c (process_alt_operands): Increase reject for
35117 reloading an input/output operand.
35119 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35122 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35123 insns to convert from signed/unsigned char/short to IEEE 128-bit
35125 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35127 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35130 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35131 subreg of inner mode for values already in registers.
35133 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35135 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35136 iteration reg is used after the loop.
35138 2017-03-14 Martin Sebor <msebor@redhat.com>
35140 PR tree-optimization/79800
35141 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35142 precision in negative-positive range.
35143 (format_floating): Call non-const overload with adjusted precision.
35145 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35148 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35151 2017-03-14 Martin Sebor <msebor@redhat.com>
35153 PR middle-end/80020
35154 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35155 * builtins.def (aligned_alloc): Use it.
35158 * Makefile.in (GTFILES): Add calls.c.
35159 * calls.c: Include "gt-calls.h".
35161 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35163 PR rtl-optimization/79728
35164 * regs.h (struct target_regs): New field
35165 x_contains_allocatable_regs_of_mode.
35166 (contains_allocatable_regs_of_mode): New macro.
35167 * reginfo.c (init_reg_sets_1): Initialize it, and change
35168 contains_reg_of_mode so it includes global regs as well.
35169 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35170 rather than contains_regs_of_mode.
35172 2017-03-14 Martin Liska <mliska@suse.cz>
35174 * doc/invoke.texi: Document options that can't be combined with
35175 -fcheck-pointer-bounds.
35177 2017-03-14 Martin Liska <mliska@suse.cz>
35179 PR middle-end/79831
35180 * doc/invoke.texi (-Wchkp): Document the option.
35182 2017-03-14 Martin Liska <mliska@suse.cz>
35184 * Makefile.in: Install gcov-dump.
35186 2017-03-14 Martin Liska <mliska@suse.cz>
35188 * multiple_target.c (expand_target_clones): Bail out for
35189 an invalid attribute.
35191 2017-03-14 Richard Biener <rguenther@suse.de>
35193 * alias.c (struct alias_set_entry): Pack properly.
35194 * cfgloop.h (struct loop): Likewise.
35195 * cse.c (struct set): Likewise.
35196 * ipa-utils.c (struct searchc_env): Likewise.
35197 * loop-invariant.c (struct invariant): Likewise.
35198 * lra-remat.c (struct cand): Likewise.
35199 * recog.c (struct change_t): Likewise.
35200 * rtl.h (struct address_info): Likewise.
35201 * symbol-summary.h (function_summary): Likewise.
35202 * tree-loop-distribution.c (struct partition): Likewise.
35203 * tree-object-size.c (struct object_size_info): Likewise.
35204 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35205 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35206 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35207 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35208 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35209 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35210 (struct _stmt_vec_info): Likewise.
35212 2017-03-14 Martin Liska <mliska@suse.cz>
35215 * multiple_target.c (create_dispatcher_calls): Check that
35216 a target can create a function dispatcher.
35218 2017-03-14 Martin Liska <mliska@suse.cz>
35221 * multiple_target.c (expand_target_clones): Drop local.local
35222 flag for default implementation.
35224 2017-03-14 Richard Biener <rguenther@suse.de>
35226 PR tree-optimization/80030
35227 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35229 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35231 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35232 gcc_fallthrough() instead of __attribute__((fallthrough));
35234 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35236 * doc/gcc.texi: Remove "up" link to (DIR).
35237 * doc/gccint.texi: Ditto.
35239 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35241 * doc/install.texi (Specific) <avr>: Remove reference to
35244 2017-03-13 Jeff Law <law@redhat.com>
35246 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35247 attribute rather than comments.
35249 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35250 match_scratch operand is highest.
35252 2017-03-13 Martin Liska <mliska@suse.cz>
35254 PR middle-end/78339
35255 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35256 is a CHKP clone, use original declaration.
35258 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35260 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35261 (arc_conditional_register_usage): Use a different allocation order
35262 when optimizing for size.
35263 * common/config/arc/arc-common.c (arc_option_optimization_table):
35264 Section anchors default on when optimizing for size.
35266 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35268 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35270 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35272 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35273 * config/arc/arc.md (cpu_facility): Add cd variant.
35274 (*movqi_insn): Add code density variant.
35275 (*movhi_insn): Likewise.
35276 (*movqi_insn): Likewise.
35277 (*addsi3_mixed): Likewise.
35278 (subsi3_insn): Likewise.
35280 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35282 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35284 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35286 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35287 expressions with MINUS and UNARY ops.
35289 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35292 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35294 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35295 between vec_select and vector argument.
35296 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35297 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35298 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35299 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35300 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35301 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35303 2017-03-13 Richard Biener <rguenther@suse.de>
35306 * params.def (vect-max-peeling-for-alignment): Fix typo.
35308 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35310 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35311 issue that only occurred with binutils below 2.18.
35313 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35315 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35316 refer to binutils 2.11/2.12 minimum.
35318 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35320 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35321 ftp.kernel.org and simplify binutils requirement.
35323 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35325 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35327 (Optimize Options): Ditto. Also remove redundancy.
35329 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35331 PR translation/79848
35332 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35334 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35335 to G_ to avoid double translation.
35337 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35339 PR translation/79923
35340 * auto-profile.c (get_combined_location): Convert leading
35341 character of diagnostics to lower case and remove trailing period.
35342 (read_profile): Likewise for various diagnostics.
35343 * config/arm/arm.c (arm_option_override): Remove trailing period
35344 from various diagnostics.
35345 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35346 (msp430_expand_delay_cycles): Likewise.
35348 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35351 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35352 full command-line argument, rather than just "str".
35353 (aarch64_validate_march): Likewise.
35354 (aarch64_validate_mtune): Likewise.
35356 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35358 PR rtl-optimization/78911
35359 * lra-assigns.c (must_not_spill_p): New function.
35360 (spill_for): Use it.
35362 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35364 PR tree-optimization/79981
35365 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35366 ATOMIC_COMPARE_EXCHANGE ifn result.
35367 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35368 IFN_ATOMIC_COMPARE_EXCHANGE.
35370 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35373 * opts.c (parse_sanitizer_options): Add missing question mark to
35374 "did you mean" message.
35376 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35378 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35380 (VMULEUH_UNS): Likewise.
35381 (VMULOUB_UNS): Likewise.
35382 (VMULOUH_UNS): Likewise.
35383 * config/rs6000/rs6000.c (builtin_function_type): Remove
35384 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35386 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35389 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35390 x with result of extra_parsing_for_operand_code_0.
35391 (function_reader::extra_parsing_for_operand_code_0): Convert
35392 return type from void to rtx, returning x. When reading
35393 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35394 larger size containing struct block_symbol.
35396 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35398 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35399 -mfloat128-hardware without -m64.
35401 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35404 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35405 entries to the case statement that marks unsigned arguments to
35406 overloaded functions.
35408 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35410 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35411 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35413 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35416 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35417 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35419 2017-03-10 Martin Liska <mliska@suse.cz>
35423 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35424 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35427 2017-03-10 Olivier Hainque <hainque@adacore.com>
35429 * tree-switch-conversion (array_value_type): Start by resetting
35430 candidate type to it's main variant.
35432 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35434 PR rtl-optimization/79909
35435 * combine.c (try_combine): Use simplify_replace_rtx on individual
35436 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35437 of the whole CALL_INSN_FUNCTION_USAGE.
35439 PR tree-optimization/79972
35440 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35441 get_range_info on SSA_NAMEs. Formatting fixes.
35443 2017-03-10 Richard Biener <rguenther@suse.de>
35444 Jakub Jelinek <jakub@redhat.com>
35446 PR tree-optimization/77975
35447 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35448 edge to be constant.
35449 (get_val_for): For constant x return it. Formatting fix.
35450 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35451 would use the same bases as the current one.
35453 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35455 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35456 instead of vec_select for V1TImode.
35457 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35459 (VSX_LE_128): Add V1TI to this mode iterator.
35460 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35461 (*vsx_le_perm_store_<mode>): Likewise.
35462 (pre-reload splitter for VSX stores): Likewise.
35463 (post-reload splitter for VSX stores): Likewise.
35464 (*vsx_xxpermdi2_le_<mode>): Likewise.
35465 (*vsx_lxvd2x2_le_<mode>): Likewise.
35466 (*vsx_stxvd2x2_le_<mode>): Likewise.
35468 2017-03-09 Michael Eager <eager@eagercon.com>
35470 Correct failures with --enable-checking=yes,rtl.
35472 * config/microblaze/microblaze.c (microblaze_expand_shift):
35473 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35474 test for const0_rtx.
35475 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35476 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35478 2017-03-09 Richard Biener <rguenther@suse.de>
35480 PR tree-optimization/79977
35481 * graphite-scop-detection.c (scop_detection::merge_sese):
35482 Handle the case of extra exits to blocks dominating the entry.
35484 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35486 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35489 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35491 PR rtl-optimization/79949
35492 * lra-constraints.c (process_alt_operands): Check memory when
35493 trying to predict a cycle. Print about the overall increase.
35495 2017-03-09 Richard Biener <rguenther@suse.de>
35497 PR middle-end/79971
35498 * gimple-expr.c (useless_type_conversion_p): Preserve
35499 TYPE_SATURATING for fixed-point types.
35501 2017-03-09 Richard Biener <rguenther@suse.de>
35504 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35505 alignment of BLKmode params.
35507 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35510 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35511 (VALL_NO_V2Q): Likewise.
35513 * config/aarch64/aarch64-simd.md
35514 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35516 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35517 VALL_NO_V2Q mode iterator.
35518 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35520 2017-03-09 Martin Liska <mliska@suse.cz>
35522 PR tree-optimization/79631
35523 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35524 tree_int_cst_sign_bit just for INTEGER constants.
35526 2017-03-09 Martin Liska <mliska@suse.cz>
35530 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35533 2017-03-09 Marek Polacek <polacek@redhat.com>
35536 * tree.c (inchash::add_expr): Handle TREE_VEC.
35538 2017-03-09 Martin Liska <mliska@suse.cz>
35541 (chkp_narrow_size_and_offset): New function.
35542 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35543 (void chkp_parse_bit_field_ref): New function.
35544 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35545 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35547 2017-03-09 Martin Liska <mliska@suse.cz>
35550 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35551 (chkp_find_bounds_1): Remove gcc_unreachable.
35553 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35556 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35557 BUILT_IN_SYNC*, determine the access type from the size suffix and
35558 always build a MEM_REF with that type. Handle forgotten
35559 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35562 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35563 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35564 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35565 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35566 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35567 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35568 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35569 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35570 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35571 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35572 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35573 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35574 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35575 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35576 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35577 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35578 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35579 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35580 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35581 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35582 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35583 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35584 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35585 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35586 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35587 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35588 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35589 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35590 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35591 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35592 definitions outside of __OPTIMIZE__ guarded section.
35595 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35596 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35597 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35598 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35601 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35603 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35604 ("vfenez<mode>"): Add missing constraints.
35606 2017-03-08 Martin Sebor <msebor@redhat.com>
35609 * config/nds32/nds32.c (nds32_option_override):
35610 Fix misspelled diagnostic.
35612 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35615 * gimplify.c (gimplify_omp_for): Replace index var in outer
35616 taskloop statement with an artificial variable and add
35617 OMP_CLAUSE_PRIVATE clause for it.
35619 2017-03-08 Richard Biener <rguenther@suse.de>
35621 PR tree-optimization/79955
35622 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35623 for accesses that are completely outside of the variable.
35625 2017-03-08 Andrew Haley <aph@redhat.com>
35627 PR tree-optimization/79943
35628 * tree-ssa-loop-split.c (compute_new_first_bound): When
35629 calculating the new upper bound, (END-BEG) should be added, not
35632 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35634 * config/avr/avr.md (setmemhi): Make sure match_dup
35635 operand number comes before match_scratch.
35637 2017-03-08 Richard Biener <rguenther@suse.de>
35639 PR tree-optimization/79920
35640 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35641 with ncopies == 1 to ...
35642 (vect_transform_slp_perm_load): ... here. Properly compute
35643 all element loads by iterating VF times over the group. Do
35644 not handle ncopies (computed in a broken way) in
35645 vect_create_mask_and_perm.
35647 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35650 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35651 is a uniform vector, use uniform_vector_p return value instead of
35652 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35654 2017-03-07 Marek Polacek <polacek@redhat.com>
35656 PR middle-end/79809
35657 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35658 (alloca_call_type): Likewise.
35660 2017-03-07 Martin Liska <mliska@suse.cz>
35662 * gcov.c (process_args): Put comment to correct location.
35664 2017-03-07 Martin Liska <mliska@suse.cz>
35666 PR middle-end/68270
35667 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35668 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35669 (chkp_narrow_bounds_for_field): Likewise.
35670 (chkp_parse_array_and_component_ref): Pass one more argument to
35673 2017-03-07 Richard Biener <rguenther@suse.de>
35675 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35678 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35680 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35681 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35683 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35686 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35687 to end of description.
35688 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35690 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35692 PR rtl-optimization/79901
35693 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35695 (*avx512f_<code><mode>3<mask_name>): ... this.
35696 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35697 iterator instead of VI8_AVX2_AVX512BW.
35699 PR rtl-optimization/79901
35700 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35701 min/max expander, expand it using expand_vec_cond_expr.
35704 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35707 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35710 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35711 to void * for PCH reasons.
35712 * dwarf2out.c (output_loc_operands, output_die): Cast
35713 v.val_vec.array to unsigned char *.
35715 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35718 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35721 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35723 PR rtl-optimization/79571
35724 * lra-constraints.c (process_alt_operands): Calculate static
35725 reject and subtract it from overall when only addresses will be
35728 2017-03-06 Julia Koval <julia.koval@intel.com>
35731 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35732 incoming stack boundary to 128 for 64-bit targets.
35734 2017-03-06 Richard Biener <rguenther@suse.de>
35736 PR tree-optimization/79894
35737 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35738 to NULL after folding it.
35740 2017-03-06 Richard Biener <rguenther@suse.de>
35742 PR tree-optimization/79824
35743 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35744 check disabling peeling for gaps.
35746 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35748 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35749 attributes): Document gettimeofday.
35751 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35753 * config/s390/s390.c (s390_option_override_internal): Set
35754 PARAM_MIN_VECT_LOOP_BOUND
35756 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35758 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35759 * config/s390/s390.md: Likewise.
35761 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35764 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35765 (<avx2_avx512>_perm<mode>): Rename to ...
35766 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35768 (<avx512>_perm<mode>_mask): Rename to ...
35769 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35771 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35772 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35773 instead of VI8F_256_512.
35774 (avx512f_perm<mode>): New define_expand.
35775 (avx512f_perm<mode>_mask): Likewise.
35776 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35777 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35779 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35781 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35782 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35784 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35786 2017-03-06 Martin Liska <mliska@suse.cz>
35789 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35790 when having a SSA NAME w/o VAR_DECL assigned to it.
35792 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35794 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35795 msa_dpsub_<su>_d): Fix MODE for vec_select.
35797 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35799 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35801 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35803 2017-03-06 Richard Biener <rguenther@suse.de>
35805 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35806 * plugin.c (register_plugin_info): Likewise.
35807 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35809 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35811 * config/i386/sse.md (sse_storehps, sse_storelps,
35812 avx_<castmode><avxsizesuffix>_<castmode>,
35813 avx512f_<castmode><avxsizesuffix>_<castmode>,
35814 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35815 in condition that at least one operand is not a MEM.
35817 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35819 PR middle-end/79805
35820 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35821 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35823 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35824 gimple_call_nothrow_p flag based on whether original builtin can throw.
35825 If it can, emit following stmts on the fallthrough edge.
35826 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35827 don't create new bb if inserting just debug stmts on the edge, try to
35828 insert them on the fallthru bb or just reset debug stmts.
35830 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35833 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35834 restore recog_data (including the operand rtxes inside it) around
35835 the call to get_insn_template.
35837 2017-03-03 Martin Sebor <msebor@redhat.com>
35839 PR tree-optimization/79699
35840 * context.c (context::~context): Free MPFR caches to avoid
35841 a memory leak on program exit.
35843 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35845 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35846 Use wide_int::ulow () instead of .elt (0).
35848 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35850 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35851 (*pushxf): Limit oF constraint to 32bit targets and add oC
35852 constraint for 64bit targets.
35853 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35854 (*pushdf): Change rmF constraint to rmC.
35856 2017-03-03 Martin Liska <mliska@suse.cz>
35858 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35859 Remove unused variable.
35861 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35864 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35865 is a memory operand, increase num_memory.
35866 (ix86_expand_args_builtin): Likewise.
35868 2017-03-03 Jan Hubicka <jh@suse.cz>
35871 * ipa-devirt.c (maybe_record_node): Properly handle
35872 __cxa_pure_virtual visibility.
35874 2017-03-03 Martin Liska <mliska@suse.cz>
35876 PR tree-optimization/79803
35877 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35879 (pass_loop_prefetch::execute): Disabled optimization if an
35880 assumption about L1 cache size is not met.
35882 2017-03-03 Martin Liska <mliska@suse.cz>
35884 PR rtl-optimization/79574
35885 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35886 (hash_scan_set): Likewise.
35887 (dump_hash_table): Likewise.
35888 (hoist_code): Likewise.
35890 2017-03-03 Richard Biener <rguenther@suse.de>
35892 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35893 in place of uhigh ().
35894 (fixed_convert_from_real): Likewise.
35896 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35899 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35901 2017-03-03 Richard Biener <rguenther@suse.de>
35903 PR middle-end/79818
35904 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35905 TYPE_OVERFLOW_UNDEFINED check.
35907 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35909 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35911 (vector_ae_<mode>_p): Likewise.
35912 (vector_nez_<mode>_p): Likewise.
35913 (vector_ne_v2di_p): Likewise.
35914 (vector_ae_v2di_p): Likewise.
35915 (vector_ne_<mode>_p): Likewise.
35916 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35918 (vsx_tsqrt<mode>2_fe): Likewise.
35920 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35923 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35925 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35927 PR rtl-optimization/79780
35928 * cprop.c (one_cprop_pass): When second and further conditional trap
35929 in a single basic block is turned into an unconditional trap, turn it
35930 into a deleted note to avoid RTL verification failures.
35932 2017-03-02 Richard Biener <rguenther@suse.de>
35934 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35936 2017-03-02 Richard Biener <rguenther@suse.de>
35938 PR tree-optimization/79345
35940 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35941 param and abort the walk, returning -1 if it is hit.
35942 (walk_aliased_vdefs): Take a limit param and pass it on.
35943 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35944 defaulting to 0 and return a signed int.
35945 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35946 (check_defs): New helper.
35947 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35948 about uninitialized memory.
35949 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35950 bogus uninitialized warning.
35951 (fixed_convert_from_real): Likewise.
35953 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35955 PR tree-optimization/66768
35956 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35957 iv_use if base object can't be determined.
35959 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35961 PR tree-optimization/79345
35962 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35963 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35964 (get_pattern_stats): Initialize it.
35965 * genemit.c (gen_expand): Verify match_scratch numbers come after
35966 match_operand/match_dup numbers.
35967 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35968 match_scratch numbers.
35969 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35971 * config/s390/s390.md (trunctdsd2): Likewise.
35973 2017-03-02 Richard Biener <rguenther@suse.de>
35975 * wide-int.h (wide_int_storage::operator=): Implement in terms
35978 2017-03-02 Richard Biener <rguenther@suse.de>
35980 PR tree-optimization/79777
35981 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35982 the to insert expression to sth existing.
35984 2017-03-01 Martin Sebor <msebor@redhat.com>
35986 PR middle-end/79692
35987 * gimple-ssa-sprintf.c
35988 (directive::known_width_and_precision): New function.
35989 (format_integer): Use it.
35990 (get_mpfr_format_length): Consider the full range of precision
35991 when computing %g output with the # flag. Set the likely byte
35992 count to 3 rather than 1 when precision is indeterminate.
35993 (format_floating): Correct the lower bound of precision.
35995 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35997 * doc/invoke.texi: Document default code model for 64-bit Linux.
35999 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36002 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36003 udiv rather than div since input pattern is unsigned.
36005 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36007 * config/i386/i386.c (print_reg): Warn for values of
36008 unsupported size in integer register.
36010 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36013 * config/rs6000/predicates.md (current_file_function_operand): Do
36014 not allow self calls to be local if the function is replaceable.
36016 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36019 * config/rs6000/altivec.h (vec_ctz and others): Change the
36020 preprocessor macro that controls conditional compilation from
36021 _ARCH_PWR9 to __POWER9_VECTOR__.
36022 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36023 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36024 control (instead of _ARCH_PWR9 control) so that template
36025 definition uses power9-specific function.
36026 (vec_any_eq): Likewise.
36027 (vec_all_ne): Change macro definition to use a power9-specific
36028 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36029 _ARCH_PWR9 control).
36030 (vec_any_eq) Likewise.
36031 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36032 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36033 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36034 support for xvcmpnedp instruction.
36035 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36036 macro expansion so that Power9 implementation of vec_all_ne does
36037 not use the AltiVec predicate framework.
36038 (VCMPNEH_P): Likewise.
36039 (VCMPNEW_P): Likewise.
36040 (VCMPNED_P): Likewise.
36041 (VCMPNEFP_P): Likewise.
36042 (VCMPNEDP_P): Likewise.
36043 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36044 implementation of vec_any_eq to not use AltiVec predicate
36046 (VCMPAEH_P): Likewise.
36047 (VCMPAEW_P): Likewise.
36048 (VCMPAED_P): Likewise.
36049 (VCMPAEFP_P): Likewise.
36050 (VCMPAEDP_P): Likewise.
36051 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36052 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36053 not use the AltiVec predicate framework.
36054 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36055 of vec_any_eq to not use AltiVec predicate framework.
36056 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36057 support for predefined __POWER9_VECTOR__ macro to indicate that
36058 Power9 instruction selection is enabled.
36059 (altivec_overloaded_builtins): Remove extraneous
36060 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36061 function argument types RS6000_BTI_bool_V16QI and
36062 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36063 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36064 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36065 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36066 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36067 Power9 for implementations of vec_cmpne. Change the signature for
36068 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36069 (representing vec_all_ne) to remove the previously described first
36070 argument of type RS6000_BTI_INTSI, as this was an artifact of
36071 reliance on the AltiVec predicate framework, which is no longer
36072 used in the implementation of these functions. Add
36073 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36074 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36075 since, unlike the AltiVec predicate framework implementation, we
36076 do not share function descriptors between vec_alle and vec_anyeq.
36077 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36078 set of modes that receive special treatment even when
36079 TARGET_P9_VECTOR is true. The special treatment emits code that
36080 does not depend on Power9 instructions.
36081 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36082 define_expand to not rely on AltiVec predicate framework.
36083 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36085 (vector_ne_v2di_p): Change this define_expand to not rely on
36086 AltiVec predicate framework.
36087 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36089 (vector_ne_<mode>_p): Change this define_expand to not rely on
36090 AltiVec predicate framework.
36091 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36093 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36094 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36095 define_insn pattern.
36096 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36097 define_insn pattern because the xvcmpne<VSs>. instruction is not
36099 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36100 instruction is not supported.
36102 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36104 * config/nvptx/nvptx.c: Include intl.h.
36106 2017-03-01 Martin Jambor <mjambor@suse.cz>
36109 * ipa-prop.h (ipa_bits): Removed field known.
36110 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36111 to pointers. Adjusted their comments to warn about their sharing.
36112 (ipcp_transformation_summary): Change bits to a vector of pointers.
36113 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36114 (ipa_get_ipa_bits_for_value): Declare.
36115 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36116 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36117 (ipa_bits_hash_table): Likewise.
36118 (ipa_vr_ggc_hash_traits): Likewise.
36119 (ipa_vr_hash_table): Likewise.
36120 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36121 being pointers and vr_known being removed.
36122 (ipa_set_jf_unknown): Likewise.
36123 (ipa_get_ipa_bits_for_value): New function.
36124 (ipa_set_jfunc_bits): Likewise.
36125 (ipa_get_value_range): New overloaded functions.
36126 (ipa_set_jfunc_vr): Likewise.
36127 (ipa_compute_jump_functions_for_edge): Use the above functions to
36128 construct bits and vr parts of jump functions.
36129 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36130 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36132 (ipcp_grow_transformations_if_necessary): Also allocate
36133 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36135 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36136 them. Fix too long lines.
36137 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36138 vr_known being removed.
36139 (ipa_read_jump_function): Use new setter functions to construct bits
36140 and vr parts of jump functions or set them to NULL.
36141 (write_ipcp_transformation_info): Adjust for bits being pointers.
36142 (read_ipcp_transformation_info): Likewise.
36143 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36145 Include gt-ipa-prop.h.
36146 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36148 (ipcp_store_bits_results): Likewise.
36149 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36150 Do not write to existing jump functions but use a temporary instead.
36152 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36155 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36156 attempt to use its first operand as BIT_FIELD_REF base.
36158 2017-03-01 Richard Biener <rguenther@suse.de>
36160 PR middle-end/79721
36161 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36162 interpolating formula in wrapping arithmetic.
36163 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36165 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36167 PR tree-optimization/79734
36168 * tree-vect-generic.c (expand_vector_condition): Optimize
36169 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36170 Handle VEC_COND_EXPR where comparison has different inner width from
36171 type's inner width.
36173 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36175 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36176 markup, and similar issues. Remove @opindex entries for things
36177 that aren't options. Add missing -mmpy-option entries.
36179 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36181 PR tree-optimization/79737
36182 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36183 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36184 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36185 instead of byte_size. Formatting fix.
36186 (shift_bytes_in_array_right): Formatting fix.
36188 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36191 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36192 condition on optimize for the leaf function test.
36194 2017-02-28 Martin Liska <mliska@suse.cz>
36197 * read-rtl-function.c (function_reader::handle_unknown_directive):
36198 Bail out when one uses -flto.
36200 2017-02-28 Martin Liska <mliska@suse.cz>
36202 * common.opt: Replace space with tabular for options of <number>
36204 * config/i386/i386.opt: Show <number> value for
36205 -mlarge-data-threshold.
36206 * opts.c (print_filtered_help): Do not display number in hexadecimal
36209 2017-02-28 Martin Liska <mliska@suse.cz>
36211 * common.opt: Fix --help=option -Q for options which are of
36214 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36216 * config/i386/i386.c (print_reg): Error out for values
36217 of 8-bit size in invalid integer register.
36219 2017-02-28 Martin Sebor <msebor@redhat.com>
36221 PR tree-optimization/79691
36222 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36224 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36227 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36228 gcc_unreachable with output_operand_lossage.
36230 2017-02-28 Richard Biener <rguenther@suse.de>
36232 PR tree-optimization/79740
36233 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36235 (visit_nary_op): Insert the nary into the hashtable if we
36236 pattern-matched sth.
36237 * tree-ssa-pre.c (eliminate_insert): Robustify.
36239 2017-02-28 Richard Biener <rguenther@suse.de>
36241 PR middle-end/79731
36242 * fold-const.c (decode_field_reference): Reject out-of-bound
36245 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36247 * config/i386/i386.c: Include intl.h.
36248 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36249 instead of just cond ? "..." : "...".
36250 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36251 * coverage.c (read_counts_file): Likewise.
36252 * omp-offload.c: Include intl.h.
36253 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36254 of just cond ? "..." : "...".
36255 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36256 of just cond ? "..." : "...".
36258 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36261 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36263 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36264 'tune for' CPU name.
36265 * config/arm/arm-cpu-data.h: Regenerated.
36267 2017-02-28 Richard Biener <rguenther@suse.de>
36269 PR tree-optimization/79732
36270 * tree-inline.c (expand_call_inline): Do not shadow var.
36272 2017-02-28 Richard Biener <rguenther@suse.de>
36274 PR tree-optimization/79723
36275 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36276 address-space properly.
36278 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36280 * doc/optinfo.texi (Optimization groups): Fix option used for
36282 * doc/invoke.texi (-fopt-info): Document "omp".
36283 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36284 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36285 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36286 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36287 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36289 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36291 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36292 OPTGROUP_OMP with "omp".
36294 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36297 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36298 for arithmetic shift of unsigned V2DI.
36300 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36302 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36303 arc/linux.h headers.
36304 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36305 (LINK_SPEC): Likewise.
36306 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36307 (EXTRA_SPECS): Likewise.
36308 (STARTFILE_SPEC): Likewise.
36309 (ENDFILE_SPEC): Likewise.
36310 (LIB_SPEC): Likewise.
36311 (TARGET_SDATA_DEFAULT): Likewise.
36312 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36313 (MULTILIB_DEFAULTS): Likewise.
36314 (DWARF2_UNWIND_INFO): Likewise.
36315 * config/arc/big.h: New file.
36316 * config/arc/elf.h: Likewise.
36317 * config/arc/linux.h: Likewise.
36318 * config/arc/t-uClibc: Remove.
36320 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36322 PR tree-optimization/77536
36323 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36324 (tree_transform_and_unroll_loop): Use above function to compute the
36325 estimated niter of unrolled loop and use it when scaling profile.
36326 Also use count info rather than frequency if it's non-zero.
36327 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36328 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36329 (vect_transform_loop): Call above function.
36331 2017-02-27 Richard Biener <rguenther@suse.de>
36333 PR tree-optimization/45397
36334 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36335 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36336 (visit_nary_op): Add pattern matching for CSEing sign-changed
36337 or truncated operations with wider ones.
36339 2017-02-27 Richard Biener <rguenther@suse.de>
36341 PR tree-optimization/79690
36342 * tree-vect-stmts.c (vectorizable_store): Use vector type
36343 built from the DR with address-space.
36345 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36347 * doc/invoke.texi (Optimize Options): Refine the description
36348 of asan-use-after-return.
36350 2017-02-25 Alan Modra <amodra@gmail.com>
36352 PR rtl-optimization/79584
36353 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36354 base, not ad->base_term, the reg within base. Remove assertion
36355 that ad->base == ad->base_term. Replace gen_int_mode using
36356 bogus mode with const0_rtx.
36358 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36360 PR middle-end/79396
36361 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36362 FMA_EXPR like tcc_binary or tcc_unary.
36364 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36367 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36369 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36370 (output_loc_operands): Handle DW_OP_call_ref and
36371 DW_OP_GNU_variable_value.
36372 (struct variable_value_struct): New type.
36373 (struct variable_value_hasher): Likewise.
36374 (variable_value_hash): New variable.
36375 (string_types): Remove.
36376 (copy_loc_descr): New function.
36377 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36378 (prepend_loc_descr_to_each): New function.
36379 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36380 instead of add_loc_descr_to_each if the first argument is single
36381 location list and the second has multiple.
36382 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36383 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36384 when looking for variable value which doesn't have other location info.
36385 (loc_list_from_tree): Formatting fix.
36386 (gen_array_type_die): Simplify DW_AT_string_length handling.
36387 (adjust_string_types): Remove.
36388 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36389 string_types. Call resolve_variable_values.
36390 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36391 (resolve_addr_in_expr): Likewise. Add A argument.
36392 (copy_deref_exprloc): Remove deref argument. Adjust for the
36393 original expression being DW_OP_GNU_variable_value with optionally
36394 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36395 optionally after it.
36396 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36397 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36398 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36399 (variable_value_hasher::hash, variable_value_hasher::equal): New
36401 (resolve_variable_value_in_expr, resolve_variable_value,
36402 resolve_variable_values, note_variable_value_in_expr,
36403 note_variable_value): New functions.
36404 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36407 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36410 * opts.h (handle_generated_option): Add GENERATED_P argument.
36411 * opts-common.c (handle_option): Adjust function comment.
36412 (handle_generated_option): Add GENERATED_P argument, pass it to
36414 (control_warning_option): Pass false to handle_generated_option
36416 * opts.c (maybe_default_option): Pass true to handle_generated_option
36418 * optc-gen.awk: Likewise.
36420 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36422 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36423 a REG, look at the REG it is a SUBREG of.
36424 (splitter for cmpeqsi_t): Ditto.
36426 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36428 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36429 the special USEs with the pattern of the insn, not the insn itself.
36431 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36434 * doc/invoke.texi: Document -mload-store-pairs.
36436 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36437 Sandra Loosemore <sandra@codesourcery.com>
36439 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36440 argument isn't a CONST_INT.
36441 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36442 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36443 (nios2_validate_compare): Bypass alternate compare logic if *op2
36444 is not a CONST_INT.
36445 (ldstwm_operation_p): Return false if first_base is not a REG or
36446 if first_offset is not a CONST_INT.
36448 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36450 * config/cris/cris.md: Use correct operand in a define_peephole2.
36452 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36454 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36456 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36458 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36459 this_insn if it is an INSN or JUMP_INSN.
36460 (force_offsettable): Look at base, not at addr.
36461 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36462 on things that aren't necessarily CONST_INTs.
36464 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36466 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36467 -mfpmath=sse is the default also for x86-32 targets with SSE2
36468 instruction set when @option{-ffast-math} is enabled
36470 2017-02-24 Jeff Law <law@redhat.com>
36472 PR rtl-optimizatoin/79286
36473 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36476 2017-02-24 Richard Biener <rguenther@suse.de>
36478 PR tree-optimization/79389
36479 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36482 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36484 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36485 function comment to reflect reality.
36486 (loop_exits_before_overflow): Fix typo in function description.
36488 2017-02-24 Richard Biener <rguenther@suse.de>
36490 PR tree-optimization/79389
36491 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36492 properly that a threading opportunity exists. Detect conditional
36493 copy/constant propagation opportunities.
36495 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36497 * config/visium/visium.md (type): Add trap.
36498 (b): New mode attribute.
36499 (*btst): Rename into...
36500 (*btst<mode>): ...this and adjust.
36501 (*cbranchsi4_btst_insn): Rename into...
36502 (*cbranch<mode>4_btst_insn): ...this and adjust.
36503 (trap): New define_insn.
36505 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36507 PR tree-optimization/79389
36508 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36509 (noce_reversed_cond_code): New function.
36510 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36511 reversed_comparison_code. Formatting fix.
36512 (noce_try_store_flag): Test rev_cond != NULL in addition to
36513 reversed_comparison_code.
36514 (noce_try_store_flag_constants): Likewise.
36515 (noce_try_store_flag_mask): Likewise.
36516 (noce_try_addcc): Use rev_cond if non-NULL instead of
36517 reversed_comparison_code.
36518 (noce_try_cmove_arith): Likewise. Formatting fixes.
36519 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36520 (noce_find_if_block): Initialize rev_cond.
36521 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36522 instead of false as last argument never attempt to reverse it
36525 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36527 PR tree-optimization/79663
36528 * tree-predcom.c (combine_chains): Process refs in reverse order
36529 only for ZERO length chains, and add explaining comment.
36531 2017-02-23 Jeff Law <law@redhat.com>
36533 PR tree-optimization/79578
36534 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36535 in call to operand_equal_p.
36537 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36540 * config/i386/cpuid.h: Fix another undefined behavior.
36542 2017-02-23 Richard Biener <rguenther@suse.de>
36544 PR tree-optimization/79683
36545 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36546 vector types for data-refs.
36548 2017-02-23 Martin Liska <mliska@suse.cz>
36550 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36552 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36554 PR middle-end/79665
36555 * internal-fn.c (get_range_pos_neg): Moved to ...
36556 * tree.c (get_range_pos_neg): ... here. No longer static.
36557 * tree.h (get_range_pos_neg): New prototype.
36558 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36559 are known to be in between 0 and signed maximum inclusive, try to
36560 expand both unsigned and signed divmod and use the cheaper one from
36563 2017-02-22 Jeff Law <law@redhat.com>
36565 PR tree-optimization/79578
36566 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36567 to compare base operands.
36569 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36572 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36573 gpc_reg_operand instead of fpr_reg_operand.
36575 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36577 * config/mips/mips.c (mips_return_in_memory): Force FP
36578 vector types to be returned in memory for o32 ABI.
36580 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36582 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36583 instead of DW_TAG_member for static data member declarations and don't
36584 set no_linkage_name for static inline data members.
36585 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36588 2017-02-22 Martin Liska <mliska@suse.cz>
36590 * doc/invoke.texi: Replace inequality signs with square brackets
36593 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36596 * lra-constraints.c (simplify_operand_subreg): Handle
36597 WORD_REGISTER_OPERATIONS targets.
36599 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36602 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36603 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36604 elimination by swapping fld*.
36606 2017-02-22 Richard Biener <rguenther@suse.de>
36608 PR tree-optimization/79673
36609 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36610 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36611 irrelevant address-space qualifiers and avoiding a
36612 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36614 2017-02-22 Richard Biener <rguenther@suse.de>
36616 PR tree-optimization/79666
36617 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36618 to not symbolically negate if that may introduce undefined
36621 2017-02-22 Martin Liska <mliska@suse.cz>
36624 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36625 * data-streamer-out.c (streamer_write_gcov_count_stream):
36627 * value-prof.c (stream_out_histogram_value): Make assert more
36628 precise based on type of counter.
36630 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36633 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36634 Use nonimmediate_operand instead of memory_operand for operand 1.
36635 (float-extend standard_x87sse_constant_load splitter): Ditto.
36637 2017-02-21 Jeff Law <law@redhat.com>
36639 PR tree-optimization/79621
36640 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36641 blocks with edges to themselves.
36643 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36646 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36647 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36648 Use gimple_call_builtin_p.
36651 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36652 on temporarily removed DEBUG_INSNs.
36654 PR tree-optimization/79649
36655 * tree-loop-distribution.c (classify_partition): Give up on
36656 non-generic address space loads/stores.
36658 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36660 * doc/loop.texi (Loop manipulation): Remove nonexistent
36661 tree_ssa_loop_version from the documentation.
36662 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36664 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36667 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36668 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36669 * config/rs6000/rs6000.c: Include except.h.
36670 (rs6000_expand_split_stack_prologue): Call
36671 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36673 2017-02-21 Martin Jambor <mjambor@suse.cz>
36676 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36677 have been analyzed.
36679 2017-02-21 Martin Jambor <mjambor@suse.cz>
36681 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36682 for backward compatibility only.
36683 * doc/invoke.texi (Option Summary): Remove all references to
36684 -fipa-cp-alignment.
36686 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36690 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36692 * lra-constraints.c (curr_insn_transform): Handle
36693 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36695 2017-02-21 Martin Liska <mliska@suse.cz>
36697 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36699 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36701 PR translation/79638
36702 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36704 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36707 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36708 (arm_function_ok_for_sibcall): Return false for an indirect call by
36709 descriptor if all the argument registers are used.
36710 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36711 alignment of the function.
36713 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36715 PR tree-optimization/61441
36716 * simplify-rtx.c (simplify_const_unary_operation): For
36717 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36718 the sNaN unmodified.
36720 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36722 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36723 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36724 instead of SYSTEM_HEADER_DIR.
36726 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36727 Martin Liška <mliska@suse.cz>
36729 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36730 Fix typos and grammar, use active voice, and clarify.
36732 2017-02-20 Marek Polacek <polacek@redhat.com>
36734 PR middle-end/79537
36735 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36738 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36740 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36743 * config/i386/i386.c (ix86_expand_builtin): Handle
36744 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36745 ix86_builtins_isa[fcode].isa as a requirement of those
36746 flags and any other flag in the bitmask.
36747 (ix86_init_mmx_sse_builtins): Use 0 instead of
36748 ~OPTION_MASK_ISA_64BIT as mask.
36749 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36750 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36751 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36752 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36754 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36757 * lra-constraints.c (split_reg): Check requested split mode
36758 is supported by the register.
36760 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36762 * lra-constraints.c (simplify_operand_subreg): Remove early
36765 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36768 * lra-constraints.c (curr_insn_transform): Tighten condition
36769 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36771 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36774 * lra-constraints.c (curr_insn_transform): Handle
36775 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36777 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36780 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36782 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36784 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36787 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36790 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36792 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36795 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36797 * final.c (last_columnnum, override_columnnum): New variables.
36798 (final_start_function): Set last_columnnum, pass it to begin_prologue
36799 hook and pass 0 to dwarf2out_begin_prologue.
36800 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36801 to source_line debug hook.
36802 (notice_source_line): Compute last_columnnum and for debug_column_info
36803 return true on column changes.
36804 * debug.h (struct gcc_debug_hooks): Add column argument to
36805 source_line and begin_prologue hooks.
36806 (debug_nothing_int_charstar_int_bool): Remove prototype.
36807 (debug_nothing_int_int_charstar,
36808 debug_nothing_int_int_charstar_int_bool): New prototypes.
36809 (dwarf2out_begin_prologue): Add column argument.
36810 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36811 begin_prologue hooks.
36812 (debug_nothing_int_charstar_int_bool): Remove.
36813 (debug_nothing_int_int_charstar,
36814 debug_nothing_int_int_charstar_int_bool): New functions.
36815 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36816 through to dwarf2out_source_line.
36817 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36818 (dwarf2out_source_line): Add column argument, emit it if requested.
36819 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36821 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36822 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36823 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36824 through to dwarf2out_begin_prologue.
36825 (vmsdbgout_source_line): Add column argument, pass it through to
36826 dwarf2out_source_line.
36827 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36828 dbxout_source_line caller.
36829 (dbxout_source_line): Add column argument.
36831 * common.opt (gno-column-info, gcolumn-info): New options.
36832 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36833 (check_die): Also test for multiple DW_AT_decl_column attributes.
36834 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36835 DW_AT_decl_column if requested.
36836 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36838 (gen_variable_die): Likewise.
36839 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36840 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36843 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36844 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36845 (ix86_handle_option): Handle OPT_m3dnowa.
36846 * doc/invoke.texi (-m3dnowa): Document.
36847 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36848 -m3dnowa instead of -m3dnow -march=athlon.
36851 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36852 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36854 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36857 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36858 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36859 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36860 generator for vsx_xxpermdi_<mode>_be.
36861 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36862 force big-endian semantics.
36863 (vsx_xxpermdi_<mode>_be): New define_expand with same
36864 implementation as previous version of vsx_xxpermdi_<mode>.
36866 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36868 PR tree-optimization/79327
36869 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36870 variable, its initialization and use.
36872 2017-02-17 Julia Koval <julia.koval@intel.com>
36874 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36875 (OPTION_MASK_ISA_PKU_UNSET): New.
36876 (ix86_handle_option): Handle -mrdpid.
36877 * config/i386/cpuid.h (bit_RDPID): New.
36878 * config/i386/driver-i386.c (host_detect_local_cpu):
36879 Detect RDPID feature.
36880 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36881 * config/i386/i386-c.c (ix86_target_macros_internal):
36883 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36884 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36885 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36886 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36887 * config/i386/i386.md (define_insn "rdpid"): New.
36888 * config/i386/i386.opt Add -mrdpid.
36889 * config/i386/immintrin.h (_rdpid_u32): New.
36891 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36893 PR rtl-optimization/79541
36894 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36895 instead of transforming it into USE.
36897 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36899 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36900 If HONOR_SNANS (SFmode) force the input to a register.
36901 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36902 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36903 an frsp or similar insn.
36905 2017-02-17 Martin Liska <mliska@suse.cz>
36907 PR rtl-optimization/79577
36908 * params.def (selsched-max-sched-times): Increase minimum to 1.
36910 2017-02-17 Martin Liska <mliska@suse.cz>
36912 PR rtl-optimization/79574
36913 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36915 2017-02-17 Martin Liska <mliska@suse.cz>
36917 PR tree-optimization/79529
36918 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36919 ssa_defined_default_def_p to handle cases which are implicitly
36921 * tree-ssa.c (ssa_defined_default_def_p): New function.
36922 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36923 which are implicitly defined.
36924 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36926 2017-02-17 Richard Biener <rguenther@suse.de>
36928 PR middle-end/79576
36929 * params.def (max-ssa-name-query-depth): Limit to 10.
36931 2017-02-17 Richard Biener <rguenther@suse.de>
36933 PR tree-optimization/79552
36934 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36937 2017-02-17 Richard Biener <rguenther@suse.de>
36940 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36942 2017-02-17 Marek Polacek <polacek@redhat.com>
36944 PR middle-end/79536
36945 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36946 (fold_negate_expr): New wrapper.
36948 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36950 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36951 Correct terminology and de-emphasize pre-standard behavior.
36953 2017-02-16 Alan Modra <amodra@gmail.com>
36955 PR rtl-optimization/79286
36956 * ira.c (def_dominates_uses): New function.
36957 (update_equiv_regs): Don't create an equivalence for insns that
36958 may trap where the register def does not dominate the use.
36960 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36962 PR rtl-optimization/78127
36963 * lra.c (lra): Call lra_eliminate before finish the loop after
36966 2017-02-16 Richard Biener <rguenther@suse.de>
36968 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36970 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36971 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36972 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36973 (isl_val_int_from_wi): New function.
36974 (extract_affine_gmp): Rename to ...
36975 (extract_affine_wi): ... this, take a widest_int.
36976 (extract_affine_int): Just wrap extract_affine_wi.
36977 (add_param_constraints): Use isl_val_int_from_wi.
36978 (add_loop_constraints): Likewise, and extract_affine_wi.
36980 2017-02-15 Jeff Law <law@redhat.com>
36982 PR middle-end/79521
36983 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36984 ira_init_register_move_cost_if_necessary.
36986 2017-02-15 Martin Sebor <msebor@redhat.com>
36988 PR middle-end/32003
36989 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36990 removed in a prior commit.
36992 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36994 PR tree-optimization/79347
36995 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36996 counters during peeling.
36998 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37000 * Makefile.in (site.exp): Remove "set ISLVER".
37002 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37005 * real.c (real_from_integer): Call real_convert even for decimal.
37007 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37010 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37012 2017-02-14 Andrew Pinski <apinski@cavium.com>
37014 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37015 cores and change the partno/implementer to be correct.
37016 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37017 the 'B" as the implementer.
37018 * config/aarch64/aarch64-tune.md: Regenerate.
37020 2017-02-14 Carl Love <cel@us.ibm.com>
37022 * config/rs6000/rs6000.c: Add case statement entry to make the
37023 xvcvuxdsp built-in argument unsigned.
37024 * config/rs6000/vsx.md: Fix the source and return operand types so they
37025 match the instruction definitions from the ISA document. Fix typo
37026 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37029 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37032 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37033 member early_clobber_alts.
37034 * lra-lives.c (reg_early_clobber_p): New.
37035 (process_bb_lives): Use it.
37036 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37037 (debug_operand_data): Initialize early_clobber_alts.
37038 (setup_operand_alternative): Set up early_clobber_alts.
37039 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37040 alternatives to new_insn_reg.
37041 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37043 (lra_update_insn_regno_info): Pass the new arg.
37045 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37047 PR middle-end/79505
37048 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37049 (new_oacc_loop_raw): Don't clear already cleared fields.
37052 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37053 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37054 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37056 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37059 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37061 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37064 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37065 the extra instruction to the right place to store 128-bit constant
37068 2017-02-14 Martin Sebor <msebor@redhat.com>
37070 PR middle-end/79448
37071 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37072 warning for strings of unknown length.
37074 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37076 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37078 2017-02-14 Jeff Law <law@redhat.com>
37081 * ira-costs.c (scan_one_insn): Initialize register move costs
37082 for pseudos seen in USE/CLOBBER insns.
37084 PR tree-optimization/79095
37085 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37086 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37087 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37088 if the operands are known to be not equal, then the resulting range
37090 (intersect_ranges): If the new range is ~[0,0] and the old range is
37091 wide, then prefer ~[0,0].
37092 * tree-vrp.c (overflow_comparison_p_1): New function.
37093 (overflow_comparison_p): New function.
37094 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37095 if NAME is used in an overflow test.
37096 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37097 overflow check that can be expressed as an equality test, then adjust
37098 ops to be that equality test.
37100 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37102 * config/s390/s390-builtin-types.def: Remove flags argument.
37103 * config/s390/s390.c (s390_init_builtins): Likewise.
37105 2017-02-14 Martin Liska <mliska@suse.cz>
37107 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37108 vector. Fix trailing white spaces.
37110 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37112 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37115 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37117 PR rtl-optimization/68664
37118 * config/arm/arm.c (arm_sched_can_speculate_insn):
37119 New function. Declare prototype.
37120 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37122 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37124 PR rtl-optimization/68664
37125 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37127 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37129 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37131 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37132 max skip bytes for function, loop and jump.
37134 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37136 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37137 ABS_EXPR for gimple dump.
37139 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37142 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37144 PR tree-optimization/79408
37145 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37146 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37147 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37148 also if rhs1 is INTEGER_CST.
37150 2017-02-14 Richard Biener <rguenther@suse.de>
37152 PR middle-end/79432
37153 * tree-into-ssa.c (insert_phi_nodes): When the function can
37154 have abnormal edges rewrite SSA names with broken use-def
37155 dominance out of SSA and register them for PHI insertion.
37157 2017-02-13 Martin Sebor <msebor@redhat.com>
37159 PR middle-end/79496
37160 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37161 clearing info.nowrite flag when snprintf size argument is a range.
37163 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37165 * cprop.c (cprop_jump): Add missing space in string literal.
37166 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37167 (get_constraint_for_component_ref): Likewise.
37168 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37169 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37170 * lra-constraints.c (process_alt_operands): Likewise.
37171 * ipa-inline.c (inline_small_functions): Likewise.
37172 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37173 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37174 * trans-mem.c (diagnose_tm_1_op): Likewise.
37175 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37176 (grid_parallel_clauses_gridifiable): Likewise.
37178 * config/nvptx/mkoffload.c (process): Add space in between
37181 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37182 "MOD4_SSE_REGS" and "ALL_REGS".
37184 * spellcheck.c (test_data): Add , in between "foo" and "food".
37186 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37189 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37190 boundary crossing check and subsequent code generation agree.
37192 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37194 * config/aarch64/aarch64.c (has_memory_op): Delete.
37195 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37198 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37200 PR rtl-optimization/79388
37201 PR rtl-optimization/79450
37202 * combine.c (distribute_notes): When removing TEM_INSN for which
37203 corresponding dest has last value recorded, invalidate that last
37206 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37208 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37209 of explicit '@'. Add missing assembly comment marker on branch costs
37212 2017-02-13 Nathan Sidwell <nathan@acm.org>
37214 * gengtype-lex.l (<in_struct>): Add '/'.
37216 2017-02-13 Martin Liska <mliska@suse.cz>
37219 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37221 2017-02-13 Richard Biener <rguenther@suse.de>
37223 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37225 * configure: Re-generate.
37226 * config.in: Likewise.
37227 * graphite-dependences.c: Simplify as if
37228 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37229 * graphite-isl-ast-to-gimple.c: Likewise.
37230 * graphite-optimize-isl.c: Likewise.
37231 * graphite-poly.c: Likewise.
37232 * graphite-sese-to-poly.c: Likewise.
37233 * graphite.h: Likewise.
37234 * toplev.c: Include isl/version.h and use isl_version () for
37235 printing the ISL version.
37236 * doc/install.texi: Update ISL requirement.
37238 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37240 * doc/standards.texi (Standards): Update reference to
37243 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37245 * doc/extend.texi (Named Address Spaces): sourceware.org now
37247 * doc/install.texi (Binaries): Ditto.
37250 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37252 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37253 standard terminology "stringize"/"stringizing" throughout.
37254 * doc/cppinternals.texi: Likewise.
37256 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37258 * doc/extend.texi: Fix some spelling mistakes and typos.
37259 * doc/invoke.texi: Likewise.
37261 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37264 * params.def (inline-min-speedup) Change from 10 to 8.
37266 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37268 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37271 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37274 * ipa-inline-analysis.c (get_minimal_bb): New function.
37275 (record_modified): Use it.
37276 (remap_edge_change_prob): Handle also ancestor functions.
37278 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37280 * doc/contrib.texi (Contributors): Remove broken link into
37281 the Mauve CVS repository.
37283 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37285 PR middle-end/79454
37286 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37287 result computation whenever lhs doesn't have vector mode, not
37288 just when it has BLKmode.
37290 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37292 * doc/makefile.texi (profiledbootstrap): Refer to the
37293 installation instructions only in textual form.
37295 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37298 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37300 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37302 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37303 (Specific): Update mingw-w64 reference.
37305 (Specific): Remove broken link to Renesas RX processor.
37307 2017-02-10 Richard Biener <rguenther@suse.de>
37309 * toplev.c (process_options): Do not mention obsolete graphite
37310 options when printing sorry message about missing graphite support.
37311 Mention -floop-nest-optimize.
37313 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37315 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37316 (vtst_p16): Likewise.
37317 (vtstq_p8): Likewise.
37318 (vtstq_p16): Likewise.
37320 (vtstq_p64): Likewise.
37321 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37322 (vset_lane_p64): New.
37323 (vsetq_lane_p64): New.
37325 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37327 PR tree-optimization/79411
37328 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37329 stmt operands are SSA_NAMEs used in abnormal phis.
37330 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37333 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37336 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37339 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37341 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37343 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37345 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37349 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37350 not arbitrary TREE_CONSTANT.
37353 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37354 "omp declare target link" attribute unless is_global_var.
37355 * omp-offload.c (find_link_var_op): Likewise.
37357 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37358 Chung-Lin Tang <cltang@codesourcery.com>
37360 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37362 (gimplify_adjust_omp_clauses): Don't delete TILE.
37363 (gimplify_omp_for): Deal with TILE.
37364 * internal-fn.c (expand_GOACC_TILE): New function.
37365 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37367 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37368 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37370 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37371 avoid DIV for outermost collapse var.
37372 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37373 Remove out of date comments, fix whitespace.
37374 * omp-general.c (omp_extract_for_data): Deal with tiling.
37375 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37376 adjust OLF_DIM_BASE value.
37377 (struct omp_for_data): Add tiling field.
37378 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37379 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37380 for auto loops. Remove default auto determining, moved to
37381 oacc_loop_fixed_partitions.
37382 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37383 stmts, add e_mask field.
37384 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37385 (oacc_thread_numbers): Use oacc_dim_call.
37386 (oacc_xform_tile): New.
37387 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37388 (finish_oacc_loop): Adjust for ifns vector.
37389 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37390 add case for GOACC_TILE fns.
37391 (oacc_loop_xform_loop): Delete.
37392 (oacc_loop_process): Iterate over call list directly, and add
37393 handling for GOACC_TILE fns.
37394 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37396 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37397 vector partitioning to outer loops. Assign 2 partitions to loops
37398 when available. Add TILE handling.
37399 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37400 (execite_oacc_device_lower): Process GOACC_TILE fns,
37401 ignore unknown specs.
37402 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37403 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37404 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37406 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37408 * configure.ac (ACX_BUGURL): Update.
37409 * configure: Regenerate.
37411 2017-02-09 Richard Biener <rguenther@suse.de>
37413 PR tree-optimization/69823
37414 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37415 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37417 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37419 * config/arc/arc-c.def: Add __NPS400__ definition.
37420 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37421 (TARGET_NPS400): Define.
37423 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37425 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37427 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37428 pointer, arch_info.
37429 (arc_cpu_types): Fill the arch_info field with a pointer into the
37430 arc_arch_types table.
37431 (arc_selected_cpu): Declare.
37432 * config/arc/arc.c (arc_selected_cpu): Make global.
37433 (arc_selected_arch): Delete.
37434 (arc_base_cpu): Delete.
37435 (arc_override_options): Remove references to deleted variables,
37436 update access to arch information.
37437 (ARC_OPT): Update access to arch information.
37438 (ARC_OPTX): Likewise.
37439 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37440 (TARGET_ARC600): Update access to arch information.
37441 (TARGET_ARC601): Likewise.
37442 (TARGET_ARC700): Likewise.
37443 (TARGET_EM): Likewise.
37444 (TARGET_HS): Likewise.
37445 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37448 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37451 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37452 condition/operands for integer GE/LE/GEU/LEU operations.
37454 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37456 PR translation/79397
37457 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37460 2017-02-08 Martin Jambor <mjambor@suse.cz>
37463 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37464 whether allocation happened.
37465 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37466 nothing was allocated.
37468 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37470 PR tree-optimization/79408
37471 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37472 constant, but SSA_NAME with a known integer range, use the minimum
37473 of that range instead of op1 to determine if modulo can be replaced
37474 with its first operand.
37476 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37478 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37480 2017-02-08 Richard Biener <rguenther@suse.de>
37482 PR tree-optimization/71824
37483 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37484 Check all loops contained in the merged region.
37486 2017-02-07 Andrew Pinski <apinski@cavium.com>
37488 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37490 2017-02-07 Andrew Pinski <apinski@cavium.com>
37492 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37493 (thunderxt88): Likewise.
37494 (thunderxt81): Disable LSE and change v8.1 to v8.
37495 (thunderxt83): Likewise.
37497 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37498 Richard Biener <rguenther@suse.de>
37500 PR middle-end/79399
37501 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37502 type from int to size_t.
37503 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37505 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37507 PR rtl-optimization/79386
37508 * cprop.c (bypass_conditional_jumps): Initialize
37509 bypass_last_basic_block already before splitting bbs after
37510 unconditional traps...
37511 (bypass_conditional_jumps): ... rather than here.
37514 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37515 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37516 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37517 fix -masm=intel patterns.
37519 2017-02-07 Richard Biener <rguenther@suse.de>
37521 PR tree-optimization/79256
37522 PR middle-end/79278
37523 * builtins.c (get_object_alignment_2): Use min_align_of_type
37524 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37525 and ADJUST_FIELD_ALIGN.
37527 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37529 * doc/tm.texi: Regenerate.
37530 * stor-layout.c (layout_decl): Adjust.
37531 (update_alignment_for_field): Likewise.
37532 (place_field): Likewise.
37533 (min_align_of_type): Likewise.
37534 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37535 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37536 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37537 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37538 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37539 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37540 * config/i386/i386.c (x86_field_alignment): Likewise.
37541 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37542 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37543 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37544 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37545 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37546 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37550 2017-01-30 Richard Biener <rguenther@suse.de>
37552 PR tree-optimization/79256
37553 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37554 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37557 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37559 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37560 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37561 builtins to SImode and emit a zero-extend, if necessary.
37563 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37565 * docs/invoke.texi (RISC-V Options): Alphabetize.
37567 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37569 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37572 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37574 * config/riscv/riscv.c: New file.
37575 * common/config/riscv/riscv-common.c: Likewise.
37576 * config.gcc: Likewise.
37577 * config/riscv/constraints.md: Likewise.
37578 * config/riscv/elf.h: Likewise.
37579 * config/riscv/generic.md: Likewise.
37580 * config/riscv/linux.h: Likewise.
37581 * config/riscv/multilib-generator: Likewise.
37582 * config/riscv/peephole.md: Likewise.
37583 * config/riscv/pic.md: Likewise.
37584 * config/riscv/predicates.md: Likewise.
37585 * config/riscv/riscv-builtins.c: Likewise.
37586 * config/riscv/riscv-c.c: Likewise.
37587 * config/riscv/riscv-ftypes.def: Likewise.
37588 * config/riscv/riscv-modes.def: Likewise.
37589 * config/riscv/riscv-opts.h: Likewise.
37590 * config/riscv/riscv-protos.h: Likewise.
37591 * config/riscv/riscv.h: Likewise.
37592 * config/riscv/riscv.md: Likewise.
37593 * config/riscv/riscv.opt: Likewise.
37594 * config/riscv/sync.md: Likewise.
37595 * config/riscv/t-elf-multilib: Likewise.
37596 * config/riscv/t-linux: Likewise.
37597 * config/riscv/t-linux-multilib: Likewise.
37598 * config/riscv/t-riscv: Likewise.
37599 * configure.ac: Likewise.
37600 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37601 Waterman as RISC-V maintainers.
37602 * doc/install.texi: Add RISC-V entries.
37603 * doc/invoke.texi: Add RISC-V options section.
37604 * doc/md.texi: Add RISC-V constraints section.
37605 * configure: Regenerated.
37607 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37610 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37611 false values to be constant vectors with all 0 or all 1 bits set.
37612 (vcondu<mode><mode>): Likewise.
37613 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37615 (fpmask_comparison_operator): Update comment.
37616 (vecint_comparison_operator): New predicate.
37617 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37618 vector conditionals when the true and false values are constant
37619 vectors with all 0 bits or all 1 bits set.
37621 2017-02-06 Martin Sebor <msebor@redhat.com>
37623 PR tree-optimization/79376
37624 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37626 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37628 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37629 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37630 to simplify split condition.
37632 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37634 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37635 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37638 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37640 PR rtl-optimization/68664
37641 * target.def (can_speculate_insn): New hook.
37642 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37643 * doc/tm.texi: Regenerate.
37644 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37645 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37646 (rs6000_sched_can_speculate_insn): New function.
37648 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37650 PR tree-optimization/79284
37651 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37652 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37653 vectorizable_mask_load_store, vectorizable_operation,
37654 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37655 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37656 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37657 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37658 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37659 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37660 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37661 is_gimple_assign (stmt). Replace another such test with
37662 is_gimple_assign (stmt).
37664 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37667 * config/avr/avr.c (rtl-iter.h): Include it.
37668 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37669 (avr_legitimate_combined_insn): ...and implementation.
37671 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37673 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37674 * config/s390/s390.c (s390_const_operand_ok)
37675 (s390_canonicalize_comparison, s390_extract_part)
37676 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37677 (s390_contiguous_bitmask_p, s390_rtx_costs)
37678 (legitimize_pic_address): Likewise.
37679 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37680 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37681 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37682 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37683 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37685 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37687 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37688 REGNO($0) == REGNO($1).
37690 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37692 * config/s390/linux.h(SIZE_TYPE): Add comment.
37694 2017-02-06 Julian Brown <julian@codesourcery.com>
37695 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37696 Virendra Pathak <virendra.pathak@broadcom.com>
37698 * config/aarch64/aarch64-cores.def: Change the scheduler
37700 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37701 * config/aarch64/thunderx2t99.md: New file.
37703 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37705 * doc/standards.texi (Go Language): Update link to language
37708 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37710 * tree-eh.c (lower_resx): Sanitize profile.
37711 (cleanup_empty_eh_move_lp): Likewise.
37713 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37716 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37718 * cfgloopmanip.h (loop_version): Update prototype.
37719 * modulo-sched.c (sms_schedule): Update call of loop_version.
37720 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37721 * tree-parloops.c (gen_parallel_loop): Likewise.
37722 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37723 * tree-ssa-loop-split.c (split_loop): Likewise.
37724 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37725 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37727 2017-02-05 Martin Liska <mliska@suse.cz>
37730 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37732 (print_operand_address): Initialize a struct to zero.
37734 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37736 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37737 garbage collector only in textual form.
37739 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37741 * doc/extend.texi (x86 specific memory model extensions for
37742 transactional memory): Simplify a phrase.
37744 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37747 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37748 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37749 (atomic_storedi_1): Likewise.
37751 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37753 PR tree-optimization/79338
37754 * tree-parloops.c (gather_scalar_reductions): Don't call
37755 vect_analyze_loop_form for loop->inner before destroying loop's
37758 2017-02-03 Martin Sebor <msebor@redhat.com>
37760 PR tree-optimization/79327
37761 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37762 when precision has resulted in leading zeros.
37763 (format_integer): Adjust the likely counter to assume an unknown
37764 argument that may be zero is non-zero.
37766 2017-02-03 Jason Merrill <jason@redhat.com>
37769 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37770 avoid copying non-taken branch.
37772 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37774 PR tree-optimization/79340
37775 * tree-vect-loop.c (vectorizable_reduction): Release
37776 vec_defs elements after safe_splicing them into other vectors.
37779 PR tree-optimization/79327
37780 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37781 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37783 (format_integer): Use wide_int_to_tree instead of build_int_cst
37784 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37785 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37786 of shortest and longest sequence.
37788 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37790 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37791 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37793 2017-02-03 Walter Lee <walt@tilera.com>
37796 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37797 after initial stackframe link reg save.
37798 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37800 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37803 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37804 wu for stxssp alternative.
37806 2017-02-03 Martin Sebor <msebor@redhat.com>
37808 PR tree-optimization/79352
37809 * gimple-fold.c (get_range_strlen): Add argument.
37810 (get_range_strlen): Change return type to bool.
37811 (get_maxval_strlen): Pass in a dummy argument.
37812 * gimple-fold.h (get_range_strlen): Change return type to bool.
37813 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37814 * tree.h (array_at_struct_end_p): Add argument.
37815 * tree.c (array_at_struct_end_p): Handle it.
37817 2017-02-03 Martin Liska <mliska@suse.cz>
37820 * multiple_target.c (create_dispatcher_calls): Redirect edge
37821 from a caller of a dispatcher.
37822 (expand_target_clones): Make the clones local.
37823 (ipa_target_clone): Do both target clones and resolvers.
37824 (ipa_dispatcher_calls): Remove the pass.
37825 (pass_dispatcher_calls::gate): Likewise.
37826 (make_pass_dispatcher_calls): Likewise.
37827 * passes.def (pass_target_clone): Put as very first IPA early
37830 2017-02-03 Martin Liska <mliska@suse.cz>
37832 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37833 in case of a function with ifunc attribute.
37835 2017-02-03 Martin Liska <mliska@suse.cz>
37837 * cgraph.c (cgraph_node::dump): Dump function version info.
37838 * symtab.c (symtab_node::dump_base): Add missing new line.
37840 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37842 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37843 (ifcombine_ifandif): Use it.
37845 2017-02-03 Martin Liska <mliska@suse.cz>
37847 * doc/invoke.texi: Document default value for
37848 use-after-scope-direct-emission-threshold.
37850 2017-02-03 Martin Liska <mliska@suse.cz>
37852 PR tree-optimization/79339
37853 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37854 (format_floating): Likewise.
37856 2017-02-03 Martin Liska <mliska@suse.cz>
37859 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37861 (ipa_node_params_t::remove): Likewise.
37862 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37863 initialization from removed ipa_node_params_t::insert.
37864 (ipa_node_params::~ipa_node_params): Move from removed
37865 ipa_node_params_t::release.
37866 * symbol-summary.h (symbol_summary::m_released): New member.
37867 Do not release a summary twice. Do not allow to call finalizer
37868 for types of a summary that live in GGC memory.
37870 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37872 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37875 2017-02-02 Martin Sebor <msebor@redhat.com>
37877 PR middle-end/79275
37878 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37879 (format_string): Tighten up the range of output for non-constant
37880 strings and correct the expected range for wide non-constant strings.
37882 2017-02-02 Martin Sebor <msebor@redhat.com>
37884 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37886 PR middle-end/32003
37887 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37889 (-fdump-tree-@var): Add to index and document how to come up
37890 with pass-specific option and dump file names.
37891 (-fdump-passes): Clarify where to look for output.
37893 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37895 PR middle-end/77445
37896 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37897 statistics of the analyzed path; allow threading for speed when
37898 any of BBs along the path are optimized for speed.
37900 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37902 PR middle-end/78468
37903 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37904 settings of the virtual registers.
37907 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37909 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37910 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37913 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37915 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37916 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37918 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37920 * config/s390/s390.md: Add missing comments with the expanded
37922 * config/s390/vector.md: Likewise.
37923 * config/s390/vx-builtins.md: Likewise.
37925 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37928 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37929 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37930 conditions on a single line.
37932 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37934 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37935 __S390_VX__ to __VX__.
37937 2017-02-01 Andrew Pinski <apinski@cavium.com>
37939 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37940 stmt_info to record_stmt_cost.
37941 (vect_get_known_peeling_cost): Pass stmt_info if known to
37943 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37944 cpu_vector_cost field into
37945 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37946 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37947 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37948 splitting of scalar_stmt_cost and vec_stmt_cost.
37949 (thunderx_vector_cost): Likewise.
37950 (cortexa57_vector_cost): LIkewise.
37951 (exynosm1_vector_cost): Likewise.
37952 (xgene1_vector_cost): Likewise.
37953 (thunderx2t99_vector_cost): Improve after the splitting of the two
37955 (aarch64_builtin_vectorization_cost): Update for the splitting of
37956 scalar_stmt_cost and vec_stmt_cost.
37958 2017-02-01 Torvald Riegel <triegel@redhat.com>
37959 Richard Henderson <rth@redhat.com>
37961 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37962 conditional on existance of a fast atomic load.
37963 * optabs-query.c (can_atomic_load_p): New function.
37964 * optabs-query.h (can_atomic_load_p): Declare it.
37965 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37966 no fast atomic load is available for the particular size of access.
37967 (expand_atomic_compare_and_swap): Likewise.
37968 (expand_atomic_load): Likewise.
37969 (expand_atomic_store): Likewise.
37970 (expand_atomic_fetch_op): Likewise.
37971 * testsuite/lib/target-supports.exp
37972 (check_effective_target_sync_int_128): Remove x86 because it provides
37973 no fast atomic load.
37974 (check_effective_target_sync_int_128_runtime): Likewise.
37976 2017-02-01 Richard Biener <rguenther@suse.de>
37978 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37979 (graphite_transform_loops): Provide opt-info for optimized nests.
37980 * tree-parloop.c (parallelize_loops): Provide opt-info for
37981 parallelized loops.
37983 2017-02-01 Richard Biener <rguenther@suse.de>
37985 PR middle-end/79315
37986 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37987 was not set before.
37989 2017-02-01 Richard Biener <rguenther@suse.de>
37991 PR tree-optimization/71824
37992 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37993 Verify the loops are valid in the merged SESE region.
37994 (scop_detection::can_represent_loop_1): Check analyzing the
37995 evolution of the number of iterations in the region succeeds.
37997 2017-01-31 Ian Lance Taylor <iant@golang.org>
37999 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38000 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38002 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38004 PR preprocessor/79210
38005 * input.c (get_substring_ranges_for_loc): Replace line_width
38006 assertion with error-handling.
38008 2017-01-31 Richard Biener <rguenther@suse.de>
38010 PR tree-optimization/77318
38011 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38012 (create_pw_aff_from_tree): Take loop parameter.
38013 (add_condition_to_pbb): Pass loop of the condition to
38014 create_pw_aff_from_tree.
38016 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38018 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38019 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38021 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38025 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38027 (convert_int_to_float128): Likewise.
38028 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38029 (convert_int_to_float128): Likewise.
38030 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38031 (UNSPEC_IEEE128_CONVERT): Likewise.
38032 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38033 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38034 Use local variables for IBM extended format.
38035 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38036 (fix_trunc<mode>si2_fprs): Likewise.
38037 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38038 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38039 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38040 to know that we can now have integers of all sizes in vector
38042 (fix<uns>_<mode>di2_hw): Likewise.
38043 (float<uns>_<mode>si2_hw): Likewise.
38044 (fix_<mode>si2_hw): Likewise.
38045 (fixuns_<mode>si2_hw): Likewise.
38046 (float<uns>_<mode>di2_hw): Likewise.
38047 (float_<mode>di2_hw): Likewise.
38048 (float_<mode>si2_hw): Likewise.
38049 (floatuns_<mode>di2_hw): Likewise.
38050 (floatuns_<mode>si2_hw): Likewise.
38051 (xscvqp<su>wz_<mode>): Delete, no longer used.
38052 (xscvqp<su>dz_<mode>): Likewise.
38053 (xscv<su>dqp_<mode>): Likewise.
38054 (ieee128_mfvsrd_64bit): Likewise.
38055 (ieee128_mfvsrd_32bit): Likewise.
38056 (ieee128_mfvsrwz): Likewise.
38057 (ieee128_mtvsrw): Likewise.
38058 (ieee128_mtvsrd_64bit): Likewise.
38059 (ieee128_mtvsrd_32bit): Likewise.
38061 2017-01-31 Martin Liska <mliska@suse.cz>
38064 * ipa-prop.c (ipa_free_all_node_params): Call release method
38065 instead of ~sumbol_summary to not to trigger double times
38068 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38070 PR tree-optimization/71691
38071 * bitmap.h (class auto_bitmap): New.
38072 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38073 is_maybe_undefined instead of ssa_undefined_value_p.
38075 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38077 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38078 __S390_ARCH_LEVEL__ to __ARCH__.
38080 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38082 PR tree-optimization/79267
38083 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38084 if should_remove_lhs_p is true.
38086 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38089 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38090 (add_alignment_attribute): New.
38091 (base_type_die): Add alignment attribute.
38092 (subrange_type_die): Likewise.
38093 (modified_type_die): Likewise.
38094 (gen_array_type_die): Likewise.
38095 (gen_descr_array_type_die: Likewise.
38096 (gen_enumeration_type_die): Likewise.
38097 (gen_subprogram_die): Likewise.
38098 (gen_variable_die): Likewise.
38099 (gen_field_die): Likewise.
38100 (gen_ptr_to_mbr_type_die): Likewise.
38101 (gen_struct_or_union_type_die): Likewise.
38102 (gen_subroutine_type_die): Likewise.
38103 (gen_typedef_die): Likewise.
38104 (base_type_cmp): Compare alignment attribute.
38106 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38109 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38110 (setb_unsigned) New pattern for setb with CCUNS.
38111 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38112 subfc./subfe sequence to avoid overflow problems. Generate a
38113 shorter sequence with cmpld/setb for power9.
38114 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38115 for generating subfc. instruction.
38116 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38117 now uses this instruction.
38119 2017-01-30 Ian Lance Taylor <iant@google.com>
38122 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38123 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38125 2017-01-30 Martin Sebor <msebor@redhat.com>
38127 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38128 Move constant to the right of a relational operator.
38129 (get_mpfr_format_length, format_character, format_string): Ditto.
38130 (should_warn_p, maybe_warn): Same.
38132 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38134 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38137 * asan.c (get_translation_unit_decl): Remove function.
38138 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38140 2017-01-30 Martin Liska <mliska@suse.cz>
38142 PR gcov-profile/79259
38143 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38144 -fprofile-generate.
38146 2017-01-30 Martin Liska <mliska@suse.cz>
38149 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38150 Initialize variables with NULL value.
38152 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38155 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38157 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38159 2017-01-30 Richard Biener <rguenther@suse.de>
38161 PR tree-optimization/79276
38162 * tree-vrp.c (process_assert_insertions): Properly adjust common
38163 when removing a duplicate.
38165 2017-01-30 Richard Biener <rguenther@suse.de>
38167 PR tree-optimization/79256
38168 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38169 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38171 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38173 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38176 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38177 ("*r<noxa>sbg_<mode>_sll_bitmask")
38178 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38179 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38180 Use contiguous_bitmask_nowrap_operand.
38182 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38185 * config/rs6000/altivec.h (vec_xl): Revise #define.
38186 (vec_xst): Likewise.
38188 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38190 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38192 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38194 PR rtl-optimization/79194
38195 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38196 traps before call to bypass_conditional_jumps.
38198 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38200 PR tree-optimization/71374
38201 * lra-constraints.c (check_conflict_input_operands): New.
38202 (match_reload): Use it.
38204 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38207 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38208 account to calculate conflict_set.
38210 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38212 PR rtl-optimization/78559
38213 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38214 other_insn in combine.
38216 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38218 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38219 uint16_type_node for BT_UINT16.
38221 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38223 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38224 "RTL Tests" to menu.
38225 (GIMPLE Tests): New node.
38226 (RTL Tests): New node.
38228 2017-01-27 Richard Biener <rguenther@suse.de>
38230 PR tree-optimization/79245
38231 * tree-loop-distribution.c (distribute_loop): Apply cost
38232 modeling also to detected patterns.
38234 2017-01-27 Richard Biener <rguenther@suse.de>
38236 PR tree-optimization/71433
38237 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38238 (compare_assert_loc): New function.
38239 (process_assert_insertions): Sort and optimize assert locations
38240 to remove duplicates and push down identical assertions on
38241 edges to their destination block.
38243 2017-01-27 Richard Biener <rguenther@suse.de>
38245 PR tree-optimization/79244
38246 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38247 out SSA names even if abnormal.
38249 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38251 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38252 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38253 instead of MPFR_RNDN.
38255 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38258 * arm.c (arm_option_override): Don't call build_target_option_node
38259 until after doing all option overrides.
38260 (arm_valid_target_attribute_tree): Likewise.
38262 2017-01-27 Martin Liska <mliska@suse.cz>
38264 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38265 for {cd}tors and C++ {cd}tors.
38267 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38269 * config/s390/s390.md ("*setmem_long_and")
38270 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38272 2017-01-26 Martin Sebor <msebor@redhat.com>
38274 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38277 2017-01-26 Martin Sebor <msebor@redhat.com>
38279 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38280 HAVE_DFmode before using XFmode or DFmode.
38281 (parse_directive): Avoid using the z length modifier to avoid
38282 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38284 PR middle-end/78703
38285 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38286 to accept adjustment as an array.
38287 (get_int_range): New function.
38288 (struct directive): Make width and prec arrays.
38289 (directive::set_width, directive::set_precision): Call get_int_range.
38290 (format_integer, format_floating): Handle width and precision ranges.
38291 (format_string, parse_directive): Same.
38293 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38296 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38297 comdat_type_p set, just clone them, but keep the children in the
38301 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38302 which have direct callers with -fvar-tracking-assignments enabled
38304 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38305 inside of type units.
38307 2017-01-26 Martin Sebor <msebor@redhat.com>
38309 PR middle-end/78703
38310 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38312 (struct format_result): Replace number_chars, number_chars_min,
38313 and number_chars_max with a single member of struct result_range.
38315 (format_result::operator+=): Adjust.
38316 (struct fmtresult): Remove bounded. Handle likely and unlikely
38318 (fmtresult::adjust_for_width_or_precision): New function.
38319 (fmtresult:type_max_digits): New function.
38320 (bytes_remaining): Handle likely and unlikely counters.
38321 (min_bytes_remaining): Remove.
38322 (format_percent): Simplify.
38323 (format_integer, format_floating): Set likely and unlikely counters.
38324 (get_string_length, format_character, format_string): Same.
38325 (format_plain, should_warn_p): New function.
38326 (maybe_warn): Call should_warn_p. Update diagnostic messages
38327 and handle those for all directives, including plain strings.
38328 (format_directive): Handle likely and unlikely counters.
38329 Remove unnecessary quoting from diagnostics. Add an informational
38331 (add_bytes): Remove.
38332 (pass_sprintf_length::compute_format_length): Simplify.
38333 (try_substitute_return_value): Handle likely and unlikely counters.
38335 2017-01-26 Carl Love <cel@us.ibm.com>
38337 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38338 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38340 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38343 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38344 endianess for subregs into account.
38345 * lra-constraints.c (lra_constraints): Do risky transformations
38346 always on the first iteration.
38347 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38348 last_call_used_reg_set.
38349 (process_bb_lives): Define and use last_call_used_reg_set.
38350 * lra.c (lra): Always continue after lra_constraints on the first
38353 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38355 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38357 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38359 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38361 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38362 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38363 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38364 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38365 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38366 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38367 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38368 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38369 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38371 2017-01-26 Marek Polacek <polacek@redhat.com>
38374 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38375 for the third operand.
38377 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38379 PR middle-end/79236
38380 * omp-low.c (struct omp_context): Add simt_stmt field.
38381 (scan_omp_for): Return omp_context *.
38382 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38383 context to the _simt_ SIMD stmt.
38384 (lower_omp_for): For combined SIMD with sibling _simt_
38385 SIMD, make sure to use the same decls in _looptemp_
38386 clauses as in the sibling.
38388 2017-01-26 David Sherwood <david.sherwood@arm.com>
38390 PR middle-end/79212
38391 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38394 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38397 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38398 emit fld b; fld a; if possible.
38400 * brig-builtins.def: Update copyright years.
38401 * config/arm/arm_acle_builtins.def: Update copyright years.
38403 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38406 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38407 constraint instead of o for the stxsd instruction.
38409 2017-01-25 Carl Love <cel@us.ibm.com>
38411 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38412 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38414 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38416 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38418 2017-01-25 Richard Biener <rguenther@suse.de>
38420 PR tree-optimization/69264
38421 * target.def (vector_alignment_reachable): Improve documentation.
38422 * doc/tm.texi: Regenerate.
38423 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38425 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38426 earlier changes with respect to TYPE_USER_ALIGN.
38427 (vector_alignment_reachable_p): Likewise. Improve dumping.
38429 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38432 * config/arm/arm.md (xordi3): Force constant operand into a register
38435 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38437 * doc/invoke.texi (-fstore-merging): Correct default optimization
38438 levels at which it is enabled.
38439 (-O): Move -fstore-merging from list to...
38442 2017-01-25 Richard Biener <rguenther@suse.de>
38445 * omp-expand.c: Include debug.h.
38446 (expand_omp_taskreg): Make sure to generate early debug before
38447 outlining anything from a function.
38448 (expand_omp_target): Likewise.
38449 (grid_expand_target_grid_body): Likewise.
38451 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38454 * asan.c (get_translation_unit_decl): New function.
38455 (asan_add_global): Extract modules file name from globals
38456 TRANSLATION_UNIT_DECL name.
38458 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38461 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38462 for long calls with APCS frame and VFP.
38464 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38466 * cfg.c (original_copy_tables_initialized_p): New function.
38467 * cfg.h (original_copy_tables_initialized_p): New decl.
38468 * cfgrtl.c (relink_block_chain): Guard the call to
38469 free_original_copy_tables with a call to
38470 original_copy_tables_initialized_p.
38471 * cgraph.h (symtab_node::native_rtl_p): New decl.
38472 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38473 (symtab_node::needed_p): Don't assert for early assembly output
38474 for __RTL functions.
38475 (cgraph_node::finalize_function): Set "force_output" for __RTL
38477 (cgraph_node::analyze): Bail out early for __RTL functions.
38478 (analyze_functions): Update assertion to support __RTL functions.
38479 (cgraph_node::expand): Bail out early for __RTL functions.
38480 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38482 * function.h (struct function): Update comment for field
38484 * gimple-expr.c: Include "tree-pass.h".
38485 (gimple_has_body_p): Return false for __RTL functions.
38486 * Makefile.in (OBJS): Add run-rtl-passes.o.
38487 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38489 (gcc::pass_manager::get_clean_slate): New accessor.
38490 * passes.c: Include "insn-addr.h".
38491 (should_skip_pass_p): Add logging. Update logic for running
38492 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38493 property-provider override so it is only done for gimple passes.
38495 (skip_pass): New function.
38496 (execute_one_pass): Call skip_pass when skipping passes.
38497 * read-md.c (md_reader::read_char): Support filtering
38498 the input to a subset of line numbers.
38499 (md_reader::md_reader): Initialize fields
38500 m_first_line and m_last_line.
38501 (md_reader::read_file_fragment): New function.
38502 * read-md.h (md_reader::read_file_fragment): New decl.
38503 (md_reader::m_first_line): New field.
38504 (md_reader::m_last_line): New field.
38505 * read-rtl-function.c (function_reader::create_function): Only
38506 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38507 curr_properties. Set DECL_INITIAL to a dummy block.
38508 (read_rtl_function_body_from_file_range): New function.
38509 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38511 * run-rtl-passes.c: New file.
38512 * run-rtl-passes.h: New file.
38514 2017-01-24 Jeff Law <law@redhat.com>
38516 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38519 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38521 PR tree-optimization/79159
38522 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38523 (record_nonwrapping_iv): Improve boundary using above function if no
38524 value range information.
38526 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38527 Martin Jambor <mjambor@suse.cz>
38529 * brig-builtins.def: New file.
38530 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38531 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38532 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38533 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38534 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38535 * builtin-types.def (BT_INT8): New.
38536 (BT_INT16): Likewise.
38537 (BT_UINT8): Likewise.
38538 (BT_UINT16): Likewise.
38539 (BT_FN_ULONG): Likewise.
38540 (BT_FN_UINT_INT): Likewise.
38541 (BT_FN_UINT_ULONG): Likewise.
38542 (BT_FN_UINT_LONG): Likewise.
38543 (BT_FN_UINT_PTR): Likewise.
38544 (BT_FN_ULONG_PTR): Likewise.
38545 (BT_FN_INT8_FLOAT): Likewise.
38546 (BT_FN_INT16_FLOAT): Likewise.
38547 (BT_FN_UINT32_FLOAT): Likewise.
38548 (BT_FN_UINT16_FLOAT): Likewise.
38549 (BT_FN_UINT8_FLOAT): Likewise.
38550 (BT_FN_UINT64_FLOAT): Likewise.
38551 (BT_FN_UINT16_UINT32): Likewise.
38552 (BT_FN_UINT32_UINT16): Likewise.
38553 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38554 (BT_FN_INT_PTR_INT): Likewise.
38555 (BT_FN_UINT_PTR_UINT): Likewise.
38556 (BT_FN_LONG_PTR_LONG): Likewise.
38557 (BT_FN_ULONG_PTR_ULONG): Likewise.
38558 (BT_FN_VOID_UINT64_UINT64): Likewise.
38559 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38560 (BT_FN_INT8_INT8_INT8): Likewise.
38561 (BT_FN_INT16_INT16_INT16): Likewise.
38562 (BT_FN_INT_INT_INT): Likewise.
38563 (BT_FN_UINT_FLOAT_UINT): Likewise.
38564 (BT_FN_FLOAT_UINT_UINT): Likewise.
38565 (BT_FN_ULONG_UINT_UINT): Likewise.
38566 (BT_FN_ULONG_UINT_PTR): Likewise.
38567 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38568 (BT_FN_UINT_UINT_UINT): Likewise.
38569 (BT_FN_VOID_UINT_PTR): Likewise.
38570 (BT_FN_UINT_UINT_PTR: Likewise.
38571 (BT_FN_UINT32_UINT64_PTR): Likewise.
38572 (BT_FN_INT_INT_UINT_UINT): Likewise.
38573 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38574 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38575 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38576 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38577 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38578 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38579 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38580 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38581 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38582 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38583 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38584 * doc/frontends.texi: List BRIG FE.
38585 * doc/install.texi (Testing): Add BRIG tesring requirements.
38586 * doc/invoke.texi (Overall Options): Mention BRIG.
38587 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38589 2017-01-24 Richard Biener <rguenther@suse.de>
38591 PR translation/79208
38592 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38594 2017-01-24 Martin Jambor <mjambor@suse.cz>
38597 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38598 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38599 and known_contexts.
38601 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38603 PR middle-end/79123
38604 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38605 casts from signed to unsigned really don't have a range.
38607 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38609 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38610 GMP_RNDx for compatiblity.
38612 2017-01-24 Martin Liska <mliska@suse.cz>
38615 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38616 that would prevent us to call alloca with -1 as argument.
38618 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38620 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38621 Avoid -Wformat-security warning.
38623 2017-01-23 Andrew Pinski <apinski@cavium.com>
38625 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38628 2017-01-23 Martin Sebor <msebor@redhat.com>
38630 PR middle-end/78703
38631 * gimple-ssa-sprintf.c (warn_level): New global.
38632 (format_integer): Use it here and throughout the rest of the file.
38633 Use the same switch to compute sign as base.
38634 (maybe_warn): New function.
38635 (format_directive): Factor out warnings into maybe_warn.
38636 Add debugging output. Use warn_level.
38637 (add_bytes): Use warn_level.
38638 (pass_sprintf_length::compute_format_length): Add debugging output.
38639 (try_substitute_return_value): Same.
38640 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38642 PR middle-end/78703
38643 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38644 (struct fmtresult, format_integer, format_floating): Adjust.
38645 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38646 (get_string_length, format_string,format_directive): Same.
38647 (pass_sprintf_length::compute_format_length): Same.
38648 (try_substitute_return_value): Simplify slightly.
38650 PR middle-end/78703
38651 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38652 (fmtresult::operator+=): Outlined.
38653 (struct fmtresult): Add ctors.
38654 (struct conversion_spec): Rename...
38655 (struct directive): ...to this. Add and remove data members.
38656 (directive::set_width, directive::set_precision): New functions.
38657 (format_percent): Use fmtresult ctor.
38658 (get_width_and_precision): Remove.
38659 (format_integer): Make naming changes. Avoid computing width and
38661 (format_floating): Same. Adjust indentation.
38662 (format_character, format_none): New functions.
38663 (format_string): Moved character handling to format_character.
38664 (format_directive): Remove arguments, change return type.
38665 (parse_directive): New function.
38666 (pass_sprintf_length::compute_format_length): Move directive
38667 parsing to parse_directive.
38669 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38671 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38672 (assign_assembler_name_if_needed): ... this.
38673 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38674 (assign_assembler_name_if_needed): ... this.
38675 (free_lang_data_in_cgraph): Adjust callers.
38676 * cgraphunit.c (cgraph_node::analyze): Likewise.
38677 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38680 2017-01-23 Richard Biener <rguenther@suse.de>
38682 PR tree-optimization/79088
38683 PR tree-optimization/79188
38684 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38685 resetting loop bounds after last path deletion. Reset loop
38686 bounds of the target loop, make code match the comments.
38687 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38688 Make sure loops need no fixups.
38690 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38692 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38693 exponent support with double type for first argument.
38694 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38695 type returned by __builtin_vec_extract_sig,
38696 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38697 functions from "vector int" to "vector unsigned int" or from
38698 "vector long long int" to "vector unsigned long long int".
38699 Changed type returned by __builtin_vec_extract_exp,
38700 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38701 functions from "vector int" to "vector unsigned int" or from
38702 "vector long long int" to "vector unsigned long long int".
38703 Changed return type of __builtin_vec_test_data_class,
38704 __builtin_vec_test_data_class_sp, and
38705 __builtin_vec_test_data_class_dp from "vector int" to
38706 "vector bool int" or from "vector long long int" to "vector bool
38707 long long int" and changed second argument type from "unsigned
38708 int" to "int". Added new overloaded function forms "vector float
38709 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38710 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38711 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38712 double, vector unsigned long long int)" and "vector double
38713 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38714 long int)". Changed return type of
38715 __builtin_scalar_test_data_class and
38716 __builtin_scalar_test_data_class_sp and
38717 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38718 int" and changed second argument from "unsigned int" to "int".
38719 Changed type returned by __builtin_scalar_test_neg,
38720 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38721 from "int" to "bool int". Added new overloaded function form
38722 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38723 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38724 exponent double-precision with floating point first argument.
38725 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38726 documentation of scalar_test_data_class, scalar_test_neg,
38727 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38728 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38729 vec_test_data_class built-in functions to reflect refinements in
38730 their type signatures.
38732 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38734 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38736 (aarch64_elf_asm_destructor): Likewise.
38738 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38740 PR rtl-optimization/78634
38741 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38742 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38743 * ifcvt.c (noce_try_cmove): Add missing cost check.
38745 PR rtl-optimization/71724
38746 * combine.c (if_then_else_cond): Look for situations where it is
38747 beneficial to undo the work of one of the recursive calls.
38749 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38751 PR tree-optimization/70754
38752 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38753 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38754 combined stmt before it if not NULL.
38755 (combine_chains): Process refs reversely and compute dominance point
38758 2017-01-23 Martin Liska <mliska@suse.cz>
38760 PR tree-optimization/79196
38761 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38762 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38764 (strlen_optimize_stmt): Call the renamed function.
38766 2017-01-23 Michael Matz <matz@suse.de>
38768 PR tree-optimization/78384
38769 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38771 2017-01-23 Richard Biener <rguenther@suse.de>
38773 PR tree-optimization/79186
38774 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38775 both incoming edges before moving an assert.
38777 2017-01-23 Martin Jambor <mjambor@suse.cz>
38779 * ipa-prop.c (load_from_param_1): Removed.
38780 (load_from_unmodified_param): Bits from load_from_param_1 put back
38782 (load_from_param): Removed.
38783 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38784 with stmt. Reverted back to use of load_from_unmodified_param.
38786 2017-01-23 Martin Jambor <mjambor@suse.cz>
38789 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38790 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38791 field a pointer to garbage collected vector, mark lattices and
38792 ipcp_orig_node with GTY((skip)).
38793 (ipa_get_param_count): Adjust to descriptors being a pointer.
38794 (ipa_get_param): Likewise.
38795 (ipa_get_type): Likewise.
38796 (ipa_get_param_move_cost): Likewise.
38797 (ipa_set_param_used): Likewise.
38798 (ipa_get_controlled_uses): Likewise.
38799 (ipa_set_controlled_uses): Likewise.
38800 (ipa_is_param_used): Likewise.
38801 (ipa_node_params_t): Move into garbage collector. New methods insert
38803 (ipa_node_params_sum): Annotate wth GTY(()).
38804 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38806 (ipa_load_from_parm_agg): Adjust declaration.
38807 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38808 * ipa-profile.c (ipa_profile): Likewise.
38809 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38810 (ipa_populate_param_decls): Make descriptors parameter garbage
38812 (ipa_dump_param): Adjust to descriptors being a pointer.
38813 (ipa_alloc_node_params): Likewise.
38814 (ipa_initialize_node_params): Likewise.
38815 (load_from_param_1): Make descriptors parameter garbage collected.
38816 (load_from_unmodified_param): Likewise.
38817 (load_from_param): Likewise.
38818 (ipa_load_from_parm_agg): Likewise.
38819 (ipa_node_params::~ipa_node_params): Removed.
38820 (ipa_free_all_node_params): Remove call to delete operator.
38821 (ipa_node_params_t::insert): New.
38822 (ipa_node_params_t::remove): Likewise.
38823 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38824 copy known_csts and known_contexts vectors.
38825 (ipa_read_node_info): Adjust to descriptors being a pointer.
38826 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38828 (ipcp_transform_function): Make descriptors variable garbage
38831 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38833 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38834 * config/i386/avx512dqintrin.h: Ditto.
38835 * config/i386/avx512fintrin.h: Ditto.
38836 * config/i386/i386.c: Handle new builtins.
38837 * config/i386/i386-builtin.def: Add new builtins.
38838 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38839 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38841 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38842 Martin Liska <mliska@suse.cz>
38844 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38845 * asan.c (asan_expand_poison_ifn): Support stores and use
38846 appropriate ASAN report function.
38847 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38848 * internal-fn.def (ASAN_POISON_USE): Declare.
38849 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38850 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38851 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38852 ASAN_POISON calls w/o LHS.
38853 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38854 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38855 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38856 * gimplify.c (asan_poison_variables): Add attribute
38857 use_after_scope_memory to variables that really needs to live
38859 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38860 having the attribute.
38862 2017-01-23 Martin Liska <mliska@suse.cz>
38864 * asan.c (create_asan_shadow_var): New function.
38865 (asan_expand_poison_ifn): Likewise.
38866 * asan.h (asan_expand_poison_ifn): New declaration.
38867 * internal-fn.c (expand_ASAN_POISON): Likewise.
38868 * internal-fn.def (ASAN_POISON): New builtin.
38869 * sanopt.c (pass_sanopt::execute): Expand
38870 asan_expand_poison_ifn.
38871 * tree-inline.c (copy_decl_for_dup_finish): Make function
38873 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38874 * tree-ssa.c (is_asan_mark_p): New function.
38875 (execute_update_addresses_taken): Rewrite local variables
38876 (identified just by use-after-scope as addressable) into SSA.
38878 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38880 * doc/install.texi (Specific): opensource.apple.com uses https
38881 now. Remove trailing slash.
38883 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38885 * README.Portability: Remove note on an Irix compatibility issue.
38887 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38889 * gcov.c (INCLUDE_ALGORITHM): Define.
38890 (INCLUDE_VECTOR): Define.
38891 No longer include <vector> and <algorithm> directly.
38893 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38895 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38897 * doc/invoke.texi (Code Gen Options): Ditto.
38899 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38902 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38904 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38906 rtl-optimization/79125
38907 * cprop.c (local_cprop_pass): Handle cases where we make an
38908 unconditional trap.
38910 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38914 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38915 read from, for big endian.
38917 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38919 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38920 register pauth builtins for LP64 only.
38922 2017-01-20 Marek Polacek <polacek@redhat.com>
38925 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38928 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38930 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38932 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38933 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38934 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38936 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38939 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38940 in big-endian mode when they are not a single duplicated value.
38942 2017-01-20 Richard Biener <rguenther@suse.de>
38944 * BASE-VER: Bump to 7.0.1.
38946 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38948 * omp-low.c (omplow_simd_context): New struct. Use it...
38949 (lower_rec_simd_input_clauses): ...here and...
38950 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38951 references to idx, lane, max_vf, is_simt.
38953 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38955 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38958 2017-01-20 Martin Jambor <mjambor@suse.cz>
38960 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38961 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38963 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38964 (GTFILES): Rename hsa.c to hsa-common.c.
38965 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38966 * hsa-dump.c: Likewise.
38967 * hsa-gen.c: Likewise.
38968 * hsa-regalloc.c: Likewise.
38969 * ipa-hsa.c: Likewise.
38970 * omp-expand.c: Likewise.
38971 * omp-low.c: Likewise.
38972 * toplev.c: Likewise.
38974 2017-01-20 Marek Polacek <polacek@redhat.com>
38977 * doc/invoke.texi: Document -Wduplicated-branches.
38978 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38979 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38980 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38981 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38982 return 0 only when not OEP_LEXICOGRAPHIC.
38983 (fold_build_cleanup_point_expr): Use the expression
38984 location when building CLEANUP_POINT_EXPR.
38985 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38986 * tree.c (add_expr): Handle error_mark_node.
38988 2017-01-20 Martin Liska <mliska@suse.cz>
38991 * tree-profile.c (init_ic_make_global_vars): Do not call
38993 (gimple_init_gcov_profiler): Likewise.
38995 2017-01-20 Martin Liska <mliska@suse.cz>
38998 * cgraph.h (maybe_create_reference): Remove argument and
39000 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39002 * ipa-cp.c (create_specialized_node): Likewise.
39003 * symtab.c (symtab_node::maybe_create_reference): Handle
39004 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39006 2017-01-20 Martin Liska <mliska@suse.cz>
39008 * read-rtl-function.c (function_reader::create_function): Use
39009 build_decl instread of build_decl_stat.
39011 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39013 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39014 * config/i386/avx512dqintrin.h: Ditto.
39015 * config/i386/avx512fintrin.h: Ditto.
39016 * config/i386/i386-builtin-types.def: Add new types.
39017 * config/i386/i386.c: Handle new types.
39018 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39019 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39020 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39021 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39022 (__builtin_ia32_kshiftridi): New.
39023 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39025 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39029 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39030 define to rs6000_init_stack_protect_guard.
39031 (rs6000_init_stack_protect_guard): New function.
39033 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39034 Yunqiang Su <yunqiang.su@imgtec.com>
39036 * config.gcc (supported_defaults): Add madd4.
39037 (with_madd4): Add validation.
39038 (all_defaults): Add madd4.
39039 * config/mips/mips.opt (mmadd4): New option.
39040 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39042 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39044 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39045 (ISA_HAS_FUSED_MADD4): Likewise.
39046 * doc/invoke.texi (-mmadd4): Document the new option.
39047 * doc/install.texi (--with-madd4): Document the new option.
39049 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39051 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39052 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39053 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39054 (aarch64_init_pauth_hint_builtins): New.
39055 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39056 (aarch64_expand_builtin): Expand new builtins.
39058 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39060 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39061 * combine-stack-adj.c (no_unhandled_cfa): Handle
39062 REG_CFA_TOGGLE_RA_MANGLE.
39063 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39064 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39065 info for return address signing.
39066 (aarch64_expand_epilogue): Likewise.
39068 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39070 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39071 * config/aarch64/aarch64-protos.h
39072 (aarch64_return_address_signing_enabled): New declaration.
39073 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39075 (aarch64_expand_prologue): Sign return address before it's pushed onto
39077 (aarch64_expand_epilogue): Authenticate return address fetched from
39079 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39080 (aarch64_attributes): New function attributes for "sign-return-address".
39081 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39082 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39083 ("*do_return"): Generate combined instructions according to key index.
39084 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39085 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39087 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39088 * config/aarch64/aarch64.opt (msign-return-address=): New.
39089 * doc/extend.texi (AArch64 Function Attributes): Documents
39090 "sign-return-address=".
39091 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39093 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39095 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39096 overall option summary.
39098 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39100 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39101 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39102 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39103 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39105 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39107 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39108 -mpower9-minmax by default for -mcpu=power9.
39109 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39110 128-bit floating point.
39112 2017-01-20 Alan Modra <amodra@gmail.com>
39114 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39115 optimizing for size.
39117 2017-01-20 Alan Modra <amodra@gmail.com>
39120 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39121 for strcmp and strncmp from corresponding builtin decl.
39123 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39125 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39126 instead of i386/rtems-64.h.
39127 * config/i386/rtems-64.h: Remove.
39129 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39133 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39135 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39137 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39139 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39140 Change int to HOST_WIDE_INT.
39141 * config/aarch64/aarch64-protos.h
39142 (aarch64_simd_gen_const_vector_dup): Likewise.
39143 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39145 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39147 * langhooks-def.h (lhd_type_for_size): New decl.
39148 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39149 * langhooks.c (lhd_type_for_size): New function, taken from
39152 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39154 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39155 define_bypass for CR latency.
39156 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39157 (power9-alu2): Add define_bypass for CR latency.
39159 (power9-mul): Update insn latency.
39160 (power9-mul-compare): Update insn latency, bypass latency and remove
39163 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39165 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39167 * config/aarch64/aarch64.md
39168 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39169 aarch64_nopcrelative_literal_loads.
39170 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39172 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39174 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39175 TARGET_LOONGSON_3A.
39176 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39178 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39181 * config.gcc (supported_defaults): Add lxc1-sxc1.
39182 (with_lxc1_sxc1): Add validation.
39183 (all_defaults): Add lxc1-sxc1.
39184 * config/mips/mips.opt (mlxc1-sxc1): New option.
39185 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39187 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39188 __mips_no_lxc1_sxc1.
39189 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39190 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39191 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39193 2017-01-19 Richard Biener <rguenther@suse.de>
39195 PR tree-optimization/72488
39196 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39197 sure to restore SSA info.
39198 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39200 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39202 PR rtl-optimization/79121
39203 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39204 of the inner type when shifting an extended value.
39206 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39209 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39210 interposable aliases.
39212 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39215 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39216 Use the evmergelohi instruction.
39217 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39218 (mov_sitf_e500_subreg8_2_be): Likewise.
39219 (mov_sitf_e500_subreg12_2_le): Likewise.
39220 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39221 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39222 (mov_sitf_e500_subreg8_2_le): Likewise.
39223 (mov_sitf_e500_subreg12_2_be): Likewise.
39225 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39227 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39228 attribute from vecsimple to vecperm.
39229 (altivec_vbpermq2): Likewise.
39231 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39234 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39236 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39237 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39238 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39239 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39240 case where N arg is SIZE_MAX.
39241 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39242 (cmpstrsi): Add pattern.
39244 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39246 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39247 __builtin_vec_revb builtins.
39248 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39249 built-in functions to support generation of the ISA 3.0 XXBR<x>
39250 vector byte reverse instructions.
39251 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39252 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39253 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39254 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39255 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39256 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39257 (P9V_BUILTIN_VEC_REVB): Likewise.
39258 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39259 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39260 (p9_xxbrq_v16qi): Likewise.
39261 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39262 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39263 (p9_xxbrh_v8hi): Likewise.
39264 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39265 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39266 vec_revb built-in functions.
39268 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39270 PR rtl-optimization/78952
39271 * config/i386/i386.md (any_extract): New code iterator.
39272 (*insvqi_2): Use any_extract for source operand.
39273 (*insvqi_3): Use any_shiftrt for source operand.
39275 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39277 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39279 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39281 2017-01-18 Matthias Klose <doko@ubuntu.com>
39283 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39285 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39287 * config/rs6000/altivec.h (vec_bperm): Change #define.
39288 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39289 (altivec_vbpermq2): New define_insn.
39290 (altivec_vbpermd): Likewise.
39291 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39292 function interface.
39293 (VBPERMD): Likewise.
39294 (VBPERM): New polymorphic function interface.
39295 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39296 Add entries for P9V_BUILTIN_VEC_VBPERM.
39297 * doc/extend.texi: Add interfaces for vec_bperm.
39299 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39301 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39302 first letter of error messages.
39303 (s390_resolve_overloaded_builtin): Likewise.
39304 * config/s390/s390.c (s390_expand_builtin): Likewise.
39305 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39306 (s390_valid_target_attribute_inner_p): Likewise.
39307 * config/s390/s390.md ("tabort"): Likewise.
39309 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39311 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39312 (ISA_AVOID_DIV_HILO): New macro.
39313 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39314 (ISA_HAS_DDIV): Likewise.
39316 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39318 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39320 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39322 * doc/invoke.texi (fabi-version): Spelling fix.
39324 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39327 * doc/invoke.texi (fabi-version): Mention mangling fix for
39330 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39333 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39335 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39338 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39339 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39341 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39343 (stack_protect_test): Ditto.
39344 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39345 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39347 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39348 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39349 -mstack-protector-guard-offset=.
39350 (RS/6000 and PowerPC Options): Ditto.
39352 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39354 * config/i386/i386.h (MASK_CLASS_P): New define.
39355 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39356 there are no registers from different register sets also when
39357 mask registers are used. Update function comment.
39358 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39359 to (*k/*r) and (*k/*km) alternatives.
39361 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39363 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39364 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39365 (EH_RETURN_HANDLER_RTX): New define.
39366 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39367 Force frame pointer in EH return functions.
39368 (aarch64_expand_epilogue): Add barrier for eh_return.
39369 (aarch64_final_eh_return_addr): Remove.
39370 (aarch64_eh_return_handler_rtx): New function.
39371 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39373 (aarch64_eh_return_handler_rtx): New prototype.
39375 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39377 * config/rs6000/altivec.h (vec_rlmi): New #define.
39378 (vec_vrlnm): Likewise.
39379 (vec_rlnm): Likewise.
39380 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39381 (UNSPEC_VRLNM): Likewise.
39382 (VIlong): New mode iterator.
39383 (altivec_vrl<VI_char>mi): New define_insn.
39384 (altivec_vrl<VI_char>nm): Likewise.
39385 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39387 (VRLDNM): Likewise.
39388 (RLNM): New polymorphic function entry.
39389 (VRLWMI): New monomorphic function entry.
39390 (VRLDMI): Likewise.
39391 (RLMI): New polymorphic function entry.
39392 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39393 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39394 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39397 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39400 * dwarf2out.c (field_byte_offset): Restore the
39401 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39402 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39403 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39406 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39409 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39411 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39414 * dwarf2out.c (add_data_member_location_attribute): For constant
39415 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39416 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39417 DW_AT_byte_size attributes.
39419 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39421 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39422 after forcing to constant memory when the code model is medium.
39424 2017-01-17 Julia Koval <julia.koval@intel.com>
39427 * config/i386/avx512fintrin.h
39428 (_mm512_i32gather_ps): Change __addr type to void const*.
39429 (_mm512_mask_i32gather_ps): Ditto.
39430 (_mm512_i32gather_pd): Ditto.
39431 (_mm512_mask_i32gather_pd): Ditto.
39432 (_mm512_i64gather_ps): Ditto.
39433 (_mm512_mask_i64gather_ps): Ditto.
39434 (_mm512_i64gather_pd): Ditto.
39435 (_mm512_mask_i64gather_pd): Ditto.
39436 (_mm512_i32gather_epi32): Ditto.
39437 (_mm512_mask_i32gather_epi32): Ditto.
39438 (_mm512_i32gather_epi64): Ditto.
39439 (_mm512_mask_i32gather_epi64): Ditto.
39440 (_mm512_i64gather_epi32): Ditto.
39441 (_mm512_mask_i64gather_epi32): Ditto.
39442 (_mm512_i64gather_epi64): Ditto.
39443 (_mm512_mask_i64gather_epi64): Ditto.
39444 (_mm512_i32scatter_ps): Change __addr type to void*.
39445 (_mm512_mask_i32scatter_ps): Ditto.
39446 (_mm512_i32scatter_pd): Ditto.
39447 (_mm512_mask_i32scatter_pd): Ditto.
39448 (_mm512_i64scatter_ps): Ditto.
39449 (_mm512_mask_i64scatter_ps): Ditto.
39450 (_mm512_i64scatter_pd): Ditto.
39451 (_mm512_mask_i64scatter_pd): Ditto.
39452 (_mm512_i32scatter_epi32): Ditto.
39453 (_mm512_mask_i32scatter_epi32): Ditto.
39454 (_mm512_i32scatter_epi64): Ditto.
39455 (_mm512_mask_i32scatter_epi64): Ditto.
39456 (_mm512_i64scatter_epi32): Ditto.
39457 (_mm512_mask_i64scatter_epi32): Ditto.
39458 (_mm512_i64scatter_epi64): Ditto.
39459 (_mm512_mask_i64scatter_epi64): Ditto.
39460 * config/i386/avx512pfintrin.h
39461 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39462 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39463 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39464 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39465 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39466 (_mm512_prefetch_i32scatter_ps): Ditto.
39467 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39468 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39469 (_mm512_prefetch_i64scatter_pd): Ditto.
39470 (_mm512_prefetch_i64scatter_ps): Ditto.
39471 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39472 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39473 * config/i386/avx512vlintrin.h
39474 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39475 (_mm_mmask_i32gather_ps): Ditto.
39476 (_mm256_mmask_i32gather_pd): Ditto.
39477 (_mm_mmask_i32gather_pd): Ditto.
39478 (_mm256_mmask_i64gather_ps): Ditto.
39479 (_mm_mmask_i64gather_ps): Ditto.
39480 (_mm256_mmask_i64gather_pd): Ditto.
39481 (_mm_mmask_i64gather_pd): Ditto.
39482 (_mm256_mmask_i32gather_epi32): Ditto.
39483 (_mm_mmask_i32gather_epi32): Ditto.
39484 (_mm256_mmask_i32gather_epi64): Ditto.
39485 (_mm_mmask_i32gather_epi64): Ditto.
39486 (_mm256_mmask_i64gather_epi32): Ditto.
39487 (_mm_mmask_i64gather_epi32): Ditto.
39488 (_mm256_mmask_i64gather_epi64): Ditto.
39489 (_mm_mmask_i64gather_epi64): Ditto.
39490 (_mm256_i32scatter_ps): Change __addr type to void*.
39491 (_mm256_mask_i32scatter_ps): Ditto.
39492 (_mm_i32scatter_ps): Ditto.
39493 (_mm_mask_i32scatter_ps): Ditto.
39494 (_mm256_i32scatter_pd): Ditto.
39495 (_mm256_mask_i32scatter_pd): Ditto.
39496 (_mm_i32scatter_pd): Ditto.
39497 (_mm_mask_i32scatter_pd): Ditto.
39498 (_mm256_i64scatter_ps): Ditto.
39499 (_mm256_mask_i64scatter_ps): Ditto.
39500 (_mm_i64scatter_ps): Ditto.
39501 (_mm_mask_i64scatter_ps): Ditto.
39502 (_mm256_i64scatter_pd): Ditto.
39503 (_mm256_mask_i64scatter_pd): Ditto.
39504 (_mm_i64scatter_pd): Ditto.
39505 (_mm_mask_i64scatter_pd): Ditto.
39506 (_mm256_i32scatter_epi32): Ditto.
39507 (_mm256_mask_i32scatter_epi32): Ditto.
39508 (_mm_i32scatter_epi32): Ditto.
39509 (_mm_mask_i32scatter_epi32): Ditto.
39510 (_mm256_i32scatter_epi64): Ditto.
39511 (_mm256_mask_i32scatter_epi64): Ditto.
39512 (_mm_i32scatter_epi64): Ditto.
39513 (_mm_mask_i32scatter_epi64): Ditto.
39514 (_mm256_i64scatter_epi32): Ditto.
39515 (_mm256_mask_i64scatter_epi32): Ditto.
39516 (_mm_i64scatter_epi32): Ditto.
39517 (_mm_mask_i64scatter_epi32): Ditto.
39518 (_mm256_i64scatter_epi64): Ditto.
39519 (_mm256_mask_i64scatter_epi64): Ditto.
39520 (_mm_i64scatter_epi64): Ditto.
39521 (_mm_mask_i64scatter_epi64): Ditto.
39522 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39523 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39524 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39525 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39526 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39527 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39528 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39529 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39530 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39531 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39532 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39533 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39534 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39535 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39536 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39537 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39538 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39539 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39540 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39541 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39542 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39543 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39544 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39545 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39546 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39547 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39548 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39549 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39550 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39551 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39552 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39553 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39554 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39555 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39556 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39557 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39558 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39559 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39560 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39561 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39562 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39563 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39564 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39565 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39566 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39567 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39568 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39569 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39570 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39571 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39572 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39573 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39574 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39575 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39576 definitions accordingly.
39578 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39579 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39582 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39585 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39588 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39589 hard regs for inner regno.
39591 2017-01-17 Martin Liska <mliska@suse.cz>
39594 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39595 assumption and add comment.
39597 2017-01-17 Nathan Sidwell <nathan@acm.org>
39599 * ipa-visibility.c (localize_node): New function, broken out of ...
39600 (function_and_variable_visibility): ... here. Call it.
39602 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39604 PR middle-end/77445
39605 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39606 correctly set frequency of oudgoing edge.
39607 (duplicate_thread_path): Fix profile updating.
39609 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39612 * configure.ac: Add GCC_BASE_VER.
39613 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39614 version from BASE-VER file.
39615 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39616 (gcc.o): Depend on $(BASEVER).
39617 * common.opt (dumpfullversion): New option.
39618 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39619 * doc/invoke.texi: Document -dumpfullversion.
39620 * doc/install.texi: Document --with-gcc-major-version-only.
39621 * configure: Regenerated.
39623 2017-01-17 Richard Biener <rguenther@suse.de>
39625 PR tree-optimization/71433
39626 * tree-vrp.c (register_new_assert_for): Merge same asserts
39627 on all incoming edges.
39628 (process_assert_insertions_for): Handle insertions at the
39631 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39633 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39634 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39636 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39639 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39642 2017-01-17 Alan Modra <amodra@gmail.com>
39645 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39646 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39647 symbolic stack limit when pic.
39649 2017-01-16 Martin Sebor <msebor@redhat.com>
39651 PR tree-optimization/78608
39652 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39654 2017-01-16 Jeff Law <law@redhat.com>
39657 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39658 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39659 for several include directories that may be relative to sysroot.
39660 * config/i386/x-mingw32 (gplus_includedir): Define.
39661 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39662 (native_system_includedir): Likewise.
39663 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39664 override if TARGET_SYSTEM_ROOT is defined.
39665 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39667 PR tree-optimization/79090
39668 PR tree-optimization/33562
39669 PR tree-optimization/61912
39670 PR tree-optimization/77485
39671 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39672 and computed trims into the dump file.
39674 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39676 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39678 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39681 * gimplify.c (gimplify_init_constructor): If want_value and
39682 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39686 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39687 sequence. Formatting fixes.
39688 (doloop_optimize): Formatting fixes.
39691 * gcc.c (debug_level_greater_than_spec_func): New function.
39692 (static_spec_functions): Add debug-level-gt spec function.
39693 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39695 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39696 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39697 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39698 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39699 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39700 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39702 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39704 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39705 QImode fixups to general and mask registers only.
39707 2017-01-16 Carl Love <cel@us.ibm.com>
39709 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39710 for built-in functions
39711 vector signed char vec_nabs (vector signed char)
39712 vector signed short vec_nabs (vector signed short)
39713 vector signed int vec_nabs (vector signed int)
39714 vector signed long long vec_nabs (vector signed long long)
39715 vector float vec_nabs (vector float)
39716 vector double vec_nabs (vector double)
39717 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39719 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39720 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39721 * doc/extend.texi: Update the documentation file for the new built-in
39724 2017-01-16 Martin Sebor <msebor@redhat.com>
39726 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39729 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39731 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39732 UNSPEC_VSX__XXSPLTD to require special splat handling.
39734 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39737 * system.h: Poison strndup.
39739 2017-01-16 Alan Modra <amodra@gmail.com>
39742 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39745 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39747 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39749 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39751 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39752 call recog here. Assert that INSN_CODE (insn) is non-negative.
39754 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39757 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39759 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39760 in the currently scheduled RTL fragment.
39762 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39764 PR rtl-optimization/78751
39765 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39768 2017-01-14 Jeff Law <law@redhat.com>
39770 PR tree-optimization/79090
39771 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39772 variable length stores.
39773 (compute_trims): Delete dead assignment to *trim_tail.
39774 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39777 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39779 PR rtl-optimization/78626
39780 PR rtl-optimization/78727
39781 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39782 of a block, and split such blocks after everything else is finished.
39784 2017-01-14 Alan Modra <amodra@gmail.com>
39787 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39788 target legitimate_combined_insn.
39789 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39790 (rs6000_legitimate_combined_insn): New function.
39791 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39793 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39794 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39795 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39797 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39799 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39801 2017-01-13 Jeff Law <law@redhat.com>
39803 PR tree-optimization/33562
39804 PR tree-optimization/61912
39805 PR tree-optimization/77485
39806 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39808 (delete_dead_assignment): Likewise.
39809 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39810 statement to delete_dead_call and delete_dead_assignment.
39812 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39815 * substring-locations.c (format_warning_va): Strengthen case 1 so
39816 that both endpoints of the substring must be within the format
39817 range for just the substring to be printed.
39819 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39821 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39822 * config/i386/i386.c (ix86_target_string): Add missing options
39823 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39824 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39825 flags_other and ix86_target_other to flags2_other. Display unknown
39827 (ix86_valid_target_attribute_inner_p): Add missing options and
39828 reorder options by implied ISAs, as in ix86_target_string.
39830 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39832 * hash-table.h (hash_table::too_empty_p): New function.
39833 (hash_table::expand): Use it.
39834 (hash_table::traverse): Likewise.
39835 (hash_table::empty_slot): Use sizeof (value_type) instead of
39836 sizeof (PTR) to convert bytes to elements. Shrink the table
39837 if the current size is excessive for the current number of
39840 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39842 * ira-costs.c (record_reg_classes): Break from the inner loop
39843 early once alt_fail is known to be true. Update outer loop
39844 handling accordingly.
39846 2017-01-13 Jeff Law <law@redhat.com>
39848 * tree-ssa-dse.c (decrement_count): New function.
39849 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39850 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39851 when we know the partially dead statement is a mem* function.
39853 PR tree-optimization/61912
39854 PR tree-optimization/77485
39855 * tree-ssa-dse.c: Include expr.h.
39856 (maybe_trim_constructor_store): New function.
39857 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39859 PR tree-optimization/33562
39860 PR tree-optimization/61912
39861 PR tree-optimization/77485
39862 * doc/invoke.texi: Document new dse-max-object-size param.
39863 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39864 * tree-ssa-dse.c: Include params.h.
39865 (dse_store_status): New enum.
39866 (initialize_ao_ref_for_dse): New, partially extracted from
39868 (valid_ao_ref_for_dse, normalize_ref): New.
39869 (setup_live_bytes_from_ref, compute_trims): Likewise.
39870 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39871 (maybe_trim_partially_dead_store): Likewise.
39872 (maybe_trim_complex_store): Likewise.
39873 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39874 Track what bytes live from the original store. Return tri-state
39875 for dead, partially dead or live.
39876 (dse_dom_walker): Add constructor, destructor and new private members.
39877 (delete_dead_call, delete_dead_assignment): New extracted from
39879 (dse_optimize_stmt): Make a member of dse_dom_walker.
39880 Use initialize_ao_ref_for_dse.
39882 PR tree-optimization/33562
39883 PR tree-optimization/61912
39884 PR tree-optimization/77485
39885 * sbitmap.h (bitmap_count_bits): Prototype.
39886 (bitmap_clear_range, bitmap_set_range): Likewise.
39887 * sbitmap.c (bitmap_clear_range): New function.
39888 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39890 2017-01-13 Martin Liska <mliska@suse.cz>
39893 * function.c (set_cfun): Add new argument force.
39894 * function.h (set_cfun): Likewise.
39895 * ipa-inline-transform.c (inline_call): Use the function when
39896 strict alising from is dropped for function we inline to.
39898 2017-01-13 Richard Biener <rguenther@suse.de>
39900 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39901 for dumping GIMPLE INTEGER_CSTs.
39903 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39905 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39906 to 201112L since C++17.
39908 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39911 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39912 if -fsanitize=kernel-address is present.
39914 2017-01-13 Richard Biener <rguenther@suse.de>
39916 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39917 as _Literal ( type ) number in case usual suffixes do not
39918 preserve all information.
39920 2017-01-13 Richard Biener <rguenther@suse.de>
39922 PR tree-optimization/77283
39923 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39924 and ssa-iterators.h.
39925 (is_feasible_trace): Implement a cost model based on joiner
39928 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39931 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39932 char or short to __float128/_Float128 directly.
39934 2017-01-12 Martin Sebor <msebor@redhat.com>
39936 to -Wformat-overflow.
39937 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39938 (min_bytes_remaining): Same.
39939 (get_string_length): Same.
39940 (format_string): Same.
39941 (format_directive): Same.
39943 (pass_sprintf_length::handle_gimple_call): Same.
39945 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39947 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39948 info.nowrite calls with no lhs that can't throw. Return bool
39949 whether gsi_remove has been called or not.
39950 (pass_sprintf_length::handle_gimple_call): Return bool whether
39951 try_substitute_return_value called gsi_remove. Formatting fix.
39952 (pass_sprintf_length::execute): Don't use gsi_remove if
39953 handle_gimple_call returned true.
39956 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39957 be removed due to side-effects, don't remove following barrier nor
39958 turn the successor edge into fallthru edge.
39960 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39963 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39964 element-reversing loads and stores as not swappable.
39966 2017-01-12 Nathan Sidwell <nathan@acm.org>
39967 Nicolai Stange <nicstange@gmail.com>
39969 * combine.c (try_combine): Don't ignore result of overlap checking
39970 loop. Combine overlap & asm check into single loop.
39972 2017-01-12 Richard Biener <rguenther@suse.de>
39974 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39975 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39977 2017-01-12 Richard Biener <rguenther@suse.de>
39979 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39980 and TS_TARGET_OPTION directly derive from TS_BASE.
39981 * tree-core.h (tree_optimization_option): Derive from tree_base.
39982 (tree_target_option): Likewise.
39984 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39986 * config/i386/i386.c (memory_address_length): Increase len
39987 only when rip_relative_addr_p returns false.
39989 2017-01-11 Julia Koval <julia.koval@intel.com>
39991 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39992 (OPTION_MASK_ISA_SGX_SET): New.
39993 (ix86_handle_option): Handle OPT_msgx.
39994 * config.gcc: Added sgxintrin.h.
39995 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39996 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39997 * config/i386/i386.c (ix86_target_string): Add -msgx.
39999 (ix86_option_override_internal): Handle new options.
40000 (ix86_valid_target_attribute_inner_p): Add sgx.
40001 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40002 * config/i386/i386.opt: Add msgx.
40003 * config/i386/sgxintrin.h: New file.
40004 * config/i386/x86intrin.h: Add sgxintrin.h.
40006 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40009 * fold-const.c (maybe_nonzero_address): Return 1 for function
40011 (tree_single_nonzero_warnv_p): Don't handle function local objects
40015 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40018 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40021 * opts.c: Include "spellcheck.h"
40022 (struct string_fragment): New struct.
40023 (struct edit_distance_traits<const string_fragment &>): New
40025 (get_closest_sanitizer_option): New function.
40026 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40028 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40030 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40032 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40033 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40034 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40035 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40036 for initial die_offset if dwarf_split_debug_info.
40037 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40038 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40040 (output_skeleton_debug_sections): Formatting fix. Use
40041 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40042 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40044 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40046 * config/arm/cortex-a53.md: Add bypasses for
40047 cortex_a53_r2f_cvt.
40048 (cortex_a53_r2f): Only use for transfers.
40049 (cortex_a53_f2r): Likewise.
40050 (cortex_a53_r2f_cvt): Add reservation for conversions.
40051 (cortex_a53_f2r_cvt): Likewise.
40053 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40055 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40056 to all inlined functions, change static to extern.
40058 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40061 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40063 (arm_assemble_integer): Likewise.
40065 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40067 * config.gcc: Use new awk script to check CPU, FPU and architecture
40068 parameters for --with-... options.
40069 * config/arm/parsecpu.awk: New file
40070 * config/arm/arm-cpus.in: New file.
40071 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40073 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40075 * config/arm/t-arm: Update dependency rules.
40076 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40077 of processing .def files.
40078 * config/arm/genopt.sh: Deleted.
40079 * config/arm/gentune.sh: Deleted.
40080 * config/arm/arm-cores.def: Deleted.
40081 * config/arm/arm-arches.def: Deleted.
40082 * config/arm/arm-fpus.def: Deleted.
40083 * config/arm/arm-tune.md: Regenerated.
40084 * config/arm/arm-tables.opt: Regenerated.
40085 * config/arm/arm-cpu.h: New generated file.
40086 * config/arm/arm-cpu-data.h: New generated file.
40087 * config/arm/arm-cpu-cdata.h: New generated file.
40089 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40092 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40094 (input_varpool_node): Unpack dynamically_initialized bit.
40096 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40098 PR rtl-optimization/79032
40099 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40100 the alignment of the adjusted memory reference against that of MODE,
40101 instead of the alignment of the original memory reference.
40103 2017-01-11 Martin Jambor <mjambor@suse.cz>
40105 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40107 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40108 decorated functions.
40110 2017-01-11 Richard Biener <rguenther@suse.de>
40112 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40113 set range/nonnull info for PHI results. Do not set it on
40114 stmts marked for removal.
40116 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40118 * expr.c (store_field): In the bitfield case, fetch the return value
40119 from the registers before applying a single big-endian adjustment.
40120 Always do a final load for a BLKmode value not larger than a word.
40122 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40125 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40126 that we correctly handle column numbers greater than
40127 LINE_MAP_MAX_COLUMN_NUMBER.
40129 2017-01-10 Martin Sebor <msebor@redhat.com>
40131 PR middle-end/78245
40132 * gimple-ssa-sprintf.c (get_destination_size): Call
40133 {init,fini}object_sizes.
40134 * tree-object-size.c (addr_object_size): Adjust.
40135 (pass_through_call): Adjust.
40136 (pass_object_sizes::execute): Adjust.
40137 * tree-object-size.h (fini_object_sizes): Declare.
40139 2017-01-10 Martin Sebor <msebor@redhat.com>
40141 PR tree-optimization/78775
40142 * builtins.c (get_size_range): Move...
40143 * calls.c: ...to here.
40144 (alloc_max_size): Accept zero argument.
40145 (operand_signed_p): Remove.
40146 (maybe_warn_alloc_args_overflow): Call get_size_range.
40147 * calls.h (get_size_range): Declare.
40149 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40151 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40152 from TI's devices.csv file as of September 2016.
40153 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40155 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40157 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40158 * doc/invoke.texi: Likewise.
40159 * doc/md.texi: Likewise.
40160 * doc/objc.texi: Likewise.
40162 2017-01-10 Joshua Conner <joshconner@google.com>
40164 * config/arm/fuchsia-elf.h: New file.
40165 * config/fuchsia.h: New file.
40166 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40167 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40169 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40171 2016-01-10 Richard Biener <rguenther@suse.de>
40173 PR tree-optimization/79034
40174 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40175 Propagate out degenerate PHIs in the joiner.
40177 2017-01-10 Martin Liska <mliska@suse.cz>
40179 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40180 (sort_congruence_classes_by_decl_uid): Likewise.
40181 (sort_congruence_class_groups_by_decl_uid): Likewise.
40182 (sem_item_optimizer::merge_classes): Sort class, groups in these
40183 classes and members in the groups by DECL_UID of declarations.
40184 This would make merge operations stable.
40186 2017-01-10 Martin Liska <mliska@suse.cz>
40188 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40189 usage of m_classes_vec.
40190 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40191 (sem_item_optimizer::get_group_by_hash): Likewise.
40192 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40193 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40194 (sem_item_optimizer::verify_classes): Likewise.
40195 (sem_item_optimizer::process_cong_reduction): Likewise.
40196 (sem_item_optimizer::dump_cong_classes): Likewise.
40197 (sem_item_optimizer::merge_classes): Likewise.
40198 * ipa-icf.h (congruence_class_hash): Rename from
40199 congruence_class_group_hash. Remove declaration of m_classes_vec.
40201 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40203 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40204 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40205 * config.gcc: Add avx512vpopcntdqintrin.h.
40206 * config/i386/avx512vpopcntdqintrin.h: New.
40207 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40208 * config/i386/i386-builtin-types.def: Add new types.
40209 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40210 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40211 __builtin_ia32_vpopcountq_v8di_mask): New.
40212 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40213 __AVX512VPOPCNTDQ__.
40214 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40215 (PTA_AVX512VPOPCNTDQ): Define.
40216 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40217 TARGET_AVX512VPOPCNTDQ_P): Define.
40218 * config/i386/i386.opt: Add mavx512vpopcntdq.
40219 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40220 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40222 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40224 PR middle-end/77484
40225 * predict.def (PRED_CALL): Set to 67.
40227 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40229 * expr.c (store_field): In the bitfield case, if the value comes from
40230 a function call and is of an aggregate type returned in registers, do
40231 not modify the field mode; extract the value in all cases if the mode
40232 is BLKmode and the size is not larger than a word.
40234 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40237 * config/i386/cpuid.h: Fix undefined behavior.
40239 2017-01-04 Jeff Law <law@redhat.com>
40241 PR tree-optimization/79007
40242 PR tree-optimization/67955
40243 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40244 conservative for pt.null when flag_non_call_exceptions is on.
40246 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40248 PR translation/79019
40249 PR translation/79020
40250 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40251 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40252 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40254 * config/avr/avr.opt (maccumulate-args): Likewise.
40255 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40256 * common.opt (freport-bug): Likewise.
40257 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40258 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40259 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40260 translatable string.
40261 * config/i386/i386.c (function_value_32): Likewise.
40262 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40263 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40265 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40266 * common/config/msp430/msp430-common.c (msp430_handle_option):
40268 * symtab.c (symtab_node::verify_base): Likewise.
40269 * opts.c (set_debug_level): Likewise.
40270 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40271 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40272 missing whitespace to translatable strings.
40273 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40274 * config/sh/superh.h: Likewise.
40275 * config/i386/xopintrin.h: Likewise.
40276 * config/i386/znver1.md: Likewise.
40277 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40278 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40279 * double-int.h (struct double_int): Likewise.
40280 * double-int.c (div_and_round_double): Likewise.
40281 * wide-int.cc: Likewise.
40282 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40283 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40284 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40285 (crossjumps_occurred): ... this.
40286 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40289 PR tree-optimization/78899
40290 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40291 returning bool return struct loop *, NULL for failure and the new
40293 (versionable_outer_loop_p): Don't version outer loop if it has
40294 dont_vectorized bit set.
40295 (tree_if_conversion): When versioning outer loop, ensure
40296 tree_if_conversion is performed also on the inner loop of the
40297 non-vectorizable outer loop copy.
40298 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40299 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40300 prevent vectorization of it.
40301 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40302 the outer loop vectorization of the non-scalar version is attempted
40303 before vectorization of the inner loop in scalar version. If
40304 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40305 vectorization of its inner loop.
40306 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40307 has 2 inner loops, rename also on edges from bb whose single pred
40308 is outer_loop->header. Fix typo in function comment.
40310 2017-01-09 Martin Sebor <msebor@redhat.com>
40313 * asan.c (asan_emit_stack_protection): Increase local buffer size
40314 to avoid snprintf truncation warning.
40316 2017-01-09 Andrew Pinski <apinski@cavium.com>
40318 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40319 to reference thunderx2t99 for the tuning structure
40320 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40322 (thunderx2t99_extra_costs): This.
40323 * config/aarch64/aarch64-tune.md: Regenerate.
40324 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40325 (vulcan_addrcost_table): This.
40326 (vulcan_regmove_cost): Rename to ...
40327 (thunderx2t99_regmove_cost): This.
40328 (vulcan_vector_cost): Rename to ...
40329 (thunderx2t99_vector_cost): this.
40330 (vulcan_branch_cost): Rename to ...
40331 (thunderx2t99_branch_cost): This.
40332 (vulcan_tunings): Rename to ...
40333 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40334 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40336 2017-01-09 Martin Jambor <mjambor@suse.cz>
40340 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40341 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40342 (propagate_vr_accross_jump_function): Use the above function for all
40343 value range computations for pass-through jump functions and type
40344 converasion from explicit value range values.
40345 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40346 parameters from TYPE_ARG_TYPES.
40347 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40348 (ipa_write_node_info): Stream type of the actual argument.
40349 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40351 2017-01-09 Martin Liska <mliska@suse.cz>
40354 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40355 (lookup_compiler): Do not show error message with have_E.
40357 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40359 PR tree-optimization/78938
40360 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40361 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40362 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40363 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40366 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40368 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40371 2017-01-09 Richard Biener <rguenther@suse.de>
40373 PR tree-optimization/78997
40374 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40375 name condition properly.
40377 2017-01-09 Richard Biener <rguenther@suse.de>
40380 * dwarf2out.c (is_cxx): New overload with context.
40381 (is_naming_typedef_decl): Use it.
40383 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40385 * invoke.texi (Option Summary): Correct spacing in option lists
40386 and add line breaks to fix over-long lines.
40388 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40390 PR middle-end/17660
40392 * extend.texi (Common Variable Attributes): Add xref to GCC
40393 Internals manual to explain mode attribute keywords.
40395 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40398 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40399 and Preprocessor Options.
40400 (Options for Linking): Document -pthread here....
40401 (RS/6000 and PowerPC Options): ...not here.
40402 (Solaris 2 Options): ...or here.
40403 * doc/cppopts.texi: Document -pthread.
40405 2017-01-08 Martin Sebor <msebor@redhat.com>
40407 PR middle-end/77708
40408 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40409 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40410 New member functions.
40411 (format_directive): Used them.
40413 (pass_sprintf_length::handle_gimple_call): Same.
40414 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40415 to avoid truncation for any argument.
40416 (extract_affine_mul): Same.
40417 * tree.c (get_file_function_name): Same.
40419 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40421 PR middle-end/77484
40422 * predict.def (PRED_INDIR_CALL): Set to 86.
40424 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40426 PR preprocessor/54124
40427 * doc/cppopts.texi: Reformat -d subtable to list the full name
40428 of the options. Add cross-reference to the docs for the general
40429 compiler -d options.
40430 * doc/invoke.texi (Developer Options): Add cross-reference to the
40431 preprocessor-specific -d option documentation.
40433 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40435 PR preprocessor/13498
40436 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40437 redudant material, and reflect new command-line options.
40438 (System Headers): Likewise.
40440 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40442 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40443 -isystem, and -idirafter. Copy-edit.
40444 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40445 default for -ftrack-macro-expansion. Delete obsolete and
40446 badly-formatted implementation details about -fdebug-cpp output.
40447 * doc/cppwarnopts.texi: Copy-edit.
40449 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40452 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40453 that the transition from a max line width >= 1<<10 to narrower
40454 lines works correctly.
40456 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40458 * doc/options.texi (PerFunction): New.
40459 * opt-functions.awk (switch_flags): Map both Optimization and
40460 PerFunction to CL_OPTIMIZATION.
40461 * opth-gen.awk: Test for PerFunction flag along with
40463 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40464 it only when the latter is present. Skip those that don't in
40465 the hash function generator.
40466 * common.opt (fvar-tracking): Mark as PerFunction instead of
40468 (fvar-tracking-assignments): Likewise.
40469 (fvar-tracking-assignments-toggle): Likewise.
40470 (fvar-tracking-uninit): Likewise.
40472 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40474 PR translation/79018
40475 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40478 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40481 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40482 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40483 TARGET_LONG_JUMP_TABLE_OFFSETS.
40484 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40485 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40486 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40487 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40488 * config/m68k/m68k.md (tablejump expander): Likewise.
40489 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40490 TARGET_LONG_JUMP_TABLE_OFFSETS.
40491 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40492 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40494 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40495 David Holsgrove <david.holsgrove@xilinx.com>
40497 * common/config/microblaze/microblaze-common.c
40498 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40499 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40501 * config/microblaze/microblaze.c (microblaze_must_save_register)
40502 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40504 (microblaze_eh_return): New function.
40505 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40506 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40507 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40508 * config/microblaze/microblaze.md (eh_return): New pattern.
40510 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40512 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40513 GCC_DIAGNOSTIC_STRINGIFY): Define.
40515 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40517 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40519 * config/arm/arm.md (<mcrr>): New.
40521 * config/arm/arm.c (arm_arch5te): New.
40522 (arm_option_override): Set arm_arch5te.
40523 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40525 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40526 (arm_mcrr_qualifiers): ... this. New.
40527 (MRRC_QUALIFIERS): Define to...
40528 (arm_mrrc_qualifiers): ... this. New.
40529 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40531 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40532 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40533 (MRRCI, mrrc, MRRC): New.
40534 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40535 VUNSPEC_MRRC2): New.
40537 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40539 * config/arm/arm.md (<mcr>): New.
40541 * config/arm/arm.c (arm_coproc_builtin_available): Add
40542 support for mcr, mrc, mcr2 and mrc2.
40543 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40544 (arm_mcr_qualifiers): ... this. New.
40545 (MRC_QUALIFIERS): Define to ...
40546 (arm_mrc_qualifiers): ... this. New.
40547 (MCR_QUALIFIERS): Define to ...
40548 (arm_mcr_qualifiers): ... this. New.
40549 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40551 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40552 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40553 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40554 VUNSPEC_MRC2): New.
40556 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40558 * config/arm/arm.md (*ldc): New.
40562 * config/arm/arm.c (arm_coproc_builtin_available): Add
40563 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40564 (arm_coproc_ldc_stc_legitimate_address): New.
40565 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40566 'qualifier_const_pointer'.
40567 (LDC_QUALIFIERS): Define to...
40568 (arm_ldc_qualifiers): ... this. New.
40569 (STC_QUALIFIERS): Define to...
40570 (arm_stc_qualifiers): ... this. New.
40571 * config/arm/arm-protos.h
40572 (arm_coproc_ldc_stc_legitimate_address): New.
40573 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40574 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40575 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40576 stc2, stcl, stc2l): New.
40577 * config/arm/constraints.md (Uz): New.
40578 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40579 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40580 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40581 VUNSPEC_STC2L): New.
40583 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40585 * config/arm/arm.md (<cdp>): New.
40586 * config/arm/arm.c (neon_const_bounds): Rename this ...
40587 (arm_const_bounds): ... this.
40588 (arm_coproc_builtin_available): New.
40589 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40590 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40591 (CDP_QUALIFIERS): Define to...
40592 (arm_cdp_qualifiers): ... this. New.
40594 (arm_expand_builtin_args): Add case for 6 arguments.
40595 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40596 (arm_const_bounds): ... this.
40597 (arm_coproc_builtin_available): New.
40598 * config/arm/arm_acle.h (__arm_cdp): New.
40600 * config/arm/arm_acle_builtins.def (cdp): New.
40602 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40603 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40604 'arm_const_bounds'.
40605 * config/arm/types.md (coproc): New.
40606 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40607 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40608 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40609 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40611 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40613 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40614 (UBINOP_QUALIFIERS): New.
40616 (acle_builtin_data): New. Change comment.
40617 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40618 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40619 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40620 arm_acle_builtins.def.
40621 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40622 (arm_init_acle_builtins): New.
40623 (CRC32_BUILTIN): Remove.
40624 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40625 crc32cb, crc32ch and crc32cw.
40626 (arm_init_crc32_builtins): Remove.
40627 (arm_init_builtins): Use arm_init_acle_builtins rather
40628 than arm_init_crc32_builtins.
40629 (arm_expand_acle_builtin): New.
40630 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40631 * config/arm/arm_acle_builtins.def: New.
40633 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40635 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40636 (arm_builtin_datum): ... this.
40637 (arm_init_neon_builtin): Rename to ...
40638 (arm_init_builtin): ... this. Add a new parameters PREFIX
40639 and USE_SIG_IN_NAME.
40640 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40641 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40642 'arm_builtin_datum'.
40643 (arm_init_vfp_builtins): Likewise.
40644 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40645 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40646 (arm_expand_neon_args): Rename to ...
40647 (arm_expand_builtin_args): ... this. Rename builtin_arg
40648 enum values and differentiate between ARG_BUILTIN_MEMORY
40649 and ARG_BUILTIN_NEON_MEMORY.
40650 (arm_expand_neon_builtin_1): Rename to ...
40651 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40652 values, arm_expand_builtin_args and add bool parameter NEON.
40653 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40654 (arm_expand_vfp_builtin): Likewise.
40655 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40657 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40659 PR middle-end/77484
40660 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40661 * predict.c (tree_estimate_probability_bb): Reverse direction of
40662 polymorphic call predictor.
40664 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40666 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40667 (determine_pass_name_match): ...this new function and...
40668 (should_skip_pass_p): ...this new function.
40670 2017-01-06 Nathan Sidwell <nathan@acm.org>
40672 * ipa-visibility.c (function_and_variable_visibility): Reformat
40673 comments and long lines. Remove extrneous if.
40674 * symtab.c (symtab_node::make_decl_local): Fix code format.
40675 (symtab_node::set_section_for_node): Fix comment typo.
40677 2017-01-06 Martin Liska <mliska@suse.cz>
40680 * lra-constraints.c: Rename invariant to lra_invariant.
40681 * predict.c (set_even_probabilities): Initialize e to NULL.
40683 2017-01-05 Martin Sebor <msebor@redhat.com>
40685 PR tree-optimization/78910
40686 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40687 (format_integer): Correct off-by-one error in the handling
40688 of precision with negative numbers in signed conversions..
40690 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40692 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40694 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40696 PR tree-optimization/71016
40697 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40698 factor_out_conditional_conversion. Formatting fix.
40699 (factor_out_conditional_conversion): Add cond_stmt argument.
40700 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40701 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40704 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40706 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40707 read-rtl-function.o, and selftest-rtl.o.
40708 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40709 (selftest::aarch64_test_loading_full_dump): New function.
40710 (selftest::aarch64_run_selftests): New function.
40711 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40712 selftest::aarch64_run_selftests.
40713 * config/i386/i386.c
40714 (selftest::ix86_test_loading_dump_fragment_1): New function.
40715 (selftest::ix86_test_loading_call_insn): New function.
40716 (selftest::ix86_test_loading_full_dump): New function.
40717 (selftest::ix86_test_loading_unspec): New function.
40718 (selftest::ix86_run_selftests): Call the new functions.
40719 * emit-rtl.c (maybe_set_max_label_num): New function.
40720 * emit-rtl.h (maybe_set_max_label_num): New decl.
40721 * function.c (instantiate_decls): Guard call to
40722 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40723 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40725 * gensupport.c (gen_reader::gen_reader): Pass "false"
40726 for new "compact" param of rtx_reader.
40727 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40728 rather than an empty string for NULL strings.
40729 * read-md.c: Potentially include config.h rather than bconfig.h.
40730 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40731 (have_error): New global, copied from errors.c.
40732 (md_reader::read_name): Rename to...
40733 (md_reader::read_name_1): ...this, adding "out_loc" param,
40734 and converting "missing name or number" to returning false, rather
40736 (md_reader::read_name): Reimplement in terms of read_name_1.
40737 (md_reader::read_name_or_nil): New function.
40738 (md_reader::read_string): Handle "(nil)" by returning NULL.
40739 (md_reader::md_reader): Add new param "compact".
40740 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40741 (md_reader::read_file): New method.
40742 * read-md.h (md_reader::md_reader): Add new param "compact".
40743 (md_reader::read_file): New method.
40744 (md_reader::is_compact): New accessor.
40745 (md_reader::read_name): Convert return type from void to file_location.
40746 (md_reader::read_name_or_nil): New decl.
40747 (md_reader::read_name_1): New decl.
40748 (md_reader::m_compact): New field.
40749 (noop_reader::noop_reader): Pass "false" for new "compact" param
40751 (rtx_reader::rtx_reader): Add new "compact" param.
40752 (rtx_reader::read_rtx_operand): Make virtual and convert return
40753 type from void to rtx.
40754 (rtx_reader::read_until): New decl.
40755 (rtx_reader::handle_any_trailing_information): New virtual function.
40756 (rtx_reader::postprocess): New virtual function.
40757 (rtx_reader::finalize_string): New virtual function.
40758 (rtx_reader::m_in_call_function_usage): New field.
40759 (rtx_reader::m_reuse_rtx_by_id): New field.
40760 * read-rtl-function.c: New file.
40761 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40762 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40763 (selftest::verify_three_block_rtl_cfg): New decl.
40764 * read-rtl-function.h: New file.
40765 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40766 For host, include function.h, memmodel.h, and emit-rtl.h.
40767 (one_time_initialization): New function.
40768 (struct compact_insn_name): New struct.
40769 (compact_insn_names): New array.
40770 (find_code): Handle insn codes in compact dumps.
40771 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40772 (bind_subst_iter_and_attr): Likewise.
40773 (add_condition_to_string): Likewise.
40774 (add_condition_to_rtx): Likewise.
40775 (apply_attribute_uses): Likewise.
40776 (add_current_iterators): Likewise.
40777 (apply_iterators): Likewise.
40778 (initialize_iterators): Guard usage of apply_subst_iterator with
40779 #ifdef GENERATOR_FILE.
40780 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40781 (md_reader::read_mapping): Likewise.
40782 (add_define_attr_for_define_subst): Likewise.
40783 (add_define_subst_attr): Likewise.
40784 (read_subst_mapping): Likewise.
40785 (check_code_iterator): Likewise.
40786 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40788 (one_time_initialization): New function.
40789 (rtx_reader::read_until): New method.
40790 (read_flags): New function.
40791 (parse_reg_note_name): New function.
40792 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40793 Handle reuse_rtx ids.
40794 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40795 Add parsing support for RTL dumps, mirroring the special-cases in
40796 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40797 values, and calling handle_any_trailing_information.
40798 (rtx_reader::read_rtx_operand): Convert return type from void
40799 to rtx, returning return_rtx. Handle case 'e'. Call
40800 finalize_string on XSTR and XTMPL fields.
40801 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40802 "(nil)" values were omitted. Call the postprocess vfunc on the
40804 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40805 class ctor. Initialize m_in_call_function_usage. Call
40806 one_time_initialization.
40807 * rtl-tests.c (selftest::test_uncond_jump): Call
40808 set_new_first_and_last_insn.
40809 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40810 * selftest-rtl.c: New file.
40811 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40812 (selftest::get_insn_by_uid): New decl.
40813 * selftest-run-tests.c (selftest::run_tests): Call
40814 read_rtl_function_c_tests.
40815 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40816 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40819 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40821 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40822 operands in a special way. Assert that pos+len <= mode precision.
40824 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40826 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40827 3 argument Alias with unlimited for the negative form.
40828 (fno-vect-cost-model): Removed.
40830 2017-01-05 Martin Liska <mliska@suse.cz>
40832 * hsa-gen.c (gen_hsa_divmod): New function.
40833 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40835 2017-01-05 Martin Liska <mliska@suse.cz>
40838 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40841 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40843 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40844 small constant length operands.
40846 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40848 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40849 between loop iterations.
40851 2017-01-05 Martin Liska <mliska@suse.cz>
40854 * gimplify.c (gimplify_decl_expr): Compare to
40855 asan_poisoned_variables instread of checking flags.
40856 (gimplify_target_expr): Likewise.
40857 (gimplify_expr): Likewise.
40858 (gimplify_function_tree): Conditionally initialize
40859 asan_poisoned_variables.
40861 2017-01-04 Jeff Law <law@redhat.com>
40863 PR tree-optimizatin/78812
40864 * rtl.h (contains_mem_rtx_p): Prototype.
40865 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40866 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40867 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40868 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40869 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40871 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40873 * input.c (assert_char_at_range): Default-initialize actual_range.
40875 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40877 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40878 to match the caller.
40880 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40882 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40883 insns after final jump in test to emit dummy move.
40885 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40887 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40888 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40890 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40892 * multiple_target.c (create_dispatcher_calls): Init e_next.
40893 * tree-ssa-loop-split.c (split_loop): Init border.
40894 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40897 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40902 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40903 (altivec_register_operand): Do not return true if the operand
40904 contains a SUBREG mixing SImode and SFmode.
40905 (vsx_register_operand): Likewise.
40906 (vsx_reg_sfsubreg_ok): New predicate.
40907 (vfloat_operand): Do not return true if the operand contains a
40908 SUBREG mixing SImode and SFmode.
40909 (vint_operand): Likewise.
40910 (vlogical_operand): Likewise.
40911 (gpc_reg_operand): Likewise.
40912 (int_reg_operand): Likewise.
40913 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40914 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40915 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40917 (rs6000_emit_move_si_sf_subreg): New helper function.
40918 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40919 fixup SUBREGs involving SImode and SFmode.
40920 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40921 numbers for the new peephole2 optimization.
40922 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40923 the GLIBC math library that do AND/IOR/XOR operations on single
40924 precision floating point.
40925 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40926 target macros to say whether we need to avoid SUBREGs mixing
40928 (TARGET_ALLOW_SF_SUBREG): Likewise.
40929 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40930 (UNSPEC_SI_FROM_SF): Likewise.
40931 (iorxor): Change spacing.
40932 (and_ior_xor): New iterator for AND, IOR, and XOR.
40933 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40934 (movdi_from_sf_zero_ext): Likewise.
40935 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40936 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40937 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40938 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40939 (fms<mode>4): Likewise.
40940 (fnma<mode>4): Likewise.
40941 (fnms<mode>4): Likewise.
40942 (nfma<mode>4): Likewise.
40943 (nfms<mode>4): Likewise.
40945 2017-01-04 Marek Polacek <polacek@redhat.com>
40948 * doc/invoke.texi: Document -Wpointer-compare.
40950 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40952 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40955 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40956 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40959 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40962 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40963 documentation of the powerpc_popcntb_ok attribute.
40964 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40965 code to issue warning messages if a requested CPU configuration is
40966 not supported by the binary (assembler and loader) toolchain.
40967 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40968 made to define a built-in function that has been disabled.
40969 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40970 made to define a built-in function that has been disabled.
40971 (altivec_init_builtins): Add comment explaining why definition
40972 of the DST built-in functions is not preceded by an assertion
40973 check. Add assertions to prevent ICE if attempts are made to
40974 define an altivec predicate or an abs* built-in function that has
40976 (htm_init_builtins): Add comment explaining why definition of the
40977 htm built-in functions is not preceded by an assertion check.
40979 2017-01-04 Jeff Law <law@redhat.com>
40981 PR tree-optimizatin/67955
40982 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40983 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40984 the points-to solution does not include pt_null. Use DECL_PT_UID
40987 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40989 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40990 Use gen_int_mode instead of gen_lopwart for const_int operands.
40992 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40994 PR tree-optimization/71563
40995 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40996 out of range value - has low bits known to be zero.
40998 2017-01-04 Alan Modra <amodra@gmail.com>
41000 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41001 * configure: Regenerate.
41002 * config.in: Regenerate.
41004 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41007 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41008 a substring of the message, but strcmp with the whole message. Ifdef
41009 ENABLE_NLS, translate the message first using dgettext.
41011 2017-01-03 Jeff Law <law@redhat.com>
41013 PR tree-optimizatin/78856
41014 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41015 (mark_threaded_blocks): Remove code to truncate thread paths that
41016 cross multiple loop headers. Instead invalidate the cached loop
41017 iteration information and handle case of a thread path walking
41018 into an irreducible region.
41020 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41023 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41024 assertions. Add support for doing the signbit if the IEEE 128-bit
41025 floating point value is in a GPR.
41026 * config/rs6000/rs6000.md (Fsignbit): Delete.
41027 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41028 Update the length attribute if the value is in a GPR.
41029 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41030 the sign or zero extension instruction, since the value is always 0/1.
41031 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41034 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41035 extracting SImode to a GPR register so that we can generate a
41036 store, limit the vector to be in a traditional Altivec register
41037 for the vextuwrx instruction.
41039 2017-01-03 Ian Lance Taylor <iant@google.com>
41041 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41043 2017-01-03 Martin Sebor <msebor@redhat.com>
41045 PR tree-optimization/78696
41046 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41047 precision. Use MPFR for %f for greater fidelity. Correct handling
41049 (pass_sprintf_length::compute_format_length): Set width and precision
41050 specified by asrerisk to void_node for vararg functions.
41051 (try_substitute_return_value): Adjust dump output.
41053 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41055 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41057 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41059 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41060 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41061 -mlra/-mno-lra was passed to the compiler.
41063 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41065 PR rtl-optimization/65618
41066 * emit-rtl.c (try_split): Move initialization of "before" and
41067 "after" to just before the call to emit_insn_after_setloc.
41069 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41071 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41073 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41075 * dwarf2out.c (gen_enumeration_type_die): When
41076 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41078 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41080 PR tree-optimization/78965
41081 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41082 Change first argument from const call_info & to call_info &. For %n
41083 set info.nowrite to false.
41085 PR middle-end/78901
41086 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41087 possibly throwing calls.
41089 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41090 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41091 and fns handling, rather than in a separate case SSA_NAME.
41093 2017-01-02 Jeff Law <law@redhat.com>
41095 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41096 fixes for first_period and second_period variables.
41098 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41101 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41102 (*insvqi_1): New insn pattern.
41103 (*insvqi_1_mem_rex64): Ditto.
41104 (*insvqi_2): Ditto.
41105 (*insvqi_3): Rename from *insvqi.
41107 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41109 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41111 * doc/cfg.texi (Edges): Remove reference to Java.
41112 (Maintaining the CFG): Ditto.
41114 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41116 PR middle-end/77674
41117 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41118 transparent aliases.
41120 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41122 PR middle-end/77484
41123 * predict.def (PRED_CALL): Update hitrate.
41124 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41125 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41126 into direct/indirect/polymorphic variants.
41128 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41130 Update copyright years.
41132 * gcc.c (process_command): Update copyright notice dates.
41133 * gcov-dump.c (print_version): Ditto.
41134 * gcov.c (print_version): Ditto.
41135 * gcov-tool.c (print_version): Ditto.
41136 * gengtype.c (create_file): Ditto.
41137 * doc/cpp.texi: Bump @copying's copyright year.
41138 * doc/cppinternals.texi: Ditto.
41139 * doc/gcc.texi: Ditto.
41140 * doc/gccint.texi: Ditto.
41141 * doc/gcov.texi: Ditto.
41142 * doc/install.texi: Ditto.
41143 * doc/invoke.texi: Ditto.
41145 Copyright (C) 2017 Free Software Foundation, Inc.
41147 Copying and distribution of this file, with or without modification,
41148 are permitted in any medium without royalty provided the copyright
41149 notice and this notice are preserved.