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 * function.h (frame_space): Change start and length from HOST_WIDE_INT
7 (get_frame_size): Return the size as a poly_int64 rather than a
9 (frame_offset_overflow): Take the offset as a poly_int64 rather
11 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
12 (assign_stack_temp): Likewise for the size.
13 * function.c (get_frame_size): Return a poly_int64 rather than
15 (frame_offset_overflow): Take the offset as a poly_int64 rather
17 (try_fit_stack_local): Take the start, length and size as poly_int64s
18 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
19 rather than a HOST_WIDE_INT.
20 (add_frame_space): Take the start and end as poly_int64s rather than
22 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
23 (assign_stack_temp): Likewise for the size.
24 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
26 (find_temp_slot_from_address): Handle polynomial offsets.
27 (combine_temp_slots): Likewise.
28 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
30 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
31 poly_int64 rather than a HOST_WIDE_INT.
32 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
34 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
35 polynomial frame offsets.
36 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
37 as a poly_int64 rather than an int.
38 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
39 * config/v850/v850-protos.h (compute_frame_size): Likewise.
40 * config/v850/v850.c (compute_frame_size): Likewise.
41 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
42 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
43 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
44 * config/pa/pa.c (pa_compute_frame_size): Likewise.
45 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
46 rather than a HOST_WIDE_INT.
47 * explow.c (get_dynamic_stack_base): Likewise.
48 * final.c (final_start_function): Use the constant lower bound
49 of the frame size for -Wframe-larger-than.
50 * ira.c (do_reload): Adjust for new get_frame_size return type.
51 * lra.c (lra): Likewise.
52 * reload1.c (reload): Likewise.
53 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
54 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
55 * rtlanal.c (get_initial_register_offset): Return the offset as
56 a poly_int64 rather than a HOST_WIDE_INT.
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 * reload1.c (elim_table): Change initial_offset, offset and
63 previous_offset from HOST_WIDE_INT to poly_int64_pod.
64 (offsets_at): Change the target array's element type from
65 HOST_WIDE_INT to poly_int64_pod.
66 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
67 (elimination_costs_in_insn, update_eliminable_offsets)
68 (verify_initial_elim_offsets, set_offsets_for_label)
69 (init_eliminable_invariants): Update after above changes.
71 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
72 Alan Hayward <alan.hayward@arm.com>
73 David Sherwood <david.sherwood@arm.com>
75 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
76 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
77 (decomposition): Change start and end from HOST_WIDE_INT
79 (decompose, immune_p): Update accordingly.
80 (find_inc_amount): Return a poly_int64 rather than an int.
81 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
84 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
85 Alan Hayward <alan.hayward@arm.com>
86 David Sherwood <david.sherwood@arm.com>
88 * tree.h (get_inner_reference): Return the bitsize and bitpos
89 as poly_int64_pods rather than HOST_WIDE_INT.
90 * fold-const.h (ptr_difference_const): Return the pointer difference
91 as a poly_int64_pod rather than a HOST_WIDE_INT.
92 * expr.c (get_inner_reference): Return the bitsize and bitpos
93 as poly_int64_pods rather than HOST_WIDE_INT.
94 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
96 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
97 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
98 (optimize_bit_field_compare): Update call to get_inner_reference.
99 (decode_field_reference): Likewise.
100 (fold_unary_loc): Track polynomial offsets and sizes.
101 (split_address_to_core_and_offset): Return the bitpos as a
102 poly_int64_pod rather than a HOST_WIDE_INT.
103 (ptr_difference_const): Likewise for the pointer difference.
104 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
105 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
106 * dbxout.c (dbxout_expand_expr): Likewise.
107 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
108 (loc_list_from_tree_1, fortran_common): Likewise.
109 * gimple-laddress.c (pass_laddress::execute): Likewise.
110 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
111 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
112 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
113 * tree-affine.c (tree_to_aff_combination): Likewise.
114 (get_inner_reference_aff): Likewise.
115 * tree-data-ref.c (split_constant_offset_1): Likewise.
116 (dr_analyze_innermost): Likewise.
117 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
118 * tree-sra.c (ipa_sra_check_caller): Likewise.
119 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
120 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
121 (instrument_bool_enum_load, instrument_object_size): Likewise.
122 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
123 to get_inner_reference.
124 * hsa-gen.c (gen_hsa_addr): Likewise.
125 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
126 * tsan.c (instrument_expr): Likewise.
127 * match.pd: Update call to ptr_difference_const.
129 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
130 Alan Hayward <alan.hayward@arm.com>
131 David Sherwood <david.sherwood@arm.com>
133 * fold-const.c (fold_comparison): Track sizes and offsets as
134 poly_int64s rather than HOST_WIDE_INTs when folding address
137 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
138 Alan Hayward <alan.hayward@arm.com>
139 David Sherwood <david.sherwood@arm.com>
141 * expr.h (get_bit_range): Return the bitstart and bitend as
142 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
143 as a poly_int64 rather than a HOST_WIDE_INT.
144 * expr.c (get_bit_range): Likewise.
145 (expand_assignment): Update call accordingly.
146 * fold-const.c (optimize_bit_field_compare): Likewise.
148 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
149 Alan Hayward <alan.hayward@arm.com>
150 David Sherwood <david.sherwood@arm.com>
152 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
153 (aarch64_legitimate_address_p): Use it instead of an rtx code,
154 as an optional final parameter.
155 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
156 (aarch64_legitimate_address_p): Likewise.
157 (aarch64_print_address_internal): Take an aarch64_addr_query_type
158 instead of an rtx code.
159 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
160 (aarch64_legitimate_address_hook_p): Likewise.
161 (aarch64_print_ldpstp_address): Likewise.
162 (aarch64_print_operand_address): Likewise.
163 (aarch64_address_cost): Likewise.
164 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
165 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
166 (aarch64_mem_pair_lanes_operand): Likewise.
168 2017-12-20 Richard Biener <rguenther@suse.de>
170 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
171 update_stmt_if_modified.
173 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
175 PR tree-optimization/83491
176 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
177 before walking uses. Improve coding style and comments.
179 2017-12-20 Tom de Vries <tom@codesourcery.com>
181 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
183 2017-12-20 Jakub Jelinek <jakub@redhat.com>
186 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
188 (pass_ipa_free_fn_summary): Add small_p private data member,
189 initialize to false in the ctor.
190 (pass_ipa_free_fn_summary::clone,
191 pass_ipa_free_fn_summary::set_pass_param,
192 pass_ipa_free_fn_summary::gate): New methods.
193 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
194 | TODO_dump_symtab if small_p.
195 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
196 entry and add another instance of the pass with false parm after
198 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
199 ipa_free_fn_summary here.
201 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
203 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
205 2017-12-20 Martin Sebor <msebor@redhat.com>
208 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
209 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
210 (expand_builtin_strncmp): Same.
212 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
215 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
216 allowed debug stmts before labels.
217 (expand_gimple_basic_block): Likewise.
218 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
219 * gimple-iterator.h (gsi_after_labels): Likewise.
220 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
221 rename reused variable, and simplify using gsi_move_before.
222 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
223 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
224 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
225 (gimple_verify_flow_info, gimple_block_label): Likewise.
226 (make_blocks): Move debug markers after adjacent labels.
227 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
228 allowed debug insns outside blocks.
229 * df-scan.c (df_insn_delete): Likewise.
230 * lra-constraints.c (update_ebb_live_info): Likewise.
231 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
232 (vt_initialize, delete_vta_debug_insns): Likewise.
233 (reemit_marker_as_note): Drop BB parm. Adjust callers.
235 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
236 Alan Hayward <alan.hayward@arm.com>
237 David Sherwood <david.sherwood@arm.com>
239 * poly-int-types.h (round_down_to_byte_boundary): New macro.
240 (round_up_to_byte_boundary): Likewise.
241 * expr.h (get_bit_range): Add temporary shim.
242 * gimple-ssa-store-merging.c (store_operand_info): Change the
243 bitsize, bitpos, bitregion_start and bitregion_end fields from
244 unsigned HOST_WIDE_INT to poly_uint64.
245 (merged_store_group): Likewise load_align_base.
246 (compatible_load_p, compatible_load_p): Update accordingly.
247 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
248 (split_group, imm_store_chain_info::output_merged_store): Likewise.
249 (mem_valid_for_store_merging): Return the bitsize, bitpos,
250 bitregion_start and bitregion_end as poly_uint64s rather than
251 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
252 (handled_load): Take the bitsize, bitpos,
253 bitregion_start and bitregion_end as poly_uint64s rather than
254 unsigned HOST_WIDE_INTs.
255 (pass_store_merging::process_store): Update call to
256 mem_valid_for_store_merging.
258 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
259 Alan Hayward <alan.hayward@arm.com>
260 David Sherwood <david.sherwood@arm.com>
262 * builtins.c (get_object_alignment_2): Track polynomial offsets
263 and sizes. Update the alignment handling.
265 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
266 Alan Hayward <alan.hayward@arm.com>
267 David Sherwood <david.sherwood@arm.com>
269 * tree.h (get_inner_reference): Add a version that returns the
270 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
271 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
272 the case in which bitpos is not associated with the first byte.
274 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
275 Alan Hayward <alan.hayward@arm.com>
276 David Sherwood <david.sherwood@arm.com>
278 * tree-affine.h (get_inner_reference_aff): Return the size as a
280 * tree-affine.c (get_inner_reference_aff): Likewise.
281 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
282 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
284 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
285 Alan Hayward <alan.hayward@arm.com>
286 David Sherwood <david.sherwood@arm.com>
288 * fold-const.c (pointer_may_wrap_p): Take the offset as a
289 HOST_WIDE_INT rather than a poly_int64.
291 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
292 Alan Hayward <alan.hayward@arm.com>
293 David Sherwood <david.sherwood@arm.com>
295 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
296 HOST_WIDE_INT to poly_int64_pod.
297 (perform_symbolic_merge): Update accordingly.
298 (bswap_replace): Likewise.
300 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
301 Alan Hayward <alan.hayward@arm.com>
302 David Sherwood <david.sherwood@arm.com>
304 * tree-affine.h (aff_tree::offset): Change from widest_int
306 (wide_int_ext_for_comb): Delete.
307 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
308 constants as poly_widest_int rather than widest_int.
309 (aff_combination_constant_multiple_p): Return the multiplier
310 as a poly_widest_int.
311 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
313 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
314 version static and add an overload for poly_widest_int.
315 (aff_combination_const, aff_combination_add_cst)
316 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
317 the constants as poly_widest_int rather than widest_int.
318 (tree_to_aff_combination): Generalize INTEGER_CST case to
320 (aff_combination_to_tree): Track offsets as poly_widest_ints.
321 (aff_combination_add_product, aff_combination_mult): Handle
323 (aff_combination_constant_multiple_p): Return the multiplier
324 as a poly_widest_int.
325 * tree-predcom.c (determine_offset): Return the offset as a
327 (split_data_refs_to_components, suitable_component_p): Update
329 (valid_initializer_p): Update call to
330 aff_combination_constant_multiple_p.
331 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
332 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
333 as a poly_int64 rather than a HOST_WIDE_INT.
334 (get_address_cost): Handle polynomial offsets.
335 (iv_elimination_compare_lt): Likewise.
336 (rewrite_use_nonlinear_expr): Likewise.
338 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
339 Alan Hayward <alan.hayward@arm.com>
340 David Sherwood <david.sherwood@arm.com>
342 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
343 as a poly_int64_pod rather than a HOST_WIDE_INT.
344 (get_addr_base_and_unit_offset): Likewise.
345 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
346 (get_addr_base_and_unit_offset): Likewise.
347 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
348 to poly_int64 in example.
349 * fold-const.c (fold_binary_loc): Update call to
350 get_addr_base_and_unit_offset.
351 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
352 (maybe_canonicalize_mem_ref_addr): Likewise.
353 (gimple_fold_stmt_to_constant_1): Likewise.
354 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
356 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
357 * match.pd: Likewise.
358 * omp-low.c (lower_omp_target): Likewise.
359 * tree-sra.c (build_ref_for_offset): Likewise.
360 (build_debug_ref_for_model): Likewise.
361 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
362 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
363 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
364 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
365 (constant_pointer_difference): Likewise.
366 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
367 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
368 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
369 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
370 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
371 (set_ssa_val_to): Likewise.
372 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
373 (maybe_diag_stxncpy_trunc): Likewise.
374 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
375 * tree.c (build_simple_mem_ref_loc): Likewise.
376 (array_at_struct_end_p): Likewise.
378 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
379 Alan Hayward <alan.hayward@arm.com>
380 David Sherwood <david.sherwood@arm.com>
382 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
383 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
384 (get_ref_base_and_extent_hwi): Declare.
385 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
386 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
387 (get_ref_base_and_extent_hwi): New function.
388 * cfgexpand.c (expand_debug_expr): Update call to
389 get_ref_base_and_extent.
390 * dwarf2out.c (add_var_loc_to_decl): Likewise.
391 * gimple-fold.c (get_base_constructor): Return the offset as a
392 poly_int64_pod rather than a HOST_WIDE_INT.
393 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
394 * ipa-polymorphic-call.c
395 (ipa_polymorphic_call_context::set_by_invariant)
396 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
397 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
398 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
399 rather than get_ref_base_and_extent.
400 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
401 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
402 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
404 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
405 call to get_ref_base_and_extent.
406 * tree-sra.c (create_access, get_access_for_expr): Likewise.
407 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
408 (stmt_kills_ref_p): Likewise.
409 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
410 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
412 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
413 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
414 when calling native_encode_expr.
415 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
416 call to get_ref_base_and_extent.
417 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
418 get_ref_base_and_extent.
419 * var-tracking.c (track_expr_p): Likewise.
421 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
422 Alan Hayward <alan.hayward@arm.com>
423 David Sherwood <david.sherwood@arm.com>
425 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
426 HOST_WIDE_INT to poly_int64_pod.
427 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
428 polynomail parameter offsets.
430 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
431 Alan Hayward <alan.hayward@arm.com>
432 David Sherwood <david.sherwood@arm.com>
434 * gengtype.c (main): Handle poly_int64_pod.
435 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
436 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
437 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
438 from HOST_WIDE_INT to poly_int64_pod.
439 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
440 (copy_cfa): New function.
441 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
442 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
443 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
444 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
445 (initial_return_save): Treat offsets as poly_ints.
446 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
447 offset is nonconstant.
448 (reg_save): Take the offset as a poly_int64. Fall back to
449 DW_CFA_expression for nonconstant offsets.
450 (queue_reg_save): Take the offset as a poly_int64.
451 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
453 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
454 Alan Hayward <alan.hayward@arm.com>
455 David Sherwood <david.sherwood@arm.com>
457 * rtl.h (operand_subword, operand_subword_force): Take the offset
458 as a poly_uint64 an unsigned int.
459 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
461 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
462 Alan Hayward <alan.hayward@arm.com>
463 David Sherwood <david.sherwood@arm.com>
465 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
466 'p' format code. Use INT_LIST rather than SUBREG as the example of
467 a code with an XINT and an XEXP. Remove the implication that
468 accessing an rtx field using XINT is expected to work.
469 * rtl.def (SUBREG): Change format from "ei" to "ep".
470 * rtl.h (rtunion::rt_subreg): New field.
471 (XCSUBREG): New macro.
472 (SUBREG_BYTE): Use it.
473 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
474 Update constructor accordingly.
475 (subreg_shape::operator ==): Update accordingly.
476 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
477 than an unsigned int.
478 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
479 a poly_uint64 rather than an unsigned int.
480 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
481 than an unsigned int.
482 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
483 (subreg_size_highpart_offset): Return a poly_uint64 rather than
484 an unsigned int. Take the sizes as poly_uint64s.
485 (subreg_offset_from_lsb): Return a poly_uint64 rather than
486 an unsigned int. Take the shift as a poly_uint64 rather than
488 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
489 as a poly_uint64 rather than an unsigned int.
490 (simplify_subreg_regno): Likewise.
491 (byte_lowpart_offset): Return the memory offset as a poly_int64
493 (subreg_memory_offset): Likewise. Take the subreg offset as a
494 poly_uint64 rather than an unsigned int.
495 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
496 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
497 poly_uint64 rather than an unsigned int.
498 * rtl.c (rtx_format): Describe 'p' in comment.
499 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
500 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
501 offset as a poly_uint64 rather than an unsigned int.
502 (byte_lowpart_offset): Return the memory offset as a poly_int64
504 (subreg_memory_offset): Likewise. Take the subreg offset as a
505 poly_uint64 rather than an unsigned int.
506 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
507 mode sizes as poly_uint64s rather than unsigned ints. Return a
508 poly_uint64 rather than an unsigned int.
509 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
510 (copy_insn_1): Handle 'p'.
511 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
512 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
513 an unsigned int. Return the shift in the same way.
514 (subreg_lsb): Return the shift as a poly_uint64 rather than an
516 (subreg_size_offset_from_lsb): Take the sizes and shift as
517 poly_uint64s rather than unsigned ints. Return the offset as
519 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
520 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
522 * rtlhash.c (add_rtx): Handle 'p'.
523 * genemit.c (gen_exp): Likewise.
524 * gengenrtl.c (type_from_format, gendef): Likewise.
525 * gensupport.c (subst_pattern_match, get_alternatives_number)
526 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
527 (subst_dup): Likewise.
528 * gengtype.c (adjust_field_rtx_def): Likewise.
529 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
530 (match_pattern_2): Likewise.
531 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
532 (rtx_test::subreg_field): New function.
533 (operator ==, safe_to_hoist_p, transition_parameter_type)
534 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
535 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
536 * genpeep.c (match_rtx): Likewise.
537 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
538 (rtx_writer::print_rtx_operand): Handle 'p'.
539 (print_value): Handle SUBREG.
540 * read-rtl.c (apply_int_iterator): Likewise.
541 (rtx_reader::read_rtx_operand): Handle 'p'.
542 * alias.c (rtx_equal_for_memref_p): Likewise.
543 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
544 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
546 * calls.c (expand_call): Likewise.
547 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
548 (make_extraction, gen_lowpart_for_combine): Likewise.
549 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
551 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
552 rather than an unsigned int. Treat subreg offsets as poly_ints.
553 (exp_equiv_p): Handle 'p'.
554 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
555 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
556 * dse.c (find_shift_sequence): Likewise.
557 * dwarf2out.c (rtl_for_decl_location): Likewise.
558 * expmed.c (extract_low_bits): Likewise.
559 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
560 (expand_expr_real_2): Likewise.
561 * final.c (alter_subreg): Likewise.
562 (leaf_renumber_regs_insn): Handle 'p'.
563 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
564 Treat subreg offsets as poly_ints.
565 * fwprop.c (forward_propagate_and_simplify): Likewise.
566 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
567 * ira.c (get_subreg_tracking_sizes): Likewise.
568 * ira-conflicts.c (go_through_subreg): Likewise.
569 * ira-lives.c (process_single_reg_class_operands): Likewise.
570 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
571 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
572 as a poly_uint64 rather than an unsigned int.
573 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
574 subreg offsets as poly_ints.
575 * lra-constraints.c (operands_match_p): Handle 'p'.
576 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
577 * lra-spills.c (assign_mem_slot): Likewise.
578 * postreload.c (move2add_valid_value_p): Likewise.
579 * recog.c (general_operand, indirect_operand): Likewise.
580 * regcprop.c (copy_value, maybe_mode_change): Likewise.
581 (copyprop_hardreg_forward_1): Likewise.
582 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
583 (record_subregs_of_mode): Likewise.
584 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
585 * reload.c (operands_match_p): Handle 'p'.
586 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
587 * reload1.c (alter_reg, choose_reload_regs): Likewise.
588 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
589 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
590 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
591 (simplify_const_poly_int_tests<N>::run): Likewise.
592 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
593 a poly_uint64 rather than an unsigned int.
594 * valtrack.c (debug_lowpart_subreg): Likewise.
595 * var-tracking.c (var_lowpart): Likewise.
596 (loc_cmp): Handle 'p'.
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 * ira.c (get_subreg_tracking_sizes): New function.
603 (init_live_subregs): Take an integer size rather than a register.
604 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
605 to init_live_subregs.
607 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
608 Alan Hayward <alan.hayward@arm.com>
609 David Sherwood <david.sherwood@arm.com>
611 * expr.c (store_constructor_field): Change bitsize from a
612 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
613 HOST_WIDE_INT to a poly_int64.
614 (store_constructor): Change size from a HOST_WIDE_INT to
616 (store_field): Likewise bitsize and bitpos.
618 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
619 Alan Hayward <alan.hayward@arm.com>
620 David Sherwood <david.sherwood@arm.com>
622 * expmed.h (store_bit_field): Change bitregion_start and
623 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
624 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
625 (store_bit_field_1, store_integral_bit_field, store_bit_field)
626 (store_fixed_bit_field, store_split_bit_field): Likewise.
627 * expr.c (store_constructor_field, store_field): Likewise.
628 (optimize_bitfield_assignment_op): Likewise. Make the same change
629 to bitsize and bitpos.
630 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
631 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
632 same change in the constructor arguments.
633 (get_best_mode): Change bitregion_start and bitregion_end from
634 unsigned HOST_WIDE_INT to poly_uint64.
635 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
636 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
638 (bit_field_mode_iterator::next_mode): Update for new types
639 of m_bitregion_start and m_bitregion_end.
640 (get_best_mode): Change bitregion_start and bitregion_end from
641 unsigned HOST_WIDE_INT to poly_uint64.
643 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
644 Alan Hayward <alan.hayward@arm.com>
645 David Sherwood <david.sherwood@arm.com>
647 * rtl.h (simplify_gen_subreg): Add a temporary overload that
648 accepts poly_uint64 offsets.
649 * expmed.h (extract_bit_field): Take bitsize and bitnum as
650 poly_uint64s rather than unsigned HOST_WIDE_INTs.
651 * expmed.c (lowpart_bit_field_p): Likewise.
652 (extract_bit_field_as_subreg): New function, split out from...
653 (extract_bit_field_1): ...here. Take bitsize and bitnum as
654 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
655 extractions, check that BITSIZE matches the size of the extracted
656 value and that BITNUM is an exact multiple of that size.
657 If all else fails, try forcing the value into memory if
658 BITNUM is variable, and adjusting the address so that the
659 offset is constant. Split the part that can only handle constant
660 bitsize and bitnum out into...
661 (extract_integral_bit_field): ...this new function.
662 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
663 rather than unsigned HOST_WIDE_INTs.
665 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
666 Alan Hayward <alan.hayward@arm.com>
667 David Sherwood <david.sherwood@arm.com>
669 * expmed.h (store_bit_field): Take bitsize and bitnum as
670 poly_uint64s rather than unsigned HOST_WIDE_INTs.
671 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
672 that returns the byte size.
673 (store_bit_field_1): Take bitsize and bitnum as
674 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
675 to simple_mem_bitfield_p. Split the part that can only handle
676 constant bitsize and bitnum out into...
677 (store_integral_bit_field): ...this new function.
678 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
679 than unsigned HOST_WIDE_INTs.
680 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
682 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
683 Alan Hayward <alan.hayward@arm.com>
684 David Sherwood <david.sherwood@arm.com>
686 * lra-int.h (lra_reg): Change offset from int to poly_int64.
687 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
689 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
690 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
691 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
692 offset as a poly_int64 rather than an int.
693 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
694 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
695 * lra-constraints.c (equiv_address_substitution): Track offsets
697 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
698 (curr_insn_transform): Handle the new form of sp_offset.
699 * lra-eliminations.c (lra_elim_table): Change previous_offset
700 and offset from HOST_WIDE_INT to poly_int64.
701 (print_elim_table, update_reg_eliminate): Update accordingly.
702 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
703 (get_elimination): Update accordingly.
704 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
705 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
706 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
707 poly_int64 offsets generally.
708 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
709 (mark_not_eliminable, init_elimination): Update accordingly.
710 (remove_reg_equal_offset_note): Return a bool and pass the new
711 offset back by pointer as a poly_int64.
712 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
713 rather than a HOST_WIDE_INT.
714 (do_remat): Track offsets poly_int64s.
715 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
717 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
718 Alan Hayward <alan.hayward@arm.com>
719 David Sherwood <david.sherwood@arm.com>
721 * rtl.h (mem_attrs): Add a default constructor. Change size and
722 offset from HOST_WIDE_INT to poly_int64.
723 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
724 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
725 (widen_memory_access): Take the sizes and offsets as poly_int64s
726 rather than HOST_WIDE_INTs.
727 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
728 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
730 (adjust_offset_for_component_ref): Change the offset from a
731 HOST_WIDE_INT to a poly_int64.
732 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
733 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
734 * dce.c (find_call_stack_args): Likewise.
735 * dse.c (record_store): Likewise.
736 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
737 * print-rtl.c (rtx_writer::print_rtx): Likewise.
738 * read-rtl-function.c (test_loading_mem): Likewise.
739 * rtlanal.c (may_trap_p_1): Likewise.
740 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
741 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
742 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
743 (mem_attrs::mem_attrs): New function.
744 (set_mem_attributes_minus_bitpos): Change bitpos from a
745 HOST_WIDE_INT to poly_int64.
746 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
747 (clear_mem_offset, clear_mem_size, change_address)
748 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
749 initialize mem_attrs.
750 (set_mem_offset, set_mem_size, adjust_address_1)
751 (adjust_automodify_address_1, offset_address, widen_memory_access):
752 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
754 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
755 Alan Hayward <alan.hayward@arm.com>
756 David Sherwood <david.sherwood@arm.com>
758 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
759 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
760 rather than 0 to represent an unknown size. Assert that the size
761 is known when the mode isn't BLKmode.
762 (may_trap_p_1): Use -1 for unknown sizes.
763 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
765 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
766 Alan Hayward <alan.hayward@arm.com>
767 David Sherwood <david.sherwood@arm.com>
769 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
770 to poly_int64. Update commentary for positions_needed.large.
771 (read_info_type): Change offset and width from HOST_WIDE_INT
773 (set_usage_bits): Likewise.
774 (canon_address): Return the offset as a poly_int64 rather than
775 a HOST_WIDE_INT. Use strip_offset_and_add.
776 (set_all_positions_unneeded, any_positions_needed_p): Use
777 positions_needed.large to track stores with non-constant widths.
778 (all_positions_needed_p): Likewise. Take the offset and width
779 as poly_int64s rather than ints. Assert that rhs is nonnull.
780 (record_store): Cope with non-constant offsets and widths.
781 Nullify the rhs of an earlier store if we can't tell which bytes
783 (find_shift_sequence): Take the access_size and shift as poly_int64s
785 (get_stored_val): Take the read_offset and read_width as poly_int64s
786 rather than HOST_WIDE_INTs.
787 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
788 non-constant offsets and widths.
790 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
791 Alan Hayward <alan.hayward@arm.com>
792 David Sherwood <david.sherwood@arm.com>
794 * inchash.h (inchash::hash::add_poly_int): New function.
795 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
796 Use poly_int64 rather than HOST_WIDE_INT.
797 (ao_ref::max_size_known_p): New function.
798 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
799 rather than HOST_WIDE_INT.
800 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
801 to temporaries until its interface is adjusted to match.
802 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
803 (aliasing_component_refs_p, decl_refs_may_alias_p)
804 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
805 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
806 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
808 * alias.c (ao_ref_from_mem): Likewise.
809 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
810 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
811 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
812 (maybe_trim_complex_store, maybe_trim_constructor_store)
813 (live_bytes_read, dse_classify_store): Likewise.
814 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
815 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
816 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
817 (vn_reference_lookup_3): Likewise.
818 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
820 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
821 Alan Hayward <alan.hayward@arm.com>
822 David Sherwood <david.sherwood@arm.com>
824 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
825 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
826 instead of ranges_overlap_p.
828 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
829 Alan Hayward <alan.hayward@arm.com>
830 David Sherwood <david.sherwood@arm.com>
832 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
833 sizes as poly_int64s rather than HOST_WIDE_INTs.
835 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
836 Alan Hayward <alan.hayward@arm.com>
837 David Sherwood <david.sherwood@arm.com>
839 * gimple-fold.h (fold_ctor_reference): Take the offset and size
840 as poly_uint64 rather than unsigned HOST_WIDE_INT.
841 * gimple-fold.c (fold_ctor_reference): Likewise.
843 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
844 Alan Hayward <alan.hayward@arm.com>
845 David Sherwood <david.sherwood@arm.com>
847 * target.def (dwarf_poly_indeterminate_value): New hook.
848 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
849 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
850 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
851 * doc/tm.texi: Regenerate.
852 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
853 offset as a poly_int64.
854 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
855 offset as a poly_int64.
856 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
857 Take the offset as a poly_int64.
858 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
859 (frame_pointer_fb_offset): Change to a poly_int64.
860 (int_loc_descriptor): Take the offset as a poly_int64. Use
861 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
862 (based_loc_descr): Take the offset as a poly_int64.
863 Use strip_offset_and_add to handle (plus X (const)).
864 Use new_reg_loc_descr instead of an open-coded version of the
865 previous implementation.
866 (mem_loc_descriptor): Handle CONST_POLY_INT.
867 (compute_frame_pointer_to_fb_displacement): Take the offset as a
868 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
870 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
871 Alan Hayward <alan.hayward@arm.com>
872 David Sherwood <david.sherwood@arm.com>
874 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
875 (gen_rtx_REG_offset): Take the offset as a poly_int64.
876 * inchash.h (inchash::hash::add_poly_hwi): New function.
877 * gengtype.c (main): Register poly_int64.
878 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
879 offset as a poly_int.
880 (reg_attr_hasher::equal): Use must_eq to compare offsets.
881 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
882 offset as a poly_int64.
883 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
884 * print-rtl.c (print_poly_int): New function.
885 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
887 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
889 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
890 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
891 (vt_get_decl_and_offset): Return the offset as a poly_int64.
892 Enforce track_offset_p for parts of a PARALLEL.
893 (vt_add_function_parameter): Use const_offset for the final
894 offset to track. Use get_tracked_reg_offset for the parts
897 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
898 Alan Hayward <alan.hayward@arm.com>
899 David Sherwood <david.sherwood@arm.com>
901 * target.def (truly_noop_truncation): Take poly_uint64s instead of
902 unsigned ints. Change default to hook_bool_puint64_puint64_true.
903 * doc/tm.texi: Regenerate.
904 * hooks.h (hook_bool_uint_uint_true): Delete.
905 (hook_bool_puint64_puint64_true): Declare.
906 * hooks.c (hook_bool_uint_uint_true): Delete.
907 (hook_bool_puint64_puint64_true): New function.
908 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
909 instead of unsigned ints.
910 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
911 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
913 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
914 Alan Hayward <alan.hayward@arm.com>
915 David Sherwood <david.sherwood@arm.com>
917 * optabs.h (expand_operand): Add an int_value field.
918 (create_expand_operand): Add an int_value parameter and use it
919 to initialize the new expand_operand field.
920 (create_integer_operand): Replace with a declaration of a function
921 that accepts poly_int64s. Move the implementation to...
922 * optabs.c (create_integer_operand): ...here.
923 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
924 the mode preserves the value of int_value, instead of calling
925 const_int_operand on the rtx. Use gen_int_mode to generate
928 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
929 Alan Hayward <alan.hayward@arm.com>
930 David Sherwood <david.sherwood@arm.com>
932 * dumpfile.h (dump_dec): Declare.
933 * dumpfile.c (dump_dec): New function.
934 * pretty-print.h (pp_wide_integer): Turn into a function and
935 declare a poly_int version.
936 * pretty-print.c (pp_wide_integer): New function for poly_ints.
938 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
939 Alan Hayward <alan.hayward@arm.com>
940 David Sherwood <david.sherwood@arm.com>
942 * doc/generic.texi (POLY_INT_CST): Document.
943 * tree.def (POLY_INT_CST): New tree code.
944 * treestruct.def (TS_POLY_INT_CST): New tree layout.
945 * tree-core.h (tree_poly_int_cst): New struct.
946 (tree_node): Add a poly_int_cst field.
947 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
948 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
949 instead of a wide_int_ref.
950 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
952 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
953 instead of an unsigned HOST_WIDE_INT.
954 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
955 (ptrdiff_tree_p): Declare.
956 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
957 extern inline implementations if the target doesn't use POLY_INT_CST.
958 (poly_int_tree_p): New function.
959 (wi::unextended_tree): New class.
960 (wi::int_traits <unextended_tree>): New override.
961 (wi::extended_tree): Add a default constructor.
962 (wi::extended_tree::get_tree): New function.
963 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
964 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
965 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
966 (wi::tree_to_poly_wide_ref): New typedefs.
967 (wi::ints_for): Provide overloads for extended_tree and
969 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
970 (wi::to_wide): New functions.
971 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
972 * tree.c (poly_int_cst_hasher): New struct.
973 (poly_int_cst_hash_table): New variable.
974 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
975 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
977 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
978 (init_ttree): Initialize poly_int_cst_hash_table.
979 (build_int_cst, build_int_cst_type, build_invariant_address): Take
980 a poly_int64 instead of a HOST_WIDE_INT.
981 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
982 instead of an unsigned HOST_WIDE_INT.
983 (wide_int_to_tree): Rename to...
984 (wide_int_to_tree_1): ...this.
985 (build_new_poly_int_cst, build_poly_int_cst): New functions.
986 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
987 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
988 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
989 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
990 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
992 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
993 (streamer_read_tree_body): Likewise.
994 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
995 (streamer_write_tree_body): Likewise.
996 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
997 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
998 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
999 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1000 * gimple-expr.h (is_gimple_constant): Likewise.
1001 * gimplify.c (maybe_with_size_expr): Likewise.
1002 * print-tree.c (print_node): Likewise.
1003 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1004 * tree-pretty-print.c (dump_generic_node): Likewise.
1005 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1006 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1007 * tree-vrp.c (compare_values_warnv): Likewise.
1008 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1009 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1010 (force_expr_to_var_cost): Likewise.
1011 * tree-ssa-loop.c (for_each_index): Likewise.
1012 * fold-const.h (build_invariant_address, size_int_kind): Take a
1013 poly_int64 instead of a HOST_WIDE_INT.
1014 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1015 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1017 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1018 (int_const_binop_2): New function, split out from...
1019 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1020 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1021 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1022 * gimple-ssa-strength-reduction.c (slsr_process_add)
1023 (slsr_process_mul): Check for INTEGER_CSTs before using them
1025 * stor-layout.c (bits_from_bytes): New function.
1026 (bit_from_pos): Use it.
1027 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1028 by BITS_PER_UNIT to get the TYPE_SIZE.
1029 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1030 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1032 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1033 Alan Hayward <alan.hayward@arm.com>
1034 David Sherwood <david.sherwood@arm.com>
1036 * doc/rtl.texi (const_poly_int): Document. Also document the
1037 rtl sharing behavior.
1038 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1039 * rtl.h (const_poly_int_def): New struct.
1040 (rtx_def::u): Add a cpi field.
1041 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1042 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1043 (wi::rtx_to_poly_wide_ref): New typedef
1044 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1045 (poly_int_rtx_p): New functions.
1046 (trunc_int_for_mode): Declare a poly_int64 version.
1047 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1048 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1050 (strip_offset): Declare.
1051 (strip_offset_and_add): New function.
1052 * rtl.def (CONST_POLY_INT): New rtx code.
1053 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1054 (shared_const_p): Use poly_int_rtx_p.
1055 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1057 (gen_int_shift_amount): Likewise.
1058 * emit-rtl.c (const_poly_int_hasher): New class.
1059 (const_poly_int_htab): New variable.
1060 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1061 (const_poly_int_hasher::hash): New function.
1062 (const_poly_int_hasher::equal): Likewise.
1063 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1064 (immed_wide_int_const): Rename to...
1065 (immed_wide_int_const_1): ...this and make static.
1066 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1067 instead of a wide_int_ref.
1068 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1069 (gen_lowpart_common): Handle CONST_POLY_INT.
1070 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1071 * cselib.c (cselib_hash_rtx): Likewise.
1072 * dwarf2out.c (const_ok_for_output_1): Likewise.
1073 * expr.c (convert_modes): Likewise.
1074 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1075 * rtlhash.c (add_rtx): Likewise.
1076 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1077 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1078 Handle existing CONST_POLY_INT rtxes.
1079 * expmed.h (expand_shift): Take a poly_int64 instead of a
1081 * expmed.c (expand_shift): Likewise.
1082 * rtlanal.c (strip_offset): New function.
1083 (commutative_operand_precedence): Give CONST_POLY_INT the same
1084 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1086 * rtl-tests.c (const_poly_int_tests): New struct.
1087 (rtl_tests_c_tests): Use it.
1088 * simplify-rtx.c (simplify_const_unary_operation): Handle
1090 (simplify_const_binary_operation): Likewise.
1091 (simplify_binary_operation_1): Fold additions of symbolic constants
1092 and CONST_POLY_INTs.
1093 (simplify_subreg): Handle extensions and truncations of
1095 (simplify_const_poly_int_tests): New struct.
1096 (simplify_rtx_c_tests): Use it.
1097 * wide-int.h (storage_ref): Add default constructor.
1098 (wide_int_ref_storage): Likewise.
1099 (trailing_wide_ints): Use GTY((user)).
1100 (trailing_wide_ints::operator[]): Add a const version.
1101 (trailing_wide_ints::get_precision): New function.
1102 (trailing_wide_ints::extra_size): Likewise.
1104 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1105 Alan Hayward <alan.hayward@arm.com>
1106 David Sherwood <david.sherwood@arm.com>
1108 * emit-rtl.h (gen_int_shift_amount): Declare.
1109 * emit-rtl.c (gen_int_shift_amount): New function.
1110 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1112 * calls.c (shift_return_value): Likewise.
1113 * cse.c (fold_rtx): Likewise.
1114 * dse.c (find_shift_sequence): Likewise.
1115 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1116 (expand_shift, expand_smod_pow2): Likewise.
1117 * lower-subreg.c (shift_cost): Likewise.
1118 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1119 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1120 (expand_vec_perm_var): Likewise.
1121 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1122 (simplify_binary_operation_1): Likewise.
1123 * combine.c (try_combine, find_split_point, force_int_to_mode)
1124 (simplify_shift_const_1, simplify_shift_const): Likewise.
1125 (change_zero_ext): Likewise. Use simplify_gen_binary.
1127 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1129 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1131 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1133 * doc/invoke.texi (ARM Options): Document accepted extension options
1134 for -march=armv8.3-a.
1136 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1139 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1140 or --with-float=softfp, set the default CPU to arm10e.
1142 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1144 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1145 * config/visium/predicates.md (const_shift_operand): Likewise.
1146 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1147 (visium_legitimize_reload_address): Likewise.
1149 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1152 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1154 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1157 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1159 2017-12-20 Julia Koval <julia.koval@intel.com>
1161 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1162 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1163 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1164 * config.gcc: Include vpclmulqdqintrin.h.
1165 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1166 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1168 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1169 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1170 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1171 * config/i386/i386.c (isa2_opts): Add -mcx16.
1172 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1173 (ix86_option_override_internal): Move mcx16 to flags2.
1174 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1175 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1176 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1177 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1178 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1179 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1180 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1181 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1182 * doc/invoke.texi: Add -mvpclmulqdq.
1184 2017-12-20 Tom de Vries <tom@codesourcery.com>
1187 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1188 * calls.c (rtx_for_static_chain): ... here. New function.
1189 * calls.h (rtx_for_static_chain): Declare.
1190 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1191 instead of targetm.calls.static_chain.
1192 * df-scan.c (df_get_entry_block_def_set): Same.
1194 2017-12-19 Tom de Vries <tom@codesourcery.com>
1196 PR tree-optimization/83493
1197 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1200 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1202 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1203 inform with hardcoded english plural handling.
1205 2017-12-18 Jeff Law <law@redhat.com>
1207 PR tree-optimization/83477
1208 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1209 a non-virtual PHI, always push a new range.
1211 2017-12-19 Martin Sebor <msebor@redhat.com>
1214 * builtins.c (compute_objsize): Handle non-constant offsets.
1216 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1218 PR tree-optimization/83444
1219 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1220 character load case, if get_stridx on MEM_REF's operand doesn't
1221 look usable, retry with get_addr_stridx.
1223 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1226 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1227 (variable_tracking_main_1): Keep markers even when VTA fails.
1230 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1231 even if there are markers before it.
1232 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1234 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1236 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1237 typical order conditions.
1238 * sel-sched.c (extract_new_fences_from): Likewise.
1239 * config/visium/constraints.md (J, K, L): Likewise.
1240 * config/visium/predicates.md (const_shift_operand): Likewise.
1241 * config/visium/visium.c (visium_legitimize_address,
1242 visium_legitimize_reload_address): Likewise.
1243 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1244 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1245 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1246 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1247 SET_DUMP_DETAIL): Likewise.
1248 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1249 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1250 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1251 avr_set_core_architecture, avr_set_current_function,
1252 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1253 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1254 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1255 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1256 avr_map_decompose, avr_fold_builtin): Likewise.
1257 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1258 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1259 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1260 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1261 * config/m32c/m32c.c (m32c_conditional_register_usage,
1262 m32c_address_cost): Likewise.
1263 * config/m32c/predicates.md (shiftcount_operand,
1264 longshiftcount_operand): Likewise.
1265 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1266 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1267 can_use_cdx_ldstw): Likewise.
1268 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1269 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1271 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1272 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1273 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1274 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1275 * config/vax/vax.c (adjacent_operands_p): Likewise.
1276 * config/ft32/constraints.md (L, b, KA): Likewise.
1277 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1279 * cfgexpand.c (expand_stack_alignment): Likewise.
1280 * gcse.c (insert_expr_in_table): Likewise.
1281 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1282 * cgraphunit.c (cgraph_node::expand): Likewise.
1283 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1284 * emit-rtl.c (add_insn): Likewise.
1285 * input.c (dump_location_info): Likewise.
1286 * passes.c (NEXT_PASS): Likewise.
1287 * read-rtl-function.c (parse_note_insn_name,
1288 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1290 * sched-rgn.c (sched_rgn_init): Likewise.
1291 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1292 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1293 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1294 * explow.c (eliminate_constant_term): Likewise.
1295 * final.c (leaf_renumber_regs_insn): Likewise.
1296 * cfgrtl.c (print_rtl_with_bb): Likewise.
1297 * genhooks.c (emit_init_macros): Likewise.
1298 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1299 * tree-data-ref.c (conflict_fn): Likewise.
1300 * selftest.c (assert_streq): Likewise.
1301 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1302 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1303 fold_binary_loc, multiple_of_p): Likewise.
1304 * reload.c (push_reload, find_equiv_reg): Likewise.
1305 * et-forest.c (et_nca, et_below): Likewise.
1306 * dbxout.c (dbxout_symbol_location): Likewise.
1307 * reorg.c (relax_delay_slots): Likewise.
1308 * dojump.c (do_compare_rtx_and_jump): Likewise.
1309 * gengtype-parse.c (type): Likewise.
1310 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1311 simplify_const_relational_operation): Likewise.
1312 * reload1.c (do_output_reload): Likewise.
1313 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1314 * gengtype.c (type_for_name): Likewise.
1315 * gimple-ssa-sprintf.c (format_directive): Likewise.
1317 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1320 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1321 accessing it. Adjust comment.
1323 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1326 * predict.c (zero_one_minusone): New function.
1327 (apply_return_prediction): Avoid return prediction for functions
1328 returning only -1, 0 and 1 values, unless they only return -1 and 0
1331 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1333 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1334 fall-through warning.
1335 (arc_compute_frame_size): Remove unused variables.
1336 (arc_print_operand): Fix fprintif format.
1337 (arc_can_follow_jump): Clean fall-through warning.
1339 2017-12-19 Marek Polacek <polacek@redhat.com>
1342 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1345 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1347 * config/arc/arc.c (overriderregs): New variable.
1348 (arc_override_options): Track fixed/call saved/call options.
1349 (arc_conditional_register_usage): Check against overrideregs
1350 variable whenever we change register properties.
1352 2017-12-19 Nathan Sidwell <nathan@acm.org>
1354 * opts.c (finish_options): Don't prefix dump_base_name if it
1355 already contains directories.
1357 2017-12-19 Martin Liska <mliska@suse.cz>
1359 PR rtl-optimization/82675
1360 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1361 more element in sbitmap.
1363 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1365 * gimplify.c (gimplify_expr): Use error_operand_p.
1367 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1370 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1373 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1375 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1376 the jump is to a label.
1378 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1380 PR tree-optimization/83444
1381 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1386 * ipa-inline.c (flatten_remove_node_hook): New function.
1387 (ipa_inline): Keep only nodes with flatten attribute at the end of
1388 the array in the order from ipa_reverse_postorder, only walk that
1389 portion of array for flattening, if there is more than one such
1390 node, temporarily register a removal hook and ignore removed nodes.
1392 PR tree-optimization/80631
1393 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1394 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1395 IFN_REDUC_MAX or IFN_REDUC_MIN.
1397 2017-12-18 Jeff Law <law@redhat.com>
1399 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1400 record symbolic equivalences from backedges in the CFG.
1403 2017-11-19 Jeff Law <law@redhat.com>
1405 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1406 of degenerates resulting from ignoring an edge.
1408 2017-12-18 Martin Sebor <msebor@redhat.com>
1411 PR tree-optimization/78450
1412 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1413 (handle_builtin_strlen): Call it.
1415 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1417 PR rtl-optimization/83424
1418 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1420 2017-12-18 Marek Polacek <polacek@redhat.com>
1423 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1424 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1427 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1429 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1430 (umaddsidi4, umaddsidi_split): Likewise.
1432 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1434 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1437 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1440 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1442 2017-12-18 Richard Biener <rguenther@suse.de>
1444 PR tree-optimization/81877
1445 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1446 (outermost_indep_loop): Adjust.
1447 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1448 (can_sm_ref_p): Adjust.
1450 2017-12-18 Richard Biener <rguenther@suse.de>
1453 * tree.c (array_at_struct_end_p): Return true if the underlying
1454 object has space for at least one element in excess of what
1455 the array domain specifies.
1457 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1459 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1460 example to avoid overfull hbox.
1461 * doc/invoke.texi (Option Summary): Add missing @gol.
1462 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1465 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1466 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1468 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1469 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1470 (AArch64 Options): Document -mverbose-cost-dump.
1471 (ARM Options): Likewise, plus -mflip-thumb.
1473 2017-12-17 Martin Sebor <msebor@redhat.com>
1476 * gimple-ssa-warn-restrict.c
1477 (builtin_memref::offset_out_of_bounds): Correct the handling of
1480 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1482 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1484 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1486 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1488 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1490 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1493 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1495 * doc/invoke.texi: Fix some typos.
1497 2017-12-16 Martin Sebor <msebor@redhat.com>
1499 PR tree-optimization/78918
1500 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1501 * builtins.c (check_sizes): Rename...
1502 (check_access): ...to this. Rename function arguments for clarity.
1503 (check_memop_sizes): Adjust names.
1504 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1505 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1506 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1507 (check_strncat_sizes, expand_builtin_strncat): Same.
1508 (expand_builtin_strncpy, expand_builtin_memset): Same.
1509 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1510 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1511 (maybe_emit_sprintf_chk_warning): Same.
1512 (expand_builtin_strcpy): Adjust.
1513 (expand_builtin_stpcpy): Same.
1514 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1515 in pointer-checking forms of memcpy, memmove, and mempcpy.
1516 (gcall_to_tree_minimal, max_object_size): Define new functions.
1517 * builtins.h (max_object_size): Declare.
1518 * calls.c (alloc_max_size): Call max_object_size instead of
1519 hardcoding ssizetype limit.
1520 (get_size_range): Handle new argument.
1521 * calls.h (get_size_range): Add a new argument.
1522 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1523 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1524 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1526 (gimple_fold_builtin_memory_chk): Same.
1527 (gimple_fold_builtin_stxcpy_chk): New function.
1528 * gimple-ssa-warn-restrict.c: New source.
1529 * gimple-ssa-warn-restrict.h: New header.
1530 * gimple.c (gimple_build_call_from_tree): Propagate location.
1531 * passes.def (pass_warn_restrict): Add new pass.
1532 * tree-pass.h (make_pass_warn_restrict): Declare.
1533 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1535 (handle_builtin_strcat): Same.
1536 (strlen_optimize_stmt): Rename...
1537 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1538 stpncpy, strncpy, and their checking forms.
1540 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1542 PR rtl-optimization/82849
1543 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1544 and get_max_loop_iterations_int.
1546 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1547 Alan Hayward <alan.hayward@arm.com>
1548 David Sherwood <david.sherwood@arm.com>
1550 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1551 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1552 size as a poly_uint64.
1553 (mode_for_vector, mode_for_int_vector): Take the number of vector
1554 elements as a poly_uint64.
1555 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1556 size as a poly_uint64.
1557 (mode_for_vector, mode_for_int_vector): Take the number of vector
1558 elements as a poly_uint64.
1560 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1561 Alan Hayward <alan.hayward@arm.com>
1562 David Sherwood <david.sherwood@arm.com>
1564 * machmode.h (MACRO_MODE): New macro.
1565 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1566 * targhooks.c (default_libcall_value, default_secondary_reload)
1567 (default_memory_move_cost, default_register_move_cost)
1568 (default_class_max_nregs): Likewise.
1570 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1571 Alan Hayward <alan.hayward@arm.com>
1572 David Sherwood <david.sherwood@arm.com>
1574 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1575 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1577 * genautomata.c (main): Likewise.
1578 * genconditions.c (write_header): Likewise.
1579 * genemit.c (main): Likewise.
1580 * genextract.c (print_header): Likewise.
1581 * genopinit.c (main): Likewise.
1582 * genoutput.c (output_prologue): Likewise.
1583 * genpeep.c (main): Likewise.
1584 * genpreds.c (write_insn_preds_c): Likewise.
1585 * genrecog.c (writer_header): Likewise.
1586 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1587 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1588 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1589 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1590 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1591 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1592 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1593 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1594 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1595 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1596 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1597 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1598 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1599 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1600 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1601 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1602 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1603 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1604 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1605 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1606 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1607 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1608 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1609 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1610 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1611 * config/darwin.c (IN_TARGET_CODE): Likewise.
1612 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1613 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1614 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1615 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1616 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1617 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1618 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1619 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1620 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1621 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1622 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1623 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1624 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1625 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1626 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1627 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1628 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1629 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1630 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1631 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1632 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1633 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1634 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1635 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1636 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1637 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1638 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1639 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1640 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1641 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1642 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1643 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1644 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1645 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1646 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1647 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1648 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1649 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1650 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1651 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1652 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1653 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1654 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1655 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1656 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1657 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1658 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1659 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1660 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1661 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1662 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1663 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1664 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1665 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1666 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1667 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1668 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1669 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1670 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1671 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1672 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1673 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1674 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1675 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1676 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1677 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1678 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1679 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1680 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1681 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1682 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1683 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1684 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1685 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1686 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1687 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1688 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1689 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1690 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1691 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1692 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1693 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1694 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1695 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1696 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1697 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1698 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1699 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1700 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1701 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1702 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1703 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1704 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1705 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1706 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1707 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1708 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1709 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1710 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1711 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1712 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1713 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1714 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1715 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1716 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1717 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1719 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1720 Alan Hayward <alan.hayward@arm.com>
1721 David Sherwood <david.sherwood@arm.com>
1723 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1726 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1727 Alan Hayward <alan.hayward@arm.com>
1728 David Sherwood <david.sherwood@arm.com>
1730 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1731 * doc/md.texi (vec_series@var{m}): Document.
1732 * tree.def (VEC_SERIES_EXPR): New tree code.
1733 * tree.h (build_vec_series): Declare.
1734 * tree.c (build_vec_series): New function.
1735 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1736 * tree-pretty-print.c (dump_generic_node): Likewise.
1737 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1738 * tree-inline.c (estimate_operator_cost): Likewise.
1739 * expr.c (expand_expr_real_2): Likewise.
1740 * optabs-tree.c (optab_for_tree_code): Likewise.
1741 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1742 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1743 * expmed.c (make_tree): Handle VEC_SERIES.
1744 * optabs.def (vec_series_optab): New optab.
1745 * optabs.h (expand_vec_series_expr): Declare.
1746 * optabs.c (expand_vec_series_expr): New function.
1747 * tree-vect-generic.c (expand_vector_operations_1): Check that
1748 the operands also have vector type.
1750 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1751 Alan Hayward <alan.hawyard@arm.com>
1752 David Sherwood <david.sherwood@arm.com>
1754 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1755 (VEC_COND_EXPR): Add missing @tindex.
1756 * doc/md.texi (vec_duplicate@var{m}): Document.
1757 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1758 * tree.c (build_vector_from_val): Add stubbed-out handling of
1759 variable-length vectors, using VEC_DUPLICATE_EXPR.
1760 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1761 * cfgexpand.c (expand_debug_expr): Likewise.
1762 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1763 * tree-inline.c (estimate_operator_cost): Likewise.
1764 * tree-pretty-print.c (dump_generic_node): Likewise.
1765 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1766 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1767 (test_vec_duplicate_folding): New function.
1768 (fold_const_c_tests): Call it.
1769 * optabs.def (vec_duplicate_optab): New optab.
1770 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1771 * optabs.h (expand_vector_broadcast): Declare.
1772 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1773 vec_duplicate_optab.
1774 * expr.c (store_constructor): Try using vec_duplicate_optab for
1776 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1778 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1781 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1782 div/mod latencies a bit.
1784 2017-12-15 Jeff Law <law@redhat.com>
1786 PR tree-optimization/36550
1787 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1788 (mark_threaded_blocks): Rewrite code to avoid block copying when
1789 optimizing for size. Don't pessimize blocks which will be
1790 copied, but all the statements will be dead.
1792 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1794 PR tree-optimization/81165
1795 * tree-ssa-threadupdate.c (uses_in_bb): New.
1796 (estimate_threading_killed_stmts): New.
1797 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1798 * tree-ssa-threadedge.c
1799 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1802 2017-12-15 Jeff Law <law@redhat.com>
1804 PR tree-optimization/83410
1805 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1806 threads when parallelizing loops.
1808 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1810 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1812 * config/alpha/alpha.c (vms_attribute_table): Swap
1813 affects_type_identity and handler fields, adjust comments.
1814 * config/mips/mips.c (mips_attribute_table): Likewise.
1815 * config/visium/visium.c (visium_attribute_table): Likewise.
1816 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1817 * config/microblaze/microblaze.c (microblaze_attribute_table):
1819 * config/spu/spu.c (spu_attribute_table): Likewise.
1820 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1821 * config/arc/arc.c (arc_attribute_table): Likewise.
1822 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1823 * config/v850/v850.c (v850_handle_interrupt_attribute,
1824 v850_handle_data_area_attribute): Formatting fixes.
1825 (v850_attribute_table): Swap affects_type_identity and handler
1826 fields, adjust comments.
1827 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1828 * config/arm/arm.c (arm_attribute_table): Likewise.
1829 * config/avr/avr.c (avr_attribute_table): Likewise.
1830 * config/s390/s390.c (s390_attribute_table): Likewise.
1831 * config/sh/sh.c (sh_attribute_table): Likewise.
1832 * config/i386/i386.c (ix86_handle_cconv_attribute,
1833 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1834 (ix86_attribute_table): Swap affects_type_identity and handler
1835 fields, adjust comments.
1836 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1837 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1838 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1839 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1840 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1841 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1842 * config/rx/rx.c (rx_attribute_table): Likewise.
1843 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1844 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1845 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1846 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1847 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1848 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1849 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1850 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1851 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1852 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1853 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1855 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1857 (register_scoped_attributes, decl_attributes): Formatting fixes.
1859 PR tree-optimization/83269
1860 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
1861 subtraction in arg0's type if type is signed and arg0 is unsigned.
1865 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
1866 convert? on @0 instead of convert. Check type of @1, not @0.
1867 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
1868 convert? on @0 instead of convert. Check type of @1, not @0.
1869 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
1870 only optimize if either both @1 and @2 types are narrower
1871 precision, or both are wider or equal precision, and in the former
1872 case only if both have undefined overflow.
1874 2017-12-15 Richard Biener <rguenther@suse.de>
1877 * internal-fn.def (IFN_NOP): Add.
1878 * internal-fn.c (expand_NOP): Do nothing.
1879 * lto-streamer-in.c (input_function): Instead of removing
1880 sanitizer calls replace them with IFN_NOP calls.
1882 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
1883 Alan Hayward <alan.hayward@arm.com>
1884 David Sherwood <david.sherwood@arm.com>
1886 * dse.c (store_info, read_info_type): Replace begin and end with
1888 (print_range): New function.
1889 (set_all_positions_unneeded, any_positions_needed_p)
1890 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
1892 (record_store): Likewise. Optimize the case in which all positions
1894 (get_stored_val): Replace read_begin and read_end with read_offset
1896 (replace_read): Update call accordingly.
1898 2017-12-15 Bin Cheng <bin.cheng@arm.com>
1900 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
1901 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
1902 (loop_cand::loop_cand): Initialize above members.
1903 (loop_cand::supported_operations): Delete.
1904 (loop_cand::can_interchange_p): Inline above function.
1905 (loop_cand::classify_simple_reduction): Record number of constant
1906 initialized simple reductions.
1907 (should_interchange_loops): New parameters. Check stmt cost of loops
1909 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
1910 Update call to should_interchange_loops.
1911 (should_interchange_loop_nest): Update call to
1912 should_interchange_loops.
1914 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
1917 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
1918 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
1919 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
1921 2017-12-15 Julia Koval <julia.koval@intel.com>
1923 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
1924 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
1925 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
1926 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
1927 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
1929 2017-12-15 Julia Koval <julia.koval@intel.com>
1931 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
1932 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
1933 * config/i386/sse.md (vaesenc_<mode>): New pattern.
1934 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
1935 _mm_aesenc_epi128): New intrinsics.
1937 2017-12-15 Julia Koval <julia.koval@intel.com>
1939 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
1940 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
1941 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
1942 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
1943 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
1945 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1947 * gimple-ssa-strength-reduction.c (analyze_increments):
1948 Distinguish replacement costs for constant strides from those for
1951 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1953 * var-tracking.c (variable_tracking_main_1): Formatting fix.
1955 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
1957 * doc/invoke.texi: Document -Wcast-function-type.
1958 * recog.h (stored_funcptr): Change signature.
1959 * tree-dump.c (dump_node): Avoid warning.
1960 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
1962 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
1965 * gimple-fold.c (get_range_strlen): Add the handling of non-member
1968 2017-12-14 David Malcolm <dmalcolm@redhat.com>
1970 PR tree-optimization/83312
1971 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
1972 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
1973 "val" param, and to cope with arbitrary basic blocks.
1974 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
1975 handle NULL_TREE for "val", dropping "bb" param.
1976 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
1977 drop "bb" param. Handle NULL_TREE for "val".
1978 (find_case_label_for_value): Make "switch_stmt" param const.
1979 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
1981 (vrp_prop::check_all_array_refs): Reimplement as...
1982 (check_array_bounds_dom_walker::before_dom_children): ...this new
1983 vfunc. Replace linear search through BB block list, excluding
1984 those with non-executable in-edges via dominator walk.
1986 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1988 * config/arm/arm.opt (mverbose-cost-dump): New option.
1989 * config/arm/arm.c (arm_rtx_costs): Use it.
1991 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
1994 * reload1.c (emit_input_reload_insns): Skip debug markers.
1996 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
1998 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2002 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2003 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2004 * config/sh/sh.c: Skip debug insns besides notes.
2005 * config/sh/sh.md: Likewise.
2006 * config/sh/sh_treg_combine.cc: Likewise.
2007 * config/sh/sync.md: Likewise.
2009 2017-12-14 Tom de Vries <tom@codesourcery.com>
2011 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2012 Add item for weak_undefined.
2014 2017-12-14 Richard Biener <rguenther@suse.de>
2016 PR tree-optimization/67842
2017 * sese.h (bb_in_region): Remove #if 0'ed code.
2019 2017-12-14 Richard Biener <rguenther@suse.de>
2021 PR tree-optimization/83326
2022 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2023 may_be_zero parameter and handle it by not marking the first
2024 peeled copy as not exiting the loop.
2025 (try_peel_loop): Likewise.
2026 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2027 to handle the case of constant or zero iterations and perform
2028 loop header copying on-the-fly.
2030 2017-12-14 Richard Biener <rguenther@suse.de>
2032 PR tree-optimization/83418
2033 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2034 Instead of asserting we don't get unfolded comparisons deal with
2037 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2040 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2041 basic blocks. Assert debug bind insns don't appear outside of bbs,
2042 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2045 PR tree-optimization/83198
2046 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2047 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2048 value if arg is a REAL_CST with incompatible type.
2050 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2051 Bin Cheng <bin.cheng@arm.com>
2054 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2056 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2059 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2061 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2062 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2064 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2065 Alan Hayward <alan.hayward@arm.com>
2066 David Sherwood <david.sherwood@arm.com>
2068 * poly-int.h: New file.
2069 * poly-int-types.h: Likewise.
2070 * coretypes.h: Include them.
2071 (POLY_INT_CONVERSION): Define.
2072 * target.def (estimated_poly_value): New hook.
2073 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2074 * doc/tm.texi: Regenerate.
2075 * doc/poly-int.texi: New file.
2076 * doc/gccint.texi: Include it.
2077 * doc/rtl.texi: Describe restrictions on subreg modes.
2078 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2079 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2080 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2081 * targhooks.h (default_estimated_poly_value): Declare.
2082 * targhooks.c (default_estimated_poly_value): New function.
2083 * target.h (estimated_poly_value): Likewise.
2084 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2085 (wi::unary_traits): Delete.
2086 (wi::binary_traits::signed_shift_result_type): Define for
2087 offset_int << HOST_WIDE_INT, etc.
2088 (generic_wide_int::operator <<=): Define for all types and use
2089 wi::lshift instead of <<.
2090 (wi::hwi_with_prec): Add a default constructor.
2091 (wi::ints_for): New class.
2092 (operator <<): Define for all wide-int types.
2093 (operator /): New function.
2094 (operator %): Likewise.
2095 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2096 (ASSERT_MAYBE_NE_AT): New macros.
2098 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2099 Dominik Vogt <vogt@linux.vnet.ibm.com>
2102 * emit-rtl.c (init_emit): Remove ??? comment.
2103 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2104 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2105 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2106 alignment of 3 virtual registers to BITS_PER_WORD.
2108 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2110 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2112 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2113 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2115 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2118 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2120 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2122 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2123 with the exception of debug begin stmt markers appear before
2127 * final.c (rest_of_handle_final): Call variable_tracking_main only
2128 if !flag_var_tracking.
2130 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2131 Jakub Jelinek <jakub@redhat.com>
2135 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2136 labels debug stmts that can only appear after labels.
2138 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2140 PR rtl-optimization/82398
2141 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2142 EXPR_USEFULNESS in priority comparison.
2144 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2146 PR rtl-optimization/83393
2147 * combine.c (move_deaths): If reg_stat points to a too new insn in
2148 last_death, do not use it: find the proper insn instead.
2150 2017-12-12 Jeff Law <law@redhat.com>
2152 PR tree-optimization/83298
2153 PR tree-optimization/83362
2154 PR tree-optimization/83383
2155 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2156 push_value_range a public interface. Add new argument to
2157 record_ranges_from_stmt.
2158 * gimple-ssa-evrp-analyze.c
2159 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2160 Update comments. Handle recording temporary equivalences.
2161 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2162 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2163 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2164 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2165 gimple-ssa-evrp-analyze.h.
2166 (record_temporary_equivalences_from_phis): Add new argument. When
2167 the PHI arg is an SSA_NAME, set the result's range to the range
2169 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2170 from statements too.
2171 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2172 Pass it down to children as needed.
2173 (thread_outgoing_edges): Likewise.
2174 (thread_across_edge): Likewise. Push/pop range state as needed.
2175 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2177 2017-12-12 Julia Koval <julia.koval@intel.com>
2179 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2180 (PTA_CANNONLAKE): Remove PTA_CLWB.
2182 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2185 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2186 (vcondv2div2df): Likewise.
2187 (vconduv2dfv2di): Likewise.
2189 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2191 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2192 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2193 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2194 ATTR_MATHFN_FPROUNDING_ERRNO.
2196 2017-12-12 Richard Biener <rguenther@suse.de>
2198 PR tree-optimization/83385
2199 * graphite-scop-detection.c (get_order, order): Remove.
2200 (bb_to_rpo): New global.
2202 (build_scops): Sort pbbs in RPO order.
2204 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2206 * combine.c (simplify_set): Do not transform subregs to zero_extends
2207 if the destination mode is a vector mode.
2209 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2211 PR tree-optimization/83359
2212 * tree-cfg.h (fold_loop_internal_call): Declare.
2213 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2214 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2215 (find_loop_dist_alias): New function.
2216 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2217 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2220 PR tree-optimization/80631
2221 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2222 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2223 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2224 hardcoding zero as the value if COND_EXPR is never true. For
2225 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2226 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2227 hardcoding MAX_EXPR as the reduction operation.
2228 (is_nonwrapping_integer_induction): Allow negative step.
2229 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2230 vect_create_epilog_for_reduction, if no value is suitable, don't
2231 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2233 2017-12-12 Richard Biener <rguenther@suse.de>
2235 PR tree-optimization/81889
2236 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2237 range info from the non-wrapping IV instead of just the range
2240 2017-12-12 Julia Koval <julia.koval@intel.com>
2242 * config.gcc: Add vaesintrin.h.
2243 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2245 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2246 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2248 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2249 * config/i386/immintrin.h: Include vaesintrin.h.
2250 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2251 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2252 _mm_aesdec_epi128): New intrinsics.
2254 2017-12-12 Julia Koval <julia.koval@intel.com>
2256 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2257 OPTION_MASK_ISA_VAES_UNSET): New.
2258 (ix86_handle_option): Handle -mvaes.
2259 * config/i386/cpuid.h: Define bit_VAES.
2260 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2261 * config/i386/i386-c.c (__VAES__): New.
2262 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2263 (ix86_valid_target_attribute_inner_p): Ditto.
2264 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2265 * config/i386/i386.opt: Add -mvaes.
2266 * doc/invoke.texi: Ditto.
2268 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2270 * debug.h (gcc_debug_hooks): Add inline_entry.
2271 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2272 * debug.c (do_nothing_debug_hooks): Likewise.
2273 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2274 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2275 (dwarf2_lineno_debug_hooks): Likewise.
2277 * common.opt (gstatement-frontiers): New, setting
2278 debug_nonbind_markers_p.
2279 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2280 * toplev.c (process_options): Autodetect value for debug statement
2282 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2283 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2285 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2286 markers. Integrate source bind into debug stmt expand loop.
2287 (pass_expand::execute): Check debug marker limit. Avoid deep
2288 TER and expand debug locations for debug bind insns only.
2289 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2291 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2292 * final.c (reemit_insn_block_notes): Take current block from
2293 nonbind markers. Declare note where it's first set.
2294 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2295 begin stmt markers if enabled.
2296 (notice_source_line): Handle nonbind markers. Fail if their
2297 location is unknown or that of builtins.
2298 (rest_of_handle_final): Convert begin stmt markers to notes if
2299 var-tracking didn't run.
2300 (rest_of_clean_state): Skip begin stmt markers.
2301 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2303 * function.c (allocate_struct_function): Set begin_stmt_markers.
2304 * function.h (struct function): Add debug_marker_count counter
2305 and debug_nonbind_markers flag.
2306 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2307 * gimple-low.c (lower_function_body): Adjust
2308 debug_nonbind_markers.
2309 (lower_stmt): Drop or skip gimple debug stmts.
2310 (lower_try_catch): Skip debug stmts.
2311 * gimple.c (gimple_build_debug_begin_stmt): New.
2312 (gimple_copy): Increment debug_marker_count if copying one.
2313 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2314 * gimplify.c (rexpr_location): New.
2315 (rexpr_has_location): New.
2316 (warn_switch_unreachable_r): Handle gimple debug stmts.
2317 (shortcut_cond_r): Call expr_location.
2319 (find_goto_label): New.
2320 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2322 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2324 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2325 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2326 (LANG_HOOKS_INITIALIZER): ... this.
2327 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2328 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2330 (update_ebb_live_info): Skip debug insn markers.
2331 * lra.c (debug_insn_static_data): Rename to...
2332 (debug_bind_static_data): ... this.
2333 (debug_marker_static_data): New.
2334 (lra_set_insn_recog_data): Select one of the above depending
2336 (lra_update_isn_regno_info): Don't assume debug insns have
2338 (push_insns): Skip debug insns.
2339 * lto-streamer-in.c (input_function): Drop debug stmts
2340 depending on active options. Adjust debug_nonbind_markers.
2341 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2342 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2343 begin stmt marker notes.
2344 (print_insn): Likewise.
2345 * recog.c (extract_insn): Recognize rtl for debug markers.
2346 * rtl.def (DEBUG_MARKER): New.
2347 * tree-inline.c: Include params.h.
2348 (remap_gimple_stmt): Handle nonbind markers.
2349 (maybe_move_debug_stmts_to_successors): Likewise.
2350 (copy_debug_stmt): Likewise.
2351 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2352 markers regardless of no side effects.
2353 (tsi_link_before): Don't update container's side effects when adding
2354 a begin stmt marker.
2355 (tsi_link_after): Likewise.
2356 (expr_first): Skip begin stmt markers.
2357 (expr_last): Likewise.
2358 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2359 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2360 Disregard nonbind markers.
2361 * tree.c (make_node_stat): Don't set side effects for begin stmt
2363 (build1_stat): Likewise.
2364 * tree.def (DEBUG_BEGIN_STMT): New.
2365 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2366 * var-tracking.c (delete_debug_insns): Renamed to...
2367 (delete_vta_debug_insns): ... this.
2368 (reemit_marker_as_note): New.
2369 (vt_initialize): Reemit markers.
2370 (delete_vta_debug_insns): Likewise.
2371 (vt_debug_insns_local): Reemit or delete markers.
2372 (variable_tracking_main_1): Likewise.
2373 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2374 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2375 (gimple_debug_nonbind_marker_p): New.
2376 (gimple_build_debug_bind): Adjust.
2377 (gimple_build_debug_begin_stmt): New.
2378 * doc/invoke.texi (max-debug-marker-count): New param.
2379 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2380 (debug_parameter_ref, debug_marker): New.
2381 (NOTE_INSN_BEGIN_STMT): New.
2382 (DEBUG_INSN): Describe begin stmt markers.
2384 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2385 without debug insns, we wouldn't, but clean up debug insns
2386 after a control flow insn nevertheless.
2388 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2389 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2390 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2391 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2392 (rtl_tidy_fallthru_edge): Likewise.
2393 (rtl_verify_fallthru): Likewise.
2394 (rtl_verify_bb_layout): Likewise.
2395 (skip_insns_after_block): Likewise.
2396 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2397 * dwarf2out.c: Include print-rtl.h.
2398 (dwarf2out_next_real_insn): New.
2399 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2400 Dump debug binds in asm comments.
2401 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2402 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2403 callers to use gsi_start_nondebug_bb instead.
2404 (gsi_after_labels): Skip gimple debug stmts.
2405 (gsi_start_nondebug): New.
2406 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2407 (proper_loop_form_for_interchange): Adjust.
2408 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2409 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2410 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2411 (collect_fallthrough_labels): Likewise.
2412 (should_warn_for_implicit_fallthrough): Likewise.
2413 (warn_implicit_fallthrough_r): Likewise.
2414 (expand_FALLTHROUGH_r): Likewise.
2415 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2416 (graphite_copy_stmts_from_block): Skip nonbind markers.
2417 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2418 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2419 * jump.c (clean_barriers): Skip debug insns.
2420 * omp-expand.c (expand_parallel_call): Skip debug insns.
2421 (expand_task_call): Likewise.
2422 (remove_exit_barrier): Likewise.
2423 (expand_omp_taskreg): Likewise.
2424 (expand_omp_for_init_counts): Likewise.
2425 (expand_omp_for_generic): Likewise.
2426 (expand_omp_for_static_nochunk): Likewise.
2427 (expand_omp_for_static_chunk): Likewise.
2428 (expand_omp_simd): Likewise.
2429 (expand_omp_taskloop_for_outer): Likewise.
2430 (expand_omp_taskloop_for_inner): Likewise.
2431 (expand_oacc_for): Likewise.
2432 (expand_omp_sections): Likewise.
2433 (expand_omp_single): Likewise.
2434 (expand_omp_synch): Likewise.
2435 (expand_omp_atomic_load): Likewise.
2436 (expand_omp_atomic_store): Likewise.
2437 (expand_omp_atomic_fetch_op): Likewise.
2438 (expand_omp_atomic_pipeline): Likewise.
2439 (expand_omp_atomic_mutex): Likewise.
2440 (expand_omp_target): Likewise.
2441 (grid_expand_omp_for_loop): Likewise.
2442 (grid_expand_target_grid_body): Likewise.
2443 (build_omp_regions_1): Likewise.
2444 * omp-low.c (check_combined_parallel): Skip debug stmts.
2445 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2446 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2447 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2449 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2450 (make_edges): Likewise.
2451 (cleanup_dead_labels): Likewise.
2452 (gimple_can_merge_blocks_p): Likewise.
2453 (stmt_starts_bb_p): Likewise.
2454 (gimple_block_label): Likewise.
2455 (gimple_redirect_edge_and_branch): Likewise.
2456 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2458 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2460 * tree-pretty-print (print_declaration): Omit initializer in slim
2462 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2464 (eliminate_unnecessary_stmts): Stabilize block removal order.
2465 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2466 * var-tracking.c (get_first_insn): New.
2467 (vt_emit_notes): Call it.
2468 (vt_initialize): Walk any insns before the first BB.
2469 (delete_debug_insns): Likewise.
2471 * gimple.h (enum gimple_debug_subcode): Add
2472 GIMPLE_DEBUG_BEGIN_STMT.
2473 (gimple_debug_begin_stmt_p): New.
2474 (gimple_debug_nonbind_marker_p): New.
2475 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2476 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2477 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2478 * insn-notes.def (BEGIN_STMT): New.
2479 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2480 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2481 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2482 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2483 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2484 (INSN_DEBUG_MARKER_KIND): New.
2485 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2486 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2487 (INSN_VAR_LOCATION_PTR): New.
2488 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2490 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2491 deps only if debug bind insns are enabled.
2492 (pass_expand::execute): Avoid deep TER and expand
2493 debug locations for debug bind insns only.
2494 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2495 debug stmts special handling down to debug bind stmts.
2496 * combine.c (try_combine): Narrow debug insns special handling
2497 down to debug bind insns.
2498 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2499 Narrow debug insns preexisting special handling down to debug
2501 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2502 handling down to debug bind insns.
2503 * function.c (instantiate_virtual_regs): Skip debug markers,
2504 adjust handling of debug binds.
2505 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2506 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2507 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2508 insns to debug bind insns.
2509 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2510 special handling of debug stmts to debug bind stmts.
2511 * ipa-split.c (split_function): Likewise.
2512 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2513 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2515 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2516 * regrename.c (build_def_use): Likewise.
2517 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2518 (pass_cprop_hardreg): Narrow special casing of debug insns to
2520 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2521 * reload1.c (reload): Likewise.
2522 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2523 casing of debug stmts to debug bind stmts.
2524 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2525 * ssa-iterators.h (num_imm_uses): Likewise.
2526 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2527 debug stmts to debug bind stmts.
2528 * tree-inline.c (tree_function_versioning): Narrow special casing
2529 of debug stmts to debug bind stmts.
2530 * tree-loop-distribution.c (generate_loops_for_partition):
2531 Narrow special casing of debug stmts to debug bind stmts.
2532 * tree-sra.c (analyze_access_subtree): Narrow special casing
2533 of debug stmts to debug bind stmts.
2534 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2535 stmts to debug bind stmts.
2536 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2537 casing of debug stmts to debug bind stmts.
2538 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2539 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2540 casing of debug stmts to debug bind stmts.
2541 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2543 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2544 debug stmts to debug bind stmts.
2545 (gimple_replace_ssa_lhs): Likewise.
2546 (insert_debug_temp_for_var_def): Likewise.
2547 (insert_debug_temps_for_defs): Likewise.
2548 (reset_debug_uses): Likewise.
2549 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2550 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2551 (adjust_debug_stmts): Likewise.
2552 (adjust_phi_and_debug_stmts): Likewise.
2553 (vect_do_peeling): Likewise.
2554 * tree-vect-loop.c (vect_transform_loop): Likewise.
2555 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2556 * var-tracking.c (adjust_mems): Narrow special casing of debug
2557 insns to debug bind insns.
2558 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2559 (compute_bb_dataflow, vt_find_locations): Likewise.
2560 (vt_expand_loc, emit_notes_for_changes): Likewise.
2561 (vt_init_cfa_base): Likewise.
2562 (vt_emit_notes): Likewise.
2563 (vt_initialize): Likewise.
2564 (vt_finalize): Likewise.
2566 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2567 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2568 (next_nonnote_nondebug_insn_bb): New.
2569 (prev_nonnote_nondebug_insn_bb): New.
2570 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2571 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2572 (prev_nonnote_nondebug_insn_bb): Declare.
2573 (next_nonnote_nondebug_insn_bb): Declare.
2574 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2575 * cfgrtl.c (get_last_bb_insn): Likewise.
2576 * lra.c (push_insns): Likewise.
2578 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2581 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2583 * selftest.c (selftest::test_runner::test_runner): New ctor.
2584 (selftest::test_runner::~test_runner): New dtor.
2585 * selftest.h (class selftest::test_runner): New class.
2587 2017-12-11 Carl Love <cel@us.ibm.com>
2589 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2590 vec_extract_fp32_from_shortl]): Add #defines.
2591 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2592 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2593 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2594 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2595 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2596 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2597 * doc/extend.texi: Add documentation for the added builtins.
2599 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2601 PR rtl-optimization/80693
2602 PR rtl-optimization/81019
2603 PR rtl-optimization/81020
2604 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2605 are not mentioned in i3. Place the REG_UNUSED note on i2,
2606 possibly modified to REG_DEAD, if it did not originate in i3.
2608 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2610 * recog.c (store_data_bypass_p_1): New function.
2611 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2612 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2614 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2616 PR rtl-optimization/83361
2617 * ifcvt.c (if_convert): Call fixup_partitions.
2619 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2621 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2622 early folding of splat_u{8,16,32}.
2624 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2626 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2627 output_operand_lossage first argument with capital letter.
2628 (aarch64_override_options): Don't start error and sorry first argument
2629 with capital letter.
2631 2017-12-11 Andi Kleen <ak@linux.intel.com>
2633 PR gcov-profile/83355
2634 * auto-profile.c (string_table::get_index_by_decl): Don't
2635 recurse when abstract origin points to itself.
2637 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2639 PR tree-optimization/83320
2640 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2641 (prune_datarefs_not_in_loop): Ditto.
2643 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2645 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2647 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2649 PR tree-optimization/83337
2650 * gimple-loop-interchange.cc (compute_access_stride): Handle
2651 bitfield DRs properly.
2653 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2655 PR tree-optimization/83338
2656 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2657 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2658 vector of unsigned integers to vector of signed integers.
2660 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2662 PR rtl-optimization/83317
2663 * lra-constraints.c (process_address_1): Add insn code check.
2665 2017-12-08 Michael Matz <matz@suse.de>
2667 Fix PR tree-optimization/83323
2668 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2669 head-controlled loops and loop BBs.
2670 * common.opt (funroll-and-jam): Remove, instead ...
2671 (floop-unroll-and-jam): ... reuse this option.
2672 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2673 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2674 (-floop-unroll-and-jam): ... this option.
2676 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2678 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2679 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2680 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2682 2017-12-08 Richard Biener <rguenther@suse.de>
2685 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2686 handle accesses outside of zero-sized vars.
2688 2017-12-08 Martin Jambor <mjambor@suse.cz>
2690 PR tree-optimization/83141
2691 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2692 test for MEM_REFs implicitely changing types with padding. Remove
2694 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2696 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2698 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2700 * config/avr/avr.c (avr_attribute_table): Likewise.
2701 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2702 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2703 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2704 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2705 of struct attribute_spec.
2706 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2708 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2711 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2713 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2716 * profile-count.c (profile_count::from_gcov_type): Move from
2717 profile-count.h; handle overflow.
2718 * profile-count.h (profile_count::from_gcov_type): Move offline.
2720 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2722 PR rtl-optimization/83304
2723 * combine.c (move_deaths): If we do not know where a register died,
2726 2017-12-08 Richard Biener <rguenther@suse.de>
2728 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2729 Provide -fopt-info-loop feedback when we interchange in a nest.
2731 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2733 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2734 for armv6 ARM CPU IDs.
2736 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2738 * common/config/arm/arm-common.c: Include <algorithm>.
2739 (INCLUDE_VECTOR): Define.
2740 (compare_opt_names): New function.
2741 (arm_rewrite_selected_arch): Only strip out extensions that can be
2742 expressed through -mfpu. Sort the remaining extensions
2745 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2747 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2748 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2749 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2750 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2752 2017-06-08 Tristan Gingold <gindold@adacore.com>
2755 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2756 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2758 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2760 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2761 of struct attribute_spec.
2763 2017-12-08 Julia Koval <julia.koval@intel.com>
2765 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2766 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2767 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2768 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2769 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2770 _mm_maskz_dpwssds_epi32): Ditto.
2772 2017-12-08 Richard Biener <rguenther@suse.de>
2774 PR tree-optimization/81303
2775 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2776 conditions try to create a comparison vector type matching
2777 the data vector type.
2778 (vectorizable_condition): Adjust.
2779 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2780 Leave invariant conditions alone in case we can vectorize those.
2782 2017-12-08 Julia Koval <julia.koval@intel.com>
2784 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2785 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2786 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2787 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2788 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2790 2017-12-08 Julia Koval <julia.koval@intel.com>
2792 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2793 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2794 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2795 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2796 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2797 _mm_maskz_dpbusds_epi32): New intrinsics.
2799 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2801 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2803 (ld<bh>io_signed): Likewise.
2804 (st<bhw>io): Likewise.
2805 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2806 SMALL_INT12 constant integer operand.
2808 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2810 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2812 (nios2_large_constant_allowed): ...this. Adjust uses.
2813 (nios2_plus_symbolic_constant_p): Rename to...
2814 (nios2_plus_large_constant_p): ...this. Adjust uses.
2815 (nios2_legitimate_address_p): Correct CONST_INT handling.
2816 (nios2_symbolic_memory_operand_p): Rename to...
2817 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2818 (nios2_large_constant_p): Check for large constant integers too.
2819 (nios2_split_large_constant): Handle constant integers.
2820 (nios2_split_symbolic_memory_operand): Rename to...
2821 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2822 (nios2_legitimize_constant_address): Handle constant integers.
2823 (r0rel_constant_p): Handle small constant integers.
2824 (nios2_print_operand_address): Handle r0-relative integer addresses.
2825 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2826 * config/nios2/nios2.md: Adjust for renamed functions.
2828 2017-12-07 Andrew Waterman <andrew@sifive.com>
2830 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2831 (TARGET_HAVE_SRODATA_SECTION): New define.
2832 (riscv_select_section): New function.
2834 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2835 Alexander Monakov <amonakov@ispras.ru>
2836 Jakub Jelinek <jakub@redhat.com>
2839 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2841 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2843 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2844 a bool success value. Don't call output_operand_lossage here.
2845 (aarch64_print_ldpstp_address): Return a bool success value.
2846 (aarch64_print_operand_address): Call output_addr_const if
2847 aarch64_print_address_internal fails.
2848 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2849 'y'; call output_operand_lossage instead. Call output_operand_lossage
2850 if aarch64_print_ldpstp_address fails.
2852 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2854 * tree-vector-builder.h
2855 (tree_vector_builder::binary_encoded_nelts): Declare.
2856 * tree-vector-builder.c
2857 (tree_vector_builder::binary_encoded_nelts): New function.
2858 * fold-const.c (negate_expr_p): Likewise.
2859 (operand_equal_p, fold_checksum_tree): Likewise.
2860 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
2861 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
2862 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
2863 (uniform_vector_p): Likewise.
2864 * varasm.c (const_hash_1, compare_constant): Likewise.
2865 * tree-ssa-ccp.c: Include tree-vector-builder.h.
2866 (valid_lattice_transition): Operate directly on the VECTOR_CST
2868 * ipa-icf.c: Include tree-vector-builder.h.
2869 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
2870 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
2872 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2874 * tree.c (build_vector): Delete.
2875 * tree.h (build_vector): Make static and move into the self-testing
2878 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2880 * vector-builder.h (vector_builder::derived): New const overload.
2881 (vector_builder::elt): New function.
2882 * tree-vector-builder.h (tree_vector_builder::type): New function.
2883 (tree_vector_builder::apply_step): Declare.
2884 * tree-vector-builder.c (tree_vector_builder::apply_step): New
2886 * gimple-fold.h (tree_vector_builder): Declare.
2887 (gimple_build_vector): Take a tree_vector_builder instead of a
2888 type and vector of elements.
2889 * gimple-fold.c (gimple_build_vector): Likewise.
2890 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
2892 (get_initial_defs_for_reduction): Likewise.
2893 (vectorizable_induction): Likewise.
2895 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2897 * tree-vector-builder.h
2898 (tree_vector_builder::new_binary_operation): Declare.
2899 * tree-vector-builder.c
2900 (tree_vector_builder::new_binary_operation): New function.
2901 * fold-const.c (fold_relational_const): Use it.
2902 (const_binop): Likewise. Check that both input vectors have
2903 the same number of elements, thus excluding things like WIDEN_SUM.
2904 Check whether it is possible to operate directly on the encodings
2907 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2909 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
2910 new_unary_operation, operating only on the encoded elements.
2911 (const_unop): Likewise.
2912 (exact_inverse): Likewise.
2913 (distributes_over_addition_p): New function.
2914 (const_binop): Use tree_vector_builder and new_unary_operation
2915 for combinations of VECTOR_CST and INTEGER_CST. Operate only
2916 on the encoded elements unless the encoding is strided and the
2917 operation does not distribute over addition.
2918 (fold_convert_const): Use tree_vector_builder and
2919 new_unary_operation. Operate only on the encoded elements
2920 for truncating integer conversions, or for non-stepped encodings.
2922 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2924 * config/sparc/sparc.c: Include tree-vector-builder.h.
2925 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
2926 * expmed.c: Include tree-vector-builder.h.
2927 (make_tree): Use tree_vector_builder instead of build_vector.
2928 * fold-const.c: Include tree-vector-builder.h.
2929 (const_binop): Use tree_vector_builder instead of build_vector.
2930 (const_unop): Likewise.
2931 (native_interpret_vector): Likewise.
2932 (fold_vec_perm): Likewise.
2933 (fold_ternary_loc): Likewise.
2934 * gimple-fold.c: Include tree-vector-builder.h.
2935 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
2937 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
2938 (simplify_vector_constructor): Use tree_vector_builder instead
2940 * tree-vect-generic.c: Include tree-vector-builder.h.
2941 (add_rshift): Use tree_vector_builder instead of build_vector.
2942 (expand_vector_divmod): Likewise.
2943 (optimize_vector_constructor): Likewise.
2944 * tree-vect-loop.c: Include tree-vector-builder.h.
2945 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
2946 of build_vector. Explicitly use a stepped encoding for
2948 * tree-vect-slp.c: Include tree-vector-builder.h.
2949 (vect_get_constant_vectors): Use tree_vector_builder instead
2951 (vect_transform_slp_perm_load): Likewise.
2952 (vect_schedule_slp_instance): Likewise.
2953 * tree-vect-stmts.c: Include tree-vector-builder.h.
2954 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
2955 (vect_gen_perm_mask_any): Likewise.
2956 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
2957 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
2959 (build_vector_from_val): Likewise. Explicitly use a duplicate
2962 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
2964 * doc/generic.texi (VECTOR_CST): Describe new representation of
2966 * vector-builder.h: New file.
2967 * tree-vector-builder.h: Likewise.
2968 * tree-vector-builder.c: Likewise.
2969 * Makefile.in (OBJS): Add tree-vector-builder.o.
2970 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
2971 * tree-core.h (tree_base): Add a vector_cst field to the u union.
2972 (tree_vector): Change the number of elements to
2973 vector_cst_encoded_nelts.
2974 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
2975 (VECTOR_CST_ELTS): Delete.
2976 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
2977 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
2978 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
2979 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
2980 (VECTOR_CST_ENCODED_ELT): Likewise.
2981 (vector_cst_encoded_nelts): New function.
2982 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2983 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2984 (vector_cst_int_elt, vector_cst_elt): Declare.
2985 * tree.c: Include tree-vector-builder.h.
2986 (tree_code_size): Abort if passed VECTOR_CST.
2987 (tree_size): Update for new VECTOR_CST layout.
2988 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2989 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2990 (build_vector): Use tree_vector_builder.
2991 (vector_cst_int_elt, vector_cst_elt): New functions.
2992 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
2993 encoded elements and then create the vector in the canonical form.
2994 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
2995 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
2996 (tree_c_tests): Call test_vector_cst_patterns.
2997 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
2999 (hash_tree): Likewise.
3000 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3001 (streamer_write_tree_header): Likewise.
3002 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3003 (streamer_alloc_tree): Likewise. Update call to make_vector.
3004 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3006 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3008 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3009 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3010 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3012 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3013 Richard Biener <rguenther@suse.de>
3015 PR tree-optimization/81303
3016 * Makefile.in (gimple-loop-interchange.o): New object file.
3017 * common.opt (floop-interchange): Reuse the option from graphite.
3018 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3019 -floop-interchange and mention it for -O3.
3020 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3021 * gimple-loop-interchange.cc: New file.
3022 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3023 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3024 * passes.def (pass_linterchange): New pass.
3025 * timevar.def (TV_LINTERCHANGE): New time var.
3026 * tree-pass.h (make_pass_linterchange): New declaration.
3027 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3028 interchange. Record IV before/after increment in new parameters.
3029 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3030 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3032 (check_reduction_path): ...New function here.
3033 * tree-vectorizer.h (check_reduction_path): New declaration.
3035 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3038 PR rtl-optimization/80818
3039 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3040 always early clobbered.
3041 * lra-lives.c (process_bb_lives): Check input hard regs for early
3042 clobbered non-operand hard reg.
3044 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3047 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3048 types_compatible_p, just that TYPE_MODE is the same.
3050 2017-12-07 Martin Sebor <msebor@redhat.com>
3053 * attribs.c (empty_attribute_table): Initialize new member of
3054 struct attribute_spec.
3055 (decl_attributes): Add argument. Handle mutually exclusive
3056 combinations of attributes.
3057 (selftests::test_attribute_exclusions): New function.
3058 (selftests::attribute_c_tests): Ditto.
3059 * attribs.h (decl_attributes): Add default argument.
3060 * selftest.h (attribute_c_tests): Declare.
3061 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3062 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3064 * doc/extend.texi (Common Function Attributes): Update const and pure.
3065 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3066 of struct attribute_spec.
3067 * config/arc/arc.c (arc_attribute_table): Same.
3068 * config/arm/arm.c (arm_attribute_table): Same.
3069 * config/avr/avr.c ( avr_attribute_table): Same.
3070 * config/bfin/bfin.c (bfin_attribute_table): Same.
3071 * config/cr16/cr16.c (cr16_attribute_table): Same.
3072 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3073 * config/h8300/h8300.c (h8300_attribute_table): Same.
3074 * config/i386/i386.c (ix86_attribute_table): Same.
3075 * config/ia64/ia64.c (ia64_attribute_table): Same.
3076 * config/m32c/m32c.c (m32c_attribute_table): Same.
3077 * config/m32r/m32r.c (m32r_attribute_table): Same.
3078 * config/m68k/m68k.c (m68k_attribute_table): Same.
3079 * config/mcore/mcore.c (mcore_attribute_table): Same.
3080 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3081 * config/mips/mips.c (mips_attribute_table): Same.
3082 * config/msp430/msp430.c (msp430_attribute_table): Same.
3083 * config/nds32/nds32.c (nds32_attribute_table): Same.
3084 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3085 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3086 * config/rl78/rl78.c (rl78__attribute_table): Same.
3087 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3088 * onfig/rx/rx.c (rx_attribute_table): Same.
3089 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3090 * config/sh/sh.c (sh_attribute_table): Same.
3091 * config/sparc/sparc.c (sparc_attribute_table): Same.
3092 * config/spu/spu.c (spu_attribute_table): Same.
3093 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3094 * config/v850/v850.c (v850_attribute_table): Same.
3095 * config/visium/visium.c (visium_attribute_table): Same.
3097 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3100 * config/arm/arm.c (INCLUDE_STRING): Define.
3101 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3102 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3105 2017-12-07 Michael Matz <matz@suse.de>
3107 Add unroll and jam pass
3109 * gimple-loop-jam.c: New file.
3110 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3111 * common.opt (funroll-and-jam): New option.
3112 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3113 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3114 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3115 * passes.def: Add pass_loop_jam.
3116 * timevar.def (TV_LOOP_JAM): Add.
3117 * tree-pass.h (make_pass_loop_jam): Declare.
3118 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3119 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3120 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3121 to flow_loop_tree_node_add.
3122 (duplicate_subloops, copy_loops_to): Append to sibling list.
3123 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3124 * doc/invoke.texi (-funroll-and-jam): Document new option.
3125 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3127 2017-12-07 Richard Biener <rguenther@suse.de>
3129 PR tree-optimization/83296
3130 PR tree-optimization/67769
3131 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3132 flow sensitive info in an unrelated BB.
3133 (value_replacement): Use reset_flow_sensitive_info.
3134 (minmax_replacement): Reset flow sensitive info on the def
3135 we move. Do not reset flow sensitive info in the whole BB
3136 we move the stmt to.
3137 (abs_replacement): Likewise.
3139 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3142 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3143 rs6000_cpu to the given -mcpu=, or to the default processor.
3145 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3147 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3148 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3149 instead of rs6000_cpu_attr.
3150 (rs6000_adjust_cost): Ditto.
3151 (is_microcoded_insn): Ditto.
3152 (rs6000_adjust_priority): Ditto.
3153 (rs6000_issue_rate): Ditto.
3154 (rs6000_use_sched_lookahead): Ditto.
3155 (rs6000_use_sched_lookahead_guard): Ditto.
3156 (rs6000_sched_reorder): Ditto.
3157 (force_new_group): Ditto.
3158 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3159 (group_ending_nop): Ditto.
3161 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3163 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3164 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3165 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3166 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3168 (rs6000_reassociation_width): Ditto.
3169 (rs6000_emit_epilogue): Ditto.
3170 (rs6000_adjust_cost): Ditto.
3171 (is_microcoded_insn): Ditto.
3172 (is_cracked_insn): Ditto.
3173 (rs6000_adjust_priority): Ditto.
3174 (rs6000_sched_reorder): Ditto.
3175 (rs6000_sched_reorder2): Ditto.
3176 (insn_must_be_first_in_group): Ditto.
3177 (insn_must_be_last_in_group): Ditto.
3178 (rs6000_register_move_cost): Ditto.
3179 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3182 2017-12-07 Julia Koval <julia.koval@intel.com>
3184 * config.gcc: Add vaesintrin.h.
3185 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3187 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3188 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3190 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3191 * config/i386/immintrin.h: Include vaesintrin.h.
3192 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3193 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3194 _mm_aesdec_epi128): New intrinsics.
3196 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3198 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3199 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3200 c-family/c-spellcheck.cc.
3201 (best_macro_match::best_macro_match): Likewise.
3203 (struct edit_distance_traits<cpp_hashnode *>): Move to
3204 c-family/c-spellcheck.h.
3205 (class best_macro_match): Likewise.
3207 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3209 PR tree-optimization/83293
3210 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3211 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3212 might insert into empty bb.
3215 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3216 simplify for plus with :c added, and pointer_plus without that.
3217 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3218 with undefined overflow and the conversion is not widening,
3219 perform negation in utype and only convert to type afterwards.
3220 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3221 simplify for plus with :c added, and pointer_plus without that.
3222 If type is integral with undefined overflow and the conversion is
3223 not widening, perform minus in utype and only convert to type
3224 afterwards. Move the last pointer_diff_expr simplify into the
3227 2017-12-06 Martin Sebor <msebor@redhat.com>
3229 PR tree-optimization/82646
3230 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3231 strncpy, not maxlen.
3233 2017-12-06 Martin Sebor <msebor@redhat.com>
3235 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3238 PR tree-optimization/83075
3239 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3240 strncat/strncpy don't change length of source string.
3242 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3245 2017-11-29 Martin Aberg <maberg@gaisler.com>
3247 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3248 to prevent b2bst errata sequence.
3249 (sqrtdf2_fix): Likewise.
3251 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3253 PR tree-optimization/81945
3254 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3255 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3256 to dest_cfun has orig_loop_num set, either remap it to the new
3257 loop number if the loop got moved too, or clear it.
3259 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3261 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3263 (thunderx2t99_nothing): New insn reservation.
3264 (thunderx2t99_mrs): New insn reservation.
3265 (thunderx2t99_multiple): New insn reservation.
3266 (thunderx2t99_alu_basi): Add bfx to reservation.
3267 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3269 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3272 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3274 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3276 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3277 tree-ssa-pre.c::remove_dead_inserted_code.
3278 * tree-ssa-dce.h: New file.
3279 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3280 (remove_dead_inserted_code): Move and rename to function
3281 tree-ssa-dce.c::simple_dce_from_worklist.
3282 (pass_pre::execute): Update use.
3284 2017-12-05 Richard Biener <rguenther@suse.de>
3286 PR tree-optimization/83277
3287 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3288 to code-gen liveout vars.
3290 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3292 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3293 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3294 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3297 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3299 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3301 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3302 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3305 2017-12-05 Richard Biener <rguenther@suse.de>
3307 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3308 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3309 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3310 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3312 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3314 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3316 2017-12-05 Martin Liska <mliska@suse.cz>
3317 Jakub Jelinek <jakub@redhat.com>
3319 * doc/invoke.texi: Document the options.
3320 * flag-types.h (enum sanitize_code): Add
3321 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3322 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3323 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3324 * opts.c: Define new sanitizer options.
3325 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3326 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3328 2017-12-05 Julia Koval <julia.koval@intel.com>
3330 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3331 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3332 (ix86_handle_option): Handle -mavx512vnni.
3333 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3334 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3335 * config/i386/i386-c (__AVX512VNNI__): New.
3336 * config/i386/i386.c (ix86_target_string): Handle new option.
3337 (ix86_valid_target_attribute_inner_p): Handle new option.
3338 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3339 * config/i386/i386.opt (mavx512vnni): New option.
3341 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3344 * athlon.md: Disable for generic.
3345 * haswell.md: Enable for generic.
3346 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3347 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3349 (ix86_adjust_cost): Move generic to haswell path.
3351 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3353 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3354 instead of MEM_P in a couple more places. Fix formatting issues.
3356 2017-12-04 Jim Wilson <jimw@sifive.com>
3358 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3359 instead of GP_REG_LAST-1.
3360 (riscv_adjust_libcall_cfi_prologue): Likewise.
3361 (riscv_adjust_libcall_cri_epilogue): Likewise.
3362 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3365 2017-12-04 Luis Machado <luis.machado@linaro.org>
3367 * ipa-pure-const.c (check_decl): Add missing newline.
3368 (state_from_flags): Likewise.
3370 2017-12-04 Jeff Law <law@redhat.com>
3372 PR tree-optimizatin/78496
3373 * gimple-ssa-evrp-analyze.h
3374 (evrp_range_analyzer::get_vr_values): Simplify.
3375 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3376 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3377 and gimple-ssa-evrp-analyze.h.
3378 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3379 (simplify_stmt_for_jump_threading): Copy a blob of code from
3380 tree-vrp.c to use ranges to simplify statements.
3381 (dom_opt_dom_walker::before_dom_children): Call
3382 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3383 (dom_opt_dom_walker::after_dom_children): Similarly for
3384 evrp_range_analyzer::leave.
3385 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3388 * gimple-ssa-evrp-analyze.c
3389 (evrp_range_analyzer::extract_range_from_stmt): Always use
3390 vr_values::update_value_range so preexisting range info is
3391 medged with new range info, even if the new range is VR_VARYING.
3393 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3395 * combine.c: Adjust comment.
3396 (use_crosses_set_p): Delete.
3397 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3398 (try_combine): Ditto.
3400 2017-12-04 Richard Biener <rguenther@suse.de>
3402 PR tree-optimization/83255
3403 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3404 Re-add zero-iteration check.
3406 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3408 PR rtl-optimization/83245
3409 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3410 hard registers as earlyclobber, also if not in an asm.
3412 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3416 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3419 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3420 rs6000_cpu based on cpu_index, not tune_index.
3422 2017-12-04 Richard Biener <rguenther@suse.de>
3424 PR tree-optimization/83238
3425 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3426 code match comment, rejecting invalid SESE regions.
3428 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3430 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3431 require base operand is a REG_POINTER prior to reload on targets
3432 with non-equivalent space registers.
3434 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3436 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3437 (update_specialized_profile): Fix updating of counts.
3438 (perhaps_add_new_callers): Likewise.
3440 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3443 * x86-tune.def: Remove obsolette FIXMEs.
3444 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3445 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3446 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3448 (X86_TUNE_PAD_RETURNS): Disable for generic.
3450 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3452 PR tree-optimization/83170
3453 PR tree-optimization/83241
3454 * gimple-ssa-store-merging.c
3455 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3456 gimple_vuse (ins_stmt) in case it has changed.
3457 (imm_store_chain_info::output_merged_store): Likewise.
3459 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3463 * tree-cfg.c (pass_warn_function_return::execute): Handle
3464 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3465 with BUILTINS_LOCATION.
3469 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3470 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3471 instead of DECL_MODE.
3473 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3474 last argument to rtx pointer.
3475 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3476 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3477 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3478 depending on the chosen ISAs.
3479 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3480 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3481 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3483 * config/i386/sse.md (mov<mode>_internal): Likewise.
3484 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3486 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3488 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3490 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3492 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3493 parameters from prototype.
3494 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3495 parameters. Don't print a comment.
3496 (emit_fusion_gpr_load): Adjust.
3497 (emit_fusion_load_store): Adjust.
3498 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3499 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3500 comment on the second line.
3502 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3505 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3506 rs6000_cpu based on cpu_index, not tune_index.
3508 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3510 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3511 which_alternative instead of which_alternative + 1.
3512 (output_asm_insn): Print an extra tab if the template is short.
3514 2017-12-01 Jim Wilson <jimw@sifive.com>
3516 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3518 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3519 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3521 * doc/tm.texi: Regenerate.
3523 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3526 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3527 whether we can allocate pseudos before trying to fix an address.
3528 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3529 memory address is indexed or indirect.
3530 (floatuns_<mode>si2_hw2): Likewise.
3532 2017-12-01 Jason Merrill <jason@redhat.com>
3534 * Makefile.in (TAGS): Add c-family/*.cc.
3536 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3538 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3539 (call_value_insn): Likewise.
3540 (sibcall_insn): Likewise.
3541 (sibcall_value_insn): Likewise.
3542 (movsi_aarch64): Likewise.
3543 (movdi_aarch64): Likewise.
3544 (add_losym_): Likewise.
3545 (ldr_got_small_): Likewise.
3546 (ldr_got_small_sidi): Likewise.
3547 (ldr_got_small_28k_): Likewise.
3548 (ldr_got_small_28k_sidi): Likewise.
3549 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3550 Move output_addr_const to symbolic case. Add error check.
3552 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3554 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3555 * config/s390/s390.md: Disable MVC merging peephole if it would
3556 disable operand forwarding.
3557 (new peephole2): Split MVCs if it would turn them into up to 2
3560 2017-12-01 Richard Biener <rguenther@suse.de>
3562 PR tree-optimization/83232
3563 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3564 detection of same access. Instead of breaking the group here
3565 do not consider the duplicate. Add comment explaining real fix.
3567 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3569 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3571 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3573 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3575 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3577 * function.h (struct function): Remove cilk_frame_decl,
3578 is_cilk_function and calls_cilk_spawn fields.
3579 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3581 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3582 cilk_elemental field.
3583 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3584 * target.def: Adjust comment.
3585 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3586 Don't test cilk_elemental.
3588 PR tree-optimization/83233
3589 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3590 bswap_stat name for the struct.
3593 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3594 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3595 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3596 SWITCH_BREAK_LABEL_P set on the label.
3597 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3598 added for default case if it was missing and not all cases covered.
3599 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3600 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3602 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3603 to avoid -Wimplicit-fallthrough warning.
3604 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3605 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3607 PR tree-optimization/83221
3608 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3610 (init_reassoc): Formatting fix.
3613 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3614 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3618 * tree-cfg.c: Include asan.h.
3619 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3621 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3623 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3625 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3627 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3628 vpcmpeqd instruction.
3630 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3632 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3633 registers type for 128bit mode.
3635 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3637 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3639 * gengtype-state.c (read_state_pair): Likewise.
3640 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3641 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3642 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3643 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3644 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3645 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3646 * builtins.c (fold_builtin_3): Likewise.
3647 * graphite-scop-detection.c
3648 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3649 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3652 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3655 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3656 parameter. Return true if ignore_decl_rtl_set_p is true and other
3657 conditions are satisfied.
3658 * asan.h (asan_protect_global): Add new parameter.
3659 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3660 to asan_protect_global calls.
3662 2017-11-30 Jim Wilson <jimw@sifive.com>
3664 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3665 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3666 -mstrict-align, add info on default value. Delete redundant lines for
3667 -mabi. Add missing -mexplicit-relocs docs.
3669 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3671 * config/arc/arc.md (trap): New pattern.
3673 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3675 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3676 instruction to end into a delay slot.
3677 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3678 can be placed into a delay slot against reg_note.
3680 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3682 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3683 labels number of usages.
3685 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3687 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3689 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3691 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3694 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3695 multiplication by power of 2 constant into two shifts + comparison.
3697 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3700 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3702 2017-11-30 Richard Biener <rguenther@suse.de>
3704 PR tree-optimization/83202
3705 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3706 allow_peel argument and guard peeling.
3707 (canonicalize_loop_induction_variables): Likewise.
3708 (canonicalize_induction_variables): Pass false.
3709 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3710 peeling from cunrolli.
3712 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3714 * combine.c (try_combine): Print a message to dump file whenever
3715 I0, I1, or I2 cannot be combined into I3.
3717 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3719 PR rtl-optimization/83156
3720 PR rtl-optimization/82621
3721 * combine.c (try_combine): Don't split an I2 if one of the dests is
3722 set again before I3. Allow unused dests.
3724 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3726 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3728 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3730 PR rtl-optimization/80818
3731 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3732 recursively. Use insn code for clobber.
3733 (lra_set_insn_recog_data): Pass the new arg to
3734 collect_non_operand_hard_regs.
3735 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3737 (lra_update_insn_regno_info): Pass insn to
3738 add_regs_to_insn_regno_info.
3740 2017-11-29 Jim Wilson <jimw@sifive.com>
3741 Andrew Waterman <andrew@sifive.com>
3743 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3744 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3745 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3746 use SINGLE_SHIFT_COST.
3747 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3748 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3750 2017-11-29 Julia Koval <julia.koval@intel.com>
3752 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3753 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3754 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3755 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3756 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3757 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3758 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3759 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3760 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3761 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3762 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3763 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3764 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3765 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3766 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3767 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3768 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3769 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3770 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3771 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3772 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3773 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3774 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3775 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3776 __builtin_ia32_vpshldv_v2di_mask,
3777 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3778 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3779 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3781 2017-11-29 Julia Koval <julia.koval@intel.com>
3783 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3784 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3785 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3786 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3787 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3788 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3789 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3790 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3791 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3792 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3793 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3794 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3795 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3796 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3797 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3798 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3799 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3800 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3801 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3802 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3803 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3805 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3807 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3808 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3811 2017-11-29 Martin Aberg <maberg@gaisler.com>
3813 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3814 to prevent b2bst errata sequence.
3815 (sqrtdf2_fix): Likewise.
3817 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3819 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3820 (div_sqrt_insn_p): New function.
3821 (sparc_do_work_around_errata): Insert NOP instructions to
3822 prevent sequences that could trigger the TN-0013 errata for
3823 certain LEON3 processors.
3824 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3825 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3826 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3827 (in_branch_delay): Prevent div and sqrt in delay slot if
3829 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3831 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3833 * config/sparc/sparc.c (atomic_insn_p): New function.
3834 (sparc_do_work_around_errata): Insert NOP instructions to
3835 prevent sequences that could trigger the TN-0010 errata for
3837 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3838 instruction referable in atomic_insns_p.
3840 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3842 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3843 (atomic_compare_and_swap_leon3_1): Likewise.
3846 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3848 * config/sparc/sparc.c (fpop_insn_p): New function.
3849 (sparc_do_work_around_errata): Insert NOP instructions to
3850 prevent sequences that could trigger the TN-0012 errata for
3852 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3853 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3854 (in_branch_annul_delay): Prevent floating-point instructions
3855 in delay slot of annulled integer branch.
3857 2017-11-29 Richard Biener <rguenther@suse.de>
3859 PR tree-optimization/83202
3860 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
3862 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
3863 nodes vectorized to the same stmts multiple times.
3864 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
3865 (vect_analyze_slp_instance): Adjust.
3866 (scalar_stmts_to_slp_tree_map_t): New typedef.
3867 (vect_schedule_slp_instance): Add a map recording the SLP node
3868 representing the vectorized stmts for a set of scalar stmts.
3869 Avoid code-generating redundancies.
3870 (vect_schedule_slp): Allocate map and pass it down.
3872 2017-11-29 Nathan Sidwell <nathan@acm.org>
3875 * tree.c (build_complex_type): Fix canonicalization. Only fill in
3878 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3880 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
3881 cases for printing LDP/STP memory addresses.
3882 (aarch64_print_address_internal): Renamed from
3883 aarch64_print_operand_address, added parameter, add Pmode check.
3884 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
3885 (aarch64_print_operand_address): Indirect to
3886 aarch64_print_address_internal.
3887 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
3888 'y' operand output specifier.
3890 2017-11-29 Jakub Jelinek <jakub@redhat.com>
3893 * tree.c (build_simple_mem_ref_loc): Handle
3894 get_addr_base_and_unit_offset returning a MEM_REF.
3897 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
3900 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
3903 2017-11-28 David Malcolm <dmalcolm@redhat.com>
3905 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
3907 (layout::move_to_column): Likewise.
3909 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3912 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
3913 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
3915 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3916 Martin Jambor <mjambor@suse.cz>
3919 * tree.h (expr_type_first_operand_type_p): Declare
3920 * tree.c (expr_type_first_operand_type_p): New function.
3921 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
3922 (ipa_value_from_jfunc): Adjust declaration.
3923 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
3924 Use it as result type for arithmetics, unless it is NULL in which case
3925 be more conservative.
3926 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
3927 ipa_get_jf_pass_through_result.
3928 (propagate_vals_across_pass_through): Likewise.
3929 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
3930 is to propagate_vals_across_pass_through.
3931 (propagate_constants_across_call): Pass PARM_TYPE to
3932 propagate_scalar_across_jump_function.
3933 (find_more_scalar_values_for_callers_subset): Pass parameter type to
3934 ipa_value_from_jfunc.
3935 (cgraph_edge_brings_all_scalars_for_node): Likewise.
3936 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
3937 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
3938 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
3939 target_type, pass it to ipa_value_from_jfunc.
3940 (update_indirect_edges_after_inlining): Pass parameter type to
3941 try_make_edge_direct_simple_call.
3943 2017-11-28 Jeff Law <law@redhat.com>
3945 * gimple-ssa-evrp-analyze.c
3946 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
3947 refine ranges if scev_initialized_p returns true.
3948 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
3950 2017-11-28 Julia Koval <julia.koval@intel.com>
3952 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
3953 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
3954 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
3955 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
3956 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
3957 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
3958 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
3959 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
3960 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
3961 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
3962 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
3963 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
3964 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
3965 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
3966 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
3967 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
3968 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
3969 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
3970 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
3971 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
3972 __builtin_ia32_vpshrd_v2di_mask): New builtins.
3973 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
3975 2017-11-28 Julia Koval <julia.koval@intel.com>
3977 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
3978 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
3979 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
3980 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
3981 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
3982 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
3983 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
3984 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
3985 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
3986 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
3987 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
3988 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
3989 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
3990 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
3991 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
3992 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
3993 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
3994 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
3995 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
3996 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
3997 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
3998 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
3999 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4000 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4001 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4002 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4003 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4004 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4005 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4006 __builtin_ia32_vpshld_v2di_mask): New builtins.
4007 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4008 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4010 2017-11-28 Richard Biener <rguenther@suse.de>
4012 PR tree-optimization/80776
4013 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4015 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4017 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4018 If the incoming edge is an effective fallthru because the other
4019 edge only reaches a __builtin_unreachable () then record ranges
4020 derived from the controlling condition in SSA info.
4021 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4022 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4024 2017-11-28 Olivier Hainque <hainque@adacore.com>
4026 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4027 on mingw build hosts.
4029 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4032 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4034 2017-11-28 Richard Biener <rguenther@suse.de>
4037 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4038 copies generated from memcpy use a character array as reference
4041 2017-11-28 Julia Koval <julia.koval@intel.com>
4042 Sebastian Peryt <sebastian.peryt@intel.com>
4044 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4045 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4046 cilk-common.o, cilk.h, cilk-common.c): Remove.
4048 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4049 * builtins.c (is_builtin_name): Remove cilkplus condition.
4050 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4051 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4052 cilk-builtins.def, cilkplus.def): Remove.
4053 * cif-code.def (CILK_SPAWN): Remove.
4054 * cilk-builtins.def: Delete.
4055 * cilk-common.c: Ditto.
4057 * cilkplus.def: Ditto.
4058 * config/darwin.h (fcilkplus): Delete.
4059 * cppbuiltin.c: Ditto.
4060 * doc/extend.texi: Remove cilkplus doc.
4061 * doc/generic.texi: Ditto.
4062 * doc/invoke.texi: Ditto.
4063 * doc/passes.texi: Ditto.
4064 * gcc.c (fcilkplus): Remove.
4065 * gengtype.c (cilk.h): Remove.
4066 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4068 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4070 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4072 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4073 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4074 cilkplus conditions.
4075 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4076 inline_read_section): Ditto.
4077 * ipa-inline-analysis.c (cilk.h): Remove.
4078 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4079 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4080 append_linker_options): Remove condition for fcilkplus.
4081 * lto/lto-lang.c (cilk.h): Remove.
4082 (lto_init): Remove condition for fcilkplus.
4083 * omp-expand.c (expand_cilk_for_call): Delete.
4084 (expand_omp_taskreg, expand_omp_for_static_chunk,
4085 expand_omp_for): Remove cilkplus
4087 (expand_cilk_for): Delete.
4088 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4089 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4090 execute_lower_omp, diagnose_sb_0): Ditto.
4091 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4092 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4093 * tree-nested.c: Ditto.
4094 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4095 (dump_generic_node): Ditto.
4096 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4097 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4099 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4101 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4103 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4104 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4105 for vector mode and !TARGET_SIMD.
4107 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4109 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4111 * tree.h (SWITCH_LABELS): Remove.
4112 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4113 assert SWITCH_BODY is non-NULL.
4114 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4116 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4118 PR tree-optimization/80788
4119 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4120 has TREE_OVERFLOW set, call drop_tree_overflow.
4122 2017-11-28 Richard Biener <rguenther@suse.de>
4124 PR tree-optimization/83158
4125 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4127 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4130 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4131 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4133 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4135 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4136 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4138 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4141 * varasm.c (compare_constant): Take the mode of the constants into
4142 account when comparing floating point constants.
4144 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4146 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4147 from explicit instantiation of debug_helper.
4148 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4150 2017-11-27 Richard Biener <rguenther@suse.de>
4152 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4155 2017-11-27 Richard Biener <rguenther@suse.de>
4157 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4158 (type_hash_canon): Also clear int_cst_hash_table entry for
4160 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4162 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4164 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4165 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4166 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4168 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4170 * hash-map.h (gt_cleare_cache): Avoid UB.
4172 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4174 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4175 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4176 (decide_unroll_runtime_iterations): Likewise.
4177 (decide_unroll_stupid): Likewise.
4179 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4182 * config/i386/i386.md: Add a loop with incssp.
4184 2017-11-27 Martin Jambor <mjambor@suse.cz>
4186 PR tree-optimization/81248
4187 * tree-sra.c (splice_param_accesses): Remove size check.
4188 (decide_one_param_reduction): Fix size check.
4189 * gimple-pretty-print.c (dump_profile): Silence warning.
4190 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4192 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4195 * dbxout.c (lastlineno): New variable.
4196 (dbx_debug_hooks): Use dbxout_switch_text_section as
4197 switch_text_section debug hook.
4198 (dbxout_function_end): Switch to current_function_section
4199 rather than function_section. If crtl->has_bb_partition,
4200 output just one N_FUN, depending on in_cold_section_p.
4201 (dbxout_source_line): Remember last lineno in lastlineno.
4202 (dbxout_switch_text_section): New function.
4203 (dbxout_function_decl): Adjust dbxout_block caller.
4204 (dbx_block_with_cold_children): New function.
4205 (dbxout_block): Return true if any LBRAC/RBRAC have been
4206 emitted. Use dbx_block_with_cold_children at depth == 0
4207 in second partition. Add PARENT_BLOCKNUM argument, pass
4208 it optionally adjusted to children. Output LBRAC/RBRAC
4209 around recursive call only if the block is in the current
4210 partition, if not and anything was output, emit empty
4212 * final.c (final_scan_insn): Compute cold_function_name
4213 before calling switch_text_section debug hook. Call
4214 that hook even if dwarf2out_do_frame if not emitting
4218 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4219 TREE_READONLY decls.
4221 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4223 PR rtl-optimization/82488
4224 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4226 2017-11-26 Julia Koval <julia.koval@intel.com>
4228 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4230 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4233 2017-11-26 Julia Koval <julia.koval@intel.com>
4235 * config/i386/driver-i386.c (host_detect_local_cpu):
4236 Detect skylake-avx512.
4238 2017-11-26 Julia Koval <julia.koval@intel.com>
4240 * config.gcc: Add -march=cannonlake.
4241 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4242 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4243 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4244 (PTA_CANNONLAKE): New.
4245 (processor_target_table): Add cannonlake.
4246 (ix86_option_override_internal): Ditto.
4247 (fold_builtin_cpu): Ditto.
4248 (get_builtin_code_for_version): Handle cannonlake.
4249 (M_INTEL_COREI7_CANNONLAKE): New.
4250 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4251 * doc/invoke.texi: Add -march=cannonlake.
4253 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4255 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4256 (try_init_one_plugin): Alternative implementation for MinGW.
4257 * Makefile.in (plugin_implib): New.
4258 (gengtype-lex.c): Fix broken AIX workaround.
4259 * configure: Regenerate.
4260 * doc/plugins.texi: Document support for MinGW.
4262 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4264 PR rtl-optimization/81553
4265 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4266 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4267 is a shift where C1 has different mode than the whole shift, use C1's
4268 mode for MULT rather than the shift's mode.
4271 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4272 builtins not enabled in the currently selected ISA.
4274 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4276 PR tree-optimization/71026
4277 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4278 (insert_reciprocals): Change to insert reciprocals before a division
4279 by a square and to insert the square of a reciprocal.
4280 (execute_cse_reciprocals_1): Change to consider division by a square.
4281 (register_division_in): Add importance parameter.
4283 2017-11-24 Richard Biener <rguenther@suse.de>
4285 PR tree-optimization/82402
4286 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4287 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4289 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4291 * match.pd (0-ptr): New transformation.
4293 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4296 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4297 when skipping checking.
4299 2017-11-24 Richard Biener <rguenther@suse.de>
4301 PR tree-optimization/83128
4302 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4303 (vn_reference_lookup_3): Likewise.
4305 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4308 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4309 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4312 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4313 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4314 check for the latter with gimple_call_builtin_p. Do not handle
4315 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4317 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4319 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4322 2017-11-23 Julia Koval <julia.koval@intel.com>
4324 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4325 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4326 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4327 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4328 _mm512_maskz_expandloadu_epi16): New intrinsics.
4329 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4330 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4331 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4332 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4333 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4334 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4335 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4336 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4337 _mm256_maskz_expandloadu_epi8): New intrinsics.
4338 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4339 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4340 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4341 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4342 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4343 * config/i386/sse.md (VI248_VLBW): New iterator.
4344 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4346 2017-11-23 Julia Koval <julia.koval@intel.com>
4348 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4349 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4350 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4351 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4352 _mm512_mask_compressstoreu_epi16): New.
4353 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4354 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4355 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4356 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4357 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4358 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4359 _mm256_mask_compressstoreu_epi8): New.
4360 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4361 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4362 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4363 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4364 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4365 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4366 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4367 __builtin_ia32_compresshi128_mask,
4368 __builtin_ia32_compressstoreuqi512_mask,
4369 __builtin_ia32_compressstoreuhi512_mask,
4370 __builtin_ia32_compressstoreuqi256_mask,
4371 __builtin_ia32_compressstoreuqi128_mask,
4372 __builtin_ia32_compressstoreuhi256_mask,
4373 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4374 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4376 (ix86_expand_special_args_builtin): Handle new types.
4377 (s4fma_expand): Handle new builtin array.
4378 * config/i386/immintrin.h: Include new headers.
4379 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4380 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4382 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4384 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4385 fixes. Declare temp and g variables at the top in order to avoid
4386 {} in most of the cases.
4388 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4390 * match.pd (ptr-0): New transformation.
4392 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4394 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4395 (struct addr_mode_cost_table): New.
4396 (struct tune_params): Add field addr_mode_costs.
4397 * config/arm/arm.c (generic_addr_mode_costs): New.
4398 (arm_slowmul_tune): Initialise addr_mode_costs field.
4399 (arm_fastmul_tune): Likewise.
4400 (arm_strongarm_tune): Likewise.
4401 (arm_xscale_tune): Likewise.
4402 (arm_9e_tune): Likewise.
4403 (arm_marvell_pj4_tune): Likewise.
4404 (arm_v6t2_tune): Likewise.
4405 (arm_cortex_tune): Likewise.
4406 (arm_cortex_a8_tune): Likewise.
4407 (arm_cortex_a7_tune): Likewise.
4408 (arm_cortex_a15_tune): Likewise.
4409 (arm_cortex_a35_tune): Likewise.
4410 (arm_cortex_a53_tune): Likewise.
4411 (arm_cortex_a57_tune): Likewise.
4412 (arm_exynosm1_tune): Likewise.
4413 (arm_xgene1_tune): Likewise.
4414 (arm_cortex_a5_tune): Likewise.
4415 (arm_cortex_a9_tune): Likewise.
4416 (arm_cortex_a12_tune): Likewise.
4417 (arm_cortex_a73_tune): Likewise.
4418 (arm_v7m_tune): Likewise.
4419 (arm_cortex_m7_tune): Likewise.
4420 (arm_v6m_tune): Likewise.
4421 (arm_fa726te_tune): Likewise.
4422 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4424 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4426 * config/arm/arm.c (arm_mem_costs): New function.
4427 (arm_rtx_costs_internal): Use arm_mem_costs.
4429 2017-11-23 Mark Wielaard <mark@klomp.org>
4431 * dwarf2out.c (init_sections_and_labels): Use generation to create
4432 unique ranges_section_label and ranges_base_label. Return generation.
4433 (output_rnglists): Add generation argument. Use generation to create
4434 unique ranges labels.
4435 (dwarf2out_finish): Get generation from init_sections_and_labels
4436 and pass generation to output_rnglists.
4438 2017-11-23 Mike Stump <mikestump@comcast.net>
4439 Eric Botcazou <ebotcazou@adacore.com>
4441 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4442 * cfgloop.h (struct loop): Add unroll field.
4443 * function.h (struct function): Add has_unroll bitfield.
4444 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4445 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4446 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4447 (pass_rtl_unroll_loops::gate): Likewise.
4448 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4449 for which loop->unroll==1.
4450 (decide_unroll_constant_iterations): Use note for consistency and
4451 take loop->unroll into account. Return early if loop->unroll is set.
4452 Fix thinko in existing test.
4453 (decide_unroll_runtime_iterations): Use note for consistency and
4454 take loop->unroll into account.
4455 (decide_unroll_stupid): Likewise.
4456 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4457 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4458 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4460 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4461 (print_loop): Print loop->unroll if set.
4462 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4463 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4464 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4466 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4467 loop->unroll is set and smaller than the trip count. Otherwise bypass
4468 entirely the heuristics if loop->unroll is set. Remove dead note.
4469 Fix off-by-one bug in other note.
4470 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4471 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4473 (tree_unroll_loops_completely): Make static.
4474 (pass_complete_unroll::execute): Use correct type for variable.
4475 (pass_complete_unrolli::execute): Fix formatting.
4476 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4478 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4480 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4481 enable when TARGET_PREFER_AVX128 is set.
4483 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4485 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4487 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4488 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4490 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4492 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4493 frequencies. Fix estimation of aggregate parameters.
4495 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4497 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4500 2017-11-23 Tom de Vries <tom@codesourcery.com>
4502 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4504 2017-11-23 Tom de Vries <tom@codesourcery.com>
4506 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4509 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4512 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4513 sibcall_value_pcrel_fdpic): Use local variable instead of
4515 (calli_tbr_rel): Add missing operand 2.
4516 (call_valuei_tbr_rel): Add missing operand 3.
4518 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4521 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4522 bitpos/bitsize covering the whole destination, use store_expr only if
4523 the complex mode is the same. Otherwise, use expand_normal and if
4524 it returns CONCAT, subreg each part separately instead of trying to
4525 subreg the whole result.
4527 2017-11-23 Richard Biener <rguenther@suse.de>
4529 PR tree-optimization/23094
4530 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4531 come here from walking over backedges in the first iteration.
4532 (vn_reference_lookup_3): Skip clobbers that store the same value.
4534 2017-11-23 Richard Biener <rguenther@suse.de>
4536 PR tree-optimization/81403
4537 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4538 a block we need a leader relative to.
4539 (phi_translate_1): For nary processing require a leader from
4540 get_representative_for given we run expression simplification
4541 using match-and-simplify. Remove previous fix.
4543 2017-11-22 Jeff Law <law@redhat.com>
4545 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4546 Use new method allocate_value_range rather than accessing the
4547 vrp_value_range_pool data member directly.
4548 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4549 to use extract_range_from_stmt method to avoid need for
4550 extract_range_from_assignment method.
4551 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4552 method rather than setting values_propgated data member directly.
4553 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4554 and values propagated data members and extract_range_from_assignment
4555 method. Reorder private data members to conform to standards.
4556 Add new methods set_lattice_propagation_complete and
4557 allocate_value_range.
4559 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4561 PR rtl-optimization/83030
4562 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4563 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4564 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4565 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4566 CROSSING_JUMP_P flag is already set before setting it.
4567 * cfgrtl.c (fixup_partition_crossing): Likewise.
4568 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4571 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4573 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4574 Handle the case where both arguments are using gen_const_vec_series.
4576 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4579 * pretty-print.c (pp_format): Move quoting implementation to
4580 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4581 to pass address of "quote" local.
4582 (pp_begin_quote): New function.
4583 (pp_end_quote): New function.
4584 * pretty-print.h (printer_fn): Convert penultimate param from bool
4586 (pp_begin_quote): New decl.
4587 (pp_end_quote): New decl.
4588 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4589 param from bool to bool *.
4590 * tree-diagnostic.h (default_tree_printer): Likewise.
4592 2017-11-22 Jeff Law <law@redhat.com>
4594 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4595 Initialize vr_values.
4596 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4597 vr_values via vr_values class instance rather than delegators.
4598 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4599 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4600 (evrp_range_analyzer::push_value_range): Likewise.
4601 (evrp_range_analyzer::pop_value_range): Likewise.
4602 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4603 most delegators. Those remaining are exposed as public interfaces.
4604 Make vr_values a pointer and private.
4605 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4607 (evrp_range_analyzer::get_vr_value): New method.
4608 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4609 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4610 (class evrp_dom_walker): Attach evrp_folder class, initialize
4611 it in the ctor. Remove temporary delegators.
4612 (evrp_dom_walker::before_dom_children): Call methods in attached
4613 evrp_range_analyzer class via class instance pointer. Use
4614 free value_range_constant_singleton to remove need for
4615 op_with_constant_singleton_value delegator method. Do not
4616 create a vrp_prop class instance for every call! Narrow
4617 scope of a couple variables.
4618 (evrp_dom_walker::cleanup): Call methods in attached
4619 evrp_range_analyzer class via class instance pointer.
4620 * vr-values.h (class vr_values): Privatize many methods and
4623 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4625 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4626 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4627 * expr.c (expand_expr_real_2): Likewise.
4628 * fold-const.c (const_unop): Likewise.
4629 * optabs-tree.c (optab_for_tree_code): Likewise.
4630 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4631 * tree-inline.c (estimate_operator_cost): Likewise.
4632 * tree-pretty-print.c (dump_generic_node): Likewise.
4633 (op_code_prio): Likewise.
4634 (op_symbol_code): Likewise.
4635 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4636 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4637 * internal-fn.c (direct_internal_fn_optab): New function.
4638 (direct_internal_fn_array, direct_internal_fn_supported_p
4639 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4640 * fold-const-call.c (fold_const_reduction): New function.
4641 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4643 * tree-vect-loop.c: Include internal-fn.h.
4644 (reduction_code_for_scalar_code): Rename to...
4645 (reduction_fn_for_scalar_code): ...this and return an internal
4647 (vect_model_reduction_cost): Take an internal_fn rather than
4649 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4651 (vectorizable_reduction): Use internal functions rather than tree
4652 codes for the reduction operation. Update calls to the functions
4654 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4655 Use calls to internal functions rather than REDUC tree codes.
4656 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4658 2017-11-22 Olivier Hainque <hainque@adacore.com>
4660 * config/vxworks.c (vxworks_override_options): Pick default
4661 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4662 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4663 DWARF_GNAT_ENCODINGS_DEFAULT.
4664 * config/vxworksae.h: Likewise.
4666 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4668 PR tree-optimization/83104
4669 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4672 2017-11-22 Marek Polacek <polacek@redhat.com>
4673 H.J. Lu <hongjiu.lu@intel.com>
4674 Jason Merrill <jason@redhat.com>
4679 * calls.c (initialize_argument_information): Call
4680 warn_parameter_passing_abi target hook.
4681 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4682 argument onto stack.
4683 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4684 * common.opt: Update -fabi-version description.
4685 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4686 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4688 (ix86_is_empty_record): New function.
4689 (ix86_warn_parameter_passing_abi): New function.
4690 (TARGET_EMPTY_RECORD_P): Redefine.
4691 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4692 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4693 * doc/tm.texi: Regenerated.
4694 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4695 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4696 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4697 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4698 instead of int_size_in_bytes.
4699 * expr.c (copy_blkmode_to_reg): Likewise.
4700 * function.c (aggregate_value_p): Return 0 for empty types.
4701 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4703 (locate_and_pad_parm): Call arg size_in_bytes instead
4705 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4706 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4707 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4709 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4710 (std_gimplify_va_arg_expr): Skip empty records. Call
4711 arg_size_in_bytes instead size_in_bytes.
4712 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4713 * tree-core.h (tree_type_common): Add empty_flag.
4714 (tree_decl_common): Update comments.
4715 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4717 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4718 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4720 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4721 * tree.c (default_is_empty_type): New function.
4722 (default_is_empty_record): New function.
4723 (arg_int_size_in_bytes): New function.
4724 (arg_size_in_bytes): New function.
4725 (get_ultimate_context): New function.
4726 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4727 TRANSLATION_UNIT_WARN_EMPTY_P.
4728 (default_is_empty_record, arg_int_size_in_bytes,
4729 arg_size_in_bytes, get_ultimate_context): Declare.
4731 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4733 * config/arm/arm.c (cmse_clear_registers): New function.
4734 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4735 code by call to cmse_clear_registers.
4736 (cmse_nonsecure_entry_clear_before_return): Likewise.
4738 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4740 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4741 (vdot_s32, vdotq_s32): New.
4742 (vdot_lane_u32, vdotq_lane_u32): New.
4743 (vdot_lane_s32, vdotq_lane_s32): New.
4746 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4749 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4750 for unsigned values with fewer HWIs than the precision.
4751 (test_overflow): New function.
4752 (wide_int_cc_tests): Call it.
4754 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4756 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4757 has the same mode class as Pmode.
4759 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4761 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4762 padding_bits_to_clear_ptr.
4763 (cmse_nonsecure_entry_clear_before_return): Likewise.
4765 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4767 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4768 auto_sbitap instead of integer bitfield to control register needing
4771 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4773 PR tree-optimization/83044
4774 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4775 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4776 that need the upper bound. Subtract offset from
4777 get_addr_base_and_unit_offset only if positive and subtract it
4778 before division by eltsize rather than after it.
4781 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4782 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4783 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4787 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4788 expand_binop with *mul_widen_optab, make sure at least one of the
4789 operands doesn't have VOIDmode.
4792 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4794 PR rtl-optimization/82044
4795 PR tree-optimization/82042
4796 * dse.c (record_store): Check for overflow.
4797 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4798 add_wild_read instead of clear_rhs_from_active_local_stores on
4799 overflow and log it into dump_file.
4801 2017-11-22 Richard Biener <rguenther@suse.de>
4803 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4804 fake edges to exit when looking for a place to insert.
4805 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4807 (insert_into_preds_of_block): Commit edge insertion immediately,
4808 assert that doesn't require new BBs.
4809 (fini_pre): Release expressions.
4810 (pass_pre::execute): Shuffle things around a bit, if the fn
4811 is too large do not compute AVAIL either as this is really the
4814 2017-11-22 Richard Biener <rguenther@suse.de>
4816 PR tree-optimization/83089
4817 * tree-if-conv.c (pass_if_conversion::execute): If anything
4818 changed reset SCEV and free the number of iteration estimates.
4820 2017-11-21 Martin Sebor <msebor@redhat.com>
4822 PR tree-optimization/82945
4823 * calls.h (warn_nonstring_bound): Remove unused function.
4825 2017-11-21 Martin Sebor <msebor@redhat.com>
4827 PR tree-optimization/82945
4828 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4829 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4830 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4832 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4833 * calls.h (get_attr_nonstring_decl): Declare new function.
4834 * doc/extend.texi (attribute nonstring): Update.
4835 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4836 get_attr_nonstring_decl and handle it.
4837 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4838 detection of nul-termination.
4839 (strlen_to_stridx): Change to a pointer.
4840 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4841 (pass_strlen::execute): Same.
4843 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4845 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4846 for the new option -mprefer-vector-width=[none|128|256|512].
4847 * config/i386/i386.c (ix86_target_string): remove old style options
4848 -mprefer-avx256 and make -mprefer-avx128 as alias.
4849 (ix86_option_override_internal): Apply defaults for the
4850 -mprefer-vector-width=[128|256] option.
4851 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4852 Implement macros to work with -mprefer-vector-width=.
4853 * config/i386/i386.opt: Implemented option
4854 -mprefer-vector-width=[none|128|256|512].
4855 * doc/invoke.texi: Documentation for
4856 -mprefer-vector-width=[none|128|256|512].
4858 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
4860 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
4862 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
4864 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
4866 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
4868 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
4870 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
4871 * expr.c (expand_expr_real_2): Likewise.
4872 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
4873 fold_binary_loc): Likewise.
4874 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
4875 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
4876 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
4877 MINUS_EXPR transformations.
4878 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
4879 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
4880 * tree-inline.c (estimate_operator_cost): Likewise.
4881 * tree-pretty-print.c (dump_generic_node, op_code_prio,
4882 op_symbol_code): Likewise.
4883 * tree-vect-stmts.c (vectorizable_operation): Likewise.
4884 * vr-values.c (extract_range_from_binary_expr): Likewise.
4885 * varasm.c (initializer_constant_valid_p_1): Likewise.
4886 * tree.def: New tree code POINTER_DIFF_EXPR.
4888 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
4890 * config/i386/i386.md (*bswap<mode>2_movbe): Add
4891 integer suffix to movbe mnemonic.
4892 (*bswaphi2_movbe): Ditto.
4893 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
4895 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4898 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
4899 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
4900 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
4902 2017-11-21 Martin Liska <mliska@suse.cz>
4904 * tree-inline.c (expand_call_inline): Remove not needed
4907 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
4908 Jakub Jelinek <jakub@redhat.com>
4911 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
4912 Remove static keyword from f variable.
4914 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4916 PR tree-optimization/83086
4917 * gimple-ssa-store-merging.c
4918 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
4919 rather than n.base_addr.
4921 2017-11-21 Martin Liska <mliska@suse.cz>
4923 PR rtl-optimization/82044
4924 PR tree-optimization/82042
4925 * dse.c (check_mem_read_rtx): Check for overflow.
4927 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4929 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
4932 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4934 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
4935 and its corresponding call arg location note.
4937 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4938 Andrew Burgess <andrew.burgess@embecosm.com>
4940 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
4942 (arc_return_slot_offset): Likewise.
4943 (arc_eh_return_address_location): New declaration.
4944 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
4945 (MUST_SAVE_REGISTER): Add exception handler case.
4946 (MUST_SAVE_RETURN_ADDR): Likewise.
4947 (arc_frame_pointer_required): Likewise.
4948 (arc_frame_pointer_needed): New function.
4949 (arc_compute_frame_size): Changed.
4950 (arc_expand_prologue): Likewise.
4951 (arc_expand_epilogue): Likewise.
4952 (arc_initial_elimination_offset): Likewise.
4953 (arc_return_slot_offset): Delete.
4954 (arc_eh_return_address_location): New function.
4955 (arc_builtin_setjmp_frame_value): Likewise.
4956 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
4957 (EH_RETURN_STACKADJ_RTX): Define.
4958 (EH_RETURN_HANDLER_RTX): Likewise.
4959 * config/arc/arc.md (eh_return): Delete.
4961 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4963 * print-tree.h (debug_vec_tree): Remove prototype.
4964 * gdbinit.in (pvt): Remove macro.
4966 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4968 PR tree-optimization/83047
4969 * gimple-ssa-store-merging.c
4970 (imm_store_chain_info::output_merged_store): If the loads with the
4971 same vuse are in different basic blocks, for load_gsi pick a load
4972 location that is dominated by the other loads.
4975 * config/i386/i386.c (ix86_memmodel_check): Start
4976 -Winvalid-memory-model diagnostics with lowercase letter.
4979 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
4980 set in_cold_section_p to the partition containing loc_list->first.
4981 When seeing loc_list->last_before_switch node, update secname and
4982 perform range_across_switch second partition handling only after that.
4985 * run-rtl-passes.c: Include debug.h.
4986 (run_rtl_passes): Call debug_hooks->assembly_start.
4987 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
4991 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
4992 OPTAB_DIRECT in calls to expand_simple_binop.
4994 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4997 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
4999 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5002 * spellcheck.h (best_match::blithely_get_best_candidate): New
5005 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5007 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5008 with lower case letter.
5010 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5012 * config/i386/i386.md (bswaphi2): New expander.
5013 (*bswaphi2_movbe): New insn pattern.
5014 (bswaphi -> rorhi pepehole2): New peephole pattern.
5016 2017-11-20 Jeff Law <law@redhat.com>
5018 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5019 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5020 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5021 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5022 gimple-ssa-evrp-analyze.h.
5024 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5025 set BB_VISITED here.
5026 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5028 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5030 * tree-predcom.c: Add general comment on Store-Store chains.
5031 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5033 (get_chain_last_ref_at): Rename into...
5034 (get_chain_last_write_at): ...this.
5035 (get_chain_last_write_before_load): New function.
5036 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5037 CT_STORE_STORE when write reference is added.
5038 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5039 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5040 (initialize_root_vars_store_elim_1): Ditto.
5041 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5042 definition is created.
5043 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5044 chain by replacing it with dominant stored value.
5046 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5048 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5050 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5052 * vr-values.c (extract_range_from_binary_expr): Use a full range
5055 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5057 * config/arm/arm.md (R4_REGNUM): Define constant.
5058 (nonsecure_call_internal): Remove r4 clobber.
5059 (nonsecure_call_value_internal): Likewise.
5060 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5061 clobber and resequence match_operands.
5062 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5063 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5064 (nonsecure_call_value_reg_thumb2): Likewise.
5066 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5068 PR tree-optimization/78821
5069 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5070 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5071 to the address of the base rather than the base itself.
5072 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5073 (find_bswap_or_nop_finalize): New function.
5074 (find_bswap_or_nop): Use it.
5075 (bswap_replace): Return a tree rather than bool, change first
5076 argument from gimple * to gimple_stmt_iterator, allow inserting
5077 into an empty sequence, allow ins_stmt to be NULL - then emit
5078 all stmts into gsi. Fix up MEM_REF address gimplification.
5079 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5080 (struct store_immediate_info): Add N and INS_STMT non-static
5082 (store_immediate_info::store_immediate_info): Initialize them
5083 from newly added ctor args.
5084 (merged_store_group::apply_stores): Formatting fixes. Sort by
5086 (stmts_may_clobber_ref_p): For stores call also
5087 refs_anti_dependent_p.
5088 (gather_bswap_load_refs): New function.
5089 (imm_store_chain_info::try_coalesce_bswap): New method.
5090 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5091 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5092 (imm_store_chain_info::output_merged_store): Fail if number of
5093 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5094 and NOP_EXPR rhs_code.
5095 (pass_store_merging::process_store): Compute n and ins_stmt, if
5096 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5097 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5099 (pass_store_merging::execute): Calculate dominators.
5101 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5102 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5103 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5104 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5105 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5106 class pass_optimize_bswap, bswap_replace,
5107 pass_optimize_bswap::execute): Moved to ...
5108 * gimple-ssa-store-merging.c: ... this file.
5109 Include optabs-tree.h.
5110 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5111 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5112 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5113 anonymous namespace, remove static keywords.
5114 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5115 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5117 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5120 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5122 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5124 * vec.h (debug_helper): New function.
5125 (DEFINE_DEBUG_VEC): New macro.
5126 * hash-set.h (debug_helper): New function.
5127 (DEFINE_DEBUG_HASH_SET): New macro.
5128 * cfg.c (debug_slim (edge)): New function.
5129 Call DEFINE_DEBUG_VEC for edges.
5130 Call DEFINE_DEBUG_HASH_SET for edges.
5131 * cfghooks.c (debug_slim (basic_block)): New function.
5132 Call DEFINE_DEBUG_VEC for basic blocks.
5133 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5134 * print-tree.c (debug_slim): New function to handle trees.
5135 Call DEFINE_DEBUG_VEC for trees.
5136 Call DEFINE_DEBUG_HASH_SET for trees.
5137 (debug (vec<tree, va_gc>) &): Remove.
5138 (debug (<vec<tree, va_gc>) *): Remove.
5139 * print-rtl.c (debug_slim): New function to handle const_rtx.
5140 Call DEFINE_DEBUG_VEC for rtx_def.
5141 Call DEFINE_DEBUG_VEC for rtx_insn.
5142 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5143 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5144 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5145 (debug (vec<rtx_insn *> *ptr): Remove.
5146 (debug_insn_vector): Remove.
5147 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5149 2017-11-20 Tom de Vries <tom@codesourcery.com>
5151 PR rtl-optimization/82020
5152 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5153 IF_THEN_ELSE condition.
5155 2017-11-19 Jeff Law <law@redhat.com>
5157 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5158 of degenerates resulting from ignoring an edge.
5160 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5163 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5165 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5168 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5169 for uninitialied counts.
5171 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5174 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5176 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5179 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5181 2017-11-19 Tom de Vries <tom@codesourcery.com>
5183 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5186 2017-11-19 Tom de Vries <tom@codesourcery.com>
5188 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5191 2017-11-19 Tom de Vries <tom@codesourcery.com>
5193 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5196 2017-11-19 Tom de Vries <tom@codesourcery.com>
5198 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5200 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5202 2017-11-19 Tom de Vries <tom@codesourcery.com>
5204 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5206 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5207 semicolon after MCORE_EXPORT_NAME call.
5209 2017-11-19 Tom de Vries <tom@codesourcery.com>
5212 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5213 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5215 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5217 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5218 (cgraph_edge::clone): Cleanup updating of profile.
5219 * ipa-cp.c (update_profiling_info): Likewise.
5220 * ipa-inline-transform.c (inline_transform): Likewise.
5221 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5222 * ipa-split.c (execute_split_functions): Do not split when function
5224 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5225 * profile-count.c (profile_count::dump): Add global0.
5226 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5228 (profile_count::to_sreal_scale): Likewise.
5229 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5230 (profile_count::combine_with_ipa_count): New function.
5231 * profile-count.h (profile_guessed_global0adjusted): New.
5232 (profile_count::adjusted_zero): New.
5233 (profile_count::global0adjusted): New.
5234 (profile_count::combine_with_ipa_count): New.
5235 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5236 correct profile of return block of split functions.
5237 (copy_cfg_body): Remove unused profile_count.
5238 (copy_body): Likewise.
5239 (expand_call_inline): Update.
5240 (tree_function_versioning): Update.
5242 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5244 * hash-set.h (hash_set::empty): New.
5245 * tree-ssa-threadbackward.h: Delete.
5246 * tree-ssa-threadbackward.c (class thread_jumps): New.
5247 Move max_threaded_paths into class.
5248 (fsm_find_thread_path): Remove arguments that are now in class.
5249 (profitable_jump_thread_path): Rename to...
5250 (thread_jumps::profitable_jump_thread_path): ...this.
5251 (convert_and_register_jump_thread_path): Rename to...
5252 (thread_jumps::convert_and_register_current_path): ...this.
5253 (check_subpath_and_update_thread_path): Rename to...
5254 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5255 (register_jump_thread_path_if_profitable): Rename to...
5256 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5257 (handle_phi): Rename to...
5258 (thread_jumps::handle_phi): ...this.
5259 (handle_assignment): Rename to...
5260 (thread_jumps::handle_assignment): ...this.
5261 (fsm_find_control_statement_thread_paths): Rename to...
5262 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5263 (find_jump_threads_backwards): Rename to...
5264 (thread_jumps::find_jump_threads_backwards): ...this.
5265 Initialize path local data.
5266 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5267 from within thread_jumps class.
5268 (pass_early_thread_jumps::execute): Same.
5270 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5272 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5274 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5276 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5277 to canonical location.
5279 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5282 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5284 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5286 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5288 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5289 rcrt1.o%s/grcrt1.o%s for -static-pie.
5291 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5293 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5294 ix86_shift_rotate_cost): Break out from ...
5295 (ix86_rtx_costs): ... here.
5296 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5299 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5301 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5302 when entry block was promoted unlikely.
5303 (estimate_bb_frequencies): Increase frequency scale.
5304 * profile-count.h (profile_count): Export precision info.
5306 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5308 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5309 disturbin profile of entry block.
5311 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5313 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5316 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5318 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5319 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5321 2017-11-17 Jeff Law <law@redhat.com>
5323 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5324 from evrp_dom_walker class. Various methods moved into new class.
5325 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5326 (evrp_range_analyzer::enter): New method.
5327 (evrp_range_analyzer::leave): New method.
5328 (evrp_dom_walker): Remove delegators no longer needed by this class.
5329 Replace vr_values data member with evrp_range_analyzer
5331 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5332 method extracted from evrp_dom_walker::before_dom_children.
5333 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5334 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5336 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5337 Add private copy constructor and move assignment operators.
5338 Privatize methods and class data where trivially possible.
5339 (evrp_dom_walker::cleanup): New function, extracted from
5340 execute_early_vrp. Simplify access to class data.
5342 * vr-values.h (get_output_for_vrp): Prototype.
5343 * vr-values.c (get_output_for_vrp): New function extracted from
5344 vrp_visit_assignment_or_call and extract_range_from_stmt.
5345 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5347 2017-11-17 Luis Machado <luis.machado@linaro.org>
5349 * config/aarch64/aarch64.c
5350 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5351 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5352 tune_params::AUTOPREFETCHER_WEAK.
5354 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5357 * config/arm/arm.c (arm_valid_target_attribute_rec):
5358 Parse "arch=" and "+<ext>".
5359 (arm_valid_target_attribute_tree): Re-init global options.
5360 (arm_option_override): Make non-static.
5361 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5362 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5363 (__ARM_FEATURE_CRC32): Support undef.
5364 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5365 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5367 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5369 * gdbinit.in (break-on-diagnostic): New command.
5371 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5373 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5374 return to indirect jump for EH return if control-flow
5375 protection is enabled. Change explicit 'false' argument in
5376 pro_epilogue_adjust_stack with a value of
5378 * config/i386/i386.md (simple_return_indirect_internal):
5379 Remove SImode restriction to support 64-bit.
5381 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5383 * combine.c (added_notes_insn): New.
5384 (try_combine): Handle added_notes_insn like added_links_insn.
5385 Rewrite return value code.
5386 (distribute_notes): Set added_notes_insn to the earliest insn we added
5389 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5391 PR rtl-optimization/82621
5392 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5393 dest of one of those SETs is unused.
5395 2017-11-17 Richard Biener <rguenther@suse.de>
5398 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5399 * tree-pretty-print.c (dump_generic_node): Handle
5400 annot_expr_parallel_kind.
5401 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5402 * gimplify.c (gimple_boolify): Likewise.
5404 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5406 * config.gcc (extra_headers): Add cet.h for x86 targets.
5407 * config/i386/cet.h: New file.
5408 * doc/install.texi: Add --enable-cet/--disable-cet.
5410 2017-11-17 Richard Biener <rguenther@suse.de>
5412 PR tree-optimization/83017
5413 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5414 (gen_parallel_loop): Properly count iterations.
5415 (parallelize_loops): Handle loop->can_be_parallel independent
5416 of flag_loop_parallelize_all. Make static profitability test match
5418 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5419 * invoke.texi (parloops-min-per-thread): Document.
5421 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5423 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5424 upstreaming review comments.
5426 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5428 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5429 with fast unaligned access.
5430 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5432 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5434 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5435 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5437 2017-11-17 Richard Biener <rguenther@suse.de>
5439 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5440 folding of references.
5442 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5445 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5446 of replacing call to strncmp with corresponding call to strcmp when
5449 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5451 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5452 option prefer-avx256 for skylake-avx512 configuration.
5453 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5454 (get_builtin_code_for_version): Ditto.
5456 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5457 Monk Chiang <sh.chiang04@gmail.com>
5459 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5460 (FIXED_REGISTERS): Reserve more register numbers.
5461 (CALL_USED_REGISTERS): Likewise.
5462 (REG_ALLOC_ORDER): Likewise.
5463 (REG_CLASS_CONTENTS): Likewise.
5464 (REGISTER_NAMES): Likewise.
5466 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5467 Kito Cheng <kito.cheng@gmail.com>
5469 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5471 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5473 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5475 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5476 parameter list for vec_splats.
5478 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5480 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5482 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5484 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5487 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5488 with the required version of automake 1.11.6.
5490 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5492 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5494 (DU_C2_3_power9): Correct reservation combinations.
5495 (FP_DIV_power9, VEC_DIV_power9): New.
5496 (power9-alu): Split out rotate/shift...
5497 (power9-rot): ...to here, correct dispatch resource.
5498 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5500 (power9-fp): Correct latency.
5501 (power9-sdiv): Add div/sqrt resource.
5502 (power9-ddiv): Correct latency, add div/sqrt resource.
5503 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5504 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5506 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5508 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5510 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5511 switch statement mapping KF built-ins to TF built-ins if we don't
5512 have the proper ISA 3.0 assembler support.
5514 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5516 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5517 (lower_emutls_function_body): Do not compute it.
5519 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5521 * ipa-split.c (split_bb_info): Turn time to sreal.
5522 (split_point): Likewise.
5523 (dump_split_point): Likewise.
5524 (fine_split_points): Likewise.
5525 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5527 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5529 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5532 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5534 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5536 (expensive_function_p): Remove useless assert.
5537 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5539 2017-11-16 Martin Sebor <msebor@redhat.com>
5541 PR tree-optimization/82588
5542 PR tree-optimization/82583
5543 * tree-vrp.c (check_array_ref): Handle flexible array members,
5544 string literals, and inner indices.
5545 (search_for_addr_array): Add detail to diagnostics.
5547 2017-11-16 Nathan Sidwell <nathan@acm.org>
5551 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5552 (SET_DECL_ASSEMBLER_NAME): Forward to
5553 overwrite_decl_assembler_name.
5554 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5555 (overwrite_decl_assembler_name): Declare.
5556 * tree.c (overwrite_decl_assembler_name): New.
5557 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5558 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5559 (LANG_HOOKS_INITIALIZER): Add it.
5560 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5561 * langhooks.c (lhd_set_decl_assembler_name): Use
5562 SET_DECL_ASSEMBLER_NAME.
5563 (lhd_overwrite_decl_assembler_name): Default implementation.
5565 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5566 Jackson Woodruff <jackson.woodruff@arm.com>
5568 PR tree-optimization/71026
5569 * match.pd: Canonicalize constant multiplies in division.
5571 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5573 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5574 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5575 Remove OPT_fomit_frame_pointer entry.
5576 * common/config/arc/arc-common.c: Likewise.
5577 * common/config/arm/arm-common.c: Likewise.
5578 * common/config/avr/avr-common.c: Likewise.
5579 * common/config/c6x/c6x-common.c: Likewise.
5580 * common/config/cr16/cr16-common.c: Likewise.
5581 * common/config/cris/cris-common.c: Likewise.
5582 * common/config/epiphany/epiphany-common.c: Likewise.
5583 * common/config/fr30/fr30-common.c: Likewise.
5584 * common/config/frv/frv-common.c: Likewise.
5585 * common/config/ia64/ia64-common.c: Likewise.
5586 * common/config/iq2000/iq2000-common.c: Likewise.
5587 * common/config/lm32/lm32-common.c: Likewise.
5588 * common/config/m32r/m32r-common.c: Likewise.
5589 * common/config/mcore/mcore-common.c: Likewise.
5590 * common/config/microblaze/microblaze-common.c: Likewise.
5591 * common/config/mips/mips-common.c: Likewise.
5592 * common/config/mmix/mmix-common.c: Likewise.
5593 * common/config/mn10300/mn10300-common.c: Likewise.
5594 * common/config/nios2/nios2-common.c: Likewise.
5595 * common/config/pa/pa-common.c: Likewise.
5596 * common/config/pdp11/pdp11-common.c: Likewise.
5597 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5598 * common/config/riscv/riscv-common.c: Likewise.
5599 * common/config/rs6000/rs6000-common.c: Likewise.
5600 * common/config/rx/rx-common.c: Likewise.
5601 * common/config/s390/s390-common.c: Likewise.
5602 * common/config/sh/sh-common.c: Likewise.
5603 * common/config/sparc/sparc-common.c: Likewise.
5604 * common/config/tilegx/tilegx-common.c: Likewise.
5605 * common/config/tilepro/tilepro-common.c: Likewise.
5606 * common/config/v850/v850-common.c: Likewise.
5607 * common/config/visium/visium-common.c: Likewise.
5608 * common/config/xstormy16/xstormy16-common.c: Likewise.
5609 * common/config/xtensa/xtensa-common.c: Likewise.
5610 * invoke.texi (-fomit-frame-pointer): Update documentation.
5612 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5614 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5616 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5618 * cfg.c (scale_bbs_frequencies_int,
5619 cale_bbs_frequencies_gcov_type): Remove.
5620 * cfg.h (scale_bbs_frequencies_int,
5621 cale_bbs_frequencies_gcov_type): Remove.
5623 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5625 * tree-ssa-loop-manip.c
5626 (scale_dominated_blocks_in_loop): Update to profile counts.
5627 (tree_transform_and_unroll_loop): Likewise.
5629 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5631 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5632 scale_bbs_frequencies_int.
5634 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5636 * final.c (compute_alignments): Use counts rather than frequencies.
5638 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5640 * cfgloopanal.c: Include sreal.h
5641 (average_num_loop_insns): Use counts and sreal for accounting.
5643 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5645 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5648 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5649 Kito Cheng <kito.cheng@gmail.com>
5651 * config/nds32/constraints.md: Provide more constraints.
5652 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5653 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5654 support constraints usage.
5656 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5658 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5660 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5661 Kito Cheng <kito.cheng@gmail.com>
5663 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5664 * config/nds32/nds32.opt: Refine the layout.
5665 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5666 TARGET_EXT_STRING): Support new options.
5667 * config/nds32/nds32.h: Likewise.
5668 * config/nds32/nds32.md: Likewise.
5669 * config/nds32/nds32-predicates.c: Likewise.
5670 * config/nds32/constraints.md: Likewise.
5671 * common/config/nds32/nds32-common.c: Likewise.
5673 2017-11-16 Julia Koval <julia.koval@intel.com>
5676 * config/i386/gfniintrin.h: Add sse check.
5677 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5679 2017-11-16 Julia Koval <julia.koval@intel.com>
5681 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5682 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5683 (ix86_handle_option): Handle -mavx512vbmi2.
5684 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5685 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5686 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5687 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5688 (ix86_valid_target_attribute_inner_p): Ditto.
5689 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5690 * config/i386/i386.opt (mavx512vbmi2): New option.
5691 * doc/invoke.texi: Add new option.
5693 2017-11-16 Julia Koval <julia.koval@intel.com>
5695 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5696 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5697 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5698 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5699 _mm512_gf2p8mul_epi8): New intrinsics.
5700 * config/i386/i386-builtin-types.def
5701 (V64QI_FTYPE_V64QI_V64QI): New type.
5702 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5703 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5704 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5705 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5706 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5707 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5709 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5711 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5712 explicitly as a stream of bytes.
5714 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5716 * config/rs6000/altivec.h (vec_xst_be): New #define.
5717 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5718 and externalize from *altivec_vperm_<mode>_internal.
5719 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5721 (XL_BE_V8HI): Likewise.
5722 (XL_BE_V4SI): Likewise.
5723 (XL_BE_V4SI): Likewise.
5724 (XL_BE_V2DI): Likewise.
5725 (XL_BE_V4SF): Likewise.
5726 (XL_BE_V2DF): Likewise.
5727 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5728 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5729 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5730 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5731 VSX_BUILTIN_VEC_XST_BE.
5732 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5733 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5735 (altivec_init_builtins): Replace conditional calls to def_builtin
5736 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5737 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5738 with unconditional calls. Remove calls to def_builtin for
5739 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5740 __builtin_vec_xst_be.
5741 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5742 to define_expand, and add alternate RTL generation for P8.
5743 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5744 vsx_ld_elemrev_v8hi.
5745 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5746 add alternate RTL generation for P8.
5747 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5748 vsx_ld_elemrev_v16qi.
5749 (vsx_st_elemrev_v8hi): Convert define_insn
5750 to define_expand, and add alternate RTL generation for P8.
5751 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5752 vsx_st_elemrev_v8hi.
5753 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5754 add alternate RTL generation for P8.
5755 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5756 vsx_st_elemrev_v16qi.
5758 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5761 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5762 TARGET_AVX512ER check.
5763 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5764 neither -mzeroupper nor -mno-zeroupper is used and
5765 TARGET_EMIT_VZEROUPPER is set.
5766 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5767 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5769 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5771 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5772 folding of vector compares.
5773 (fold_build_vec_cmp): New helper function.
5774 (fold_compare_helper): New helper function.
5775 (builtin_function_type): Add compare builtins to the list of functions
5776 having unsigned arguments. Cosmetic updates to comment indentation.
5777 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5778 the not+eq combination.
5780 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5782 PR tree-optimization/82726
5783 PR tree-optimization/70754
5784 * tree-predcom.c (order_drefs_by_pos): New function.
5785 (combine_chains): Move code setting has_max_use_after to...
5786 (try_combine_chains): ...here. New parameter. Sort combined chains
5787 according to position information.
5788 (tree_predictive_commoning_loop): Update call to above function.
5789 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5791 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5793 PR tree-optimization/82726
5795 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5797 PR tree-optimization/70754
5798 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5799 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5800 combined stmt before it if not NULL.
5801 (combine_chains): Process refs reversely and compute dominance point
5805 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5807 PR tree-optimization/79663
5808 * tree-predcom.c (combine_chains): Process refs in reverse order
5809 only for ZERO length chains, and add explaining comment.
5811 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5813 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5814 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5815 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5817 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5819 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5821 2017-11-15 Martin Liska <mliska@suse.cz>
5823 * tree-cfg.c (pass_warn_function_return::execute):
5824 Compare warn_return_type for greater than zero.
5826 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5830 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5831 to return true on Xeon and not on Xeon Phi.
5832 (ix86_check_avx256_register): Changed to ...
5833 (ix86_check_avx_upper_register): ... this. Add extra check for
5834 VALID_AVX512F_REG_OR_XI_MODE.
5835 (ix86_avx_u128_mode_needed): Changed
5836 ix86_check_avx256_register to ix86_check_avx_upper_register.
5837 (ix86_check_avx256_stores): Changed to ...
5838 (ix86_check_avx_upper_stores): ... this. Changed
5839 ix86_check_avx256_register to ix86_check_avx_upper_register.
5840 (ix86_avx_u128_mode_after): Changed
5841 avx_reg256_found to avx_upper_reg_found. Changed
5842 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5843 (ix86_avx_u128_mode_entry): Changed
5844 ix86_check_avx256_register to ix86_check_avx_upper_register.
5845 (ix86_avx_u128_mode_exit): Ditto.
5846 * config/i386/i386.h: (host_detect_local_cpu): New define.
5848 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5850 * config/arm/xgene1.md (xgene1): Split into automatons
5851 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5852 (xgene1_f_load): Adjust reservations and/or types.
5853 (xgene1_f_store): Likewise.
5854 (xgene1_load_pair): Likewise.
5855 (xgene1_store_pair): Likewise.
5856 (xgene1_fp_load1): Likewise.
5857 (xgene1_load1): Likewise.
5858 (xgene1_store1): Likewise.
5859 (xgene1_move): Likewise.
5860 (xgene1_alu): Likewise.
5861 (xgene1_simd): Likewise.
5862 (xgene1_bfm): Likewise.
5863 (xgene1_neon_load1): Likewise.
5864 (xgene1_neon_store1): Likewise.
5865 (xgene1_neon_logic): Likewise.
5866 (xgene1_neon_st1): Likewise.
5867 (xgene1_neon_ld1r): Likewise.
5868 (xgene1_alu_cond): Added.
5869 (xgene1_shift_reg): Likwise.
5870 (xgene1_bfx): Likewise.
5871 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
5873 2017-11-15 Jakub Jelinek <jakub@redhat.com>
5876 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
5878 (can_widen_mult_without_libcall): New function.
5879 (expand_mul_overflow): If only checking unsigned mul overflow,
5880 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
5881 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
5882 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
5883 (expand_DIVMOD): Formatting fix.
5884 * expmed.h (expand_mult): Add NO_LIBCALL argument.
5885 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
5886 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
5888 PR tree-optimization/82977
5889 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5890 constructed temporary to strlen_to_stridx.put.
5892 2017-11-15 Martin Liska <mliska@suse.cz>
5894 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
5895 * configure: Regenerate.
5897 2017-11-15 Martin Liska <mliska@suse.cz>
5900 * config/sh/sh-mem.cc: Use proper probability for
5903 2017-11-14 Jeff Law <law@redhat.com>
5905 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
5906 the red zone for stack_clash_protection_final_dynamic_probe targets
5907 when the total dynamic stack size is zero bytes.
5909 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
5910 blocks is post order.
5912 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
5914 * dumpfile.h (TDF_COMPARE_DEBUG): New.
5915 * final.c (rest_of_clean_state): Set it for the
5916 -fcompare-debug dump.
5917 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
5918 class when TDF_COMPARE_DEBUG is set.
5920 * dwarf2out.c (gen_producer_string): Discard
5923 2017-11-15 Joseph Myers <joseph@codesourcery.com>
5926 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
5927 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
5928 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
5929 (__TGMATH_REAL_2_3): Remove macros.
5930 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
5931 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
5933 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
5935 (remquo): Define using __TGMATH_REAL_3.
5937 2017-11-14 Jeff Law <law@redhat.com>
5939 * vr-values.c: New file with contents extracted from tree-vrp.c.
5940 * Makefile.in (OBJS): Add vr-values.o
5941 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
5942 (set_value_range, set_and_canonicalize_value_range): Likewise.
5943 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
5944 (value_range_constant_singleton, symbolic_range_p): Likewise.
5945 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
5946 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
5947 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
5948 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
5949 (find_case_label_range, find_case_label_index): Likewise.
5950 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
5951 (range_int_cst_singleton_p, value_inside_range): Likewise.
5952 (get_single_symbol): Likewise.
5953 (switch_update): Move structure definition here.
5954 (to_remove_edges, to_update_switch_stmts): Provide externs.
5955 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
5956 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
5957 (vrp_val_is_min, set_value_range): Likewise.
5958 (set_and_canonicalize_value_range, copy_value_range): Likewise.
5959 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
5960 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
5961 (range_is_nonnull, range_int_cst_p): Likewwise.
5962 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
5963 (get_single_symbol, operand_less_p): Likewise
5964 (compare_values_warnv, compare_values): Likewise.
5965 (value_inside_range, value_range_constant_singleton): Likewise.
5966 (zero_nonzero_bitgs_from_vr): Likewise.
5967 (extract_range_from_binary_expr_1): Likewise.
5968 (overflow_comparison_p): Likewise.
5969 (to_remove_edges, to_update_switch_stmts): Likewise.
5970 (find_case_label-index, find_case_label_range): Likewise.
5971 (switch_update, set_value_range_to_nonnegative): Remove.
5972 (set_value_range_to_truthvalue): Likewise.
5973 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
5974 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
5975 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
5976 (find_case_label_ranges, test_for_singularity): Likewise.
5977 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
5978 (x_vr_values): Move to its remaining use site.
5980 2017-11-10 Jeff Law <law@redhat.com>
5982 * vr-values.h (VR_INITIALIZER): Move #define here.
5983 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
5984 * Makefile.in (OBJS): Add tree-evrp.o
5985 * tree-vrp.h (assert_info): Move structure definition here.
5986 (set_value_range_to_varying): Prototype.
5987 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5988 (infer_value_range, register_edge_assert_for): Likewise.
5989 (stmt_interesting_for_vrp): Likewise.
5990 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
5991 (set_value_range_to_varying): No longer static.
5992 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5993 (infer_value_range, register_edge_assert_for): Likewise.
5995 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5997 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
5998 generate the XXBRD instruction.
6000 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6001 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6002 as being compatible if -mabi=ieeelongdouble.
6003 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6004 to setup float128 built-ins with hardware support.
6005 (BU_FLOAT128_HW_2): Likewise.
6006 (BU_FLOAT128_HW_3): Likewise.
6007 (BU_FLOAT128_HW_VSX_1): Likewise.
6008 (BU_FLOAT128_HW_VSX_2): Likewise.
6009 (scalar_extract_expq): Change float128 built-in functions to
6010 accommodate having both KFmode and TFmode functions. Use the
6011 KFmode variant as the default.
6012 (scalar_extract_sigq): Likewise.
6013 (scalar_test_neg_qp): Likewise.
6014 (scalar_insert_exp_q): Likewise.
6015 (scalar_insert_exp_qp): Likewise.
6016 (scalar_test_data_class_qp): Likewise.
6017 (sqrtf128_round_to_odd): Delete processing the round to odd
6018 built-in functions as special built-in functions, and define them
6019 as float128 built-ins. Use the KFmode variant as the default.
6020 (truncf128_round_to_odd): Likewise.
6021 (addf128_round_to_odd): Likewise.
6022 (subf128_round_to_odd): Likewise.
6023 (mulf128_round_to_odd): Likewise.
6024 (divf128_round_to_odd): Likewise.
6025 (fmaf128_round_to_odd): Likewise.
6026 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6027 support for KFmode and TFmode xststdcqp calls.
6028 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6029 point, switch the built-in handlers for the get/set float128
6030 exponent, get float128 mantissa, float128 test built-ins, and the
6031 float128 round to odd built-in functions. Eliminate creating the
6032 float128 round to odd built-in functions as special built-ins.
6033 (rs6000_init_builtins): Eliminate special creation of the float128
6034 round to odd built-in functions.
6035 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6036 function insns to support both TFmode and KFmode variants.
6037 (xsxsigqp_<mode>): Likewise.
6038 (xsiexpqpf_<mode>): Likewise.
6039 (xsiexpqp_<mode>): Likewise.
6040 (xststdcqp_<mode>): Likewise.
6041 (xststdcnegqp_<mode>): Likewise.
6042 (xststdcqp_<mode>): Likewise.
6044 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6046 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6047 unused path_in_freq_ptr parameter.
6048 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6050 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6052 * ipa-inline.c (edge_badness): Dump sreal frequency.
6053 (compute_inlined_call_time): Match natural implementaiton ...
6054 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6055 forgotten division by CGRAPH_FREQ_BASE.
6057 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6059 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6060 Solaris 11. Update comment.
6061 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6063 * config/sol2.h (STARTFILE_SPEC): Likewise.
6064 * configure: Regenerate.
6066 2017-11-14 Carl Love <cel@us.ibm.com>
6068 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6069 le_ and be_ prefixes to swap* variables. Remove
6070 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6073 2017-11-14 Jason Merrill <jason@redhat.com>
6075 Support GTY((cache)) on hash_map.
6076 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6077 (ggc_cache_remove): Override it instead of ggc_mx.
6078 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6079 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6080 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6081 (simple_cache_map_traits): Override maybe_mx.
6082 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6083 (hash_map): Friend gt_cleare_cache.
6084 (gt_cleare_cache): New.
6085 * tree.h (tree_cache_traits): New hash_map traits class.
6086 (tree_cache_map): New typedef.
6088 2017-11-14 Richard Biener <rguenther@suse.de>
6090 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6091 paramter and handling.
6092 (cleanup_control_flow_bb): Likewise.
6093 (cleanup_control_flow_pre): New helper performing a DFS walk
6094 to call cleanup_control_flow_bb in PRE order.
6095 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6096 via cleanup_control_flow_pre.
6098 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6100 * config/aarch64/aarch64-simd.md
6101 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6102 (*aarch64_simd_bsl<mode>_alt): Likewise.
6103 (aarch64_simd_bsldi_internal): New.
6104 (aarch64_simd_bsldi_alt): Likewise.
6106 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6108 * tracer.c (better_p): Do not compare frequencies.
6109 * reg-stack.c (better_edge): Likewise.
6110 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6113 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6115 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6116 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6117 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6118 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6119 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6120 * predict.c (maybe_hot_frequency_p): Inline to ...
6121 (maybe_hot_count_p): ... here; rewrite to counts.
6122 (counts_to_freqs): Rename to ...
6123 (update_max_bb_count): ... this one.
6124 (expensive_function_p): Use counts.
6125 (estimate_bb_frequencies): Update.
6126 (rebuild_frequencies): Update.
6127 * predict.h (counts_to_freqs): Rename to ...
6128 (update_max_bb_count): ... this one.
6129 * profile.c (compute_branch_probabilities): Add debug info
6130 * tree-inline.c (expand_call_inline): Update debug info.
6131 (optimize_inline_calls): Use update_max_bb_count..
6132 (tree_function_versioning): Use update_max_bb_count..
6133 * value-prof.c (gimple_value_profile_transformations):
6134 Do not use update_max_bb_count.
6136 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6138 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6139 always use frequencies.
6141 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6143 * bb-reorder.c: Remove frequencies from comments.
6144 (better_edge_p): Use profile counts.
6145 (find_traces): Dump profile counts.
6146 (rotate_loop): Use profile counts.
6147 (find_traces_1_round): Likewise.
6148 (connect_better_edge_p): Use counts instead of probabilities for
6150 (copy_bb_p): Drop early check for non-0 frequency.
6151 (sanitize_hot_paths): Update comments.
6153 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6155 * ipa-split.c (struct split_point): Add count.
6156 (consider_split): Do not compute incoming frequency; compute incoming
6157 count and store it to split_point.
6158 (split_function): Set count of the call to split part correctly.
6160 2017-11-13 Carl Love <cel@us.ibm.com>
6162 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6164 2017-11-13 Tom Tromey <tom@tromey.com>
6166 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6168 2017-11-13 Carl Love <cel@us.ibm.com>
6170 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6171 Add support for builtins:
6172 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6173 vector {un,}signed {char,int,short},
6174 vector {un,}signed {char,int,short}) arguments.
6175 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6176 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6177 Add BU_P9V_AV_2 expansions for the builtins.
6178 * config/rs6000/altivec.h (vec_first_match_index,
6179 vec_first_mismatch_index, vec_first_match_or_eos_index,
6180 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6181 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6182 new extern declaration.
6183 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6184 * config/rs6000/vsx.md (first_match_index_<mode>,
6185 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6186 first_mismatch_or_eos_index_<mode>): Add define expand.
6187 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6188 * doc/extend.texi: Update the built-in documenation file for the new
6191 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6193 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6194 into the min/max operations for _Float<N> and _Float<N>X types.
6196 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6199 * dwarf2out.c (do_eh_frame): New static variable.
6200 (dwarf2out_begin_prologue): Set it.
6201 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6203 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6205 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6207 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6209 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6212 * debug.h (dwarf2out_do_eh_frame): Declare.
6213 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6214 (dwarf2out_do_frame): Use it.
6215 (dwarf2out_do_cfi_asm): Likewise.
6216 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6217 (dwarf2out_assembly_start): Likewise.
6218 (dwarf2out_begin_prologue): Fix comment.
6219 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6220 if the target needs either debug or unwind DWARF2 info.
6221 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6222 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6223 -ftrapping-math, -ftrapv and -fwrapv.
6225 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6227 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6228 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6229 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6230 (estimate_edge_size_and_time): Likewise.
6231 (ipa_merge_fn_summary_after_inlining): Likewise.
6232 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6233 (compute_uninlined_call_time): Use sreal_frequency.
6234 (compute_inlined_call_time): Likewise.
6235 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6236 * profile-count.c: Include sreal.h.
6237 (profile_count::to_sreal_scale): New.
6238 * profile-count.h: Forward declare sreal.
6239 (profile_count::to_sreal_scale): Declare.
6241 2017-11-13 Nathan Sidwell <nathan@acm.org>
6243 * diagnostic.c (maybe_line_and_column): New.
6244 (diagnostic_get_location_text): Use it.
6245 (diagnostic_report_current_module): Likewise.
6246 (test_diagnostic_get_location_text): Add tests.
6248 2017-11-13 Luis Machado <luis.machado@linaro.org>
6250 * doc/md.texi (Specifying processor pipeline description): Fix
6251 incorrect latency for the div instruction example.
6253 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6255 PR tree-optimization/78821
6256 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6257 that bit_not_p is the same.
6258 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6259 (split_group): Count precisely bit_not_p bits in each statement.
6260 (invert_op): New function.
6261 (imm_store_chain_info::output_merged_store): Use invert_op to
6262 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6263 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6265 2017-11-13 Martin Liska <mliska@suse.cz>
6267 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6268 (struct source_info): Likewise.
6269 (add_branch_counts): Likewise.
6270 (add_line_counts): Likewise.
6271 (function_summary): Likewise.
6272 (output_intermediate_line): Likewise.
6273 (generate_results): Likewise.
6275 2017-11-13 Martin Liska <mliska@suse.cz>
6277 * gcov.c (struct block_info): Remove typedef for block_t.
6278 (struct line_info): Likewise.
6279 (line_info::has_block): Likewise.
6280 (EXIT_BLOCK): Likewise.
6281 (unblock): Likewise.
6282 (circuit): Likewise.
6283 (get_cycles_count): Likewise.
6284 (process_file): Likewise.
6285 (read_graph_file): Likewise.
6286 (solve_flow_graph): Likewise.
6287 (find_exception_blocks): Likewise.
6288 (add_line_counts): Likewise.
6289 (accumulate_line_info): Likewise.
6290 (output_line_details): Likewise.
6292 2017-11-13 Martin Liska <mliska@suse.cz>
6294 * gcov.c (struct arc_info): Remove typedef for arc_t.
6295 (struct line_info): Likewise.
6296 (add_branch_counts): Likewise.
6297 (output_branch_count): Likewise.
6298 (function_info::~function_info): Likewise.
6299 (circuit): Likewise.
6300 (output_intermediate_line): Likewise.
6301 (read_graph_file): Likewise.
6302 (solve_flow_graph): Likewise.
6303 (find_exception_blocks): Likewise.
6304 (add_line_counts): Likewise.
6305 (accumulate_line_info): Likewise.
6306 (output_line_details): Likewise.
6307 (output_function_details): Likewise.
6309 2017-11-13 Martin Liska <mliska@suse.cz>
6311 * gcov.c (struct function_info): Remove typedef for function_t.
6312 (struct source_info): Likewise.
6313 (source_info::get_functions_at_location): Likewise.
6314 (solve_flow_graph): Likewise.
6315 (find_exception_blocks): Likewise.
6316 (add_line_counts): Likewise.
6317 (output_intermediate_file): Likewise.
6318 (process_file): Likewise.
6319 (generate_results): Likewise.
6320 (release_structures): Likewise.
6321 (read_graph_file): Likewise.
6322 (read_count_file): Likewise.
6323 (accumulate_line_counts): Likewise.
6324 (output_lines): Likewise.
6326 2017-11-13 Martin Liska <mliska@suse.cz>
6328 * gcov.c (function_info::function_info): Remove num_counts
6329 and add vector<gcov_type>.
6330 (function_info::~function_info): Use the vector.
6331 (process_file): Likewise.
6332 (read_graph_file): Likewise.
6333 (read_count_file): Likewise.
6334 (solve_flow_graph): Likewise.
6336 2017-11-13 Martin Liska <mliska@suse.cz>
6338 * gcov.c (function_info::is_artificial): New function.
6339 (process_file): Erase all artificial early.
6340 (generate_results): Skip as all artificial are already
6343 2017-11-13 Martin Liska <mliska@suse.cz>
6345 * gcov.c (read_graph_file): Store to global vector of functions.
6346 (read_count_file): Iterate the vector.
6347 (process_file): Likewise.
6348 (generate_results): Likewise.
6349 (release_structures): Likewise.
6351 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6353 PR tree-optimization/82954
6354 * gimple-ssa-store-merging.c
6355 (imm_store_chain_info::coalesce_immediate_stores): If
6356 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6358 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6360 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6361 Upddate call to ENDIAN_LANE_N.
6362 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6363 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6364 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6365 and use aarch64_endian_lane_rtx.
6366 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6368 2017-11-12 Tom de Vries <tom@codesourcery.com>
6370 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6372 2017-11-12 Tom de Vries <tom@codesourcery.com>
6374 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6376 2017-11-12 Tom de Vries <tom@codesourcery.com>
6378 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6380 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6381 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6382 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6384 2017-11-11 Martin Sebor <msebor@redhat.com>
6387 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6390 * prefic.c (translate_name): Replace strncpy with memcpy to
6391 avoid -Wstringop-truncation.
6393 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6395 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6397 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6399 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6401 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6403 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6406 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6408 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6409 enable generating XXBRH if the value is in a vector register.
6410 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6411 value is in a vector register.
6412 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6413 register to register bswap64's instead of doing the GPR sequence
6414 used on previous machines.
6415 (bswapdi2_xxbrd): New insn.
6416 (bswapdi2_reg): Disallow on ISA 3.0.
6417 (register to register bswap64 splitter): Do not split the insn on
6418 ISA 3.0 systems that use XXBRD.
6420 2017-11-10 Martin Sebor <msebor@redhat.com>
6423 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6425 (find_subframework_file): Same.
6427 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6429 * auto-profile.c (afdo_indirect_call): Drop frequency.
6430 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6431 (cgraph_node::create_edge): Drop frequency argument.
6432 (cgraph_node::create_indirect_edge): Drop frequency argument.
6433 (cgraph_edge::make_speculative): Drop frequency arguments.
6434 (cgraph_edge::resolve_speculation): Do not update frequencies
6435 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6436 (cgraph_node::dump): Check consistency in IPA mode.
6437 (cgraph_edge::maybe_hot_p): Use IPA counter.
6438 (cgraph_edge::verify_count_and_frequency): Rename to ...
6439 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6440 (cgraph_node::verify_node): Update.
6441 * cgraph.h (struct cgraph_edge): Drop frequency.
6442 (cgraph_edge::frequency): New function.
6443 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6445 (cgraph_edge::rebuild_edges): Likewise.
6446 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6447 (duplicate_thunk_for_node): Do not pass frequency.
6448 (cgraph_node::create_clone): Scale only counts.
6449 (cgraph_node::create_virtual_clone): Do not pass frequency.
6450 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6451 (cgraph_node::create_version_clone): Do not pass frequency.
6452 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6453 (cgraph_node::expand_thunk): Do not pass frequency.
6454 (cgraph_node::create_wrapper): Do not pass frequency.
6455 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6457 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6458 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6459 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6460 (gather_caller_stats): Use frequency function.
6461 (ipcp_cloning_candidate_p): Use frequency function.
6462 (ipcp_propagate_stage): Use frequency function.
6463 (get_info_about_necessary_edges): Use frequency function.
6464 (update_profiling_info): Update only IPA profile.
6465 (update_specialized_profile): Use frequency functoin.
6466 (perhaps_add_new_callers): Update only IPA profile.
6467 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6468 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6469 (dump_ipa_call_summary): Use frequency function.
6470 (estimate_edge_size_and_time): Use frequency function.
6471 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6472 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6473 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6474 (update_noncloned_counts): ... ths one; scale counts only.
6475 (clone_inlined_nodes): Do not scale frequency.
6476 (inline_call): Do not pass frequency.
6477 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6478 (compute_inlined_call_time): Use IPA profile.
6479 (want_inline_small_function_p): Use IPA profile.
6480 (want_inline_self_recursive_call_p): Use IPA profile.
6481 (edge_badness): Use IPA profile.
6482 (lookup_recursive_calls): Use IPA profile.
6483 (recursive_inlining): Do not pass frequency.
6484 (resolve_noninline_speculation): Do not update frequency.
6485 (inline_small_functions): Collect max of IPA profile.
6486 (dump_overall_stats): Dump IPA porfile.
6487 (dump_inline_stats): Dump IPA porfile.
6488 (ipa_inline): Collect IPA stats.
6489 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6490 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6491 (ipa_propagate_frequency): Use frequency function.
6492 (ipa_profile): Cleanup.
6493 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6494 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6495 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6496 (input_node): Do not stream frequency.
6497 (input_edge): Do not stream frequency.
6498 (merge_profile_summaries): Scale only IPA profiles.
6499 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6500 * predict.c (drop_profile): Do not recompute frequency.
6501 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6502 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6503 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6504 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6505 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6506 * tree-inline.c (copy_bb): Do not scale frequency.
6507 (expand_call_inline): Do not scale frequency.
6508 (tree_function_versioning): Do not scale frequency.
6509 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6511 2017-11-10 Julia Koval <julia.koval@intel.com>
6513 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6514 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6515 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6516 (_mm256_maskz_gf2p8affine_epi64_epi8)
6517 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6518 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6519 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6520 (__builtin_ia32_vgf2p8affineqb_v32qi)
6521 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6522 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6524 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6528 (arm_option_override): Refactor.
6529 (arm_option_reconfigure_globals): New.
6530 (arm_options_perform_arch_sanity_checks): New.
6531 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6533 (arm_options_perform_arch_sanity_checks): Likewise
6535 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6537 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6538 (power9-qpmul): New.
6539 * rs6000/rs6000.md ("type" attr): Add qmul.
6540 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6541 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6542 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6544 2017-11-10 Martin Sebor <msebor@redhat.com>
6547 * builtins.c (compute_objsize): Handle arrays that
6548 compute_builtin_object_size likes to fail for. Make extern.
6549 * builtins.h (compute_objsize): Declare.
6550 (check_strncpy_sizes): New function.
6551 (expand_builtin_strncpy): Call check_strncpy_sizes.
6552 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6553 -Wstringop-truncation.
6554 (gimple_fold_builtin_strncat): Same.
6555 * gimple.c (gimple_build_call_from_tree): Set call location.
6556 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6557 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6558 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6559 (handle_builtin_strlen): Use strlen_to_stridx.
6560 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6562 Use strlen_to_stridx.
6563 (pass_strlen::execute): Release strlen_to_stridx.
6564 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6565 (-Wstringop-truncation): Document new option.
6567 2017-11-10 Martin Liska <mliska@suse.cz>
6569 PR gcov-profile/82702
6570 * gcov.c (main): Handle intermediate files in a different
6572 (get_gcov_intermediate_filename): New function.
6573 (output_gcov_file): Remove support of intermediate files.
6574 (generate_results): Allocate intermediate file.
6575 (release_structures): Clean-up properly fn_end.
6576 (output_intermediate_file): Start iterating with line 1.
6578 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6580 PR tree-optimization/82929
6581 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6582 ops_swapped_p non-static data member.
6583 (store_immediate_info::store_immediate_info): Clear it.
6584 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6585 ops set ops_swapped_p.
6586 (count_multiple_uses): Handle ops_swapped_p.
6588 2017-11-10 Martin Liska <mliska@suse.cz>
6590 * coverage.c (coverage_init): Stream information about
6591 support of has_unexecuted_blocks.
6592 * doc/gcov.texi: Document that.
6593 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6594 * gcov.c (read_graph_file): Likewise.
6595 (output_line_beginning): Fix a small issue with
6598 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6600 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6601 reference of trivial component.
6603 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6606 * gimple-ssa-store-merging.c
6607 (pass_store_merging::terminate_all_aliasing_chains): For
6608 gimple_store_p stmts also call refs_output_dependent_p.
6610 PR rtl-optimization/82913
6611 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6614 2017-11-09 Jeff Law <law@redhat.com>
6616 * vr-values.h: New file with vr_values class.
6617 * tree-vrp.c: Include vr-values.h
6618 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6619 data objects into the vr_values class.
6620 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6621 (get_value_range): Make it a member function within vr_values class.
6622 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6623 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6624 (op_with_constant_singleton_value_range): Likewise.
6625 (extract_range_for_var_from_comparison_expr): Likewise.
6626 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6627 (extract_range_from_binary_expr): Likewise.
6628 (extract_range_from_unary_expr): Likewise.
6629 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6630 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6631 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6632 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6633 (compare_name_with_value, compare_names): Likewise.
6634 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6635 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6636 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6637 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6638 (extract_range_from_phi_node): Likewise.
6639 (simplify_truth_ops_using_ranges): Likewise.
6640 (simplify_div_or_mod_using_ranges): Likewise.
6641 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6642 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6643 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6644 (simplify_float_conversion_using_ranges): Likewise.
6645 (simplify_internal_call_using_ranges): Likewise.
6646 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6647 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6648 poitner via x_vr_values for calls into gimple folder.
6649 (vrp_initialize_lattice): Make this the vr_values ctor.
6650 (vrp_free_lattice): Make this the vr_values dtor.
6651 (set_vr_value): New function.
6652 (class vrp_prop): Add vr_values data member. Add various member
6653 functions as well as member functions that delegate to vr_values.
6654 (check_array_ref): Make a member function within vrp_prop class.
6655 (search_for_addr_array, vrp_initialize): Likewise.
6656 (vrp_finalize): Likewise. Revamp to avoid direct access to
6657 vr_value, values_propagated, etc.
6658 (check_array_bounds): Extract vrp_prop class instance pointer from
6659 walk info structure. Use it to call member functions.
6660 (check_all_array_refs): Make a member function within vrp_prop class.
6661 Smuggle class instance pointer via walk info structure.
6662 (x_vr_values): New local static.
6663 (vrp_valueize): Use x_vr_values to get class instance.
6664 (vr_valueize_1): Likewise.
6665 (class vrp_folder): Add vr_values data member. Add various member
6666 functions as well as member functions that delegate to vr_values.
6667 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6668 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6669 class instance from vr_values. Use it to call member functions.
6670 (vrp_dom_walker): Add vr_values data member.
6671 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6672 instance via x_vr_values.
6673 (identify_jump_threads): Accept vr_values as argument. Store
6674 it into the walker structure.
6675 (evrp_dom_walker): Add vr_values class data member. Add various
6677 (evrp_dom_walker::try_find_new_range): Use vr_values data
6678 member to access the memory allocator.
6679 (evrp_dom_walker::before_dom_children): Store vr_values class
6680 instance into the vrp_folder class.
6681 (evrp_dom_walker::push_value_range): Rework to avoid direct
6682 access to num_vr_values and vr_value.
6683 (evrp_dom_walker::pop_value_range): Likewise.
6684 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6685 Use vr_values to get to dump_all_value_ranges member function.
6686 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6687 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6688 Pass vr_values class instance down to identify_jump_threads.
6689 Remove call to vrp_free_lattice.
6690 (debug_all_value_ranges): Remove.
6692 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6693 (vrp_folder): Likewise.
6695 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6696 Get it from the existing bitmap instead.
6697 (vrp_intersect_ranges_1): Likewise.
6699 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6701 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6703 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6704 set bit_not_p to it.
6705 (imm_store_chain_info::coalesce_immediate_stores): Break group
6706 if bit_not_p is different.
6707 (count_multiple_uses, split_group,
6708 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6709 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6710 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6711 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6712 to store_immediate_info ctor.
6714 2017-11-09 Jim Wilson <jimw@sifive.com>
6716 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6717 (scan_prog_file): Likewise.
6719 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6721 * bb-reorder.c (max_entry_frequency): Remove.
6722 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6723 connect_traces, push_to_next_round_p): Remove prototypes.
6724 (find_traces_1_round): Use counts only.
6725 (push_to_next_round_p): Likewise.
6726 (find_traces): Likewise.
6727 (rotate_loop): Likewise.
6728 (find_traces_1_round): Likewise.
6729 (connect_traces): Likewise.
6730 (edge_order): Likewise.
6732 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6734 * config/arm/arm.c (output_return_instruction): Add comments to
6735 indicate requirement for cmse_nonsecure_entry return to account
6736 for the size of clearing instruction output here.
6737 (thumb_exit): Likewise.
6738 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6739 return in hardfloat mode.
6741 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6743 * config/rs6000/rs6000.c (machine_function): Add a bool,
6744 "toc_is_wrapped_separately".
6745 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6746 if it wasn't explicitly set or unset, we are optimizing for speed, and
6747 doing separate shrink-wrapping.
6748 (rs6000_get_separate_components): Enable the TOC component if
6749 saving the TOC register in the prologue.
6750 (rs6000_components_for_bb): Handle the TOC component.
6751 (rs6000_emit_prologue_components): Store the TOC register where needed.
6752 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6753 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6755 2017-11-09 Martin Jambor <mjambor@suse.cz>
6757 * ipa-param-manipulation.c: New file.
6758 * ipa-param-manipulation.h: Likewise.
6759 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6760 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6761 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6762 (ipa_parm_adjustment): Likewise.
6763 (ipa_parm_adjustment_vec): Likewise.
6764 (ipa_get_vector_of_formal_parms): Moved declaration to
6765 ipa-param-manipulation.h.
6766 (ipa_get_vector_of_formal_parm_types): Likewise.
6767 (ipa_modify_formal_parameters): Likewise.
6768 (ipa_modify_call_arguments): Likewise.
6769 (ipa_combine_adjustments): Likewise.
6770 (ipa_dump_param_adjustments): Likewise.
6771 (ipa_modify_expr): Likewise.
6772 (ipa_get_adjustment_candidate): Likewise.
6773 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6774 ipa-param-manipulation.c.
6775 (ipa_get_vector_of_formal_parm_types): Likewise.
6776 (ipa_modify_formal_parameters): Likewise.
6777 (ipa_modify_call_arguments): Likewise.
6778 (ipa_modify_expr): Likewise.
6779 (get_ssa_base_param): Likewise.
6780 (ipa_get_adjustment_candidate): Likewise.
6781 (index_in_adjustments_multiple_times_p): Likewise.
6782 (ipa_combine_adjustments): Likewise.
6783 (ipa_dump_param_adjustments): Likewise.
6784 * tree-sra.c: Also include ipa-param-manipulation.h
6785 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6788 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6789 Alan Hayward <alan.hayward@arm.com>
6790 David Sherwood <david.sherwood@arm.com>
6792 * doc/sourcebuild.texi (vect_masked_store): Document.
6794 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6795 Alan Hayward <alan.hayward@arm.com>
6796 David Sherwood <david.sherwood@arm.com>
6798 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6800 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6801 Alan Hayward <alan.hayward@arm.com>
6802 David Sherwood <david.sherwood@arm.com>
6804 * doc/sourcebuild.texi (vect_variable_length): Document.
6806 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6807 Alan Hayward <alan.hayward@arm.com>
6808 David Sherwood <david.sherwood@arm.com>
6810 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6812 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6813 Alan Hayward <alan.hayward@arm.com>
6814 David Sherwood <david.sherwood@arm.com>
6816 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
6818 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6819 Alan Hayward <alan.hayward@arm.com>
6820 David Sherwood <david.sherwood@arm.com>
6822 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6823 previously undocumented selectors.
6824 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6826 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6828 * doc/rtl.texi (const_vector): Say that elements can be
6829 const_wide_ints too.
6830 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6831 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6832 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6833 * optabs.c (expand_vector_broadcast): Likewise.
6835 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6836 Alan Hayward <alan.hayward@arm.com>
6837 David Sherwood <david.sherwood@arm.com>
6839 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6840 scaled index even if the unscaled address was invalid.
6841 Don't increase the complexity of using a scale in that case.
6843 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6844 Alan Hayward <alan.hayward@arm.com>
6845 David Sherwood <david.sherwood@arm.com>
6847 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6848 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6849 UNITS_PER_WORD bytes.
6850 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6851 based on REGMODE_NATURAL_SIZE of the inner mode.
6852 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6853 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6854 * expmed.c (lowpart_bit_field_p): Divide the value up into
6855 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6856 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6857 whether something is likely to occupy more than one register.
6859 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6862 * ipa-inline-transform.c (update_noncloned_frequencies): Use
6863 profile_count::adjust_for_ipa_scaling.
6864 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
6865 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
6867 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
6869 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6871 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
6872 (split_group): Add total_orig and total_new arguments, estimate the
6873 number of statements related to the store group without store merging
6874 and with store merging.
6875 (imm_store_chain_info::output_merged_store): Adjust split_group
6876 callers, punt if estimated number of statements with store merging
6877 is not smaller than estimated number of statements without it.
6879 (handled_load): Remove has_single_use checks.
6880 (pass_store_merging::process_store): Likewise.
6882 2017-11-09 Richard Biener <rguenther@suse.de>
6884 PR tree-optimization/82902
6885 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
6887 2017-11-09 Martin Liska <mliska@suse.cz>
6890 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
6893 2017-11-09 Martin Liska <mliska@suse.cz>
6895 PR tree-optimization/82669
6896 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
6898 2017-11-09 Martin Liska <mliska@suse.cz>
6900 PR gcov-profile/48463
6901 * coverage.c (coverage_begin_function): Output also end locus
6902 of a function and information whether the function is
6904 * gcov-dump.c (tag_function): Parse and print the information.
6905 * gcov.c (INCLUDE_MAP): Add include.
6906 (INCLUDE_SET): Likewise.
6907 (struct line_info): Move earlier in the source file because
6908 of vector<line_info> in function_info structure.
6909 (line_info::line_info): Likewise.
6910 (line_info::has_block): Likewise.
6911 (struct source_info): Add new member index.
6912 (source_info::get_functions_at_location): New function.
6913 (function_info::group_line_p): New function.
6914 (output_intermediate_line): New function.
6915 (output_intermediate_file): Use the mentioned function.
6916 (struct function_start): New.
6917 (struct function_start_pair_hash): Likewise.
6918 (process_file): Add code that identifies group functions.
6919 Assign lines either to global or function scope.
6920 (generate_results): Skip artificial functions.
6921 (find_source): Assign index for each source file.
6922 (read_graph_file): Read new flag artificial and end_line.
6923 (add_line_counts): Assign it either to global of function scope.
6924 (accumulate_line_counts): Isolate core of the function to
6925 accumulate_line_info and call it for both function and global
6927 (accumulate_line_info): New function.
6928 (output_line_beginning): Fix GNU coding style.
6929 (print_source_line): New function.
6930 (output_line_details): Likewise.
6931 (output_function_details): Likewise.
6932 (output_lines): Iterate both source (global) scope and function
6934 (struct function_line_start_cmp): New class.
6935 * doc/gcov.texi: Reflect changes in documentation.
6937 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6940 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
6941 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6942 and similarly for not instead of neg.
6944 2017-11-08 Andi Kleen <ak@linux.intel.com>
6946 * config/i386/i386.opt: Add -mforce-indirect-call.
6947 * config/i386/predicates.md: Check for flag_force_indirect_call.
6948 * doc/invoke.texi: Document -mforce-indirect-call
6950 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
6952 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
6954 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6955 config/riscv/riscv.c (predict.h): New include.
6956 (riscv_slow_unaligned_access_p): No longer static.
6957 (riscv_block_move_straight): Add require.
6958 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6960 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6963 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
6964 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
6965 nonimmediate_operand predicate for operand 1 instead of
6968 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6970 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
6972 * config/aarch64/constraints.md (Uml): New constraint.
6973 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
6976 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6978 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6979 of two vec_duplicates into a vec_concat.
6981 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6983 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6984 Simplify vec_merge of vec_duplicate and vec_concat.
6985 * config/aarch64/constraints.md (Utq): New constraint.
6986 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
6989 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6991 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6992 Simplify vec_merge of vec_duplicate and const_vector.
6993 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
6995 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
6996 mode iterator. Update predicate on operand 1 to
6997 handle non-const_vec constants. Delete constraints.
6998 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7000 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7002 PR tree-optimization/78821
7003 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7005 (store_operand_info::store_operand_info): Initialize it to false.
7006 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7007 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7008 store in the group, and if chain_info is non-NULL, to ignore altogether
7010 (compatible_load_p): Fail if bit_not_p does not match.
7011 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7012 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7013 (pass_store_merging::process_store): Adjust
7014 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7015 call terminate_all_aliasing_chains newly when adding a store into
7016 a chain with non-NULL chain_info.
7018 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7020 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7022 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7024 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7026 (aarch64_layout_frame): Initialise emit_frame_chain.
7027 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7028 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7030 2017-11-08 Martin Liska <mliska@suse.cz>
7032 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7033 of gimple_call_internal_p.
7035 2017-11-07 Tom de Vries <tom@codesourcery.com>
7037 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7039 2017-11-07 Tom de Vries <tom@codesourcery.com>
7041 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7044 2017-11-08 Martin Liska <mliska@suse.cz>
7047 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7049 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7051 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7053 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7056 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7057 (*cmp<mode>_ccz_1): New insn with $k alternative.
7060 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7061 RTX_COMM_COMPARE as commutative as well.
7062 (ix86_binary_operator_ok): Formatting fix.
7063 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7064 *<code><mode>3<mask_name><round_saeonly_name>,
7065 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7066 *<s>mul<mode>3_highpart<mask_name>,
7067 *vec_widen_umult_even_v16si<mask_name>,
7068 *vec_widen_umult_even_v8si<mask_name>,
7069 *vec_widen_umult_even_v4si<mask_name>,
7070 *vec_widen_smult_even_v16si<mask_name>,
7071 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7072 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7073 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7074 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7075 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7076 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7077 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7078 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7079 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7080 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7081 ix86_binary_operator_ok. Formatting fixes.
7082 (*<plusminus_insn><mode>3<mask_name><round_name>,
7083 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7086 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7088 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7089 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7090 eq and ne if TARGET_ISEL.
7091 (cmp): New code_iterator.
7092 (UNS, UNSU_, UNSIK): New code_attrs.
7093 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7094 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7096 ("eq<mode>3"): ... this.
7097 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7099 ("ne<mode>3"): ... this.
7101 2017-11-07 Julia Koval <julia.koval@intel.com>
7104 * common/config/i386/i386-common.c
7105 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7106 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7107 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7108 * config/i386/i386.opt: Ditto.
7109 * config/i386/i386.c (ix86_target_string): Ditto.
7110 (ix86_option_override_internal): Ditto.
7111 (ix86_init_mpx_builtins): Move MPX to args2.
7112 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7113 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7114 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7115 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7116 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7117 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7118 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7120 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7123 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7124 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7125 (zero-extendsidi peephole2): Remove peephole.
7127 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7130 * stor-layout.c: Include attribs.h.
7131 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7132 explicit lookup of "aligned" attribute.
7134 2017-11-07 Andrew Waterman <andrew@sifive.com>
7136 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7137 (riscv_expand_block_move): Likewise.
7138 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7140 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7141 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7142 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7143 (riscv_adjust_block_mem): Likewise.
7144 (riscv_block_move_loop): Likewise.
7145 (riscv_expand_block_move): Likewise.
7146 * config/riscv/riscv.md (movmemsi): New pattern.
7148 2017-11-07 Michael Clark <michaeljclark@mac.com>
7150 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7151 (MUSL_DYNAMIC_LINKER): Likewise.
7153 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7155 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7156 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7159 2017-11-07 Richard Biener <rguenther@suse.de>
7161 * match.pd: Fix build.
7163 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7164 Jackson Woodruff <jackson.woodruff@arm.com>
7166 PR tree-optimization/71026
7167 * match.pd: Canonicalize negate in division.
7169 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7172 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7174 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7176 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7177 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7179 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7181 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7182 non-scalar integral types.
7183 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7184 (-(A-B), -(~A)): New transformations.
7186 2017-11-07 Tom de Vries <tom@codesourcery.com>
7188 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7189 semicolon after "do {} while (0)".
7190 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7191 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7192 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7193 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7194 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7196 2017-11-07 Tom de Vries <tom@codesourcery.com>
7198 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7199 after "do {} while (0)".
7200 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7201 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7202 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7203 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7204 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7206 2017-11-07 Tom de Vries <tom@codesourcery.com>
7209 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7211 (arm_rtx_costs_internal): Add missing semicolon after
7212 HANDLE_NARROW_SHIFT_ARITH call.
7214 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7216 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7217 disable isel if it was not set explicitly.
7219 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7221 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7222 (add_type_attribute) likewise.
7224 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7226 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7227 of struct ix86_frame.
7228 (ix86_initial_elimination_offset): Likewise.
7229 (ix86_expand_split_stack_prologue): Likewise.
7231 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7233 * tree-vrp.h (enum value_range_type): Update stale comment.
7235 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7236 Alan Hayward <alan.hayward@arm.com>
7237 David Sherwood <david.sherwood@arm.com>
7239 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7240 (aarch64_expand_vec_perm_const): Take the number of units too.
7241 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7242 (aarch64_expand_vec_perm_const): Likewise.
7243 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7244 (vec_perm<mode>): Update accordingly.
7246 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7247 Alan Hayward <alan.hayward@arm.com>
7248 David Sherwood <david.sherwood@arm.com>
7250 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7251 Take the number of units too.
7252 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7253 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7254 but check for a vector mode before rather than after the call.
7255 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7256 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7257 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7258 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7259 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7260 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7261 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7262 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7263 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7264 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7265 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7266 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7267 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7268 (aarch64_sqdmull2_n<mode>): Update accordingly.
7270 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7271 Alan Hayward <alan.hayward@arm.com>
7272 David Sherwood <david.sherwood@arm.com>
7274 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7275 the number of units too.
7276 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7277 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7278 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7279 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7280 (vec_store_lanesxi<mode>): Update accordingly.
7282 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7283 Alan Hayward <alan.hayward@arm.com>
7284 David Sherwood <david.sherwood@arm.com>
7286 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7287 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7288 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7289 of units rather than the mode.
7290 * config/aarch64/iterators.md (nunits): New mode attribute.
7291 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7292 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7293 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7294 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7295 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7296 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7297 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7298 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7299 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7300 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7301 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7302 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7303 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7304 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7305 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7306 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7307 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7308 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7309 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7310 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7311 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7312 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7313 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7314 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7315 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7316 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7317 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7318 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7319 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7320 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7321 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7322 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7323 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7324 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7325 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7326 (aarch64_simd_vec_setv2di): Likewise.
7328 2017-11-06 Carl Love <cel@us.ibm.com>
7330 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7332 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7333 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7335 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7337 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7338 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7339 Add power 8 macro expansions.
7340 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7341 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7342 power 8 instructions. (VSX_XXBR): Add iterator.
7344 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7346 * config/arm/arm.md (predicable_short_it): Change default to "no",
7347 improve documentation, remove uses that are identical to the default.
7348 (enabled_for_depr_it): Rename to enabled_for_short_it.
7349 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7350 Remove default uses.
7351 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7352 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7353 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7354 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7356 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7359 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7360 float128 helper macros, which are no longer used after deleting
7361 the old 'q' built-in functions, and moving the round to odd
7362 built-in functions to being special built-in functions.
7363 (BU_FLOAT128_2): Likewise.
7364 (BU_FLOAT128_1_HW): Likewise.
7365 (BU_FLOAT128_2_HW): Likewise.
7366 (BU_FLOAT128_3_HW): Likewise.
7367 (FABSQ): Delete old 'q' built-in functions.
7368 (COPYSIGNQ): Likewise.
7369 (SQRTF128_ODD): Move round to odd built-in functions to be
7370 special built-in functions, so that we can handle
7371 -mabi=ieeelongdouble.
7372 (TRUNCF128_ODD): Likewise.
7373 (ADDF128_ODD): Likewise.
7374 (SUBF128_ODD): Likewise.
7375 (MULF128_ODD): Likewise.
7376 (DIVF128_ODD): Likewise.
7377 (FMAF128_ODD): Likewise.
7378 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7379 built-in names to 'f128'.
7380 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7381 old 'q' built-in functions, as the machine independent code for
7382 'f128' built-in functions handles this.
7383 (rs6000_expand_builtin): Add expansion for float128 round to odd
7384 functions, keying off on -mabi=ieeelongdouble of whether to use
7385 the KFmode or TFmode variant.
7386 (rs6000_init_builtins): Initialize the _Float128 round to odd
7388 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7389 _Float128 'q' built-in functions are now mapped into the new
7390 'f128' built-in functions.
7392 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7394 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7395 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7397 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7400 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7401 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7402 * toplev.c: Include "ipa-fnsummary.h".
7403 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7405 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7407 PR tree-optimization/82838
7408 * gimple-ssa-store-merging.c
7409 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7410 on a separate gimple_seq which is then appended to seq.
7412 2017-11-06 Jeff Law <law@redhat.com>
7415 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7416 (get_probe_interval): New functions.
7417 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7418 (ix86_adjust_stack_and_probe): Likewise.
7419 (output_adjust_stack_and_probe): Likewise.
7420 (ix86_emit_probe_stack_range): Likewise.
7421 (ix86_expand_prologue): Likewise.
7423 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7425 PR tree-optimization/82816
7426 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7427 if the modes of the two types are the same.
7428 (convert_plusminus_to_widen): Likewise.
7430 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7432 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7434 (usadv16qi): New define_expand.
7435 (usadv8hi): New define_expand.
7437 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7440 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7442 (inline_transform): Likewise.
7443 * predict.c (counts_to_freqs): Remove useless conditional.
7444 * profile-count.h (profile_count::apply_scale): Move sanity check.
7445 * tree-inline.c (copy_bb): Always scale.
7446 (copy_cfg_body): Likewise.
7448 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7451 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7453 (*cmp_ite0): Add enabled_for_depr_it attribute.
7454 (*cmp_ite1): Likewise.
7456 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7458 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7461 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7463 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7464 return the value by pointer.
7465 (extract_range_from_multiplicative_op_1): Update accordingly.
7466 Return as soon as an operation fails.
7468 2017-11-05 Tom de Vries <tom@codesourcery.com>
7471 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7472 (DEF_SANITIZER_BUILTIN): ... here.
7473 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7474 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7476 2017-11-05 Tom de Vries <tom@codesourcery.com>
7479 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7481 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7482 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7483 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7485 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7486 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7487 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7489 2017-11-05 Tom de Vries <tom@codesourcery.com>
7492 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7495 2017-11-04 Michael Clark <michaeljclark@mac.com>
7497 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7498 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7500 (*addsi3_extended): Likewise.
7501 (*addsi3_extended2): Likewise.
7502 (<optab>si3): Likewise.
7503 (<optab>di3): Likewise.
7504 (<optab><mode>3): Likewise.
7505 (<*optabe>si3_internal): Likewise.
7506 (zero_extendqi<SUPERQI:mode>2): Likewise.
7507 (*add<mode>hi3): Likewise.
7508 (*xor<mode>hi3): Likewise.
7509 (<optab>di3): Likewise.
7510 (*<optab>si3_extend): Likewise.
7511 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7512 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7513 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7515 2017-11-04 Andrew Waterman <andrew@sifive.com>
7517 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7518 TARGET_STRICT_ALIGN based upon -mtune argument.
7520 2017-11-04 Andrew Waterman <andrew@sifive.com>
7522 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7524 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7526 * config/i386/i386.c (choose_basereg): Use optional scratch
7527 register and add assertion.
7528 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7529 needed, and don't allocate stack.
7530 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7531 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7532 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7534 2017-11-03 Jeff Law <law@redhat.com>
7536 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7537 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7538 to probe at the start of a noreturn function.
7540 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7542 PR tree-optimization/78821
7543 * gimple-ssa-store-merging.c: Update the file comment.
7544 (MAX_STORE_ALIAS_CHECKS): Define.
7545 (struct store_operand_info): New type.
7546 (store_operand_info::store_operand_info): New constructor.
7547 (struct store_immediate_info): Add rhs_code and ops data members.
7548 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7549 and op1r arguments to the ctor, initialize corresponding data members.
7550 (struct merged_store_group): Add load_align_base and load_align
7552 (merged_store_group::merged_store_group): Initialize them.
7553 (merged_store_group::do_merge): Update them.
7554 (merged_store_group::apply_stores): Pick the constant for
7555 encode_tree_to_bitpos from one of the two operands, or skip
7556 encode_tree_to_bitpos if neither operand is a constant.
7557 (class pass_store_merging): Add process_store method decl. Remove
7558 bool argument from terminate_all_aliasing_chains method decl.
7559 (pass_store_merging::terminate_all_aliasing_chains): Remove
7560 var_offset_p argument and corresponding handling.
7561 (stmts_may_clobber_ref_p): New function.
7562 (compatible_load_p): New function.
7563 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7564 if there is overlap and rhs_code is not INTEGER_CST. For
7565 non-overlapping stores terminate group if rhs is not mergeable.
7566 (get_alias_type_for_stmts): Change first argument from
7567 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7568 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7569 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7571 (get_location_for_stmts): Change first argument from
7572 auto_vec<gimple *> & to vec<gimple *> &.
7573 (struct split_store): Remove orig_stmts data member, add orig_stores.
7574 (split_store::split_store): Create orig_stores rather than orig_stmts.
7575 (find_constituent_stmts): Renamed to ...
7576 (find_constituent_stores): ... this. Change second argument from
7577 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7578 to info structures rather than the statements.
7579 (split_group): Rename ALLOW_UNALIGNED argument to
7580 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7581 it. Adjust find_constituent_stores caller.
7582 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7583 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7584 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7585 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7586 (mem_valid_for_store_merging): New function.
7587 (handled_load): New function.
7588 (pass_store_merging::process_store): New method.
7589 (pass_store_merging::execute): Use process_store method. Adjust
7590 terminate_all_aliasing_chains caller.
7592 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7594 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7595 Return true for more constants, symbols and label references.
7596 (aarch64_valid_floating_const): Remove unused function.
7598 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7601 * config/aarch64/aarch64.c (aarch64_layout_frame):
7602 Undo forcing of LR at bottom of frame.
7604 2017-11-03 Jeff Law <law@redhat.com>
7607 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7608 for int_registers_saved.
7610 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7611 extracted from tree-ssa-dom.c.
7612 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7613 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7614 (record_equivalences_from_incoming_edge): Add additional argument
7615 to single_pred_edge_ignoring_loop_edges call.
7616 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7617 (uncprop_dom_walker::before_dom_children): Add additional argument
7618 to single_pred_edge_ignoring_loop_edges call.
7619 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7620 single_pred_edge_ignoring_loop_edges rather than open coding.
7621 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7623 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7625 * match.pd (-(-A)): Rewrite.
7627 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7629 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7630 (rs6000_emit_int_cmove): New declaration.
7631 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7632 (rs6000_emit_sISEL): Delete.
7633 (rs6000_emit_int_cmove): Make non-static.
7634 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7635 instead of rs6000_emit_sISEL.
7637 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7639 * asan.c (create_cond_insert_point): Maintain profile.
7640 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7642 * basic-block.h (struct basic_block_def): Remove frequency.
7643 (EDGE_FREQUENCY): Use to_frequency
7644 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7646 (find_traces): Update to use to_frequency.
7647 (find_traces_1_round): Likewise; use only IPA counts.
7648 (bb_to_key): Likewise.
7649 (connect_traces): Use IPA counts only.
7650 (copy_bb_p): Update to use to_frequency.
7651 (fix_up_crossing_landing_pad): Likewise.
7652 (sanitize_hot_paths): Likewise.
7653 * bt-load.c (basic_block_freq): Likewise.
7654 * cfg.c (init_flow): Set count_max to uninitialized.
7655 (check_bb_profile): Remove frequencies; check counts.
7656 (dump_bb_info): Do not dump frequencies.
7657 (update_bb_profile_for_threading): Update counts only.
7658 (scale_bbs_frequencies_int): Likewise.
7659 (MAX_SAFE_MULTIPLIER): Remove.
7660 (scale_bbs_frequencies_gcov_type): Update counts only.
7661 (scale_bbs_frequencies_profile_count): Update counts only.
7662 (scale_bbs_frequencies): Update counts only.
7663 * cfg.h (struct control_flow_graph): Add count-max.
7664 (update_bb_profile_for_threading): Update prototype.
7665 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7666 (find_many_sub_basic_blocks): Likewise.
7667 * cfgcleanup.c (try_forward_edges): Likewise.
7668 (try_crossjump_to_edge): Likewise.
7669 * cfgexpand.c (expand_gimple_cond): Likewise.
7670 (expand_gimple_tailcall): Likewise.
7671 (construct_init_block): Likewise.
7672 (construct_exit_block): Likewise.
7673 * cfghooks.c (verify_flow_info): Check consistency of counts.
7674 (dump_bb_for_graph): Do not dump frequencies.
7675 (split_block_1): Do not update frequencies.
7676 (split_edge): Do not update frequencies.
7677 (make_forwarder_block): Do not update frequencies.
7678 (duplicate_block): Do not update frequencies.
7679 (account_profile_record): Do not update frequencies.
7680 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7681 for global heuristics.
7682 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7683 (expected_loop_iterations_unbounded): Use counts only.
7684 * cfgloopmanip.c (scale_loop_profile): Simplify.
7685 (create_empty_loop_on_edge): Simplify
7687 (duplicate_loop_to_header_edge): Simplify
7688 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7689 (update_br_prob_note): Take care of removing note when profile
7691 (relink_block_chain): Do not dump frequency.
7692 (rtl_account_profile_record): Use to_frequency.
7693 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7694 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7695 (cgraph_update_edges_for_call_stmt_node): Likewise.
7696 (cgraph_edge::verify_count_and_frequency): Update.
7697 (cgraph_node::verify_node): Temporarily disable frequency verification.
7698 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7699 to_cgraph_frequency.
7700 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7701 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7703 (cgraph_node::expand_thunk): Update profile.
7704 * except.c (dw2_build_landing_pads): Do not update frequency.
7705 * final.c (compute_alignments): Use to_frequency.
7706 (dump_basic_block_info): Do not dump frequency.
7707 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7708 (dump_gimple_bb_header): Do not dump frequency.
7709 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7711 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7712 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7713 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7714 (init_before_recovery): Do not update frequency.
7715 (sched_create_recovery_edges): Do not update frequency.
7716 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7717 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7718 (ipa_cp_c_finalize): Set max_count to uninitialized.
7719 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7720 (param_change_prob): Use counts.
7721 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7723 * ipa-split.c (consider_split): Use to_frequency.
7724 (split_function): Use to_frequency.
7725 * ira-build.c (loop_compare_func): Likewise.
7726 (mark_loops_for_removal): Likewise.
7727 (mark_all_loops_for_removal): Likewise.
7728 * loop-doloop.c (doloop_modify): Do not update frequency.
7729 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7731 * lto-streamer-in.c (input_function): Update count_max.
7732 * omp-expand.c (expand_omp_taskreg): Update count_max.
7733 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7734 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7735 (maybe_hot_count_p): Use ipa counts only.
7736 (maybe_hot_bb_p): Simplify.
7737 (maybe_hot_edge_p): Simplify.
7738 (probably_never_executed): Do not take frequency argument.
7739 (probably_never_executed_bb_p): Do not pass frequency.
7740 (probably_never_executed_edge_p): Likewise.
7741 (combine_predictions_for_bb): Check that profile is nonzero.
7742 (propagate_freq): Do not set frequency.
7743 (drop_profile): Simplify.
7744 (counts_to_freqs): Simplify.
7745 (expensive_function_p): Use to_frequency.
7746 (propagate_unlikely_bbs_forward): Simplify.
7747 (determine_unlikely_bbs): Simplify.
7748 (estimate_bb_frequencies): Add hack to silence graphite issues.
7749 (compute_function_frequency): Use ipa counts.
7750 (pass_profile::execute): Update.
7751 (rebuild_frequencies): Use counts only.
7752 (force_edge_cold): Use counts only.
7753 * profile-count.c (profile_count::dump): Dump new count types.
7754 (profile_count::differs_from_p): Check compatiblity.
7755 (profile_count::to_frequency): New function.
7756 (profile_count::to_cgraph_frequency): New function.
7757 * profile-count.h (struct function): Declare.
7758 (enum profile_quality): Add profile_guessed_local and
7759 profile_guessed_global0.
7760 (class profile_proability): Decrease number of bits to 29;
7761 update from_reg_br_prob_note and to_reg_br_prob_note.
7762 (class profile_count: Update comment; decrease number of bits
7763 to 61. Check compatibility.
7764 (profile_count::compatible_p): New private member function.
7765 (profile_count::ipa_p): New member function.
7766 (profile_count::operator<): Handle global zero correctly.
7767 (profile_count::operator>): Handle global zero correctly.
7768 (profile_count::operator<=): Handle global zero correctly.
7769 (profile_count::operator>=): Handle global zero correctly.
7770 (profile_count::nonzero_p): New member function.
7771 (profile_count::force_nonzero): New member function.
7772 (profile_count::max): New member function.
7773 (profile_count::apply_scale): Handle IPA scalling.
7774 (profile_count::guessed_local): New member function.
7775 (profile_count::global0): New member function.
7776 (profile_count::ipa): New member function.
7777 (profile_count::to_frequency): Declare.
7778 (profile_count::to_cgraph_frequency): Declare.
7779 * profile.c (OVERLAP_BASE): Delete.
7780 (compute_frequency_overlap): Delete.
7781 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7782 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7783 * sched-ebb.c (rank): Use counts only.
7784 * shrink-wrap.c (handle_simple_exit): Use counts only.
7785 (try_shrink_wrapping): Use counts only.
7786 (place_prologue_for_one_component): Use counts only.
7787 * tracer.c (find_best_predecessor): Use to_frequency.
7788 (find_trace): Use to_frequency.
7789 (tail_duplicate): Use to_frequency.
7790 * trans-mem.c (expand_transaction): Do not update frequency.
7791 * tree-call-cdce.c: Do not update frequency.
7792 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7793 (gimple_merge_blocks): Likewise.
7794 (gimple_split_edge): Likewise.
7795 (gimple_duplicate_sese_region): Likewise.
7796 (gimple_duplicate_sese_tail): Likewise.
7797 (move_sese_region_to_fn): Likewise.
7798 (gimple_account_profile_record): Likewise.
7799 (insert_cond_bb): Likewise.
7800 * tree-complex.c (expand_complex_div_wide): Likewise.
7801 * tree-eh.c (lower_resx): Update profile.
7802 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7804 (initialize_cfun): Do not initialize frequencies
7805 (freqs_to_counts): Delete.
7806 (copy_cfg_body): Ignore count parameter.
7807 (copy_body): Update.
7808 (expand_call_inline): Update count_max.
7809 (optimize_inline_calls): Update count_max.
7810 (tree_function_versioning): Update count_max.
7811 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7812 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7814 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7815 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7816 (try_peel_loop): Likewise.
7817 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7819 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7820 (tree_transform_and_unroll_loop): Do not use frequencies
7821 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7822 Use reliable prediction only.
7823 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7824 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7825 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7826 probability scaling.
7827 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7828 not update frequency
7829 (any_remaining_duplicated_blocks): Likewise.
7830 (update_profile): Likewise.
7831 (estimated_freqs_path): Delete.
7832 (freqs_to_counts_path): Delete.
7833 (clear_counts_path): Delete.
7834 (ssa_fix_duplicate_block_edges): Likewise.
7835 (duplicate_thread_path): Likewise.
7836 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7837 * tree-tailcall.c (decrease_profile): Do not update frequency.
7838 (eliminate_tail_call): Likewise.
7839 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7840 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7841 (optimize_mask_stores): Likewise.
7842 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7843 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7844 (ubsan_expand_ptr_ifn): Update profile.
7845 * value-prof.c (gimple_ic): Simplify.
7846 * value-prof.h (gimple_ic): Update prototype.
7847 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7848 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7850 (want_inline_self_recursive_call_p): Likewise.
7851 (resolve_noninline_speculation): Only cummulate defined counts.
7852 (inline_small_functions): Use nonzero_p.
7853 (ipa_inline): Do not access freed node.
7855 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7857 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7858 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
7860 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
7862 * config/riscv/riscv.c (riscv_legitimize_move): Handle
7863 non-legitimate address.
7865 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7867 * config/rs6000/rs6000.md (*lt0_disi): Delete.
7868 (*lt0_<mode>di, *lt0_<mode>si): New.
7870 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7872 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
7873 TARGET_PAIRED_FLOAT.
7875 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
7876 Jim Wilson <jim.wilson@linaro.org>
7878 * config/aarch64/aarch64-cores.def (saphira): New CPU.
7879 * config/aarch64/aarch64-tune.md: Regenerated.
7880 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
7881 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
7883 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
7885 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
7886 (arc_expand_prologue): Restore blink for millicode.
7887 * config/arc/linux.h (LINK_EH_SPEC): Defined.
7889 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
7892 * config/i386/i386.c (ix86_vector_duplicate_value): Use
7893 gen_vec_duplicate after forcing the scalar into a register.
7895 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
7897 * combine (try_combine): Print the insns input to try_combine to the
7900 2017-11-02 Steve Ellcey <sellcey@cavium.com>
7903 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
7904 Remove second argument from aarch64_process_target_attr call.
7905 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
7907 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
7909 (aarch64_handle_attr_arch): Remove second argument.
7910 (aarch64_handle_attr_cpu): Ditto.
7911 (aarch64_handle_attr_tune): Ditto.
7912 (aarch64_handle_attr_isa_flags): Ditto.
7913 (aarch64_process_one_target_attr): Ditto.
7914 (aarch64_process_target_attr): Ditto.
7915 (aarch64_option_valid_attribute_p): Remove second argument.
7916 on aarch64_process_target_attr call.
7918 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7920 * diagnostic.c: Include "selftest-diagnostic.h".
7921 (selftest::assert_location_text): New function.
7922 (selftest::test_diagnostic_get_location_text): New function.
7923 (selftest::diagnostic_c_tests): Call it.
7925 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7927 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
7928 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
7929 (class selftest::test_diagnostic_context): Move to...
7930 * selftest-diagnostic.c: New file.
7931 * selftest-diagnostic.h: New file.
7933 2017-11-02 James Bowman <james.bowman@ftdichip.com>
7935 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
7936 offset range for FT32B.
7937 * config/ft32/ft32.h: option "mcompress" enables relaxation.
7938 * config/ft32/ft32.md: Add TARGET_NOPM.
7939 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
7940 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
7942 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
7944 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
7946 2017-11-02 Jeff Law <law@redhat.com>
7948 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
7949 virtual keyword on FINAL OVERRIDE members.
7951 * tree-ssa-propagate.h (ssa_propagation_engine): Group
7952 virtuals together. Add virtual destructor.
7953 (substitute_and_fold_engine): Similarly.
7955 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
7957 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
7959 2017-11-02 Richard Biener <rguenther@suse.de>
7961 PR tree-optimization/82795
7962 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
7964 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7966 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
7968 (gcc_GAS_CHECK_FEATURE): Remove.
7969 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
7970 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
7971 * configure: Regenerate.
7973 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
7975 * config/arc/arc.c (hwloop_optimize): Account for empty
7978 2017-11-02 Richard Biener <rguenther@suse.de>
7981 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
7982 Truncate ARRAY_REF index and element size.
7984 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7986 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
7988 2017-11-01 Jeff Law <law@redhat.com>
7990 * tree-ssa-ccp.c (ccp_folder): New class derived from
7991 substitute_and_fold_engine.
7992 (ccp_folder::get_value): New member function.
7993 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
7994 (ccp_fold_stmt): Remove prototype.
7995 (ccp_finalize): Call substitute_and_fold from the ccp_class.
7996 * tree-ssa-copy.c (copy_folder): New class derived from
7997 substitute_and_fold_engine.
7998 (copy_folder::get_value): Renamed from get_value.
7999 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8000 * tree-vrp.c (vrp_folder): New class derived from
8001 substitute_and_fold_engine.
8002 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8003 (vrp_folder::get_value): New member function.
8004 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8005 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8006 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8007 provide a class interface to folder/substitute routines.
8008 (ssa_prop_fold_stmt_fn): Remove typedef.
8009 (ssa_prop_get_value_fn): Likewise.
8010 (subsitute_and_fold): Remove prototype.
8011 (replace_uses_in): Likewise.
8012 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8013 Renamed from replace_uses_in. Call the virtual member function
8014 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8015 (substitute_and_fold_dom_walker): Remove initialization of
8016 data member entries for calbacks. Add substitute_and_fold_engine
8017 member and initialize it.
8018 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8019 member functions for get_value, replace_phi_args_in c
8020 replace_uses_in, and fold_stmt calls.
8021 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8022 substitute_and_fold. Remove assert. Update ctor call.
8024 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8025 (ssa_prop_visit_phi_fn): Likewise.
8026 (class ssa_propagation_engine): New class to provide an interface
8028 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8030 (ssa_prop_visit_phi): Likewise.
8031 (ssa_propagation_engine::simulate_stmt): Moved into class.
8032 Call visit_phi/visit_stmt from the class rather than via
8033 file scoped static variables.
8034 (ssa_propagation_engine::simulate_block): Moved into class.
8035 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8036 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8037 set file scoped statics for the visit_stmt/visit_phi callbacks.
8038 * tree-complex.c (complex_propagate): New class derived from
8039 ssa_propagation_engine.
8040 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8041 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8042 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8044 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8045 ssa_propagation_engine.
8046 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8047 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8048 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8049 * tree-ssa-copy.c (copy_prop): New class derived from
8050 ssa_propagation_engine.
8051 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8052 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8053 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8054 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8055 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8056 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8057 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8059 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8061 PR rtl-optimization/82778
8062 PR rtl-optimization/82597
8063 * compare-elim.c (struct comparison): Add in_a_setter field.
8064 (find_comparison_dom_walker::before_dom_children): Remove killed
8065 bitmap and df_simulate_find_defs call, instead walk the defs.
8066 Compute last_setter and initialize in_a_setter. Merge definitions
8067 with first initialization for a few variables.
8068 (try_validate_parallel): Use insn_invalid_p instead of
8069 recog_memoized. Return insn rather than just the pattern.
8070 (try_merge_compare): Fix up comment. Don't uselessly test if
8071 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8073 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8076 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8077 Alan Hayward <alan.hayward@arm.com>
8078 David Sherwood <david.sherwood@arm.com>
8080 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8081 aarch64_hard_regno_nregs to get the number of registers
8084 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8085 Alan Hayward <alan.hayward@arm.com>
8086 David Sherwood <david.sherwood@arm.com>
8088 * config/aarch64/constraints.md (Upl): Rename to...
8090 * config/aarch64/aarch64.md
8091 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8094 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8095 Alan Hayward <alan.hayward@arm.com>
8096 David Sherwood <david.sherwood@arm.com>
8098 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8099 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8102 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8103 Alan Hayward <alan.hayward@arm.com>
8104 David Sherwood <david.sherwood@arm.com>
8106 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8107 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8108 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8110 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8111 of a single element.
8112 * config/aarch64/iterators.md: Add a comment above the permute
8113 unspecs to say that they are generated directly by
8114 aarch64_expand_vec_perm_const.
8115 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8117 2017-11-01 Nathan Sidwell <nathan@acm.org>
8119 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8121 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8123 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8124 and medany code models, and describe what they do.
8126 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8128 Revert accidental duplicate:
8130 * combine.c (can_change_dest_mode): Reject changes in
8131 REGMODE_NATURAL_SIZE.
8133 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8135 PR rtl-optimization/64682
8136 PR rtl-optimization/69567
8137 PR rtl-optimization/69737
8138 PR rtl-optimization/82683
8139 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8140 register mentioned in the note, drop the note, unless it came from I3,
8141 in which case it should go to I3 again.
8143 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8145 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8146 and return false if not.
8147 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8149 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8151 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8152 range is known to be empty.
8154 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8155 Alan Hayward <alan.hayward@arm.com>
8156 David Sherwood <david.sherwood@arm.com>
8158 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8159 and CONST_VECTOR_NUNITS instead of computing the number of units from
8160 the byte sizes of the vector and element.
8161 (simplify_binary_operation_1): Likewise.
8162 (simplify_const_binary_operation): Likewise.
8163 (simplify_ternary_operation): Likewise.
8165 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8166 Alan Hayward <alan.hayward@arm.com>
8167 David Sherwood <david.sherwood@arm.com>
8169 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8170 (int_mem_offset): ...this new function.
8171 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8172 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8173 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8176 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8177 Alan Hayward <alan.hayward@arm.com>
8178 David Sherwood <david.sherwood@arm.com>
8180 * lower-subreg.c (interesting_mode_p): New function.
8181 (compute_costs, find_decomposable_subregs, decompose_register)
8182 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8183 (resolve_clobber, dump_choices): Use it.
8185 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8186 Alan Hayward <alan.hayward@arm.com>
8187 David Sherwood <david.sherwood@arm.com>
8189 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8191 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8192 Alan Hayward <alan.hayward@arm.com>
8193 David Sherwood <david.sherwood@arm.com>
8195 * alias.c (find_base_value, find_base_term): Only process integer
8196 truncations. Check the precision rather than the size.
8198 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8199 Alan Hayward <alan.hayward@arm.com>
8200 David Sherwood <david.sherwood@arm.com>
8202 * machmode.h (is_narrower_int_mode): New function
8203 * optabs.c (expand_float, expand_fix): Use it.
8204 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8206 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8207 Alan Hayward <alan.hayward@arm.com>
8208 David Sherwood <david.sherwood@arm.com>
8210 * rtl.h (narrower_subreg_mode): New function.
8211 * ira-color.c (update_costs_from_allocno): Use it.
8213 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8214 Alan Hayward <alan.hayward@arm.com>
8215 David Sherwood <david.sherwood@arm.com>
8217 * optabs-query.h (convert_optab_p): New function, split out from...
8218 (convert_optab_handler): ...here.
8219 (widening_optab_handler): Delete.
8220 (find_widening_optab_handler): Remove permit_non_widening parameter.
8221 (find_widening_optab_handler_and_mode): Likewise. Provide an
8222 override that operates on mode class wrappers.
8223 * optabs-query.c (widening_optab_handler): Delete.
8224 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8225 parameter. Assert that the two modes are the same class and that
8226 the "from" mode is narrower than the "to" mode. Use
8227 convert_optab_handler instead of widening_optab_handler.
8228 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8229 instead of widening_optab_handler.
8230 * expr.c (expand_expr_real_2): Update calls to
8231 find_widening_optab_handler.
8232 * optabs.c (expand_widen_pattern_expr): Likewise.
8233 (expand_binop_directly): Take the insn_code as a parameter.
8234 (expand_binop): Only call find_widening_optab_handler for
8235 conversion optabs; use optab_handler otherwise. Update calls
8236 to find_widening_optab_handler and expand_binop_directly.
8237 Use convert_optab_handler instead of widening_optab_handler.
8238 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8239 find_widening_optab_handler and use scalar_mode rather than
8241 (convert_plusminus_to_widen): Likewise.
8243 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8244 Alan Hayward <alan.hayward@arm.com>
8245 David Sherwood <david.sherwood@arm.com>
8247 * machmode.h (fixed_size_mode): New class.
8248 * rtl.h (get_pool_mode): Return fixed_size_mode.
8249 * gengtype.c (main): Add fixed_size_mode.
8250 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8251 (get_raw_arg_mode): Likewise.
8252 * doc/tm.texi: Regenerate.
8253 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8254 * targhooks.c (default_get_reg_raw_mode): Likewise.
8255 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8256 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8257 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8258 (msp430_get_raw_result_mode): Likewise.
8259 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8260 * dbxout.c (dbxout_parms): Require fixed-size modes.
8261 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8262 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8263 * omp-low.c (lower_oacc_reductions): Likewise.
8264 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8265 (simplify_subreg): Update accordingly.
8266 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8267 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8268 that aren't fixed-size.
8269 (get_pool_mode): Return a fixed_size_mode.
8270 (output_constant_pool_2): Take a fixed_size_mode.
8272 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8273 Alan Hayward <alan.hayward@arm.com>
8274 David Sherwood <david.sherwood@arm.com>
8276 * doc/rtl.texi (vec_series): Document.
8277 (const): Say that the operand can be a vec_series.
8278 * rtl.def (VEC_SERIES): New rtx code.
8279 * rtl.h (const_vec_series_p_1): Declare.
8280 (const_vec_series_p): New function.
8281 * emit-rtl.h (gen_const_vec_series): Declare.
8282 (gen_vec_series): Likewise.
8283 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8284 (gen_vec_series): Likewise.
8285 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8286 * simplify-rtx.c (simplify_unary_operation): Handle negations
8288 (simplify_binary_operation_series): New function.
8289 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8290 (test_vector_ops_series): New function.
8291 (test_vector_ops): Call it.
8292 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8293 gen_const_vec_series.
8294 (altivec_lvsr): Likewise.
8295 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8297 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8298 Alan Hayward <alan.hayward@arm.com>
8299 David Sherwood <david.sherwood@arm.com>
8301 * doc/rtl.texi (const): Update description of address constants.
8302 Say that vector constants are allowed too.
8303 * common.md (E, F): Use CONSTANT_P instead of checking for
8305 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8306 checking for CONST_VECTOR.
8307 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8309 * expr.c (expand_expr_real_2): Check for vector modes instead
8310 of checking for CONST_VECTOR.
8311 * rtl.h (const_vec_p): New function.
8312 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8313 (unwrap_const_vec_duplicate): Handle them here too.
8315 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8316 David Malcolm <dmalcolm@redhat.com>
8317 Alan Hayward <alan.hayward@arm.com>
8318 David Sherwood <david.sherwood@arm.com>
8320 * rtl.h (vec_duplicate_p): New function.
8321 * selftest-rtl.c (assert_rtx_eq_at): New function.
8322 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8323 (assert_rtx_eq_at): Declare.
8324 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8325 * selftest-run-tests.c (selftest::run_tests): Call it.
8326 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8327 (simplify_unary_operation_1): Recursively handle vector duplicates.
8328 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8330 (simplify_subreg): Handle subregs of vector duplicates.
8331 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8332 (selftest::simplify_rtx_c_tests): New functions.
8334 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8335 Alan Hayward <alan.hayward@arm.com>
8336 David Sherwood <david.sherwood@arm.com>
8338 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8339 (gen_vec_duplicate): Likewise.
8340 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8342 (gen_const_vector): ...here.
8343 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8344 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8345 whose elements are all equal.
8346 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8347 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8348 (simplify_relational_operation): Likewise.
8349 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8351 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8352 (aarch64_expand_vector_init): Likewise.
8353 * config/arm/arm.c (neon_vdup_constant): Likewise.
8354 (neon_expand_vector_init): Likewise.
8355 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8356 (arm_block_set_unaligned_vect): Likewise.
8357 (arm_block_set_aligned_vect): Likewise.
8358 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8359 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8360 (expand_vec_perm_even_odd_pack): Likewise.
8361 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8362 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8363 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8364 gen_const_vec_duplicate.
8365 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8366 * config/mips/mips.c (mips_gen_const_int_vector): Use
8367 gen_const_vec_duplicate.
8368 (mips_expand_vector_init): Use CONST0_RTX.
8369 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8370 (define_split): Use gen_const_vec_duplicate.
8371 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8372 (define_split): Use gen_const_vec_duplicate.
8373 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8374 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8375 * config/spu/spu.c (spu_const): Likewise.
8377 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8378 Alan Hayward <alan.hayward@arm.com>
8379 David Sherwood <david.sherwood@arm.com>
8381 * combine.c (can_change_dest_mode): Reject changes in
8382 REGMODE_NATURAL_SIZE.
8384 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8386 * configure.ac (--enable-libssp): New.
8387 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8388 trying to determine target-specific default. Adjust indentation.
8389 * configure: Regenerated.
8390 * doc/install.texi (Configuration): Expand --disable-libssp
8393 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8395 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8398 2017-10-31 Martin Jambor <mjambor@suse.cz>
8401 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8403 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8405 * auto-profile.c (autofdo_source_profile::read): Use
8406 UNKNOWN_LOCATION rather than 0.
8407 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8408 (warning_at): ...this overload.
8409 (warning_at_rich_loc_n): Rename to...
8410 (warning_n): ...this overload.
8411 (error_at_rich_loc): Rename to...
8412 (error_at): ...this overload.
8413 (pedwarn_at_rich_loc): Rename to...
8414 (pedwarn): ...this overload.
8415 (permerror_at_rich_loc): Rename to...
8416 (permerror): ...this overload.
8417 (inform_at_rich_loc): Rename to...
8418 (inform): ...this overload.
8419 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8420 (diagnostic_n_impl_richloc): Rename to...
8421 (diagnostic_n_impl): ...this rich_location *-based decl.
8422 (inform_at_rich_loc): Rename to...
8423 (inform): ...this, and add an assertion.
8424 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8425 (warning_at_rich_loc): Rename to...
8426 (warning_at): ...this, and add an assertion.
8427 (warning_at_rich_loc_n): Rename to...
8428 (warning_n): ...this, and add an assertion.
8429 (warning_n): Update location_t-based implementation for removal of
8430 location_t-based diagnostic_n_impl.
8431 (pedwarn_at_rich_loc): Rename to...
8432 (pedwarn): ...this, and add an assertion.
8433 (permerror_at_rich_loc): Rename to...
8434 (permerror): ...this, and add an assertion.
8435 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8436 (error_at_rich_loc): Rename to...
8437 (error_at): ...this, and add an assertion.
8438 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8439 (driver::do_spec_on_infiles): Likewise.
8440 * substring-locations.c (format_warning_va): Update for renaming
8441 of inform_at_rich_loc.
8443 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8445 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8446 _Float<N>X built-in functions so that the variant without the
8447 "__builtin_" prefix is only enabled for the GNU C and Objective C
8448 languages when they are in non-strict ANSI/ISO mode.
8449 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8450 * target.def (floatn_builtin_p): Add a target hook to control
8451 whether _Float<N> and _Float<N>X built-in functions without the
8452 "__builtin_" prefix are enabled, and return true for C and
8453 Objective C in the default hook. Include langhooks.h in
8455 * targhooks.h (default_floatn_builtin_p): Likewise.
8456 * targhooks.c (default_floatn_builtin_p): Likewise.
8457 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8458 floatn_builtin_p target hook.
8459 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8461 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8462 Eric Botcazou <ebotcazou@adacore.com>
8464 PR rtl-optimization/81803
8465 * lra-constraints.c (curr_insn_transform): Also reload the whole
8466 register for a strict subreg no wider than a word if this is for
8467 a WORD_REGISTER_OPERATIONS target.
8469 2017-10-31 Jason Merrill <jason@redhat.com>
8471 * gdbinit.in: Skip over inlines from timevar.h.
8473 2017-10-31 Martin Liska <mliska@suse.cz>
8475 * doc/gcov.texi: Document new option.
8476 * gcov.c (print_usage): Likewise print it.
8477 (process_args): Support the argument.
8478 (format_count): New function.
8479 (format_gcov): Use the function.
8481 2017-10-31 Martin Liska <mliska@suse.cz>
8483 * gcov.c (struct name_map): do not use typedef.
8484 Define operator== and operator<.
8485 (name_search): Remove.
8486 (name_sort): Remove.
8487 (main): Do not allocate names.
8488 (process_file): Add vertical space.
8489 (generate_results): Use std::find.
8490 (release_structures): Do not release memory.
8491 (find_source): Use std::find.
8493 2017-10-31 Martin Liska <mliska@suse.cz>
8495 * gcov.c (struct line_info): Remove it's typedef.
8496 (line_info::line_info): Add proper ctor.
8497 (line_info::has_block): Do not use a typedef.
8498 (struct source_info): Do not use typedef.
8499 (circuit): Likewise.
8500 (get_cycles_count): Likewise.
8501 (output_intermediate_file): Iterate via vector iterator.
8502 (add_line_counts): Use std::vector methods.
8503 (accumulate_line_counts): Likewise.
8504 (output_lines): Likewise.
8506 2017-10-31 Martin Liska <mliska@suse.cz>
8508 * gcov.c (struct source_info): Remove typedef.
8509 (source_info::source_info): Add proper ctor.
8510 (accumulate_line_counts): Use struct, not it's typedef.
8511 (output_gcov_file): Likewise.
8512 (output_lines): Likewise.
8513 (main): Do not allocate an array.
8514 (output_intermediate_file): Use size of vector container.
8515 (process_file): Resize the vector.
8516 (generate_results): Do not preallocate, use newly added vector
8518 (release_structures): Do not release sources.
8519 (find_source): Use vector methods.
8520 (add_line_counts): Do not use typedef.
8522 2017-10-31 Martin Liska <mliska@suse.cz>
8524 * doc/gcov.texi: Document that.
8525 * gcov.c (add_line_counts): Mark lines with a non-executed
8527 (output_line_beginning): Handle such lines.
8528 (output_lines): Pass new argument.
8529 (output_intermediate_file): Print it in intermediate format.
8531 2017-10-31 Martin Liska <mliska@suse.cz>
8533 * color-macros.h: New file.
8534 * diagnostic-color.c: Factor out color related to macros to
8536 * doc/gcov.texi: Document -k option.
8537 * gcov.c (INCLUDE_STRING): Include string.h.
8538 (print_usage): Add -k option.
8539 (process_args): Parse it.
8540 (pad_count_string): New function.
8541 (output_line_beginning): Likewise.
8542 (DEFAULT_LINE_START): New macro.
8543 (output_lines): Support color output.
8545 2017-10-31 Martin Liska <mliska@suse.cz>
8547 PR gcov-profile/82633
8548 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8549 their interaction with GCOV infrastructure.
8550 * configure.ac: Add -fkeep-{inline,static}-functions to
8552 * configure: Regenerate.
8554 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8557 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8559 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8562 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8563 into a register if it does not fit into an immediate offset field.
8565 2017-10-31 Olivier Hainque <hainque@adacore.com>
8567 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8569 2017-10-31 Julia Koval <julia.koval@intel.com>
8571 * config.gcc: Add gfniintrin.h.
8572 * config/i386/gfniintrin.h: New.
8573 * config/i386/i386-builtin-types.def
8574 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8575 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8576 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8577 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8578 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8579 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8580 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8581 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8582 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8583 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8584 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8585 * config/i386/immintrin.h: Include gfniintrin.h.
8586 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8588 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8590 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8592 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8594 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8595 (arm_ashldi3_1bit): Remove pattern.
8596 (ashrdi3): Remove shift by 1 expansion.
8597 (arm_ashrdi3_1bit): Remove pattern.
8598 (lshrdi3): Remove shift by 1 expansion.
8599 (arm_lshrdi3_1bit): Remove pattern.
8600 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8601 cost of ashldi3 by 1.
8602 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8603 (<shift>di3_neon): Likewise.
8605 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8607 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8608 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8609 and (*aarch64_simd_mov<VQ:mode>).
8610 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8611 pattern alternative.
8612 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8613 attributes to match pattern alternative.
8615 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8617 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8618 * config/rs6000/emmintrin.h: New file.
8619 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8621 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8623 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8624 * (movdi_vfp_cortexa8): Remove pattern.
8626 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8628 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8630 (Specific, alpha*-dec-osf5.1): Remove.
8631 (Specific, mips-sgi-irix5): Remove.
8632 (Specific, mips-sgi-irix6): Remove.
8634 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8637 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8638 arguments to clear_bit_region_be.
8640 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8642 * gimplify.c: Include memmodel.h.
8644 2017-10-30 Martin Jambor <mjambor@suse.cz>
8646 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8647 condition whether loop should be executed at all.
8649 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8651 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8652 gimple folding of vec_madd() intrinsics.
8653 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8654 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8655 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8657 2017-10-30 Richard Biener <rguenther@suse.de>
8659 PR tree-optimization/82762
8661 2017-10-23 Richard Biener <rguenther@suse.de>
8663 PR tree-optimization/82129
8665 2017-08-01 Richard Biener <rguenther@suse.de>
8667 PR tree-optimization/81181
8668 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8669 (compute_antic): ... end of iteration here.
8671 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8673 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8674 -std=iso9899:2017 and -std=gnu17.
8675 * doc/standards.texi (C Language): Document C17 support.
8676 * doc/cpp.texi (Overview): Mention -std=c17.
8677 (Standard Predefined Macros): Document C11 and C17 values of
8678 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8679 * doc/extend.texi (Inline): Do not list individual options for
8680 standards newer than C99.
8681 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8683 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8686 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8688 * asan.c (asan_finish_file): Align asan globals array by shadow
8691 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8694 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8695 (struct store_immediate_info): Add bitregion_start and bitregion_end
8697 (store_immediate_info::store_immediate_info): Add brs and bre
8698 arguments and initialize bitregion_{start,end} from those.
8699 (struct merged_store_group): Add bitregion_start, bitregion_end,
8700 align_base and mask fields. Drop unnecessary struct keyword from
8701 struct store_immediate_info. Add do_merge method.
8702 (clear_bit_region_be): Use memset instead of loop storing zeros.
8703 (merged_store_group::do_merge): New method.
8704 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8705 stores as long as the surrounding bitregions have no gaps.
8706 (merged_store_group::merge_overlapping): Use do_merge.
8707 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8708 is byte aligned, rather than requiring that start and width are
8709 byte aligned. Drop unnecessary struct keyword from
8710 struct store_immediate_info. Allocate and populate also mask array.
8711 Make start of the arrays relative to bitregion_start rather than
8712 start and size them according to bitregion_{end,start} difference.
8713 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8714 struct store_immediate_info.
8715 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8716 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8717 struct keyword from struct store_immediate_info.
8718 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8719 between stores as long as the surrounding bitregions have no gaps.
8721 (struct split_store): Add orig non-static data member.
8722 (split_store::split_store): Initialize orig to false.
8723 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8724 if there is exactly a single original stmt. Change stmts argument
8725 to pointer from reference, if NULL, don't push anything to it. Add
8726 first argument, use it to optimize skipping over orig stmts that
8727 are known to be before bitpos already. Simplify.
8728 (split_group): Return unsigned int count how many stores are or
8729 would be needed rather than a bool. Add allow_unaligned argument.
8730 Change split_stores argument from reference to pointer, if NULL,
8731 only do a dry run computing how many stores would be produced.
8732 Rewritten algorithm to use both alignment and misalign if
8733 !allow_unaligned and handle bitfield stores with gaps.
8734 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8735 from bitregion_start instead of start. Compute allow_unaligned
8736 here, if true, do 2 split_group dry runs to compute which one
8737 produces fewer stores and prefer aligned if equal. Punt if
8738 new count is bigger or equal than original before emitting any
8739 statements, rather than during that. Remove no longer needed
8740 new_ssa_names tracking. Replace num_stmts with
8741 split_stores.length (). Use 32-bit stack allocated entries
8742 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8743 if possible. Handle bitfields with gaps.
8744 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8745 Compute bitregion_{start,end} for the stores and construct
8746 store_immediate_info with that. Formatting fixes.
8748 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8751 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8752 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8754 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8756 * gimplify.c: Include tm_p.h.
8758 * common.opt (gcoff): Re-add as ignored option.
8759 (gcoff1, gcoff2, gcoff3): Likewise.
8761 * Makefile.in (OBJS): Delete sdbout.o.
8762 (GTFILES): Delete $(srcdir)/sdbout.c.
8763 * debug.h: Delete sdb_debug_hooks.
8764 * final.c: Delete sdbout.h include.
8765 (final_scan_insn): Delete SDB_DEBUG check.
8766 (rest_of_clean_state): Likewise.
8767 * output.h: Delete sdb_begin_function_line.
8770 * toplev.c: Delete sdbout.h include.
8771 (process_options): Delete SDB_DEBUG check.
8772 * tree-core.h (tree_type_common): Delete pointer field of
8774 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8775 TYPE_SYMTAB_POINTER.
8776 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8777 (TYPE_SYMTAB_IS_POINTER): Delete.
8778 (TYPE_SYMTAB_IS_DIE): Renumber.
8779 * xcoffout.c: Refer to former sdbout.c file.
8780 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8782 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8783 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8784 (Debugging Options): Delete -gcoff.
8785 (-gstabs): Delete SDB reference.
8787 (-gcoff@var{level}): Delete.
8788 * doc/passes.texi (Debugging information output): Delete SDB and
8789 sdbout.c references.
8790 * doc/tm.texi: Regenerate.
8791 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8792 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8794 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8795 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8796 (SDB_DEBUGGING_INFO): Delete.
8797 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8798 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8799 * target.def (output_source_filename): Delete COFF reference.
8801 * common.opt (gcoff): Delete.
8802 (gxcoff+): Update Negative chain.
8803 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8805 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8806 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8807 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8809 (expand_function_start): Change sdb reference to past tense.
8810 (expand_function_end): Change sdb reference to past tense.
8811 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8812 * opts.c (debug_type_names): Delete coff entry.
8813 (common_handle_option): Delete OPT_gcoff case.
8814 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8816 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8817 * config/cris/cris.h: Delete SDB reference in comment.
8818 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8819 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8820 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8821 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8823 (ix86_expand_prologue): Likewise.
8824 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8825 * config/ia64/ia64.h: Likewise.
8826 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8827 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8829 * config/mmix/mmix.h: Likewise.
8830 * config/nds32/nds32.c: Likewise.
8831 * config/stormy/storym16.h: Likewise.
8832 * config/visium/visium.h: Likewise.
8833 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8835 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8837 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8838 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8839 FRAME_POINTER_REGNUM point at high end of local var area.
8841 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8843 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8844 Move comment around. Do not reset best_edge for a copiable
8845 destination if the copy would cause a partition change.
8846 (better_edge_p): Remove redundant check.
8848 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8850 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8852 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8854 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8855 for math functions that have _Float<N> and _Float<N>X variants.
8856 (mathfn_built_in_2): Add support for math functions that have
8857 _Float<N> and _Float<N>X variants.
8858 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
8859 (expand_builtin_mathfn_ternary): Add support for fma with
8860 _Float<N> and _Float<N>X variants.
8861 (expand_builtin): Likewise.
8862 (fold_builtin_3): Likewise.
8863 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
8864 create math function _Float<N> and _Float<N>X variants as external
8866 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
8867 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
8868 the __builtin_ prefix and if not strict ansi, without the prefix.
8869 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
8870 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
8871 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
8872 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
8873 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
8874 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
8875 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
8876 function signatures for fma _Float<N> and _Float<N>X variants.
8877 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
8878 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
8879 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
8880 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
8881 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
8882 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
8883 * gencfn-macros.c (print_case_cfn): Add support for math functions
8884 that have _Float<N> and _Float<N>X variants.
8885 (print_define_operator_list): Likewise.
8886 (fltfn_suffixes): Likewise.
8888 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
8889 for math functions that have _Float<N> and _Float<N>X variants.
8890 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
8891 and _Float<N>X variants.
8892 (COPYSIGN): Likewise.
8895 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
8896 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
8898 (integer_valued_read_call_p): Likewise.
8899 * fold-const-call.c (fold_const_call_ss): Likewise.
8900 (fold_const_call_sss): Add support for copysign, fmin, and fmax
8901 _Float<N> and _Float<N>X variants.
8902 (fold_const_call_ssss): Add support for fma _Float<N> and
8903 _Float<N>X variants.
8904 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
8905 support for copysign and fma _Float<N> and _Float<N>X variants.
8906 (backprop::process_builtin_call_use): Likewise.
8907 * tree-call-cdce.c (can_test_argument_range); Add support for
8908 sqrt _Float<N> and _Float<N>X variants.
8909 (edom_only_function): Likewise.
8910 (get_no_error_domain): Likewise.
8911 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
8912 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
8913 copysign _Float<N> and _Float<N>X variants.
8914 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
8915 handled by machine independent code.
8916 (FMAF128): Likewise.
8917 * doc/cpp.texi (Common Predefined Macros): Document defining
8918 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
8919 fma _Float<N> and _Float<N>X variants.
8921 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8924 * config/i386/i386-modes.def (CCFPU): Remove definition.
8925 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
8926 (ix86_cc_modes_compatible): Ditto.
8927 (ix86_expand_carry_flag_compare): Ditto.
8928 (ix86_expand_int_movcc): Ditto.
8929 (ix86_expand_int_addcc): Ditto.
8930 (ix86_reverse_condition): Ditto.
8931 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
8932 Return true/false for unordered/ordered fp comparisons.
8933 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
8934 (ix86_prepare_fp_compare_args): Update for rename.
8935 (ix86_expand_fp_compare): Update for rename. Generate unordered
8936 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
8937 (ix86_expand_sse_compare_and_jump): Ditto.
8938 * config/i386/predicates.md (fcmov_comparison_operator):
8939 Remove CCFPU mode handling.
8940 (ix86_comparison_operator): Ditto.
8941 (ix86_carry_flag_operator): Ditto.
8942 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
8943 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
8944 (*cmpu<mode>_cc_i387): Ditto.
8945 (FPCMP): Remove mode iterator.
8946 (unord): Remove mode attribute.
8947 (unord_subst): New define_subst transformation
8948 (unord): New define_subst attribute.
8950 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
8951 (*cmpi<unord>xf_i387): Ditto.
8952 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
8953 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
8954 using unord_subst transformation.
8955 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
8956 (round_saeonly): Also handle CCFP mode.
8957 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
8958 Remove UNSPEC_SAHF unspec handling.
8960 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8962 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
8964 2017-10-27 Jeff Law <law@redhat.com>
8966 * gimple-ssa-sprintf.c: Include domwalk.h.
8967 (class sprintf_dom_walker): New class, derived from dom_walker.
8968 (sprintf_dom_walker::before_dom_children): New function.
8969 (struct call_info): Moved into sprintf_dom_walker class
8970 (compute_formath_length, handle_gimple_call): Likewise.
8971 (sprintf_length::execute): Call the dominator walker rather
8972 than walking the statements.
8974 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
8975 gimple statement locations.
8976 (check_array_bounds): Corresponding changes. Get the statement's
8977 location directly from wi->stmt.
8979 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
8982 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
8984 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8986 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
8987 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
8989 2017-10-27 Jakub Jelinek <jakub@redhat.com>
8992 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
8993 * config/i386/i386.c (maybe_get_pool_constant): Removed.
8994 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
8995 maybe_get_pool_constant.
8996 * config/i386/predicates.md (zero_extended_scalar_load_operand):
8999 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9001 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9002 2.26 caveat. Update gas and gld versions.
9003 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9006 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9008 * cgraph.h (set_malloc_flag): Declare.
9009 * cgraph.c (set_malloc_flag_1): New function.
9010 (set_malloc_flag): Likewise.
9011 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9012 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9014 (read_ipa_call_summary): Add support for reading is_return_callee.
9015 (write_ipa_call_summary): Stream is_return_callee.
9016 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9017 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9018 ipa-prop.h, ipa-fnsummary.h.
9019 (pure_const_names): Change to static.
9020 (malloc_state_e): Define.
9021 (malloc_state_names): Define.
9022 (funct_state_d): Add field malloc_state.
9023 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9024 (check_retval_uses): New function.
9025 (malloc_candidate_p): Likewise.
9026 (analyze_function): Add support for malloc attribute.
9027 (pure_const_write_summary): Stream malloc_state.
9028 (pure_const_read_summary): Add support for reading malloc_state.
9029 (dump_malloc_lattice): New function.
9030 (propagate_malloc): New function.
9031 (warn_function_malloc): New function.
9032 (ipa_pure_const::execute): Call propagate_malloc and
9033 ipa_free_fn_summary.
9034 (pass_local_pure_const::execute): Add support for malloc attribute.
9035 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9036 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9038 2017-10-27 Martin Liska <mliska@suse.cz>
9040 PR gcov-profile/82457
9041 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9042 for fork-like functions to be properly instrumented.
9044 2017-10-27 Richard Biener <rguenther@suse.de>
9047 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9048 info when we redirected EH.
9050 2017-10-26 Michael Collison <michael.collison@arm.com>
9052 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9054 (<optab>_trunchf<GPI:mode>2: New pattern.
9055 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9056 * config/aarch64/iterators.md (wv): New mode attribute.
9057 (vf, VF): New mode attributes.
9058 (vgp, VGP): New mode attributes.
9059 (s): Update attribute with SImode and DImode prefixes.
9061 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9063 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9064 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9065 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9066 (nios2_option_overide): Initialize it. Don't allow R0-relative
9067 addressing with PIC.
9068 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9069 (nios2_symbolic_constant_p): Likewise.
9070 (nios2_legitimate_address_p): Likewise.
9071 (nios2_r0rel_section_name_p): New.
9072 (nios2_symbol_ref_in_r0rel_data_p): New.
9073 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9074 (r0rel_constant_p): New.
9075 (nios2_print_operand_address): Handle r0rel_constant_p.
9076 (nios2_cdx_narrow_form_p): Likewise.
9077 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9078 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9079 (Nios II Options): Document -mr0rel-sec.
9081 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9083 * config/nios2/nios2.c: Include xregex.h.
9084 (nios2_gprel_sec_regex): New.
9085 (nios2_option_overide): Initialize it. Don't allow GP-relative
9086 addressing with PIC.
9087 (nios2_small_section_name_p): Check for regex match.
9088 * config/nios2/nios2.opt (mgprel-sec=): New option.
9089 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9090 (Nios II Options): Document -mgprel-sec.
9092 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9094 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9096 2017-10-26 Tom de Vries <tom@codesourcery.com>
9098 PR tree-optimization/82707
9099 * gimple.c (gimple_copy): Fix unsharing of
9100 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9102 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9104 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9105 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9106 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9107 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9108 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9109 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9110 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9111 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9112 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9113 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9114 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9115 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9116 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9117 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9118 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9119 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9120 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9122 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9124 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9126 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9127 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9128 warning to rs6000.c. Remove the Undocumented flag, since it has
9130 (-mabi=ibmlongdouble): Likewise.
9131 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9132 already set, set the default format for long double.
9133 (rs6000_debug_reg_global): Print whether long double is IBM or
9135 (rs6000_option_override_internal): Rework setting long double
9136 format. Only warn if the user is changing the long double default
9137 and they did not use -Wno-psabi.
9138 * doc/invoke.texi (PowerPC options): Update the documentation for
9139 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9141 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9142 Alan Hayward <alan.hayward@arm.com>
9143 David Sherwood <david.sherwood@arm.com>
9145 * rtl.h (wider_subreg_mode): New function.
9146 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9147 rather than an unsigned int *.
9148 * ira-color.c (regno_max_ref_width): Replace with...
9149 (regno_max_ref_mode): ...this new variable.
9150 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9151 Use wider_subreg_mode.
9152 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9153 rather than an unsigned int *.
9154 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9155 (process_alt_operands): Likewise.
9156 (invariant_p): Likewise.
9157 * lra-spills.c (assign_mem_slot): Likewise.
9158 (add_pseudo_to_slot): Likewise.
9159 * lra.c (collect_non_operand_hard_regs): Likewise.
9160 (add_regs_to_insn_regno_info): Likewise.
9161 * reload1.c (regno_max_ref_width): Replace with...
9162 (regno_max_ref_mode): ...this new variable.
9163 (reload): Update accordingly. Update call to
9164 ira_sort_regnos_for_alter_reg.
9165 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9166 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9167 (scan_paradoxical_subregs): Likewise.
9169 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9171 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9172 (aarch64_frame): Add emit_frame_chain boolean.
9173 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9174 Move eh_return case to aarch64_layout_frame.
9175 (aarch64_layout_frame): Initialize emit_frame_chain.
9176 (aarch64_expand_prologue): Use emit_frame_chain.
9178 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9180 * config/aarch64/aarch64.c (aarch64_layout_frame):
9181 Ensure LR is always stored at the bottom of the callee-saves.
9182 Remove rarely used frame layout which saves callee-saves at top of
9183 frame, so the store of LR can be used as a valid probe in all cases.
9185 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9187 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9188 Improve unaligned TImode/TFmode base/offset split.
9190 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9191 Alan Hayward <alan.hayward@arm.com>
9192 David Sherwood <david.sherwood@arm.com>
9194 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9195 * combine.c (find_single_use_1): Likewise.
9196 (expand_field_assignment): Likewise.
9197 (move_deaths): Likewise.
9198 * lra-constraints.c (simplify_operand_subreg): Likewise.
9199 (curr_insn_transform): Likewise.
9200 * lra.c (collect_non_operand_hard_regs): Likewise.
9201 (add_regs_to_insn_regno_info): Likewise.
9202 * rtlanal.c (reg_referenced_p): Likewise.
9203 (covers_regno_no_parallel_p): Likewise.
9205 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9207 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9208 Don't print any bits outside the precision of the value.
9209 * wide-int.cc (test_printing): Add some new tests.
9211 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9213 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9214 supports -xbrace_comment option.
9215 * configure: Regenerate.
9216 * config.in: Regenerate.
9217 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9218 (ASM_CPU_SPEC): Use it.
9220 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9222 * target.def (static_rtx_alignment): New hook.
9223 * targhooks.h (default_static_rtx_alignment): Declare.
9224 * targhooks.c (default_static_rtx_alignment): New function.
9225 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9226 * doc/tm.texi: Regenerate.
9227 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9228 instead of targetm.constant_alignment. Remove call to
9230 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9231 (cris_preferred_mininum_alignment): New function, split out from...
9232 (cris_constant_alignment): ...here.
9233 (cris_static_rtx_alignment): New function.
9234 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9236 (ix86_constant_alignment): ...here.
9237 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9238 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9239 (mmix_static_rtx_alignment): New function.
9240 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9241 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9243 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9246 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9247 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9249 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9251 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9252 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9253 mark as a sign-extending load.
9254 (local_pic_load_u): Define.
9256 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9259 * fold-const.c (operand_equal_for_comparison_p): Also return true
9260 if ARG0 is a simple variant of ARG1 with narrower precision.
9261 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9263 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9265 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9267 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9268 scatter_static, scatter_per_elt.
9269 * x86-tune-costs.h: Add new cost entries.
9271 2017-10-25 Richard Biener <rguenther@suse.de>
9273 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9274 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9275 class pass_fre): Move to ...
9276 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9277 class pass_fre): ... here and adjust for statistics.
9279 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9282 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9283 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9284 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9285 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9288 2017-10-25 Richard Biener <rguenther@suse.de>
9290 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9291 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9292 eliminate_push_avail, eliminate_insert): Move inside...
9293 (class eliminate_dom_walker): ... this class in preparation
9295 (fini_eliminate): Remove by merging with ...
9296 (eliminate): ... this function. Adjust for class changes.
9297 (pass_pre::execute): Remove fini_eliminate call.
9298 (pass_fre::execute): Likewise.
9300 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9303 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9304 (VPERMI2, VPERMI2I): New mode iterators.
9305 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9306 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9308 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9309 mode iterator. Remove 3 old define_insn patterns.
9310 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9311 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9312 VPERMI2 mode iterator, remove the other two expanders.
9313 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9314 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9315 instructions, remove the other two patterns.
9316 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9317 mode iterator, remove the other two patterns.
9318 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9319 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9320 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9321 and adjust argument order accordingly.
9322 (ix86_expand_vec_perm): Adjust caller.
9323 (expand_vec_perm_1): Likewise.
9324 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9325 (expand_vec_perm_vpermt2_vpshub2): ... this.
9326 (ix86_expand_vec_perm_const_1): Adjust caller.
9327 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9330 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9331 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9332 (vec_shl_<mode>): Remove unused expander.
9333 (avx512bw_<shift_insn><mode>3): New define_insn.
9334 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9335 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9337 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9340 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9343 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9345 PR rtl-optimization/82396
9346 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9347 (autopref_multipass_init): Simplify initialization.
9348 (autopref_rank_data): Simplify sort order.
9349 * gcc/sched-int.h (autopref_multipass_data_): Remove
9350 multi_mem_insn_p, min_offset and max_offset.
9352 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9355 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9356 Check special value of flag_omit_frame_pointer.
9357 (aarch64_can_eliminate): Likewise.
9358 (aarch64_override_options_after_change_1): Simplify handling of
9359 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9361 2017-10-24 Richard Biener <rguenther@suse.de>
9363 PR tree-optimization/82697
9364 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9365 zero for conditional load and unconditional store.
9367 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9369 * doc/install.texi: Document bootstrap-cet.
9371 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9374 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9375 ENDBR instruction at function entrance if function is only
9378 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9381 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9382 patterns to better describe from which operation the CF is computed.
9383 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9384 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9385 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9386 is 0, use _0 suffixed expanders instead of emitting a comparison
9389 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9391 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9392 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9394 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9397 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9398 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9399 * loop-iv.c (iv_get_reaching_def): Likewise.
9400 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9401 variable is promoted and the partition contains undefined values.
9403 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9405 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9407 (nios2_address_cost): Define.
9408 (nios2_legitimize_address): Recognize (exp + constant) directly.
9409 (TARGET_ADDRESS_COST): Define.
9411 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9413 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9414 (nios2_symbolic_memory_operand_p): Declare.
9415 (nios2_split_large_constant): Declare.
9416 (nios2_split_symbolic_memory_operand): Declare.
9417 * config/nios2/nios2.c: Adjust includes.
9418 (nios2_symbolic_constant_allowed): New.
9419 (nios2_symbolic_constant_p): New.
9420 (nios2_plus_symbolic_constant_p): New.
9421 (nios2_valid_addr_expr_p): Recognize addresses involving
9423 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9424 (nios2_symbolic_memory_operand_p): New.
9425 (nios2_large_constant_p): New.
9426 (nios2_split_large_constant): New.
9427 (nios2_split_plus_large_constant): New.
9428 (nios2_split_symbolic_memory_operand): New.
9429 (nios2_legitimize_address): Code refactoring. Handle addresses
9430 involving symbolic constants.
9431 (nios2_emit_move_sequence): Likewise.
9432 (nios2_print_operand): Improve error output.
9433 (nios2_print_operand_address): Handle LO_SUM.
9434 (nios2_cdx_narrow_form_p): Likewise.
9435 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9436 operands involving symbolic constants.
9437 (movhi_internal, movsi_internal): Likewise.
9438 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9439 (extendhisi2, extendqi<mode>2): Likewise.
9441 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9443 * tree-pass.h (PROP_rtl_split_insns): Define.
9444 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9446 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9448 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9450 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9453 * target.def (const_not_ok_for_debug_p): Default to
9454 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9455 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9456 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9457 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9458 which targetm.const_not_ok_for_debug_p returned true.
9459 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9461 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9463 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9464 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9465 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9467 (ix86_const_not_ok_for_debug_p): New function.
9468 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9469 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9471 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9474 * system.h: Conditionally include "unique-ptr.h" if
9475 INCLUDE_UNIQUE_PTR is defined.
9476 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9477 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9479 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9481 * config/rl78/rl78.md: New define_expand "subdi3".
9483 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9486 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9487 DF_REF_INSN if DF_REF_INSN_INFO is false.
9489 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9491 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9492 xmm_move instead of sse_move.
9493 (sse_store_index): New function.
9494 (ix86_register_move_cost): Be more sensible about mismatch stall;
9495 model AVX moves correctly; make difference between sse->integer and
9497 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9498 moves; make difference between SSE and AVX.
9499 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9500 and zmm_move. Increase size of sse load and store tables;
9501 add unaligned load and store tables; add ssemmx_to_integer.
9502 * x86-tune-costs.h: Update all entries according to real
9503 move latencies from Agner Fog's manual and chip documentation.
9505 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9508 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9509 * config/i386/constraints.md (Wf): New constraint.
9510 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9511 (cmp<dwi>_doubleword): Removed.
9512 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9513 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9514 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9515 expand with cmp<dwi>_doubleword. For LTU and GEU use
9516 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9518 * common.opt (gcolumn-info): Enable by default.
9519 * doc/invoke.texi (gcolumn-info): Document new default.
9521 2017-10-23 Richard Biener <rguenther@suse.de>
9523 PR tree-optimization/82672
9524 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9525 Fold the stmt if we propagated into it.
9527 2017-10-23 Richard Biener <rguenther@suse.de>
9529 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9530 (bitmap_remove_expr_from_set): ... this. All callers call this
9531 for non-constant values.
9532 (bitmap_set_subtract): Rename to...
9533 (bitmap_set_subtract_expressions): ... this. Adjust and
9535 (bitmap_set_contains_value): Remove superfluous check.
9536 (bitmap_set_replace_value): Inline into single caller ...
9537 (bitmap_value_replace_in_set): ... here and simplify.
9538 (dependent_clean): Merge into ...
9539 (clean): ... this using an overload. Adjust.
9540 (prune_clobbered_mems): Adjust.
9541 (compute_antic_aux): Likewise.
9542 (compute_partial_antic_aux): Likewise.
9544 2017-10-23 Richard Biener <rguenther@suse.de>
9546 PR tree-optimization/82129
9548 2017-08-01 Richard Biener <rguenther@suse.de>
9550 PR tree-optimization/81181
9551 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9552 (compute_antic): ... end of iteration here.
9554 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9556 * target.def (starting_frame_offset): New hook.
9557 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9558 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9559 * doc/tm.texi.in: Regenerate.
9560 * hooks.h (hook_hwi_void_0): Declare.
9561 * hooks.c (hook_hwi_void_0): New function.
9562 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9563 STARTING_FRAME_OFFSET.
9564 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9565 * reload1.c (reload): Likewise.
9566 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9567 instead of STARTING_FRAME_OFFSET.
9568 * function.c (try_fit_stack_local): Likewise.
9569 (assign_stack_local_1): Likewise
9570 (instantiate_virtual_regs): Likewise.
9571 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9572 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9573 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9574 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9575 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9576 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9577 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9578 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9579 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9580 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9581 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9582 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9583 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9584 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9585 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9586 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9587 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9588 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9589 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9590 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9591 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9592 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9593 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9594 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9595 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9596 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9597 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9598 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9599 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9600 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9601 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9602 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9603 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9604 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9605 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9606 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9607 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9608 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9609 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9610 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9611 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9612 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9613 return a HOST_WIDE_INT.
9614 (avr_builtin_setjmp_frame_value): Use it instead of
9615 STARTING_FRAME_OFFSET.
9616 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9617 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9618 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9620 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9621 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9622 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9623 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9624 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9625 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9626 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9627 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9628 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9629 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9630 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9631 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9633 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9634 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9635 * config/mips/mips.c (mips_compute_frame_info): Refer to
9636 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9637 (mips_starting_frame_offset): New function.
9638 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9639 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9640 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9641 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9642 and return a HOST_WIDE_INT.
9643 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9644 (mmix_initial_elimination_offset): Refer to
9645 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9646 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9647 * config/pa/pa.c (pa_starting_frame_offset): New function.
9648 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9649 (pa_expand_prologue): Likewise.
9650 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9651 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9652 !FRAME_GROWS_DOWNWARD handling to...
9653 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9654 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9655 !FRAME_GROWS_DOWNWARD handling to...
9656 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9657 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9658 !FRAME_GROWS_DOWNWARD handling to...
9659 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9660 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9662 (rs6000_starting_frame_offset): New function.
9663 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9664 !FRAME_GROWS_DOWNWARD handling to...
9665 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9666 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9667 !FRAME_GROWS_DOWNWARD handling to...
9668 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9669 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9670 !FRAME_GROWS_DOWNWARD handling to...
9671 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9672 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9673 (rs6000_starting_frame_offset): New function.
9674 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9675 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9676 * config/vax/vax.c (vax_starting_frame_offset): New function.
9677 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9678 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9679 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9680 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9681 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9682 * system.h (STARTING_FRAME_OFFSET): Poison.
9684 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9686 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9687 SCALAR_TYPE_MODE instead of TYPE_MODE.
9689 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9690 Alan Hayward <alan.hayward@arm.com>
9691 David Sherwood <david.sherwood@arm.com>
9693 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9695 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9696 Alan Hayward <alan.hayward@arm.com>
9697 David Sherwood <david.sherwood@arm.com>
9699 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9701 2017-10-23 Richard Biener <rguenther@suse.de>
9703 PR tree-optimization/82129
9704 * tree-ssa-pre.c (bitmap_set_and): Remove.
9705 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9706 canonicalizing expressions in the set to those with lowest
9707 ID rather than taking that from the first edge.
9709 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9711 * combine.c (rtx_equal_for_field_assignment_p): Use
9712 byte_lowpart_offset.
9714 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9715 Alan Hayward <alan.hayward@arm.com>
9716 David Sherwood <david.sherwood@arm.com>
9718 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9719 to a SUBREG_PROMOTED_VAR.
9721 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9722 Alan Hayward <alan.hayward@arm.com>
9723 David Sherwood <david.sherwood@arm.com>
9725 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9726 (expand_debug_source_expr): Likewise.
9727 * combine.c (combine_simplify_rtx): Likewise.
9728 * cse.c (fold_rtx): Likewise.
9729 * optabs.c (expand_float): Likewise.
9730 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9731 (simplify_binary_operation_1): Likewise.
9733 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9734 Alan Hayward <alan.hayward@arm.com>
9735 David Sherwood <david.sherwood@arm.com>
9737 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9738 (record_promoted_value): Likewise.
9739 * expr.c (expand_expr_real_2): Likewise.
9740 * ree.c (update_reg_equal_equiv_notes): Likewise.
9741 (combine_set_extension): Likewise.
9742 * rtlanal.c (low_bitmask_len): Likewise.
9743 * simplify-rtx.c (neg_const_int): Likewise.
9744 (simplify_binary_operation_1): Likewise.
9746 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9747 Alan Hayward <alan.hayward@arm.com>
9748 David Sherwood <david.sherwood@arm.com>
9750 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9751 * regcprop.c (maybe_mode_change): Likewise.
9752 * reload1.c (alter_reg): Likewise.
9754 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9756 * inchash.h (inchash::hash::add_wide_int): New function.
9757 * lto-streamer-out.c (hash_tree): Use it.
9759 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9761 * inchash.h (inchash::hash::add_wide_int): Rename to...
9762 (inchash::hash::add_hwi): ...this.
9763 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9764 (polymorphic_call_target_hasher::hash): Likewise.
9765 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9766 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9767 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9768 * lto-streamer-out.c (hash_tree): Likewise.
9769 * optc-save-gen.awk: Likewise.
9770 * tree.c (add_expr): Likewise.
9772 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9775 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9776 for ordered inequality comparisons even with TARGET_IEEE_FP.
9778 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9781 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9782 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9783 Expand with cmp<dwi>_doubleword.
9785 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9787 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9788 List CET intrinsics.
9789 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9790 specific to -fcf-protection option.
9792 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9794 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9795 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9796 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9797 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9798 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9799 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9800 (extra_objs): Add cet.o for Linux/x86 targets.
9801 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9802 * config/i386/cet.c: New file.
9803 * config/i386/cetintrin.h: Likewise.
9804 * config/i386/t-cet: Likewise.
9805 * config/i386/cpuid.h (bit_SHSTK): New.
9806 (bit_IBT): Likewise.
9807 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9808 pass IBT and SHSTK bits.
9809 * config/i386/i386-builtin-types.def
9810 (VOID_FTYPE_UNSIGNED_PVOID): New.
9811 (VOID_FTYPE_UINT64_PVOID): Likewise.
9812 * config/i386/i386-builtin.def: Add CET intrinsics.
9813 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9814 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9815 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9816 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9818 * config/i386/i386.c (rest_of_insert_endbranch): New.
9819 (pass_data_insert_endbranch): Likewise.
9820 (pass_insert_endbranch): Likewise.
9821 (make_pass_insert_endbranch): Likewise.
9822 (ix86_notrack_prefixed_insn_p): Likewise.
9823 (ix86_target_string): Add -mibt, -mshstk flags.
9824 (ix86_option_override_internal): Add flag_cf_protection
9826 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9827 (ix86_print_operand): Add 'notrack' prefix output.
9828 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9829 (ix86_expand_builtin): Expand CET intrinsics.
9830 (x86_output_mi_thunk): Add 'endbranch' instruction.
9831 * config/i386/i386.h (TARGET_IBT): New.
9832 (TARGET_IBT_P): Likewise.
9833 (TARGET_SHSTK): Likewise.
9834 (TARGET_SHSTK_P): Likewise.
9835 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9836 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9837 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9838 (builtin_setjmp_setup): New pattern.
9839 (builtin_longjmp): Likewise.
9840 (rdssp<mode>): Likewise.
9841 (incssp<mode>): Likewise.
9842 (saveprevssp): Likewise.
9843 (rstorssp): Likewise.
9844 (wrss<mode>): Likewise.
9845 (wruss<mode>): Likewise.
9846 (setssbsy): Likewise.
9847 (clrssbsy): Likewise.
9848 (nop_endbr): Likewise.
9849 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9851 * config/i386/immintrin.h: Include <cetintrin.h>.
9852 * config/i386/linux-common.h
9853 (file_end_indicate_exec_stack_and_cet): New prototype.
9854 (TARGET_ASM_FILE_END): New.
9856 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9858 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
9859 latencies instead of having separate table; make difference between
9860 integer and float costs.
9861 * i386.h (processor_costs): Remove scalar_stmt_cost,
9862 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
9863 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
9865 * x86-tune-costs.h: Remove entries which has been removed in
9866 procesor_costs from all tables; make cond_taken_branch_cost
9867 and cond_not_taken_branch_cost COST_N_INSNS based.
9869 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9871 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
9873 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9875 * config/i386/i386.md (isa): Remove fma_avx512f.
9876 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9877 <avx512>_fmadd_<mode>_mask3<round_name>,
9878 <avx512>_fmsub_<mode>_mask<round_name>,
9879 <avx512>_fmsub_<mode>_mask3<round_name>,
9880 <avx512>_fnmadd_<mode>_mask<round_name>,
9881 <avx512>_fnmadd_<mode>_mask3<round_name>,
9882 <avx512>_fnmsub_<mode>_mask<round_name>,
9883 <avx512>_fnmsub_<mode>_mask3<round_name>,
9884 <avx512>_fmaddsub_<mode>_mask<round_name>,
9885 <avx512>_fmaddsub_<mode>_mask3<round_name>,
9886 <avx512>_fmsubadd_<mode>_mask<round_name>,
9887 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
9888 (*vec_widen_umult_even_v16si<mask_name>,
9889 *vec_widen_smult_even_v16si<mask_name>): Likewise.
9890 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
9892 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9894 * extend.texi: Add 'nocf_check' documentation.
9895 * gimple.texi: Add second parameter to
9896 gimple_build_call_from_tree.
9897 * invoke.texi: Add -fcf-protection documentation.
9898 * rtl.texi: Add REG_CALL_NOTRACK documenation.
9900 2017-10-20 Richard Biener <rguenther@suse.de>
9902 PR tree-optimization/82473
9903 * tree-vect-loop.c (vectorizable_reduction): Properly get at
9904 the largest input type.
9906 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9908 * c-attribs.c (handle_nocf_check_attribute): New function.
9909 (c_common_attribute_table): Add 'nocf_check' handling.
9910 * gimple-parser.c: Add second argument NULL to
9911 gimple_build_call_from_tree.
9912 * attrib.c (comp_type_attributes): Check nocf_check attribute.
9913 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
9915 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
9916 * common.opt: Add fcf-protection flag.
9917 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
9918 * flag-types.h: Add enum cf_protection_level.
9919 * gimple.c (gimple_build_call_from_tree): Add second parameter.
9920 Add 'nocf_check' attribute propagation to gimple call.
9921 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
9922 (gimple_build_call_from_tree): Update prototype.
9923 (gimple_call_nocf_check_p): New function.
9924 (gimple_call_set_nocf_check): Likewise.
9925 * gimplify.c: Add second argument to gimple_build_call_from_tree.
9926 * ipa-icf.c: Add nocf_check attribute in statement hash.
9927 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
9928 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
9929 * toplev.c (process_options): Add flag_cf_protection handling.
9931 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9933 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
9935 2017-10-20 Richard Biener <rguenther@suse.de>
9937 PR tree-optimization/82603
9938 * tree-if-conv.c (predicate_mem_writes): Make sure to only
9939 remove false predicated stores.
9941 2017-10-20 Richard Biener <rguenther@suse.de>
9943 * graphite-isl-ast-to-gimple.c
9944 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
9945 Remove return value and simplify, dump copied stmt after lhs
9947 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
9948 Reduce dump verbosity.
9949 (gsi_insert_earliest): Likewise.
9950 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
9951 * graphite.c (print_global_statistics): Adjust dumping.
9952 (print_graphite_scop_statistics): Likewise.
9953 (print_graphite_statistics): Do not dump loops here.
9954 (graphite_transform_loops): But here.
9956 2017-10-20 Nicolas Roche <roche@adacore.com>
9958 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
9959 * configure: Regenerate.
9961 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9964 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
9965 functions when optimizing replace GIMPLE_RETURN stmts with
9966 calls to __builtin_unreachable ().
9969 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
9970 for -fsanitize=thread link of executables.
9971 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
9972 link of executables.
9975 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
9977 (<shift_insn><mode>3<mask_name>): Change the last of the 3
9978 define_insns for logical vector shifts to use VI248_AVX512BW
9979 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
9980 condition, useless isa and prefix attributes. Change the first
9981 2 of these define_insns to ...
9982 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
9983 define_insn for avx512vl.
9984 (<shift_insn><mode>3): ... and this, new define_insn without
9985 masking for non-avx512vl.
9988 * config/i386/sse.md (*andnot<mode>3,
9989 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
9990 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
9991 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
9992 not applied use empty suffix even for TARGET_AVX512VL.
9993 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
9994 is applied, supply evex,evex or evex,evex,evex instead of just
9997 2017-10-20 Julia Koval <julia.koval@intel.com>
9999 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10000 (OPTION_MASK_ISA_GFNI_UNSET): New.
10001 (ix86_handle_option): Handle OPT_mgfni.
10002 * config/i386/cpuid.h (bit_GFNI): New.
10003 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10004 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10005 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10006 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10007 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10008 * config/i386/i386.opt: Add mgfni.
10010 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10012 * config/msp430/msp430.c (msp430_option_override): Disable
10013 -fdelete-null-pointer-checks.
10014 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10016 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10018 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10019 of x87 and SSE instructions.
10021 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10023 * asan.c (create_cond_insert_point): Do not update edge count.
10024 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10025 (afdo_propagate_circuit): Likewise.
10026 (afdo_calculate_branch_prob): Likewise.
10027 (afdo_annotate_cfg): Likewise.
10028 * basic-block.h (struct edge_def): Remove count.
10029 (edge_def::count): New accessor.
10030 * bb-reorder.c (rotate_loop): Update.
10031 (find_traces_1_round): Update.
10032 (connect_traces): Update.
10033 (sanitize_hot_paths): Update.
10034 * cfg.c (unchecked_make_edge): Update.
10035 (make_single_succ_edge): Update.
10036 (check_bb_profile): Update.
10037 (dump_edge_info): Update.
10038 (update_bb_profile_for_threading): Update.
10039 (scale_bbs_frequencies_int): Update.
10040 (scale_bbs_frequencies_gcov_type): Update.
10041 (scale_bbs_frequencies_profile_count): Update.
10042 (scale_bbs_frequencies): Update.
10043 * cfganal.c (connect_infinite_loops_to_exit): Update.
10044 * cfgbuild.c (compute_outgoing_frequencies): Update.
10045 (find_many_sub_basic_blocks): Update.
10046 * cfgcleanup.c (try_forward_edges): Update.
10047 (try_crossjump_to_edge): Update
10048 * cfgexpand.c (expand_gimple_cond): Update
10049 (expand_gimple_tailcall): Update
10050 (construct_exit_block): Update
10051 * cfghooks.c (verify_flow_info): Update
10052 (redirect_edge_succ_nodup): Update
10053 (split_edge): Update
10054 (make_forwarder_block): Update
10055 (duplicate_block): Update
10056 (account_profile_record): Update
10057 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10058 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10059 * cfgloopmanip.c (scale_loop_profile): Update.
10061 (lv_adjust_loop_entry_edge): Update.
10062 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10063 (force_nonfallthru_and_redirect): Update.
10064 (purge_dead_edges): Update.
10065 (rtl_flow_call_edges_add): Update.
10066 * cgraphunit.c (init_lowered_empty_function): Update.
10067 (cgraph_node::expand_thunk): Update.
10068 * gimple-pretty-print.c (dump_probability): Update.
10069 (dump_edge_probability): Update.
10070 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10071 * haifa-sched.c (sched_create_recovery_edges): Update.
10072 * hsa-gen.c (convert_switch_statements): Update.
10073 * ifcvt.c (dead_or_predicable): Update.
10074 * ipa-inline-transform.c (inline_transform): Update.
10075 * ipa-split.c (split_function): Update.
10076 * ipa-utils.c (ipa_merge_profiles): Update.
10077 * loop-doloop.c (add_test): Update.
10078 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10079 * lto-streamer-in.c (input_cfg): Update.
10080 (input_function): Update.
10081 * lto-streamer-out.c (output_cfg): Update.
10082 * modulo-sched.c (sms_schedule): Update.
10083 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10084 * predict.c (maybe_hot_edge_p): Update.
10085 (unlikely_executed_edge_p): Update.
10086 (probably_never_executed_edge_p): Update.
10087 (dump_prediction): Update.
10088 (drop_profile): Update.
10089 (propagate_unlikely_bbs_forward): Update.
10090 (determine_unlikely_bbs): Update.
10091 (force_edge_cold): Update.
10092 * profile.c (compute_branch_probabilities): Update.
10093 * reg-stack.c (better_edge): Update.
10094 * shrink-wrap.c (handle_simple_exit): Update.
10095 * tracer.c (better_p): Update.
10096 * trans-mem.c (expand_transaction): Update.
10097 (split_bb_make_tm_edge): Update.
10098 * tree-call-cdce.c: Update.
10099 * tree-cfg.c (gimple_find_sub_bbs): Update.
10100 (gimple_split_edge): Update.
10101 (gimple_duplicate_sese_region): Update.
10102 (gimple_duplicate_sese_tail): Update.
10103 (gimple_flow_call_edges_add): Update.
10104 (insert_cond_bb): Update.
10105 (execute_fixup_cfg): Update.
10106 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10107 * tree-complex.c (expand_complex_div_wide): Update.
10108 * tree-eh.c (lower_resx): Update.
10109 (unsplit_eh): Update.
10110 (cleanup_empty_eh_move_lp): Update.
10111 * tree-inline.c (copy_edges_for_bb): Update.
10112 (freqs_to_counts): Update.
10113 (copy_cfg_body): Update.
10114 * tree-ssa-dce.c (remove_dead_stmt): Update.
10115 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10116 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10117 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10118 (unloop_loops): Update.
10119 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10120 * tree-ssa-loop-split.c (connect_loops): Update.
10121 (split_loop): Update.
10122 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10123 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10124 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10125 * tree-ssa-reassoc.c (branch_fixup): Update.
10126 * tree-ssa-tail-merge.c (replace_block_by): Update.
10127 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10128 (compute_path_counts): Update.
10129 (update_profile): Update.
10130 (recompute_probabilities): Update.
10131 (update_joiner_offpath_counts): Update.
10132 (estimated_freqs_path): Update.
10133 (freqs_to_counts_path): Update.
10134 (clear_counts_path): Update.
10135 (ssa_fix_duplicate_block_edges): Update.
10136 (duplicate_thread_path): Update.
10137 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10138 (case_bit_test_cmp): Update.
10139 (collect_switch_conv_info): Update.
10140 (gen_inbound_check): Update.
10141 (do_jump_if_equal): Update.
10142 (emit_cmp_and_jump_insns): Update.
10143 * tree-tailcall.c (decrease_profile): Update.
10144 (eliminate_tail_call): Update.
10145 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10146 (vect_do_peeling): Update.
10147 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10148 * ubsan.c (ubsan_expand_null_ifn): Update.
10149 (ubsan_expand_ptr_ifn): Update.
10150 * value-prof.c (gimple_divmod_fixed_value): Update.
10151 (gimple_mod_pow2): Update.
10152 (gimple_mod_subtract): Update.
10153 (gimple_ic): Update.
10154 (gimple_stringop_fixed_value): Update.
10156 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10159 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10161 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10163 PR rtl-optimization/82395
10164 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10165 based on non_spilled_static_chain_regno_p.
10167 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10169 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10170 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10171 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10173 2017-10-19 Martin Sebor <msebor@redhat.com>
10175 PR tree-optimization/82596
10176 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10178 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10180 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10181 (get_mem_refs_of_builtin_call): Likewise.
10182 * builtins.c (expand_builtin_apply): Adjust call to
10183 allocate_dynamic_stack_space.
10184 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10185 the third argument to allocate_dynamic_stack_space, otherwise -1.
10186 (expand_builtin): Deal with all alloca variants.
10187 (is_inexpensive_builtin): Likewise.
10188 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10189 * calls.c (special_function_p): Deal with all alloca variants.
10190 (initialize_argument_information): Adjust call to
10191 allocate_dynamic_stack_space.
10192 (expand_call): Likewise.
10193 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10194 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10195 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10196 use it for the stack usage computation.
10197 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10198 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10199 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10200 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10201 (in_loop_p): Remove first argument and useless check.
10202 (pass_walloca::execute): Remove useless test and adjust call to above.
10203 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10204 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10205 (gimplify_call_expr): Deal with all alloca variants.
10206 * hsa-gen.c (gen_hsa_alloca): Likewise.
10207 (gen_hsa_insns_for_call): Likewise.
10208 * ipa-pure-const.c (special_builtin_state): Likewise.
10209 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10210 * tree-object-size.c (alloc_object_size): Likewise.
10211 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10212 (call_may_clobber_ref_p_1): Likewise.
10213 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10214 (ccp_fold_stmt): Likewise.
10215 (optimize_stack_restore): Likewise.
10216 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10217 (mark_all_reaching_defs_necessary_1): Likewise.
10218 (propagate_necessity): Likewise.
10219 (eliminate_unnecessary_stmts): Likewise.
10220 * tree.c (build_common_builtin_nodes): Build
10221 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10222 (build_alloca_call_expr): New function.
10223 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10224 (CASE_BUILT_IN_ALLOCA): Likewise.
10225 (build_alloca_call_expr): Declare.
10226 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10228 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10231 * dwarf2out.c (new_die_raw): New static inline function.
10232 (new_die): Use it to create the DIE.
10233 (add_AT_external_die_ref): Likewise.
10234 (clone_die): Likewise.
10235 (clone_as_declaration): Likewise.
10236 (dwarf2out_vms_debug_main_pointer): Likewise.
10237 (base_type_die): Likewise. Remove early return for corner cases.
10238 Do not call add_pubtype on the DIE here.
10239 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10240 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10241 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10242 native order exists for base types, attach the DIE manually and call
10243 add_pubtype on it. Do not equate a reverse order DIE to the type.
10245 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10247 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10248 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10250 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10252 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10254 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10255 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10256 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10257 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10258 * opts.c (sanitizer_opts): Add builtin.
10259 * ubsan.c (instrument_builtin): New function.
10260 (pass_ubsan::execute): Call it.
10261 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10262 * doc/invoke.texi: Document -fsanitize=builtin.
10264 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10265 builtins, store max (log2 (align), 0) into uchar field instead of
10266 align into uptr field.
10267 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10268 store uchar 0 field instead of uptr 0 field.
10269 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10270 instead of passing one address of struct with 2 locations pass
10271 two addresses of structs with 1 location each.
10272 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10273 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10274 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10275 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10276 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10277 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10278 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10279 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10281 2017-10-19 Martin Liska <mliska@suse.cz>
10284 * file-find.c (remove_prefix): Remove.
10285 * file-find.h (remove_prefix): Likewise.
10286 * gcc-ar.c: Remove smartness of lookup.
10288 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10290 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10291 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10292 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10294 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10297 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10298 (setcc + and to xor + setcc): New peephole2.
10300 2017-10-19 Tom de Vries <tom@codesourcery.com>
10302 * doc/sourcebuild.texi (Test Directives, Variants of
10303 dg-require-support): Add dg-require-stack-size.
10305 2017-10-19 Martin Liska <mliska@suse.cz>
10308 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10309 that have a large alignment.
10310 (gimplify_target_expr): Likewise.
10312 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10314 PR rtl-optimization/82602
10315 * ira.c (rtx_moveable_p): Return false for volatile asm.
10317 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10320 * config/i386/i386-modes.def (CCGZ): New CC mode.
10321 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10322 * config/i386/predicates.md (ix86_comparison_operator):
10324 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10325 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10326 with double-word subtraction.
10327 (put_condition_code): Handle CCGZmode.
10329 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10331 * wide-int.cc (debug (const wide_int &)): New.
10332 (debug (const wide_int *)): New.
10333 (debug (const widest_int &)): New.
10334 (debug (const widest_int *)): New.
10336 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10338 PR middle-end/82556
10339 * lra-constraints.c (curr_insn_transform): Use non-input operand
10340 instead of output one for matched reload.
10342 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10344 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10345 (tree-ssa-loop-ivopts.h): New header file.
10346 (struct builtin_info): New fields.
10347 (classify_builtin_1): Compute and record base and offset parts for
10348 memset builtin partition by calling strip_offset.
10349 (offset_cmp, fuse_memset_builtins): New functions.
10350 (finalize_partitions): Fuse adjacent memset partitions by calling
10352 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10353 Expose the interface.
10354 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10356 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10358 PR tree-optimization/82574
10359 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10360 that data reference must be executed exactly once per iteration
10361 against the outermost loop in nest.
10362 (classify_partition): Update call to above function.
10364 2017-10-18 Richard Biener <rguenther@suse.de>
10366 PR tree-optimization/82591
10367 * graphite.c (graphite_transform_loops): Move code gen message
10369 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10370 Here. Handle scop_to_isl_ast failing.
10371 (scop_to_isl_ast): Limit the number of ISL operations.
10373 2017-10-18 Richard Biener <rguenther@suse.de>
10375 * graphite-isl-ast-to-gimple.c
10376 (translate_isl_ast_to_gimple::set_rename): Simplify.
10377 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10378 (graphite_copy_stmts_from_block): ... here.
10379 (copy_bb_and_scalar_dependences): Simplify.
10380 (add_parameters_to_ivs_params): Canonicalize.
10381 (generate_entry_out_of_ssa_copies): Simplify.
10382 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10383 by passing in ISL dimension.
10384 (parameter_index_in_region_1): Rename to ...
10385 (parameter_index_in_region): ... this.
10386 (extract_affine): Adjust assert, pass down parameter index.
10387 (add_param_constraints): Use range-info when available.
10388 (build_scop_context): Adjust.
10389 * sese.c (new_sese_info): Adjust.
10390 (free_sese_info): Likewise.
10391 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10392 Remove unused typedefs.
10393 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10395 2017-10-18 Martin Liska <mliska@suse.cz>
10397 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10399 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10401 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10402 (s390_sched_init): Do not reset s390_sched_state if we entered the
10403 current basic block via a fallthru edge and all others are unlikely.
10405 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10407 * config/s390/s390.c (NUM_SIDES): New variable.
10408 (LONGRUNNING_THRESHOLD): New variable.
10409 (LATENCY_FACTOR): New variable.
10410 (s390_sched_score): Decrease score for long-running instructions on
10412 (s390_sched_variable_issue): Perform bookkeeping for long-running
10415 2017-10-18 Richard Biener <rguenther@suse.de>
10417 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10418 Simplify with removal of the parameter rename map.
10419 (set_rename): Likewise.
10420 (should_copy_to_new_region): Likewise.
10421 (graphite_copy_stmts_from_block): Likewise.
10422 (copy_bb_and_scalar_dependences): Remove initialization of
10423 unused copied_bb_map.
10424 (copy_def): Remove.
10425 (copy_internal_parameters): Likewise.
10426 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10427 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10428 Use INTEGRAL_TYPE_P.
10429 (parameter_index_in_region_1): Rename to ...
10430 (assign_parameter_index_in_region): ... this. Assert we have
10431 a parameter we handle.
10432 (scan_tree_for_params): Adjust.
10433 * sese.h (parameter_rename_map_t): Remove.
10434 (struct sese_info_t): Remove unused parameter_rename_map and
10435 copied_bb_map members.
10436 * sese.c (new_sese_info): Adjust.
10437 (free_sese_info): Likewise.
10439 2017-10-18 Martin Liska <mliska@suse.cz>
10442 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10443 on an abnormal edge.
10445 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10447 * doc/invoke.texi (ffunction-sections and fdata-sections):
10450 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10452 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10453 the use statement can throw internally.
10455 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10457 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10458 any RTX present on the RHS of a SET.
10459 * compare-elim.c (try_eliminate_compare): Restore comment.
10461 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10463 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10464 may be also called on tcc_type nodes.
10465 * langhooks.c (lhd_tree_size): Likewise.
10467 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10469 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10470 format_warning_at_substring.
10471 (maybe_warn): Convert source_range * param to a location_t. Pass
10472 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10473 (format_directive): Remove code to extract source_ranges and
10474 source_range * in favor of just a location_t.
10475 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10477 * substring-locations.c (format_warning_va): Convert
10478 source_range * param to a location_t.
10479 (format_warning_at_substring): Likewise.
10480 * substring-locations.h (format_warning_va): Likewise.
10481 (format_warning_at_substring): Likewise.
10483 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10485 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10487 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10488 and scatter/gather ops.
10490 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10491 vec_gather_load and vec_scatter_store.
10492 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10493 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10494 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10495 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10496 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10497 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10499 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10501 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10502 Detect FTST instruction and handle its register pops. Only pop
10503 second operand if can_pop_second_op is true.
10504 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10505 set can_pop_second_op to false in the compare_for_stack_reg call.
10507 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10508 output_fp_compare for stack register operands.
10509 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10510 instructions here. Do not emit stack register pops here. Assert
10511 that FCOMPP pops next to top stack register. Rewrite function.
10513 2017-10-17 Nathan Sidwell <nathan@acm.org>
10515 PR middle-end/82577
10516 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10517 use DECL_ASSEMBLER_NAME_RAW.
10519 PR middle-end/82546
10520 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10523 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10524 Wilco Dijkstra <wilco.dijkstra@arm.com>
10526 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10527 converstion to Pmode from the buf_addr.
10529 2017-10-17 Richard Biener <rguenther@suse.de>
10531 * graphite-dependences.c (scop_get_reads_and_writes): Change
10532 output parameters to references.
10534 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10536 PR 71026/tree-optimization
10537 * fold-const.c (distribute_real_division): Removed.
10538 (fold_binary_loc): Remove calls to distribute_real_divison.
10540 2017-10-17 Richard Biener <rguenther@suse.de>
10542 * graphite-scop-detection.c
10543 (scop_detection::stmt_has_simple_data_refs_p): Always use
10544 the full nest as region.
10545 (try_generate_gimple_bb): Likewise.
10546 * sese.c (scalar_evolution_in_region): Simplify now that
10547 SCEV can handle instantiation in regions.
10548 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10549 in the non-loop part of a function if requested.
10551 2017-10-17 Richard Biener <rguenther@suse.de>
10553 PR tree-optimization/82563
10554 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10556 (graphite_regenerate_ast_isl): Call it.
10557 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10559 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10561 PR tree-optimization/82549
10562 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10563 Formatting fixes. Instead of calling make_bit_field_ref with negative
10566 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10568 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10569 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10570 _mm_maskz_reduce_ss): New.
10571 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10572 __builtin_ia32_reducess_mask): Ditto..
10573 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10574 * config/i386/sse.md (reduces<mode>): Renamed to ...
10575 (reduces<mode><mask_scalar_name>): ... this.
10576 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10578 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10579 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10581 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10583 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10584 * selftest-run-tests.c (selftest::run_tests): Call
10585 selftest::unique_ptr_tests_cc_tests.
10586 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10587 * unique-ptr-tests.cc: New file.
10589 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10592 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10594 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10595 (make_hard_regno_born, make_hard_regno_dead): Update
10596 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10598 2017-10-16 Jeff Law <law@redhat.com>
10600 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10602 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10604 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10606 2017-10-16 Olivier Hainque <hainque@adacore.com>
10608 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10609 with_cpu if we were configured for an e500v2 target cpu name.
10611 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10613 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10614 * doc/invoke.texi: Document +nodsp as a valid extension for
10617 2017-10-16 Martin Liska <mliska@suse.cz>
10619 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10620 (test_set_range): Likewise.
10621 (test_range_functions): Rename to ...
10622 (test_bit_in_range): ... this.
10623 (sbitmap_c_tests): Add new test.
10625 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10627 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10629 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10630 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10632 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10634 * config/aarch64/aarch64-builtins.c
10635 (aarch64_types_quadopu_lane_qualifiers): New.
10636 (TYPES_QUADOPU_LANE): New.
10637 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10638 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10639 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10640 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10641 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10642 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10643 (Vdottype, DOTPROD): New.
10644 (sur): Add SDOT and UDOT.
10646 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10648 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10649 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10650 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10651 Add TARGET_DOTPROD.
10652 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10653 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10654 Enable TARGET_DOTPROD.
10655 (cortex-a75.cortex-a55): Likewise.
10656 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10658 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10660 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10661 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10663 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10665 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10666 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10667 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10668 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10669 * config/arm/types.md (neon_dot, neon_dot_q): New.
10670 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10672 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10674 * config/arm/arm.h (TARGET_DOTPROD): New.
10675 * config/arm/arm.c (arm_arch_dotprod): New.
10676 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10677 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10678 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10679 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10680 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10681 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10682 * doc/invoke.texi (armv8.2-a): Document dotprod
10684 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10686 * i386.c (ix86_vec_cost): New function.
10687 (ix86_rtx_costs): Handle vector operations better.
10688 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10689 * x86-tune-costs.h: Add new costs to all tables.
10691 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10693 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10695 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10696 divsd, sqrtss and sqrtsd
10697 * x86-tune-costs.h: Add new entries to all costs.
10698 (znver1_cost): Fix to match real instruction latencies.
10700 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10701 Michael Collison <michael.collison@arm.com>
10703 * compare-elim.c: Include emit-rtl.h.
10704 (can_merge_compare_into_arith): New function.
10705 (try_validate_parallel): Likewise.
10706 (try_merge_compare): Likewise.
10707 (try_eliminate_compare): Call the above when no previous clobber
10709 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10712 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10714 PR middle-end/62263
10715 PR middle-end/82498
10716 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10717 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10719 PR middle-end/62263
10720 PR middle-end/82498
10721 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10722 to be any operand_equal_p operands. For & (B - 1) require
10723 B to be power of 2. Recognize
10724 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10726 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10729 * optabs.c (expand_memory_blockage): Fix call of
10730 targetm.have_memory_blockage.
10732 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10735 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10736 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10737 objects to extra_objs instead of overwriting it.
10739 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10741 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10742 Use any_fp_register_operand as operand[3] predicate. Simplify
10743 equality test for operands[2] and operands[4] memory location.
10744 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10745 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10746 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10747 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10748 any_fp_register_operand as operand[1] predicate. Simplify
10749 equality test for operands[0] and operands[3] memory location.
10750 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10751 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10752 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10754 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10756 * target-insns.def: Add memory_blockage.
10757 * optabs.c (expand_memory_blockage): New function.
10758 (expand_asm_memory_barrier): Rename ...
10759 (expand_asm_memory_blockage): ... to this.
10760 (expand_mem_thread_fence): Call expand_memory_blockage
10761 instead of expand_asm_memory_barrier.
10762 (expand_mem_singnal_fence): Ditto.
10763 (expand_atomic_load): Ditto.
10764 (expand_atomic_store): Ditto.
10765 * doc/md.texi (Standard Pattern Names For Generation):
10766 Document memory_blockage instruction pattern.
10768 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10770 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10771 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10772 * config/rl78/rl78.md: New define_expand "adddi3".
10774 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10776 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10779 2017-10-13 Jeff Law <law@redhat.com>
10781 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10783 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10786 * internal-fn.c (expand_mul_overflow): If both operands have
10787 the same highpart of -1 or 0 and the topmost bit of lowpart
10788 is different, overflow is if res <= 0 rather than res < 0.
10790 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10792 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10793 TARGET_P9_VECTOR code for unaligned_load case.
10795 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10797 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10799 2017-10-13 Nathan Sidwell <nathan@acm.org>
10801 * tree-core.h (tree_contains_struct): Make bool.
10802 * tree.c (tree_contains_struct): Likewise.
10803 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10804 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10805 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10806 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10807 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10809 2017-10-13 Richard Biener <rguenther@suse.de>
10811 * graphite-isl-ast-to-gimple.c
10812 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10813 parameters and dominance check.
10814 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10815 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10816 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10817 Do not update SSA form here or do intermediate IL verification.
10818 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10819 (graphite_initialize): Remove check on the number of loops in
10820 the function and inline into graphite_transform_loops.
10821 (graphite_finalize): Inline into graphite_transform_loops.
10822 (graphite_transform_loops): Perform SSA update and IL verification
10824 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10826 2017-10-13 Richard Biener <rguenther@suse.de>
10828 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10829 graphite_expression_type_precision): Avoid global constructor
10831 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10832 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10833 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10834 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10835 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10836 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10838 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10841 * config/i386/i386.h (ix86_red_zone_size): New.
10842 * config/i386/i386.md (push peephole2s): Replace
10843 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10845 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10846 Alan Hayward <alan.hayward@arm.com>
10847 David Sherwood <david.sherwood@arm.com>
10849 * combine.c (can_change_dest_mode): Reject changes in
10850 REGMODE_NATURAL_SIZE.
10852 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10853 Alan Hayward <alan.hayward@arm.com>
10854 David Sherwood <david.sherwood@arm.com>
10856 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10857 (expand_debug_source_expr): Likewise.
10858 * combine.c (combine_simplify_rtx): Likewise.
10859 * cse.c (fold_rtx): Likewise.
10860 * fwprop.c (canonicalize_address): Likewise.
10861 * targhooks.c (default_shift_truncation_mask): Likewise.
10863 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10864 Alan Hayward <alan.hayward@arm.com>
10865 David Sherwood <david.sherwood@arm.com>
10867 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
10868 (widened_mode): Likewise.
10869 (expand_unop): Likewise.
10870 * ree.c (transform_ifelse): Likewise.
10871 (merge_def_and_ext): Likewise.
10872 (combine_reaching_defs): Likewise.
10873 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10875 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10876 Alan Hayward <alan.hayward@arm.com>
10877 David Sherwood <david.sherwood@arm.com>
10879 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
10880 * combine.c (gen_lowpart_for_combine): Likewise.
10881 * dwarf2out.c (rtl_for_decl_location): Likewise.
10882 * final.c (alter_subreg): Likewise.
10883 * rtlhooks.c (gen_lowpart_general): Likewise.
10884 (gen_lowpart_if_possible): Likewise.
10886 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10887 Alan Hayward <alan.hayward@arm.com>
10888 David Sherwood <david.sherwood@arm.com>
10890 * calls.c (expand_call): Use subreg_lowpart_offset.
10891 * cse.c (cse_insn): Likewise.
10892 * regcprop.c (copy_value): Likewise.
10893 (copyprop_hardreg_forward_1): Likewise.
10895 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10898 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
10899 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
10900 =Q constraints to +Q and into insn condition add check
10901 that operands[0] and operands[1] are equal.
10902 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
10903 =Q constraints to +Q and into insn condition add check
10904 that operands[0] is equal to either operands[1] or operands[2].
10907 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
10908 instead of handling MINUS_EXPR twice (once for each argument),
10909 canonicalize operand order and handle just once, use rtype where
10910 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
10913 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
10914 any values of __C while still being pattern recognizable as a simple
10915 rotate instruction.
10917 2017-10-13 Richard Biener <rguenther@suse.de>
10919 PR tree-optimization/82451
10921 2017-10-02 Richard Biener <rguenther@suse.de>
10923 PR tree-optimization/82355
10924 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10925 a mapping for the enclosing loop but avoid generating one for
10926 the loop tree root.
10927 (copy_bb_and_scalar_dependences): Remove premature codegen
10928 error on PHIs in blocks duplicated into multiple places.
10929 * graphite-scop-detection.c
10930 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10931 in the region use it as loop and nest to analyze the DR in.
10932 (try_generate_gimple_bb): Likewise.
10933 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10934 (add_loop_constraints): For blocks in a loop not in the region
10935 create a dimension with a single iteration.
10936 * sese.h (gbb_loop_at_index): Remove assert.
10938 * cfgloop.c (loop_preheader_edge): For the loop tree root
10939 return the single successor of the entry block.
10940 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10941 Reset the SCEV hashtable and niters.
10942 * graphite-scop-detection.c
10943 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
10944 assert that we only have POLYNOMIAL_CHREC that vary in loops
10945 contained in the region.
10946 (scop_detection::graphite_can_represent_expr): Adjust.
10947 (scop_detection::stmt_has_simple_data_refs_p): For loops
10948 not in the region set loop to NULL. The nest is now the
10949 entry edge to the region.
10950 (try_generate_gimple_bb): Likewise.
10951 * sese.c (scalar_evolution_in_region): Adjust for
10952 instantiate_scev change.
10953 * tree-data-ref.h (graphite_find_data_references_in_stmt):
10954 Make nest parameter the edge into the region.
10955 (create_data_ref): Likewise.
10956 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
10957 entry edge into a region and adjust instantiate_scev calls.
10958 (create_data_ref): Likewise.
10959 (graphite_find_data_references_in_stmt): Likewise.
10960 (find_data_references_in_stmt): Pass the loop preheader edge
10961 from the nest argument.
10962 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
10963 parameter the edge into the region.
10964 (instantiate_parameters): Use the loop preheader edge as entry.
10965 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
10967 (get_instantiated_value_entry): Make instantiate_below parameter
10968 the edge into the region.
10969 (instantiate_scev_name): Likewise. Adjust dominance checks,
10970 when we cannot use loop-based instantiation instantiate by
10971 walking use-def chains.
10972 (instantiate_scev_poly): Adjust.
10973 (instantiate_scev_binary): Likewise.
10974 (instantiate_scev_convert): Likewise.
10975 (instantiate_scev_not): Likewise.
10976 (instantiate_array_ref): Remove.
10977 (instantiate_scev_3): Likewise.
10978 (instantiate_scev_2): Likewise.
10979 (instantiate_scev_1): Likewise.
10980 (instantiate_scev_r): Do not blindly handle N-operand trees.
10981 Do not instantiate array-refs. Handle all constants and invariants.
10982 (instantiate_scev): Make instantiate_below parameter
10983 the edge into the region.
10984 (resolve_mixers): Use the loop preheader edge for the region
10985 parameter to instantiate_scev_r.
10986 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
10988 2017-10-13 Richard Biener <rguenther@suse.de>
10990 PR tree-optimization/82525
10991 * graphite-isl-ast-to-gimple.c
10992 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
10994 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
10995 Fail code generation when we cannot represent the isl integer.
10996 (binary_op_to_tree): Elide modulo operations that are no-ops
10997 in the type we code generate. Remove now superfluous code
11000 2017-10-13 Richard Biener <rguenther@suse.de>
11002 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11003 (scop_detection::harmful_loop_in_region): Remove premature
11004 IV type restriction.
11005 (scop_detection::graphite_can_represent_scev): We can handle
11006 pointer IVs just fine.
11008 2017-10-13 Alan Modra <amodra@gmail.com>
11010 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11011 "Clobbers and Scratch Registers". Add paragraph on
11012 alternative to clobbers for scratch registers and OpenBLAS
11015 2017-10-13 Alan Modra <amodra@gmail.com>
11017 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11018 example of a memory input for a string of known length. Move
11019 commentary out of table. Add a number of new examples
11020 covering array memory inputs.
11022 2017-10-12 Martin Liska <mliska@suse.cz>
11024 PR tree-optimization/82493
11025 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11026 (test_range_functions): New function.
11027 (sbitmap_c_tests): Likewise.
11028 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11029 * selftest.h (sbitmap_c_tests): New function.
11031 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11033 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11035 * config/rs6000/amo.h: Fix spacing issue.
11037 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11040 * config/i386/i386.md (*ashl<mode>3_mask_1,
11041 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11042 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11045 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11047 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11048 (profile_probability): Set max_probability
11049 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11051 * profile-count.c (profile_probability::differs_from_p): Do not
11052 rely on max_probaiblity == 10000
11054 2017-10-12 Jeff Law <law@redhat.com>
11056 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11059 2017-10-12 Martin Sebor <msebor@redhat.com>
11063 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11064 (handle_alias_pairs): Call it.
11065 * common.opt (-Wattribute-alias): New option.
11066 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11067 * doc/invoke.texi (-Wattribute-alias): Document.
11069 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11072 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11074 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11076 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11077 (make_hard_regno_born, make_hard_regno_dead): Update
11078 bb_killed_pseudos and bb_gen_pseudos.
11080 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11082 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11084 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11086 * config/alpha/alpha.c (alpha_split_conditional_move):
11087 Use std::swap instead of manually swapping.
11088 (alpha_stdarg_optimize_hook): Ditto.
11089 (alpha_canonicalize_comparison): Ditto.
11091 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11093 * tree-loop-distribution.c (struct builtin_info): New struct.
11094 (struct partition): Refactor fields into struct builtin_info.
11095 (partition_free): Free struct builtin_info.
11096 (build_size_arg_loc, build_addr_arg_loc): Delete.
11097 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11098 information from struct builtin_info.
11099 (find_single_drs): New function refactored from classify_partition.
11100 Also moved builtin validity checks to this function.
11101 (compute_access_range, alloc_builtin): New functions.
11102 (classify_builtin_st, classify_builtin_ldst): New functions.
11103 (classify_partition): Refactor code into functions find_single_drs,
11104 classify_builtin_st and classify_builtin_ldst.
11105 (distribute_loop): Don't do runtime alias check when distributing
11107 (find_seed_stmts_for_distribution): New function.
11108 (pass_loop_distribution::execute): Refactor code finding seed
11109 stmts into above function. Support distribution for the innermost
11110 two-level loop nest. Adjust dump information.
11112 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11114 * tree-loop-distribution.c: Adjust the general comment.
11115 (NUM_PARTITION_THRESHOLD): New macro.
11116 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11117 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11118 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11119 in call to build_partition_graph.
11120 (finalize_partitions): New parameter. Make loop distribution more
11121 conservative by fusing more partitions.
11122 (distribute_loop): Don't do runtime alias check in case of loop nest
11124 (find_seed_stmts_for_distribution): New function.
11125 (prepare_perfect_loop_nest): New function.
11126 (pass_loop_distribution::execute): Refactor code finding seed stmts
11127 and loop nest into above functions. Support loop nest distribution.
11128 Adjust dump information accordingly.
11130 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11132 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11133 and set PTYPE_SEQUENTIAL for merged partition.
11135 2017-10-12 Richard Biener <rguenther@suse.de>
11137 PR tree-optimization/69728
11139 2017-09-19 Richard Biener <rguenther@suse.de>
11141 PR tree-optimization/69728
11142 * graphite-sese-to-poly.c (schedule_error): New global.
11143 (add_loop_schedule): Handle empty domain by failing the
11145 (build_original_schedule): Handle schedule_error.
11147 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11148 domain by returning an unchanged schedule.
11150 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11152 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11153 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11155 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11157 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11160 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11163 * expr.c (store_field): Don't optimize away bitsize == 0 store
11164 from CALL_EXPR with addressable return type.
11166 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11168 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11169 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11170 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11171 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11172 TARGET_ISEL instead of TARGET_ISEL<sel>.
11174 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11176 * config/rs6000/rs6000.c
11177 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11179 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11181 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11183 (reg_or_cint_operand): Fix comment.
11184 (reg_or_zero_operand): New predicate.
11185 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11186 * config/rs6000/rs6000.c (output_isel): Delete.
11187 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11188 instead of reg_or_cint_operand. Output instruction directly (not via
11190 (isel_unsigned_<mode>): Ditto.
11191 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11192 gpc_reg_operand. Add an instruction alternative for this. Output
11193 instruction directly.
11194 (*isel_reversed_unsigned_<mode>): Ditto.
11196 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11198 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11199 (TARGET_CANONICALIZE_COMPARISON): Define.
11201 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11204 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11205 Check whether the dest is REG before adding REG_EQUIV note.
11207 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11210 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11212 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11213 (make_hard_regno_born, make_hard_regno_dead): Update
11214 bb_killed_pseudos and bb_gen_pseudos.
11216 2017-10-11 Nathan Sidwell <nathan@acm.org>
11218 * incpath.h (enum incpath_kind): Name enum, prefix values.
11219 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11220 * incpath.c (heads, tails): Use INC_MAX.
11221 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11222 (merge_include_chains, split_quote_chain,
11223 register_include_chains): Update incpath_kind names.
11224 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11225 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11227 (add_framework_path, darwin_register_objc_includes): Likewise.
11228 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11230 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11232 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11233 Do not use float_operator operator predicate.
11234 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11235 * config/i386/predicates.md (float_operator): Remove predicate.
11237 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11239 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11240 (*jccxf_i387): Ditto.
11241 (*jcc<mode>_i387): Ditto.
11242 (*jccu<mode>_i387): Ditto.
11243 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11244 (*jcc_*_i387 splitters): Remove.
11245 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11246 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11247 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11250 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11252 * profile-count.h (slow_safe_scale_64bit): New function.
11253 (safe_scale_64bit): New inline.
11254 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11255 * profile-count.c: Include wide-int.h
11256 (slow_safe_scale_64bit): New.
11258 2017-10-11 Nathan Sidwell <nathan@acm.org>
11260 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11261 HAS_DECL_ASSEMBLER_NAME_P.
11262 * gimple-expr.c (gimple_decl_printable_name: Check
11263 HAS_DECL_ASSEMBLER_NAME_P too.
11264 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11265 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11266 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11267 * passes.c (rest_of_decl_compilation): Check
11268 HAS_DECL_ASSEMBLER_NAME_P too.
11269 * recog.c (verify_changes): Likewise.
11270 * tree-pretty-print.c (dump_decl_name): Likewise.
11271 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11273 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11274 (SET_DECL_ASSEMBLER_NAME): Use it.
11275 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11276 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11277 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11279 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11281 * config.gcc (i386, x86_64): Add extra objects.
11282 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11283 (ix86_min_insn_size): Declare.
11284 (ix86_issue_rate): Declare.
11285 (ix86_adjust_cost): Declare.
11286 (ia32_multipass_dfa_lookahead): Declare.
11287 (ix86_macro_fusion_p): Declare.
11288 (ix86_macro_fusion_pair_p): Declare.
11289 (ix86_bd_has_dispatch): Declare.
11290 (ix86_bd_do_dispatch): Declare.
11291 (ix86_core2i7_init_hooks): Declare.
11292 (ix86_atom_sched_reorder): Declare.
11293 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11294 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11295 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11296 (rip_relative_addr_p): Rename to ...
11297 (ix86_rip_relative_addr_p): ... this one; export.
11298 (memory_address_length): Update.
11299 (ix86_issue_rate): Move to x86-tune-sched.c.
11300 (ix86_flags_dependent): Move to x86-tune-sched.c.
11301 (ix86_agi_dependent): Move to x86-tune-sched.c.
11302 (exact_dependency_1): Move to x86-tune-sched.c.
11303 (exact_store_load_dependency): Move to x86-tune-sched.c.
11304 (ix86_adjust_cost): Move to x86-tune-sched.c.
11305 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11306 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11307 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11308 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11309 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11310 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11311 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11312 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11313 (min_insn_size): Rename to ...
11314 (ix86_min_insn_size): ... this one; export.
11315 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11316 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11317 (core2i7_first_cycle_multipass_backtrack): Move to
11318 x86-tune-sched-core.c.
11319 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11320 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11321 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11322 (ix86_avoid_jump_mispredicts): Update.
11323 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11324 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11325 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11326 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11327 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11328 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11329 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11330 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11331 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11332 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11333 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11334 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11335 (BIG): Move to ix86-tune-sched-bd.c.
11336 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11337 (enum insn_path): Move to ix86-tune-sched-bd.c.
11338 (get_mem_group): Move to ix86-tune-sched-bd.c.
11339 (is_cmp): Move to ix86-tune-sched-bd.c.
11340 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11341 (is_branch): Move to ix86-tune-sched-bd.c.
11342 (is_prefetch): Move to ix86-tune-sched-bd.c.
11343 (init_window): Move to ix86-tune-sched-bd.c.
11344 (allocate_window): Move to ix86-tune-sched-bd.c.
11345 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11346 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11347 (process_end_window): Move to ix86-tune-sched-bd.c.
11348 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11349 (find_constant): Move to ix86-tune-sched-bd.c.
11350 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11351 (has_immediate): Move to ix86-tune-sched-bd.c.
11352 (get_insn_path): Move to ix86-tune-sched-bd.c.
11353 (get_insn_group): Move to ix86-tune-sched-bd.c.
11354 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11355 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11356 (add_insn_window): Move to ix86-tune-sched-bd.c.
11357 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11358 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11359 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11360 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11361 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11362 (do_dispatch): Move to ix86-tune-sched-bd.c.
11363 (has_dispatch): Move to ix86-tune-sched-bd.c.
11364 * config/i386/t-i386: Add new object files.
11365 * config/i386/x86-tune-costs.h: New file.
11366 * config/i386/x86-tune-sched-atom.c: New file.
11367 * config/i386/x86-tune-sched-bd.c: New file.
11368 * config/i386/x86-tune-sched-core.c: New file.
11369 * config/i386/x86-tune-sched.c: New file.
11371 2017-10-11 Liu Hao <lh_mouse@126.com>
11373 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11374 the generic version below instead.
11375 (should_colorize): Recognize Windows consoles as terminals
11377 * pretty-print.c [__MINGW32__] (write_all): New function.
11378 [__MINGW32__] (find_esc_head): Likewise.
11379 [__MINGW32__] (find_esc_terminator): Likewise.
11380 [__MINGW32__] (eat_esc_sequence): Likewise.
11381 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11383 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11386 2017-10-11 Richard Biener <rguenther@suse.de>
11388 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11389 Properly call analyze_scalar_evolution with the loop of the stmt.
11391 2017-10-11 Richard Biener <rguenther@suse.de>
11393 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11394 * tree-core.h (tree_base): Add chrec_var union member.
11395 * tree.h (CHREC_VAR): Remove.
11396 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11397 * tree-chrec.h (build_polynomial_chrec): Adjust.
11398 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11399 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11401 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11403 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11404 * match.pd: ... here.
11405 ((T) X == (T) Y): Relax condition.
11407 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11409 PR tree-optimization/82472
11410 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11412 (break_alias_scc_partitions): Update postorder number.
11414 2017-10-11 Martin Liska <mliska@suse.cz>
11417 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11419 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11421 2017-10-11 Martin Liska <mliska@suse.cz>
11423 * print-rtl.c (print_insn): Move declaration of idbuf
11424 to same scope as name.
11426 2017-10-11 Martin Liska <mliska@suse.cz>
11431 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11432 volatile arguments.
11434 2017-10-11 Martin Liska <mliska@suse.cz>
11437 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11438 volatile arguments.
11440 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11442 * config.gcc (default_gnu_indirect_function): Default to yes for
11443 arm*-*-linux* with glibc.
11445 2017-10-11 Richard Biener <rguenther@suse.de>
11447 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11448 default-defs and types we do not want to analyze.
11449 (interpret_loop_phi): Replace unreachable code with an assert.
11450 (compute_scalar_evolution_in_loop): Remove and inline ...
11451 (analyze_scalar_evolution_1): ... here, replacing condition with
11452 what makes the intent clearer. Remove handling of cases
11453 get_scalar_evolution now handles.
11455 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11457 PR rtl-optimization/81434
11458 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11459 comment for main loop. In sched_group_found if, also add checks for
11460 pass and min_cost_group.
11462 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11464 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11465 (rs6000_insn_cost): New function.
11466 * config/rs6000/rs6000.md (cost): New attribute.
11468 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11469 H.J. Lu <hongjiu.lu@intel.com>
11473 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11474 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11475 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11476 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11477 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11478 OPTION_MASK_ISA_MMX and some other bit set require both
11479 mmx and the ISAs without the mmx bit.
11480 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11481 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11482 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11483 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11484 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11485 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11486 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11487 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11488 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11489 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11490 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11491 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11492 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11493 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11494 Add OPTION_MASK_ISA_MMX.
11496 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11498 * config.gcc (armv7*-*-freebsd*): New target.
11499 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11501 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11503 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11504 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11507 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11509 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11510 a template parameter.
11511 (WIDE_INT_REF_FOR): Update accordingly.
11512 * tree.h (wi::int_traits <const_tree>): Delete.
11513 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11514 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11515 (wi::tree_to_wide_ref): New typedef.
11516 (wi::to_wide): New function.
11517 * calls.c (get_size_range): Use wi::to_wide when operating on
11518 trees as wide_ints.
11519 * cgraph.c (cgraph_node::create_thunk): Likewise.
11520 * config/i386/i386.c (ix86_data_alignment): Likewise.
11521 (ix86_local_alignment): Likewise.
11522 * dbxout.c (stabstr_O): Likewise.
11523 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11524 * expr.c (const_vector_from_tree): Likewise.
11525 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11526 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11527 (fold_negate_expr_1, int_const_binop_1, const_binop)
11528 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11529 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11530 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11531 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11532 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11533 (fold_not_const, round_up_loc): Likewise.
11534 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11535 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11536 (alloca_call_type): Likewise.
11537 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11538 * godump.c (go_output_typedef): Likewise.
11539 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11540 * internal-fn.c (get_min_precision): Likewise.
11541 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11542 * ipa-polymorphic-call.c
11543 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11544 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11545 (ipa_modify_call_arguments): Likewise.
11546 * match.pd: Likewise.
11547 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11548 * print-tree.c (print_node_brief, print_node): Likewise.
11549 * stmt.c (expand_case): Likewise.
11550 * stor-layout.c (layout_type): Likewise.
11551 * tree-affine.c (tree_to_aff_combination): Likewise.
11552 * tree-cfg.c (group_case_labels_stmt): Likewise.
11553 * tree-data-ref.c (dr_analyze_indices): Likewise.
11554 (prune_runtime_alias_test_list): Likewise.
11555 * tree-dump.c (dequeue_and_dump): Likewise.
11556 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11557 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11558 * tree-pretty-print.c (dump_generic_node): Likewise.
11559 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11560 (simple_iv_with_niters): Likewise.
11561 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11562 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11563 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11564 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11565 (refine_value_range_using_guard, number_of_iterations_ne_max)
11566 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11567 (get_cst_init_from_scev, record_nonwrapping_iv)
11568 (scev_var_range_cant_overflow): Likewise.
11569 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11570 * tree-ssa-pre.c (compute_avail): Likewise.
11571 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11572 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11573 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11574 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11575 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11576 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11577 (dump_case_nodes, try_switch_expansion): Likewise.
11578 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11579 (vect_do_peeling): Likewise.
11580 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11581 * tree-vect-stmts.c (vectorizable_load): Likewise.
11582 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11583 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11584 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11585 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11586 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11587 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11588 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11589 (evrp_dom_walker::before_dom_children): Likewise.
11590 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11591 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11592 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11593 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11594 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11595 (get_range_pos_neg): Likewise.
11596 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11597 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11598 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11599 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11600 * config/avr/avr.c (avr_fold_builtin): Likewise.
11601 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11602 * config/msp430/msp430.c (msp430_attr): Likewise.
11603 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11604 * config/powerpcspe/powerpcspe-c.c
11605 (altivec_resolve_overloaded_builtin): Likewise.
11606 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11607 (rs6000_expand_ternop_builtin): Likewise.
11608 * config/rs6000/rs6000-c.c
11609 (altivec_resolve_overloaded_builtin): Likewise.
11610 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11611 (rs6000_expand_ternop_builtin): Likewise.
11612 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11614 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11616 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11617 when copying loop nest with only one inner loop.
11619 2017-10-10 Richard Biener <rguenther@suse.de>
11621 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11622 blocks if SCEV is active.
11623 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11625 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11626 (scev_initialize): Assert we are not yet initialized.
11628 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11630 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11631 inner loop's exit stmt by making it always exit the loop, otherwise
11632 we would generate an infinite empty loop.
11634 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11636 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11637 renaming variables in new preheader if it's deleted.
11639 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11641 * tree-loop-distribution.c (struct partition): Remove unused field
11642 loops of the structure.
11643 (partition_alloc, partition_free): Ditto.
11644 (build_rdg_partition_for_vertex): Ditto.
11646 2017-10-09 Jeff Law <law@redhat.com>
11648 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11649 return type to match prototype and documentation.
11651 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11653 * config/rs6000/rs6000.c (processor_costs): Move to ...
11654 * config/rs6000/rs6000.h: ... here.
11655 (rs6000_cost): Declare.
11657 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11659 * except.c (setjmp_fn): New global variable.
11660 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11661 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11662 if DONT_USE_BUILTIN_SETJMP is defined.
11664 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11666 * target.def (insn_cost): New hook.
11667 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11668 * doc/tm.texi: Regenerate.
11669 * rtlanal.c (insn_cost): Use the new hook.
11671 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11673 * combine.c (combine_validate_cost): Compute the new insn_cost,
11674 not just pattern_cost.
11675 (try_combine): Adjust comment.
11677 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11679 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11681 * combine.c (uid_insn_cost): Adjust comment.
11682 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11684 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11685 * dse.c (find_shift_sequence): Ditto.
11686 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11687 (bb_valid_for_noce_process_p): Use pattern_cost.
11688 * rtl.h (insn_rtx_cost): Delete.
11689 (pattern_cost): New prototype.
11690 (insn_cost): New prototype.
11691 * rtlanal.c (insn_rtx_cost): Rename to...
11692 (pattern_cost): ... this.
11695 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11697 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11698 (*jcc<mode>_0_r_i387): Ditto.
11699 (*jccxf_r_i387): Ditto.
11700 (*jcc<mode>_r_i387): Ditto.
11701 (*jccu<mode>_r_i387): Ditto.
11702 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11703 (*jcc): Rename from *jcc_1.
11705 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11707 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11708 deferred rescans after the lvx/stvx recombination pre-pass.
11710 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11712 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11713 memory operation instruction support.
11714 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11715 (rs6000-ibm-aix[789]*): Likewise.
11716 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11717 Document new functions.
11719 2017-10-09 Richard Biener <rguenther@suse.de>
11721 PR tree-optimization/82397
11722 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11723 equality only for semantically equal trees.
11725 2017-10-09 Richard Biener <rguenther@suse.de>
11727 PR tree-optimization/82449
11728 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11729 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11730 allow constant addresses.
11731 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11734 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11736 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11739 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11742 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11745 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11748 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11750 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11753 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11754 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11756 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11758 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11759 (WI_BINARY_PREDICATE_RESULT): Likewise.
11760 (wi::binary_traits::operator_result): New type.
11761 (wi::binary_traits::predicate_result): Likewise.
11762 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11763 (generic_wide_int::operator==, generic_wide_int::operator!=)
11764 (generic_wide_int::operator&, generic_wide_int::and_not)
11765 (generic_wide_int::operator|, generic_wide_int::or_not)
11766 (generic_wide_int::operator^, generic_wide_int::operator+
11767 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11769 (operator~, unary operator-, operator==, operator!=, operator&)
11770 (operator|, operator^, operator+, binary operator-, operator*): New
11772 * expr.c (get_inner_reference): Use wi::bit_and_not.
11773 * fold-const.c (fold_binary_loc): Likewise.
11774 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11775 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11776 (bit_value_binop): Likewise.
11777 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11778 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11779 (extract_range_from_binary_expr_1): Likewise.
11780 (masked_increment): Likewise.
11781 (simplify_bit_ops_using_ranges): Likewise.
11783 2017-10-09 Martin Jambor <mjambor@suse.cz>
11786 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11787 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11788 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11790 (hsa_fixup_mov_insn_type): New function.
11791 (hsa_op_with_type::get_in_type): Use it.
11792 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11793 immediates in an assert.
11794 (hsa_op_with_type::extend_int_to_32bit): New method.
11795 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11796 types. Convert to dest type if necessary.
11797 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11798 (reg_for_gimple_ssa): Pass false as min32int to
11799 hsa_type_for_scalar_tree_type.
11800 (gen_hsa_addr): Fixup type when creating addresable temporary.
11801 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11802 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11803 necessary. Call hsa_fixup_mov_insn_type.
11804 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11805 extend operands and convert to dest type if necessary.
11806 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11807 to dest type if necessary.
11808 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11809 if conversion nt necessary and size matches.
11810 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11811 to dest type if necessary.
11812 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11813 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11815 (gen_hsa_clrsb): Likewise.
11816 (gen_hsa_ffs): Likewise.
11817 (gen_hsa_divmod): Extend operands and convert to dest type if
11819 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11821 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11823 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11824 Remove empty default arguments. Use a brace block as output
11826 (conditional return): Ditto.
11828 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11829 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11831 (group_ending_nop): Ditto.
11832 (doloop_end): Ditto.
11833 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11834 (splitters for those): Ditto.
11836 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11838 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11839 a conditional jump (and the compare for it) so that pc_rtx is the
11841 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11842 for the deleted and renamed ctr<mode>_internal[234] patterns.
11843 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11844 Delete second conditional return pattern.
11845 (ctr<mode>_internal2): Delete this second bdnz pattern.
11846 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11847 (ctr<mode>_internal4): Delete this second bdz pattern.
11849 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11851 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11852 (always_initialized_rtx_for_ssa_name_p): New predicate.
11853 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11854 (finish_out_of_ssa): Free new field of SA.
11855 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11856 * tree-ssa-coalesce.c: Include tree-ssa.h.
11857 (get_parm_default_def_partitions): Remove extern keyword.
11858 (get_undefined_value_partitions): New function.
11859 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
11860 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
11861 uninitialized bits.
11862 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
11864 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11866 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
11868 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11870 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
11871 for targets that preffer 128bit.
11873 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11875 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
11877 2017-10-08 Olivier Hainque <hainque@adacore.com>
11879 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
11880 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
11882 (thumb_set_return_address): Likewise.
11884 2017-10-08 Olivier Hainque <hainque@adacore.com>
11886 * common/config/arm/arm-common.c (arm_except_unwind_info):
11887 Handle DWARF2_UNWIND_INFO.
11889 2017-10-07 Michael Collison <michael.collison@arm.com>
11891 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
11894 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
11896 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
11897 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
11898 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
11899 defined, force the creation of a new block for a dispatch label.
11901 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
11903 * invoke.texi (Wsuggest-attribute=cold): Document.
11904 * common.opt (Wsuggest-attribute=cold): New
11905 * ipa-pure-const.c (warn_function_cold): New function.
11906 * predict.c (compute_function_frequency): Use it.
11907 * predict.h (warn_function_cold): Declare.
11909 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11911 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
11914 2017-10-06 Martin Liska <mliska@suse.cz>
11916 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
11917 keyword for member functions.
11918 (struct sanopt_tree_couple): New struct.
11919 (struct sanopt_tree_couple_hash): New function.
11920 (struct sanopt_ctx): Add new hash_map.
11921 (has_dominating_ubsan_ptr_check): New function.
11922 (record_ubsan_ptr_check_stmt): Likewise.
11923 (maybe_optimize_ubsan_ptr_ifn): Likewise.
11924 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
11925 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
11927 2017-10-06 Sudakshina Das <sudi.das@arm.com>
11930 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
11931 aarch64_simd_valid_immediate on CONST_VECTORs.
11932 (aarch64_reg_or_bic_imm): Likewise.
11934 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
11936 PR rtl-optimization/82396
11937 * haifa-sched.c (ready_sort_real): Disable qsort checking.
11939 2017-10-06 Sebastian Pop <sebpop@gmail.com>
11941 * graphite-dependences.c (scop_get_reads): Move code to...
11942 (scop_get_must_writes): Move code to...
11943 (scop_get_may_writes): Move code to...
11944 (scop_get_reads_and_writes): ... here.
11945 (scop_get_dependences): Call scop_get_reads_and_writes.
11947 2017-10-06 Jakub Jelinek <jakub@redhat.com>
11949 PR tree-optimization/82434
11950 * fold-const.h (can_native_encode_type_p,
11951 can_native_encode_string_p): Remove.
11952 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
11953 don't encode anything, just return what would be otherwise returned.
11954 (native_encode_fixed, native_encode_complex, native_encode_vector):
11956 (native_encode_string): Likewise. Inline by hand
11957 can_native_encode_string_p.
11958 (can_native_encode_type_p): Remove.
11959 (can_native_encode_string_p): Remove.
11960 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
11961 STRING_CSTs using can_native_encode_string_p, test all
11962 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
11963 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
11964 argument from native_encode_expr.
11965 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
11966 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
11969 2017-10-06 Richard Biener <rguenther@suse.de>
11971 PR tree-optimization/82397
11972 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
11973 operand_equal_p but rely on data_ref_compare_tree for detecting
11975 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
11976 to match up with dr_group_sort_cmp.
11978 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11981 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
11983 * config/s390/s390-builtin-types.def: Regenerate.
11985 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11988 * config/s390/s390-builtin-types.def: Regenerate.
11989 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
11990 Change flag from B_VXE to B_VX.
11991 (s390_vec_min_dbl): Remove B_VXE flag.
11993 2017-10-06 Richard Biener <rguenther@suse.de>
11995 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
11996 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
11997 translate_isl_ast_to_gimple::is_valid_rename,
11998 translate_isl_ast_to_gimple::get_rename,
11999 translate_isl_ast_to_gimple::get_def_bb_for_const,
12000 translate_isl_ast_to_gimple::get_new_name,
12001 translate_isl_ast_to_gimple::collect_all_ssa_names,
12002 translate_isl_ast_to_gimple::copy_loop_phi_args,
12003 translate_isl_ast_to_gimple::collect_all_ssa_names,
12004 translate_isl_ast_to_gimple::copy_loop_phi_args,
12005 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12006 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12007 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12008 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12009 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12010 translate_isl_ast_to_gimple::copy_cond_phi_args,
12011 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12012 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12013 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12014 translate_isl_ast_to_gimple::rename_uses,
12015 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12016 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12017 (set_rename_for_each_def): Likewise.
12018 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12019 here. Handle rewriting SCEV analyzable uses here.
12020 (copy_bb_and_scalar_dependences): Generate code for PHI
12022 (graphite_regenerate_ast_isl): Adjust.
12023 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12024 (add_write, add_read): New functions.
12025 (build_cross_bb_scalars_def): Use it and simplify.
12026 (build_cross_bb_scalars_use): Likewise.
12027 (graphite_find_cross_bb_scalar_vars): Inline into...
12028 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12029 simulating out-of-SSA. Compute liveout and add dependencies.
12030 (build_scops): Force an empty entry block.
12031 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12033 (sese_build_liveouts): Declare.
12034 (sese_trivially_empty_bb_p): Likewise.
12035 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12036 compute liveout and debug_liveout.
12037 (sese_bad_liveouts_use): Remove.
12038 (sese_reset_debug_liveouts_bb): Likewise.
12039 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12040 (sese_build_liveouts): Build liveout and debug_liveout and store
12042 (new_sese_info): Adjust.
12043 (free_sese_info): Likewise.
12044 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12045 do not build liveout here.
12046 (move_sese_in_condition): Adjust region entry.
12047 (scev_analyzable_p): Match up with chrec_apply requirements.
12048 (sese_trivially_empty_bb_p): New.
12049 * tree-into-ssa.c (get_reaching_def): Properly support generating
12050 default-defs for incremental rewrite of anonymous names.
12052 2017-10-06 Richard Biener <rguenther@suse.de>
12054 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12055 precision do not perform modulo reduction.
12057 2017-10-06 Richard Biener <rguenther@suse.de>
12059 PR tree-optimization/82436
12060 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12061 conservatively choose the vectorization factor when checking
12062 whether we can perform the required load permutation.
12063 (vect_transform_slp_perm_load): Assert when we may not fail.
12065 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12067 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12068 message for incompatible -msdata=* and -mcall-* options.
12070 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12072 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12073 rate for post-reload scheduling.
12075 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12077 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12079 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12081 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12082 to improve monte carlo in scimark.
12084 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12086 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12087 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12088 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12089 pentium4_cost, nocona_cost): Set reassociation width to 1.
12090 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12091 width to 2 for fp operations and 1 otherwise.
12092 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12094 (atom_cost): Set reassociation width to 2.
12095 (slm_cost, generic_cost): Set fp reassociation width
12096 to 2 and 1 otherwise.
12097 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12098 (core_cost): Set fp reassociation width to 4 and vector to 2.
12099 (ix86_reassociation_width): Rewrite using cost table; special case
12100 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12101 and TARGET_AVX128_OPTIMAL.
12102 * config/i386/i386.h (processor_costs): Add
12103 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12104 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12105 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12106 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12107 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12108 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12110 2017-10-05 Nathan Sidwell <nathan@acm.org>
12112 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12114 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12116 * config/arm/arm.c (arm_test_fpu_data): New.
12117 (arm_run_selftests): Call arm_test_fpu_data.
12119 2017-10-04 Nathan Sidwell <nathan@acm.org>
12121 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12123 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12125 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12127 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12128 check type for aarch64_simd_valid_immediate.
12129 (aarch64_output_simd_mov_immediate): Update prototype.
12130 (aarch64_simd_valid_immediate): Update prototype.
12131 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12132 support for ORR-immediate.
12133 (and<mode>3): modified pattern to add support for BIC-immediate.
12134 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12135 now checks for valid immediate for BIC and ORR based on new enum
12137 (aarch64_output_simd_mov_immediate): Function now used to output
12138 BIC/ORR imm as well based on new enum argument.
12139 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12141 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12142 (aarch64_reg_or_bic_imm): Likewise.
12144 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12146 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12147 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12149 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12153 PR rtl-optimization/82396
12154 * haifa-sched.c (autopref_multipass_init): Simplify
12156 (autopref_rank_data): Simplify sort order.
12157 * sched-int.h (autopref_multipass_data_): Remove
12158 multi_mem_insn_p, min_offset and max_offset.
12160 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12162 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12164 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12166 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12167 vect_doubleint_cvt.
12169 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12171 * doc/sourcebuild.texi: Document vect_long_mult.
12173 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12175 PR tree-optimization/82413
12176 * fold-const.c (build_range_check): Use widest_int when comparing
12177 the maximum ETYPE value with HIGH.
12179 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12181 PR rtl-optimization/82396
12182 * haifa-sched.c (autopref_multipass_init): Simplify
12184 (autopref_rank_data): Simplify sort order.
12185 * sched-int.h (autopref_multipass_data_): Remove
12186 multi_mem_insn_p, min_offset and max_offset.
12188 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12190 PR tree-optimization/82381
12191 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12192 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12195 PR tree-optimization/82374
12196 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12197 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12198 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12199 current_function_decl to the new decl.
12201 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12203 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12204 helper macro for IEEE float128 hardware built-in functions.
12205 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12207 (TRUNCF128_ODD): Likewise.
12208 (ADDF128_ODD): Likewise.
12209 (SUBF128_ODD): Likewise.
12210 (MULF128_ODD): Likewise.
12211 (DIVF128_ODD): Likewise.
12212 (FMAF128_ODD): Likewise.
12213 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12214 UNSPEC_TRUNC_ROUND_TO_ODD.
12215 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12216 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12217 floating point round to odd instructions.
12218 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12219 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12220 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12221 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12222 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12223 (trunc<mode>sf2_hw): Change the truncate with round to odd
12224 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12225 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12226 to odd hardware instructions.
12227 (sub<mode>3_odd): Likewise.
12228 (mul<mode>3_odd): Likewise.
12229 (div<mode>3_odd): Likewise.
12230 (sqrt<mode>2_odd): Likewise.
12231 (fma<mode>4_odd): Likewise.
12232 (fms<mode>4_odd): Likewise.
12233 (nfma<mode>4_odd): Likewise.
12234 (nfms<mode>4_odd): Likewise.
12235 (trunc<mode>df2_odd): Change the truncate with round to odd
12236 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12238 * doc/extend.texi (PowerPC built-in functions): Update documentation
12239 for existing IEEE float128-bit built-in functions. Add built-in
12240 functions that generate the IEEE 128-bit floating point round to
12243 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12245 PR rtl-optimization/77729
12246 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12247 to (X&(C1&~C2))|C2 transformations.
12249 2017-10-03 Martin Jambor <mjambor@suse.cz>
12251 PR tree-optimization/82363
12252 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12253 mismatch, mark lacc written regardless of racc.
12255 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12257 PR tree-optimization/82381
12258 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12259 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12260 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12261 fallthrough into reassoc_stmt_dominates_stmt_p.
12264 * combine.c (combine_instructions): Don't combine in unreachable
12267 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12270 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12271 function to not use the have_cpu variable. Do not set cpu_index,
12272 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12273 or the default cpu.
12274 (rs6000_valid_attribute_p): Remove duplicate initializations of
12275 old_optimize and func_optimize.
12276 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12277 (rs6000_activate_target_options): Make global.
12278 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12281 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12283 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12284 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12285 Set *poffset to 0 and *psize and *pmax_size to -1 if
12286 *poffset + *psize overflows in HOST_WIDE_INT.
12288 PR tree-optimization/82387
12289 PR tree-optimization/82388
12290 PR tree-optimization/82389
12291 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12292 instead of live_bytes non-NULL.
12294 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12297 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12300 2017-10-02 Richard Biener <rguenther@suse.de>
12302 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12303 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12304 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12306 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12308 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12309 requested precision matches the type's.
12310 * calls.c (alloc_max_size): Calculate the new candidate size as
12311 a widest_int and use wi::to_widest when comparing it with the
12312 current candidate size.
12313 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12314 zero rather than integer_zero_node.
12315 * match.pd: Check for a no-op conversion before using wi::add
12316 rather than after. Use tree_to_uhwi when summing small shift
12317 counts into an unsigned int.
12319 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12320 Alan Hayward <alan.hayward@arm.com>
12321 David Sherwood <david.sherwood@arm.com>
12324 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12325 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12326 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12327 POINTER_AND_FP_REGS.
12329 2017-10-02 Richard Biener <rguenther@suse.de>
12331 PR tree-optimization/82355
12332 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12333 a mapping for the enclosing loop but avoid generating one for
12334 the loop tree root.
12335 (copy_bb_and_scalar_dependences): Remove premature codegen
12336 error on PHIs in blocks duplicated into multiple places.
12337 * graphite-scop-detection.c
12338 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12339 in the region use it as loop and nest to analyze the DR in.
12340 (try_generate_gimple_bb): Likewise.
12341 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12342 (add_loop_constraints): For blocks in a loop not in the region
12343 create a dimension with a single iteration.
12344 * sese.h (gbb_loop_at_index): Remove assert.
12346 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12348 * omp-expand.c (adjust_context_scope): New function.
12349 (expand_parallel_call): Call adjust_context_scope.
12351 2017-10-01 Jeff Law <law@redhat.com>
12353 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12354 dom_opt_dom_walker class with direct access to private members.
12355 Add comments. Call test_for_singularity.
12356 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12357 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12358 m_dummy_cond anymore.
12359 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12361 (pass_dominator:execute): Build the dummy_cond here and pass it
12362 to the dom_opt_dom_walker ctor.
12363 (test_for_singularity): New function.
12365 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12366 Maya Rashish <coypu@sdf.org>
12368 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12369 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12370 (alpha*-*-netbsd*) Use nbsd_tm_file.
12371 (arm*-*-netbsdelf*) Likewise.
12372 (i[34567]86-*-netbsdelf*) Likewise.
12373 (x86_64-*-netbsd*) Likewise.
12374 (mips*-*-netbsd*) Likewise.
12375 (powerpc-*-netbsd*) Likewise.
12376 (sh*-*-netbsd*) Likewise.
12377 (sparc-*-netbsdelf*) Likewise.
12378 (sparc64-*-netbsd*) Likewise.
12379 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12381 (vax-*-netbsdelf*) Likewise.
12382 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12383 (UINT_FAST8_TYPE) Likewise.
12384 (INT_FAST16_TYPE) Check CHAR_FAST16.
12385 (UINT_FAST16_TYPE) Likewise.
12387 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12390 * config/i386/i386.md
12391 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12392 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12393 *divmodsi4_zext_2): New define_insn_and_split.
12394 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12395 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12396 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12397 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12398 New define_insn_and_split.
12399 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12400 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12401 operands[1] having DImode when mode is SImode.
12403 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12404 always SImode for DIV and MOD in REG_EQUAL notes.
12406 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12408 PR middle-end/82319
12409 * match.pd: Fix handling of NaNs in pattern.
12411 2017-09-29 Jeff Law <law@redhat.com>
12413 * sbitmap.c (bitmap_bit_in_range_p): New function.
12414 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12415 * tree-ssa-dse.c (live_bytes_read): New function.
12416 (dse_classify_store): Ignore reads of dead bytes.
12418 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12419 typos and whitespace errors.
12420 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12421 * config/s390/s390.c (s390_emit_prologue): Likewise.
12423 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12426 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12427 with a symbol for LRA.
12429 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12431 PR rtl-optimization/82338
12432 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12434 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12436 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12437 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12438 (qsort_chk): Declare.
12439 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12440 (qsort_chk): New function.
12442 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12444 PR tree-optimization/82337
12445 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12446 phi definition if the PHI result appears in an abnormal PHI.
12447 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12448 basis appears in an abnormal PHI.
12450 2017-09-29 Richard Biener <rguenther@suse.de>
12452 * graphite-isl-ast-to-gimple.c
12453 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12454 (binary_op_to_tree): Use it.
12455 (get_rename_from_scev): Likewise.
12456 (copy_loop_phi_nodes): Likewise.
12457 (copy_bb_and_scalar_dependences): Likewise.
12458 (translate_pending_phi_nodes): Likewise.
12460 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12463 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12464 for movabsq $(i32 << shift), r64.
12466 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12468 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12469 index when encoding %esp as %rsp to avoid 0x67 prefix.
12471 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12473 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12474 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12476 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12478 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12479 Extensions with more than 16 double VFP registers.
12480 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12481 to_clear_mask and all code related to it. Replace the remaining
12482 entry by a sbitmap and adapt code accordingly.
12484 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12486 * brig-builtins.def: Change pure attributes to const.
12488 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12490 * config.gcc (default_gnu_indirect_function): Default to yes for
12491 sparc*-*-linux* with glibc.
12493 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12495 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12496 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12497 when creating .init_array and .fini_array sections with priority
12500 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12503 * config/aarch64/aarch64.c
12504 (aarch64_builtin_support_vector_misalignment): Always return false
12505 when misalignment is unknown.
12507 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12509 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12510 this function to return false if the definition used by the swap
12511 instruction is artificial, or if the memory address from which the
12512 constant value is loaded is not represented by a base address held
12513 in a register or if the base address register is a frame or stack
12514 pointer. Additionally, return false if the base address of the
12515 loaded constant is a SYMBOL_REF but is not considered to be a
12517 (replace_swapped_load_constant): New function.
12518 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12519 loaded constant vector with a load of a swapped constant vector.
12521 2017-09-27 Carl Love <cel@us.ibm.com>
12523 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12524 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12525 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12528 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12530 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12531 first, always call autopref_rank_data otherwise.
12533 2017-09-27 Richard Biener <rguenther@suse.de>
12535 * graphite-scop-detection.c (find_scop_parameters): Move
12536 loop bound handling ...
12537 (gather_bbs::before_dom_children): ... here, avoiding the need
12538 to build scop_info->loop_nest.
12539 (record_loop_in_sese): Remove.
12540 * sese.h (sese_info_t::loop_nest): Remove.
12541 * sese.c (new_sese_info): Do not allocate loop_nest.
12542 (free_sese_info): Do not free loop_nest.
12544 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12547 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12548 lhs from calls if the lhs has addressable type.
12550 2017-09-27 Richard Biener <rguenther@suse.de>
12552 * graphite.h (scop::max_alias_set): New member.
12553 * graphite-scop-detection.c: Remove references to non-existing
12554 --param in comments.
12555 (build_alias_sets): Record the maximum alias set used for drs.
12556 (build_scops): Support zero as unlimited for
12557 --param graphite-max-arrays-per-scop.
12558 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12559 and inline into ...
12560 (build_poly_sr_1): ... here. Compute alias set based on the
12561 maximum alias set used for drs rather than
12562 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12564 2017-09-27 Richard Biener <rguenther@suse.de>
12566 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12567 --param loop-block-tile-size=0 to disable tiling.
12569 2017-09-27 Richard Biener <rguenther@suse.de>
12571 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12572 (graphite-max-nb-scop-params): Document special value zero.
12573 * domwalk.h (dom_walker::STOP): New symbolical constant.
12574 (dom_walker::dom_walker): Add optional parameter for bb to
12576 (dom_walker::~dom_walker): Declare.
12577 (dom_walker::before_dom_children): Document STOP return value.
12578 (dom_walker::m_user_bb_to_rpo): New member.
12579 (dom_walker::m_bb_to_rpo): Likewise.
12580 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12581 mapping here if not provided by the user.
12582 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12583 provided by the user.
12584 (dom_walker::STOP): Define.
12585 (dom_walker::walk): Do not compute bb to RPO mapping here.
12586 Support STOP return value from before_dom_children to stop
12588 * graphite-optimize-isl.c (optimize_isl): If the schedule
12589 is the same still generate code if -fgraphite-identity
12590 or -floop-parallelize-all are given.
12591 * graphite-scop-detection.c: Include cfganal.h.
12592 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12594 (gather_bbs::before_dom_children): Return STOP for BBs
12596 (build_scops): Compute bb to RPO mapping and pass it to
12597 the domwalk. Treat --param graphite-max-nb-scop-params=0
12598 as not limiting the number of params.
12599 * graphite.c (graphite_initialize): Remove limit on the
12600 number of basic-blocks in a function.
12601 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12602 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12603 default value of 10.
12605 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12607 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12608 Adjust code to eliminate needing to do the shift right 32-bits
12609 operation after XSCVDPSPN.
12611 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12613 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12614 ((X / Y) != 0 -> X >= Y): Likewise.
12616 2017-09-26 Carl Love <cel@us.ibm.com>
12618 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12619 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12620 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12621 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12622 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12623 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12624 definitions and overloading.
12625 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12626 statement for P9V_BUILTIN_XST_LEN_R.
12627 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12628 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12629 define_expand and define_insn for the instructions and builtins.
12630 * doc/extend.texi: Update the built-in documentation file for the new
12631 built-in functions.
12632 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12633 define_insn for the instructions
12635 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12638 * gcc/config/netbsd-protos.h: New file.
12639 * gcc/config/netbsd.c: New file.
12640 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12641 * gcc/config/t-netbsd: New file.
12642 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12643 (tmake_file) Add t-netbsd.
12644 (extra_objs) Add netbsd.o.
12646 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12650 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12652 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12654 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12655 sign extension from a vector register to a GPR by doing a 32-bit
12656 direct move and then an EXTSW.
12657 (extendsi<mode>2 splitter): Likewise.
12658 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12659 right or vector extract after doing XSCVDPSPN. Use
12660 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12662 (movdi_from_sf_zero_ext): Likewise.
12663 (reload_gpr_from_vsxsf): Likewise.
12664 (p8_mfvsrd_4_disf): Delete, no longer used.
12665 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12666 then needing to move the SFmode to a GPR to use the XSCVDPSP
12667 instruction instead of FRSP and XSCVDPSPN.
12668 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12669 it is adjacent to the other XSCVSPDP insns.
12670 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12671 SFmode to be in traditional Altivec registers.
12672 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12673 (vsx_xscvspdpn): Likewise.
12674 (vsx_xscvspdpn_scalar): Likewise.
12676 2017-09-26 Martin Jambor <mjambor@suse.cz>
12678 * tree-sra.c (compare_access_positions): Put integral types first,
12679 stabilize sorting of integral types, remove conditions putting
12680 non-full-precision integers last.
12681 (sort_and_splice_var_accesses): Disable scalarization if a
12682 non-integert would be represented by a non-full-precision integer.
12684 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12686 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12687 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12688 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12689 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12690 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12691 conditionals inside the function instead of around it. Call
12692 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12693 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12695 2017-09-26 Richard Biener <rguenther@suse.de>
12697 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12699 (scop_detection::build_scop_breadth): ... this. Removed.
12700 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12701 (scop_detection::harmful_stmt_in_bb): Likewise.
12702 (scop_detection::graphite_can_represent_stmt): Likewise.
12703 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12704 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12705 (scop_detection::can_represent_loop_1): ... this. Removed.
12706 (scop_detection::harmful_loop_in_region): Simplify after inlining
12707 the above and remove more quadraticness.
12708 (build_scops): Adjust.
12709 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12712 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12715 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12716 REGNO (base) == SP_REG if base is a REG.
12718 PR middle-end/35691
12719 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12720 if it is different SSA_NAME.
12721 (optimize_range_tests_cmp_bitwise): New function.
12722 (optimize_range_tests): Call it.
12724 2017-09-26 Richard Biener <rguenther@suse.de>
12726 PR tree-optimization/82321
12727 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12728 for the def being inside the loop.
12730 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12732 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12734 * config/s390/s390-builtins.def: Fix constraint on op4.
12736 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12738 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12739 independent expanders.
12740 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12741 ("vec_ordered", "vec_unordered"): New expanders.
12743 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12745 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12748 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12750 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12751 vec_unpacks_lo_v16qi.
12752 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12754 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12756 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12757 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12758 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12760 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12762 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12764 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12766 ("*vec_slb<mode>"): New insn pattern.
12767 ("vec_shr_<mode>"): New expander.
12768 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12769 and force the shift count operand to V16QImode.
12770 ("vec_srb<mode>"): Set shift count mode to V16QI.
12772 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12774 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12775 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12776 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12778 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12781 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12783 2017-09-26 Richard Biener <rguenther@suse.de>
12785 PR tree-optimization/82320
12786 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12789 2017-09-25 Jeff Law <law@redhat.com>
12791 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12792 prototype for new argument.
12793 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12794 mostly extracted from rs6000_emit_allocate_stack.
12795 (rs6000_emit_probe_stack_range_stack_clash): New function.
12796 (rs6000_emit_allocate_stack): Call
12797 rs6000_emit_probe_stack_range_stack_clash as needed.
12798 (rs6000_emit_probe_stack_range): Add additional argument
12799 to call to gen_probe_stack_range{si,di}.
12800 (output_probe_stack_range): New.
12801 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12802 (output_probe_stack_range_stack_clash): New.
12803 (rs6000_emit_prologue): Emit notes into dump file as requested.
12804 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12805 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12806 Add additional operand and pass it to output_probe_stack_range.
12808 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12810 PR tree-optimization/82163
12811 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12812 (checking_verify_loop_closed_ssa): New parameter.
12813 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12814 (check_loop_closed_ssa_stmt): Delete.
12815 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12816 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12817 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12820 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12822 * brig-builtins.def: Treat HSAIL barrier builtins as
12823 setjmp/longjump style functions.
12825 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12827 * target.def (constant_alignment): New hook.
12828 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12829 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12830 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12831 * doc/tm.texi: Regenerate.
12832 * targhooks.h (default_constant_alignment): Declare.
12833 (constant_alignment_word_strings): Likewise.
12834 * targhooks.c (default_constant_alignment): New function.
12835 (constant_alignment_word_strings): Likewise.
12836 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12837 instead of CONSTANT_ALIGNMENT.
12838 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12839 (force_const_mem): Likewise.
12840 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12841 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12842 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12843 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12844 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12846 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12847 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12848 constant_alignment_word_strings.
12849 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12850 (CONSTANT_ALIGNMENT): Likewise.
12851 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12852 (arm_constant_alignment): New function.
12853 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12854 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12855 constant_alignment_word_strings.
12856 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12857 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12858 constant_alignment_word_strings.
12859 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
12860 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12861 (cris_constant_alignment): New function.
12862 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
12863 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12864 (epiphany_constant_alignment): New function.
12865 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
12866 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12867 constant_alignment_word_strings.
12868 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
12869 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12870 constant_alignment_word_strings.
12871 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
12872 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12873 constant_alignment_word_strings.
12874 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
12875 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
12876 * config/i386/i386.c (ix86_constant_alignment): Make static.
12877 Use the same interface as the target hook.
12878 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12879 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
12880 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12881 constant_alignment_word_strings.
12882 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
12883 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
12884 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12885 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
12886 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12887 constant_alignment_word_strings.
12888 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
12889 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12890 constant_alignment_word_strings.
12891 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
12892 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12893 constant_alignment_word_strings.
12894 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
12895 * config/microblaze/microblaze.c (microblaze_constant_alignment):
12897 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12898 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
12899 * config/mips/mips.c (mips_constant_alignment): New function.
12900 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12901 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
12902 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
12903 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12904 (mmix_constant_alignment): Make static. Use the same interface
12905 as the target hook.
12906 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
12907 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12908 constant_alignment_word_strings.
12909 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
12910 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12911 constant_alignment_word_strings.
12912 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
12913 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12914 constant_alignment_word_strings.
12915 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
12916 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12917 (rs6000_constant_alignment): New function.
12918 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
12919 * config/riscv/riscv.c (riscv_constant_alignment): New function.
12920 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12921 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
12922 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12923 (rs6000_constant_alignment): New function.
12924 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
12925 * config/s390/s390.c (s390_constant_alignment): New function.
12926 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12927 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
12928 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12929 constant_alignment_word_strings.
12930 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
12931 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12932 (sparc_constant_alignment): New function.
12933 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
12934 * config/spu/spu.c (spu_constant_alignment): New function.
12935 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12936 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
12937 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12938 constant_alignment_word_strings.
12939 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
12940 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12941 constant_alignment_word_strings.
12942 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
12943 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12944 constant_alignment_word_strings.
12945 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
12946 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12947 (visium_constant_alignment): New function.
12948 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
12949 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12950 (xtensa_constant_alignment): New function.
12951 * system.h (CONSTANT_ALIGNMENT): Poison.
12953 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
12955 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12956 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
12957 (rs6000_builtin_valid_without_lhs): New helper function.
12958 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12959 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
12961 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12963 * target.h (vec_perm_indices): Use unsigned short rather than
12965 (auto_vec_perm_indices): Likewise.
12966 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
12967 Use unsigned int rather than unsigned char.
12968 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
12970 2017-09-25 Richard Biener <rguenther@suse.de>
12972 * cfgloop.h (sort_sibling_loops): Declare.
12973 * cfgloop.c (sort_sibling_loops_cmp): New helper.
12974 (sort_sibling_loops): New function sorting the sibling loop list
12976 * graphite.c (graphite_transform_loops): Sort sibling loops.
12978 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
12980 * target.def (vec_perm_const_ok): Change sel parameter to
12982 * optabs-query.c (can_vec_perm_p): Update accordingly.
12983 * doc/tm.texi: Regenerate.
12984 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
12985 auto_vec_perm_indices and remove separate nelt field.
12986 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
12987 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
12988 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
12989 (aarch64_expand_vec_perm_const): Update accordingly.
12990 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
12991 to vec_perm_indices.
12992 * config/arm/arm.c (expand_vec_perm_d): Change perm to
12993 auto_vec_perm_indices and remove separate nelt field.
12994 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
12995 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
12996 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
12998 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
12999 to vec_perm_indices.
13000 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13001 sel to vec_perm_indices.
13002 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13003 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13004 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13006 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13009 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13012 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13013 on the FUNCTION_DECL function context if it has a DIE that is a
13016 2017-09-25 Richard Biener <rguenther@suse.de>
13018 PR tree-optimization/82285
13019 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13022 2017-09-25 Tom de Vries <tom@codesourcery.com>
13026 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13029 2017-09-25 Richard Biener <rguenther@suse.de>
13031 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13032 ISL errors other than isl_error_quota happen. Dump if the
13033 schedule is the same.
13034 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13035 errors instead of aborting inside ISL.
13037 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13040 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13041 of libgcc_eh for m64.
13042 * config/i386/darwin64.h: Likewise.
13044 2017-09-25 Richard Biener <rguenther@suse.de>
13046 PR middle-end/82144
13047 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13048 attribute for incomplete types nor twice for complete ones.
13050 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13053 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13054 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13057 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13060 * config/i386/i386.opt (mprefer-avx256): Use
13061 ix86_target_flags variable.
13062 * config/i386/i386.c (ix86_target_string): Move
13063 -mprefer-avx256 to flag2_opts.
13065 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13067 PR middle-end/35691
13068 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13069 and x != -1 | y != -1 into (x & y) != -1.
13071 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13073 * config.gcc: Add new case statement to set
13074 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13075 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13076 s390x-*-linux* case statements. Added aarch64 to the list of
13077 supported architectures.
13079 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13081 PR tree-optimization/82289
13082 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13083 STMT_VINFO_RELEVANT_P.
13085 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13086 Alan Hayward <alan.hayward@arm.com>
13087 David Sherwood <david.sherwood@arm.com>
13089 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13090 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13091 (extract_range_from_binary_expr_1): Don't call
13092 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13094 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13095 Alan Hayward <alan.hayward@arm.com>
13096 David Sherwood <david.sherwood@arm.com>
13098 * target.def (preferred_vector_alignment): New hook.
13099 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13101 * doc/tm.texi: Regenerate.
13102 * targhooks.h (default_preferred_vector_alignment): Declare.
13103 * targhooks.c (default_preferred_vector_alignment): New function.
13104 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13106 (DR_TARGET_ALIGNMENT): New macro.
13107 (aligned_access_p): Update commentary.
13108 (vect_known_alignment_in_bytes): New function.
13109 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13111 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13112 Calculate the misalignment based on the target alignment rather than
13114 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13115 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13116 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13117 the target alignment, rather than masking the element misalignment
13118 with the number of elements in a vector. Also use the target
13119 alignment when calculating the maximum number of peels.
13120 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13121 instead of TYPE_ALIGN_UNIT.
13122 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13123 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13124 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13125 (vect_create_data_ref_ptr): Likewise.
13126 (vect_setup_realignment): Realign by ANDing with
13127 -DR_TARGET_MISALIGNMENT.
13128 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13129 the number of peels based on DR_TARGET_ALIGNMENT.
13130 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13131 with the guaranteed alignment boundary when deciding whether
13133 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13134 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13135 (ensure_base_align): Remove stmt_info parameter. Get the
13136 target base alignment from DR_TARGET_ALIGNMENT.
13137 (vectorizable_store): Update call accordingly. Interpret
13138 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13140 (vectorizable_load): Likewise.
13142 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13143 Alan Hayward <alan.hayward@arm.com>
13144 David Sherwood <david.sherwood@arm.com>
13146 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13147 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13148 (vect_enhance_data_refs_alignment): Likewise.
13150 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13152 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13153 error. Only quit immediately if parsing is complete.
13154 (BEGIN): Initialize fatal_err and parse_done.
13155 (begin fpu, end fpu): Check number of arguments.
13156 (begin arch, end arch): Likewise.
13157 (begin cpu, end cpu): Likewise.
13158 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13159 (optalias): Likewise.
13161 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13163 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13164 * config/arm/arm-isa.h: Delete. Move definitions to ...
13165 * arm-cpus.in: ... here. Use new feature and fgroup values.
13166 * config/arm/arm.c (arm_option_override): Use lower case for feature
13168 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13169 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13170 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13172 (print_isa_bits_for): New function.
13173 (gen_isa): New function.
13174 (gen_comm_data): Use print_isa_bits_for.
13175 (define feature): New keyword.
13176 (define fgroup): New keyword.
13177 * config/arm/t-arm (TM_H): Remove.
13178 (GTM_H): Add arm-isa.h.
13179 (arm-isa.h): Add rule to generate file.
13180 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13181 case for feature bit names.
13183 2017-09-22 Richard Biener <rguenther@suse.de>
13185 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13187 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13188 print of no dependency loops ...
13189 * graphite.c (graphite_transform_loops): ... here.
13190 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13192 (same_close_phi_node, remove_duplicate_close_phi,
13193 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13194 (canonicalize_loop_closed_ssa): ... here and simplify.
13195 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13196 (optimize_isl): Use dump_printf_loc to tell when we stopped
13197 optimizing because of an ISL timeout.
13199 2017-09-22 Richard Biener <rguenther@suse.de>
13201 PR tree-optimization/82291
13202 * tree-if-conv.c (predicate_mem_writes): Make sure to
13203 remove writes in blocks predicated with false.
13205 2017-09-22 Richard Biener <rguenther@suse.de>
13207 * sese.c: Include cfganal.h.
13208 (if_region_set_false_region): Remove.
13209 (create_if_region_on_edge): Likewise.
13210 (move_sese_in_condition): Re-implement without destroying
13213 2017-09-22 Richard Biener <rguenther@suse.de>
13215 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13216 Verify both BBs contain loop PHI nodes before dispatching to
13217 copy_loop_phi_args.
13218 (graphite_regenerate_ast_isl): Do not recompute dominators,
13219 do not verify three times. Restructure for clarity.
13220 * graphite-scop-detection.c (same_close_phi_node,
13221 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13222 defined_in_loop_p, canonicalize_loop_closed_ssa,
13223 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13224 checking and SSA rewrite, move to ...
13225 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13226 (graphite_initialize): Do not pass in ctx, do not reset the
13227 SCEV cache, compute only dominators.
13228 (graphite_transform_loops): Allocate ISL ctx after
13229 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13230 Maintain post-dominators only around build_scops.
13231 * sese.c (if_region_set_false_region): Make static. Free
13232 and recompute dominators.
13233 (move_sese_in_condition): Assert we don't get called with
13234 post-dominators computed.
13235 * sese.h (if_region_set_false_region): Remove.
13237 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13239 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13240 mode attribute for TARGET_AVX512VL.
13242 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13244 * config/i386/i386.opt (mprefer-avx256): New option.
13245 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13247 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13248 for TARGET_PREFER_AVX256.
13249 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13251 2017-09-21 Jeff Law <law@redhat.com>
13253 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13254 Fix dump output if the only stack space is for pushed registers.
13256 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13258 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13261 2017-09-21 Martin Sebor <msebor@redhat.com>
13264 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13265 code (in C++) or code that triggers warnings.
13267 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13269 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13271 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13273 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13274 * sched-rgn.c: Ditto.
13275 * sel-sched-ir.c: Ditto.
13277 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13279 * toplev.h (set_random_seed): Adjust return type.
13280 * toplev.c (init_local_tick): Move eager initialization of random_seed
13281 to get_random_seed. Adjust comment.
13282 (init_random_seed): Inline to get_random_seed, delete.
13283 (get_random_seed): Initialize random_seed lazily.
13284 (set_random_seed): Do not return previous value.
13285 (print_switch_value): Do not call get_random_seed.
13287 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13289 * cgraph.c (delete_function_version): New, broken out from...
13290 (cgraph_node::delete_function_version): ...here. Rename to
13291 cgraph_node::delete_function_version_by_decl. Update all uses.
13292 (cgraph_node::remove): Call delete_function_version.
13294 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13297 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13298 VAR_DECLs to which addressable non-volatile parameters are mapped
13299 and for id->retvar after the return value assignment. Clear
13300 id->retval and id->retbnd after inlining.
13302 2017-09-21 Richard Biener <rguenther@suse.de>
13304 PR tree-optimization/82276
13305 PR tree-optimization/82244
13306 * tree-vrp.c (build_assert_expr_for): Set
13307 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13309 (remove_range_assertions): Revert earlier change.
13311 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13314 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13316 2017-09-21 Richard Biener <rguenther@suse.de>
13318 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13319 Restore valid IL after code generation errors.
13320 * graphite.c (graphite_transform_loops): Diagnose code
13321 generation issues as MSG_MISSED_OPTIMIZATION and continue
13322 with processing SCOPs.
13324 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13325 Alan Hayward <alan.hayward@arm.com>
13326 David Sherwood <david.sherwood@arm.com>
13328 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13329 rather than choosing between plus_constant and gen_rtx_<CODE>.
13330 * expr.c (emit_push_insn): Likewise.
13331 (expand_expr_real_2): Likewise.
13333 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13334 Alan Hayward <alan.hayward@arm.com>
13335 David Sherwood <david.sherwood@arm.com>
13337 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13339 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13340 Alan Hayward <alan.hayward@arm.com>
13341 David Sherwood <david.sherwood@arm.com>
13343 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13344 calling tree_to_uhwi.
13346 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13347 Alan Hayward <alan.hayward@arm.com>
13348 David Sherwood <david.sherwood@arm.com>
13350 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13351 INTEGER_CST rather than a negative test for ADDR_EXPR.
13353 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13354 Alan Hayward <alan.hayward@arm.com>
13355 David Sherwood <david.sherwood@arm.com>
13357 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13358 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13360 2017-09-21 Richard Biener <rguenther@suse.de>
13362 PR tree-optimization/71351
13363 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13364 graphite_create_new_loop_guard): Remove, fold remaining parts
13366 (translate_isl_ast_node_for): ... here and simplify.
13368 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13371 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13372 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13373 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13374 alternative always use QI mode, for -Os imov (=R,R) alternative
13375 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13378 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13379 Jeff Law <law@redhat.com>
13381 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13382 (allocate_stack_space): New function, partially extracted from
13383 s390_emit_prologue.
13384 (s390_emit_prologue): Track offset to most recent stack probe.
13385 Code to allocate space moved into allocate_stack_space.
13386 Dump actions when no stack is allocated.
13387 (s390_prologue_plus_offset): New function.
13388 (s390_emit_stack_probe): Likewise.
13390 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13392 * common.opt (Wa, Wl, Wp, g, gz=): Add
13394 (gno-column-info): Remove.
13395 (gcolumn-info): Drop RejectNegative.
13396 (gno-): New prefix.
13397 (gno-record-gcc-switches): Remove.
13398 (grecord-gcc-switches): Drop RejectNegative.
13399 (gno-split-dwarf): Remove.
13400 (gsplit-dwarf): Drop RejectNegative.
13401 (gno-strict-dwarf): Remove.
13402 (gstrict-dwarf): Drop RejectNegative.
13403 * config/darwin.opt (gfull, gused): Add RejectNegative.
13404 * dwarf2out.c (gen_producer_string): Drop
13405 gno-record-gcc-switches handler.
13406 * optc-gen.awk: Add g to prefixes with negative forms.
13407 * opts-common.c (remapping_prefix_p): New.
13408 (find_opt): Check it.
13409 (generate_canonical_option): Test g prefix.
13410 (option_map): Add -gno- mapping.
13411 (add_misspelling_candidates): Check remapping_prefix_p.
13413 2017-09-20 Jeff Law <law@redhat.com>
13415 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13416 thinko in stack clash protection support.
13418 * explow.c (compute_stack_clash_protection_loop_data): Use
13419 CONST_INT_P instead of explicit test. Verify object is a
13420 CONST_INT_P before looking at INTVAL.
13421 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13422 instead of explicit test.
13424 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13427 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13428 address instead of to r1 and r11.
13430 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13432 * config.gcc: Support "knm".
13433 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13434 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13436 * config/i386/i386.c (m_KNM): Define.
13437 (processor_target_table): Add "knm".
13439 (ix86_option_override_internal): Add "knm".
13440 (ix86_issue_rate): Add PROCESSOR_KNM.
13441 (ix86_adjust_cost): Ditto.
13442 (ia32_multipass_dfa_lookahead): Ditto.
13443 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13444 (fold_builtin_cpu): Add M_INTEL_KNM.
13445 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13446 (processor_type): Add PROCESSOR_KNM.
13447 * config/i386/x86-tune.def: Add m_KNM.
13448 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13450 2017-09-20 Richard Biener <rguenther@suse.de>
13452 PR tree-optimization/80213
13453 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13454 are allowed in empty BBs as well.
13455 (canonicalize_loop_closed_ssa): Also look for other complex
13457 (scop_detection::get_sese): Include the loop-closed PHI block
13459 (scop_detection::merge_sese): Remove code adding extra blocks.
13460 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13461 (build_scops): Assert the final returned scop is invalid.
13463 2017-09-20 Richard Biener <rguenther@suse.de>
13465 PR tree-optimization/82264
13466 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13468 (vn_phi_lookup): Likewise.
13469 (vn_phi_insert): Likewise.
13471 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13473 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13474 that fits into uhwi or shwi, add DW_AT_const_value regardless
13475 of early_dwarf without going through RTL, using add_AT_unsigned
13478 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13479 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13480 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13481 (DEBUG_MACRO_SECTION): Likewise.
13482 (DEBUG_LTO_MACRO_SECTION): Likewise.
13483 (DEBUG_STR_DWO_SECTION): Likewise.
13484 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13485 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13486 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13487 (DEBUG_LTO_LINE_STR_SECTION): Define.
13488 (init_sections_and_labels): Initialize debug_line_str_section
13489 variable. Initialize debug_loc_section for -gdwarf-5 to
13490 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13492 2017-09-20 Richard Biener <rguenther@suse.de>
13494 * graphite-sese-to-poly.c (extract_affine): Properly handle
13495 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13497 2017-09-20 Richard Biener <rguenther@suse.de>
13499 PR tree-optimization/81373
13500 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13501 Force SESE live-out defs to be handled even if they are
13504 2017-09-19 Jeff Law <law@redhat.com>
13506 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13507 nothing for stack adjustments with REG_STACK_CHECK.
13508 * sched-deps.c (parse_add_or_inc): Reject insns with
13509 REG_STACK_CHECK from dependency breaking.
13510 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13511 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13512 * reg-notes.def (STACK_CHECK): New note.
13514 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13515 (ix86_expand_prologue): Dump stack clash info as needed.
13516 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13518 * function.c (dump_stack_clash_frame_info): New function.
13519 * function.h (dump_stack_clash_frame_info): Prototype.
13520 (enum stack_clash_probes): New enum.
13522 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13523 flag_stack_clash_protection.
13524 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13525 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13526 (arm_frame_pointer_required): Likewise.
13527 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13528 (ia64_expand_prologue): Likewise.
13529 * config/mips/mips.c (mips_expand_prologue): Likewise.
13530 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13531 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13532 (sparc_flat_expand_prologue): Likewise.
13533 * config/spu/spu.c (spu_expand_prologue): Likewise.
13535 * explow.c: Include "params.h".
13536 (anti_adjust_stack_and_probe_stack_clash): New function.
13537 (get_stack_check_protect): Likewise.
13538 (compute_stack_clash_protection_loop_data): Likewise.
13539 (emit_stack_clash_protection_loop_start): Likewise.
13540 (emit_stack_clash_protection_loop_end): Likewise.
13541 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13542 Use anti_adjust_stack_and_probe_stack_clash.
13543 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13544 (emit_stack_clash_protection_loop_start): Likewise.
13545 (emit_stack_clash_protection_loop_end): Likewise.
13546 * rtl.h (get_stack_check_protect): Prototype.
13547 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13548 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13549 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13551 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13553 * doc/tm.texi: Rebuilt.
13554 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13555 get_stack_check_protect.
13556 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13557 * config/arm/arm.c (arm_expand_prologue): Likewise.
13558 (arm_frame_pointer_required): Likewise.
13559 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13560 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13561 * config/mips/mips.c (mips_expand_prologue): Likewise.
13562 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13563 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13564 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13565 (sparc_flat_expand_prologue): Likewise.
13567 * common.opt (-fstack-clash-protection): New option.
13568 * flag-types.h (enum stack_check_type): Note difference between
13569 -fstack-check= and -fstack-clash-protection.
13570 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13571 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13572 * toplev.c (process_options): Issue warnings/errors for cases
13573 not handled with -fstack-clash-protection.
13574 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13575 (-fstack-check): Note additional problem with -fstack-check=generic.
13576 Note that -fstack-check is primarily for Ada and refer users
13577 to -fstack-clash-protection for stack-clash-protection.
13578 Document new params for stack clash protection.
13580 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13582 * config/i386/i386.c (ix86_split_long_move): Do not handle
13583 address used for LEA in a special way.
13585 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13587 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13589 2017-09-19 Martin Sebor <msebor@redhat.com>
13592 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13593 of incompatible types.
13595 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13597 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13598 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13599 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13600 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13602 2017-09-19 Richard Biener <rguenther@suse.de>
13604 PR tree-optimization/82244
13605 * tree-vrp.c (remove_range_assertions): Do not propagate
13606 a constant to abnormals but replace the assert with a copy.
13608 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13610 PR rtl-optimization/57878
13611 PR rtl-optimization/68988
13612 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13613 avoidance test involving non_reload_pseudos. Move frequency test
13614 below the general fragmentation avoidance test.
13616 2017-09-19 Richard Biener <rguenther@suse.de>
13618 PR tree-optimization/69728
13619 * graphite-sese-to-poly.c (schedule_error): New global.
13620 (add_loop_schedule): Handle empty domain by failing the
13622 (build_original_schedule): Handle schedule_error.
13624 2017-09-19 Richard Biener <rguenther@suse.de>
13626 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13627 Do not iterate to sibling loops but only to siblings of inner
13630 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13633 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13634 that the registers are different.
13636 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13638 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13639 to processor_model and "amdfam17h" to arch_names_table.
13640 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13642 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13645 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13647 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13648 Alan Hayward <alan.hayward@arm.com>
13649 David Sherwood <david.sherwood@arm.com>
13651 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13653 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13655 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13656 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13657 vect_schedule_slp_instance.
13658 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13659 Update call to vect_slp_analyze_node_operations. Simplify return
13661 (vect_slp_analyze_bb_1): Update call accordingly.
13662 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13663 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13664 (vect_schedule_slp): Update call accordingly.
13666 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13667 Alan Hayward <alan.hayward@arm.com>
13668 David Sherwood <david.sherwood@arm.com>
13670 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13671 with types that aren't in fact scalar.
13673 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13675 * tree-vect-slp.c (vect_record_max_nunits): New function,
13677 (vect_build_slp_tree_1): ...here.
13678 (vect_build_slp_tree_2): Call it for phis too.
13680 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13682 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13683 to vect_get_vec_def_for_operand when getting the mask operand.
13685 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13686 Alan Hayward <alan.hayward@arm.com>
13687 David Sherwood <david.sherwood@arm.com>
13689 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13692 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13693 Alan Hayward <alan.hayward@arm.com>
13694 David Sherwood <david.sherwood@arm.com>
13696 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13697 calculation for vector booleans.
13699 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13700 Alan Hayward <alan.hayward@arm.com>
13701 David Sherwood <david.sherwood@arm.com>
13703 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13705 (vect_transform_stmt): ...here.
13706 (vect_analyze_stmt): Use it instead of calling
13707 vectorizable_live_operation directly.
13709 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13711 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13712 non-SIMT targets in acc vector loops.
13714 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13716 * configure.ac: Add arc and check if assembler supports gdwarf2.
13717 * configure: Regenerate.
13719 2017-09-18 Richard Biener <rguenther@suse.de>
13721 PR tree-optimization/82220
13722 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13723 epilogue niters from the min_profitable_iters compute.
13725 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13728 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13729 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13730 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13732 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13735 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13736 switching to a new text section.
13738 2017-09-18 Richard Biener <rguenther@suse.de>
13740 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13742 (build_alias_set): Reject aliases with no access function.
13744 2017-09-18 Richard Biener <rguenther@suse.de>
13746 PR tree-optimization/79622
13747 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13749 (build_cross_bb_scalars_use): Likewise.
13751 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13753 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13755 2017-09-18 Alan Modra <amodra@gmail.com>
13758 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13759 stack_pointer_rtx for count 0. Update comments. Break up
13760 large rtl expression.
13762 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13764 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13765 Increase to 20 bytes.
13766 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13767 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13768 or avx version of the stub.
13770 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13773 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13774 compute the minimum stack alignment. Also update preferred stack
13775 boundary for leaf functions.
13777 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13779 PR tree-optimization/82228
13780 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13783 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13785 * common/config/nds32/nds32-common.c
13786 (nds32_option_optimization_table): Refine formatting.
13787 (nds32_option_optimization_table): Use -fsched-pressure and
13788 -fomit-frame-pointer for specific optimization level.
13790 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13792 * config/nds32/nds32.c: Refine formatting and comments.
13793 * config/nds32/nds32.h: Likewise.
13794 * config/nds32/nds32.md: Likewise.
13795 * config/nds32/nds32-cost.c: Likewise.
13796 * config/nds32/nds32-isr.c: Likewise.
13797 * config/nds32/nds32-md-auxiliary.c: Likewise.
13798 * config/nds32/nds32-multiple.md: Likewise.
13799 * config/nds32/nds32-predicates.c: Likewise.
13801 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13802 Jakub Jelinek <jakub@redhat.com>
13804 Add support for -std=c++2a.
13805 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13807 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13809 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13812 * doc/extend.texi (Common Function Attributes): Add
13813 references to ARM, AArch64, and S/390 specific attributes.
13814 (Function Specific Option Pragmas): Add AArch64 and S/390
13815 to list of back ends that support the target pragma.
13817 2017-09-15 Nathan Sidwell <nathan@acm.org>
13819 * doc/standards.texi: Fix C++17 description. Update URLs for
13822 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13824 * common.opt (Wcast-align=strict): New warning option.
13825 * doc/invoke.texi: Document -Wcast-align=strict.
13827 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13829 * cgraph.h (cgraph_thunk_info): Add comments.
13830 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13831 assert for VIRTUAL_* arguments stricter.
13833 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13835 PR tree-optimization/71026
13836 * match.pd: Move RDIV patterns from fold-const.c
13837 * fold-const.c (distribute_real_division): Removed.
13838 (fold_binary_loc): Remove calls to distribute_real_divison.
13840 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13842 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13843 c++1z and gnu++1z as deprecated. Change other references to
13844 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13845 Change -Wc++1z-compat to -Wc++17-compat.
13846 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13847 * dwarf2out.c (highest_c_language): Handle C++17.
13848 (gen_compile_unit_die): Likewise.
13850 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13852 PR rtl-optimization/82192
13853 * combine.c (make_extraction): Don't look through non-paradoxical
13854 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13857 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13858 Alan Hayward <alan.hayward@arm.com>
13859 David Sherwood <david.sherwood@arm.com>
13861 * target.def (function_arg_offset): New hook.
13862 * targhooks.h (default_function_arg_offset): Declare.
13863 * targhooks.c (default_function_arg_offset): New function.
13864 * function.c (locate_and_pad_parm): Use
13865 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
13866 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
13867 (TARGET_FUNCTION_ARG_OFFSET): ...this.
13868 * doc/tm.texi: Regenerate.
13869 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
13870 * config/spu/spu.c (spu_function_arg_offset): New function.
13871 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
13872 * system.h (FUNCTION_ARG_OFFSET): Poison.
13874 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13875 Alan Hayard <alan.hayward@arm.com>
13876 David Sherwood <david.sherwood@arm.com>
13878 * target.def (truly_noop_truncation): New hook.
13879 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
13880 than TRULY_NOOP_TRUNCATION.
13881 * hooks.h (hook_bool_uint_uint_true): Declare.
13882 * hooks.c (hook_bool_uint_uint_true): New function.
13883 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
13884 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
13885 * doc/tm.texi: Regenerate.
13886 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
13887 rather than TRULY_NOOP_TRUNCATION in comments.
13888 (simplify_comparison): Likewise.
13889 (record_truncated_value): Likewise.
13890 * expmed.c (extract_bit_field_1): Likewise.
13891 (extract_split_bit_field): Likewise.
13892 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
13893 instead of TRULY_NOOP_TRUNCATION.
13894 * function.c (assign_parm_setup_block): Likewise.
13895 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
13896 * rtlhooks.c: Include target.h.
13897 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
13898 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
13899 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
13900 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
13901 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
13902 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
13903 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
13904 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
13905 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
13906 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
13907 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
13908 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
13909 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
13910 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
13911 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
13912 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
13913 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
13914 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
13915 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
13916 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
13917 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
13918 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
13919 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
13920 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
13921 * config/mips/mips.c (mips_truly_noop_truncation): New function.
13922 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13923 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13924 rather than TRULY_NOOP_TRUNCATION in comments.
13925 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
13926 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
13927 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
13928 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
13929 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
13930 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
13931 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
13932 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
13933 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
13934 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
13935 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
13936 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13937 rather than TRULY_NOOP_TRUNCATION in comments.
13938 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
13939 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
13940 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
13941 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
13942 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
13943 TRULY_NOOP_TRUNCATION condition.
13944 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
13945 (TRULY_NOOP_TRUNCATION): Delete.
13946 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
13947 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
13948 * config/spu/spu.c (spu_truly_noop_truncation): New function.
13949 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13950 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
13951 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
13952 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
13953 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13954 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13955 rather than TRULY_NOOP_TRUNCATION in comments.
13956 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
13957 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
13958 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
13959 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
13960 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
13961 * system.h (TRULY_NOOP_TRUNCATION): Poison.
13963 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
13966 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
13967 (*cmp_ior): Likewise.
13968 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
13969 (*ior_scc_scc_cmp): Likewise.
13970 (*and_scc_scc): Likewise.
13971 (*and_scc_scc_cmp): Likewise.
13973 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13974 Alan Hayard <alan.hayward@arm.com>
13975 David Sherwood <david.sherwood@arm.com>
13977 * target.def (can_change_mode_class): New hook.
13978 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
13979 (hard_regno_nregs): Likewise.
13980 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
13981 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
13982 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
13983 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
13984 (LOAD_EXTEND_OP): Update accordingly.
13985 * doc/tm.texi: Regenerate.
13986 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
13987 CANNOT_CHANGE_MODE_CLASS.
13988 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
13989 (REG_CAN_CHANGE_MODE_P): ...this new macro.
13990 * combine.c (simplify_set): Update accordingly.
13991 * emit-rtl.c (validate_subreg): Likewise.
13992 * recog.c (general_operand): Likewise.
13993 * regcprop.c (mode_change_ok): Likewise.
13994 * reload1.c (choose_reload_regs): Likewise.
13995 (inherit_piecemeal_p): Likewise.
13996 * rtlanal.c (simplify_subreg_regno): Likewise.
13997 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
13998 instead of CANNOT_CHANGE_MODE_CLASS.
13999 (reload_cse_simplify_operands): Likewise.
14000 * reload.c (push_reload): Use targetm.can_change_mode_class
14001 instead of CANNOT_CHANGE_MODE_CLASS.
14002 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14003 REG_CANNOT_CHANGE_MODE_P.
14004 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14005 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14006 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14007 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14008 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14009 (arm_can_change_mode_class): New function.
14010 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14011 than CANNOT_CHANGE_MODE_CLASS in comments.
14012 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14013 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14014 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14015 (ix86_can_change_mode_class): ...this new function, inverting the
14016 sense of the return value.
14017 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14018 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14019 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14020 (ia64_can_change_mode_class): New function.
14021 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14022 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14023 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14024 (m32c_can_change_mode_class): ...this new function, inverting the
14025 sense of the return value.
14026 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14027 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14028 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14029 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14030 (mips_can_change_mode_class): ...this new function, inverting the
14031 sense of the return value.
14032 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14033 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14034 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14035 (msp430_can_change_mode_class): New function.
14036 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14037 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14038 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14039 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14040 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14041 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14042 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14043 (pa_cannot_change_mode_class): Replace with...
14044 (pa_can_change_mode_class): ...this new function, inverting the
14045 sense of the return value.
14046 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14047 than CANNOT_CHANGE_MODE_CLASS in comments.
14048 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14049 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14050 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14051 (pdp11_cannot_change_mode_class): Replace with...
14052 (pdp11_can_change_mode_class): ...this new function, inverting the
14053 sense of the return value.
14054 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14055 * config/powerpcspe/powerpcspe-protos.h
14056 (rs6000_cannot_change_mode_class_ptr): Delete.
14057 * config/powerpcspe/powerpcspe.c
14058 (rs6000_cannot_change_mode_class_ptr): Delete.
14059 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14060 (rs6000_option_override_internal): Assign to
14061 targetm.can_change_mode_class instead of
14062 rs6000_cannot_change_mode_class_ptr.
14063 (rs6000_cannot_change_mode_class): Replace with...
14064 (rs6000_can_change_mode_class): ...this new function, inverting the
14065 sense of the return value.
14066 (rs6000_debug_cannot_change_mode_class): Replace with...
14067 (rs6000_debug_can_change_mode_class): ...this new function.
14068 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14069 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14070 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14071 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14072 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14074 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14075 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14076 (rs6000_option_override_internal): Assign to
14077 targetm.can_change_mode_class instead of
14078 rs6000_cannot_change_mode_class_ptr.
14079 (rs6000_cannot_change_mode_class): Replace with...
14080 (rs6000_can_change_mode_class): ...this new function, inverting the
14081 sense of the return value.
14082 (rs6000_debug_cannot_change_mode_class): Replace with...
14083 (rs6000_debug_can_change_mode_class): ...this new function.
14084 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14085 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14086 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14087 (s390_can_change_mode_class): ...this new function, inverting the
14088 sense of the return value.
14089 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14090 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14091 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14092 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14093 (sh_cannot_change_mode_class): Replace with...
14094 (sh_can_change_mode_class): ...this new function, inverting the
14095 sense of the return value.
14096 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14097 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14098 (sparc_can_change_mode_class): New function.
14099 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14100 * config/spu/spu.c (spu_can_change_mode_class): New function.
14101 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14102 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14103 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14104 (visium_can_change_mode_class): New function.
14105 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14107 2017-09-15 Richard Biener <rguenther@suse.de>
14109 PR tree-optimization/82217
14110 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14111 but not undefined case.
14113 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14116 * postreload.c (reload_cse_simplify_operands): Skip
14117 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14119 2017-09-15 Richard Biener <rguenther@suse.de>
14121 PR tree-optimization/68823
14122 * graphite-scop-detection.c (build_alias_set): If we have a
14123 possible dependence check whether we can handle them by just
14124 looking at the DRs DR_ACCESS_FNs.
14125 (build_scops): If build_alias_set fails, fail the SCOP.
14127 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14129 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14130 to support float128 built-in functions that require the ISA 3.0
14132 (BU_FLOAT128_3_HW): Likewise.
14133 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14134 built-in functions.
14135 (FMAF128): Likewise.
14137 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14138 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14139 floating point instructions.
14140 (rs6000_invalid_builtin): Likewise.
14141 (rs6000_builtin_mask_names): Likewise.
14142 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14143 (RS6000_BTM_FLOAT128_HW): Likewise.
14144 (RS6000_BTM_COMMON): Likewise.
14145 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14147 * doc/extend.texi (RS/6000 built-in functions): Document the
14148 IEEE 128-bit floating point square root and fused multiply-add
14149 built-in functions.
14151 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14153 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14154 reg (r2) isn't in the set of registers defined in the prologue.
14156 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14157 Alan Hayward <alan.hayward@arm.com>
14158 David Sherwood <david.sherwood@arm.com>
14160 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14161 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14162 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14163 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14164 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14166 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14167 max_vectorization_factor.
14168 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14170 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14171 Alan Hayward <alan.hayward@arm.com>
14172 David Sherwood <david.sherwood@arm.com>
14174 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14175 (vect_worthwhile_without_simd_p): Declare.
14176 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14177 (vectorizable_reduction): Use it.
14178 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14179 (vectorizable_operation): Likewise.
14181 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14182 Alan Hayward <alan.hayward@arm.com>
14183 David Sherwood <david.sherwood@arm.com>
14185 * tree-vectorizer.h (vect_get_num_copies): New function.
14186 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14187 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14188 (vectorizable_induction): Likewise.
14189 (vectorizable_live_operation): Likewise.
14190 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14191 (vectorizable_bswap): Likewise.
14192 (vectorizable_call): Likewise.
14193 (vectorizable_conversion): Likewise.
14194 (vectorizable_assignment): Likewise.
14195 (vectorizable_shift): Likewise.
14196 (vectorizable_operation): Likewise.
14197 (vectorizable_store): Likewise.
14198 (vectorizable_load): Likewise.
14199 (vectorizable_condition): Likewise.
14200 (vectorizable_comparison): Likewise.
14201 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14203 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14204 Alan Hayward <alan.hayward@arm.com>
14205 David Sherwood <david.sherwood@arm.com>
14207 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14208 of vect_init_vector.
14210 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14211 Alan Hayward <alan.hayward@arm.com>
14212 David Sherwood <david.sherwood@arm.com>
14214 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14215 an inline wrapper that provides a location.
14216 (gimple_build_vector): Likewise.
14217 * gimple-fold.c (gimple_build_vector_from_val): New function.
14218 (gimple_build_vector): Likewise.
14219 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14220 functions to build the initial value. Always return a gimple value.
14221 (get_initial_defs_for_reduction): Likewise. Only compute
14223 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14224 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14225 (vectorizable_induction): Use gimple_build_vector rather than
14228 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14229 Alan Hayward <alan.hayward@arm.com>
14230 David Sherwood <david.sherwood@arm.com>
14232 * target.h (vec_perm_indices): New typedef.
14233 (auto_vec_perm_indices): Likewise.
14234 * optabs-query.h: Include target.h
14235 (can_vec_perm_p): Take a vec_perm_indices *.
14236 * optabs-query.c (can_vec_perm_p): Likewise.
14237 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14238 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14239 * tree-vect-generic.c (lower_vec_perm): Likewise.
14240 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14241 (vect_grouped_load_supported): Likewise.
14242 (vect_shift_permute_load_chain): Likewise.
14243 (vect_permute_store_chain): Use auto_vec_perm_indices.
14244 (vect_permute_load_chain): Likewise.
14245 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14246 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14247 Update uses of can_vec_perm_p.
14248 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14249 mode with a number of elements. Take a vec_perm_indices *.
14250 (vect_create_epilog_for_reduction): Update accordingly.
14251 Use auto_vec_perm_indices.
14252 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14253 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14254 (vect_transform_slp_perm_load): Likewise.
14255 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14256 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14257 (vect_gen_perm_mask_checked): Likewise.
14258 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14259 (vect_gen_perm_mask_checked): Likewise.
14260 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14261 (vectorizable_store): Likewise.
14262 (vectorizable_load): Likewise.
14263 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14264 (vectorizable_bswap): Likewise.
14266 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14267 Alan Hayward <alan.hayward@arm.com>
14268 David Sherwood <david.sherwood@arm.com>
14270 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14271 * tree.c (build_vector): Likewise.
14272 (build_vector_from_ctor): Update accordingly.
14273 (build_vector_from_val): Likewise.
14274 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14275 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14276 * tree-vect-generic.c (add_rshift): Likewise.
14277 (expand_vector_divmod): Likewise.
14278 (optimize_vector_constructor): Likewise.
14279 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14280 (vect_transform_slp_perm_load): Likewise.
14281 (vect_schedule_slp_instance): Likewise.
14282 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14283 (vectorizable_call): Likewise.
14284 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14285 * expmed.c (make_tree): Likewise.
14286 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14287 a vector passed to build_vector.
14288 (fold_convert_const): Likewise.
14289 (exact_inverse): Likewise.
14290 (fold_ternary_loc): Likewise.
14291 (fold_relational_const): Likewise.
14292 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14293 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14294 (const_unop): Likewise. Store the reduction accumulator in a
14295 variable rather than an array.
14296 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14297 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14298 the new vector, rather than constructing it after the input arrays.
14299 (native_interpret_vector): Use auto_vec<tree> when building
14300 a vector passed to build_vector. Add elements in order.
14301 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14302 auto_vec<tree> when building a vector passed to build_vector.
14303 (vect_create_epilog_for_reduction): Likewise.
14304 (vectorizable_induction): Likewise.
14305 (get_initial_def_for_reduction): Likewise. Fix indentation of
14307 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14309 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14310 passed to build_vector.
14312 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14313 Alan Hayward <alan.hayward@arm.com>
14314 David Sherwood <david.sherwood@arm.com>
14316 * tree-core.h (tree_base::u): Add an "nelts" field.
14317 (tree_vector): Use VECTOR_CST_NELTS as the length.
14318 * tree.c (tree_size): Likewise.
14319 (make_vector): Initialize VECTOR_CST_NELTS.
14320 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14321 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14322 TYPE_VECTOR_SUBPARTS.
14323 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14324 as the number of units, setting it from VECTOR_CST_NELTS.
14325 (const_vector_from_tree): Likewise.
14326 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14327 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14328 (fold_negate_expr_1): Likewise.
14329 (fold_convert_const): Likewise.
14330 (const_binop): Likewise. Differentiate the number of output and
14332 (const_unop): Likewise.
14333 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14334 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14335 in cases that did the opposite.
14337 2017-09-14 Richard Biener <rguenther@suse.de>
14339 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14342 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14344 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14346 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14349 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14350 if and where to split a bb, except for splitting before debug insn
14351 sequences followed by non-label real insn. Delete debug insns
14352 in between basic blocks.
14354 * combine.c (make_compound_operation_int): Formatting fixes.
14356 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14357 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14358 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14359 * config/sol2.h (LINK_EH_SPEC): Likewise.
14360 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14361 * config/s390/linux.h (LINK_SPEC): Likewise.
14362 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14363 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14364 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14365 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14366 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14367 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14368 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14369 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14371 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14373 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14375 (ENDFILE_LINUX_SPEC): Likewise.
14376 (LINK_EH_SPEC): Likewise.
14377 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14378 (LINK_OS_LINUX_SPEC32): Likewise.
14379 (LINK_OS_LINUX_SPEC64): Likewise.
14380 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14381 (LINK_OS_LINUX_SPEC): Likewise.
14383 2017-09-13 Martin Liska <mliska@suse.cz>
14385 PR middle-end/82154
14386 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14387 CASE_HIGH is NULL_TREE.
14389 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14390 Alan Hayward <alan.hayward@arm.com>
14391 David Sherwood <david.sherwood@arm.com>
14393 * target.def (secondary_memory_needed): New hook.
14394 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14395 instead of SECONDARY_MEMORY_NEEDED.
14396 (secondary_memory_needed_mode): Likewise.
14397 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14398 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14399 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14400 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14401 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14402 * doc/tm.texi: Regenerate.
14403 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14404 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14405 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14406 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14407 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14408 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14409 mode argument first and change the reg_class arguments to reg_class_t.
14410 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14411 Make static. Update the call to inline_secondary_memory_needed.
14412 (ix86_register_move_cost): Update the call to
14413 inline_secondary_memory_needed.
14414 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14415 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14417 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14418 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14420 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14421 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14422 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14423 and match hook interface. Add comment from mips.h.
14424 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14425 * config/mmix/mmix.md (truncdfsf2): Refer to
14426 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14428 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14429 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14430 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14431 (pa_secondary_memory_needed): New function.
14432 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14433 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14434 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14435 (pdp11_secondary_memory_needed): Make static and match hook interface.
14436 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14437 * config/powerpcspe/powerpcspe-protos.h
14438 (rs6000_secondary_memory_needed_ptr): Delete.
14439 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14441 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14442 (rs6000_option_override_internal): Assign to
14443 targetm.secondary_memory_needed rather than
14444 rs6000_secondary_memory_needed_ptr.
14445 (rs6000_secondary_memory_needed): Match hook interface.
14446 (rs6000_debug_secondary_memory_needed): Likewise.
14447 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14448 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14449 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14450 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14451 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14452 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14454 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14455 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14456 (rs6000_option_override_internal): Assign to
14457 targetm.secondary_memory_needed rather than
14458 rs6000_secondary_memory_needed_ptr.
14459 (rs6000_secondary_memory_needed): Match hook interface.
14460 (rs6000_debug_secondary_memory_needed): Likewise.
14461 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14462 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14463 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14464 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14465 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14466 (sparc_secondary_memory_needed): New function.
14467 * lra-constraints.c (check_and_process_move): Refer to
14468 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14470 (curr_insn_transform): Likewise.
14471 (process_alt_operands): Use targetm.secondary_memory_needed
14472 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14473 (check_secondary_memory_needed_p): Likewise.
14474 (choose_split_class): Likewise.
14475 * reload.c: Unconditionally include code that was previously
14476 conditional on SECONDARY_MEMORY_NEEDED.
14477 (push_secondary_reload): Use targetm.secondary_memory_needed
14478 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14479 (push_reload): Likewise.
14480 * reload1.c: Unconditionally include code that was previously
14481 conditional on SECONDARY_MEMORY_NEEDED.
14482 (choose_reload_regs): Use targetm.secondary_memory_needed
14483 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14484 (gen_reload): Likewise.
14485 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14487 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14488 Alan Hayward <alan.hayward@arm.com>
14489 David Sherwood <david.sherwood@arm.com>
14491 * target.def (secondary_memory_needed_mode): New hook:
14492 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14493 * targhooks.h (default_secondary_memory_needed_mode): New function.
14494 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14495 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14496 * doc/tm.texi: Regenerate.
14497 * lra-constraints.c (check_and_process_move): Use
14498 targetm.secondary_memory_needed_mode instead of
14499 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14500 (curr_insn_transform): Likewise.
14501 * reload.c (get_secondary_mem): Likewise.
14502 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14503 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14505 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14506 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14507 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14508 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14509 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14511 * config/powerpcspe/powerpcspe-protos.h
14512 (rs6000_secondary_memory_needed_mode): Delete.
14513 * config/powerpcspe/powerpcspe.c
14514 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14515 (rs6000_secondary_memory_needed_mode): Make static.
14516 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14517 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14519 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14521 (rs6000_secondary_memory_needed_mode): Make static.
14522 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14523 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14524 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14525 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14526 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14528 (sparc_secondary_memory_needed_mode): New function.
14529 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14531 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14533 * config/aarch64/constraints.md (Umq): New constraint.
14534 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14536 (mov<mode>): Update condition.
14538 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14540 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14541 when bitposition is the same.
14543 2017-09-13 Richard Biener <rguenther@suse.de>
14545 * dwarf2out.c (output_die_symbol): Remove.
14546 (output_die): Do not output a DIEs symbol.
14548 2017-09-13 Richard Biener <rguenther@suse.de>
14550 PR middle-end/82128
14551 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14552 default-def to avoid breaking iterator update with the weird
14553 interaction with cgraph_update_edges_for_call_stmt_node.
14555 2017-09-13 Richard Biener <rguenther@suse.de>
14557 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14558 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14559 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14560 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14562 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14564 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14565 Disable pc relative literal load irrespective of
14566 TARGET_FIX_ERR_A53_84341 for default.
14568 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14570 * config/sparc/sparc.c (output_return): Output the source location of
14571 the insn in the delay slot, if any.
14572 (output_sibcall): Likewise.
14574 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14577 * common.opt (-static-pie): New alias.
14578 (shared): Negate static-pie.
14579 (-no-pie): Update help text.
14581 (static-pie): New option.
14582 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14583 -static-pie support.
14584 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14585 (LINK_EH_SPEC): Likewise.
14586 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14587 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14588 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14589 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14590 (init_gcc_specs): Likewise.
14591 (init_spec): Likewise.
14592 (display_help): Update help message for -pie.
14593 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14596 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14598 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14599 (movdi_aarch64): Likewise.
14600 (movti_aarch64): Likewise.
14602 2017-09-12 Simon Wright <simon@pushface.org>
14605 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14606 calculation of the minor version, always output as 0.
14608 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14611 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14612 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14613 on it early, rather than manual conversion late. For
14614 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14615 instead of performing manual conversion.
14617 2017-09-12 Carl Love <cel@us.ibm.com>
14619 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14620 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14622 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14623 VMULOSW): Add definitions.
14624 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14625 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14626 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14627 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14628 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14630 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14632 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14634 (movti_aarch64): Likewise.
14635 (movdf_aarch64): Likewise.
14636 (movtf_aarch64): Likewise.
14637 (load_pairdi): Likewise.
14638 (store_pairdi): Likewise.
14639 (load_pairdf): Likewise.
14640 (store_pairdf): Likewise.
14641 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14642 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14643 (ldr_got_small_<mode>): Likewise.
14644 (ldr_got_small_28k_<mode>): Likewise.
14645 (ldr_got_tiny): Likewise.
14646 * config/aarch64/iterators.md (ldst_sz): New.
14647 (ldpstp_sz): Likewise.
14648 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14650 (thunderx_load): Split load_8 to load_16.
14651 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14653 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14654 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14655 (xgene1_store_pair): Split store_8 to store_16.
14656 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14657 (falkor_st_0_st_sd): Split store_8 to store_16.
14659 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14661 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14662 and store1/2/3/4 to store_4/8/12/16.
14663 * config/aarch64/aarch64.md: Update for rename.
14664 * config/arm/arm.md: Likewise.: Likewise.
14665 * config/arm/arm.c: Likewise.
14666 * config/arm/thumb1.md: Likewise.
14667 * config/arm/thumb2.md: Likewise.
14668 * config/arm/vfp.md: Likewise.
14669 * config/arm/arm-generic.md: Likewise.
14670 * config/arm/arm1020e.md: Likewise.
14671 * config/arm/arm1026ejs.md: Likewise.
14672 * config/arm/arm1136jfs.md: Likewise.
14673 * config/arm/arm926ejs.md: Likewise.
14674 * config/arm/cortex-a15.md: Likewise.
14675 * config/arm/cortex-a17.md: Likewise.
14676 * config/arm/cortex-a5.md: Likewise.
14677 * config/arm/cortex-a53.md: Likewise.
14678 * config/arm/cortex-a57.md: Likewise.
14679 * config/arm/cortex-a7.md: Likewise.
14680 * config/arm/cortex-a8.md: Likewise.
14681 * config/arm/cortex-a9.md: Likewise.
14682 * config/arm/cortex-m4.md: Likewise.
14683 * config/arm/cortex-m7.md: Likewise.
14684 * config/arm/cortex-r4.md: Likewise.
14685 * config/arm/exynos-m1.md: Likewise.
14686 * config/arm/fa526.md: Likewise.
14687 * config/arm/fa606te.md: Likewise.
14688 * config/arm/fa626te.md: Likewise.
14689 * config/arm/fa726te.md: Likewise.
14690 * config/arm/fmp626.md: Likewise.
14691 * config/arm/iwmmxt.md: Likewise.
14692 * config/arm/ldmstm.md: Likewise.
14693 * config/arm/marvell-pj4.md: Likewise.
14694 * config/arm/xgene1.md: Likewise.
14695 * config/aarch64/thunderx.md: Likewise.
14696 * config/aarch64/thunderx2t99.md: Likewise.
14697 * config/aarch64/falkor.md: Likewise.
14699 2017-09-12 Martin Liska <mliska@suse.cz>
14701 * attribs.c (private_lookup_attribute): New function.
14702 * attribs.h (private_lookup_attribute): Declared here.
14703 (lookup_attribute): Called from this place.
14705 2017-09-12 Richard Biener <rguenther@suse.de>
14707 PR tree-optimization/82157
14708 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14709 stmts with side-effects.
14711 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14712 Alan Hayward <alan.hayward@arm.com>
14713 David Sherwood <david.sherwood@arm.com>
14715 * target.def (hard_regno_nregs): New hook.
14716 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14717 * targhooks.h (default_hard_regno_nregs): Declare.
14718 * targhooks.c (default_hard_regno_nregs): New function.
14719 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14720 (TARGET_HARD_REGNO_NREGS): ...this hook.
14721 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14722 (CLASS_MAX_NREGS): Likewise.
14723 * doc/tm.texi: Regenerate.
14724 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14725 instead of HARD_REGNO_NREGS.
14726 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14727 HARD_REGNO_NREGS in the comment.
14728 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14729 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14730 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14731 Return an unsigned int.
14732 (TARGET_HARD_REGNO_NREGS): Redefine.
14733 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14734 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14735 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14736 (arc_hard_regno_nregs): New function.
14737 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14738 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14739 (arm_hard_regno_nregs): New function.
14740 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14741 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14742 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14743 (TARGET_HARD_REGNO_NREGS): Redefine.
14744 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14745 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14746 (HARD_REGNO_NREGS): Delete.
14747 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14748 (cr16_hard_regno_nregs): New function.
14749 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14750 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14751 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14752 (cris_hard_regno_nregs): New function.
14753 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14754 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14755 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14756 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14757 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14758 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14759 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14760 (frv_hard_regno_nregs): Make static. Take and return an
14762 (frv_class_max_nregs): Remove outdated copy of documentation.
14763 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14764 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14765 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14766 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14767 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14768 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14769 (TARGET_HARD_REGNO_NREGS): Redefine.
14770 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14771 (CLASS_MAX_NREGS): Update comment.
14772 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14773 (ia64_hard_regno_nregs): New function.
14774 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14775 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14776 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14777 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14778 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14780 (m32c_hard_regno_nregs): Likewise. Make static.
14781 (TARGET_HARD_REGNO_NREGS): Redefine.
14782 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14783 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14784 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14785 (m68k_hard_regno_nregs): New function.
14786 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14787 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14788 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14789 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14790 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14791 Take and return an unsigned int.
14792 (TARGET_HARD_REGNO_NREGS): Redefine.
14793 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14794 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14795 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14796 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14797 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14798 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14799 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14800 (msp430_hard_regno_nregs): Make static. Take and return an
14802 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14803 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14804 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14805 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14806 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14807 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14808 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14809 (TARGET_HARD_REGNO_NREGS): Redefine.
14810 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14811 (PA_HARD_REGNO_NREGS): ...this.
14812 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14813 (PA_HARD_REGNO_NREGS): ...this.
14814 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14815 (pa_hard_regno_nregs): New function.
14816 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14817 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14818 (pdp11_hard_regno_nregs): New function.
14819 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14820 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14821 (rs6000_hard_regno_nregs_hook): New function.
14822 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14823 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14824 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14825 Take and return an unsigned int. Move earlier in file.
14826 (TARGET_HARD_REGNO_NREGS): Redefine.
14827 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14828 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14829 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14830 (rl78_hard_regno_nregs): Make static. Take and return an
14832 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14833 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14834 (rs6000_hard_regno_nregs_hook): New function.
14835 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14836 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14837 (TARGET_HARD_REGNO_NREGS): Redefine.
14838 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14839 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14840 instead of HARD_REGNO_NREGS.
14841 (s390_hard_regno_nregs): New function.
14842 (s390_hard_regno_mode_ok): Add comment from s390.h.
14843 (TARGET_HARD_REGNO_NREGS): Redefine.
14844 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14845 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14846 (sh_hard_regno_nregs): New function.
14847 (sh_pass_in_reg_p): Use it.
14848 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14849 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14850 (sparc_hard_regno_nregs): New function.
14851 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14852 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14853 (spu_function_arg_advance): Use it, supplying a valid register number.
14854 (TARGET_HARD_REGNO_NREGS): Redefine.
14855 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14856 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14857 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14858 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
14859 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
14860 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
14861 (CLASS_MAX_NREGS): Remove copy of old documentation.
14862 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
14863 (visium_hard_regno_nregs): New function.
14864 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
14865 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
14866 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14867 xtensa_hard_regno_nregs): New function.
14868 * system.h (HARD_REGNO_NREGS): Poison.
14870 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14872 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
14873 hard_regno_nregs instead of HARD_REGNO_NREGS.
14874 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
14875 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
14876 (c6x_expand_epilogue): Likewise.
14877 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
14878 (frv_read_iacc_argument): Likewise.
14879 * config/sh/sh.c: Include regs.h.
14880 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14881 (regs_used): Likewise.
14882 (output_stack_adjust): Likewise.
14883 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
14884 * expmed.c: Include regs.h.
14885 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14886 * ree.c: Include regs.h.
14887 (combine_reaching_defs): Use hard_regno_nregs instead of
14889 (add_removable_extension): Likewise.
14891 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14893 * regs.h (hard_regno_nregs): Turn into a function.
14894 (end_hard_regno): Update accordingly.
14895 * caller-save.c (setup_save_areas): Likewise.
14896 (save_call_clobbered_regs): Likewise.
14897 (replace_reg_with_saved_mem): Likewise.
14898 (insert_restore): Likewise.
14899 (insert_save): Likewise.
14900 * combine.c (can_change_dest_mode): Likewise.
14901 (move_deaths): Likewise.
14902 (distribute_notes): Likewise.
14903 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
14904 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
14905 (rs6000_split_multireg_move): Likewise.
14906 (rs6000_register_move_cost): Likewise.
14907 (rs6000_memory_move_cost): Likewise.
14908 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
14909 (rs6000_split_multireg_move): Likewise.
14910 (rs6000_register_move_cost): Likewise.
14911 (rs6000_memory_move_cost): Likewise.
14912 * cselib.c (cselib_reset_table): Likewise.
14913 (cselib_lookup_1): Likewise.
14914 * emit-rtl.c (set_mode_and_regno): Likewise.
14915 * function.c (aggregate_value_p): Likewise.
14916 * ira-color.c (setup_profitable_hard_regs): Likewise.
14917 (check_hard_reg_p): Likewise.
14918 (calculate_saved_nregs): Likewise.
14919 (assign_hard_reg): Likewise.
14920 (improve_allocation): Likewise.
14921 (calculate_spill_cost): Likewise.
14922 * ira-emit.c (modify_move_list): Likewise.
14923 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
14924 (ira_hard_reg_in_set_p): Likewise.
14925 * ira.c (setup_reg_mode_hard_regset): Likewise.
14926 (clarify_prohibited_class_mode_regs): Likewise.
14927 (check_allocation): Likewise.
14928 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14929 (lra_setup_reg_renumber): Likewise.
14930 (setup_try_hard_regno_pseudos): Likewise.
14931 (spill_for): Likewise.
14932 (assign_hard_regno): Likewise.
14933 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14934 * lra-constraints.c (in_class_p): Likewise.
14935 (lra_constraint_offset): Likewise.
14936 (simplify_operand_subreg): Likewise.
14937 (lra_constraints): Likewise.
14938 (split_reg): Likewise.
14939 (split_if_necessary): Likewise.
14940 (invariant_p): Likewise.
14941 (inherit_in_ebb): Likewise.
14942 * lra-lives.c (process_bb_lives): Likewise.
14943 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
14944 (get_hard_regs): Likewise.
14945 (do_remat): Likewise.
14946 * lra-spills.c (assign_spill_hard_regs): Likewise.
14947 * mode-switching.c (create_pre_exit): Likewise.
14948 * postreload.c (reload_combine_recognize_pattern): Likewise.
14949 * recog.c (peep2_find_free_register): Likewise.
14950 * regcprop.c (kill_value_regno): Likewise.
14951 (set_value_regno): Likewise.
14952 (copy_value): Likewise.
14953 (maybe_mode_change): Likewise.
14954 (find_oldest_value_reg): Likewise.
14955 (copyprop_hardreg_forward_1): Likewise.
14956 * regrename.c (check_new_reg_p): Likewise.
14957 (regrename_do_replace): Likewise.
14958 * reload.c (push_reload): Likewise.
14959 (combine_reloads): Likewise.
14960 (find_dummy_reload): Likewise.
14961 (operands_match_p): Likewise.
14962 (find_reloads): Likewise.
14963 (find_equiv_reg): Likewise.
14964 (reload_adjust_reg_for_mode): Likewise.
14965 * reload1.c (count_pseudo): Likewise.
14966 (count_spilled_pseudo): Likewise.
14967 (find_reg): Likewise.
14968 (clear_reload_reg_in_use): Likewise.
14969 (free_for_value_p): Likewise.
14970 (allocate_reload_reg): Likewise.
14971 (choose_reload_regs): Likewise.
14972 (reload_adjust_reg_for_temp): Likewise.
14973 (emit_reload_insns): Likewise.
14974 (delete_output_reload): Likewise.
14975 * rtlanal.c (subreg_get_info): Likewise.
14976 * sched-deps.c (sched_analyze_reg): Likewise.
14977 * sel-sched.c (init_regs_for_mode): Likewise.
14978 (mark_unavailable_hard_regs): Likewise.
14979 (choose_best_reg_1): Likewise.
14980 (verify_target_availability): Likewise.
14981 * valtrack.c (dead_debug_insert_temp): Likewise.
14982 * var-tracking.c (track_loc_p): Likewise.
14983 (emit_note_insn_var_location): Likewise.
14984 * varasm.c (make_decl_rtl): Likewise.
14985 * reginfo.c (choose_hard_reg_mode): Likewise.
14986 (init_reg_modes_target): Refer directly to
14987 this_target_regs->x_hard_regno_nregs.
14989 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14991 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
14992 instead of hard_regno_nregs.
14994 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14996 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
14997 end_hard_regno instead of hard_regno_nregs.
14998 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
14999 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15000 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15001 * ira-color.c (improve_allocation): Likewise.
15002 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15003 * lra-lives.c (mark_regno_live): Likewise.
15004 (mark_regno_dead): Likewise.
15005 * lra-remat.c (operand_to_remat): Likewise.
15006 * lra.c (collect_non_operand_hard_regs): Likewise.
15007 * postreload.c (reload_combine_note_store): Likewise.
15008 (move2add_valid_value_p): Likewise.
15009 * reload.c (regno_clobbered_p): Likewise.
15011 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15013 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15015 * config/v850/v850.c (v850_reorg): Likewise.
15016 * reload.c (refers_to_regno_for_reload_p): Likewise.
15017 (find_equiv_reg): Likewise.
15018 * reload1.c (reload_reg_reaches_end_p): Likewise.
15020 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15022 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15024 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15025 * config/arm/arm.c (output_move_neon): Likewise.
15026 (arm_attr_length_move_neon): Likewise.
15027 (neon_split_vcombine): Likewise.
15028 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15029 (c6x_mark_reg_written): Likewise.
15030 (c6x_dwarf_register_span): Likewise.
15031 * config/i386/i386.c (ix86_save_reg): Likewise.
15032 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15033 (rws_access_reg): Likewise.
15034 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15035 * mode-switching.c (create_pre_exit): Likewise.
15036 * ree.c (combine_reaching_defs): Likewise.
15037 (add_removable_extension): Likewise.
15038 * regcprop.c (find_oldest_value_reg): Likewise.
15039 (copyprop_hardreg_forward_1): Likewise.
15040 * reload.c (reload_inner_reg_of_subreg): Likewise.
15041 (push_reload): Likewise.
15042 (combine_reloads): Likewise.
15043 (find_dummy_reload): Likewise.
15044 (reload_adjust_reg_for_mode): Likewise.
15045 * reload1.c (find_reload_regs): Likewise.
15046 (forget_old_reloads_1): Likewise.
15047 (reload_reg_free_for_value_p): Likewise.
15048 (reload_adjust_reg_for_temp): Likewise.
15049 (emit_reload_insns): Likewise.
15050 (delete_output_reload): Likewise.
15051 * sel-sched.c (choose_best_reg_1): Likewise.
15052 (choose_best_pseudo_reg): Likewise.
15054 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15055 Alan Hayward <alan.hayward@arm.com>
15056 David Sherwood <david.sherwood@arm.com>
15058 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15059 * target.def (slow_unaligned_access): New hook.
15060 * targhooks.h (default_slow_unaligned_access): Declare.
15061 * targhooks.c (default_slow_unaligned_access): New function.
15062 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15063 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15064 * doc/tm.texi: Regenerate.
15065 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15066 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15067 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15069 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15070 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15072 (rs6000_slow_unaligned_access): New function.
15073 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15074 (expand_block_compare): Likewise.
15075 (expand_strn_compare): Likewise.
15076 (rs6000_rtx_costs): Likewise.
15077 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15078 (riscv_slow_unaligned_access): Likewise.
15079 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15080 (riscv_slow_unaligned_access_p): ...this and make static.
15081 (riscv_option_override): Update accordingly.
15082 (riscv_slow_unaligned_access): New function.
15083 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15084 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15085 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15086 (rs6000_slow_unaligned_access): New function.
15087 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15088 (rs6000_rtx_costs): Likewise.
15089 * config/rs6000/rs6000-string.c (expand_block_compare)
15090 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15091 of SLOW_UNALIGNED_ACCESS.
15092 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15093 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15094 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15095 of SLOW_UNALIGNED_ACCESS.
15096 * expmed.c (simple_mem_bitfield_p): Likewise.
15097 * expr.c (alignment_for_piecewise_move): Likewise.
15098 (emit_group_load_1): Likewise.
15099 (emit_group_store): Likewise.
15100 (copy_blkmode_from_reg): Likewise.
15101 (emit_push_insn): Likewise.
15102 (expand_assignment): Likewise.
15103 (store_field): Likewise.
15104 (expand_expr_real_1): Likewise.
15105 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15106 * lra-constraints.c (simplify_operand_subreg): Likewise.
15107 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15108 * gimple-ssa-store-merging.c: Likewise in block comment at start
15110 * tree-ssa-strlen.c: Include target.h.
15111 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15112 of SLOW_UNALIGNED_ACCESS.
15113 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15115 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15117 PR rtl-optimization/82185
15118 * expmed.c (emit_store_flag_int): Only test tem if it has been
15121 2017-09-12 Richard Biener <rguenther@suse.de>
15123 PR middle-end/82149
15124 * match.pd ((FTYPE) N CMP CST): Fix typo.
15126 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15128 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15130 (mips_near_type_p): Add 'short_call' attribute as a synonym
15132 * doc/extend.texi (short_call): Document new function attribute.
15134 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15137 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15138 assertion check that in the condition.
15139 (get_atomic_generic_size): Likewise. Before testing if parameter
15140 has pointer type, if it has array type, call for C++
15141 default_conversion to perform array-to-pointer conversion.
15143 2017-09-12 Richard Biener <rguenther@suse.de>
15145 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15146 for operations we cannot scalarize.
15148 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15150 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15151 vectors heap vectors. Clean up comments.
15152 Make visited_bbs a reference.
15153 (profitable_jump_thread_path): Make GC
15154 vectors heap vectors. Clean up comments.
15156 (convert_and_register_jump_thread_path): Make GC vectors heap
15158 (check_subpath_and_update_thread_path): Same. Clean up comments.
15159 Make visited_bbs a reference.
15160 (handle_phi): Abstract common code to to
15161 register_jump_thread_path_if_profitable.
15162 Rename VAR_BB to DEF_BB.
15164 Make GC vectors heap vectors.
15165 Make visited_bbs a reference.
15166 (handle_assignment): Same.
15167 (register_jump_thread_path_if_profitable): New.
15168 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15170 Make GC vectors heap vectors. Clean up comments.
15171 Make visited_bbs a reference.
15172 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15173 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15176 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15179 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15180 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15182 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15184 Revert r251800 and r251799.
15186 2017-09-11 Martin Jambor <mjambor@suse.cz>
15189 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15190 arguments in advance.
15191 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15192 use it to find predecessor edges.
15193 (naive_outof_ssa): Collect vector of predecessors.
15195 2017-09-08 Jason Merrill <jason@redhat.com>
15197 PR c++/70029 - ICE with ref-qualifier and -flto
15198 * langhooks.h (struct lang_hooks_for_types): Add
15199 copy_lang_qualifiers.
15200 * attribs.c (build_type_attribute_qual_variant): Use it.
15201 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15203 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15204 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15206 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15209 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15210 (*mulsi3_sp64): New instruction.
15211 (mulsi3): New expander.
15213 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15215 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15217 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15219 * sancov.c: Include memmodel.h.
15221 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15224 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15227 2017-09-07 Carl Love <cel@us.ibm.com>
15229 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15230 the sldi instruction.
15232 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15234 * sancov.c: Include tm_p.h.
15236 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15239 * output.h (switch_to_other_text_partition): New declaration.
15240 * varasm.c (switch_to_other_text_partition): New function.
15241 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15242 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15243 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15244 to the other text partition before emitting LCL label and switch back
15245 after emitting the word after it.
15247 2017-09-07 Richard Biener <rguenther@suse.de>
15249 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15250 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15251 critical edges here, after loop init.
15252 (pass_data_pre): Remove PROP_no_crit_edges flags.
15253 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15254 for valueization of call args to avoid leaking VN_TOP.
15255 (visit_use): Assert we do not visit default defs.
15256 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15257 Use error_mark_node to more easily detect leaking VN_TOP.
15258 All default-defs are varying, not VN_TOP. Mark them visited.
15259 (run_scc_vn): Make code match comment.
15261 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15263 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15264 OPTION_MASK_FLOAT128_KEYWORD.
15265 (POWERPC_MASKS): Likewise.
15266 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15267 support for the -mfloat128-type option, and make -mfloat128
15268 default on PowerPC Linux systems. Define or undefine
15269 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15270 Define __float128 to be __ieee128 if IEEE 128-bit support is
15271 enabled, or undefine it.
15272 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15273 Delete defining __FLOAT128_TYPE__.
15274 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15275 -mfloat128-type option and make -mfloat128 default on PowerPC
15277 (TARGET_FLOAT128_TYPE): Likewise.
15278 (-mfloat128-type): Likewise.
15279 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15280 Delete the -mfloat128-type option and make -mfloat128 default on
15281 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15282 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15283 128-bit floating point is enabled. Change tests from using
15284 -mfloat128-type to -mfloat128.
15285 (rs6000_mangle_type): Use the correct mangling for the __float128
15286 type even if normal long double is restricted to 64-bits.
15287 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15289 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15290 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15291 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15293 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15294 documentation for -mfloat128.
15296 2017-09-06 Olivier Hainque <hainque@adacore.com>
15298 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15300 2017-09-06 Wish Wu <wishwu007@gmail.com>
15301 Jakub Jelinek <jakub@redhat.com>
15303 * asan.c (initialize_sanitizer_builtins): Add
15304 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15305 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15306 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15307 BT_FN_VOID_UINT64_PTR variables.
15308 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15309 (BT_FN_VOID_UINT16_UINT16): Likewise.
15310 (BT_FN_VOID_UINT32_UINT32): Likewise.
15311 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15312 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15313 (BT_FN_VOID_UINT64_PTR): Likewise.
15314 * common.opt (flag_sanitize_coverage): New variable.
15315 (fsanitize-coverage=trace-pc): Remove.
15316 (fsanitize-coverage=): Add.
15317 * flag-types.h (enum sanitize_coverage_code): New enum.
15318 * fold-const.c (fold_range_test): Disable non-short-circuit
15319 optimization if flag_sanitize_coverage.
15320 (fold_truth_andor): Likewise.
15321 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15322 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15323 (coverage_sanitizer_opts): New array.
15324 (get_closest_sanitizer_option): Add OPTS argument, handle also
15325 OPT_fsanitize_coverage_.
15326 (parse_sanitizer_options): Adjusted to also handle
15327 OPT_fsanitize_coverage_.
15328 (common_handle_option): Add OPT_fsanitize_coverage_.
15329 * sancov.c (instrument_comparison, instrument_switch): New function.
15330 (sancov_pass): Add trace-cmp support.
15331 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15332 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15333 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15334 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15335 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15336 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15337 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15338 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15339 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15340 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15342 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15344 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15345 error. Only quit immediately if parsing is complete.
15346 (BEGIN): Initialize fatal_err and parse_done.
15347 (begin fpu, end fpu): Check number of arguments.
15348 (begin arch, end arch): Likewise.
15349 (begin cpu, end cpu): Likewise.
15350 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15351 (optalias): Likewise.
15353 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15355 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15356 * config/arm/arm-isa.h: Delete. Move definitions to ...
15357 * arm-cpus.in: ... here. Use new feature and fgroup values.
15358 * config/arm/arm.c (arm_option_override): Use lower case for feature
15360 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15361 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15362 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15364 (print_isa_bits_for): New function.
15365 (gen_isa): New function.
15366 (gen_comm_data): Use print_isa_bits_for.
15367 (define feature): New keyword.
15368 (define fgroup): New keyword.
15369 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15370 (arm-isa.h): Add rule to generate file.
15371 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15372 case for feature bit names.
15374 2017-09-06 Richard Biener <rguenther@suse.de>
15376 * tree-ssa-pre.c (NECESSARY): Remove.
15377 (create_expression_by_pieces): Do not touch pass-local flags.
15378 (insert_into_preds_of_block): Likewise.
15379 (do_pre_regular_insertion): Likewise.
15380 (eliminate_insert): Likewise.
15381 (eliminate_dom_walker::before_dom_children): Likewise.
15382 (fini_eliminate): Do not look at inserted_exprs.
15383 (mark_operand_necessary): Remove.
15384 (remove_dead_inserted_code): Replace with simple work-list
15385 algorithm based on inserted_exprs and SSA uses.
15386 (pass_pre::execute): Re-order fini_eliminate and
15387 remove_dead_inserted_code.
15389 2017-09-06 Olivier Hainque <hainque@adacore.com>
15391 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15392 for VxWorks 7. Adjust surrounding comments.
15394 2017-09-06 Richard Biener <rguenther@suse.de>
15396 * gimple-ssa-strength-reduction.c
15397 (find_candidates_dom_walker::before_dom_children): Also allow
15400 2017-09-06 Richard Biener <rguenther@suse.de>
15402 PR tree-optimization/82108
15403 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15404 for gap in the non-permutation SLP case.
15406 2017-09-06 Martin Jambor <mjambor@suse.cz>
15408 PR tree-optimization/82078
15409 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15410 add_access_to_work_queue...
15411 (build_accesses_from_assign): ...here.
15412 (propagate_all_subaccesses): Make sure racc is the group
15413 representative, if there is one.
15415 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15417 PR middle-end/82095
15418 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15421 2017-09-06 Richard Biener <rguenther@suse.de>
15423 * gimple-ssa-strength-reduction.c
15424 (find_candidates_dom_walker::before_doom_children): Use a
15425 type and not a mode check.
15427 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15430 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15432 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15434 2017-09-05 Jeff Law <law@redhat.com>
15436 PR tree-optimization/64910
15437 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15438 cases where we have 3 or more operands.
15440 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15442 PR middle-end/81768
15443 * omp-low.c (lower_omp_for): Recompute tree invariant if
15444 gimple_omp_for_initial/final is ADDR_EXPR.
15446 PR middle-end/81768
15447 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15448 into gimple val before gimplification fo the COND_EXPR.
15450 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15452 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15453 REGION_COPY argument.
15454 (thread_through_all_blocks): Remove unused argument to
15455 duplicate_thread_path.
15457 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15458 Alan Hayward <alan.hayward@arm.com>
15459 David Sherwood <david.sherwood@arm.com>
15461 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15462 Take a scalar_mode rather than a machine_mode.
15463 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15464 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15465 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15466 (aarch64_gen_adjusted_ldpstp): Likewise.
15467 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15469 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15470 Alan Hayward <alan.hayward@arm.com>
15471 David Sherwood <david.sherwood@arm.com>
15473 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15474 Take a scalar_int_mode instead of a machine_mode.
15475 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15476 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15477 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15478 (aarch64_simd_attr_length_rglist): Delete.
15479 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15480 a scalar_int_mode instead of a machine_mode.
15481 (aarch64_add_offset): Likewise.
15482 (aarch64_internal_mov_immediate): Likewise
15483 (aarch64_add_constant_internal): Likewise.
15484 (aarch64_add_constant): Likewise.
15485 (aarch64_movw_imm): Likewise.
15486 (aarch64_rtx_arith_op_extract_p): Likewise.
15487 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15488 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15489 Remove assert that the mode isn't a vector.
15490 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15491 (aarch64_expand_mov_immediate): Update calls after above changes.
15492 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15493 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15494 (aarch64_move_imm): Likewise.
15495 (aarch64_can_const_movi_rtx_p): Likewise.
15496 (aarch64_strip_extend): Likewise.
15497 (aarch64_extr_rtx_p): Likewise.
15498 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15499 a CONST_INT when the mode parameter is VOIDmode.
15500 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15502 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15504 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15505 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15506 (bitwise_type_for_mode): Update accordingly.
15508 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15510 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15511 * stor-layout.c (mode_for_size_tree): Likewise.
15512 (mode_for_array): Update accordingly.
15513 (layout_decl): Likewise.
15514 (compute_record_mode): Likewise. Only set the mode once.
15516 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15518 * target.def (get_mask_mode): Change return type to opt_mode.
15520 * doc/tm.texi: Regenerate.
15521 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15522 * targhooks.c (default_get_mask_mode): Likewise.
15523 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15524 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15525 targetm.get_mask_mode.
15526 * tree.c (build_truth_vector_type): Likewise.
15528 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15530 * machmode.h (mode_for_vector): Return an opt_mode.
15531 * stor-layout.c (mode_for_vector): Likewise.
15532 (mode_for_int_vector): Update accordingly.
15533 (layout_type): Likewise.
15534 * config/i386/i386.c (emit_memmov): Likewise.
15535 (ix86_expand_set_or_movmem): Likewise.
15536 (ix86_expand_vector_init): Likewise.
15537 (ix86_get_mask_mode): Likewise.
15538 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15540 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15541 * expmed.c (extract_bit_field_1): Likewise.
15542 * expr.c (expand_expr_real_2): Likewise.
15543 * optabs-query.c (can_vec_perm_p): Likewise.
15544 (can_vec_mask_load_store_p): Likewise.
15545 * optabs.c (expand_vec_perm): Likewise.
15546 * targhooks.c (default_get_mask_mode): Likewise.
15547 * tree-vect-stmts.c (vectorizable_store): Likewise.
15548 (vectorizable_load): Likewise.
15549 (get_vectype_for_scalar_type_and_size): Likewise.
15551 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15553 * machmode.h (mode_for_int_vector): New function.
15554 * stor-layout.c (mode_for_int_vector): Likewise.
15555 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15556 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15557 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15558 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15559 (s390_expand_vcond): Likewise.
15561 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15563 * machmode.h (opt_machine_mode): New type.
15564 (opt_mode<T>): Allow construction from anything that can be
15566 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15567 (mode_for_size): Return an opt_machine_mode.
15568 * stor-layout.c (mode_for_size): Likewise.
15569 (mode_for_size_tree): Update call accordingly.
15570 (bitwise_mode_for_mode): Likewise.
15571 (make_fract_type): Likewise.
15572 (make_accum_type): Likewise.
15573 * caller-save.c (replace_reg_with_saved_mem): Update call
15575 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15576 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15577 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15578 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15579 * expmed.c (extract_bit_field_1): Likewise.
15580 * reload.c (get_secondary_mem): Likewise.
15581 * varasm.c (assemble_integer): Likewise.
15582 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15585 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15587 * machmode.h (decimal_float_mode_for_size): New function.
15588 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15589 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15590 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15591 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15592 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15593 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15595 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15597 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15598 (get_builtin_sync_mode): Likewise.
15599 (expand_ifn_atomic_compare_exchange): Likewise.
15600 (expand_builtin_atomic_clear): Likewise.
15601 (expand_builtin_atomic_test_and_set): Likewise.
15602 (fold_builtin_atomic_always_lock_free): Likewise.
15603 * calls.c (compute_argument_addresses): Likewise.
15604 (emit_library_call_value_1): Likewise.
15605 (store_one_arg): Likewise.
15606 * combine.c (combine_instructions): Likewise.
15607 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15608 * config/arm/arm.c (arm_function_value): Likewise.
15609 (aapcs_allocate_return_reg): Likewise.
15610 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15611 * config/i386/i386.c (construct_container): Likewise.
15612 (ix86_gimplify_va_arg): Likewise.
15613 (ix86_expand_sse_cmp): Likewise.
15614 (emit_memmov): Likewise.
15615 (emit_memset): Likewise.
15616 (expand_small_movmem_or_setmem): Likewise.
15617 (ix86_expand_pextr): Likewise.
15618 (ix86_expand_pinsr): Likewise.
15619 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15620 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15622 * config/mips/mips.c (mips_function_value_1) Likewise.
15623 (mips_block_move_straight): Likewise.
15624 (mips_expand_ins_as_unaligned_store): Likewise.
15625 * config/powerpcspe/powerpcspe.c
15626 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15627 (rs6000_darwin64_record_arg_flush): Likewise.
15628 * config/rs6000/rs6000.c
15629 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15630 (rs6000_darwin64_record_arg_flush): Likewise.
15631 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15632 (sparc_function_value_1): Likewise.
15633 * config/spu/spu.c (adjust_operand): Likewise.
15634 (spu_emit_branch_or_set): Likewise.
15635 (arith_immediate_p): Likewise.
15636 * emit-rtl.c (gen_lowpart_common): Likewise.
15637 * expr.c (expand_expr_real_1): Likewise.
15638 * function.c (assign_parm_setup_block): Likewise.
15639 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15640 * reload1.c (alter_reg): Likewise.
15641 * stor-layout.c (mode_for_vector): Likewise.
15642 (layout_type): Likewise.
15644 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15646 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15647 (spu_convert_move): Likewise.
15648 * lower-subreg.c (resolve_simple_move): Likewise.
15650 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15653 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15654 define_insn to a define_expand.
15655 (altivec_vsum2sws_direct): New define_insn.
15656 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15658 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15660 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15663 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15668 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15669 realign stack if stack alignment needed is less than incoming
15672 2017-09-05 Marek Polacek <polacek@redhat.com>
15675 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15678 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15680 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15682 2017-09-05 Richard Biener <rguenther@suse.de>
15684 PR tree-optimization/82084
15685 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15687 2017-09-05 Richard Biener <rguenther@suse.de>
15689 PR tree-optimization/82102
15690 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15692 2017-09-05 Martin Liska <mliska@suse.cz>
15694 PR tree-optimization/82032
15695 * tree-cfg.c (generate_range_test): New function.
15696 * tree-cfg.h (generate_range_test): Declared here.
15697 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15698 (cleanup_control_expr_graph): Use it.
15699 * tree-switch-conversion.c (try_switch_expansion): Remove
15701 (emit_case_nodes): Use generate_range_test.
15703 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15706 * config/i386/i386.md (*<btsc><mode>_mask): Add
15707 TARGET_USE_BT to insn constraint.
15708 (*btr<mode>_mask): Ditto.
15710 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15712 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15713 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15715 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15718 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15719 TARGET_NEON and TARGET_IWMMXT.
15720 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15721 TARGET_NEON and TARGET_IWMMXT.
15722 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15724 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15726 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15727 (ix86_rewrite_tls_address): Ditto.
15728 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15729 (ix86_rewrite_tls_address_1): Ditto.
15730 (ix86_rewrite_tls_address): Ditto.
15731 * config/i386/predicates.md (tls_address_pattern): New predicate.
15732 * config/i386/i386.md (TLS address splitter): New splitter.
15734 2017-09-04 Richard Biener <rguenther@suse.de>
15736 PR tree-optimization/82084
15737 * fold-const.h (can_native_encode_string_p): Declare.
15738 * fold-const.c (can_native_encode_string_p): Factor out from ...
15739 (native_encode_string): ... here.
15740 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15741 vectorizing stores from constants we later cannot handle.
15743 2017-09-04 Marek Polacek <polacek@redhat.com>
15746 * doc/invoke.texi: Update -Wtautological-compare documentation.
15748 2017-09-04 Jeff Law <law@redhat.com>
15750 PR tree-optimization/64910
15751 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15752 swap the first and last operand if the last is a constant.
15754 2017-09-04 Marek Polacek <polacek@redhat.com>
15757 * convert.c (do_narrow): When sanitizing signed integer overflows,
15758 bail out for signed types.
15759 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15761 2017-09-04 Richard Biener <rguenther@suse.de>
15763 PR tree-optimization/82060
15764 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15765 Move devirtualization after stmt folding and before EH/AB/noreturn
15766 cleanup to get the stmt refs canonicalized. Use a bool instead
15767 of gimple_modified_p since that doesn't work for NOPs. Schedule
15768 NOPs generated by folding for removal.
15770 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15771 Alan Hayward <alan.hayward@arm.com>
15772 David Sherwood <david.sherwood@arm.com>
15774 * coretypes.h (pad_direction): New enum.
15775 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15776 (FUNCTION_ARG_PADDING): Likewise.
15777 * target.def (function_arg_padding): New hook.
15778 * targhooks.h (default_function_arg_padding): Declare.
15779 * targhooks.c (default_function_arg_padding): New function.
15780 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15781 (TARGET_FUNCTION_ARG_PADDING): ...this.
15782 * doc/tm.texi: Regenerate.
15783 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15784 instead of direction.
15785 (compute_argument_addresses): Likewise.
15786 (load_register_parameters): Likewise.
15787 (emit_library_call_value_1): Likewise.
15788 (store_one_arg): Use targetm.calls.function_arg_padding instead
15789 of FUNCTION_ARG_PADDING.
15790 (must_pass_in_stack_var_size_or_pad): Likewise.
15791 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15792 (emit_group_store): Likewise.
15793 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15794 instead of FUNCTION_ARG_PADDING.
15795 (emit_push_insn): Likewise, and propagate enum change throughout
15797 * function.h (direction): Delete.
15798 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15800 * function.c (assign_parm_find_stack_rtl): Likewise.
15801 (assign_parm_setup_block_p): Likewise.
15802 (assign_parm_setup_block): Likewise.
15803 (gimplify_parameters): Likewise.
15804 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15805 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15807 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15808 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15809 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15810 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15811 (aarch64_function_arg_padding): ...this new function.
15812 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15813 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15814 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15815 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15816 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15817 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15818 (arm_pad_arg_upward): Replace with...
15819 (arm_function_arg_padding): ...this new function.
15820 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15822 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15823 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15824 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15825 (ia64_hpux_function_arg_padding): Replace with...
15826 (ia64_function_arg_padding): ...this new function. Use pad_direction
15827 instead of direction. Check for TARGET_HPUX.
15828 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15829 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15830 (iq2000_function_arg_padding): New function.
15831 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15832 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15833 (mips_function_arg_padding): ...this new function.
15834 (mips_pad_reg_upward): Update accordingly.
15835 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15836 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15837 targetm.calls.function_arg_padding.
15838 (FUNCTION_ARG_PADDING): Delete.
15839 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15840 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15841 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15842 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15843 (nios2_block_reg_padding): Return pad_direction instead of direction.
15844 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15845 instead of direction.
15846 (nios2_function_arg_padding): Likewise. Make static.
15847 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15848 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15849 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15850 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15851 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15852 (pa_function_arg_padding): Make static. Return pad_direction instead
15854 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15855 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15856 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15857 instead of direction. Use targetm.calls.function_arg_padding.
15858 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
15859 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15860 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
15861 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
15862 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
15864 (function_arg_padding): Rename to...
15865 (rs6000_function_arg_padding): ...this. Make static. Return
15866 pad_direction instead of direction.
15867 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15868 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
15869 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15870 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
15871 instead of direction. Use targetm.calls.function_arg_padding.
15872 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
15873 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15874 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
15875 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
15876 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15877 (function_arg_padding): Rename to...
15878 (rs6000_function_arg_padding): ...this. Make static. Return
15879 pad_direction instead of direction.
15880 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15881 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
15882 * config/s390/s390.c (s390_function_arg_padding): New function.
15883 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15884 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
15885 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
15886 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15887 (function_arg_padding): Rename to...
15888 (sparc_function_arg_padding): ...this. Make static. Return
15889 pad_direction instead of direction.
15890 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
15891 * config/spu/spu.c (spu_function_arg_padding): New function.
15892 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15893 * system.h (FUNCTION_ARG_PADDING): Poison.
15895 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15896 Alan Hayward <alan.hayward@arm.com>
15897 David Sherwood <david.sherwood@arm.com>
15899 * target.def (modes_tieable_p): New hook.
15900 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
15901 (TARGET_MODES_TIEABLE_P): ...this.
15902 * doc/tm.texi.in: Regenerate.
15903 * hooks.h (hook_bool_mode_mode_true): Declare.
15904 * hooks.c (hook_bool_mode_mode_true): New function.
15905 * combine.c (subst): Use targetm.modes_tieable_p instead of
15907 * dse.c (find_shift_sequence): Likewise.
15908 * expmed.c (extract_low_bits): Likewise.
15909 * lower-subreg.c: Include target.h.
15910 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
15912 * rtlanal.c (rtx_cost): Likewise.
15913 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
15914 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
15915 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
15916 (TARGET_MODES_TIEABLE_P): Redefine.
15917 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
15918 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
15919 (TARGET_MODES_TIEABLE_P): Redefine.
15920 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
15921 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
15922 (arc_modes_tieable_p): New function.
15923 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
15924 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
15925 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
15926 (arm_modes_tieable_p): Make static.
15927 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
15928 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
15929 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
15930 (TARGET_MODES_TIEABLE_P): Redefine.
15931 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
15932 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
15933 (TARGET_MODES_TIEABLE_P): Redefine.
15934 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
15935 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
15936 (cr16_modes_tieable_p): New function.
15937 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
15938 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
15939 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
15940 (TRULY_NOOP_TRUNCATION): Update comment.
15941 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
15942 (TRULY_NOOP_TRUNCATION): Update comment.
15943 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
15944 (frv_modes_tieable_p): New function.
15945 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
15946 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
15947 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
15948 (TARGET_MODES_TIEABLE_P): Redefine.
15949 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
15950 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
15951 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
15952 (TARGET_MODES_TIEABLE_P): Redefine.
15953 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
15954 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
15955 (ia64_modes_tieable_p): New function.
15956 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
15957 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
15958 (iq2000_modes_tieable_p): New function.
15959 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
15960 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
15961 (lm32_modes_tieable_p): New function.
15962 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
15963 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
15964 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
15965 (TARGET_MODES_TIEABLE_P): Redefine.
15966 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
15967 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
15968 (m32r_modes_tieable_p): New function.
15969 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
15970 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
15971 (m68k_modes_tieable_p): New function.
15972 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
15973 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
15974 (mcore_modes_tieable_p): New function.
15975 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
15976 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
15978 (TARGET_MODES_TIEABLE_P): Redefine.
15979 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
15980 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
15981 * config/mips/mips.c (mips_modes_tieable_p): Make static.
15982 (TARGET_MODES_TIEABLE_P): Redefine.
15983 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
15984 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
15985 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
15986 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
15987 (mn10300_modes_tieable_p): ...this and make static.
15988 (TARGET_MODES_TIEABLE_P): Redefine.
15989 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
15990 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
15991 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
15992 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
15993 (msp430_modes_tieable_p): Make static.
15994 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
15995 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
15996 (TARGET_MODES_TIEABLE_P): Redefine.
15997 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
15998 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
15999 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16000 (TARGET_MODES_TIEABLE_P): Redefine.
16001 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16002 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16003 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16004 (TARGET_MODES_TIEABLE_P): Redefine.
16005 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16006 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16007 (pdp11_modes_tieable_p): New function.
16008 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16009 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16010 (rs6000_modes_tieable_p): New function.
16011 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16012 * config/powerpcspe/powerpcspe.md: Update comment.
16013 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16014 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16015 (TARGET_MODES_TIEABLE_P): Redefine.
16016 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16017 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16018 (rl78_modes_tieable_p): New function.
16019 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16020 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16021 (rs6000_modes_tieable_p): New function.
16022 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16023 * config/rs6000/rs6000.md: Update comment.
16024 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16025 * config/rx/rx.c (rx_modes_tieable_p): New function.
16026 (TARGET_MODES_TIEABLE_P): Redefine.
16027 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16028 * config/s390/s390.c (s390_modes_tieable_p): New function.
16029 (TARGET_MODES_TIEABLE_P): Redefine.
16030 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16031 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16032 (sh_modes_tieable_p): New function.
16033 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16034 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16035 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16036 (sparc_modes_tieable_p): Make static.
16037 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16038 * config/spu/spu.c (spu_modes_tieable_p): New function.
16039 (TARGET_MODES_TIEABLE_P): Redefine.
16040 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16041 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16042 (TARGET_MODES_TIEABLE_P): Redefine.
16043 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16044 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16045 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16046 * config/v850/v850.c (v850_modes_tieable_p): New function.
16047 (TARGET_MODES_TIEABLE_P): Redefine.
16048 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16049 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16050 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16051 (visium_modes_tieable_p): New function.
16052 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16053 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16054 (xtensa_modes_tieable_p): New function.
16055 * system.h (MODES_TIEABLE_P): Poison.
16057 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16058 Alan Hayward <alan.hayward@arm.com>
16059 David Sherwood <david.sherwood@arm.com>
16061 * target.def (hard_regno_mode_ok): New hook.
16062 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16063 (TARGET_HARD_REGNO_MODE_OK): ...this.
16064 * doc/tm.texi.in: Regenerate.
16065 * hooks.h (hook_bool_uint_mode_true): Declare.
16066 * hooks.c (hook_bool_uint_mode_true): New function.
16067 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16068 HARD_REGNO_MODE_OK.
16069 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16070 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16071 instead of HARD_REGNO_MODE_OK.
16072 * caller-save.c: Include target.h.
16073 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16074 HARD_REGNO_MODE_OK.
16075 * combine.c (can_combine_p): Likewise.
16076 (combinable_i3pat): Likewise.
16077 (can_change_dest_mode): Likewise.
16078 * expr.c (init_expr_target): Likewise.
16079 (convert_move): Likewise.
16080 (convert_modes): Likewise.
16081 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16082 (setup_prohibited_mode_move_regs): Likewise.
16083 * ira.h (target_ira): Likewise.
16084 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16085 * lra-constraints.c (process_alt_operands): Likewise.
16086 (split_reg): Likewise.
16087 * recog.c (peep2_find_free_register): Likewise.
16088 * ree.c (combine_reaching_defs): Likewise.
16089 * regcprop.c (maybe_mode_change): Likewise.
16090 * reginfo.c (init_reg_sets_1): Likewise.
16091 (choose_hard_reg_mode): Likewise.
16092 (simplifiable_subregs): Likewise.
16093 * regrename.c (check_new_reg_p): Likewise.
16094 * reload.c (find_valid_class): Likewise.
16095 (find_valid_class_1): Likewise.
16096 (reload_inner_reg_of_subreg): Likewise.
16097 (push_reload): Likewise.
16098 (combine_reloads): Likewise.
16099 (find_dummy_reload): Likewise.
16100 (find_reloads): Likewise.
16101 * reload1.c (find_reg): Likewise.
16102 (set_reload_reg): Likewise.
16103 (allocate_reload_reg): Likewise.
16104 (choose_reload_regs): Likewise.
16105 (reload_adjust_reg_for_temp): Likewise.
16106 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16107 (simplify_subreg_regno): Likewise.
16108 * sel-sched.c (init_regs_for_mode): Likewise.
16109 * varasm.c (make_decl_rtl): Likewise.
16110 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16111 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16112 HARD_REGNO_MODE_OK.
16113 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16114 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16115 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16116 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16117 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16118 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16119 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16120 (arc_mode_class): Delete.
16121 (HARD_REGNO_MODE_OK): Delete.
16122 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16123 (arc_hard_regno_mode_ok): Rename old array to...
16124 (arc_hard_regno_mode_ok_modes): ...this.
16125 (arc_conditional_register_usage): Update accordingly.
16126 (arc_mode_class): Make static.
16127 (arc_hard_regno_mode_ok): New function.
16128 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16129 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16130 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16131 (arm_hard_regno_mode_ok): Make static.
16132 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16133 HARD_REGNO_MODE_OK.
16134 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16135 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16136 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16138 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16139 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16140 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16141 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16142 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16143 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16144 * config/bfin/predicates.md (valid_reg_operand): Use
16145 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16146 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16147 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16148 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16149 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16150 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16151 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16152 (cr16_hard_regno_mode_ok): Make static and return a bool.
16153 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16154 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16155 (cris_hard_regno_mode_ok): New function.
16156 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16157 (epiphany_mode_class): Delete.
16158 (HARD_REGNO_MODE_OK): Delete.
16159 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16160 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16161 (hard_regno_mode_ok): Rename to...
16162 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16163 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16164 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16165 HARD_REGNO_MODE_OK.
16166 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16167 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16168 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16169 (frv_hard_regno_mode_ok): Make static and return a bool.
16170 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16171 HARD_REGNO_MODE_OK.
16172 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16173 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16174 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16175 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16177 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16178 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16179 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16180 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16182 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16183 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16184 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16185 (ia64_hard_regno_mode_ok): New function.
16186 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16187 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16188 (iq2000_hard_regno_mode_ok): New function.
16189 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16190 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16191 (lm32_hard_regno_mode_ok): New function.
16192 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16193 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16194 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16195 instead of HARD_REGNO_MODE_OK.
16196 (m32c_hard_regno_ok): Rename to...
16197 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16198 (m32c_cannot_change_mode_class): Update accordingly.
16199 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16200 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16201 (m32r_mode_class): Delete.
16202 (HARD_REGNO_MODE_OK): Delete.
16203 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16204 (m32r_hard_regno_mode_ok): Rename to...
16205 (m32r_hard_regno_modes): ...this.
16206 (m32r_mode_class): Make static.
16207 (m32r_hard_regno_mode_ok): New function.
16208 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16209 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16210 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16211 (m68k_hard_regno_mode_ok): Make static.
16212 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16213 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16214 (mcore_hard_regno_mode_ok): New function.
16215 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16216 (HARD_REGNO_MODE_OK): Delete.
16217 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16219 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16220 (microblaze_hard_regno_mode_ok): New function.
16221 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16222 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16223 (mips_hard_regno_mode_ok): Delete.
16224 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16225 (mips_hard_regno_mode_ok_p): ...this and make static.
16226 (mips_hard_regno_mode_ok_p): Rename to...
16227 (mips_hard_regno_mode_ok_uncached): ...this.
16228 (mips_hard_regno_mode_ok): New function.
16229 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16230 of HARD_REGNO_MODE_OK.
16231 (mips_option_override): Update after above name changes.
16232 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16233 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16234 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16235 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16236 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16237 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16238 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16239 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16240 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16241 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16242 (msp430_hard_regno_mode_ok): Make static and return a bool.
16243 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16244 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16245 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16247 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16248 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16249 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16250 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16251 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16252 (PA_HARD_REGNO_MODE_OK): ...this
16253 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16254 (PA_HARD_REGNO_MODE_OK): ...this.
16255 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16256 (pa_hard_regno_mode_ok): New function.
16257 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16258 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16259 (pdp11_hard_regno_mode_ok): New function.
16260 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16261 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16263 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16265 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16266 (rs6000_hard_regno_mode_ok): Rename to...
16267 (rs6000_hard_regno_mode_ok_uncached): ...this.
16268 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16269 (rs6000_hard_regno_mode_ok): New function.
16270 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16271 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16272 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16273 (riscv_hard_regno_mode_ok): ...this and make static.
16274 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16275 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16276 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16277 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16278 (rl78_hard_regno_mode_ok): Make static and return bool.
16279 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16280 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16282 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16283 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16284 (rs6000_hard_regno_mode_ok): Rename to...
16285 (rs6000_hard_regno_mode_ok_uncached): ...this.
16286 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16287 (rs6000_hard_regno_mode_ok): New function.
16288 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16289 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16290 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16291 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16292 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16293 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16294 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16295 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16296 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16297 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16298 (sh_hard_regno_mode_ok): Make static.
16299 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16300 instead of HARD_REGNO_MODE_OK.
16301 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16302 (sparc_mode_class): Delete.
16303 (HARD_REGNO_MODE_OK): Delete.
16304 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16305 (hard_regno_mode_classes): Make static.
16306 (sparc_mode_class): Likewise.
16307 (sparc_hard_regno_mode_ok): New function.
16308 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16309 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16310 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16312 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16313 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16314 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16315 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16316 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16317 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16318 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16319 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16320 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16321 (visium_hard_regno_mode_ok): New function.
16322 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16323 instead of HARD_REGNO_MODE_OK.
16324 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16325 (HARD_REGNO_MODE_OK): Delete.
16326 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16327 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16328 (xtensa_option_override): Update accordingly.
16329 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16330 (xtensa_hard_regno_mode_ok): New function.
16331 * system.h (HARD_REGNO_MODE_OK): Poison.
16333 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16334 Alan Hayward <alan.hayward@arm.com>
16335 David Sherwood <david.sherwood@arm.com>
16337 * target.def (hard_regno_call_part_clobbered): New hook.
16338 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16339 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16340 * doc/tm.texi: Regenerate.
16341 * hooks.h (hook_bool_uint_mode_false): Declare.
16342 * hooks.c (hook_bool_uint_mode_false): New function.
16343 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16344 * cselib.c (cselib_process_insn): Use
16345 targetm.hard_regno_call_part_clobbered instead of
16346 HARD_REGNO_CALL_PART_CLOBBERED.
16347 * ira-conflicts.c (ira_build_conflicts): Likewise.
16348 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16349 * lra-constraints.c (need_for_call_save_p): Likewise.
16350 * lra-lives.c: Include target.h.
16351 (check_pseudos_live_through_calls): Use
16352 targetm.hard_regno_call_part_clobbered instead of
16353 HARD_REGNO_CALL_PART_CLOBBERED.
16354 * regcprop.c: Include target.h.
16355 (copyprop_hardreg_forward_1): Use
16356 targetm.hard_regno_call_part_clobbered instead of
16357 HARD_REGNO_CALL_PART_CLOBBERED.
16358 * reginfo.c (choose_hard_reg_mode): Likewise.
16359 * regrename.c (check_new_reg_p): Likewise.
16360 * reload.c (find_equiv_reg): Likewise.
16361 * reload1.c (emit_reload_insns): Likewise.
16362 * sched-deps.c (deps_analyze_insn): Likewise.
16363 * sel-sched.c (init_regs_for_mode): Likewise.
16364 (mark_unavailable_hard_regs): Likewise.
16365 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16366 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16367 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16369 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16370 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16371 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16373 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16375 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16376 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16377 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16379 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16380 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16381 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16383 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16384 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16386 * config/powerpcspe/powerpcspe.c
16387 (rs6000_hard_regno_call_part_clobbered): New function.
16388 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16389 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16390 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16392 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16393 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16394 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16396 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16397 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16398 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16400 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16401 Alan Hayward <alan.hayward@arm.com>
16402 David Sherwood <david.sherwood@arm.com>
16404 * rtl.h (subreg_memory_offset): Declare.
16405 * emit-rtl.c (subreg_memory_offset): New function.
16406 * expmed.c (store_bit_field_1): Use it.
16407 * expr.c (undefined_operand_subword_p): Likewise.
16408 * simplify-rtx.c (simplify_subreg): Likewise.
16410 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16412 PR rtl-optimization/57448
16415 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16416 using atomic_load pattern.
16417 (expand_atomic_store): Likewise.
16419 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16422 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16423 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16424 handling. Use replace_call_with_value with NULL instead of
16425 gsi_replace, unlink_stmt_vdef and release_defs.
16427 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16430 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16432 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16435 * rtl.h (emit_library_call_value_1): Declare.
16436 (emit_library_call): Replace declaration with a series of overloads.
16437 Remove the parameter count argument.
16438 (emit_library_call_value): Likewise.
16439 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16440 with an "rtx_mode_t *".
16441 (emit_library_call_value): Delete.
16442 (emit_library_call): Likewise.
16443 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16444 (asan_emit_allocas_unpoison): Likewise.
16445 * builtins.c (expand_builtin_powi): Likewise.
16446 (expand_asan_emit_allocas_unpoison): Likewise.
16447 * cfgexpand.c (expand_main_function): Likewise.
16448 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16449 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16450 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16451 * config/arm/arm.c (arm_trampoline_init): Likewise.
16452 (arm_call_tls_get_addr): Likewise.
16453 (arm_expand_divmod_libfunc): Likewise.
16454 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16455 (smulsi3_highpart): Likewise.
16456 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16457 (c6x_expand_compare): Likewise.
16458 (c6x_expand_movmem): Likewise.
16459 * config/frv/frv.c (frv_trampoline_init): Likewise.
16460 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16461 (ix86_expand_divmod_libfunc): Likewise.
16462 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16463 (ia64_expand_compare): Likewise.
16464 (ia64_profile_hook): Likewise.
16465 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16466 (nonlocal_goto): Likewise.
16467 (restore_stack_nonlocal): Likewise.
16468 * config/m32r/m32r.c (block_move_call): Likewise.
16469 (m32r_trampoline_init): Likewise.
16470 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16471 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16472 (m68k_call_m68k_read_tp): Likewise.
16473 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16474 (microblaze_expand_divide): Likewise.
16475 * config/mips/mips.h (mips_args): Likewise.
16476 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16477 (MIPS_ICACHE_SYNC): Likewise.
16478 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16479 (nios2_trampoline_init): Likewise.
16480 * config/pa/pa.c (hppa_tls_call): Likewise.
16481 (pa_trampoline_init): Likewise.
16482 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16483 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16484 (expand_strn_compare): Likewise.
16485 (rs6000_generate_compare): Likewise.
16486 (rs6000_expand_float128_convert): Likewise.
16487 (output_profile_hook): Likewise.
16488 (rs6000_trampoline_init): Likewise.
16489 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16490 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16491 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16492 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16493 (rs6000_generate_compare): Likewise.
16494 (rs6000_expand_float128_convert): Likewise.
16495 (output_profile_hook): Likewise.
16496 (rs6000_trampoline_init): Likewise.
16497 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16498 * config/sh/sh.c (sh_trampoline_init): Likewise.
16499 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16500 (sparc_emit_float_lib_cmp): Likewise.
16501 (sparc32_initialize_trampoline): Likewise.
16502 (sparc64_initialize_trampoline): Likewise.
16503 (sparc_profile_hook): Likewise.
16504 * config/spu/spu.c (ea_load_store): Likewise.
16505 * config/spu/spu.md (floatunssidf2): Likewise.
16506 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16507 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16508 * config/visium/visium.c (expand_block_move_4): Likewise.
16509 (expand_block_move_2): Likewise.
16510 (expand_block_move_1): Likewise.
16511 (expand_block_set_4): Likewise.
16512 (expand_block_set_2): Likewise.
16513 (expand_block_set_1): Likewise.
16514 (visium_trampoline_init): Likewise.
16515 (visium_profile_hook): Likewise.
16516 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16517 (xtensa_setup_frame_addresses): Likewise.
16518 (xtensa_trampoline_init): Likewise.
16519 * except.c (sjlj_emit_function_enter): Likewise.
16520 (sjlj_emit_function_exit): Likewise.
16521 * explow.c (allocate_dynamic_stack_space): Likewise.
16522 (probe_stack_range): Likewise.
16523 * expr.c (convert_mode_scalar): Likewise.
16524 * optabs.c (expand_binop): Likewise.
16525 (expand_twoval_binop_libfunc): Likewise.
16526 (expand_unop): Likewise.
16527 (prepare_cmp_insn): Likewise.
16528 (prepare_float_lib_cmp): Likewise.
16529 (expand_float): Likewise.
16530 (expand_fix): Likewise.
16531 (expand_fixed_convert): Likewise.
16532 (maybe_emit_sync_lock_test_and_set): Likewise.
16533 (expand_atomic_compare_and_swap): Likewise.
16534 (expand_mem_thread_fence): Likewise.
16535 (expand_atomic_fetch_op): Likewise.
16537 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16539 * doc/generic.texi (OpenACC): Adjust URL.
16540 * doc/invoke.texi (C Dialect Options): Ditto.
16542 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16544 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16545 predicate for operand 1. Add (m,<S>) constraint.
16546 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16547 Prevent memory operand 1 with register operand 2.
16549 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16551 PR rtl-optimization/82024
16552 * combine.c (try_combine): If the combination result is a PARALLEL,
16553 and we only need to retain the SET in there that would be placed
16554 at I2, check that we can place that at I3 instead, before doing so.
16556 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16559 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16561 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16562 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16565 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16566 Jeff Law <law@redhat.com>
16568 * varasm.c (bss_initializer_p): Do not put constants into .bss
16569 (categorize_decl_for_section): Handle bss_initializer_p returning
16570 false when DECL_INITIAL is NULL.
16572 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16575 * config/s390/s390.c (s390_can_inline_p): New function.
16577 2017-09-01 Jeff Law <law@redhat.com>
16579 PR tree-optimization/82052
16580 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16581 Always initialize the returned slot after a hash table miss
16582 when INSERT is true.
16584 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16586 * config/s390/s390.md (mem_signal_fence): Remove.
16587 * doc/md.texi (mem_signal_fence): Remove.
16588 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16590 * target-insns.def (mem_signal_fence): Remove.
16592 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16595 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16598 * asan.c (create_odr_indicator): Strip name encoding from assembler
16599 name before appending it after __odr_asan_.
16601 2017-09-01 Martin Liska <mliska@suse.cz>
16603 PR tree-optimization/82059
16604 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16605 frequency only when an edge is redirected.
16607 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16609 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16610 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16611 (arc_conditional_register_usage): Remove ARC600 lp_count
16613 (arc_file_start): Emit Tag_ARC_CPU_variation.
16614 (arc_can_use_doloop_p): New conditions to use ZOLs.
16615 (hwloop_fail): New function.
16616 (hwloop_optimize): Likewise.
16617 (hwloop_pattern_reg): Likewise.
16618 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16619 (arc_reorg_loops): New function, calls reorg_loops.
16620 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16621 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16623 (arc_loop_hazard): Remove function, functionality moved into
16625 (arc_hazard): Remove arc_loop_hazard call.
16626 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16627 into hwloop_optimize.
16628 (arc_label_align): Remove ZOL handling.
16629 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16630 * config/arc/arc.md (doloop_begin): Remove pattern.
16631 (doloop_begin_i): Likewise.
16632 (doloop_end_i): Likewise.
16633 (doloop_fallback): Likewise.
16634 (doloop_fallback_m): Likewise.
16635 (doloop_end): Reimplement expand.
16636 (arc_lp): New pattern for LP instruction.
16637 (loop_end): New pattern.
16638 (loop_fail): Likewise.
16639 (decrement_and_branch_until_zero): Likewise.
16640 * config/arc/arc.opt (mlpc-width): New option.
16641 * doc/invoke.texi (mlpc-width): Document option.
16643 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16645 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16646 (arc_ccfsm_advance): Fix checking for delay slots.
16647 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16649 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16651 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16653 (movhi_insn): Update store instruction constraint which are saving
16654 6-bit short immediates.
16655 (movsi_insn): Consider also short scaled load operations.
16656 (zero_extendhisi2_i): Use Usd constraint instead of T.
16657 (extendhisi2_i): Add q constraint.
16658 (arc_clzsi2): Add type and length attributes.
16659 (arc_ctzsi2): Likewise.
16660 * config/arc/constraints.md (Usc): Update constraint, the
16661 assembler can parse two relocations for a single instruction.
16663 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16665 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16666 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16668 2017-08-31 Olivier Hainque <hainque@adacore.com>
16670 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16671 match as powerpc-wrs-vxworks*.
16673 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16675 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16676 register constraint for by-element operand.
16677 (aarch64_mls_elt_merge<mode>): Likewise.
16679 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16681 * config/arc/arc.c (arc_can_follow_jump): Check for short
16684 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16686 * config.gcc: Use g.opt for arc.
16687 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16688 functionality moved to ...
16689 (legitimate_scaled_address_p): New function, ...here.
16690 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16691 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16692 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16694 (arc_override_options): Handle G option.
16695 (arc_output_pic_addr_const): Correct function definition.
16696 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16697 (arc_decl_anon_ns_mem_p): Delete.
16698 (arc_in_small_data_p): Overhaul this function to take into
16699 consideration the value given via G option.
16700 (arc_rewrite_small_data_1): Renamed and corrected old
16701 arc_rewrite_small_data function.
16702 (arc_rewrite_small_data): New function.
16703 (small_data_pattern): Don't use pic_offset_table_rtx.
16704 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16705 * config/arc/simdext.md (movmisalignv2hi): Use
16706 prepare_move_operands function.
16708 (movmisalign*): Likewise.
16709 * doc/invoke.texi (ARC options): Document -G option.
16711 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16713 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16715 * config/arc/arc.c (arc_print_operand): Output scalled address for
16716 sdata whenever is possible.
16717 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16718 load/stores are available.
16719 (compact_sda_memory_operand): Check for the alignment required by
16720 code density instructions.
16721 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16723 * config/arc/constraints.md (Usd): Update constraint.
16724 (Us0): New constraint.
16725 (Usc): Update constraint.
16727 2017-08-31 Richard Biener <rguenther@suse.de>
16729 PR middle-end/82054
16730 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16731 function only once.
16733 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16735 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16736 Resize type_signature.
16738 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16739 Alan Hayward <alan.hayward@arm.com>
16740 David Sherwood <david.sherwood@arm.com>
16742 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16743 subregs whose inner modes can be stored in GPRs.
16744 (aarch64_classify_index): Likewise.
16746 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16747 Alan Hayward <alan.hayward@arm.com>
16748 David Sherwood <david.sherwood@arm.com>
16750 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16751 (V_INT_EQUIV): ...this.
16752 (v_cmp_result): Rename to...
16753 (v_int_equiv): ...this.
16754 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16755 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16756 (copysign<mode>3): Likewise.
16757 (aarch64_simd_bsl<mode>_internal): Likewise.
16758 (aarch64_simd_bsl<mode>): Likewise.
16759 (vec_cmp<mode><mode>): Likewise.
16760 (vcond<mode><mode>): Likewise.
16761 (vcond<v_cmp_mixed><mode>): Likewise.
16762 (vcondu<mode><v_cmp_mixed>): Likewise.
16763 (aarch64_cm<optab><mode>): Likewise.
16764 (aarch64_cmtst<mode>): Likewise.
16765 (aarch64_fac<optab><mode>): Likewise.
16766 (vec_perm_const<mode>): Likewise.
16767 (vcond_mask_<mode><v_cmp_result>): Rename to...
16768 (vcond_mask_<mode><v_int_equiv>): ...this.
16769 (vec_cmp<mode><v_cmp_result>): Rename to...
16770 (vec_cmp<mode><v_int_equiv>): ...this.
16772 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16773 Alan Hayward <alan.hayward@arm.com>
16774 David Sherwood <david.sherwood@arm.com>
16776 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16778 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16779 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16780 (UNSPEC_LD4_DREG): New unspecs.
16781 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16782 (aarch64_ld2<mode>_dreg_be): Replace with...
16783 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16785 (aarch64_ld3<mode>_dreg_le)
16786 (aarch64_ld3<mode>_dreg_be): Replace with...
16787 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16789 (aarch64_ld4<mode>_dreg_le)
16790 (aarch64_ld4<mode>_dreg_be): Replace with...
16791 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16794 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16796 PR tree-optimization/81987
16797 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16798 insert an initializer in a location not dominated by the stride
16801 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16803 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16804 on the entire header of the finally block in the fallthru case.
16806 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16808 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16810 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16812 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16813 rs6000_emit_move_from_cr and call renamed function.
16814 (rs6000_emit_prologue): Call renamed functions.
16815 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16816 movesi_from_cr, remove volatile CRs.
16818 2017-08-30 Jon Beniston <jon@beniston.com>
16819 Richard Biener <rguenther@suse.de>
16821 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16822 of VECTOR_MODE_P check.
16823 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16824 element vector types.
16826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16828 * df.h (df_read_modify_subreg_p): Remove in favor of...
16829 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16830 const_rtx instead of an rtx.
16831 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16832 * df-problems.c (df_word_lr_mark_ref): Likewise.
16833 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16834 (mark_pseudo_reg_dead): Likewise.
16835 (mark_ref_dead): Likewise.
16836 * reginfo.c (init_subregs_of_mode): Likewise.
16837 * sched-deps.c (sched_analyze_1): Likewise.
16838 * df-scan.c (df_def_record_1): Likewise.
16839 (df_uses_record): Likewise.
16840 (df_read_modify_subreg_p): Remove in favor of...
16841 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16842 const_rtx instead of an rtx.
16844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16845 Alan Hayward <alan.hayward@arm.com>
16846 David Sherwood <david.sherwood@arm.com>
16848 * rtl.h (partial_subreg_p): New function.
16849 * caller-save.c (save_call_clobbered_regs): Use it.
16850 * calls.c (expand_call): Likewise.
16851 * combine.c (combinable_i3pat): Likewise.
16852 (simplify_set): Likewise.
16853 (make_extraction): Likewise.
16854 (make_compound_operation_int): Likewise.
16855 (gen_lowpart_or_truncate): Likewise.
16856 (force_to_mode): Likewise.
16857 (make_field_assignment): Likewise.
16858 (reg_truncated_to_mode): Likewise.
16859 (record_truncated_value): Likewise.
16860 (move_deaths): Likewise.
16861 * cse.c (record_jump_cond): Likewise.
16862 (cse_insn): Likewise.
16863 * cselib.c (cselib_lookup_1): Likewise.
16864 * expmed.c (extract_bit_field_using_extv): Likewise.
16865 * function.c (assign_parm_setup_reg): Likewise.
16866 * ifcvt.c (noce_convert_multiple_sets): Likewise.
16867 * ira-build.c (create_insn_allocnos): Likewise.
16868 * lra-coalesce.c (merge_pseudos): Likewise.
16869 * lra-constraints.c (match_reload): Likewise.
16870 (simplify_operand_subreg): Likewise.
16871 (curr_insn_transform): Likewise.
16872 * lra-lives.c (process_bb_lives): Likewise.
16873 * lra.c (new_insn_reg): Likewise.
16874 (lra_substitute_pseudo): Likewise.
16875 * regcprop.c (mode_change_ok): Likewise.
16876 (maybe_mode_change): Likewise.
16877 (copyprop_hardreg_forward_1): Likewise.
16878 * reload.c (push_reload): Likewise.
16879 (find_reloads): Likewise.
16880 (find_reloads_subreg_address): Likewise.
16881 * reload1.c (alter_reg): Likewise.
16882 (eliminate_regs_1): Likewise.
16883 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16885 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
16887 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
16888 back to if statements, including unpack.
16890 2017-08-30 Martin Liska <mliska@suse.cz>
16892 PR inline-asm/82001
16893 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
16895 (func_checker::compare_asm_inputs_outputs): ... this function.
16896 (func_checker::compare_gimple_asm): Use the function to compare
16897 also ASM constrains.
16898 * ipa-icf-gimple.h: Rename the function.
16900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16901 Alan Hayward <alan.hayward@arm.com>
16902 David Sherwood <david.sherwood@arm.com>
16904 * coretypes.h (complex_mode): New type.
16905 * gdbhooks.py (build_pretty_printer): Handle it.
16906 * machmode.h (complex_mode): New class.
16907 (complex_mode::includes_p): New function.
16908 (is_complex_int_mode): Likewise.
16909 (is_complex_float_mode): Likewise.
16910 * genmodes.c (get_mode_class): Handle complex mode classes.
16911 * function.c (expand_function_end): Use is_complex_int_mode.
16913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16914 Alan Hayward <alan.hayward@arm.com>
16915 David Sherwood <david.sherwood@arm.com>
16917 * coretypes.h (scalar_mode_pod): New typedef.
16918 * gdbhooks.py (build_pretty_printer): Handle it.
16919 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
16920 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
16921 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
16922 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
16923 as_a <scalar_mode>.
16925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16926 Alan Hayward <alan.hayward@arm.com>
16927 David Sherwood <david.sherwood@arm.com>
16929 * machmode.h (mode_for_vector): Take a scalar_mode instead
16931 * stor-layout.c (mode_for_vector): Likewise.
16932 * explow.c (promote_mode): Use as_a <scalar_mode>.
16933 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
16935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16936 Alan Hayward <alan.hayward@arm.com>
16937 David Sherwood <david.sherwood@arm.com>
16939 * target.def (preferred_simd_mode): Take a scalar_mode
16940 instead of a machine_mode.
16941 * targhooks.h (default_preferred_simd_mode): Likewise.
16942 * targhooks.c (default_preferred_simd_mode): Likewise.
16943 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
16944 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
16945 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
16946 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
16947 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
16948 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
16949 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
16950 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
16952 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
16953 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
16954 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
16955 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
16956 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
16957 * doc/tm.texi: Regenerate.
16958 * optabs-query.c (can_vec_mask_load_store_p): Return false for
16961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16962 Alan Hayward <alan.hayward@arm.com>
16963 David Sherwood <david.sherwood@arm.com>
16965 * target.def (scalar_mode_supported_p): Take a scalar_mode
16966 instead of a machine_mode.
16967 * targhooks.h (default_scalar_mode_supported_p): Likewise.
16968 * targhooks.c (default_scalar_mode_supported_p): Likewise.
16969 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
16970 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16971 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
16972 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
16973 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
16974 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
16975 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
16976 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
16977 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
16978 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
16979 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
16980 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
16982 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
16983 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
16984 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16985 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16986 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
16988 * doc/tm.texi: Regenerate.
16990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16991 Alan Hayward <alan.hayward@arm.com>
16992 David Sherwood <david.sherwood@arm.com>
16994 * coretypes.h (opt_scalar_mode): New typedef.
16995 * gdbhooks.py (build_pretty_printers): Handle it.
16996 * machmode.h (mode_iterator::get_2xwider): Add overload for
16998 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17000 * expr.c (convert_mode_scalar): Likewise.
17001 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17002 * optabs.c (expand_float): Likewise.
17003 (expand_fix): Likewise.
17004 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17007 Alan Hayward <alan.hayward@arm.com>
17008 David Sherwood <david.sherwood@arm.com>
17010 * optabs.c (expand_float): Explicitly check for scalars before
17011 using a branching expansion.
17012 (expand_fix): Likewise.
17014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17015 Alan Hayward <alan.hayward@arm.com>
17016 David Sherwood <david.sherwood@arm.com>
17018 * expr.c (convert_mode): Split scalar handling out into...
17019 (convert_mode_scalar): ...this new function. Treat the modes
17022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17023 Alan Hayward <alan.hayward@arm.com>
17024 David Sherwood <david.sherwood@arm.com>
17026 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17028 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17031 Alan Hayward <alan.hayward@arm.com>
17032 David Sherwood <david.sherwood@arm.com>
17034 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17035 rather than a machine_mode.
17036 (fixed_from_string): Likewise.
17037 (fixed_convert): Likewise.
17038 (fixed_convert_from_int): Likewise.
17039 (fixed_convert_from_real): Likewise.
17040 (real_convert_from_fixed): Likewise.
17041 * fixed-value.c (fixed_from_double_int): Likewise.
17042 (fixed_from_string): Likewise.
17043 (fixed_convert): Likewise.
17044 (fixed_convert_from_int): Likewise.
17045 (fixed_convert_from_real): Likewise.
17046 (real_convert_from_fixed): Likewise.
17047 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17050 Alan Hayward <alan.hayward@arm.com>
17051 David Sherwood <david.sherwood@arm.com>
17053 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17054 of separate mode class checks. Do not allow vector modes here.
17055 (immed_wide_int_const): Use as_a <scalar_mode>.
17056 * explow.c (trunc_int_for_mode): Likewise.
17057 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17058 (wi::shwi): Likewise.
17059 (wi::min_value): Likewise.
17060 (wi::max_value): Likewise.
17061 * dwarf2out.c (loc_descriptor): Likewise.
17062 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17063 for CONST_WIDE_INT.
17065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17066 Alan Hayward <alan.hayward@arm.com>
17067 David Sherwood <david.sherwood@arm.com>
17069 * tree.h (SCALAR_TYPE_MODE): New macro.
17070 * expr.c (expand_expr_addr_expr_1): Use it.
17071 (expand_expr_real_2): Likewise.
17072 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17073 (fold_convert_const_fixed_from_int): Likewise.
17074 (fold_convert_const_fixed_from_real): Likewise.
17075 (native_encode_fixed): Likewise
17076 (native_encode_complex): Likewise
17077 (native_encode_vector): Likewise.
17078 (native_interpret_fixed): Likewise.
17079 (native_interpret_real): Likewise.
17080 (native_interpret_complex): Likewise.
17081 (native_interpret_vector): Likewise.
17082 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17083 (simd_clone_adjust_argument_types): Likewise.
17084 (simd_clone_init_simd_arrays): Likewise.
17085 (simd_clone_adjust): Likewise.
17086 * stor-layout.c (layout_type): Likewise.
17087 * tree.c (build_minus_one_cst): Likewise.
17088 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17089 * tree-inline.c (estimate_move_cost): Likewise.
17090 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17091 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17092 (vectorizable_reduction): Likewise.
17093 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17094 (vect_recog_mixed_size_cond_pattern): Likewise.
17095 (check_bool_pattern): Likewise.
17096 (adjust_bool_pattern): Likewise.
17097 (search_type_for_mask_1): Likewise.
17098 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17099 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17100 (vectorizable_load): Likewise.
17101 (vectorizable_store): Likewise.
17102 * ubsan.c (ubsan_encode_value): Likewise.
17103 * varasm.c (output_constant): Likewise.
17105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17106 Alan Hayward <alan.hayward@arm.com>
17107 David Sherwood <david.sherwood@arm.com>
17109 * coretypes.h (scalar_mode): New class.
17110 * machmode.h (scalar_mode): Likewise.
17111 (scalar_mode::includes_p): New function.
17112 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17113 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17114 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17115 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17116 * expmed.c (store_bit_field_1): Likewise.
17117 (extract_bit_field_1): Likewise.
17118 * expr.c (write_complex_part): Likewise.
17119 (read_complex_part): Likewise.
17120 (emit_move_complex_push): Likewise.
17121 (expand_expr_real_2): Likewise.
17122 * function.c (assign_parm_setup_reg): Likewise.
17123 (assign_parms_unsplit_complex): Likewise.
17124 * optabs.c (expand_binop): Likewise.
17125 * rtlanal.c (subreg_get_info): Likewise.
17126 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17127 * varasm.c (output_constant_pool_2): Likewise.
17129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17130 Alan Hayward <alan.hayward@arm.com>
17131 David Sherwood <david.sherwood@arm.com>
17133 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17135 (expmed_mult_highpart_optab): Likewise.
17136 (expmed_mult_highpart): Likewise.
17138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17139 Alan Hayward <alan.hayward@arm.com>
17140 David Sherwood <david.sherwood@arm.com>
17142 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17143 instead of a machine_mode.
17144 (builtin_memset_read_str): Likewise.
17145 * builtins.c (c_readstr): Likewise.
17146 (builtin_memcpy_read_str): Likewise.
17147 (builtin_strncpy_read_str): Likewise.
17148 (builtin_memset_read_str): Likewise.
17149 (builtin_memset_gen_str): Likewise.
17150 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17151 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17152 instead of a machine_mode.
17153 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17155 (make_extraction): Likewise.
17156 (try_widen_shift_mode): Take and return scalar_int_modes instead
17158 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17159 a scalar_int_mode instead of a machine_mode.
17160 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17161 (avr_addr_space_pointer_mode): Likewise.
17162 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17163 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17164 (msp430_unwind_word_mode): Likewise.
17165 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17166 (spu_addr_space_pointer_mode): Likewise.
17167 (spu_addr_space_address_mode): Likewise.
17168 (spu_libgcc_cmp_return_mode): Likewise.
17169 (spu_libgcc_shift_count_mode): Likewise.
17170 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17171 (rl78_addr_space_pointer_mode): Likewise.
17172 (fl78_unwind_word_mode): Likewise.
17173 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17175 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17176 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17177 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17178 (mips_valid_pointer_mode): Likewise.
17179 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17180 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17181 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17183 (ft32_addr_space_address_mode): Likewise.
17184 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17185 scalar_int_mode instead of a machine_mode.
17186 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17188 (m32c_addr_space_address_mode): Likewise.
17189 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17190 (rs6000_eh_return_filter_mode): Likewise.
17191 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17192 (rs6000_eh_return_filter_mode): Likewise.
17193 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17194 (s390_libgcc_shift_count_mode): Likewise.
17195 (s390_unwind_word_mode): Likewise.
17196 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17198 * target.def (mode_rep_extended): Likewise.
17199 (valid_pointer_mode): Likewise.
17200 (addr_space.valid_pointer_mode): Likewise.
17201 (eh_return_filter_mode): Return a scalar_int_mode rather than
17203 (libgcc_cmp_return_mode): Likewise.
17204 (libgcc_shift_count_mode): Likewise.
17205 (unwind_word_mode): Likewise.
17206 (addr_space.pointer_mode): Likewise.
17207 (addr_space.address_mode): Likewise.
17208 * doc/tm.texi: Regenerate.
17209 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17211 (do_jump): Use scalar_int_mode for local variables.
17212 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17213 rather than a machine_mode.
17214 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17215 (scompare_loc_descriptor_wide): Likewise.
17216 (scompare_loc_descriptor_narrow): Likewise.
17217 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17219 * except.c (sjlj_emit_dispatch_table): Likewise.
17220 (expand_builtin_eh_copy_values): Likewise.
17221 * explow.c (convert_memory_address_addr_space_1): Likewise.
17222 Take a scalar_int_mode rather than a machine_mode.
17223 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17224 than a machine_mode.
17225 (memory_address_addr_space): Use scalar_int_mode for local variables.
17226 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17227 rather than a machine_mode.
17228 * expmed.c (mask_rtx): Likewise.
17229 (init_expmed_one_conv): Likewise.
17230 (expand_mult_highpart_adjust): Likewise.
17231 (extract_high_half): Likewise.
17232 (expmed_mult_highpart_optab): Likewise.
17233 (expmed_mult_highpart): Likewise.
17234 (expand_smod_pow2): Likewise.
17235 (expand_sdiv_pow2): Likewise.
17236 (emit_store_flag_int): Likewise.
17237 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17239 (extract_low_bits): Likewise.
17240 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17242 * expr.c (pieces_addr::adjust): Likewise.
17243 (can_store_by_pieces): Likewise.
17244 (store_by_pieces): Likewise.
17245 (clear_by_pieces_1): Likewise.
17246 (expand_expr_addr_expr_1): Likewise.
17247 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17248 (expand_expr_real_1): Likewise.
17249 (try_casesi): Likewise.
17250 * final.c (shorten_branches): Likewise.
17251 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17252 type of "mode" to machine_mode.
17253 * internal-fn.c (expand_arith_overflow_result_store): Take a
17254 scalar_int_mode rather than a machine_mode.
17255 (expand_mul_overflow): Use scalar_int_mode for local variables.
17256 * loop-doloop.c (doloop_modify): Likewise.
17257 (doloop_optimize): Likewise.
17258 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17259 than a machine_mode.
17260 (expand_doubleword_shift_condmove): Likewise.
17261 (expand_doubleword_shift): Likewise.
17262 (expand_doubleword_clz): Likewise.
17263 (expand_doubleword_popcount): Likewise.
17264 (expand_doubleword_parity): Likewise.
17265 (expand_absneg_bit): Use scalar_int_mode for local variables.
17266 (prepare_float_lib_cmp): Likewise.
17267 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17268 rather than a machine_mode.
17269 (convert_memory_address_addr_space): Likewise.
17270 (get_mode_bounds): Likewise.
17271 (get_address_mode): Return a scalar_int_mode rather than a
17273 * rtlanal.c (get_address_mode): Likewise.
17274 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17275 than a machine_mode.
17276 * targhooks.c (default_mode_rep_extended): Likewise.
17277 (default_valid_pointer_mode): Likewise.
17278 (default_addr_space_valid_pointer_mode): Likewise.
17279 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17280 than a machine_mode.
17281 (default_libgcc_cmp_return_mode): Likewise.
17282 (default_libgcc_shift_count_mode): Likewise.
17283 (default_unwind_word_mode): Likewise.
17284 (default_addr_space_pointer_mode): Likewise.
17285 (default_addr_space_address_mode): Likewise.
17286 * targhooks.h (default_eh_return_filter_mode): Likewise.
17287 (default_libgcc_cmp_return_mode): Likewise.
17288 (default_libgcc_shift_count_mode): Likewise.
17289 (default_unwind_word_mode): Likewise.
17290 (default_addr_space_pointer_mode): Likewise.
17291 (default_addr_space_address_mode): Likewise.
17292 (default_mode_rep_extended): Take a scalar_int_mode rather than
17294 (default_valid_pointer_mode): Likewise.
17295 (default_addr_space_valid_pointer_mode): Likewise.
17296 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17298 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17299 rather than a machine_mode.
17300 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17301 for local variables.
17302 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17303 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17304 than a machine_mode.
17306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17307 Alan Hayward <alan.hayward@arm.com>
17308 David Sherwood <david.sherwood@arm.com>
17310 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17311 the mode argument to scalar_int_mode.
17312 (do_jump_by_parts_zero_rtx): Likewise.
17313 (do_jump_by_parts_equality_rtx): Likewise.
17314 (do_jump_by_parts_greater): Take a mode argument.
17315 (do_jump_by_parts_equality): Likewise.
17316 (do_jump_1): Update calls accordingly.
17318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17319 Alan Hayward <alan.hayward@arm.com>
17320 David Sherwood <david.sherwood@arm.com>
17322 * is-a.h (safe_dyn_cast): New function.
17323 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17324 (jump_table_for_label): Likewise.
17325 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17326 instead of an rtx_insn *.
17327 (shorten_branches): Use dyn_cast instead of LABEL_P and
17328 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17329 rtx_jump_table_data::get_data_mode.
17330 (final_scan_insn): Likewise.
17332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17333 Alan Hayward <alan.hayward@arm.com>
17334 David Sherwood <david.sherwood@arm.com>
17336 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17337 trying to combine a full-register integer set with a subreg
17340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17341 Alan Hayward <alan.hayward@arm.com>
17342 David Sherwood <david.sherwood@arm.com>
17344 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17345 that is always either address_mode or pointer_mode.
17347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17348 Alan Hayward <alan.hayward@arm.com>
17349 David Sherwood <david.sherwood@arm.com>
17351 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17352 when the two are known to be equal.
17354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17355 Alan Hayward <alan.hayward@arm.com>
17356 David Sherwood <david.sherwood@arm.com>
17358 * simplify-rtx.c (simplify_const_unary_operation): Use
17359 is_a <scalar_int_mode> instead of checking for a nonzero
17360 precision. Forcibly convert op_mode to a scalar_int_mode
17361 in that case. More clearly differentiate the operand and
17362 result modes and use the former when deciding what the value
17363 of a count-bits operation should be. Use is_int_mode instead
17364 of checking for a MODE_INT. Remove redundant check for whether
17365 this mode has a zero precision.
17367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17368 Alan Hayward <alan.hayward@arm.com>
17369 David Sherwood <david.sherwood@arm.com>
17371 * optabs.c (widen_leading): Change the type of the mode argument
17372 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17373 (widen_bswap): Likewise.
17374 (expand_parity): Likewise.
17375 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17376 (expand_ffs): Likewise.
17377 (epand_unop): Check for scalar integer modes before calling the
17380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17381 Alan Hayward <alan.hayward@arm.com>
17382 David Sherwood <david.sherwood@arm.com>
17384 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17386 (expand_expr_real_1): Update call accordingly.
17388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17389 Alan Hayward <alan.hayward@arm.com>
17390 David Sherwood <david.sherwood@arm.com>
17392 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17393 value_mode arguments. Use scalar_int_mode internally.
17394 (store_bit_field_1): Rename the new integer mode from imode
17395 to op0_mode and use it instead of GET_MODE (op0). Update calls
17396 to store_split_bit_field, store_bit_field_using_insv and
17397 store_fixed_bit_field.
17398 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17399 Use scalar_int_mode internally. Use a bit count rather than a mode
17400 when calculating the largest bit size for get_best_mode.
17401 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17402 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17403 Remove assertion that OP0 has a scalar integer mode.
17404 (store_split_bit_field): Add op0_mode and value_mode arguments.
17405 Update calls to extract_fixed_bit_field.
17406 (extract_bit_field_using_extv): Add an op0_mode argument.
17407 Use scalar_int_mode internally.
17408 (extract_bit_field_1): Rename the new integer mode from imode to
17409 op0_mode and use it instead of GET_MODE (op0). Update calls to
17410 extract_split_bit_field, extract_bit_field_using_extv and
17411 extract_fixed_bit_field.
17412 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17413 to extract_split_bit_field and extract_fixed_bit_field_1.
17414 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17415 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17416 on the target mode.
17417 (extract_split_bit_field): Add an op0_mode argument. Update call
17418 to extract_fixed_bit_field.
17420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17421 Alan Hayward <alan.hayward@arm.com>
17422 David Sherwood <david.sherwood@arm.com>
17424 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17425 * explow.c (hard_function_value): Likewise.
17426 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17427 convert_to_mode call outside the loop.
17428 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17429 for the mode iterator. Require the mode specified by max_pieces
17431 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17433 (copy_blkmode_to_reg): Likewise.
17434 (set_storage_via_setmem): Likewise.
17435 * optabs.c (prepare_cmp_insn): Likewise.
17436 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17437 * stor-layout.c (finish_bitfield_representative): Likewise.
17439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17440 Alan Hayward <alan.hayward@arm.com>
17441 David Sherwood <david.sherwood@arm.com>
17443 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17444 * expr.c (convert_move): Use them.
17445 (convert_modes): Likewise.
17446 (store_expr_with_bounds): Likewise.
17448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17449 Alan Hayward <alan.hayward@arm.com>
17450 David Sherwood <david.sherwood@arm.com>
17452 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17453 parameter for the mode of "x". Remove the "known_x", "known_mode"
17454 and "known_ret" arguments. Change the type of the mode argument
17455 to scalar_int_mode.
17456 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17457 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17458 (reg_num_sign_bit_copies_for_combine): Likewise.
17459 * rtlanal.c (nonzero_bits1): Likewise.
17460 (num_sign_bit_copies1): Likewise.
17461 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17462 (reg_num_sign_bit_copies_general): Likewise.
17463 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17464 (reg_nonzero_bits_general): Likewise.
17466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17467 Alan Hayward <alan.hayward@arm.com>
17468 David Sherwood <david.sherwood@arm.com>
17470 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17471 than in subroutines. Return 1 for non-integer modes.
17472 (cached_num_sign_bit_copies): Change the type of the mode parameter
17473 to scalar_int_mode.
17474 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17475 classes. Handle CONST_INT_P first and then check whether X also
17476 has a scalar integer mode. Check the same thing for inner registers
17477 of a SUBREG and for values that are being extended or truncated.
17479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17480 Alan Hayward <alan.hayward@arm.com>
17481 David Sherwood <david.sherwood@arm.com>
17483 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17484 in subroutines. Return the mode mask for non-integer modes.
17485 (cached_nonzero_bits): Change the type of the mode parameter
17486 to scalar_int_mode.
17487 (nonzero_bits1): Likewise. Remove early exit for other mode
17488 classes. Handle CONST_INT_P first and then check whether X
17489 also has a scalar integer mode.
17491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17492 Alan Hayward <alan.hayward@arm.com>
17493 David Sherwood <david.sherwood@arm.com>
17495 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17496 Return a scalar_int_mode and assert that the size is greater than
17498 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17499 (op_by_pieces_d::op_by_pieces_d): Likewise.
17500 (op_by_pieces_d::run): Likewise.
17501 (can_store_by_pieces): Likewise.
17503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17504 Alan Hayward <alan.hayward@arm.com>
17505 David Sherwood <david.sherwood@arm.com>
17507 * combine.c (extract_left_shift): Add a mode argument and update
17509 (make_compound_operation_int): Change the type of the mode parameter
17510 to scalar_int_mode and update the call to extract_left_shift.
17512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17513 Alan Hayward <alan.hayward@arm.com>
17514 David Sherwood <david.sherwood@arm.com>
17516 * combine.c (simplify_and_const_int): Change the type of the mode
17517 parameter to scalar_int_mode.
17518 (simplify_and_const_int_1): Likewise. Update recursive call.
17520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17521 Alan Hayward <alan.hayward@arm.com>
17522 David Sherwood <david.sherwood@arm.com>
17524 * combine.c (simplify_compare_const): Check that the mode is a
17525 scalar_int_mode (rather than VOIDmode) before testing its
17527 (simplify_comparison): Move COMPARISON_P handling out of the
17528 loop and restrict the latter part of the loop to scalar_int_modes.
17529 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17530 and when considering SUBREG_REGs. Use is_int_mode instead of
17531 checking GET_MODE_CLASS against MODE_INT.
17533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17534 Alan Hayward <alan.hayward@arm.com>
17535 David Sherwood <david.sherwood@arm.com>
17537 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17538 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17539 shift_unit_mode and for modes involved in scalar shifts.
17541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17542 Alan Hayward <alan.hayward@arm.com>
17543 David Sherwood <david.sherwood@arm.com>
17545 * combine.c (force_int_to_mode): New function, split out from...
17546 (force_to_mode): ...here. Keep xmode up-to-date and use it
17547 instead of GET_MODE (x).
17549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17550 Alan Hayward <alan.hayward@arm.com>
17551 David Sherwood <david.sherwood@arm.com>
17553 * optabs-query.h (extraction_insn::struct_mode): Change type to
17554 opt_scalar_int_mode and update comment.
17555 (extraction_insn::field_mode): Change type to scalar_int_mode.
17556 (extraction_insn::pos_mode): Likewise.
17557 * combine.c (make_extraction): Update accordingly.
17558 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17559 (get_optab_extraction_insn): Likewise.
17560 * recog.c (simplify_while_replacing): Likewise.
17561 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17562 parameter to opt_scalar_int_mode.
17564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17565 Alan Hayward <alan.hayward@arm.com>
17566 David Sherwood <david.sherwood@arm.com>
17568 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17569 to a scalar_int_mode instead of a machine_mode.
17570 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17571 (get_best_mode): Return a boolean and use a pointer argument to store
17572 the selected mode. Replace the limit mode parameter with a bit limit.
17573 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17574 for the values returned by bit_field_mode_iterator::next_mode.
17575 (store_bit_field): Update call to get_best_mode.
17576 (store_fixed_bit_field): Likewise.
17577 (extract_fixed_bit_field): Likewise.
17578 * expr.c (optimize_bitfield_assignment_op): Likewise.
17579 * fold-const.c (optimize_bit_field_compare): Likewise.
17580 (fold_truth_andor_1): Likewise.
17581 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17582 Update for new type of m_mode.
17583 (get_best_mode): As above.
17585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17586 Alan Hayward <alan.hayward@arm.com>
17587 David Sherwood <david.sherwood@arm.com>
17589 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17590 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17591 (store_bit_field): Check is_a <scalar_int_mode> before calling
17592 strict_volatile_bitfield_p.
17593 (extract_bit_field): Likewise.
17595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17596 Alan Hayward <alan.hayward@arm.com>
17597 David Sherwood <david.sherwood@arm.com>
17599 * target.def (cstore_mode): Return a scalar_int_mode.
17600 * doc/tm.texi: Regenerate.
17601 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17602 * targhooks.h (default_cstore_mode): Likewise.
17603 * targhooks.c (default_cstore_mode): Likewise, using a forced
17605 * expmed.c (emit_cstore): Expect the target of the cstore to be
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 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17614 (niter_desc): Likewise mode.
17615 (iv_analyze): Add a mode parameter.
17617 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17618 and change its type to scalar_int_mode.
17619 * loop-iv.c: Update commentary at head of file.
17620 (iv_constant): Pass the mode paraeter before the rtx it describes
17621 and change its type to scalar_int_mode. Remove VOIDmode handling.
17622 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17623 (iv_extend): Likewise.
17624 (shorten_into_mode): Likewise.
17625 (iv_add): Use scalar_int_mode.
17626 (iv_mult): Likewise.
17627 (iv_shift): Likewise.
17628 (canonicalize_iv_subregs): Likewise.
17629 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17630 it describes and change its mode to scalar_int_mode. Also change
17631 the type of the returned inner_mode to scalar_int_mode.
17632 (get_biv_step): Likewise, turning outer_mode from a pointer
17633 into a direct parameter. Update call to get_biv_step_1.
17634 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17635 iv_constant and get_biv_step.
17636 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17637 and change its type to scalar_int_mode. Don't initialise iv->mode
17638 to VOIDmode and remove later checks for its still being VOIDmode.
17639 Update calls to iv_analyze_op and iv_analyze_expr. Check
17640 is_a <scalar_int_mode> when changing the mode under consideration.
17641 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17642 Update call to iv_analyze_expr.
17643 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17644 inner register is not also a scalar_int_mode. Update call to
17646 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17647 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17648 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17649 separate mode class checks. Update calls to iv_analyze. Remove
17650 fix-up of VOIDmodes after iv_analyze_biv.
17651 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17652 don't have a scalar_int_mode. Update call to biv_p.
17654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17655 Alan Hayward <alan.hayward@arm.com>
17656 David Sherwood <david.sherwood@arm.com>
17658 * cfgexpand.c (convert_debug_memory_address): Use
17659 as_a <scalar_int_mode>.
17660 * combine.c (expand_compound_operation): Likewise.
17661 (make_extraction): Likewise.
17662 (change_zero_ext): Likewise.
17663 (simplify_comparison): Likewise.
17664 * cse.c (cse_insn): Likewise.
17665 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17666 (mem_loc_descriptor): Likewise.
17667 (loc_descriptor): Likewise.
17668 * expmed.c (init_expmed_one_mode): Likewise.
17669 (synth_mult): Likewise.
17670 (emit_store_flag_1): Likewise.
17671 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17672 of a comparison with size.
17673 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17674 (reduce_to_bit_field_precision): Likewise.
17675 * function.c (expand_function_end): Likewise.
17676 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17677 * loop-doloop.c (doloop_modify): Likewise.
17678 * optabs.c (expand_binop): Likewise.
17679 (expand_unop): Likewise.
17680 (expand_copysign_absneg): Likewise.
17681 (prepare_cmp_insn): Likewise.
17682 (maybe_legitimize_operand): Likewise.
17683 * recog.c (const_scalar_int_operand): Likewise.
17684 * rtlanal.c (get_address_mode): Likewise.
17685 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17686 (simplify_cond_clz_ctz): Likewise.
17687 * tree-nested.c (get_nl_goto_field): Likewise.
17688 * tree.c (build_vector_type_for_mode): Likewise.
17689 * var-tracking.c (use_narrower_mode): Likewise.
17691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17692 Alan Hayward <alan.hayward@arm.com>
17693 David Sherwood <david.sherwood@arm.com>
17695 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17696 * builtins.c (expand_builtin_signbit): Use it.
17697 * cfgexpand.c (expand_debug_expr): Likewise.
17698 * dojump.c (do_jump): Likewise.
17699 (do_compare_and_jump): Likewise.
17700 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17701 * expmed.c (make_tree): Likewise.
17702 * expr.c (expand_expr_real_2): Likewise.
17703 (expand_expr_real_1): Likewise.
17704 (try_casesi): Likewise.
17705 * fold-const-call.c (fold_const_call_ss): Likewise.
17706 * fold-const.c (unextend): Likewise.
17707 (extract_muldiv_1): Likewise.
17708 (fold_single_bit_test): Likewise.
17709 (native_encode_int): Likewise.
17710 (native_encode_string): Likewise.
17711 (native_interpret_int): Likewise.
17712 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17713 * internal-fn.c (expand_addsub_overflow): Likewise.
17714 (expand_neg_overflow): Likewise.
17715 (expand_mul_overflow): Likewise.
17716 (expand_arith_overflow): Likewise.
17717 * match.pd: Likewise.
17718 * stor-layout.c (layout_type): Likewise.
17719 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17720 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17721 * tree-ssanames.c (get_range_info): Likewise.
17722 * tree-switch-conversion.c (array_value_type) Likewise.
17723 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17724 (vect_recog_divmod_pattern): Likewise.
17725 (vect_recog_mixed_size_cond_pattern): Likewise.
17726 * tree-vrp.c (extract_range_basic): Likewise.
17727 (simplify_float_conversion_using_ranges): Likewise.
17728 * tree.c (int_fits_type_p): Likewise.
17729 * ubsan.c (instrument_bool_enum_load): Likewise.
17730 * varasm.c (mergeable_string_section): Likewise.
17731 (narrowing_initializer_constant_valid_p): Likewise.
17732 (output_constant): Likewise.
17734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17735 Alan Hayward <alan.hayward@arm.com>
17736 David Sherwood <david.sherwood@arm.com>
17738 * machmode.h (NARROWEST_INT_MODE): New macro.
17739 * expr.c (alignment_for_piecewise_move): Use it instead of
17740 GET_CLASS_NARROWEST_MODE (MODE_INT).
17741 (push_block): Likewise.
17742 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17744 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17747 Alan Hayward <alan.hayward@arm.com>
17748 David Sherwood <david.sherwood@arm.com>
17750 * postreload.c (move2add_valid_value_p): Change the type of the
17751 mode parameter to scalar_int_mode.
17752 (move2add_use_add2_insn): Add a mode parameter and use it instead
17754 (move2add_use_add3_insn): Likewise.
17755 (reload_cse_move2add): Update accordingly.
17757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17758 Alan Hayward <alan.hayward@arm.com>
17759 David Sherwood <david.sherwood@arm.com>
17761 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17763 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17764 * optabs.c (expand_unop): Likewise.
17766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17767 Alan Hayward <alan.hayward@arm.com>
17768 David Sherwood <david.sherwood@arm.com>
17770 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17771 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17772 (popcount_loc_descriptor): Likewise.
17773 (bswap_loc_descriptor): Likewise.
17774 (rotate_loc_descriptor): Likewise.
17775 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17776 calling the functions above.
17778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17779 Alan Hayward <alan.hayward@arm.com>
17780 David Sherwood <david.sherwood@arm.com>
17782 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17784 (try_combine): Likewise.
17785 (simplify_if_then_else): Likewise.
17786 * cse.c (cse_insn): Likewise.
17787 * dwarf2out.c (mem_loc_descriptor): Likewise.
17788 * emit-rtl.c (gen_lowpart_common): Likewise.
17789 * simplify-rtx.c (simplify_truncation): Likewise.
17790 (simplify_binary_operation_1): Likewise.
17791 (simplify_const_relational_operation): Likewise.
17792 (simplify_ternary_operation): Likewise.
17793 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
17795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17796 Alan Hayward <alan.hayward@arm.com>
17797 David Sherwood <david.sherwood@arm.com>
17799 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17800 * reload.c (push_reload): Likewise.
17801 (find_reloads): Likewise.
17803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17804 Alan Hayward <alan.hayward@arm.com>
17805 David Sherwood <david.sherwood@arm.com>
17807 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17808 (make_compound_operation_int): Likewise.
17809 (change_zero_ext): Likewise.
17810 * expr.c (convert_move): Likewise.
17811 (convert_modes): Likewise.
17812 * fwprop.c (forward_propagate_subreg): Likewise.
17813 * loop-iv.c (get_biv_step_1): Likewise.
17814 * optabs.c (widen_operand): Likewise.
17815 * postreload.c (move2add_valid_value_p): Likewise.
17816 * recog.c (simplify_while_replacing): Likewise.
17817 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17818 (simplify_binary_operation_1): Likewise. Remove redundant
17819 mode equality check.
17821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17822 Alan Hayward <alan.hayward@arm.com>
17823 David Sherwood <david.sherwood@arm.com>
17825 * combine.c (combine_simplify_rtx): Add checks for
17826 is_a <scalar_int_mode>.
17827 (simplify_if_then_else): Likewise.
17828 (make_field_assignment): Likewise.
17829 (simplify_comparison): Likewise.
17830 * ifcvt.c (noce_try_bitop): Likewise.
17831 * loop-invariant.c (canonicalize_address_mult): Likewise.
17832 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17835 Alan Hayward <alan.hayward@arm.com>
17836 David Sherwood <david.sherwood@arm.com>
17838 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17839 is_a <scalar_int_mode> instead of != BLKmode.
17841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17842 Alan Hayward <alan.hayward@arm.com>
17843 David Sherwood <david.sherwood@arm.com>
17845 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17846 instead of != VOIDmode.
17847 * combine.c (if_then_else_cond): Likewise.
17848 (change_zero_ext): Likewise.
17849 * dwarf2out.c (mem_loc_descriptor): Likewise.
17850 (loc_descriptor): Likewise.
17851 * rtlanal.c (canonicalize_condition): Likewise.
17852 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17855 Alan Hayward <alan.hayward@arm.com>
17856 David Sherwood <david.sherwood@arm.com>
17858 * simplify-rtx.c (simplify_binary_operation_1): Use
17859 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
17861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17862 Alan Hayward <alan.hayward@arm.com>
17863 David Sherwood <david.sherwood@arm.com>
17865 * wide-int.h (int_traits<unsigned char>) New class.
17866 (int_traits<unsigned short>) Likewise.
17867 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
17868 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17870 * combine.c (set_nonzero_bits_and_sign_copies): Use
17871 is_a <scalar_int_mode>.
17872 (find_split_point): Likewise.
17873 (combine_simplify_rtx): Likewise.
17874 (simplify_logical): Likewise.
17875 (expand_compound_operation): Likewise.
17876 (expand_field_assignment): Likewise.
17877 (make_compound_operation): Likewise.
17878 (extended_count): Likewise.
17879 (change_zero_ext): Likewise.
17880 (simplify_comparison): Likewise.
17881 * dwarf2out.c (scompare_loc_descriptor): Likewise.
17882 (ucompare_loc_descriptor): Likewise.
17883 (minmax_loc_descriptor): Likewise.
17884 (mem_loc_descriptor): Likewise.
17885 (loc_descriptor): Likewise.
17886 * expmed.c (init_expmed_one_mode): Likewise.
17887 * lra-constraints.c (lra_constraint_offset): Likewise.
17888 * optabs.c (prepare_libcall_arg): Likewise.
17889 * postreload.c (move2add_note_store): Likewise.
17890 * reload.c (operands_match_p): Likewise.
17891 * rtl.h (load_extend_op): Likewise.
17892 * rtlhooks.c (gen_lowpart_general): Likewise.
17893 * simplify-rtx.c (simplify_truncation): Likewise.
17894 (simplify_unary_operation_1): Likewise.
17895 (simplify_binary_operation_1): Likewise.
17896 (simplify_const_binary_operation): Likewise.
17897 (simplify_const_relational_operation): Likewise.
17898 (simplify_subreg): Likewise.
17899 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17900 * var-tracking.c (adjust_mems): Likewise.
17901 (prepare_call_arguments): Likewise.
17903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17904 Alan Hayward <alan.hayward@arm.com>
17905 David Sherwood <david.sherwood@arm.com>
17907 * machmode.h (is_int_mode): New fuction.
17908 * combine.c (find_split_point): Use it.
17909 (combine_simplify_rtx): Likewise.
17910 (simplify_if_then_else): Likewise.
17911 (simplify_set): Likewise.
17912 (simplify_shift_const_1): Likewise.
17913 (simplify_comparison): Likewise.
17914 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
17915 * cse.c (notreg_cost): Likewise.
17916 (cse_insn): Likewise.
17917 * cselib.c (cselib_lookup_1): Likewise.
17918 * dojump.c (do_jump_1): Likewise.
17919 (do_compare_rtx_and_jump): Likewise.
17920 * dse.c (get_call_args): Likewise.
17921 * dwarf2out.c (rtl_for_decl_init): Likewise.
17922 (native_encode_initializer): Likewise.
17923 * expmed.c (emit_store_flag_1): Likewise.
17924 (emit_store_flag): Likewise.
17925 * expr.c (convert_modes): Likewise.
17926 (store_field): Likewise.
17927 (expand_expr_real_1): Likewise.
17928 * fold-const.c (fold_read_from_constant_string): Likewise.
17929 * gimple-ssa-sprintf.c (get_format_string): Likewise.
17930 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
17931 * optabs.c (expand_binop): Likewise.
17932 (expand_unop): Likewise.
17933 (expand_abs_nojump): Likewise.
17934 (expand_one_cmpl_abs_nojump): Likewise.
17935 * simplify-rtx.c (mode_signbit_p): Likewise.
17936 (val_signbit_p): Likewise.
17937 (val_signbit_known_set_p): Likewise.
17938 (val_signbit_known_clear_p): Likewise.
17939 (simplify_relational_operation_1): Likewise.
17940 * tree.c (vector_type_mode): Likewise.
17942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17943 Alan Hayward <alan.hayward@arm.com>
17944 David Sherwood <david.sherwood@arm.com>
17946 * machmode.h (smallest_mode_for_size): Fix formatting.
17947 (smallest_int_mode_for_size): New function.
17948 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
17949 instead of smallest_mode_for_size.
17950 * combine.c (make_extraction): Likewise.
17951 * config/arc/arc.c (arc_expand_movmem): Likewise.
17952 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
17953 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17954 * config/s390/s390.c (s390_expand_insv): Likewise.
17955 * config/sparc/sparc.c (assign_int_registers): Likewise.
17956 * config/spu/spu.c (spu_function_value): Likewise.
17957 (spu_function_arg): Likewise.
17958 * coverage.c (get_gcov_type): Likewise.
17959 (get_gcov_unsigned_t): Likewise.
17960 * dse.c (find_shift_sequence): Likewise.
17961 * expmed.c (store_bit_field_1): Likewise.
17962 * expr.c (convert_move): Likewise.
17963 (store_field): Likewise.
17964 * internal-fn.c (expand_arith_overflow): Likewise.
17965 * optabs-query.c (get_best_extraction_insn): Likewise.
17966 * optabs.c (expand_twoval_binop_libfunc): Likewise.
17967 * stor-layout.c (layout_type): Likewise.
17968 (initialize_sizetypes): Likewise.
17969 * targhooks.c (default_get_mask_mode): Likewise.
17970 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
17972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17973 Alan Hayward <alan.hayward@arm.com>
17974 David Sherwood <david.sherwood@arm.com>
17976 * machmode.h (opt_mode::else_blk): New function.
17977 (int_mode_for_mode): Declare.
17978 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
17979 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
17981 * cfgexpand.c (expand_debug_expr): Likewise.
17982 * combine.c (gen_lowpart_or_truncate): Likewise.
17983 (gen_lowpart_for_combine): Likewise.
17984 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
17985 * config/avr/avr.c (avr_to_int_mode): Likewise.
17986 (avr_out_plus_1): Likewise.
17987 (avr_out_plus): Likewise.
17988 (avr_out_round): Likewise.
17989 * config/i386/i386.c (ix86_split_to_parts): Likewise.
17990 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17991 (s390_expand_vcond): Likewise.
17992 * config/spu/spu.c (spu_split_immediate): Likewise.
17993 (spu_expand_mov): Likewise.
17994 * dse.c (get_stored_val): Likewise.
17995 * expmed.c (store_bit_field_1): Likewise.
17996 (convert_extracted_bit_field): Use int_mode_for_mode instead of
17998 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
17999 (extract_low_bits): Likewise.
18000 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18001 handling rather than repeating the check.
18002 (emit_group_store): Likewise.
18003 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18004 * optabs.c (expand_absneg_bit): Likewise.
18005 (expand_copysign_absneg): Likewise.
18006 (expand_copysign_bit): Likewise.
18007 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18008 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18009 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18010 * var-tracking.c (prepare_call_arguments): Likewise.
18011 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18012 int_mode_for_mode instead of mode_for_size.
18013 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): 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 * machmode.h (int_mode_for_size): New function.
18020 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18021 instead of mode_for_size.
18022 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18024 * combine.c (expand_field_assignment): Use int_mode_for_size
18025 instead of mode_for_size.
18026 (make_extraction): Likewise.
18027 (simplify_shift_const_1): Likewise.
18028 (simplify_comparison): Likewise.
18029 * dojump.c (do_jump): Likewise.
18030 * dwarf2out.c (mem_loc_descriptor): Likewise.
18031 * emit-rtl.c (init_derived_machine_modes): Likewise.
18032 * expmed.c (flip_storage_order): Likewise.
18033 (convert_extracted_bit_field): Likewise.
18034 * expr.c (copy_blkmode_from_reg): Likewise.
18035 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18036 * internal-fn.c (expand_mul_overflow): Likewise.
18037 * lower-subreg.c (simple_move): Likewise.
18038 * optabs-libfuncs.c (init_optabs): Likewise.
18039 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18040 * tree.c (vector_type_mode): Likewise.
18041 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18042 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18043 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18044 * tree-vect-stmts.c (vectorizable_load): Likewise.
18045 (vectorizable_store): Likewise.
18047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18048 Alan Hayward <alan.hayward@arm.com>
18049 David Sherwood <david.sherwood@arm.com>
18051 * coretypes.h (pod_mode): New type.
18052 (scalar_int_mode_pod): New typedef.
18053 * machmode.h (pod_mode): New class.
18054 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18055 * genmodes.c (emit_mode_int_n): Update accordingly.
18056 * lower-subreg.h (target_lower_subreg): Change type to
18057 scalar_int_mode_pod.
18058 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18059 scalar_int_mode_pod.
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 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18066 machine_mode to scalar_int_mode.
18067 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18068 (rs6000_option_override_internal): Remove cast to int.
18069 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18070 machine_mode to scalar_int_mode.
18071 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18072 (rs6000_option_override_internal): Remove cast to int.
18073 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18074 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18076 * config/s390/s390.c (s390_expand_builtin): Likewise.
18077 * coretypes.h (scalar_int_mode): New type.
18078 (opt_scalar_int_mode): New typedef.
18079 * machmode.h (scalar_int_mode): New class.
18080 (scalar_int_mode::includes_p): New function.
18081 (byte_mode): Change type to scalar_int_mode.
18082 (word_mode): Likewise.
18083 (ptr_mode): Likewise.
18084 * emit-rtl.c (byte_mode): Likewise.
18085 (word_mode): Likewise.
18086 (ptr_mode): Likewise.
18087 (init_derived_machine_modes): Update accordingly.
18088 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18089 and MODE_PARTIAL_INT.
18090 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18091 opt_scalar_int_mode.
18093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18094 Alan Hayward <alan.hayward@arm.com>
18095 David Sherwood <david.sherwood@arm.com>
18097 * target.def (libgcc_floating_mode_supported_p): Take a
18099 * doc/tm.texi: Regenerate.
18100 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18102 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18103 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18107 Alan Hayward <alan.hayward@arm.com>
18108 David Sherwood <david.sherwood@arm.com>
18110 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18111 * doc/tm.texi: Regenerate.
18112 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18113 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18114 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18115 * targhooks.h (default_floatn_mode): Likewise.
18116 * targhooks.c (default_floatn_mode): Likewise.
18117 * tree.c (build_common_tree_nodes): Update accordingly.
18119 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18120 Alan Hayward <alan.hayward@arm.com>
18121 David Sherwood <david.sherwood@arm.com>
18123 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18124 (mode_iterator::iterate_p): Likewise.
18125 (mode_iterator::get_wider): Likewise.
18126 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18129 Alan Hayward <alan.hayward@arm.com>
18130 David Sherwood <david.sherwood@arm.com>
18132 * coretypes.h (opt_scalar_float_mode): New typedef.
18133 * machmode.h (float_mode_for_size): New function.
18134 * emit-rtl.c (double_mode): Delete.
18135 (init_emit_once): Use float_mode_for_size.
18136 * stor-layout.c (layout_type): Likewise.
18137 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18140 Alan Hayward <alan.hayward@arm.com>
18141 David Sherwood <david.sherwood@arm.com>
18143 * output.h (assemble_real): Take a scalar_float_mode.
18144 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18145 * config/arm/arm.md (consttable_4): Likewise.
18146 (consttable_8): Likewise.
18147 (consttable_16): Likewise.
18148 * config/mips/mips.md (consttable_float): Likewise.
18149 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18150 * varasm.c (assemble_real): Take a scalar_float_mode.
18151 (output_constant_pool_2): Update accordingly.
18152 (output_constant): Likewise.
18154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18155 Alan Hayward <alan.hayward@arm.com>
18156 David Sherwood <david.sherwood@arm.com>
18158 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18159 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18160 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18161 (native_encode_real): Likewise.
18162 (native_interpret_real): Likewise.
18163 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18164 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18167 Alan Hayward <alan.hayward@arm.com>
18168 David Sherwood <david.sherwood@arm.com>
18170 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18171 <scalar_float_mode>. Simplify.
18172 (gen_extend_conv_libfunc): Likewise.
18174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18175 Alan Hayward <alan.hayward@arm.com>
18176 David Sherwood <david.sherwood@arm.com>
18178 * coretypes.h (scalar_float_mode): New type.
18179 * machmode.h (mode_traits::from_int): Use machine_mode if
18180 USE_ENUM_MODES is defined.
18181 (is_a): New function.
18183 (dyn_cast): Likewise.
18184 (scalar_float_mode): New class.
18185 (scalar_float_mode::includes_p): New function.
18186 (is_float_mode): Likewise.
18187 * gdbhooks.py (MachineModePrinter): New class.
18188 (build_pretty_printer): Use it for scalar_float_mode.
18189 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18190 (format_helper::format_helper): Turn into a template.
18191 * genmodes.c (get_mode_class): New function.
18192 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18193 or machine_mode if none.
18194 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18195 as_a <scalar_float_mode>.
18196 * dwarf2out.c (mem_loc_descriptor): Likewise.
18197 (insert_float): Likewise.
18198 (add_const_value_attribute): Likewise.
18199 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18200 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18201 (expand_unop): Update accordingly.
18202 (expand_abs_nojump): Likewise.
18203 (expand_copysign_absneg): Take a scalar_float_mode.
18204 (expand_copysign_bit): Likewise.
18205 (expand_copysign): Update accordingly.
18207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18208 Alan Hayward <alan.hayward@arm.com>
18209 David Sherwood <david.sherwood@arm.com>
18211 * coretypes.h (opt_mode): New class.
18212 * machmode.h (opt_mode): Likewise.
18213 (opt_mode::else_void): New function.
18214 (opt_mode::require): Likewise.
18215 (opt_mode::exists): Likewise.
18216 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18217 (GET_MODE_2XWIDER_MODE): Likewise.
18218 (mode_iterator::get_wider): Update accordingly.
18219 (mode_iterator::get_2xwider): Likewise.
18220 (mode_iterator::get_known_wider): Likewise, turning into a template.
18221 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18222 forcing a wider mode to exist.
18223 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18224 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18225 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18226 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18227 * lower-subreg.c (init_lower_subreg): Likewise.
18228 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18229 on the final iteration.
18230 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18231 a wider mode exists before asking for a move pattern.
18232 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18233 forcing a wider mode to exist.
18234 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18235 returning false if no such mode exists.
18236 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18237 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18238 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18239 Avoid checking for a MODE_INT if we already know the mode is not a
18241 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18242 forcing a wider mode to exist.
18243 (expmed_mult_highpart_optab): Likewise.
18244 (expmed_mult_highpart): Likewise.
18245 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18247 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18248 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18249 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18250 * internal-fn.c (expand_mul_overflow): Update use of
18251 GET_MODE_2XWIDER_MODE.
18252 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18253 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18254 GET_MODE_WIDER_MODE.
18255 (convert_plusminus_to_widen): Likewise.
18256 * tree-switch-conversion.c (array_value_type): Likewise.
18257 * var-tracking.c (emit_note_insn_var_location): Likewise.
18258 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18259 Return false inside rather than outside the loop if no wider mode
18261 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18262 and GET_MODE_2XWIDER_MODE
18263 (can_compare_p): Use else_void.
18264 * gdbhooks.py (OptMachineModePrinter): New class.
18265 (build_pretty_printer): Use it for opt_mode.
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 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18272 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18275 Alan Hayward <alan.hayward@arm.com>
18276 David Sherwood <david.sherwood@arm.com>
18278 * machmode.h (mode_traits): New structure.
18279 (get_narrowest_mode): New function.
18280 (mode_iterator::start): Likewise.
18281 (mode_iterator::iterate_p): Likewise.
18282 (mode_iterator::get_wider): Likewise.
18283 (mode_iterator::get_known_wider): Likewise.
18284 (mode_iterator::get_2xwider): Likewise.
18285 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18286 (FOR_EACH_MODE): Likewise.
18287 (FOR_EACH_MODE_FROM): Likewise.
18288 (FOR_EACH_MODE_UNTIL): Likewise.
18289 (FOR_EACH_WIDER_MODE): Likewise.
18290 (FOR_EACH_2XWIDER_MODE): Likewise.
18291 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18292 * combine.c (simplify_comparison): Likewise
18293 * config/i386/i386.c (type_natural_mode): Likewise.
18294 * cse.c (cse_insn): Likewise.
18295 * dse.c (find_shift_sequence): Likewise.
18296 * emit-rtl.c (init_derived_machine_modes): Likewise.
18297 (init_emit_once): Likewise.
18298 * explow.c (hard_function_value): Likewise.
18299 * expmed.c (extract_fixed_bit_field_1): Likewise.
18300 (extract_bit_field_1): Likewise.
18301 (expand_divmod): Likewise.
18302 (emit_store_flag_1): Likewise.
18303 * expr.c (init_expr_target): Likewise.
18304 (convert_move): Likewise.
18305 (alignment_for_piecewise_move): Likewise.
18306 (widest_int_mode_for_size): Likewise.
18307 (emit_block_move_via_movmem): Likewise.
18308 (copy_blkmode_to_reg): Likewise.
18309 (set_storage_via_setmem): Likewise.
18310 (compress_float_constant): Likewise.
18311 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18312 * optabs-query.c (get_best_extraction_insn): Likewise.
18313 * optabs.c (expand_binop): Likewise.
18314 (expand_twoval_unop): Likewise.
18315 (expand_twoval_binop): Likewise.
18316 (widen_leading): Likewise.
18317 (widen_bswap): Likewise.
18318 (expand_parity): Likewise.
18319 (expand_unop): Likewise.
18320 (prepare_cmp_insn): Likewise.
18321 (prepare_float_lib_cmp): Likewise.
18322 (expand_float): Likewise.
18323 (expand_fix): Likewise.
18324 (expand_sfix_optab): Likewise.
18325 * postreload.c (move2add_use_add2_insn): Likewise.
18326 * reg-stack.c (reg_to_stack): Likewise.
18327 * reginfo.c (choose_hard_reg_mode): Likewise.
18328 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18329 * stor-layout.c (mode_for_size): Likewise.
18330 (smallest_mode_for_size): Likewise.
18331 (mode_for_vector): Likewise.
18332 (finish_bitfield_representative): Likewise.
18333 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18334 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18335 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18336 * var-tracking.c (prepare_call_arguments): Likewise.
18338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18339 Alan Hayward <alan.hayward@arm.com>
18340 David Sherwood <david.sherwood@arm.com>
18342 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18343 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18344 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18345 * machmode.h (mode_size): Move earlier in file.
18346 (mode_precision): Likewise.
18347 (mode_inner): Likewise.
18348 (mode_nunits): Likewise.
18349 (mode_unit_size): Likewise.
18350 (unit_unit_precision): Likewise.
18351 (mode_wider): Likewise.
18352 (mode_2xwider): Likewise.
18353 (machine_mode): New class.
18354 (mode_to_bytes): New function.
18355 (mode_to_bits): Likewise.
18356 (mode_to_precision): Likewise.
18357 (mode_to_inner): Likewise.
18358 (mode_to_unit_size): Likewise.
18359 (mode_to_unit_precision): Likewise.
18360 (mode_to_nunits): Likewise.
18361 (GET_MODE_SIZE): Use mode_to_bytes.
18362 (GET_MODE_BITSIZE): Use mode_to_bits.
18363 (GET_MODE_PRECISION): Use mode_to_precision.
18364 (GET_MODE_INNER): Use mode_to_inner.
18365 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18366 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18367 (GET_MODE_NUNITS): Use mode_to_nunits.
18368 * system.h (ALWAYS_INLINE): New macro.
18369 * config/powerpcspe/powerpcspe-c.c
18370 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18371 int for arg1_mode and arg2_mode.
18373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18374 Alan Hayward <alan.hayward@arm.com>
18375 David Sherwood <david.sherwood@arm.com>
18377 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18378 Prefix mode names with E_ in case statements.
18379 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18380 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18381 (aarch64_split_simd_move): Likewise.
18382 (aarch64_gen_storewb_pair): Likewise.
18383 (aarch64_gen_loadwb_pair): Likewise.
18384 (aarch64_gen_store_pair): Likewise.
18385 (aarch64_gen_load_pair): Likewise.
18386 (aarch64_get_condition_code_1): Likewise.
18387 (aarch64_constant_pool_reload_icode): Likewise.
18388 (get_rsqrte_type): Likewise.
18389 (get_rsqrts_type): Likewise.
18390 (get_recpe_type): Likewise.
18391 (get_recps_type): Likewise.
18392 (aarch64_gimplify_va_arg_expr): Likewise.
18393 (aarch64_simd_container_mode): Likewise.
18394 (aarch64_emit_load_exclusive): Likewise.
18395 (aarch64_emit_store_exclusive): Likewise.
18396 (aarch64_expand_compare_and_swap): Likewise.
18397 (aarch64_gen_atomic_cas): Likewise.
18398 (aarch64_emit_bic): Likewise.
18399 (aarch64_emit_atomic_swap): Likewise.
18400 (aarch64_emit_atomic_load_op): Likewise.
18401 (aarch64_evpc_trn): Likewise.
18402 (aarch64_evpc_uzp): Likewise.
18403 (aarch64_evpc_zip): Likewise.
18404 (aarch64_evpc_ext): Likewise.
18405 (aarch64_evpc_rev): Likewise.
18406 (aarch64_evpc_dup): Likewise.
18407 (aarch64_gen_ccmp_first): Likewise.
18408 (aarch64_gen_ccmp_next): Likewise.
18409 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18410 (alpha_emit_xfloating_libcall): Likewise.
18411 (emit_insxl): Likewise.
18412 (alpha_arg_type): Likewise.
18413 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18414 (arc_preferred_simd_mode): Likewise.
18415 (arc_secondary_reload): Likewise.
18416 (get_arc_condition_code): Likewise.
18417 (arc_print_operand): Likewise.
18418 (arc_legitimate_constant_p): Likewise.
18419 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18420 * config/arc/arc.md (casesi_load): Likewise.
18421 (casesi_compact_jump): Likewise.
18422 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18423 (cc_use_register): Likewise.
18424 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18425 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18426 (arm_init_iwmmxt_builtins): Likewise.
18427 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18428 (neon_expand_vector_init): Likewise.
18429 (arm_attr_length_move_neon): Likewise.
18430 (maybe_get_arm_condition_code): Likewise.
18431 (arm_emit_vector_const): Likewise.
18432 (arm_preferred_simd_mode): Likewise.
18433 (arm_output_iwmmxt_tinsr): Likewise.
18434 (thumb1_output_casesi): Likewise.
18435 (thumb2_output_casesi): Likewise.
18436 (arm_emit_load_exclusive): Likewise.
18437 (arm_emit_store_exclusive): Likewise.
18438 (arm_expand_compare_and_swap): Likewise.
18439 (arm_evpc_neon_vuzp): Likewise.
18440 (arm_evpc_neon_vzip): Likewise.
18441 (arm_evpc_neon_vrev): Likewise.
18442 (arm_evpc_neon_vtrn): Likewise.
18443 (arm_evpc_neon_vext): Likewise.
18444 (arm_validize_comparison): Likewise.
18445 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18446 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18447 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18448 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18449 (c6x_preferred_simd_mode): Likewise.
18450 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18451 (epiphany_rtx_costs): Likewise.
18452 * config/epiphany/predicates.md (proper_comparison_operator):
18454 * config/frv/frv.c (condexec_memory_operand): Likewise.
18455 (frv_emit_move): Likewise.
18456 (output_move_single): Likewise.
18457 (output_condmove_single): Likewise.
18458 (frv_hard_regno_mode_ok): Likewise.
18459 (frv_matching_accg_mode): Likewise.
18460 * config/h8300/h8300.c (split_adds_subs): Likewise.
18461 (h8300_rtx_costs): Likewise.
18462 (h8300_print_operand): Likewise.
18463 (compute_mov_length): Likewise.
18464 (output_logical_op): Likewise.
18465 (compute_logical_op_length): Likewise.
18466 (compute_logical_op_cc): Likewise.
18467 (h8300_shift_needs_scratch_p): Likewise.
18468 (output_a_shift): Likewise.
18469 (compute_a_shift_length): Likewise.
18470 (compute_a_shift_cc): Likewise.
18471 (expand_a_rotate): Likewise.
18472 (output_a_rotate): Likewise.
18473 * config/i386/i386.c (classify_argument): Likewise.
18474 (function_arg_advance_32): Likewise.
18475 (function_arg_32): Likewise.
18476 (function_arg_64): Likewise.
18477 (function_value_64): Likewise.
18478 (ix86_gimplify_va_arg): Likewise.
18479 (ix86_legitimate_constant_p): Likewise.
18480 (put_condition_code): Likewise.
18481 (split_double_mode): Likewise.
18482 (ix86_avx256_split_vector_move_misalign): Likewise.
18483 (ix86_expand_vector_logical_operator): Likewise.
18484 (ix86_split_idivmod): Likewise.
18485 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18486 (ix86_build_const_vector): Likewise.
18487 (ix86_build_signbit_mask): Likewise.
18488 (ix86_match_ccmode): Likewise.
18489 (ix86_cc_modes_compatible): Likewise.
18490 (ix86_expand_branch): Likewise.
18491 (ix86_expand_sse_cmp): Likewise.
18492 (ix86_expand_sse_movcc): Likewise.
18493 (ix86_expand_int_sse_cmp): Likewise.
18494 (ix86_expand_vec_perm_vpermi2): Likewise.
18495 (ix86_expand_vec_perm): Likewise.
18496 (ix86_expand_sse_unpack): Likewise.
18497 (ix86_expand_int_addcc): Likewise.
18498 (ix86_split_to_parts): Likewise.
18499 (ix86_vectorize_builtin_gather): Likewise.
18500 (ix86_vectorize_builtin_scatter): Likewise.
18501 (avx_vpermilp_parallel): Likewise.
18502 (inline_memory_move_cost): Likewise.
18503 (ix86_tieable_integer_mode_p): Likewise.
18504 (x86_maybe_negate_const_int): Likewise.
18505 (ix86_expand_vector_init_duplicate): Likewise.
18506 (ix86_expand_vector_init_one_nonzero): Likewise.
18507 (ix86_expand_vector_init_one_var): Likewise.
18508 (ix86_expand_vector_init_concat): Likewise.
18509 (ix86_expand_vector_init_interleave): Likewise.
18510 (ix86_expand_vector_init_general): Likewise.
18511 (ix86_expand_vector_set): Likewise.
18512 (ix86_expand_vector_extract): Likewise.
18513 (emit_reduc_half): Likewise.
18514 (ix86_emit_i387_round): Likewise.
18515 (ix86_mangle_type): Likewise.
18516 (ix86_expand_round_sse4): Likewise.
18517 (expand_vec_perm_blend): Likewise.
18518 (canonicalize_vector_int_perm): Likewise.
18519 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18520 (expand_vec_perm_1): Likewise.
18521 (expand_vec_perm_interleave3): Likewise.
18522 (expand_vec_perm_even_odd_pack): Likewise.
18523 (expand_vec_perm_even_odd_1): Likewise.
18524 (expand_vec_perm_broadcast_1): Likewise.
18525 (ix86_vectorize_vec_perm_const_ok): Likewise.
18526 (ix86_expand_vecop_qihi): Likewise.
18527 (ix86_expand_mul_widen_hilo): Likewise.
18528 (ix86_expand_sse2_abs): Likewise.
18529 (ix86_expand_pextr): Likewise.
18530 (ix86_expand_pinsr): Likewise.
18531 (ix86_preferred_simd_mode): Likewise.
18532 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18533 * config/i386/sse.md (*andnot<mode>3): Likewise.
18534 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18535 (*<code><mode>3): Likewise.
18536 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18537 (ia64_expand_atomic_op): Likewise.
18538 (ia64_arg_type): Likewise.
18539 (ia64_mode_to_int): Likewise.
18540 (ia64_scalar_mode_supported_p): Likewise.
18541 (ia64_vector_mode_supported_p): Likewise.
18542 (expand_vec_perm_broadcast): Likewise.
18543 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18544 (iq2000_function_arg_advance): Likewise.
18545 (iq2000_function_arg): Likewise.
18546 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18547 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18548 (m68k_libcall_value): Likewise.
18549 (m68k_function_value): Likewise.
18550 (sched_attr_op_type): Likewise.
18551 * config/mcore/mcore.c (mcore_output_move): Likewise.
18552 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18554 (microblaze_function_arg): Likewise.
18555 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18556 (mips_print_operand): Likewise.
18557 (mips_mode_ok_for_mov_fmt_p): Likewise.
18558 (mips_vector_mode_supported_p): Likewise.
18559 (mips_preferred_simd_mode): Likewise.
18560 (mips_expand_vpc_loongson_even_odd): Likewise.
18561 (mips_expand_vec_unpack): Likewise.
18562 (mips_expand_vi_broadcast): Likewise.
18563 (mips_expand_vector_init): Likewise.
18564 (mips_expand_vec_reduc): Likewise.
18565 (mips_expand_msa_cmp): Likewise.
18566 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18567 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18568 (cc_flags_for_mode): Likewise.
18569 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18570 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18571 (nds32_output_casesi_pc_relative): Likewise.
18572 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18573 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18574 (nvptx_gen_unpack): Likewise.
18575 (nvptx_gen_pack): Likewise.
18576 (nvptx_gen_shuffle): Likewise.
18577 (nvptx_gen_wcast): Likewise.
18578 (nvptx_preferred_simd_mode): Likewise.
18579 * config/pa/pa.c (pa_secondary_reload): Likewise.
18580 * config/pa/predicates.md (base14_operand): Likewise.
18581 * config/powerpcspe/powerpcspe-c.c
18582 (altivec_resolve_overloaded_builtin): Likewise.
18583 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18585 (rs6000_preferred_simd_mode): Likewise.
18586 (output_vec_const_move): Likewise.
18587 (rs6000_expand_vector_extract): Likewise.
18588 (rs6000_split_vec_extract_var): Likewise.
18589 (reg_offset_addressing_ok_p): Likewise.
18590 (rs6000_legitimate_offset_address_p): Likewise.
18591 (rs6000_legitimize_address): Likewise.
18592 (rs6000_emit_set_const): Likewise.
18593 (rs6000_const_vec): Likewise.
18594 (rs6000_emit_move): Likewise.
18595 (spe_build_register_parallel): Likewise.
18596 (rs6000_darwin64_record_arg_recurse): Likewise.
18597 (swap_selector_for_mode): Likewise.
18598 (spe_init_builtins): Likewise.
18599 (paired_init_builtins): Likewise.
18600 (altivec_init_builtins): Likewise.
18601 (do_load_for_compare): Likewise.
18602 (rs6000_generate_compare): Likewise.
18603 (rs6000_expand_float128_convert): Likewise.
18604 (emit_load_locked): Likewise.
18605 (emit_store_conditional): Likewise.
18606 (rs6000_output_function_epilogue): Likewise.
18607 (rs6000_handle_altivec_attribute): Likewise.
18608 (rs6000_function_value): Likewise.
18609 (emit_fusion_gpr_load): Likewise.
18610 (emit_fusion_p9_load): Likewise.
18611 (emit_fusion_p9_store): Likewise.
18612 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18613 (fusion_gpr_mem_load): Likewise.
18614 (fusion_addis_mem_combo_load): Likewise.
18615 (fusion_addis_mem_combo_store): Likewise.
18616 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18617 (fusion_gpr_mem_load): Likewise.
18618 (fusion_addis_mem_combo_load): Likewise.
18619 (fusion_addis_mem_combo_store): Likewise.
18620 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18622 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18623 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18624 (rs6000_preferred_simd_mode): Likewise.
18625 (output_vec_const_move): Likewise.
18626 (rs6000_expand_vector_extract): Likewise.
18627 (rs6000_split_vec_extract_var): Likewise.
18628 (reg_offset_addressing_ok_p): Likewise.
18629 (rs6000_legitimate_offset_address_p): Likewise.
18630 (rs6000_legitimize_address): Likewise.
18631 (rs6000_emit_set_const): Likewise.
18632 (rs6000_const_vec): Likewise.
18633 (rs6000_emit_move): Likewise.
18634 (rs6000_darwin64_record_arg_recurse): Likewise.
18635 (swap_selector_for_mode): Likewise.
18636 (paired_init_builtins): Likewise.
18637 (altivec_init_builtins): Likewise.
18638 (rs6000_expand_float128_convert): Likewise.
18639 (emit_load_locked): Likewise.
18640 (emit_store_conditional): Likewise.
18641 (rs6000_output_function_epilogue): Likewise.
18642 (rs6000_handle_altivec_attribute): Likewise.
18643 (rs6000_function_value): Likewise.
18644 (emit_fusion_gpr_load): Likewise.
18645 (emit_fusion_p9_load): Likewise.
18646 (emit_fusion_p9_store): Likewise.
18647 * config/rx/rx.c (rx_gen_move_template): Likewise.
18648 (flags_from_mode): Likewise.
18649 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18650 (s390_slb_comparison): Likewise.
18651 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18652 (s390_vector_mode_supported_p): Likewise.
18653 (s390_cc_modes_compatible): Likewise.
18654 (s390_match_ccmode_set): Likewise.
18655 (s390_canonicalize_comparison): Likewise.
18656 (s390_emit_compare_and_swap): Likewise.
18657 (s390_branch_condition_mask): Likewise.
18658 (s390_rtx_costs): Likewise.
18659 (s390_secondary_reload): Likewise.
18660 (__SECONDARY_RELOAD_CASE): Likewise.
18661 (s390_expand_cs): Likewise.
18662 (s390_preferred_simd_mode): Likewise.
18663 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18664 * config/sh/sh.c (sh_print_operand): Likewise.
18665 (dump_table): Likewise.
18666 (sh_secondary_reload): Likewise.
18667 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18668 * config/sh/sh.md (casesi_worker_1): Likewise.
18669 (casesi_worker_2): Likewise.
18670 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18671 (fcc_comparison_operator): Likewise.
18672 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18673 (emit_soft_tfmode_cvt): Likewise.
18674 (sparc_preferred_simd_mode): Likewise.
18675 (output_cbranch): Likewise.
18676 (sparc_print_operand): Likewise.
18677 (sparc_expand_vec_perm_bmask): Likewise.
18678 (vector_init_bshuffle): Likewise.
18679 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18680 (spu_vector_mode_supported_p): Likewise.
18681 (spu_expand_insv): Likewise.
18682 (spu_emit_branch_or_set): Likewise.
18683 (spu_handle_vector_attribute): Likewise.
18684 (spu_builtin_splats): Likewise.
18685 (spu_builtin_extract): Likewise.
18686 (spu_builtin_promote): Likewise.
18687 (spu_expand_sign_extend): Likewise.
18688 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18689 (tilegx_simd_int): Likewise.
18690 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18691 (tilepro_simd_int): Likewise.
18692 * config/v850/v850.c (const_double_split): Likewise.
18693 (v850_print_operand): Likewise.
18694 (ep_memory_offset): Likewise.
18695 * config/vax/vax.c (vax_rtx_costs): Likewise.
18696 (vax_output_int_move): Likewise.
18697 (vax_output_int_add): Likewise.
18698 (vax_output_int_subtract): Likewise.
18699 * config/visium/predicates.md (visium_branch_operator): Likewise.
18700 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18701 (visium_print_operand_address): Likewise.
18702 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18703 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18704 (xtensa_expand_conditional_branch): Likewise.
18705 (xtensa_copy_incoming_a7): Likewise.
18706 (xtensa_output_literal): Likewise.
18707 * dfp.c (decimal_real_maxval): Likewise.
18708 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18711 Alan Hayward <alan.hayward@arm.com>
18712 David Sherwood <david.sherwood@arm.com>
18714 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18715 (mode_nunits_inline): Likewise.
18716 (mode_inner_inline): Likewise.
18717 (mode_unit_size_inline): Likewise.
18718 (mode_unit_precision_inline): Likewise.
18719 (emit_insn_modes_h): Likewise. Also emit a #define of the
18721 (emit_mode_wider): Add an E_ prefix to mode names.
18722 (emit_mode_complex): Likewise.
18723 (emit_mode_inner): Likewise.
18724 (emit_mode_adjustments): Likewise.
18725 (emit_mode_int_n): Likewise.
18726 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18727 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18728 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18729 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18730 (CRC32_BUILTIN, ENTRY): Likewise.
18731 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18732 (aarch64_pop_regs): Likewise.
18733 (aarch64_process_components): Likewise.
18734 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18735 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18736 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18737 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18738 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18739 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18741 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18742 (pcmpistr): Likewise.
18743 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18744 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18745 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18747 * config/rl78/rl78.c (mduc_regs): Likewise.
18748 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18749 (htm_expand_builtin): Likewise.
18750 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18751 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18752 * config/xtensa/xtensa.c (print_operand): Likewise.
18753 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18754 (NUM_MODE_VECTOR_INT): Likewise.
18755 * genoutput.c (null_operand): Likewise.
18756 (output_operand_data): Likewise.
18757 * genrecog.c (print_parameter_value): Likewise.
18758 * lra.c (debug_operand_data): Likewise.
18760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18762 * dwarf2out.c (scompare_loc_descriptor_wide)
18763 (scompare_loc_descriptor_narrow): New functions, split out from...
18764 (scompare_loc_descriptor): ...here.
18765 * expmed.c (emit_store_flag_int): New function, split out from...
18766 (emit_store_flag): ...here.
18768 2017-08-30 Richard Biener <rguenther@suse.de>
18770 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18771 (dwarf2out_early_finish): Move setting of AT_pubnames from
18772 early debug output to early finish.
18774 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18776 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18777 and -mdata-region to the assembler.
18779 2017-08-30 Richard Biener <rguenther@suse.de>
18781 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18783 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18784 (dwarf2out_early_global_decl): Only generate a DIE for the
18785 abstract origin if it doesn't already exist or is a declaration DIE.
18786 (resolve_addr): Do not add the linkage name twice when
18787 generating a stub DIE for the DW_TAG_GNU_call_site target.
18789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18791 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18792 Use machine_mode rather than int for arg1_mode.
18794 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18797 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18798 that the second argument of the built-in functions to unpack
18799 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18800 switch statement instead a lot of if statements.
18801 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18802 Allow 64-bit values to be in Altivec registers as well as
18803 traditional floating point registers.
18804 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18806 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18808 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18809 MAX_REGS_PER_ADDRESS == 1.
18811 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18813 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18814 * config/i386/i386.c (ix86_option_override_internal): Simplify
18815 setting of opts->x_flag_entry.
18817 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18818 Jakub Jelinek <jakub@redhat.com>
18819 Richard Biener <rguenther@suse.de>
18821 PR tree-optimization/81503
18822 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18823 folded constant fits in the target type; reorder tests for clarity.
18825 2017-08-29 Martin Liska <mliska@suse.cz>
18827 * passes.def: Include pass_lower_switch.
18828 * stmt.c (dump_case_nodes): Remove and move to
18829 tree-switch-conversion.
18830 (case_values_threshold): Likewise.
18831 (expand_switch_as_decision_tree_p): Likewise.
18832 (emit_case_decision_tree): Likewise.
18833 (expand_case): Likewise.
18834 (balance_case_nodes): Likewise.
18835 (node_has_low_bound): Likewise.
18836 (node_has_high_bound): Likewise.
18837 (node_is_bounded): Likewise.
18838 (emit_case_nodes): Likewise.
18839 (struct simple_case_node): New struct.
18840 (add_case_node): Remove.
18841 (emit_case_dispatch_table): Use vector instead of case_list.
18842 (reset_out_edges_aux): Remove.
18843 (compute_cases_per_edge): Likewise.
18844 (expand_case): Build list of simple_case_node.
18845 (expand_sjlj_dispatch_table): Use it.
18846 * tree-switch-conversion.c (struct case_node): Moved from
18847 stmt.c and adjusted.
18848 (emit_case_nodes): Likewise.
18849 (node_has_low_bound): Likewise.
18850 (node_has_high_bound): Likewise.
18851 (node_is_bounded): Likewise.
18852 (case_values_threshold): Likewise.
18853 (reset_out_edges_aux): Likewise.
18854 (compute_cases_per_edge): Likewise.
18855 (add_case_node): Likewise.
18856 (dump_case_nodes): Likewise.
18857 (balance_case_nodes): Likewise.
18858 (expand_switch_as_decision_tree_p): Likewise.
18859 (emit_jump): Likewise.
18860 (emit_case_decision_tree): Likewise.
18861 (try_switch_expansion): Likewise.
18862 (do_jump_if_equal): Likewise.
18863 (emit_cmp_and_jump_insns): Likewise.
18864 (fix_phi_operands_for_edge): New function.
18865 (record_phi_operand_mapping): Likewise.
18866 (class pass_lower_switch): New pass.
18867 (pass_lower_switch::execute): New function.
18868 (make_pass_lower_switch): Likewise.
18869 (conditional_probability):
18870 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
18871 * tree-pass.h: Add make_pass_lower_switch.
18873 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18876 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
18879 2017-08-29 Richard Biener <rguenther@suse.de>
18881 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
18882 we do not add a DW_AT_inline attribute twice.
18883 (gen_subprogram_die): Remove code setting DW_AT_inline on
18884 DECL_ABSTRACT_P nodes.
18886 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
18888 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
18889 calls to internal functions.
18890 (gimplify_modify_expr): Likewise.
18891 * tree-call-cdce.c (use_internal_fn): Likewise.
18892 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
18893 (convert_to_divmod): Set the nothrow flag.
18894 * tree-if-conv.c (predicate_mem_writes): Likewise.
18895 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18896 (vectorizable_call): Likewise.
18897 (vectorizable_store): Likewise.
18898 (vectorizable_load): Likewise.
18899 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
18900 (vect_recog_mask_conversion_pattern): Likewise.
18902 2017-08-29 Martin Liska <mliska@suse.cz>
18905 * gcc.c (driver_handle_option): Add new argument.
18906 * opts-common.c (handle_option): Pass
18907 target_option_override_hook.
18908 * opts-global.c (lang_handle_option): Add new option.
18909 (set_default_handlers): Add new argument.
18910 (decode_options): Likewise.
18911 * opts.c (target_handle_option): Likewise.
18912 (common_handle_option): Call target_option_override_hook.
18913 * opts.h (struct cl_option_handler_func): Add hook for
18914 target option override.
18915 (struct cl_option_handlers): Likewise.
18916 (set_default_handlers): Add new argument.
18917 (decode_options): Likewise.
18918 (common_handle_option): Likewise.
18919 (target_handle_option): Likewise.
18920 * toplev.c (toplev::main): Pass targetm.target_option.override
18923 2017-08-29 Richard Biener <rguenther@suse.de>
18924 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
18926 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
18927 life to the active subtree.
18929 2017-08-28 Jeff Law <law@redhat.com>
18931 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
18932 derive_equivalences.
18933 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
18934 Code moved into....
18935 (edge_info::derive_equivalences): New private member function
18937 * tree-ssa-dom.c (class edge_info): Changed from a struct
18938 to a class. Add ctor/dtor, methods and data members.
18939 (edge_info::edge_info): Renamed from allocate_edge_info.
18940 Initialize additional members.
18941 (edge_info::~edge_info): New.
18942 (free_dom_edge_info): Delete the edge info.
18943 (record_edge_info): Use new class & associated member functions.
18944 Tighten forms for testing for edge equivalences.
18945 (record_temporary_equivalences): Iterate over the simple
18946 equivalences rather than assuming there's only one per edge.
18947 (cprop_into_successor_phis): Iterate over the simple
18948 equivalences rather than assuming there's only one per edge.
18949 (optimize_stmt): Use operand_equal_p rather than pointer
18950 equality for mini-DSE code.
18952 2017-08-28 Nathan Sidwell <nathan@acm.org>
18954 * gcc.c (execute): Fold SIGPIPE handling into switch
18955 statement. Adjust internal error message.
18957 2017-08-28 Richard Biener <rguenther@suse.de>
18960 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
18961 Do nothing for removed DIEs.
18963 2017-08-28 Richard Biener <rguenther@suse.de>
18965 PR tree-optimization/81977
18966 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
18969 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
18972 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
18973 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
18974 using targetm.gen_mem_thread_fence.
18976 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
18979 * config/i386/i386.md (*<btsc><mode>): Change operand 2
18980 predicate to register_operand. Reorder operands.
18981 (*btr<mode>): Ditto.
18982 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
18983 (*btr<mode>_mask): Ditto.
18985 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
18987 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
18988 * config/rs6000/xmmintrin.h: New file.
18989 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
18991 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18994 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
18995 parameter and_insn and return it.
18996 (recombine_lvx_pattern): Insert a copy to ensure availability of
18997 the base register of the copied masking operation at the point of
18998 the instruction replacement.
18999 (recombine_stvx_pattern): Likewise.
19001 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19003 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19004 undocumented switches.
19005 (-mpower9-dform-vector): Likewise.
19006 (-mpower9-dform): Likewise.
19007 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19008 comments to delete references to -mpower9-dform* switches.
19009 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19010 Delete reference to -mpower9-dform* switches, test for
19011 -mpower9-vector instead.
19012 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19013 (OTHER_P9_VECTOR_MASKS): Likewise.
19014 (POWERPC_MASKS): Likewise.
19015 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19016 tests against -mpower9-dform* to -mpower9-vector. Delete code
19017 that checked for -mpower9-dform* consistancy with other options.
19018 Add test for -mpower9-misc to enable other power9 switches.
19019 (rs6000_init_hard_regno_mode_ok): Likewise.
19020 (rs6000_option_override_internal): Likewise.
19021 (rs6000_emit_prologue): Likewise.
19022 (rs6000_emit_epilogue): Likewise.
19023 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19024 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19025 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19027 (emit_fusion_p9_store): Likewise.
19028 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19029 resetting these macros if the assembler does not support ISA 3.0
19031 (TARGET_P9_DFORM_VECTOR): Likewise.
19032 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19033 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19035 2017-08-25 Alan Modra <amodra@gmail.com>
19037 PR rtl-optimization/81747
19038 * cse.c (cse_extended_basic_block): Don't attempt to record
19039 equivalences for degenerate conditional branches that branch
19040 to their fall-through.
19042 2017-08-24 Martin Sebor <msebor@redhat.com>
19044 PR middle-end/81908
19045 * gimple-fold.c (size_must_be_zero_p): New function.
19046 (gimple_fold_builtin_memory_op): Call it.
19048 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19050 * config/rs6000/mm_malloc.h: New file.
19052 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19054 PR tree-optimization/81913
19055 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19056 analysis when either IVs in condition can wrap.
19058 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19060 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19061 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19063 2017-08-24 Richard Biener <rguenther@suse.de>
19066 * targhooks.c (default_target_can_inline_p): Properly
19067 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19068 is present and always compare.
19069 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19070 infer -mfpmath=sse from TARGET_SSE_P.
19071 (ix86_can_inline_p): Properly use target_option_default_node when
19072 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19074 2017-08-24 Richard Biener <rguenther@suse.de>
19077 * dwarf2out.c (output_die): Handle flag_generate_offload like
19079 (output_comp_unit): Likewise.
19080 (gen_array_type_die): Likewise.
19081 (dwarf2out_early_finish): Likewise.
19082 (note_variable_value_in_expr): Likewise.
19083 (dwarf2out_finish): Likewise. Adjust assert.
19084 * cgraphunit.c (symbol_table::compile): Move setting of
19085 flag_generate_offload earlier ...
19086 (symbol_table::finalize_compilation_unit): ... here, before
19087 early debug finalization.
19089 2017-08-24 Richard Biener <rguenther@suse.de>
19091 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19092 and ipa-fnsummary.h.
19093 (ix86_can_inline_p): When ix86_fpmath flags do not match
19094 check whether the callee uses FP math at all.
19096 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19098 PR middle-end/81931
19099 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19100 instead of TYPE_PRECISION.
19102 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19103 Alan Hayward <alan.hayward@arm.com>
19104 David Sherwood <david.sherwood@arm.com>
19106 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19108 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19110 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19112 (ix86_frame::stack_realign_allocate): New field.
19113 (struct machine_frame_state): Modify comments.
19114 (machine_frame_state::sp_realigned_fp_end): New field.
19115 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19116 layout calculation.
19117 (sp_valid_at): Add assertion to assure no attempt to access invalid
19118 offset of a realigned stack.
19119 (fp_valid_at): Likewise.
19120 (choose_baseaddr): Modify comments.
19121 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19122 ix86_expand_prologue.
19123 (ix86_expand_prologue): Modify stack realignment and allocation.
19124 (ix86_expand_epilogue): Modify comments.
19125 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19126 avx2_runtime, avx512f, and avx512f_runtime.
19128 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19130 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19131 (mstackrealign): Do not init to -1.
19132 * config/i386/i386.c (ix86_option_override_internal):
19133 Check opts_set, not opts when setting default value of
19134 opts->x_ix86_force_align_arg_pointer.
19136 2017-08-23 Richard Biener <rguenther@suse.de>
19138 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19139 lang_hooks.decl_printable_name.
19140 * print-rtl-function.c (print_rtx_function): Likewise.
19141 * tree-pretty-print.c (dump_function_header): Likewise.
19143 2017-08-23 Richard Biener <rguenther@suse.de>
19146 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19147 -g0 at compile-time.
19149 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19151 PR middle-end/19706
19152 * doc/sourcebuild.texi (Other hardware attributes):
19155 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19157 PR middle-end/19706
19158 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19159 Fix single-use check.
19161 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19163 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19165 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19167 * doc/install.texi: Modify to add more details on running selected
19170 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19172 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19173 is combined with -mabi=ms.
19174 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19177 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19179 PR tree-optimization/81488
19180 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19181 and cached_basis fields.
19182 (MAX_SPREAD): New constant.
19183 (alloc_cand_and_find_basis): Initialize new fields.
19184 (clear_visited): New function.
19185 (create_phi_basis_1): Rename from create_phi_basis, set visited
19186 and cached_basis fields.
19187 (create_phi_basis): New wrapper function.
19188 (phi_add_costs_1): Rename from phi_add_costs, add spread
19189 parameter, set visited field, short-circuit when limits reached.
19190 (phi_add_costs): New wrapper function.
19191 (record_phi_increments_1): Rename from record_phi_increments, set
19193 (record_phi_increments): New wrapper function.
19194 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19195 (phi_incr_cost): New wrapper function.
19196 (all_phi_incrs_profitable_1): Rename from
19197 all_phi_incrs_profitable, set visited field.
19198 (all_phi_incrs_profitable): New wrapper function.
19200 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19201 Alan Hayward <alan.hayward@arm.com>
19202 David Sherwood <david.sherwood@arm.com>
19204 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19205 that takes the outer and inner modes.
19206 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19207 comparison as the canonical test for a paradoxical subreg.
19208 * combine.c (simplify_set): Use paradoxical_subreg_p.
19209 (make_extraction): Likewise.
19210 (force_to_mode): Likewise.
19211 (rtx_equal_for_field_assignment_p): Likewise.
19212 (gen_lowpart_for_combine): Likewise.
19213 (simplify_comparison): Likewise.
19214 * cse.c (equiv_constant): Likewise.
19215 * expmed.c (store_bit_field_1): Likewise.
19216 * final.c (alter_subreg): Likewise.
19217 * fwprop.c (propagate_rtx): Likewise.
19218 (forward_propagate_subreg): Likewise.
19219 * ira-conflicts.c (ira_build_conflicts): Likewise.
19220 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19221 * lra-constraints.c (curr_insn_transform): Likewise.
19222 (split_reg): Likewise.
19223 * lra-eliminations.c (move_plus_up): Likewise.
19224 (lra_eliminate_regs_1): Likewise.
19225 * recog.c (general_operand): Likewise.
19226 * ree.c (combine_reaching_defs): Likewise.
19227 * reload.c (push_reload): Likewise.
19228 (find_reloads): Likewise.
19229 * reload1.c (elimination_effects): Likewise.
19230 (compute_reload_subreg_offset): Likewise.
19231 (choose_reload_regs): Likewise.
19232 * rtlanal.c (subreg_lsb_1): Likewise.
19233 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19234 (simplify_subreg): Likewise.
19235 * var-tracking.c (track_loc_p): Likewise.
19236 * emit-rtl.c (byte_lowpart_offset): Likewise.
19237 (paradoxical_subreg_p): Delete out-of-line definition.
19239 2017-08-22 Jeff Law <law@redhat.com>
19241 PR tree-optimization/81741
19242 PR tree-optimization/71947
19243 * tree-ssa-dom.c: Include tree-inline.h.
19244 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19245 equivalences if one is more expensive to compute than the other.
19246 * tree-ssa-scopedtables.h (class const_or_copies): Make
19247 record_const_or_copy_raw method private.
19248 (class avail_exprs_stack): New method simplify_binary_operation.
19249 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19250 avail_exprs_stack::simplify_binary_operation as needed.
19251 (avail_exprs_stack::simplify_binary_operation): New function.
19253 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19255 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19256 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19257 (DOT_SYMBOLS): Likewise.
19258 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19259 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19260 (RS6000_ABI_NAME): Likewise.
19261 (TARGET_CMODEL): Likewise.
19262 (TOC_SECTION_ASM_OP): Likewise.
19263 (SET_CMODEL): New macro.
19264 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19266 2017-08-22 Richard Biener <rguenther@suse.de>
19268 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19269 to allow for free-lang-data replacements similar to verify_type_variant.
19271 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19272 Alan Hayward <alan.hayward@arm.com>
19273 David Sherwood <david.sherwood@arm.com>
19275 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19276 VOIDmode for the LABEL_REF.
19278 2017-08-22 Richard Biener <rguenther@suse.de>
19280 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19282 2017-08-22 Richard Biener <rguenther@suse.de>
19284 * common.opt (feliminate-dwarf2-dups): Ignore.
19285 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19286 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19287 same_die_p_wrap, compute_section_prefix,
19288 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19289 (comdat_symbol_id, comdat_symbol_number): Likewise.
19290 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19292 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19293 (output_die): Mark unreachable path unreachable.
19294 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19295 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19296 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19297 (dwarf2out_early_finish): Likewise.
19299 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19301 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19303 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19306 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19307 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19308 (avr_attribute_table) <io, io_low, address>: Initialize
19309 .decl_required with true.
19311 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19313 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19314 undocumented debugging options.
19315 (-mvsx-scalar-double): Likewise.
19316 (-mallow-df-permute): Likewise.
19317 (-mvectorize-builtins): Likewise.
19318 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19319 (rs6000_builtin_vectorized_function): Likewise.
19320 (rs6000_builtin_md_vectorized_function): Likewise.
19321 (rs6000_opt_vars): Likewise.
19323 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19326 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19327 (*btrq_imm): Rename from *btrq.
19328 (*btcq_imm): Rename from *btcq.
19329 (btsc): New code attribute.
19330 (*<btsc><mode>): New insn pattern.
19331 (*btr<mode>): Ditto.
19332 (*<btsc><mode>_mask): New insn_and_split pattern.
19333 (*btr<mode>_mask): Ditto.
19335 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19336 Alan Hayward <alan.hayward@arm.com>
19337 David Sherwood <david.sherwood@arm.com>
19339 * function.c (pad_below): Simplify padding calculation.
19341 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19342 Alan Hayward <alan.hayward@arm.com>
19343 David Sherwood <david.sherwood@arm.com>
19345 * target.def (function_prologue): Remove frame size argument.
19346 (function_epilogue): Likewise.
19347 * doc/tm.texi: Regenerate.
19348 * final.c (final_start_function): Update call to function_prologue.
19349 (final_end_function): Update call to function_epilogue.
19350 (default_function_pro_epilogue): Remove frame size argument.
19351 * output.h (default_function_pro_epilogue): Likewise.
19352 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19353 (arm_output_function_prologue): Likewise.
19354 * config/frv/frv.c (frv_function_prologue): Likewise.
19355 (frv_function_epilogue): Likewise.
19356 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19357 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19358 (ia64_output_function_epilogue): Likewise.
19359 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19360 (m32r_output_function_epilogue): Likewise.
19361 * config/microblaze/microblaze.c (microblaze_function_prologue)
19362 (microblaze_function_epilogue): Likewise.
19363 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19364 (mips_output_function_epilogue): Likewise.
19365 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19366 (mmix_target_asm_function_epilogue): Likewise.
19367 * config/msp430/msp430.c (msp430_start_function): Likewise.
19368 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19369 (nds32_asm_function_epilogue): Likewise.
19370 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19371 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19372 (pa_output_function_epilogue): Likewise.
19373 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19374 (rs6000_output_function_epilogue): Likewise.
19375 * config/rl78/rl78.c (rl78_start_function): Likewise.
19376 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19377 (rs6000_output_function_epilogue): Likewise.
19378 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19379 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19380 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19381 (sparc_asm_function_epilogue): Likewise.
19383 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19385 * tree.h (type_has_mode_precision_p): New function.
19386 * convert.c (convert_to_integer_1): Use it.
19387 * expr.c (expand_expr_real_2): Likewise.
19388 (expand_expr_real_1): Likewise.
19389 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19390 * match.pd: Likewise.
19391 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19392 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19393 * tree-tailcall.c (process_assignment): Likewise.
19394 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19395 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19396 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19397 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19398 (vectorizable_assignment): Likewise.
19399 (vectorizable_shift): Likewise.
19400 (vectorizable_operation): Likewise.
19401 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19403 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19405 * match.pd: Add pow (C, x) simplification.
19407 2017-08-21 Richard Biener <rguenther@suse.de>
19409 PR tree-optimization/81900
19410 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19411 for blocks with abnormal predecessors.
19412 (compute_antic): Do not set visited flag prematurely.
19414 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19417 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19419 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19421 * stor-layout.h (vector_type_mode): Move to...
19422 * tree.h (vector_type_mode): ...here.
19423 * stor-layout.c (vector_type_mode): Move to...
19424 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19426 2017-08-21 Richard Biener <rguenther@suse.de>
19428 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19429 register_external_die hooks.
19430 (debug_false_tree_charstarstar_uhwistar): Declare.
19431 (debug_nothing_tree_charstar_uhwi): Likewise.
19432 * debug.c (do_nothing_debug_hooks): Adjust.
19433 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19434 (debug_nothing_tree_charstar_uhwi): Likewise.
19435 * dbxout.c (dbx_debug_hooks): Adjust.
19436 (xcoff_debug_hooks): Likewise.
19437 * sdbout.c (sdb_debug_hooks): Likewise.
19438 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19439 * dwarf2out.c (macinfo_label_base): New global.
19440 (dwarf2out_register_external_die): New function for the
19441 register_external_die hook.
19442 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19443 (dwarf2_debug_hooks): Use them.
19444 (dwarf2_lineno_debug_hooks): Adjust.
19445 (struct die_struct): Add with_offset flag.
19446 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19447 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19448 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19449 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19450 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19451 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19452 defining section names for the early LTO debug variants.
19453 (reset_indirect_string): New helper.
19454 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19455 (print_dw_val): Add support for offsetted symbol references.
19456 (get_ultimate_context): Split out from is_cxx.
19457 (is_cxx): Use get_ultimate_context.
19458 (is_fortran): Add decl overload.
19459 (compute_comp_unit_symbol): Split out worker from
19460 compute_section_prefix.
19461 (compute_section_prefix): Call compute_comp_unit_symbol and
19462 set comdat_type_p here.
19463 (output_die): Skip DIE symbol output for the LTO added one.
19464 Handle DIE symbol references with offset.
19465 (output_comp_unit): Guard section name mangling properly.
19466 For LTO debug sections emit a symbol at the section beginning
19467 which we use to refer to its DIEs.
19468 (add_abstract_origin_attribute): For DIEs registered via
19469 dwarf2out_register_external_die directly refer to the early
19470 DIE rather than indirectly through the shadow one we created.
19471 Remove obsolete call to dwarf2out_abstract_function for
19472 non-function/block origins.
19473 (gen_array_type_die): When generating early LTO debug do
19474 not emit DW_AT_string_length.
19475 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19476 late when in LTO. As suggested place a gcc_unreachable for
19477 the DECL_ABSTRACT_P case.
19478 (gen_subprogram_die): Avoid another specification DIE
19479 for early built declarations/definitions for the late LTO case.
19480 (gen_variable_die): Add type references for late duplicated VLA dies
19482 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19483 we have the abstract instance already.
19484 (process_scope_var): Adjust decl DIE contexts in LTO which
19485 first puts them in limbo.
19486 (gen_decl_die): Do not generate type DIEs late apart from
19487 types for VLAs or for decls we do not yet have a DIE. Do not
19488 call dwarf2out_abstract_function late.
19489 (dwarf2out_early_global_decl): Make sure to create DIEs
19490 for abstract instances of a decl first.
19491 (dwarf2out_late_global_decl): Adjust comment.
19492 (output_macinfo_op): With multiple macro sections use
19493 macinfo_label_base to distinguish labels.
19494 (output_macinfo): Likewise. Update macinfo_label_base.
19495 Pass in the line info label.
19496 (note_variable_value_in_expr): When generating LTO resolve
19497 all variable values here by generating DIEs as needed.
19498 (init_sections_and_labels): Add early LTO debug flag parameter
19499 and generate different sections and names if set. Add generation
19500 counter for the labels so we can have multiple of them.
19501 (reset_dies): Helper to allow DIEs to be output multiple times.
19502 (dwarf2out_finish): When outputting DIEs to the fat part of an
19503 LTO object first reset DIEs.
19504 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19505 (modified_type_die): Check for decl_ultimate_origin being self
19507 (gen_type_die_with_usage): Likewise.
19508 (gen_typedef_die): Allow decl_ultimate_origin being self.
19509 (set_decl_abstract_flags): Remove.
19510 (set_block_abstract_flags): Likewise.
19511 (dwarf2out_abstract_function): Treat the early generated DIEs
19512 as the abstract copy and only add DW_AT_inline and
19513 DW_AT_artificial here and call set_decl_origin_self.
19514 If the DIE has an abstract origin don't do anything.
19515 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19516 if we have none yet (Go fails to build one, PR78628).
19517 (variably_modified_type_p): Prevent endless recursion for Ada
19518 cyclic pointer types.
19519 * lto-streamer-in.c: Include debug.h.
19520 (dref_queue): New global.
19521 (lto_read_tree_1): Stream in DIE references.
19522 (lto_input_tree): Register DIE references.
19523 (input_function): Stream DECL_DEBUG_ARGS.
19524 * lto-streamer-out.c: Include debug.h.
19525 (lto_write_tree_1): Output DIE references.
19526 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19527 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19528 (output_function): Stream DECL_DEBUG_ARGS.
19529 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19530 Stream DECL_ABSTRACT_ORIGIN.
19531 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19532 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19533 DECL_CONTEXT for file-scope decls.
19534 * lto-streamer.h (struct dref_entry): Declare.
19535 (dref_queue): Likewise.
19536 * cfgexpand.c (pass_expand::execute): Do not call the
19537 outlining_inline_function hook here.
19538 * lto-wrapper.c (debug_obj): New global.
19539 (tool_cleanup): Unlink it if required.
19540 (debug_objcopy): New function.
19541 (run_gcc): Handle early debug sections in the IL files by
19542 extracting them to separate files, partially linkin them and
19543 feeding the result back as result to the linker.
19544 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19545 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19546 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19547 sections into a separate segment.
19548 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19550 (darwin_asm_dwarf_section): Likewise.
19551 (darwin_asm_output_dwarf_offset): Likewise.
19552 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19554 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19555 Alan Hayward <alan.hayward@arm.com>
19556 David Sherwood <david.sherwood@arm.com>
19558 * read-md.h (md_reader::record_potential_iterator_use): Replace
19559 pointer argument with an rtx and an index.
19560 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19561 (apply_mode_iterator): Likewise.
19562 (apply_code_iterator): Likewise.
19563 (apply_int_iterator): Likewise.
19564 (apply_subst_iterator): Likewise.
19565 (record_iterator_use): Likewise.
19566 (record_attribute_use): Likewise.
19567 (md_reader::record_potential_iterator_use): Likewise. Update calls
19568 to record_iterator_use and apply_iterator.
19569 (iterator_use): Replace ptr with x and index.
19570 (attribute_use): Likewise.
19571 (apply_attribute_uses): Update calls to apply_iterator.
19572 (apply_iterators): Likewise. Update initialization of iterator_use.
19573 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19574 and record_potential_iterator_use.
19575 (rtx_reader::read_rtx_operand): Likewise.
19577 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19578 Alan Hayward <alan.hayward@arm.com>
19579 David Sherwood <david.sherwood@arm.com>
19581 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19584 2017-08-21 Richard Biener <rguenther@suse.de>
19586 PR middle-end/81884
19587 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19588 at struct end conservatively when comparing common bases.
19590 2017-08-21 Richard Biener <rguenther@suse.de>
19592 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19593 (mem_ref_in_stmt): Remove.
19594 (determine_max_movement): Use ref index to get at the reference.
19595 (invariantness_dom_walker::before_dom_children): Deal with
19596 lim data already initialized.
19597 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19599 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19601 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19602 (TARGET_ISA_ROUND): Ditto.
19603 (TARGET_ROUND): Ditto.
19604 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19605 * config/i386/i386.md: Ditto.
19606 * config/i386/sse.md: Ditto.
19607 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19608 with OPTION_MASK_ISA_SSE4_1.
19610 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19613 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19614 __builtin_ia32_lzcnt_u16.
19616 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19619 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19620 (rs6000_set_current_function): Rewrite function to use it.
19622 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19625 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19626 and TYPE_WARN_IF_NOT_ALIGN.
19627 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19628 (handle_warn_if_not_align): New.
19629 (place_union_field): Call handle_warn_if_not_align.
19630 (place_field): Call handle_warn_if_not_align.
19631 Copy TYPE_WARN_IF_NOT_ALIGN.
19632 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19633 (layout_type): Likewise.
19634 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19636 (tree_decl_common): Add warn_if_not_align.
19637 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19638 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19639 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19640 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19641 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19642 * doc/extend.texi: Document warn_if_not_aligned attribute.
19643 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19645 2017-08-17 Martin Liska <mliska@suse.cz>
19648 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19649 (get_data_dependence): Use it as pointer type.
19650 (distribute_loop): Likewise.
19652 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19654 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19655 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19656 (p8_vmrgow_<mode>_direct): New define_insn.
19657 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19658 handle endianness for vmrgew and vmrgow permute patterns.
19660 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19662 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19663 * config/rs6000/rs6000-cpus.def: Remove comment.
19664 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19665 (POWERPC_MASKS): Likewise.
19666 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19667 use of TARGET_VSX_TIMODE.
19668 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19669 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19670 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19671 (rs6000_option_override_internal): Remove dead code.
19672 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19673 (rs6000_legitimize_reload_address): Likewise.
19674 (rs6000_legitimate_address_p): Likewise.
19675 (rs6000_opt_masks): Delete "vsx-timode".
19676 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19677 from function comment.
19678 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19679 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19680 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19682 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19683 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19684 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19685 (VSX_TI): Likewise.
19687 (define_peephole2): Likewise.
19689 2017-08-17 Martin Sebor <msebor@redhat.com>
19692 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19693 past the end of an array.
19694 (test_pp_format): Add test cases.
19696 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19698 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19699 missing ECF_NOTHROW flags.
19701 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19704 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19705 operands residing in integer registers.
19706 (*vsx_le_perm_load_<mode>): Likewise.
19707 (*vsx_le_perm_store_<mode>): Likewise.
19708 (define_peephole2): Add peepholes to optimize the above.
19710 2017-08-17 Marek Polacek <polacek@redhat.com>
19712 PR middle-end/81814
19713 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19714 to mimic what shorten_compare did. Change the return type to bool.
19715 (fold_cond_expr_with_comparison): Update call to
19716 operand_equal_for_comparison_p.
19717 (fold_ternary_loc): Likewise.
19719 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19721 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19723 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19725 2017-08-17 Richard Biener <rguenther@suse.de>
19727 * tree-ssa-structalias.c (solve_graph): When propagating
19728 to successors update the graphs succ edges and avoid duplicate work.
19730 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19733 * config/i386/i386.c (ix86_option_override_internal): Save target
19734 specific options after ix86_stack_protector_guard_reg was changed.
19736 2017-08-17 Richard Biener <rguenther@suse.de>
19738 PR tree-optimization/81827
19739 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19740 (new_var_info): Initialize it conservatively.
19741 (get_call_vi): Mark register vars.
19742 (new_scalar_tmp_constraint_exp): Likewise.
19743 (handle_rhs_call): Likewise.
19744 (handle_const_call): Likewise.
19745 (create_function_info_for): Likewise.
19746 (solve_constraints): Sort varinfos to separate register from
19747 non-register vars to pack points-to solution bitmaps during
19750 2017-08-17 Marek Polacek <polacek@redhat.com>
19752 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19754 2017-08-17 Richard Biener <rguenther@suse.de>
19756 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19757 to true when overflow is undefined and we saturated the result.
19759 2017-08-17 Alan Modra <amodra@gmail.com>
19762 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19763 Don't use store multiple if only one reg needs saving.
19764 (interesting_frame_related_regno): New function.
19765 (rs6000_frame_related): Don't emit frame info for regs that
19767 (rs6000_emit_epilogue): Likewise.
19769 2017-08-16 Nathan Sidwell <nathan@acm.org>
19771 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19772 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19773 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19774 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19775 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19776 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19777 (verify_type): Adjust for TYPE_BINFO move.
19778 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19779 process TYPE_BINFO directly.
19780 (hash_tree): Likewise.
19781 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19783 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19786 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19788 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19790 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19793 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19794 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19795 Add ix86_binary_operator_ok to insn constraint.
19796 (*iordi_1_bts): Ditto.
19797 (*xordi_1_btc): Ditto.
19798 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19799 Update corresponding peephole2 pattern.
19803 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19805 PR tree-optimization/81832
19806 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19807 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19809 2017-08-16 Marek Polacek <polacek@redhat.com>
19812 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19813 regarding size_zero_node.
19815 2017-08-16 Martin Liska <mliska@suse.cz>
19818 * config.gcc: Respect previously set extra_objs in case
19821 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19823 PR tree-optimization/81835
19824 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19825 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19826 not depend on the phi.
19828 2017-08-16 Alan Modra <amodra@gmail.com>
19830 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19833 2017-08-16 Alan Modra <amodra@gmail.com>
19835 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19836 (save_reg_p): ..into this. Update all callers.
19837 (first_reg_to_save): Simplify.
19839 2017-08-16 Alan Modra <amodra@gmail.com>
19841 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19844 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19848 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19849 constant count if that count is less than 32.
19851 2017-08-15 Nathan Sidwell <nathan@acm.org>
19853 * gcc.c (execute): Emit friendlier message if inferior is killed
19854 by an external cause.
19856 2017-08-15 Richard Biener <rguenther@suse.de>
19858 PR tree-optimization/81790
19859 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
19860 CONSTRUCTORs from simplifying and VN.
19862 2017-08-14 Martin Sebor <msebor@redhat.com>
19864 * builtin-attrs.def: Add comments.
19866 2017-08-14 Martin Sebor <msebor@redhat.com>
19869 * doc/extend.texi (attribute nonstring): Document new attribute.
19871 2017-08-14 Martin Sebor <msebor@redhat.com>
19874 * tree-diagnostic.c (default_tree_printer): Handle %G.
19875 * gimple-pretty-print.h (percent_G_format): Declare new function.
19876 * gimple-pretty-print.c (percent_G_format): Define.
19877 * tree-pretty-print.c (percent_K_format): Add argument.
19879 2017-08-14 Martin Sebor <msebor@redhat.com>
19881 PR translation/79998
19882 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
19883 Remove a stray space.
19885 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
19888 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
19889 (*iordi_1_bts): Ditto.
19890 (*xordi_1_btc): Ditto.
19892 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19895 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
19896 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19898 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
19899 quoted strings, and make more translator-friendly.
19900 (darwin_rs6000_override_options): Likewise.
19901 (rs6000_option_override_internal): Likewise.
19902 (rs6000_return_in_memory): Fix overlong line.
19903 (init_cmulative_args): Use quoted strings, and make more
19904 translator-friendly.
19905 (rs6000_pass_by_reference): Fix overlong line.
19906 (def_builtin): Use quoted strings.
19907 (altivec_expand_predicate_builtin): Use quoted strings, and make
19908 more translator-friendly.
19909 (htm_expand_builtin): Use quoted strings.
19910 (cpu_expand_builtin): Use quoted strings, and make more
19911 translator-friendly.
19912 (altivec_expand_builtin): Likewise.
19913 (paired_expand_predicate_builtin): Likewise.
19914 (rs6000_invalid_builtin): Likewise.
19915 (builtin_function_type): Use quoted strings.
19916 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
19917 more translator-friendly.
19918 (rs6000_trampoline_init): Likewise.
19919 (rs6000_handle_altivec_attribute): Likewise.
19920 (rs6000_inner_target_options): Use quoted strings.
19921 (rs6000_disable_incompatible_switches): Likewise.
19922 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
19923 strings, and make more translator-friendly.
19924 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
19926 2017-08-14 Bin Cheng <bin.cheng@arm.com>
19928 PR tree-optimization/81799
19929 * tree-loop-distribution.c (version_loop_by_alias_check): Force
19930 cond_expr to simple gimple operand.
19932 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
19934 PR middle-end/46932
19935 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
19937 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
19941 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
19942 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
19943 TARGET_GASISR_PROLOGUES.
19944 * config/avr/avr.c (avr_option_override): Same.
19945 (avr_pass_pre_proep::execute): Same.
19947 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
19950 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
19951 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
19953 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
19955 * config/i386/i386.md (*load_tp_<mode>): Redefine as
19956 define_insn_and_split. Split to a memory load from 0 in
19957 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
19958 using PTR mode iterator.
19959 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
19960 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
19961 (*add_tp_<mode>): Redefine as define_insn_and_split.
19962 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
19963 address space. Merge with *add_tp_x32 using PTR mode iterator.
19964 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
19965 Split to an add with a memory load from 0 in
19966 DEFAULT_TLS_SEG_REG address space.
19968 2017-08-12 Andrew Pinski <apinski@cavium.com>
19970 * config/aarch64/aarch64-option-extensions.def (rdma):
19971 Fix feature string to what Linux prints out in /proc/cpuinfo.
19973 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
19976 * dwarf2out.c (modified_type_die): For C typedef types that have
19977 an ultimate origin, process the ultimate origin instead of the
19979 (gen_typedef_die): Assert that input DECLs have no ultimate
19981 (gen_type_die_with_usage): For typedef variants that have an
19982 ultimate origin, just call gen_decl_die on the original DECL.
19983 (process_scope_var): Avoid creating DIEs for local typedefs and
19984 concrete static variables.
19986 2017-08-12 Alan Modra <amodra@gmail.com>
19990 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
19991 match gnu-user.h startfile.
19992 (ENDFILE_LINUX_SPEC): Similarly.
19994 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
19997 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
19999 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20001 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20002 * config/aarch64/aarch64.md (mov<mode>): Change.
20003 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20004 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20005 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20007 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20009 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20010 for storage order barriers.
20012 2017-08-11 Martin Liska <mliska@suse.cz>
20015 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20016 variables of void type.
20018 2017-08-11 Martin Liska <mliska@suse.cz>
20020 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20021 TARGET_SUPPORTS_ALIASES.
20022 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20023 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20024 (optimize_weakref): Likewise.
20025 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20026 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20027 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20029 2017-08-11 Martin Liska <mliska@suse.cz>
20032 * config/i386/i386.c (make_resolver_func): Do complete
20033 refactoring of the function.
20035 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20038 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20039 * config/i386/i386.c (ix86_stack_protect_guard): Use
20040 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20041 * doc/invoke.texi (x86 Options): Document
20042 -mstack-protector-guard-symbol= option.
20044 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20046 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20047 * config/i386/i386.c (ix86_split_stack_guard): New function.
20048 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20049 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20050 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20051 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20052 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20053 (split_stack_space_check): Call ix86_split_stack_guard.
20055 2017-08-10 Martin Sebor <msebor@redhat.com>
20057 * print-tree.c (print_node): Print location using the established
20059 Replace spaces with colons.
20060 (debug_raw, debug): Ditto.
20062 2017-08-10 Martin Sebor <msebor@redhat.com>
20065 * pretty-print.c (pp_format): Correct the handling of %s precision.
20067 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20070 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20072 (ix86_finalize_stack_frame_flags): This. Also clear
20073 frame_pointer_needed if -fno-omit-frame-pointer is used without
20075 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20076 with ix86_finalize_stack_frame_flags.
20077 (ix86_expand_epilogue): Likewise.
20078 (ix86_expand_split_stack_prologue): Likewise.
20079 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20081 2017-08-10 Martin Liska <mliska@suse.cz>
20084 * c-attribs.c (handle_target_attribute):
20085 Report warning for an empty string argument of target attribute.
20087 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20090 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20092 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20093 or DECL_NONLOCAL labels.
20094 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20095 or DECL_NONLOCAL labels here.
20097 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20099 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20100 to indicate when early gimple folding has been disabled.
20101 (rs6000_gimple_fold_builtin): Add debug content.
20102 (rs6000_invalid_builtin): Fix whitespace.
20103 (rs6000_expand_builtin): Fix whitespace.
20104 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20106 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20109 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20110 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20112 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20114 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20115 (qdf24xx): Likewise.
20116 * config/aarch64/aarch64.md: Include falkor.md.
20117 * config/aarch64/falkor.md: New.
20119 2017-08-09 Marek Polacek <polacek@redhat.com>
20122 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20123 * diagnostic.c (emit_diagnostic): Add a comment.
20124 (emit_diagnostic_valist): New function.
20126 2017-08-09 Marek Polacek <polacek@redhat.com>
20129 * input.c (make_location): New overload.
20130 * input.h (make_location): Declare.
20132 2017-08-08 Alan Modra <amodra@gmail.com>
20133 H.J. Lu <hongjiu.lu@intel.com>
20136 * gcc.c (NO_PIE_SPEC): Delete.
20137 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20139 (LINK_PIE_SPEC): ..to here.
20140 (LINK_COMMAND_SPEC): Support -no-pie.
20141 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20142 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20144 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20145 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20146 (ENDFILE_CRTEND_SPEC): Similarly.
20148 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20151 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20152 (mstack-protector-guard-offset=): Ditto.
20153 * config/i386/i386.c (ix86_option_override): Handle
20154 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20156 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20157 ix86_stack_protect_guard_offset variables.
20158 (TARGET_STACK_PROTECT_GUARD): Always define.
20159 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20160 and -mstack-protector-guard-offset= options.
20162 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20164 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20165 boundary case for the last candidate.
20167 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20169 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20170 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20172 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20174 PR middle-end/19706
20175 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20176 * config/aarch64/aarch64-builtins.c
20177 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20178 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20179 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20181 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20182 Andrew Pinski <pinskia@gmail.com>
20184 PR middle-end/19706
20185 * internal-fn.def (XORSIGN): New.
20186 * optabs.def (xorsign_optab): New.
20187 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20188 (convert_expand_mult_copysign): New.
20189 (pass_optimize_widening_mul::execute): Call
20190 convert_expand_mult_copysign.
20192 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20194 PR tree-optimization/81354
20195 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20196 Insert on edges rather than explicitly creating landing pads.
20197 (analyze_candidates_and_replace): Commit edge inserts.
20199 2017-08-08 Richard Biener <rguenther@suse.de>
20201 PR middle-end/81719
20202 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20203 (expand_simple_operations): Also look through ADDR_EXPRs with
20204 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20206 2017-08-08 Richard Biener <rguenther@suse.de>
20208 PR tree-optimization/81723
20209 * tree-vect-slp.c (struct bst_traits): New hash traits.
20210 (bst_fail): New global.
20211 (vect_build_slp_tree_2): New worker, split out from ...
20212 (vect_build_slp_tree): ... this now wrapping it with using
20213 bst_fail set to cache SLP tree build fails. Properly handle
20215 (vect_analyze_slp_instance): Allocate and free bst_fail.
20217 2017-08-08 Martin Liska <mliska@suse.cz>
20220 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20221 LABEL_DECLs that can be from a different function.
20223 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20225 PR tree-optimization/81744
20226 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20227 loop's number of iterations.
20229 2017-08-08 Martin Liska <mliska@suse.cz>
20231 * asan.c: Include header files.
20232 * attribs.c (build_decl_attribute_variant): New function moved
20234 (build_type_attribute_qual_variant): Likewise.
20235 (cmp_attrib_identifiers): Likewise.
20236 (simple_cst_list_equal): Likewise.
20237 (omp_declare_simd_clauses_equal): Likewise.
20238 (attribute_value_equal): Likewise.
20239 (comp_type_attributes): Likewise.
20240 (build_type_attribute_variant): Likewise.
20241 (lookup_ident_attribute): Likewise.
20242 (remove_attribute): Likewise.
20243 (merge_attributes): Likewise.
20244 (merge_type_attributes): Likewise.
20245 (merge_decl_attributes): Likewise.
20246 (merge_dllimport_decl_attributes): Likewise.
20247 (handle_dll_attribute): Likewise.
20248 (attribute_list_equal): Likewise.
20249 (attribute_list_contained): Likewise.
20250 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20251 (lookup_attribute_by_prefix): Likewise.
20252 * bb-reorder.c: Include header files.
20253 * builtins.c: Likewise.
20254 * calls.c: Likewise.
20255 * cfgexpand.c: Likewise.
20256 * cgraph.c: Likewise.
20257 * cgraphunit.c: Likewise.
20258 * convert.c: Likewise.
20259 * dwarf2out.c: Likewise.
20260 * final.c: Likewise.
20261 * fold-const.c: Likewise.
20262 * function.c: Likewise.
20263 * gimple-expr.c: Likewise.
20264 * gimple-fold.c: Likewise.
20265 * gimple-pretty-print.c: Likewise.
20266 * gimple.c: Likewise.
20267 * gimplify.c: Likewise.
20268 * hsa-common.c: Likewise.
20269 * hsa-gen.c: Likewise.
20270 * internal-fn.c: Likewise.
20271 * ipa-chkp.c: Likewise.
20272 * ipa-cp.c: Likewise.
20273 * ipa-devirt.c: Likewise.
20274 * ipa-fnsummary.c: Likewise.
20275 * ipa-inline.c: Likewise.
20276 * ipa-visibility.c: Likewise.
20278 * lto-cgraph.c: Likewise.
20279 * omp-expand.c: Likewise.
20280 * omp-general.c: Likewise.
20281 * omp-low.c: Likewise.
20282 * omp-offload.c: Likewise.
20283 * omp-simd-clone.c: Likewise.
20284 * opts-global.c: Likewise.
20285 * passes.c: Likewise.
20286 * predict.c: Likewise.
20287 * sancov.c: Likewise.
20288 * sanopt.c: Likewise.
20289 * symtab.c: Likewise.
20290 * toplev.c: Likewise.
20291 * trans-mem.c: Likewise.
20292 * tree-chkp.c: Likewise.
20293 * tree-eh.c: Likewise.
20294 * tree-into-ssa.c: Likewise.
20295 * tree-object-size.c: Likewise.
20296 * tree-parloops.c: Likewise.
20297 * tree-profile.c: Likewise.
20298 * tree-ssa-ccp.c: Likewise.
20299 * tree-ssa-live.c: Likewise.
20300 * tree-ssa-loop.c: Likewise.
20301 * tree-ssa-sccvn.c: Likewise.
20302 * tree-ssa-structalias.c: Likewise.
20303 * tree-ssa.c: Likewise.
20304 * tree-streamer-in.c: Likewise.
20305 * tree-vectorizer.c: Likewise.
20306 * tree-vrp.c: Likewise.
20307 * tsan.c: Likewise.
20308 * ubsan.c: Likewise.
20309 * varasm.c: Likewise.
20310 * varpool.c: Likewise.
20311 * tree.c: Remove functions moved to attribs.[ch].
20312 * tree.h: Likewise.
20313 * config/aarch64/aarch64.c: Add attrs.h header file.
20314 * config/alpha/alpha.c: Likewise.
20315 * config/arc/arc.c: Likewise.
20316 * config/arm/arm.c: Likewise.
20317 * config/avr/avr.c: Likewise.
20318 * config/bfin/bfin.c: Likewise.
20319 * config/c6x/c6x.c: Likewise.
20320 * config/cr16/cr16.c: Likewise.
20321 * config/cris/cris.c: Likewise.
20322 * config/darwin.c: Likewise.
20323 * config/epiphany/epiphany.c: Likewise.
20324 * config/fr30/fr30.c: Likewise.
20325 * config/frv/frv.c: Likewise.
20326 * config/ft32/ft32.c: Likewise.
20327 * config/h8300/h8300.c: Likewise.
20328 * config/i386/winnt.c: Likewise.
20329 * config/ia64/ia64.c: Likewise.
20330 * config/iq2000/iq2000.c: Likewise.
20331 * config/lm32/lm32.c: Likewise.
20332 * config/m32c/m32c.c: Likewise.
20333 * config/m32r/m32r.c: Likewise.
20334 * config/m68k/m68k.c: Likewise.
20335 * config/mcore/mcore.c: Likewise.
20336 * config/microblaze/microblaze.c: Likewise.
20337 * config/mips/mips.c: Likewise.
20338 * config/mmix/mmix.c: Likewise.
20339 * config/mn10300/mn10300.c: Likewise.
20340 * config/moxie/moxie.c: Likewise.
20341 * config/msp430/msp430.c: Likewise.
20342 * config/nds32/nds32-isr.c: Likewise.
20343 * config/nds32/nds32.c: Likewise.
20344 * config/nios2/nios2.c: Likewise.
20345 * config/nvptx/nvptx.c: Likewise.
20346 * config/pa/pa.c: Likewise.
20347 * config/pdp11/pdp11.c: Likewise.
20348 * config/powerpcspe/powerpcspe.c: Likewise.
20349 * config/riscv/riscv.c: Likewise.
20350 * config/rl78/rl78.c: Likewise.
20351 * config/rx/rx.c: Likewise.
20352 * config/s390/s390.c: Likewise.
20353 * config/sh/sh.c: Likewise.
20354 * config/sol2.c: Likewise.
20355 * config/sparc/sparc.c: Likewise.
20356 * config/spu/spu.c: Likewise.
20357 * config/stormy16/stormy16.c: Likewise.
20358 * config/tilegx/tilegx.c: Likewise.
20359 * config/tilepro/tilepro.c: Likewise.
20360 * config/v850/v850.c: Likewise.
20361 * config/vax/vax.c: Likewise.
20362 * config/visium/visium.c: Likewise.
20363 * config/xtensa/xtensa.c: Likewise.
20365 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20368 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20369 constraints since the -mupper-regs-* switches have been
20371 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20372 into a vector from a double word element that was extracted from
20373 another vector, and eliminate extra XXPERMDI instructions.
20374 (vsx_concat_<mode>_2): Likewise.
20375 (vsx_concat_<mode>_3): Likewise.
20376 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20377 concat to allow optimizing inserts from previous extracts.
20379 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20381 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20382 memory reference to a SSP offset in TLS address space.
20383 (ix86_print_operand) <case '@'>: Remove.
20384 (ix86_print_operand_punct_valid_p): Remove '@' code.
20385 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20386 UNSPEC_SP_TLS_TEST.
20387 (stack_tls_protect_set_<mode>): Remove.
20388 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20389 (stack_tls_protect_test_<mode>): Remove.
20390 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20392 2017-08-07 Olivier Hainque <hainque@adacore.com>
20395 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20397 2017-08-07 Douglas Rupp <rupp@adacore.com>
20399 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20400 variable was referenced as multidir in command.
20402 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20405 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20408 2017-08-07 Martin Liska <mliska@suse.cz>
20410 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20411 * config/rl78/rl78.c: Add include of attribs.h.
20412 * config/sh/sh.c: Likewise.
20413 * config/v850/v850.c: Likewise.
20415 2017-08-07 Tom de Vries <tom@codesourcery.com>
20417 PR middle-end/78266
20418 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20420 2017-08-07 Martin Liska <mliska@suse.cz>
20422 * config/mips/mips.c: Include attribs.h.
20424 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20427 * doc/invoke.texi: Document change in behvaior for -Ofast for
20430 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20432 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20434 (aarch64_pop_regs): Likewise.
20435 (aarch64_gen_load_pair): Likewise.
20436 (aarch64_save_callee_saves): Likewise.
20437 (aarch64_restore_callee_saves): Likewise.
20439 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20441 * config/i386/i386.c: Revert the last change.
20443 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20446 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20448 (ix86_finalize_stack_frame_flags): This. Also clear
20449 frame_pointer_needed if -fno-omit-frame-pointer is used without
20451 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20452 with ix86_finalize_stack_frame_flags.
20453 (ix86_expand_epilogue): Likewise.
20454 (ix86_expand_split_stack_prologue): Likewise.
20456 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20459 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20460 to P_AES for Westmere.
20462 2017-08-07 Jonathan Yong <10walls@gmail.com>
20464 * config/i386/mingw.opt (fset-stack-executable): Removed.
20465 * config/i386/cygming.opt (fset-stack-executable): Moved
20467 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20469 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20471 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20473 2017-08-07 Marek Polacek <polacek@redhat.com>
20475 PR middle-end/81737
20476 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20478 2017-08-07 Martin Liska <mliska@suse.cz>
20480 * attribs.h (canonicalize_attr_name): New function.
20481 (cmp_attribs): Move from c-format.c and adjusted.
20482 (is_attribute_p): Moved from tree.h.
20483 * tree-inline.c: Add new includes.
20484 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20485 (private_is_attribute_p): Remove.
20486 (private_lookup_attribute): Likewise.
20487 (private_lookup_attribute_by_prefix): Simplify.
20488 (remove_attribute): Use is_attribute_p.
20489 * tree.h: Remove removed declarations.
20491 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20493 PR middle-end/81698
20494 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20495 instead of computing it in the function. Formatting fix.
20496 (expand_case): Don't rely on default_edge being the first edge,
20497 clear it if removing it, pass default_edge to
20498 emit_case_dispatch_table.
20499 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20502 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20504 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20505 insn in the function, emit NOP after the insn.
20507 2017-08-06 Tom de Vries <tom@codesourcery.com>
20509 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20512 2017-08-06 Tom de Vries <tom@codesourcery.com>
20514 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20517 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20519 PR tree-optimization/57371
20520 * match.pd: New pattern.
20522 2017-08-04 Marek Polacek <polacek@redhat.com>
20524 PR middle-end/81695
20525 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20526 perform the computation in offset_int.
20528 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20530 PR tree-optimization/81136
20531 * tree-vectorizer.h: Include tree-hash-traits.h.
20532 (vec_base_alignments): New typedef.
20533 (vec_info): Add a base_alignments field.
20534 (vect_record_base_alignments): Declare.
20535 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20537 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20538 (create_data_ref): Add an is_conditional_in_stmt argument.
20539 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20540 the is_conditional_in_stmt field.
20541 (data_ref_loc): Add an is_conditional_in_stmt field.
20542 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20543 (find_data_references_in_stmt): Update call to create_data_ref.
20544 (graphite_find_data_references_in_stmt): Likewise.
20545 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20546 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20547 (vect_record_base_alignment): New function.
20548 (vect_record_base_alignments): Likewise.
20549 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20550 for nested statements even if we fail to compute a misalignment.
20551 Use pooled base alignments for unconditional references.
20552 (vect_find_same_alignment_drs): Compare base addresses instead
20554 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20555 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20557 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20559 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20560 Add an explicit name for the enum. Use auto_vec for slp_instances
20561 and grouped_stores.
20562 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20564 (_bb_vec_info): Add a constructor and destructor.
20565 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20566 (destroy_loop_vec_info): Delete.
20567 (vect_destroy_datarefs): Likewise.
20568 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20569 (vec_info::vec_info): New function.
20570 (vec_info::~vec_info): Likewise.
20571 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20572 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20573 destroy_loop_vec_info.
20574 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20575 (_loop_vec_info::_loop_vec_info): ...this.
20576 (destroy_loop_vec_info): Replace with...
20577 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20578 the stmt_vec_infos. Leave handling of vec_info information to its
20579 destructor. Remove explicit vector releases.
20580 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20581 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20582 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20583 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20584 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20585 (destroy_bb_vec_info): Replace with...
20586 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20587 information to its destructor.
20588 (vect_slp_analyze_bb_1): Use new and delete instead of
20589 new_bb_vec_info and destroy_bb_vec_info.
20590 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20593 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20595 * tree-data-ref.h (subscript): Add access_fn field.
20596 (data_dependence_relation): Add could_be_independent_p.
20597 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20598 (same_access_functions): Move to tree-data-ref.c.
20599 * tree-data-ref.c (ref_contains_union_access_p): New function.
20600 (access_fn_component_p): Likewise.
20601 (access_fn_components_comparable_p): Likewise.
20602 (dr_analyze_indices): Add a reference to access_fn_component_p.
20603 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20605 (constant_access_functions): Likewise.
20606 (add_other_self_distances): Likewise.
20607 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20608 (initialize_data_dependence_relation): Use XCNEW and remove
20609 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20610 of access functions that have the same type. Allow the
20611 subsequence to end with different bases in some circumstances.
20612 Record the chosen access functions in SUB_ACCESS_FN.
20613 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20614 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20615 (subscript_dependence_tester_1): Likewise dra and drb.
20616 (build_classic_dist_vector): Update calls accordingly.
20617 (subscript_dependence_tester): Likewise.
20618 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20619 DDR_COULD_BE_INDEPENDENT_P.
20620 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20621 comp_alias_ddrs instead of may_alias_ddrs.
20622 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20624 (vect_analyze_data_ref_dependence): Use it if
20625 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20626 distance vectors if that fails.
20627 (dependence_distance_ge_vf): New function.
20628 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20629 LOOP_VINFO_MAY_ALIAS_DDRS.
20631 2017-08-04 Richard Biener <rguenther@suse.de>
20633 PR middle-end/81705
20634 * fold-const.c (fold_binary_loc): Properly restrict
20635 minus_var0 && minus_var1 case when associating undefined overflow
20638 2017-08-04 Tom de Vries <tom@codesourcery.com>
20640 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20642 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20644 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20645 Don't start diagnostic messages with a capital letter.
20646 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20648 (rs6000_invalid_builtin): Likewise.
20649 (rs6000_trampoline_init): Likewise.
20651 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20654 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20655 after setting changeable df flags.
20657 2017-08-03 Richard Biener <rguenther@suse.de>
20659 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20660 up if the use is in USE - X.
20662 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20664 * toplev.c (dumpfile.h): New include.
20665 (internal_error_reentered): New static function. Use it...
20666 (internal_error_function): ...here to handle reentered internal_error.
20668 2017-08-03 Richard Biener <rguenther@suse.de>
20670 PR middle-end/81148
20671 * fold-const.c (split_tree): Add minus_var and minus_con
20672 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20673 here but always use minus_*.
20674 (associate_trees): Assert we never associate with MINUS_EXPR
20675 and NULL first operand. Do not recurse for PLUS_EXPR operands
20676 when associating as MINUS_EXPR either.
20677 (fold_binary_loc): Track minus_var and minus_con.
20679 2017-08-03 Tom de Vries <tom@codesourcery.com>
20682 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20683 ACCEL_COMPILER, apply finish_options on
20684 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20686 2017-08-03 Tom de Vries <tom@codesourcery.com>
20689 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20690 function_entry_patch_area_size > 0.
20692 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20695 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20696 instead of 10??LU, perform unit multiplication in wide_int,
20697 don't change alloc_object_size_limit if the limit is larger
20700 PR tree-optimization/81655
20701 PR tree-optimization/81588
20702 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20703 the case when ranges[i].low and high are 1 for unsigned type with
20706 PR middle-end/81052
20707 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20708 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20710 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20712 * tree-vrp.h: Add include guard.
20714 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20717 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20718 (ud2): New insn pattern.
20719 * config/i386/i386.c (ix86_expand_epilogue):
20720 For naked functions, generate ud2 instead of trap insn.
20722 2017-08-02 Marek Polacek <polacek@redhat.com>
20725 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20727 2017-08-02 Tom de Vries <tom@codesourcery.com>
20728 Cesar Philippidis <cesar@codesourcery.com>
20730 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20731 Add missing edge probabilities.
20733 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20735 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20736 Correct endianness.
20738 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20740 PR middle-end/79499
20741 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20742 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20743 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20744 prologue_seq sequences - if any.
20746 2017-08-02 Richard Biener <rguenther@suse.de>
20748 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20749 via vectors if supported, integer extracts via punning if supported
20750 or otherwise vector extracts.
20752 2017-08-02 Richard Biener <rguenther@suse.de>
20754 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20756 (bitmap_insert_into_set): ... this.
20758 2017-08-02 Richard Biener <rguenther@suse.de>
20760 PR tree-optimization/81633
20762 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20764 PR tree-optimization/71752
20765 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20767 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20769 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20770 (machine_function::call_ms2sysv_pad_out): Remove field.
20771 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20772 (ix86_compute_frame_layout): Likewise.
20774 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20777 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20778 attribute with interrupt attribute.
20780 2017-08-01 Andrew Pinski <apinski@cavium.com>
20782 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20783 BIT_INSERT_EXPR's operand 1
20784 to see if the types precision matches.
20786 2017-08-01 Martin Liska <mliska@suse.cz>
20788 PR middle-end/70140
20789 * builtins.c (expand_builtin_memcpy_args): Remove.
20790 (expand_builtin_memcpy): Call newly added function
20791 expand_builtin_memory_copy_args.
20792 (expand_builtin_memcpy_with_bounds): Likewise.
20793 (expand_builtin_mempcpy): Remove last argument.
20794 (expand_builtin_mempcpy_with_bounds): Likewise.
20795 (expand_builtin_memory_copy_args): New function created from
20796 expand_builtin_mempcpy_args with small modifications.
20797 (expand_builtin_mempcpy_args): Remove.
20798 (expand_builtin_stpcpy): Remove unused argument.
20799 (expand_builtin): Likewise.
20800 (expand_builtin_with_bounds): Likewise.
20802 2017-08-01 Martin Liska <mliska@suse.cz>
20805 Make mempcpy more optimal (PR middle-end/70140).
20807 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20810 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20811 __builtin_vec_cmpne verify both arguments are compatible vectors
20812 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20813 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20814 move computation of aligned to after checking the argument types.
20818 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20819 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20822 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20823 Doug Rupp <rupp@adacore.com>
20824 Olivier Hainque <hainque@adacore.com>
20826 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20827 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20829 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20830 for TARGET_OS_CPP_BUILTIN.
20831 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20832 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20834 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20835 passing required abi options to the assembler for EABI configurations.
20836 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20837 of .text.hot and .text.unlikely sections for kernel modules when
20838 using ARM style exceptions.
20839 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20840 options. Add EXTRA_CC1_SPEC.
20841 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20843 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20845 (ARM_TARGET2_DWARF_FORMAT): Define.
20846 * config/arm/t-vxworks: Adjust multilib control to removal of the
20847 Diab command line options.
20849 2017-08-01 Martin Liska <mliska@suse.cz>
20851 PR gcov-profile/81561
20852 * gcov.c (unblock): Make unblocking safe as we need to preserve
20853 index correspondence of blocks and block_lists.
20855 2017-08-01 Richard Biener <rguenther@suse.de>
20857 PR tree-optimization/81181
20858 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
20859 (compute_antic): ... end of iteration here.
20861 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
20863 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
20864 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
20865 (ftree-slp-vectorize): Likewise.
20866 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
20867 can no longer be set independent of flag_tree_loop_vectorize.
20868 * omp-general.c (emp_max_vf): Likewise.
20869 * opts.c (enable_fdo_optimizations): Remove references to
20870 flag_tree_vectorize, these are now implicit.
20871 (common_handle_option): Remove handling for OPT_ftree_vectorize,
20872 and leave it for the options machinery.
20874 2017-08-01 Martin Liska <mliska@suse.cz>
20876 PR middle-end/70140
20877 * builtins.c (expand_builtin_memcpy_args): Remove.
20878 (expand_builtin_memcpy): Call newly added function
20879 expand_builtin_memory_copy_args.
20880 (expand_builtin_memcpy_with_bounds): Likewise.
20881 (expand_builtin_mempcpy): Remove last argument.
20882 (expand_builtin_mempcpy_with_bounds): Likewise.
20883 (expand_builtin_memory_copy_args): New function created from
20884 expand_builtin_mempcpy_args with small modifications.
20885 (expand_builtin_mempcpy_args): Remove.
20886 (expand_builtin_stpcpy): Remove unused argument.
20887 (expand_builtin): Likewise.
20888 (expand_builtin_with_bounds): Likewise.
20890 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20893 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
20894 print "ds:" only for immediates in generic address space.
20896 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20899 * config/i386/i386.c (ix86_funciton_naked): New prototype.
20900 (ix86_function_ok_for_sibcall): Return false for naked functions.
20902 2017-08-01 Richard Biener <rguenther@suse.de>
20904 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
20905 (compute_antic): Seed worklist with exit block predecessors.
20906 * cfganal.c (dfs_find_deadend): For a cycle return the source
20907 of the edge closing it.
20909 2017-08-01 Tamar Christina <tamar.christina@arm.com>
20911 * config/aarch64/aarch64.c
20912 (aarch64_can_const_movi_rtx_p): Move 0 check.
20914 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20916 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
20917 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
20919 * match.pd: Ditto in address comparison pattern.
20921 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20923 PR tree-optimization/81627
20924 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
20925 closed ssa form for store-store chain.
20927 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20929 PR tree-optimization/81620
20930 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
20931 for store-store chain.
20933 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20935 PR tree-optimization/81588
20936 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
20937 ranges[i].in_p, invert comparison code ccode. For >/>=,
20938 swap rhs1 and rhs2 and comparison code unconditionally,
20939 for </<= don't do that. Don't swap rhs1/rhs2 again if
20940 ranges[i].in_p, instead invert comparison code ccode if
20941 opcode or oe->rank is BIT_IOR_EXPR.
20944 * optabs.def (vec_extract_optab, vec_init_optab): Change from
20945 a direct optab to conversion optab.
20946 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
20947 with GET_MODE_INNER as last argument instead of optab_handler.
20948 * expmed.c (extract_bit_field_1): Likewise. Use vector from
20949 vector extraction if possible and optab is available.
20950 * expr.c (store_constructor): Use convert_optab_handler instead
20951 of optab_handler. Use vector initialization from smaller
20952 vectors if possible and optab is available.
20953 * tree-vect-stmts.c (vectorizable_load): Likewise.
20954 * doc/md.texi (vec_extract, vec_init): Document that the optabs
20955 now have two modes.
20956 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
20957 of vec_init from half-sized vectors with the same element mode.
20958 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
20959 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20960 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
20961 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
20962 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
20963 after mode in gen_vec_extract* calls.
20964 (vec_extract<mode>): Renamed to ...
20965 (vec_extract<mode><ssescalarmodelower>): ... this.
20966 (vec_extract<mode><ssehalfvecmodelower>): New expander.
20967 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
20968 element mode after mode in gen_vec_init* calls.
20969 (VEC_INIT_HALF_MODE): New mode iterator.
20970 (vec_init<mode>): Renamed to ...
20971 (vec_init<mode><ssescalarmodelower>): ... this.
20972 (vec_init<mode><ssehalfvecmodelower>): New expander.
20973 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
20974 (vec_extractv2sfsf): ... this.
20975 (vec_initv2sf): Renamed to ...
20976 (vec_initv2sfsf): ... this.
20977 (vec_extractv2si): Renamed to ...
20978 (vec_extractv2sisi): ... this.
20979 (vec_initv2si): Renamed to ...
20980 (vec_initv2sisi): ... this.
20981 (vec_extractv4hi): Renamed to ...
20982 (vec_extractv4hihi): ... this.
20983 (vec_initv4hi): Renamed to ...
20984 (vec_initv4hihi): ... this.
20985 (vec_extractv8qi): Renamed to ...
20986 (vec_extractv8qiqi): ... this.
20987 (vec_initv8qi): Renamed to ...
20988 (vec_initv8qiqi): ... this.
20989 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
20990 (vec_init<mode>): Renamed to ...
20991 (vec_init<mode><VEC_base_l>): ... this.
20992 (vec_extract<mode>): Renamed to ...
20993 (vec_extract<mode><VEC_base_l>): ... this.
20994 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
20995 (vec_initv2sfsf): ... this.
20996 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
20997 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20998 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
20999 element mode after mode in gen_vec_init* calls.
21000 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21001 (vec_init<mode><Vel>): ... this.
21002 (vec_extract<mode>): Renamed to ...
21003 (vec_extract<mode><Vel>): ... this.
21004 * config/aarch64/iterators.md (Vel): New mode attribute.
21005 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21006 Add element mode after mode in gen_vec_extract* calls.
21007 * config/s390/vector.md (non_vec_l): New mode attribute.
21008 (vec_extract<mode>): Renamed to ...
21009 (vec_extract<mode><non_vec_l>): ... this.
21010 (vec_init<mode>): Renamed to ...
21011 (vec_init<mode><non_vec_l>): ... this.
21012 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21013 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21015 * config/arm/iterators.md (V_elem_l): New mode attribute.
21016 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21017 (vec_extract<mode><V_elem_l>): ... this.
21018 (vec_extractv2di): Renamed to ...
21019 (vec_extractv2didi): ... this.
21020 (vec_init<mode>): Renamed to ...
21021 (vec_init<mode><V_elem_l>): ... this.
21022 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21023 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21024 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21025 Add element mode after gen_vec_extract* calls.
21026 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21027 (vec_init<mode><unitmode>): ... this.
21028 (vec_extract<mode>): Renamed to ...
21029 (vec_extract<mode><unitmode>): ... this.
21030 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21031 (vec_init<mode><unitmode>): ... this.
21032 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21033 (vec_initv2sfsf): ... this.
21034 (vec_extractv2sf): Renamed to ...
21035 (vec_extractv2sfsf): ... this.
21036 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21037 Add element mode after gen_vec_extract* calls.
21038 * config/mips/mips.md (unitmode): New mode iterator.
21039 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21040 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21041 * config/spu/spu.md (inner_l): New mode attribute.
21042 (vec_init<mode>): Renamed to ...
21043 (vec_init<mode><inner_l>): ... this.
21044 (vec_extract<mode>): Renamed to ...
21045 (vec_extract<mode><inner_l>): ... this.
21046 * config/sparc/sparc.md (veltmode): New mode iterator.
21047 (vec_init<VMALL:mode>): Renamed to ...
21048 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21049 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21050 (vec_initv2sisi): ... this.
21051 (vec_initv2sf): Renamed to ...
21052 (vec_initv2sfsf): ... this.
21053 (vec_extractv2sf): Renamed to ...
21054 (vec_extractv2sfsf): ... this.
21055 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21056 (vec_init<mode>): Renamed to ...
21057 (vec_init<mode><VEC_base_l>): ... this.
21058 (vec_extract<mode>): Renamed to ...
21059 (vec_extract<mode><VEC_base_l>): ... this.
21060 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21061 (vec_initv2sfsf): ... this.
21062 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21063 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21064 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21065 gen_vec_init* calls.
21067 2017-08-01 Richard Biener <rguenther@suse.de>
21069 PR tree-optimization/81297
21070 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21071 TREE_OVERFLOW from INTEGER_CSTs.
21073 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21075 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21077 2017-07-31 Carl Love <cel@us.ibm.com>
21079 * config/rs6000/rs6000-c: Add support for built-in functions
21080 vector signed char vec_xl_be (signed long long, signed char *);
21081 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21082 vector signed int vec_xl_be (signed long long, signed int *);
21083 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21084 vector signed long long vec_xl_be (signed long long, signed long long *);
21085 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21086 vector signed short vec_xl_be (signed long long, signed short *);
21087 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21088 vector double vec_xl_be (signed long long, double *);
21089 vector float vec_xl_be (signed long long, float *);
21090 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21091 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21092 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21094 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21095 (altivec_expand_builtin): Add switch statement to call
21096 altivec_expand_xl_be for each builtin.
21097 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21098 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21099 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21100 __builtin_vsx_le_be_v16qi.
21101 * doc/extend.texi: Update the built-in documentation file for the
21102 new built-in functions.
21104 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21107 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21109 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21111 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21113 * config.gcc: Add z14.
21114 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21115 CPU model numbers for z13s and z14.
21116 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21118 * config/s390/s390-opts.h (enum processor_type): Rename
21119 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21120 * config/s390/s390.c (processor_table): Add field for CPU name to
21121 be passed to Binutils.
21122 (s390_asm_output_machine_for_arch): Use the new field in
21123 processor_table for Binutils.
21124 (s390_expand_builtin): Replace arch12 with z14.
21125 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21126 (s390_get_sched_attrmask): Likewise.
21127 (s390_get_unit_mask): Likewise.
21128 * config/s390/s390.opt: Add z14 to processor_type enum.
21130 2017-07-31 Martin Jambor <mjambor@suse.cz>
21133 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21134 regardless of optimization level.
21136 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21137 Martin Liska <mliska@suse.cz>
21139 * predict.def: Remove old comment and adjust probability.
21140 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21141 PREDICT statements.
21143 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21146 * config/i386/i386.c (ix86_function_naked): New function.
21147 (ix86_can_use_return_insn_p): Return false for naked functions.
21148 (ix86_expand_prologue): Skip prologue for naked functions.
21149 (ix86_expand_epilogue): Skip epilogue for naked functions
21150 and emit trap instruction.
21151 (ix86_warn_func_return): New function.
21152 (ix86_attribute_table): Add "naked" attribute specification.
21153 (TARGET_WARN_FUNC_RETURN): Define.
21154 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21156 2017-07-31 Martin Liska <mliska@suse.cz>
21158 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21159 (dump_gimple_bb_header): Always dump BB info.
21160 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21162 2017-07-31 Martin Liska <mliska@suse.cz>
21165 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21166 also with current_function_decl non-null equality.
21168 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21171 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21172 change type to the element type, instead add eltype variable and
21173 use it where we are interested in the element type.
21175 PR tree-optimization/81603
21176 * ipa-polymorphic-call.c
21177 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21178 offset arithmetic in offset_int, bail out if the resulting bit offset
21179 doesn't fit into shwi.
21181 2017-07-31 Martin Liska <mliska@suse.cz>
21183 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21184 (gimplify_save_expr): Fix comment.
21186 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21189 * config/i386/i386.c (ix86_function_arg): Update arguments for
21191 (ix86_compute_frame_layout): Set the initial stack offset to
21192 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21193 INCOMING_FRAME_SP_OFFSET.
21194 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21195 stack before exception handler returns.
21196 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21197 the 'ERROR_CODE' for exception handler.
21199 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21201 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21202 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21203 (ASM_OUTPUT_REG_POP): Ditto.
21204 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21205 instead of asm_fprintf to output pure string.
21207 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21209 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21210 to imported_module_or_decl hook.
21211 (debug_nothing_tree_tree_tree_bool): Remove.
21212 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21213 * debug.c (do_nothing_debug_hooks): Use
21214 debug_nothing_tree_tree_tree_bool_bool instead of
21215 debug_nothing_tree_tree_tree_bool.
21216 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21217 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21218 * sdbout.c (sdb_debug_hooks): Likewise.
21219 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21220 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21222 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21223 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21224 attribute, don't add anything.
21226 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21228 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21229 (fold_build2_stat_loc): Likewise.
21230 (fold_build3_stat_loc): Likewise.
21231 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21232 (fold_build1_loc): Remove macro.
21233 (fold_build2_loc): Likewise.
21234 (fold_build3_loc): Likewise.
21236 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21238 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21239 (gimple_build_debug_bind_source_stat): Likewise.
21240 * gimple.h (gimple_build_debug_bind): Remove macro.
21241 (gimple_build_debug_bind_source): Likewise.
21243 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21245 * bitmap.c (bitmap_alloc): Adjust.
21246 (bitmap_gc_alloc): Likewise.
21247 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21249 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21251 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21252 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21253 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21254 (bitmap_gc_alloc_stat): Likewise.
21255 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21257 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21259 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21260 * rtl.h (shallow_copy_rtx): Remove macro.
21262 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21264 * emit-rtl.c (gen_raw_REG): Adjust.
21265 * gengenrtl.c (gendef): Likewise.
21266 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21267 * rtl.h (rtx_alloc): Remove macro.
21269 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21271 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21272 (build_tree_list_stat): Likewise.
21273 * tree.h (build_tree_list): Remove macro.
21274 (build_tree_list_vec): Likewise.
21276 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21278 * tree.c (make_vector_stat): Remove _stat from name.
21279 (build_vector_stat): Likewise.
21280 * tree.h (make_vector_stat): Remove macro.
21281 (build_vector_stat): Likewise.
21283 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21285 * tree.h (build_var_debug_value): Remove prototype.
21287 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21289 * tree.c (tree_cons_stat): Remove _stat from name.
21290 * tree.h (tree_cons): Remove macro.
21292 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21294 * tree.c (build_vl_exp_stat): Remove _stat from name.
21295 * tree.h (build_vl_exp): Remove macro.
21297 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21299 * tree.c (build_decl_stat): Remove _stat from name.
21300 * tree.h (build_decl): Remove macro.
21302 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21304 * gimple.c (gimple_build_with_ops_stat): Adjust.
21305 (gimple_alloc_stat): Remove _stat from name.
21306 * gimple.h (gimple_alloc): Remove macro.
21308 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21310 * tree.c (make_tree_vec_stat): Remove _stat from name.
21311 (grow_tree_vec_stat): Likewise.
21312 * tree.h (make_tree_vec_stat): Adjust prototype.
21313 (grow_tree_vec_stat): Likewise.
21314 (make_tree_vec): Remove macro.
21315 (grow_tree_vec): Likewise.
21317 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21319 * fold-const.c (fold_build1_stat_loc): Adjust.
21320 (fold_build2_stat_loc): Likewise.
21321 (fold_build3_stat_loc): Likewise.
21322 * tree.c (build0_stat): Remove _stat from name.
21323 (build1_stat): Likewise.
21324 (build2_stat): Likewise.
21325 (build3_stat): Likewise.
21326 (build4_stat): Likewise.
21327 (build5_stat): Likewise.
21328 * tree.h (build1_loc): Remove macro, and rename _stat function
21330 (build2_loc): Likewise.
21331 (build3_loc): Likewise.
21332 (build4_loc): Likewise.
21333 (build5_loc): Likewise.
21335 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21337 * tree.c (make_int_cst_stat): Remove _stat from name.
21338 * tree.h (make_int_cst_stat): Adjust prototype.
21339 (make_int_cst): Remove macro.
21341 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21343 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21344 * tree.h (make_tree_binfo_stat): Adjust prototype.
21345 (make_tree_binfo): Remove.
21347 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21349 * tree.c (copy_node_stat): Rename to copy_node.
21350 (build_distinct_type_copy): Adjust.
21351 * tree.h (copy_node_stat): Adjust prototype.
21352 (copy_node): Remove macro.
21354 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21356 * tree.c (make_node_stat): rename to make_node.
21357 (build_tree_list_stat): Adjust.
21358 (build0_stat): Likewise.
21359 (build2_stat): Likewise.
21360 (build3_stat): Likewise.
21361 (build4_stat): Likewise.
21362 (build5_stat): Likewise.
21363 (build_decl_stat): Likewise.
21364 * tree.h (make_node_stat): Adjust prototype.
21365 (make_node): remove macro.
21367 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21369 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21370 (PPC_FEATURE2_SCV): Likewise.
21371 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21373 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21375 * config/aarch64/aarch64.c
21376 (aarch64_internal_mov_immediate): Add new special pattern.
21377 * config/aarch64/aarch64.md (*movdi_aarch64):
21378 Add reg/32bit const mov case.
21380 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21381 Richard Sandiford <richard.sandiford@linaro.org>
21383 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21384 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21385 Add integer and movi cases.
21386 (movi-split-hf-df-sf split, fp16): New.
21387 (enabled): Added TARGET_FP_F16INST.
21388 * config/aarch64/iterators.md (GPF_HF): New.
21389 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21391 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21393 * config/aarch64/aarch64.c
21394 (aarch64_simd_container_mode): Add prototype.
21395 (aarch64_expand_mov_immediate): Add HI support.
21396 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21397 (aarch64_can_const_movi_rtx_p): New.
21398 (aarch64_preferred_reload_class):
21399 Remove restrictions of using FP registers for certain SIMD operations.
21400 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21401 (aarch64_valid_floating_const): Add integer move validation.
21402 (aarch64_simd_imm_scalar_p): Remove.
21403 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21404 (aarch64_legitimate_constant_p): Expand list of supported cases.
21405 * config/aarch64/aarch64-protos.h
21406 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21407 (aarch64_reinterpret_float_as_int): New.
21408 (aarch64_simd_imm_scalar_p): Remove.
21409 * config/aarch64/constraints.md (Uvi): New.
21410 (Dd): Split into Ds and new Dd.
21411 * config/aarch64/aarch64.md (*movsi_aarch64):
21413 (*movdi_aarch64): Add SIMD mov case.
21415 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21417 * tree-predcom.c: (struct chain): Handle store-store chain in which
21418 stores for elimination only store loop invariant values.
21419 (execute_pred_commoning_chain): Ditto.
21420 (prepare_initializers_chain_store_elim): Ditto.
21421 (prepare_finalizers): Ditto.
21422 (is_inv_store_elimination_chain): New function.
21423 (initialize_root_vars_store_elim_1): New function.
21425 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21427 * tree-predcom.c: Revise general description of the pass.
21428 (enum chain_type): New enum type for store elimination.
21429 (struct chain): New field supporting store elimination.
21430 (struct component): Ditto.
21431 (dump_chain): Dump store-stores chain.
21432 (release_chain): Release resources.
21433 (split_data_refs_to_components): Compute and create component
21434 contains only stores for elimination.
21435 (get_chain_last_ref_at): New function.
21436 (make_invariant_chain): Initialization.
21437 (make_rooted_chain): Specify chain type in parameter and record it.
21438 (add_looparound_copies): Skip for store-stores chain.
21439 (determine_roots_comp): Compute type of chain and pass it to
21441 (initialize_root_vars_store_elim_2): New function.
21442 (finalize_eliminated_stores): New function.
21443 (remove_stmt): Handle store for elimination.
21444 (execute_pred_commoning_chain): Execute predictive commoning on
21445 store-store chains.
21446 (determine_unroll_factor): Skip unroll for store-stores chain.
21447 (prepare_initializers_chain_store_elim): New function.
21448 (prepare_initializers_chain): Hanlde store-store chain.
21449 (prepare_finalizers_chain, prepare_finalizers): New function.
21450 (tree_predictive_commoning_loop): Return integer value indicating
21451 if loop is unrolled or lcssa form is corrupted.
21452 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21454 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21456 * tree-predcom.c (initialize_root): Delete.
21457 (execute_pred_commoning_chain): Initialize root vars and replace
21458 reference of non-combined chain directly, rather than call above
21461 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21463 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21464 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21466 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21468 * tree-predcom.c (struct chain): New field init_seq.
21469 (release_chain): Release init_seq.
21470 (prepare_initializers_chain): Record intialization stmts in above
21472 (insert_init_seqs): New function.
21473 (tree_predictive_commoning_loop): Call insert_init_seqs.
21475 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21477 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21479 2017-07-28 Richard Biener <rguenther@suse.de>
21481 * match.pd: Remove superfluous :c.
21482 * genmatch.c (simplify::id): Add member.
21483 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21485 (current_id): New global.
21486 (dt_node::parent): Move from ...
21487 (dt_operand::parent): ... here. Add for_id member.
21488 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21489 (decision_tree::find_node): Relax order requirement when
21490 merging DT_TRUE nodes to ones inbetween the current simplify
21491 and the one we try to merge with. Add diagnostic whenever
21492 we need to enforce pattern order by not merging.
21493 (decision_tree::insert): Set current_id.
21494 (decision_tree::print_node): Dump parent node and for_id.
21495 (parser::last_id): Add member.
21496 (parser::push_simplify): Assign unique id.
21497 (parser::parser): Initialize last_id.
21499 2017-07-28 Martin Liska <mliska@suse.cz>
21502 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21503 gimple_build_debug_bind.
21505 2017-07-28 Richard Biener <rguenther@suse.de>
21507 PR tree-optimization/81502
21508 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21510 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21512 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21513 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21514 for BIT_FIELD_REF args.
21515 * fold-const.c (make_bit_field_ref): Likewise.
21516 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21518 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21521 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21522 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21523 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21524 Or it into SANITIZER_UNDEFINED.
21525 * ubsan.c: Include gimple-fold.h and varasm.h.
21526 (ubsan_expand_ptr_ifn): New function.
21527 (instrument_pointer_overflow): New function.
21528 (maybe_instrument_pointer_overflow): New function.
21529 (instrument_object_size): Formatting fix.
21530 (pass_ubsan::execute): Call instrument_pointer_overflow
21531 and maybe_instrument_pointer_overflow.
21532 * internal-fn.c (expand_UBSAN_PTR): New function.
21533 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21534 * sanitizer.def (__ubsan_handle_pointer_overflow,
21535 __ubsan_handle_pointer_overflow_abort): New builtins.
21536 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21537 * internal-fn.def (UBSAN_PTR): New internal function.
21538 * opts.c (sanitizer_opts): Add pointer-overflow.
21539 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21540 * fold-const.c (build_range_check): Compute pointer range check in
21541 integral type if pointer arithmetics would be needed. Formatting
21544 2017-07-28 Martin Liska <mliska@suse.cz>
21547 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21548 parameters that are of a variable-length.
21550 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21552 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21554 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21555 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21556 (CRT_CALL_STATIC_FUNCTION): Likewise.
21557 (ASM_DEFAULT_SPEC): New define.
21558 (ASM_SPEC32): Likewise.
21559 (ASM_SPEC64): Likewise.
21560 (ASM_SPEC_COMMON): Likewise.
21561 (ASM_SPEC): Likewise.
21562 (INVALID_64BIT): Likewise.
21563 (LINK_OS_DEFAULT_SPEC): Likewise.
21564 (LINK_OS_SPEC32): Likewise.
21565 (LINK_OS_SPEC64): Likewise.
21566 (POWERPC_LINUX): Likewise.
21567 (PTRDIFF_TYPE): Likewise.
21568 (RESTORE_FP_PREFIX): Likewise.
21569 (RESTORE_FP_SUFFIX): Likewise.
21570 (SAVE_FP_PREFIX): Likewise.
21571 (SAVE_FP_SUFFIX): Likewise.
21572 (SIZE_TYPE): Likewise.
21573 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21574 (TARGET_64BIT): Likewise.
21575 (TARGET_64BIT): Likewise.
21576 (TARGET_AIX): Likewise.
21577 (WCHAR_TYPE_SIZE): Likewise.
21578 (WCHAR_TYPE): Undefine.
21579 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21580 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21581 (CPP_OS_RTEMS_SPEC): Delete.
21582 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21583 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21585 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21587 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21589 PR tree-optimization/81578
21590 * tree-parloops.c (build_new_reduction): Bail out if
21591 reduction_code isn't one of the standard OpenMP reductions.
21592 Move the details printing after that decision.
21594 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21596 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21597 related to reload_in_progress.
21598 (splat_input_operand): Likewise.
21599 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21601 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21603 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21604 (TARGET_INSTANTIATE_DECLS): Likewise.
21605 (legitimate_indexed_address_p): Delete reload_in_progress code.
21606 (rs6000_debug_legitimate_address_p): Likewise.
21607 (rs6000_eliminate_indexed_memrefs): Likewise.
21608 (rs6000_emit_le_vsx_store): Likewise.
21609 (rs6000_emit_move_si_sf_subreg): Likewise.
21610 (rs6000_emit_move): Likewise.
21611 (register_to_reg_type): Likewise.
21612 (rs6000_pre_atomic_barrier): Likewise.
21613 (rs6000_machopic_legitimize_pic_address): Likewise.
21614 (rs6000_allocate_stack_temp): Likewise.
21615 (rs6000_address_for_fpconvert): Likewise.
21616 (rs6000_address_for_altivec): Likewise.
21617 (rs6000_secondary_memory_needed_rtx): Delete function.
21618 (rs6000_check_sdmode): Likewise.
21619 (rs6000_alloc_sdmode_stack_slot): Likewise.
21620 (rs6000_instantiate_decls): Likewise.
21621 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21622 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21623 Delete reload_in_progress.
21624 (*vec_reload_and_plus_<mptrsize>): Likewise.
21625 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21626 (vsx_div_v2di): Likewise.
21627 (vsx_udiv_v2di): Likewise.
21629 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21631 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21632 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21633 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21634 (rs6000_debug_reg_global): Delete print of LRA status.
21635 (rs6000_option_override_internal): Delete dead LRA related code.
21636 (rs6000_lra_p): Delete function.
21637 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21639 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21641 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21642 * config/riscv/rtems.h: New file.
21644 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21645 Sudakshina Das <sudi.das@arm.com>
21647 * config/aarch64/aarch64.md
21648 (define_split for and<mode>3nr_compare): Move
21649 non aarch64_logical_operand to a register.
21650 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21651 register immediate operand to a register.
21652 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21654 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21656 PR middle-end/81564
21657 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21659 2017-07-27 Richard Biener <rguenther@suse.de>
21661 PR tree-optimization/81573
21662 PR tree-optimization/81494
21663 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21664 multi defuse cycle case.
21666 2017-07-27 Richard Biener <rguenther@suse.de>
21668 PR tree-optimization/81571
21669 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21672 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21674 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21675 earlier and only if MASK_FPU is set. Adjust formatting.
21677 2017-07-27 Martin Liska <mliska@suse.cz>
21679 * opt-functions.awk: Add validation of value of Init.
21680 * optc-gen.awk: Pass new argument.
21682 2017-07-27 Martin Liska <mliska@suse.cz>
21684 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21685 Fix wrong condition.
21687 2017-07-27 Martin Liska <mliska@suse.cz>
21689 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21690 BBs and edges seen by autoFDO.
21692 2017-07-27 Richard Biener <rguenther@suse.de>
21694 PR tree-optimization/81502
21695 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21696 with incompatible but same sized type.
21697 (execute_update_addresses_taken): Likewise.
21699 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21701 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21702 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21704 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21707 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21708 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21709 Change s_operand to memory_operand.
21711 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21713 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21714 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21715 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21716 Emit instructions rather than returning an expression. Handle TFmode
21717 and KFmode by casting to TImode.
21718 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21719 (rs6000_emit_le_vsx_store): Likewise.
21720 * config/rs6000/vsx.md (VSX_TI): New iterator.
21721 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21722 (*vsx_le_undo_permute_<mode>): Likewise.
21723 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21724 emit the split sequence.
21725 (*vsx_le_perm_store_<mode>): Likewise.
21727 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21729 PR tree-optimization/81555
21730 PR tree-optimization/81556
21731 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21732 if true, force CHANGED for the recursive invocation.
21733 (reassociate_bb): Remember original length of ops array, pass
21734 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21736 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21737 attributes for noipa attribute. For naked attribute use
21738 lookup_attribute first before lookup_attribute_spec.
21739 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21741 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21742 for functions with noipa attribute.
21743 (cgraph_externally_visible_p): Return true for functions with noipa
21745 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21746 for functions with noipa attribute.
21747 * doc/extend.texi: Document noipa function attribute.
21748 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21749 also for functions with noipa attribute.
21750 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21752 2017-07-26 Andrew Pinski <apinski@cavium.com>
21754 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21755 vec_unalign_load_cost and vec_unalign_store_cost.
21757 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21759 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21760 -mvsx-small-integer option.
21761 (ISA_3_0_MASKS_IEEE): Likewise.
21762 (OTHER_VSX_VECTOR_MASKS): Likewise.
21763 (POWERPC_MASKS): Likewise.
21764 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21765 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21766 code, only testing for DImode being allowed in non-VSX floating
21768 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21769 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21771 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21772 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21773 TARGET_P8_VECTOR test.
21774 (rs6000_secondary_reload_simple_move): Likewise.
21775 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21776 since TARGET_P9_VECTOR was already tested.
21777 (rs6000_opt_masks): Remove -mvsx-small-integer.
21778 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21779 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21781 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21782 test for TARGET_VEXTRACTUB was used, and that uses
21784 (p9 extract splitter): Likewise.
21785 (vsx_extract_<mode>_di_p9): Likewise.
21786 (vsx_extract_<mode>_store_p9): Likewise.
21787 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21788 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21789 the elimination of TARGET_VSX_SMALL_INTEGER.
21790 (vsx_extract_<mode>_p8): Likewise.
21791 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21792 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21793 (vsx_set_<mode>_p9): Likewise.
21794 (vsx_set_v4sf_p9): Likewise.
21795 (vsx_set_v4sf_p9_zero): Likewise.
21796 (vsx_insert_extract_v4sf_p9): Likewise.
21797 (vsx_insert_extract_v4sf_p9_2): Likewise.
21798 * config/rs6000/rs6000.md (sign extend splitter): Change
21799 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21800 (floatsi<mode>2_lfiwax_mem): Likewise.
21801 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21802 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21803 since a test for TARGET_P9_VECTOR was used.
21804 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21805 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21806 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21807 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21808 TARGET_P8_VECTOR test.
21809 (fix_trunc<mode>si2_stfiwx): Likewise.
21810 (fix_trunc<mode>si2_internal): Likewise.
21811 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21812 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21814 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21815 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21816 TARGET_P8_VECTOR test.
21817 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21818 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21819 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21821 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21822 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21823 since a test for TARGET_P9_VECTOR was used.
21824 (splitter for loading small constants): Likewise.
21826 2017-07-26 Andrew Pinski <apinski@cavium.com>
21828 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21831 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21834 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21835 (fp_valid_at): Likewise.
21837 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21839 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21840 (qdf24xx_addrcost_table): Likewise.
21841 (cortexa57_tunings): Update to use generic_branch_cost.
21842 (cortexa72_tunings): Likewise.
21843 (cortexa73_tunings): Likewise.
21844 (qdf24xx_tunings): Likewise.
21846 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21848 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21849 (thunderx2t99_branch_cost): Likewise.
21850 (cortexa35_tunings): Update to use generic_branch_cost.
21851 (cortexa53_tunings): Likewise.
21852 (cortexa57_tunings): Likewise.
21853 (cortexa72_tunings): Likewise.
21854 (cortexa73_tunings): Likewise.
21855 (thunderx2t99_tunings): Likewise.
21857 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21859 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
21860 (sparc_option_override): Honour MASK_FSMULD.
21861 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
21862 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
21863 * config/sparc/sparc.opt (mfsmuld): New option.
21864 * doc/invoke.texi (mfsmuld): Document option.
21866 2017-07-26 Marek Polacek <polacek@redhat.com>
21868 PR middle-end/70992
21869 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
21871 2017-07-26 Richard Biener <rguenther@suse.de>
21873 * gimple-match-head.c (do_valueize): Return OP if valueize
21875 (get_def): New helper to get at the def stmt of a SSA name
21876 if valueize allows.
21877 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
21878 do_valueize to get at the def stmt.
21879 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
21881 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
21883 PR middle-end/46932
21884 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
21886 2017-07-26 Martin Liska <mliska@suse.cz>
21889 * function.c (expand_function_start): Make expansion of
21890 nonlocal_goto_save_area after parm_birth_insn.
21892 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21894 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
21895 from all CPU target flags enable members.
21897 2017-07-26 Richard Biener <rguenther@suse.de>
21899 * genmatch.c (dt_simplify::gen): Make iterator vars const.
21900 (decision_tree::gen): Make 'type' const.
21901 (write_predicate): Likewise.
21903 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21905 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
21906 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
21907 (rs6000_option_override_internal): Likewise.
21908 (rs6000_expand_vector_set): Likewise.
21909 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
21910 (TARGET_UPPER_REGS_SF): Likewise.
21911 (TARGET_UPPER_REGS_DI): Likewise.
21912 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
21913 (TARGET_DIRECT_MOVE_64BIT): Likewise.
21914 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
21915 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21916 (Splitters for DI constants in Altivec registers): Likewise.
21917 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
21918 (vsx_set_v4sf_p9): Likewise.
21919 (vsx_set_v4sf_p9_zero): Likewise.
21920 (vsx_insert_extract_v4sf_p9): Likewise.
21921 (vsx_insert_extract_v4sf_p9_2): Likewise.
21923 2017-07-25 Carl Love <cel@us.ibm.com>
21925 * doc/extend.texi: Update the built-in documentation file for the
21926 existing built-in functions
21927 vector signed char vec_cnttz (vector signed char);
21928 vector unsigned char vec_cnttz (vector unsigned char);
21929 vector signed short vec_cnttz (vector signed short);
21930 vector unsigned short vec_cnttz (vector unsigned short);
21931 vector signed int vec_cnttz (vector signed int);
21932 vector unsigned int vec_cnttz (vector unsigned int);
21933 vector signed long long vec_cnttz (vector signed long long);
21934 vector unsigned long long vec_cnttz (vector unsigned long long);
21936 2017-07-25 Andrew Pinski <apinski@cavium.com>
21938 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
21939 accesses where the use is for the first operand of a BIT_INSERT.
21941 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21944 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21945 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21947 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21949 * config/i386/gstabs.h: Delete.
21950 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
21952 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
21954 * config/i386/i386.c (ix86_decompose_address): Do not check for
21955 register RTX when looking at index_reg or base_reg.
21956 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
21958 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
21960 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21961 to update EH info here.
21963 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21965 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
21967 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21969 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
21971 2017-07-25 Torsten Duwe <duwe@suse.de>
21973 * common.opt: Introduce -fpatchable-function-entry
21974 command line option, and its variables function_entry_patch_area_size
21975 and function_entry_patch_area_start.
21976 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
21977 including a two-value parser.
21978 * target.def (print_patchable_function_entry): New target hook.
21979 * targhooks.h (default_print_patchable_function_entry): New function.
21980 * targhooks.c (default_print_patchable_function_entry): Likewise.
21981 * toplev.c (process_options): Switch off IPA-RA if
21982 patchable function entries are being generated.
21983 * varasm.c (assemble_start_function): Look at the
21984 patchable-function-entry command line switch and current
21985 function attributes and maybe generate NOP instructions by
21986 calling the print_patchable_function_entry hook.
21987 * doc/extend.texi: Document patchable_function_entry attribute.
21988 * doc/invoke.texi: Document -fpatchable_function_entry
21989 command line option.
21990 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
21992 * doc/tm.texi: Re-generate.
21994 2017-07-25 Jakub Jelinek <jakub@redhat.com>
21997 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
21998 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22000 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22002 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22004 2017-07-25 Richard Biener <rguenther@suse.de>
22006 PR tree-optimization/81455
22007 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22008 not walk in cycles when looking for guards.
22010 2017-07-25 Richard Biener <rguenther@suse.de>
22012 PR tree-optimization/81529
22013 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22014 when optimizing backedge uses.
22016 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22018 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22020 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22021 to dl_section_ref. On AIX, append an expression to subtract
22022 the size of the section length to dl_section_ref.
22024 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22026 * configure.ac: If any of the config.* scripts fail, exit 1.
22027 * configure: Regenerate.
22029 2017-07-25 Richard Biener <rguenther@suse.de>
22031 PR middle-end/81546
22032 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22033 of immediate uses, be more verbose on errors.
22035 2017-07-25 Richard Biener <rguenther@suse.de>
22037 PR tree-optimization/81510
22038 * tree-vect-loop.c (vect_is_simple_reduction): When the
22039 reduction stmt is not inside the loop bail out.
22041 2017-07-25 Richard Biener <rguenther@suse.de>
22043 PR tree-optimization/81303
22044 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22045 profitability check against LOOP_VINFO_NITERSM1.
22047 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22049 * domwalk.c (cmp_bb_postorder): Simplify.
22050 (sort_bbs_postorder): New function. Use it...
22051 (dom_walker::walk): ...here to optimize common cases.
22053 2017-07-25 Martin Liska <mliska@suse.cz>
22056 * ipa-visibility.c (function_and_variable_visibility): Make the
22057 redirection just on target that supports aliasing.
22058 Fix GNU coding style.
22060 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22063 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22065 * config/arm/rtems.h: Likewise.
22066 * config/bfin/rtems.h: Likewise.
22067 * config/i386/rtemself.h: Likewise.
22068 * config/lm32/rtems.h: Likewise.
22069 * config/m32c/rtems.h: Likewise.
22070 * config/m68k/rtemself.h: Likewise.
22071 * config/microblaze/rtems.h: Likewise.
22072 * config/mips/rtems.h: Likewise.
22073 * config/moxie/rtems.h: Likewise.
22074 * config/nios2/rtems.h: Likewise.
22075 * config/powerpcspe/rtems.h: Likewise.
22076 * config/rs6000/rtems.h: Likewise.
22077 * config/rtems.h: Likewise.
22078 * config/sh/rtems.h: Likewise.
22079 * config/sh/rtemself.h: Likewise.
22080 * config/sparc/rtemself.h: Likewise.
22082 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22085 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22086 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22087 * tree-ssa-structalias.c (alias_get_name): Same.
22089 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22092 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22093 instructions if no du chain is found.
22095 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22097 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22099 2017-07-25 Richard Biener <rguenther@suse.de>
22101 PR middle-end/81505
22102 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22105 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22107 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22108 upper-regs options.
22109 (ISA_2_7_MASKS_SERVER): Likewise.
22110 (ISA_3_0_MASKS_IEEE): Likewise.
22111 (OTHER_P8_VECTOR_MASKS): Likewise.
22112 (OTHER_VSX_VECTOR_MASKS): Likewise.
22113 (POWERPC_MASKS): Likewise.
22114 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22115 duplicate list of options.
22116 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22117 explicit -mupper-regs options.
22118 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22119 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22120 alias for -mupper-regs-df.
22121 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22122 (rs6000_init_hard_regno_mode_ok): Likewise.
22123 (rs6000_option_override_internal): Likewise.
22124 (rs6000_opt_masks): Likewise.
22125 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22126 options in terms of whether -mvsx or -mpower8-vector was used.
22127 (TARGET_UPPER_REGS_DI): Likewise.
22128 (TARGET_UPPER_REGS_SF): Likewise.
22129 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22130 -mupper-regs-* options.
22132 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22134 * passes.c (emergency_dump_function): Print some empty lines and a
22135 header before the RTL dump.
22137 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22139 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22141 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22144 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22145 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22147 2017-07-24 Carl Love <cel@us.ibm.com>
22149 * config/rs6000/rs6000-c.c: Add support for built-in functions
22150 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22151 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22152 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22153 vec_extract_fp_from_shortl): Add defines for the two builtins.
22154 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22155 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22157 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22158 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22159 * doc/extend.texi: Update the built-in documentation file for the
22160 new built-in function.
22162 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22165 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22167 * doc/generic.texi: Likewise.
22168 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22169 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22171 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22173 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22174 (aarch64_mls_elt_merge<mode>): Likewise.
22176 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22178 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22179 having __cxa_atexit.
22181 2017-07-23 Michael Collison <michael.collison@arm.com>
22183 * config/arm/arm.c (arm_option_override): Deprecate
22184 use of -mstructure-size-boundary.
22185 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22186 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22188 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22191 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22192 Reduce cost estimate for direct moves.
22194 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22197 * config/i386/i386.c (ix86_option_override_internal): Disable
22198 BMI, BMI2 and TBM instructions for -m16.
22200 2017-07-21 Carl Love <cel@us.ibm.com>
22202 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22203 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22204 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22205 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22206 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22207 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22208 VMULOSW): New enum "unspec" values.
22209 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22210 altivec_vmulosw): New patterns.
22211 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22212 VMULOSW): Add definitions.
22214 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22216 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22217 (qdf24xx): Likewise.
22218 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22219 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22220 (AARCH64_FL_V8_1): Renumber.
22221 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22222 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22223 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22224 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22225 rdma to feature modifiers list.
22227 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22229 PR middle-end/56727
22230 * ipa-visibility (function_and_variable_visibility): Convert
22231 recursive PLT call to direct call if appropriate.
22233 2017-07-21 Andrew Pinski <apinski@cavium.com>
22235 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22236 operand 1 to see if the types precision matches.
22237 * fold-const.c (operand_equal_p): Likewise.
22239 2017-07-21 Richard Biener <rguenther@suse.de>
22241 PR tree-optimization/81303
22242 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22243 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22244 (vect_peeling_hash_get_lowest_cost): Adjust.
22245 (vect_enhance_data_refs_alignment): Likewise. Use
22246 vect_get_peeling_costs_all_drs to compute the penalty for no
22247 peeling to match up costs.
22249 2017-07-21 Richard Biener <rguenther@suse.de>
22251 PR tree-optimization/81500
22252 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22253 we didn't identify a reduction path.
22255 2017-07-21 Tom de Vries <tom@codesourcery.com>
22256 Cesar Philippidis <cesar@codesourcery.com>
22258 PR gcov-profile/81442
22259 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22262 2017-07-21 Tom de Vries <tom@codesourcery.com>
22265 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22267 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22268 nvptx_override_options_after_change.
22270 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22272 * dwarf2out.c (output_file_names): Avoid double testing for
22273 dwarf_version >= 5.
22275 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22277 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22279 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22281 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22283 (try_crossjump_to_edge): Do not punt on partitioned functions.
22285 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22287 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22288 Put all BBs reachable only via paths crossing cold region to cold
22290 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22292 2016-07-21 Richard Biener <rguenther@suse.de>
22294 PR tree-optimization/81303
22295 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22296 into account prologue and epilogue iterations when raising
22297 min_profitable_iters to sth at least covering one vector iteration.
22299 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22301 * config/arm/arm.c (arm_test_cpu_arch_dat):
22304 2017-07-20 Nathan Sidwell <nathan@acm.org>
22306 Remove TYPE_METHODS.
22307 * tree.h (TYPE_METHODS): Delete.
22308 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22309 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22310 (dbxout_type_methods): Scan TYPE_FIELDS.
22311 (dbxout_type): Don't check TYPE_METHODS here.
22312 * function.c (use_register_for_decl): Always ignore register for
22313 class types when not optimizing.
22314 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22315 * tree.c (free_lang_data_in_type): Stitch out member functions and
22316 templates from TYPE_FIELDS.
22317 (build_distinct_type_copy, verify_type_variant,
22318 verify_type): Member fns are on TYPE_FIELDS.
22319 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22320 * tree-pretty-print.c (dump_generic_node): Likewise.
22322 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22325 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22326 V2TImode and V4TImode.
22327 (ix86_expand_vector_extract): Likewise.
22328 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22329 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22330 (ssescalarmode): Handle V4TImode and V2TImode.
22331 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22332 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22333 (VEXTRACTI128_MODE): New mode iterator.
22334 (splitter for *vec_extractv?ti first element): New.
22335 (VEC_INIT_MODE): New mode iterator.
22336 (vec_init<mode>): Consolidate 3 expanders into one using
22337 VEC_INIT_MODE mode iterator.
22339 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22341 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22342 non_spilled_static_chain_regno_p.
22344 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22346 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22348 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22350 * bb-reorder.c (connect_traces): Allow copying of blocks within
22353 2017-07-20 Richard Biener <rguenther@suse.de>
22355 * gimple.h (gimple_phi_result): Add gphi * overload.
22356 (gimple_phi_result_ptr): Likewise.
22357 (gimple_phi_arg): Likewise. Adjust index assert to only
22358 allow actual argument accesses rather than all slots available
22360 (gimple_phi_arg_def): Add gphi * overload.
22361 * tree-phinodes.c (make_phi_node): Initialize only actual
22363 (resize_phi_node): Clear memory not covered by old node,
22364 do not initialize excess argument slots.
22365 (reserve_phi_args_for_new_edge): Initialize new argument slot
22368 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22370 PR tree-optimization/81388
22372 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22374 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22375 by removing computation of may_be_zero.
22377 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22378 Tom de Vries <tom@codesourcery.com>
22380 PR middle-end/81030
22381 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22382 when gimple level profile disagrees with what RTL expander did.
22384 2017-07-20 Richard Biener <rguenther@suse.de>
22386 PR tree-optimization/61171
22387 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22388 (vect_analyze_stmt): Add slp instance parameter.
22389 (vectorizable_reduction): Likewise.
22390 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22391 (vect_is_simple_reduction): Deal with chains not detected
22392 as SLP reduction chain, specifically not properly associated
22393 chains containing a mix of plus/minus.
22394 (get_reduction_op): Remove.
22395 (get_initial_defs_for_reduction): Simplify, pass in whether
22396 this is a reduction chain, pass in the SLP node for the PHIs.
22397 (vect_create_epilog_for_reduction): Get the SLP instance as
22399 (vectorizable_reduction): Get the SLP instance as arg.
22400 During analysis remember the SLP node with the PHIs in the
22401 instance. Simplify getting at the vectorized reduction PHIs.
22402 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22403 through SLP instance.
22404 (vect_slp_analyze_operations): Likewise.
22405 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22406 (vect_transform_stmt): Likewise.
22408 2017-07-20 Tom de Vries <tom@codesourcery.com>
22410 PR tree-optimization/81489
22411 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22412 read of phi arg location to before loop that modifies phi.
22414 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22416 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22419 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22421 PR middle-end/81331
22422 * except.c (execute): Fix ordering issue.
22424 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22426 PR rtl-optimization/81423
22427 * combine.c (make_compound_operation_int): Don't try to optimize
22428 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22430 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22432 PR rtl-optimization/81423
22433 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22434 with a constant that is -1 in the truncated to mode.
22436 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22438 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22439 (determine_unlikely_bbs): ... here.
22440 * predict.h (propagate_unlikely_bbs_forward): Declare.
22441 * cfgexpand.c (pass_expand::execute): Use it.
22442 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22444 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22445 propagate_unlikely_bbs_forward.
22447 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22449 PR middle-end/81331
22450 * except.c (maybe_add_nop_after_section_switch): New function.
22453 2017-07-19 Tom de Vries <tom@codesourcery.com>
22455 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22457 2017-07-19 Tom de Vries <tom@codesourcery.com>
22459 * gimple.h (gimple_phi_arg): Make assert more strict.
22461 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22463 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22464 * config/rs6000/mmintrin.h: New file.
22465 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22467 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22469 PR tree-optimization/81346
22470 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22472 2017-07-19 Tom de Vries <tom@codesourcery.com>
22474 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22476 2017-07-19 Tom de Vries <tom@codesourcery.com>
22478 * config/nvptx/nvptx-modes.def: Add V2DImode.
22479 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22480 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22481 (nvptx_output_mov_insn): Handle lack of mov.b128.
22482 (nvptx_print_operand): Handle 'H' and 'L' codes.
22483 (nvptx_vector_mode_supported): Allow V2DImode.
22484 (nvptx_preferred_simd_mode): New function.
22485 (nvptx_data_alignment): New function.
22486 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22487 nvptx_preferred_simd_mode.
22488 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22490 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22492 2017-07-19 Tom de Vries <tom@codesourcery.com>
22494 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22495 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22496 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22497 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22498 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22499 (mov<VECIM>_insn): New define_insn.
22500 (define_expand "mov<VECIM>): New define_expand.
22502 2017-07-19 Tom de Vries <tom@codesourcery.com>
22504 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22506 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22508 PR tree-optimization/81346
22509 * fold-const.h (fold_div_compare, range_check_type): Declare.
22510 * fold-const.c (range_check_type): New function.
22511 (build_range_check): Use range_check_type.
22512 (fold_div_compare): No longer static, rewritten into
22513 a match.pd helper function.
22514 (fold_comparison): Don't call fold_div_compare here.
22515 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22516 as helper function.
22518 2017-07-19 Nathan Sidwell <nathan@acm.org>
22520 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22521 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22522 * tree.c (find_decls_types_r, verify_type): Use
22523 TYPE_{MIN,MAX}_VALUE_RAW.
22524 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22525 (hash_tree): Likewise.
22526 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22528 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22531 2017-07-18 Tom de Vries <tom@codesourcery.com>
22533 PR middle-end/81464
22534 * omp-expand.c (expand_omp_for_static_chunk): Handle
22535 equal-argument loop exit phi.
22537 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22540 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22541 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22542 operand 2 predicate.
22543 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22544 operand 2 predicate.
22545 (ror,rol -> rorx splitters): Use const_int_operand as
22546 operand 2 predicate.
22548 2017-06-18 Richard Biener <rguenther@suse.de>
22550 PR tree-optimization/81410
22551 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22552 the gap in the ! slp_perm SLP case after each group.
22554 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22556 PR middle-end/81463
22557 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22560 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22562 PR middle-end/81462
22563 * predict.c (set_even_probabilities): Cleanup; do not affect
22564 probabilities that are already known.
22565 (combine_predictions_for_bb): Call even when count is set.
22567 2017-07-18 Nathan Sidwell <nathan@acm.org>
22569 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22572 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22575 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22576 optimization for loop niter analysis.
22578 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22581 * config/avr/avr.c (avr_optimize_casesi): Don't use
22582 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22584 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22586 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22587 body_cost_vec from _vect_peel_extended_info.
22588 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22589 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22592 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22594 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22596 2017-07-18 Richard Biener <rguenther@suse.de>
22598 PR tree-optimization/80620
22599 PR tree-optimization/81403
22600 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22601 info when re-using a VN table entry.
22603 2017-07-18 Richard Biener <rguenther@suse.de>
22605 PR tree-optimization/81418
22606 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22607 vectype_in. Verify that with lane-reducing reduction operations
22608 we have a single def-use cycle.
22610 2017-07-17 Carl Love <cel@us.ibm.com>
22612 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22614 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22615 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22616 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22617 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22618 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22619 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22620 VMULOSW): New enum "unspec" values.
22621 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22622 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22623 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22624 altivec_vmulosw): New patterns.
22625 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22626 VMULOSW): Add definitions.
22628 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22630 * config/alpha/alpha.c: Include predict.h.
22632 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22634 * tree-vrp.c (compare_assert_loc): Fix comparison function
22635 to return predictable results.
22637 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22639 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22641 (subdi3): Likewise.
22642 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22643 * doc/invoke.texi (mexpand-adddi): Update text.
22645 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22647 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22648 that also clobbers the CC register. The old expand code is moved
22650 (*arc_clzsi2): ... here.
22651 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22652 the CC register. The old expand code is moved to ...
22653 (arc_ctzsi2): ... here.
22655 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22657 * config/arc/arc.opt (mindexed-loads): Use initial value
22658 TARGET_INDEXED_LOADS_DEFAULT.
22659 (mauto-modify-reg): Use initial value
22660 TARGET_AUTO_MODIFY_REG_DEFAULT.
22661 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22662 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22663 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22664 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22666 2017-07-17 Martin Liska <mliska@suse.cz>
22669 * opts.c (finish_options): Do not allow -fgnu-tm
22670 w/ -fsanitize={kernel-,}address. Say sorry.
22672 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22675 * tree-loop-distribution.c (classify_partition): Only assert on
22676 numer of iterations.
22677 (merge_dep_scc_partitions): Delete prameter. Update function call.
22678 (distribute_loop): Remove code handling loop with unknown niters.
22679 (pass_loop_distribution::execute): Skip loop with unknown niters.
22681 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22684 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22685 function sort_partitions_by_post_order.
22687 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22689 PR tree-optimization/81374
22690 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22691 the max index of basic blocks, rather than number of basic blocks.
22693 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22695 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22697 (arc_legitimate_pic_operand_p): Likewise.
22698 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22700 (arc_needs_pcl_p): Likewise.
22701 (arc_legitimate_pc_offset_p): Likewise.
22702 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22703 function is also used in constrains.md.
22704 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22705 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22706 PLUS. Only return true/false in known cases, otherwise assert.
22707 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22708 is already called in arc_legitimate_constant_p.
22709 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22711 (LEGITIMATE_PIC_OPERAND_P): Use
22712 arc_raw_symbolic_reference_mentioned_p function.
22713 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22718 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22719 Andrew Burgess <andrew.burgess@embecosm.com>
22721 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22722 (arc_return_address_register): New function.
22723 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22724 (arc_handle_fndecl_attribute): Add naked attribute.
22725 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22726 (TARGET_WARN_FUNC_RETURN): Likewise.
22727 (arc_allocate_stack_slots_for_args): New function.
22728 (arc_warn_func_return): Likewise.
22729 (machine_function): Change type fn_type.
22730 (arc_compute_function_type): Consider new naked function type,
22731 change function return type.
22732 (arc_must_save_register): Adapt to handle new
22733 arc_compute_function_type's return type.
22734 (arc_expand_prologue): Likewise.
22735 (arc_expand_epilogue): Likewise.
22736 (arc_return_address_regs): Delete.
22737 (arc_return_address_register): New function.
22738 (arc_epilogue_uses): Use above function.
22739 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22740 (arc_function_type): Change encoding, add naked type.
22741 (ARC_INTERRUPT_P): Change to handle the new encoding.
22742 (ARC_FAST_INTERRUPT_P): Likewise.
22743 (ARC_NORMAL_P): Define.
22744 (ARC_NAKED_P): Likewise.
22745 (arc_compute_function_type): Delete prototype.
22746 * config/arc/arc.md (in_ret_delay_slot): Use
22747 arc_return_address_register function.
22748 (simple_return): Likewise.
22749 (p_return_i): Likewise.
22751 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22753 PR tree-optimization/81428
22754 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22755 can't be built for those types.
22757 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22759 Remove stuff dead since r239246.
22761 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22762 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22763 (avr_inform_devices): Remove dead stuff.
22765 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22767 * config/arm/arm_neon.h: Fix softp typo.
22769 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22771 PR tree-optimization/81365
22772 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22773 aggregate moves onto bb predecessor edges, make sure there are no
22774 loads that could alias the lhs in between the start of bb and the
22777 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22780 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22781 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22782 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22784 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22786 PR tree-optimization/81396
22787 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22788 (init_symbolic_number): Initialize it to 1.
22789 (perform_symbolic_merge): Add n_ops from both operands into the new
22791 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22792 without base_addr as useless if they need more than one operation.
22793 (bswap_replace): Handle !bswap case for NULL base_addr.
22795 2017-07-17 Tom de Vries <tom@codesourcery.com>
22798 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22801 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22803 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22804 conditional builtin define __FIX_LEON3FT_B2BST.
22806 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22808 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22809 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22812 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22814 PR rtl-optimization/81424
22815 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22816 to remove potential trapping from operands if -fnon-call-exceptions.
22818 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22820 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22821 profile_proability for scalling.
22822 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22824 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22826 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22828 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22830 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22831 fixpoint arithmetics.
22833 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22835 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22836 fixpoint arithmetics.
22838 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22840 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22841 fixpoint arithmetics.
22843 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22845 * profile-count.h (profile_probability::from_reg_br_prob_note,
22846 profile_probability::to_reg_br_prob_note): New functions.
22847 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22848 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22849 * predict.c (probability_reliable_p): Update.
22850 (edge_probability_reliable_p): Update.
22851 (br_prob_note_reliable_p): Update.
22852 (invert_br_probabilities): Update.
22853 (add_reg_br_prob_note): New function.
22854 (combine_predictions_for_insn): Update.
22855 * asan.c (asan_clear_shadow): Update.
22856 * cfgbuild.c (compute_outgoing_frequencies): Update.
22857 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22858 (update_br_prob_note): Update.
22859 (rtl_verify_edges): Update.
22860 (purge_dead_edges): Update.
22861 (fixup_reorder_chain): Update.
22862 * emit-rtl.c (try_split): Update.
22863 * ifcvt.c (cond_exec_process_insns): Update.
22864 (cond_exec_process_if_block): Update.
22865 (dead_or_predicable): Update.
22866 * internal-fn.c (expand_addsub_overflow): Update.
22867 (expand_neg_overflow): Update.
22868 (expand_mul_overflow): Update.
22869 * loop-doloop.c (doloop_modify): Update.
22870 * loop-unroll.c (compare_and_jump_seq): Update.
22871 * optabs.c (emit_cmp_and_jump_insn_1): Update.
22872 * predict.h: Update.
22873 * reorg.c (mostly_true_jump): Update.
22875 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
22876 * config/alpha/alpha.c (emit_unlikely_jump): Update.
22877 * config/arc/arc.c: (emit_unlikely_jump): Update.
22878 * config/arm/arm.c: (emit_unlikely_jump): Update.
22879 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
22880 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
22881 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
22882 (ix86_print_operand): Update.
22883 (ix86_split_fp_branch): Update.
22884 (predict_jump): Update.
22885 * config/ia64/ia64.c (ia64_print_operand): Update.
22886 * config/mmix/mmix.c (mmix_print_operand): Update.
22887 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
22888 (rs6000_expand_split_stack_prologue): Update.
22889 * config/rs6000/rs6000.c: Update.
22890 * config/s390/s390.c (s390_expand_vec_strlen): Update.
22891 (s390_expand_vec_movstr): Update.
22892 (s390_expand_cs_tdsi): Update.
22893 (s390_expand_split_stack_prologue): Update.
22894 * config/sh/sh.c (sh_print_operand): Update.
22895 (expand_cbranchsi4): Update.
22896 (expand_cbranchdi4): Update.
22897 * config/sparc/sparc.c (output_v9branch): Update.
22898 * config/spu/spu.c (get_branch_target): Update.
22899 (ea_load_store_inline): Update.
22900 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
22901 * config/tilepro/tilepro.c: Update.
22903 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22905 * gimplify.c (mostly_copy_tree_r): Revert latest change.
22906 (gimplify_save_expr): Likewise.
22908 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22910 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
22912 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22914 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
22916 * timevar.def (TV_IPA_FNSUMMARY): Define.
22918 2017-07-16 Daniel Cederman <cederman@gaisler.com>
22920 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
22921 to back store errata sensitive sequence from being generated.
22922 (sqrtdf2_fix): Likewise.
22924 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22926 * tree-ssa-threadupdate.c (compute_path_counts,
22927 update_joiner_offpath_counts): Use profile_probability.
22929 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
22932 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22934 * config/arm/arm-c.c (arm_cpu_builtins): Define
22935 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22937 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22939 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22940 array entries to represent __ieee128 versions of the
22941 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
22942 scalar_extract_sig, and scalar_insert_exp built-in functions.
22943 (altivec_resolve_overloaded_builtin): Add special case handling
22944 for the __builtin_scalar_insert_exp function, as represented by
22945 the P9V_BUILTIN_VEC_VSIEDP constant.
22946 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
22947 exponent support for __ieee128 argument.
22948 (VSESQP): Add scalar extract signature support for __ieee128
22950 (VSTDCNQP): Add scalar test negative support for __ieee128
22952 (VSIEQP): Add scalar insert exponent support for __int128 argument
22953 with __ieee128 result.
22954 (VSIEQPF): Add scalar insert exponent support for __ieee128
22955 argument with __ieee128 result.
22956 (VSTDCQP): Add scalar test data class support for __ieee128
22958 (VSTDCNQP): Add overload support for scalar test negative with
22959 __ieee128 argument.
22960 (VSTDCQP): Add overload support for scalar test data class
22961 __ieee128 argument.
22962 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
22963 UNSPEC_VSX_SXSIGDP.
22964 (UNSPEC_VSX_SIEXPQP): New constant.
22965 (xsxexpqp): New insn for VSX scalar extract exponent quad
22967 (xsxsigqp): New insn for VSX scalar extract significand quad
22969 (xsiexpqpf): New insn for VSX scalar insert exponent quad
22970 precision with floating point argument.
22971 (xststdcqp): New expand for VSX scalar test data class quad
22973 (xststdcnegqp): New expand for VSX scalar test negative quad
22975 (xststdcqp): New insn to match expansions for VSX scalar test data
22976 class quad precision and VSX scalar test negative quad precision.
22977 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
22978 special case operand checking to enforce that second operand of
22979 VSX scalar test data class with quad precision argument is a 7-bit
22981 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22982 prototypes and descriptions of __ieee128 versions of
22983 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
22984 scalar_test_data_class, and scalar_test_neg built-in functions.
22986 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22988 PR tree-optimization/81162
22989 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
22990 replace a negate with an add.
22992 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
22994 * doc/invoke.texi (arm/-mcpu): Document +crypto.
22996 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22998 * config/arm/arm-c.c (arm_cpu_builtins): Define
22999 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23001 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23003 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23004 (armv8-r): Set ARM Cortex-R52 as default CPU.
23005 * config/arm/arm-tables.opt: Regenerate.
23006 * config/arm/arm-tune.md: Regenerate.
23007 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23009 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23010 extension for -mcpu=cortex-r52.
23012 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23014 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23015 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23016 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23017 (fp-armv8): Define it as FP_ARMv8 only.
23018 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23019 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23021 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23022 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23023 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23024 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23025 than TARGET_FPU_ARMV8.
23026 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23027 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23028 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23030 * config/arm/neon.md (neon_vrint): Likewise.
23031 (neon_vcvt): Likewise.
23032 (neon_<fmaxmin_op><mode>): Likewise.
23033 (<fmaxmin><mode>3): Likewise.
23034 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23035 * config/arm/predicates.md (arm_cond_move_operator): Check against
23036 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23038 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23040 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23041 to top of function.
23043 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23045 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23046 loop in comment with memset.
23048 2017-07-14 Martin Liska <mliska@suse.cz>
23050 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23051 * dwarf2out.c (is_java): Remove the function.
23052 (output_pubname): Remove usage of the function.
23053 (lower_bound_default): Remove usage of DW_LANG_Java.
23054 (gen_compile_unit_die): Likewise.
23055 * gcc.c: Remove compiler defaults for .java and .zip files.
23056 * gimple-expr.c (remove_suffix): Change as there's no longer
23057 extension than 4-letter one.
23058 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23059 (gimplify_save_expr): Likewise.
23060 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23061 as it's possible even for other languages than Java.
23062 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23063 * lto-opts.c (lto_write_options): Remove reference to Java.
23064 * opts.c (strip_off_ending): Update file extension handling.
23065 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23066 * tree-eh.c (lower_resx): Likewise.
23067 * tree.c (free_lang_data_in_type): Remove dead code.
23068 (find_decls_types_r): Likewise.
23069 (build_common_builtin_nodes): Remove Java from a comment.
23070 (verify_type): Remove dead code.
23071 * varasm.c (assemble_external): Remove Java from a comment.
23073 2017-07-14 Martin Liska <mliska@suse.cz>
23075 * opts.c (finish_options): Add quotes.
23076 (common_handle_option): Likewise.
23078 2017-07-14 Martin Liska <mliska@suse.cz>
23080 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23081 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23082 Remove N_SO_PASCAL.
23083 * dwarf2out.c (lower_bound_default): Do not handle
23085 (gen_compile_unit_die): Likewise.
23086 * gcc.c: Remove default extension binding for GNU Pascal.
23087 * stmt.c: Remove Pascal language from a comment.
23088 * xcoffout.c: Likewise.
23090 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23093 * diagnostic-show-locus.c (fixit_cmp): New function.
23094 (layout::layout): Sort m_fixit_hints.
23095 (column_range::column_range): Assert that the values are valid.
23096 (struct char_span): New struct.
23097 (correction::overwrite): New method.
23098 (struct source_line): New struct.
23099 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23100 calls in terms of classes source_line and char_span, and
23101 correction::overwrite.
23102 (selftest::test_overlapped_fixit_printing_2): New function.
23103 (selftest::diagnostic_show_locus_c_tests): Call it.
23105 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23107 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23108 early if there is no lhs.
23110 2017-07-13 Martin Liska <mliska@suse.cz>
23112 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23113 (gen_reference_type_die): Likewise.
23114 * stor-layout.c: Remove Pascal-related comment.
23116 2017-07-13 Martin Liska <mliska@suse.cz>
23118 * opts.c (finish_options): Add quotes to error messages.
23119 (parse_sanitizer_options): Likewise.
23121 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23123 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23125 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23127 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23129 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23131 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23133 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23135 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23138 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23139 provides the hardware capability bits, define the macro
23140 __BUILTIN_CPU_SUPPORTS__.
23141 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23142 if GLIBC does not provide the hardware capability bits. Add a
23143 gcc_unreachable call if the built-in cpu function is neither
23144 __builtin_cpu_is nor __builtin_cpu_supports.
23145 (rs6000_get_function_versions_dispatcher): Change the warning
23146 that an old GLIBC is used which does not export the capability
23147 bits to be an error.
23148 * doc/extend.texi (target_clones attribute): Document the
23149 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23150 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23151 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23152 the macros defined by GCC if the newer GLIBC is available.
23154 2017-07-12 Jeff Law <law@redhat.com>
23156 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23157 remaining includes slightly.
23158 * config/riscv/riscv-builtins.c: Include profile-count.h.
23160 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23163 * config/avr/avr.c (avr_set_current_function): In diagnostic
23164 messages: Quote keywords and (parts of) identifiers.
23165 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23168 2017-07-12 Carl Love <cel@us.ibm.com>
23170 * config/rs6000/rs6000-c.c: Add support for built-in functions
23171 vector bool char vec_revb (vector bool char);
23172 vector bool short vec_revb (vector short char);
23173 vector bool int vec_revb (vector bool int);
23174 vector bool long long vec_revb (vector bool long long);
23175 * doc/extend.texi: Update the built-in documentation file for the
23176 new built-in functions.
23178 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23180 * config/s390/s390.md: Remove movcc splitter.
23182 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23184 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23185 load/store on condition.
23187 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23190 * config/avr/avr.c (avr_encode_section_info)
23191 [progmem && !TREE_READONLY]: Error if progmem object needs
23194 2017-07-11 Michael Collison <michael.collison@arm.com>
23196 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23199 2017-07-11 Carl Love <cel@us.ibm.com>
23201 * config/rs6000/rs6000-c.c: Add support for builtins
23202 vector unsigned int vec_parity_lsbb (vector signed int);
23203 vector unsigned int vec_parity_lsbb (vector unsigned int);
23204 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23205 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23206 vector unsigned long long vec_parity_lsbb (vector signed long long);
23207 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23208 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23209 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23210 * doc/extend.texi: Update the built-in documentation file for the
23211 new built-in functions.
23213 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23215 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23216 (layout::m_primary_loc): New field.
23217 (layout::layout): Initialize new field. Move location filtering
23218 logic from here to...
23219 (layout::maybe_add_location_range): ...this new method. Add
23220 support for filtering to just the lines already specified by other
23222 (layout::will_show_line_p): New method.
23223 (gcc_rich_location::add_location_if_nearby): New method.
23224 (selftest::test_add_location_if_nearby): New test function.
23225 (selftest::diagnostic_show_locus_c_tests): Call it.
23226 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23229 2017-07-11 Tom de Vries <tom@codesourcery.com>
23231 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23232 (bb_first_real_insn): New function.
23233 (nvptx_single): Add extra initialization of broadcasted condition
23236 2017-07-11 Nathan Sidwell <nathan@acm.org>
23238 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23240 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23242 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23243 Binutils doc as TEXI will mess them up.
23244 * doc/invoke.texi (AVR Options): Same here.
23246 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23248 * config/sparc/sparc.opt (mfix-ut700): New option.
23249 (mfix-gr712rc): Likewise.
23250 (sparc_fix_b2bst): New variable.
23251 * doc/invoke.texi (SPARC options): Document them.
23252 (ARM options): Fix warnings.
23253 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23254 instructions to prevent sequences that can trigger the store-store
23255 errata for certain LEON3FT processors.
23256 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23257 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23258 * config/sparc/sparc.md (fix_b2bst): New attribute.
23259 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23261 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23264 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23266 (*rsqrtsf2_sse): Ditto.
23268 (div<mode>3): Macroize insn from divdf3 and divsf3
23269 using MODEF mode iterator.
23271 2017-07-10 Martin Sebor <msebor@redhat.com>
23273 PR tree-optimization/80397
23274 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23275 instead of testing for equality to INTEGER_TYPE.
23277 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23279 * config.gcc: Remove uclibc from arc target spec.
23281 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23283 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23285 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23288 * lto-wrapper.c (remove_option): New function.
23289 (merge_and_complain): Merge PIC/PIE options more realistically.
23291 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23293 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23297 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23298 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23299 * config.in: Regenerate.
23300 * configure: Regenerate.
23301 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23302 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23303 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23304 (TARGET_GASISR_PROLOGUES): ...target mask.
23305 * common/config/avr/avr-common.c
23306 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23307 Set -mgas-isr-prologues.
23308 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23309 INSERT_PASS_BEFORE for it.
23310 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23311 * config/avr/avr.c (avr_option_override)
23312 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23313 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23314 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23315 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23316 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23318 (make_avr_pass_pre_proep): New function.
23319 (emit_push_sfr) <treg>: Add argument to function and use it
23320 instead of TMP_REG.
23321 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23322 and set machine->gasisr.yes.
23323 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23324 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23325 __gcc_isr.n_pushed to .L__stack_usage.
23326 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23327 (avr_asm_final_postscan_insn): ...this new static function.
23328 * config/avr/avr.h (machine_function)
23329 <is_no_gccisr, use_L__stack_usage>: New fields.
23330 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23331 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23332 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23333 (gasisr, *gasisr): New expander and insn.
23334 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23335 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23336 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23338 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23340 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23343 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23345 Move jump-tables out of .text again.
23348 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23349 (ASM_OUTPUT_ADDR_VEC): New function.
23350 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23351 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23352 INSN_ADDRESSes as asm comment.
23353 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23354 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23355 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23356 * config/avr/avr.md (*tablejump): Adjust comment.
23357 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23358 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23360 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23361 (avr_output_addr_vec): New proto.
23362 (avr_log_t) <insn_addresses>: New field.
23364 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23367 * config/i386/i386.c (ix86_function_arg_advance): Set
23368 outgoing_args_on_stack to true if there are outgoing arguments
23370 (ix86_function_arg): Likewise.
23371 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23372 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23373 * config/i386/i386.h (machine_function): Add
23374 outgoing_args_on_stack.
23376 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23378 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23379 supporting pthreds.
23380 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23382 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23384 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23385 (REAL_H): Remove $(MACHMODE_H).
23386 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23388 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23389 $(MACHMODE_H) and double-int.h.
23390 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23392 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23395 2017-07-07 Andrew Pinski <apinski@cavium.com>
23397 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23398 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23400 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23402 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23403 Add warning if GCC was not configured to link against a GLIBC that
23404 exports the hardware capability bits.
23405 (make_resolver_func): Make resolver function private and not a
23406 COMDAT function. Create the name with clone_function_name instead
23407 of make_unique_name.
23410 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23411 correct operand in doing the split.
23413 2017-07-07 Carl Love <cel@us.ibm.com>
23415 * config/rs6000/rs6000-c: Add support for built-in function
23416 vector unsigned short vec_pack_to_short_fp32 (vector float,
23418 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23419 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23420 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23421 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23422 (convert_4f32_8i16): Add define_expand.
23423 * doc/extend.texi: Update the built-in documentation file for the
23424 new built-in function.
23426 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23428 * config/sparc/m8.md: New file.
23429 * config/sparc/sparc.md: Include m8.md.
23431 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23433 * config/sparc/sparc.opt: New option -mvis4b.
23434 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23435 (sparc_option_override): Handle VIS4B.
23436 (enum sparc_builtins): Define
23437 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23438 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23439 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23440 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23441 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23442 (check_constant_argument): New function.
23443 (sparc_vis_init_builtins): Define builtins
23444 __builtin_vis_dictunpack{8,16,32},
23445 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23446 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23447 __builtin_vis_fpcmpde{8,16,32}shl and
23448 __builtin_vis_fpcmpur{8,16,32}shl.
23449 (sparc_expand_builtin): Check that the constant operands to
23450 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23451 constant and in range.
23452 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23454 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23455 (SPARC_IMM5_P): Likewise.
23456 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23457 (enabled): Handle vis4b.
23458 (UNSPEC_DICTUNPACK): New unspec.
23459 (UNSPEC_FPCMPSHL): Likewise.
23460 (UNSPEC_FPUCMPSHL): Likewise.
23461 (UNSPEC_FPCMPDESHL): Likewise.
23462 (UNSPEC_FPCMPURSHL): Likewise.
23463 (cpu_feature): New CPU feature `vis4b'.
23464 (dictunpack{8,16,32}): New insns.
23465 (FPCSMODE): New mode iterator.
23466 (fpcscond): New code iterator.
23467 (fpcsucond): Likewise.
23468 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23469 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23470 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23471 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23472 * config/sparc/constraints.md: Define constraints `q' for unsigned
23473 2-bit integer constants and `t' for unsigned 5-bit integer
23475 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23477 (imm5_operand_dictunpack16): Likewise.
23478 (imm5_operand_dictunpack32): Likewise.
23479 (imm2_operand): Likewise.
23480 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23481 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23482 ditunpack* and fpcmp*shl builtins.
23484 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23486 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23487 * config.in: Add HAVE_AS_SPARC6 define.
23488 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23490 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23492 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23493 (CPP_CPU_SPEC): Handle m8.
23494 (ASM_CPU_SPEC): Likewise.
23495 * config/sparc/sparc-opts.h (enum processor_type): Add
23497 * config/sparc/sparc.c (m8_costs): New struct.
23498 (sparc_option_override): Handle TARGET_CPU_m8.
23499 (sparc32_initialize_trampoline): Likewise.
23500 (sparc64_initialize_trampoline): Likewise.
23501 (sparc_issue_rate): Likewise.
23502 (sparc_register_move_cost): Likewise.
23503 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23504 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23505 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23506 (CPP_CPU_SPEC): Handle M8.
23507 (ASM_CPU_SPEC): Likewise.
23508 (AS_M8_FLAG): Define.
23509 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23510 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23511 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23513 * configure: Regenerate.
23514 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23517 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23519 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23521 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23522 ("*movdi_insn_sp32"): Do not set v3pipe.
23523 ("*movsi_insn"): Likewise.
23524 ("*movdi_insn_sp64"): Likewise.
23525 ("*movsf_insn"): Likewise.
23526 ("*movdf_insn_sp32"): Likewise.
23527 ("*movdf_insn_sp64"): Likewise.
23528 ("*zero_extendsidi2_insn_sp64"): Likewise.
23529 ("*sign_extendsidi2_insn"): Likewise.
23530 ("*mov<VM32:mode>_insn"): Likewise.
23531 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23532 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23533 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23534 ("<vlop:code><VL:mode>3"): Likewise.
23535 ("*not_<vlop:code><VL:mode>3"): Likewise.
23536 ("*nand<VL:mode>_vis"): Likewise.
23537 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23538 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23539 ("one_cmpl<VL:mode>2"): Likewise.
23540 ("faligndata<VM64:mode>_vis"): Likewise.
23541 ("alignaddrsi_vis"): Likewise.
23542 ("alignaddrdi_vis"): Likweise.
23543 ("alignaddrlsi_vis"): Likewise.
23544 ("alignaddrldi_vis"): Likewise.
23545 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23546 ("bmaskdi_vis"): Likewise.
23547 ("bmasksi_vis"): Likewise.
23548 ("bshuffle<VM64:mode>_vis"): Likewise.
23549 ("cmask8<P:mode>_vis"): Likewise.
23550 ("cmask16<P:mode>_vis"): Likewise.
23551 ("cmask32<P:mode>_vis"): Likewise.
23552 ("pdistn<P:mode>_vis"): Likewise.
23553 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23555 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23557 * config/sparc/sparc.md ("subtype"): New insn attribute.
23558 ("*wrgsr_sp64"): Set insn subtype.
23559 ("*rdgsr_sp64"): Likewise.
23560 ("alignaddrsi_vis"): Likewise.
23561 ("alignaddrdi_vis"): Likewise.
23562 ("alignaddrlsi_vis"): Likewise.
23563 ("alignaddrldi_vis"): Likewise.
23564 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23565 ("fexpand_vis"): Likewise.
23566 ("fpmerge_vis"): Likewise.
23567 ("faligndata<VM64:mode>_vis"): Likewise.
23568 ("bshuffle<VM64:mode>_vis"): Likewise.
23569 ("cmask8<P:mode>_vis"): Likewise.
23570 ("cmask16<P:mode>_vis"): Likewise.
23571 ("cmask32<P:mode>_vis"): Likewise.
23572 ("fchksm16_vis"): Likewise.
23573 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23574 ("fmean16_vis"): Likewise.
23575 ("fp<plusminus_insn>64_vis"): Likewise.
23576 ("<plusminus_insn>v8qi3"): Likewise.
23577 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23578 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23579 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23580 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23581 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23582 ("*movqi_insn"): Likewise.
23583 ("*movhi_insn"): Likewise.
23584 ("*movsi_insn"): Likewise.
23585 ("movsi_pic_gotdata_op"): Likewise.
23586 ("*movdi_insn_sp32"): Likewise.
23587 ("*movdi_insn_sp64"): Likewise.
23588 ("movdi_pic_gotdata_op"): Likewise.
23589 ("*movsf_insn"): Likewise.
23590 ("*movdf_insn_sp32"): Likewise.
23591 ("*movdf_insn_sp64"): Likewise.
23592 ("*zero_extendhisi2_insn"): Likewise.
23593 ("*zero_extendqihi2_insn"): Likewise.
23594 ("*zero_extendqisi2_insn"): Likewise.
23595 ("*zero_extendqidi2_insn"): Likewise.
23596 ("*zero_extendhidi2_insn"): Likewise.
23597 ("*zero_extendsidi2_insn_sp64"): Likewise.
23598 ("ldfsr"): Likewise.
23599 ("prefetch_64"): Likewise.
23600 ("prefetch_32"): Likewise.
23601 ("tie_ld32"): Likewise.
23602 ("tie_ld64"): Likewise.
23603 ("*tldo_ldub_sp32"): Likewise.
23604 ("*tldo_ldub1_sp32"): Likewise.
23605 ("*tldo_ldub2_sp32"): Likewise.
23606 ("*tldo_ldub_sp64"): Likewise.
23607 ("*tldo_ldub1_sp64"): Likewise.
23608 ("*tldo_ldub2_sp64"): Likewise.
23609 ("*tldo_ldub3_sp64"): Likewise.
23610 ("*tldo_lduh_sp32"): Likewise.
23611 ("*tldo_lduh1_sp32"): Likewise.
23612 ("*tldo_lduh_sp64"): Likewise.
23613 ("*tldo_lduh1_sp64"): Likewise.
23614 ("*tldo_lduh2_sp64"): Likewise.
23615 ("*tldo_lduw_sp32"): Likewise.
23616 ("*tldo_lduw_sp64"): Likewise.
23617 ("*tldo_lduw1_sp64"): Likewise.
23618 ("*tldo_ldx_sp64"): Likewise.
23619 ("*mov<VM32:mode>_insn"): Likewise.
23620 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23621 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23623 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23625 * config/sparc/sparc.md ("type"): New insn type viscmp.
23626 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23628 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23629 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23630 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23631 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23633 ("n7_vis_logical_11cycle"): Likewise.
23634 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23635 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23636 * config/sparc/niagara.md ("niag_vis"): Likewise.
23637 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23638 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23640 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23642 * config/sparc/sparc.md: New instruction type `bmask'.
23643 (bmaskdi_vis): Use the `bmask' type.
23644 (bmasksi_vis): Likewise.
23645 * config/sparc/ultra3.md (us3_array): Likewise.
23646 * config/sparc/niagara7.md (n7_array): Likewise.
23647 * config/sparc/niagara4.md (n4_array): Likewise.
23648 * config/sparc/niagara2.md (niag2_vis): Likewise.
23649 (niag3_vis): Likewise.
23650 * config/sparc/niagara.md (niag_vis): Likewise.
23652 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23654 * ipa-comdats.c: Remove optimize check from gate.
23655 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23656 for functions not optimized.
23657 (ipa_fn_summary_read): Skip optimize check.
23658 (ipa_fn_summary_write): Likewise.
23659 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23661 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23663 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23664 (check_callers): Give up on caller which is not optimized.
23665 (inline_small_functions): Likewise.
23666 (ipa_inline): Do not give up when not optimizing.
23667 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23668 away unoptimizes cdtors.
23669 (whole_program_function_and_variable_visibility): Do
23670 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23671 * ipa.c (process_references): Do not check optimize.
23672 (symbol_table::remove_unreachable_nodes): Update optimize check.
23673 (set_writeonly_bit): Update optimize check.
23674 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23675 (pass_ipa_single_use::gate): Remove.
23677 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23679 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23680 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23681 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23682 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23683 permute_load, permute_store, adjust_extract, adjust_splat,
23684 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23685 replace_swap_with_copy, dump_swap_insn_table,
23686 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23687 recombine_lvx_pattern, recombine_stvx_pattern,
23688 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23689 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23690 to file rs6000-p8swap.c.
23691 * config/rs6000/rs6000-p8swap.c: New file.
23692 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23693 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23694 and rs6000*-*-* targets.
23696 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23698 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23700 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23702 * lto-wrapper.c (merge_and_complain): Do not merge
23703 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23704 fsigned_zeros, ftrapping_math, fwrapv.
23705 (append_compiler_options): Do not track these options.
23706 (append_linker_options): Likewie
23708 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23710 * cgraphunit.c (cgraph_node::finalize_function): When
23711 !flag_toplevel_reorde set no_reorder flag.
23712 (varpool_node::finalize_decl): Likewise.
23713 (symbol_table::compile): Drop no toplevel reorder path.
23715 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23717 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23718 edges; zero probability is not better than uninitialized.
23720 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23722 * asan.h (asan_sanitize_allocas_p): Declare.
23723 * asan.c (asan_sanitize_allocas_p): New function.
23724 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23725 (handle_builtin_alloca): Likewise.
23726 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23727 if !asan_sanitize_allocas_p.
23728 * params.def (asan-instrument-allocas): Add new option.
23729 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23730 * opts.c (common_handle_option): Disable allocas sanitization for
23733 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23735 * asan.c: Include gimple-fold.h.
23736 (get_last_alloca_addr): New function.
23737 (handle_builtin_stackrestore): Likewise.
23738 (handle_builtin_alloca): Likewise.
23739 (asan_emit_allocas_unpoison): Likewise.
23740 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23741 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23742 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23743 (instrument_builtin_call): Pass gimple iterator to
23744 get_mem_refs_of_builtin_call.
23745 (last_alloca_addr): New global.
23746 * asan.h (asan_emit_allocas_unpoison): Declare.
23747 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23748 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23749 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23750 if function calls alloca.
23751 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23752 * gimple-fold.h (replace_call_with_value): Declare.
23753 * internal-fn.c: Include asan.h.
23754 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23755 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23757 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23759 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23760 (C_SELFTEST_FLAGS): New.
23761 (CPP_SELFTEST_FLAGS): New.
23762 (SELFTEST_DEPS): New, from deps of s-selftest.
23763 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23764 (CPP_SELFTEST_DEPS): New.
23765 (selftest): Add dependency on s-selftest-c++.
23766 (s-selftest): Rename to...
23767 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23768 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23769 than SELFTEST_FLAGS.
23770 (selftest-gdb): Rename to...
23771 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23773 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23774 (selftest-valgrind): Rename to...
23775 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23777 (selftest-valgrind): Reintroduce as an alias for
23778 selftest-c-valgrind.
23779 (s-selftest-c++): New.
23780 (selftest-c++-gdb): New.
23781 (selftest-c++-valgrind): New.
23783 2017-07-06 Olivier Hainque <hainque@adacore.com>
23785 * gcc.c (process_command): When deciding if undefined variables
23786 should be ignored when processing specs, accept "gcc -v" as well.
23788 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23790 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23791 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23793 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23795 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23796 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23797 * config/arm/arm-tables.opt: Regenerate.
23798 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23800 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23802 2017-07-06 Carl Love <cel@us.ibm.com>
23804 * ChangeLog: Clean up from mid air collision
23806 2017-07-06 Carl Love <cel@us.ibm.com>
23808 * config/rs6000/rs6000-c.c: Add support for built-in functions
23809 vector signed int vec_subc (vector signed int, vector signed int);
23810 vector signed __int128 vec_subc (vector signed __int128,
23811 vector signed __int128);
23812 vector unsigned __int128 vec_subc (vector unsigned __int128,
23813 vector unsigned __int128);
23814 vector signed int vec_sube (vector signed int, vector signed int,
23815 vector signed int);
23816 vector unsigned int vec_sube (vector unsigned int,
23817 vector unsigned int,
23818 vector unsigned int);
23819 vector signed __int128 vec_sube (vector signed __int128,
23820 vector signed __int128,
23821 vector signed__int128);
23822 vector unsigned __int128 vec_sube (vector unsigned __int128,
23823 vector unsigned __int128,
23824 vector unsigned __int128);
23825 vector signed int vec_subec (vector signed int, vector signed int,
23826 vector signed int);
23827 vector unsigned int vec_subec (vector unsigned int,
23828 vector unsigned int,
23829 vector unsigned int);
23830 vector signed __int128 vec_subec (vector signed __int128,
23831 vector signed __int128,
23832 vector signed__int128);
23833 vector unsigned __int128 vec_subec (vector unsigned __int128,
23834 vector unsigned __int128,
23835 vector unsigned __int128);
23836 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23837 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23838 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23839 BU_ALTIVEC_OVERLOAD_X definitions.
23840 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23841 * doc/extend.texi: Update the built-in documentation file for the new
23842 built-in functions.
23844 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23847 * diagnostic-show-locus.c (layout::layout): Use start and finish
23848 spelling location for the start and finish of each range.
23849 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23850 Add unused aspect param.
23851 * input.c (expand_location_1): Add "aspect" param, and use it
23852 to access the correct part of the location.
23853 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23855 (expand_location_to_spelling_point): Likewise.
23856 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23857 param, and pass it to expand_location_1.
23859 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
23861 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
23862 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
23863 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
23864 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
23865 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
23866 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
23867 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
23868 _mm_maskz_getmant_ss): New intrinsics.
23869 (__builtin_ia32_getexpss128_mask): Changed to ...
23870 __builtin_ia32_getexpss128_round ... this.
23871 (__builtin_ia32_getexpsd128_mask): Changed to ...
23872 __builtin_ia32_getexpsd128_round ... this.
23873 * config/i386/i386-builtin-types.def
23874 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
23875 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
23876 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
23877 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
23878 __builtin_ia32_getmantss_mask_round): New builtins.
23879 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
23880 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
23881 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
23882 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
23883 * config/i386/sse.md
23884 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
23885 avx512f_sgetexp<mode><mask_scalar_name>
23886 <round_saeonly_scalar_name> ... this.
23887 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
23888 %0, %1, %2<round_saeonly_op3>}): Changed to ...
23889 vgetexp<ssescalarmodesuffix>
23890 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23891 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
23892 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
23893 avx512f_vgetmant<mode><mask_scalar_name>
23894 <round_saeonly_scalar_name> ... this.
23895 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23896 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
23897 vgetmant<ssescalarmodesuffix>
23898 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
23899 %0<mask_scalar_operand4>, %1, %2
23900 <round_saeonly_scalar_mask_op4>, %3} ... this.
23901 * config/i386/subst.md (mask_scalar_operand4,
23902 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
23903 round_saeonly_scalar_nimm_predicate): New subst attributes.
23905 2017-07-06 Julia Koval <julia.koval@intel.com>
23907 * config/i386/i386.c (ix86_erase_embedded_rounding):
23908 Remove code for old rounding pattern.
23910 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
23912 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
23914 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
23916 * doc/sourcebuild.texi (Test Directives, Variants of
23917 dg-require-support): Add documentation for dg-require-stack-check.
23919 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
23921 * config/i386/subst.md (mask_scalar, round_scalar,
23922 round_saeonly_scalar): New meta-templates.
23923 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
23924 round_scalar_mask_operand3, round_scalar_mask_op3,
23925 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
23926 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
23927 round_saeonly_scalar_constraint,
23928 round_saeonly_scalar_prefix): New subst attribute.
23929 * config/i386/sse.md
23930 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
23931 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
23932 <round_scalar_name> ... this.
23933 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
23934 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
23935 <round_scalar_name> ... this.
23936 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
23937 <sse>_vm<code><mode>3<mask_scalar_name>
23938 <round_saeonly_scalar_name> ... this.
23939 (v<plusminus_mnemonic><ssescalarmodesuffix>
23940 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23941 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23942 v<plusminus_mnemonic><ssescalarmodesuffix>
23943 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23944 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23945 (v<multdiv_mnemonic><ssescalarmodesuffix>
23946 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23947 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23948 v<multdiv_mnemonic><ssescalarmodesuffix>
23949 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23950 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23951 (v<maxmin_float><ssescalarmodesuffix>
23952 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
23953 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
23954 v<maxmin_float><ssescalarmodesuffix>
23955 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23956 %0<mask_scalar_operand3>, %1, %<iptr>2
23957 <round_saeonly_scalar_mask_op3>} ... this.
23959 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
23961 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
23962 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
23964 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
23965 Alan Hayward <alan.hayward@arm.com>
23966 David Sherwood <david.sherwood@arm.com>
23968 * combine.c (simplify_if_then_else): Remove "enum" before
23970 * compare-elim.c (can_eliminate_compare): Likewise.
23971 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
23973 (aarch64_lookup_simd_builtin_type): Likewise.
23974 (aarch64_simd_builtin_type): Likewise.
23975 (aarch64_init_simd_builtin_types): Likewise.
23976 (aarch64_simd_expand_args): Likewise.
23977 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
23979 (aarch64_reverse_mask): Likewise.
23980 (aarch64_simd_emit_reg_reg_move): Likewise.
23981 (aarch64_gen_adjusted_ldpstp): Likewise.
23982 (aarch64_ccmp_mode_to_code): Likewise.
23983 (aarch64_operands_ok_for_ldpstp): Likewise.
23984 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23985 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
23987 (aarch64_min_divisions_for_recip_mul): Likewise.
23988 (aarch64_reassociation_width): Likewise.
23989 (aarch64_get_condition_code_1): Likewise.
23990 (aarch64_simd_emit_reg_reg_move): Likewise.
23991 (aarch64_simd_attr_length_rglist): Likewise.
23992 (aarch64_reverse_mask): Likewise.
23993 (aarch64_operands_ok_for_ldpstp): Likewise.
23994 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23995 (aarch64_gen_adjusted_ldpstp): Likewise.
23996 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
23998 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
23999 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24000 (arm_lookup_simd_builtin_type): Likewise.
24001 (arm_simd_builtin_type): Likewise.
24002 (arm_init_simd_builtin_types): Likewise.
24003 (arm_expand_builtin_args): Likewise.
24004 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24005 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24006 (ft32_setup_incoming_varargs): Likewise.
24007 (ft32_function_arg): Likewise.
24008 (ft32_function_arg_advance): Likewise.
24009 (ft32_pass_by_reference): Likewise.
24010 (ft32_arg_partial_bytes): Likewise.
24011 (ft32_valid_pointer_mode): Likewise.
24012 (ft32_addr_space_pointer_mode): Likewise.
24013 (ft32_addr_space_legitimate_address_p): Likewise.
24014 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24016 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24017 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24018 (iamcu_alignment): Likewise.
24019 (canonicalize_vector_int_perm): Likewise.
24020 (ix86_noce_conversion_profitable_p): Likewise.
24021 (ix86_mpx_bound_mode): Likewise.
24022 (ix86_operands_ok_for_move_multiple): Likewise.
24023 * config/microblaze/microblaze-protos.h
24024 (microblaze_expand_conditional_branch_reg): Likewise.
24025 * config/microblaze/microblaze.c
24026 (microblaze_expand_conditional_branch_reg): Likewise.
24027 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24029 (rs6000_reassociation_width): Likewise.
24030 (rs6000_invalid_binary_op): Likewise.
24031 (fusion_p9_p): Likewise.
24032 (emit_fusion_p9_load): Likewise.
24033 (emit_fusion_p9_store): Likewise.
24034 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24036 (riscv_hard_regno_mode_ok_p): Likewise.
24037 (riscv_address_insns): Likewise.
24038 (riscv_split_symbol): Likewise.
24039 (riscv_legitimize_move): Likewise.
24040 (riscv_function_value): Likewise.
24041 (riscv_hard_regno_nregs): Likewise.
24042 (riscv_expand_builtin): Likewise.
24043 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24044 (riscv_build_integer): Likewise.
24045 (riscv_split_integer): Likewise.
24046 (riscv_legitimate_constant_p): Likewise.
24047 (riscv_cannot_force_const_mem): Likewise.
24048 (riscv_regno_mode_ok_for_base_p): Likewise.
24049 (riscv_valid_base_register_p): Likewise.
24050 (riscv_valid_offset_p): Likewise.
24051 (riscv_valid_lo_sum_p): Likewise.
24052 (riscv_classify_address): Likewise.
24053 (riscv_legitimate_address_p): Likewise.
24054 (riscv_address_insns): Likewise.
24055 (riscv_load_store_insns): Likewise.
24056 (riscv_force_binary): Likewise.
24057 (riscv_split_symbol): Likewise.
24058 (riscv_force_address): Likewise.
24059 (riscv_legitimize_address): Likewise.
24060 (riscv_move_integer): Likewise.
24061 (riscv_legitimize_const_move): Likewise.
24062 (riscv_legitimize_move): Likewise.
24063 (riscv_address_cost): Likewise.
24064 (riscv_subword): Likewise.
24065 (riscv_output_move): Likewise.
24066 (riscv_canonicalize_int_order_test): Likewise.
24067 (riscv_emit_int_order_test): Likewise.
24068 (riscv_function_arg_boundary): Likewise.
24069 (riscv_pass_mode_in_fpr_p): Likewise.
24070 (riscv_pass_fpr_single): Likewise.
24071 (riscv_pass_fpr_pair): Likewise.
24072 (riscv_get_arg_info): Likewise.
24073 (riscv_function_arg): Likewise.
24074 (riscv_function_arg_advance): Likewise.
24075 (riscv_arg_partial_bytes): Likewise.
24076 (riscv_function_value): Likewise.
24077 (riscv_pass_by_reference): Likewise.
24078 (riscv_setup_incoming_varargs): Likewise.
24079 (riscv_print_operand): Likewise.
24080 (riscv_elf_select_rtx_section): Likewise.
24081 (riscv_save_restore_reg): Likewise.
24082 (riscv_for_each_saved_reg): Likewise.
24083 (riscv_register_move_cost): Likewise.
24084 (riscv_hard_regno_mode_ok_p): Likewise.
24085 (riscv_hard_regno_nregs): Likewise.
24086 (riscv_class_max_nregs): Likewise.
24087 (riscv_memory_move_cost): Likewise.
24088 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24089 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24090 (rl78_addr_space_address_mode): Likewise.
24091 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24093 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24094 (rs6000_reassociation_width): Likewise.
24095 (rs6000_invalid_binary_op): Likewise.
24096 (fusion_p9_p): Likewise.
24097 (emit_fusion_p9_load): Likewise.
24098 (emit_fusion_p9_store): Likewise.
24099 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24100 (ok_for_simple_move_operands): Likewise.
24101 (ok_for_simple_move_strict_operands): Likewise.
24102 (ok_for_simple_arith_logic_operands): Likewise.
24103 (visium_legitimize_reload_address): Likewise.
24104 (visium_select_cc_mode): Likewise.
24105 (output_cbranch): Likewise.
24106 (visium_split_double_move): Likewise.
24107 (visium_expand_copysign): Likewise.
24108 (visium_expand_int_cstore): Likewise.
24109 (visium_expand_fp_cstore): Likewise.
24110 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24111 (visium_function_arg): Likewise.
24112 (visium_function_arg_advance): Likewise.
24113 (visium_libcall_value): Likewise.
24114 (visium_setup_incoming_varargs): Likewise.
24115 (visium_legitimate_constant_p): Likewise.
24116 (visium_legitimate_address_p): Likewise.
24117 (visium_legitimize_address): Likewise.
24118 (visium_secondary_reload): Likewise.
24119 (visium_register_move_cost): Likewise.
24120 (visium_memory_move_cost): Likewise.
24121 (prepare_move_operands): Likewise.
24122 (ok_for_simple_move_operands): Likewise.
24123 (ok_for_simple_move_strict_operands): Likewise.
24124 (ok_for_simple_arith_logic_operands): Likewise.
24125 (visium_function_value_1): Likewise.
24126 (rtx_ok_for_offset_p): Likewise.
24127 (visium_legitimize_reload_address): Likewise.
24128 (visium_split_double_move): Likewise.
24129 (visium_expand_copysign): Likewise.
24130 (visium_expand_int_cstore): Likewise.
24131 (visium_expand_fp_cstore): Likewise.
24132 (visium_split_cstore): Likewise.
24133 (visium_select_cc_mode): Likewise.
24134 (visium_split_cbranch): Likewise.
24135 (output_cbranch): Likewise.
24136 (visium_print_operand_address): Likewise.
24137 * expmed.c (flip_storage_order): Likewise.
24138 * expmed.h (emit_cstore): Likewise.
24139 (flip_storage_order): Likewise.
24140 * genrecog.c (validate_pattern): Likewise.
24141 * hsa-gen.c (gen_hsa_addr): Likewise.
24142 * internal-fn.c (expand_arith_overflow): Likewise.
24143 * ira-color.c (allocno_copy_cost_saving): Likewise.
24144 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24145 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24146 (process_invariant_for_inheritance): Likewise.
24147 * lra-eliminations.c (move_plus_up): Likewise.
24148 * omp-low.c (lower_oacc_reductions): Likewise.
24149 * simplify-rtx.c (simplify_subreg): Likewise.
24150 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24151 (TARGET_CHKP_BOUND_MODE): Likewise..
24152 * targhooks.c (default_chkp_bound_mode): Likewise.
24153 (default_setup_incoming_vararg_bounds): Likewise.
24154 * targhooks.h (default_chkp_bound_mode): Likewise.
24155 (default_setup_incoming_vararg_bounds): Likewise.
24156 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24157 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24158 (have_whole_vector_shift): Likewise.
24159 * tree-vect-stmts.c (vectorizable_load): Likewise.
24160 * doc/tm.texi: Regenerate.
24162 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24164 Graceful degrade if Binutils PR21472 is not available.
24167 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24168 .rodata in flash test fails.
24169 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24170 * confgure: Regenerate.
24171 * config.in: Regenerate.
24172 * config/avr/avr.c (avr_asm_named_section)
24173 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24174 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24175 (avr_asm_init_sections): Same.
24177 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24179 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24180 (fma<VH:mode>4_intrinsic): Likewise.
24181 (*fmsub<VCVTF:mode>4): Likewise.
24182 (*fmsub<VH:mode>4_intrinsic): Likewise.
24184 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24187 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24188 Don't depend on "optimize > 0".
24189 (out_movhi_r_mr, out_movqi_mr_r): Same.
24190 (out_movhi_mr_r, out_movqi_r_mr): Same.
24191 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24192 io_address_operand on "optimize > 0".
24194 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24196 * tree-loop-distribution.c: Add general explanantion on the pass.
24197 (generate_loops_for_partition): Mark distributed loop.
24198 (pg_add_dependence_edges): New parameter. Handle alias data
24199 dependence specially and record it in the parameter if asked.
24200 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24201 (init_partition_graph_vertices, add_partition_graph_edge): New.
24202 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24203 (free_partition_graph_vdata, build_partition_graph): New.
24204 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24205 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24206 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24207 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24208 (version_for_distribution_p, finalize_partitions): New.
24209 (distribute_loop): Handle alias data dependence specially. Factor
24210 out loop fusion code as functions and call these functions.
24212 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24214 * tree-loop-distribution.c (classify_partition): New parameter and
24215 better handle reduction statement.
24216 (rdg_build_partitions): Revise comment.
24217 (distribute_loop): Compute statements in all partitions and pass it
24218 to classify_partition.
24220 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24222 * tree-loop-distribution.c (enum partition_type): New.
24223 (struct partition): New field type.
24224 (partition_merge_into): Add parameter. Update partition type.
24225 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24226 (build_rdg_partition_for_vertex): Compute partition type.
24227 (rdg_build_partitions): Dump partition type.
24228 (distribute_loop): Update calls to partition_merge_into.
24230 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24232 * tree-loop-distribution.c (struct ddr_hasher): New.
24233 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24235 (classify_partition): Call get_data_dependence.
24236 (pg_add_dependence_edges): Ditto.
24237 (distribute_loop): Release data dependence hash table.
24239 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24241 * tree-loop-distribution.c (ref_base_address): Delete.
24242 (similar_memory_accesses): Rename ...
24243 (share_memory_accesses): ... to this. Check if partitions access
24244 the same memory reference.
24245 (distribute_loop): Call share_memory_accesses.
24247 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24249 * tree-loop-distribution.c (struct partition): New field recording
24250 its data reference.
24251 (partition_alloc, partition_free): Init and release data refs.
24252 (partition_merge_into): Merge data refs.
24253 (build_rdg_partition_for_vertex): Collect data refs for partition.
24254 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24256 (distribute_loop): Remve data refs from vertice data of partition
24259 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24261 * tree-loop-distribution.c (params.h): Include header file.
24262 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24263 (datarefs_vec): New global var.
24264 (create_rdg_vertices): Use datarefs_vec directly.
24265 (free_rdg): Don't free data references.
24266 (build_rdg): Update use. Don't free data references.
24267 (distribute_loop): Compute global variable for data references.
24268 Bail out if there are too many data references.
24270 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24272 * tree-loop-distribution.c (loop_nest): New global var.
24273 (build_rdg): Use loop directly, rather than loop nest.
24274 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24276 (distribute_loop): Compute global variable loop nest. Update use.
24278 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24280 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24281 (partition_merge_into): New parameter. Dump reason for fusion.
24282 (distribute_loop): Update use of partition_merge_into.
24284 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24286 * tree-loop-distribution.c (bb_top_order_index): New.
24287 (bb_top_order_index_size, bb_top_order_cmp): New.
24288 (stmts_from_loop): Use topological order.
24289 (pass_loop_distribution::execute): Compute and release topological
24290 order for basic blocks.
24292 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24294 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24297 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24299 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24300 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24301 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24302 * internal-fn.def (LOOP_DIST_ALIAS): New.
24303 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24304 (fold_loop_internal_call): ... this.
24305 (vect_loop_dist_alias_call): New function.
24306 (set_uid_loop_bbs): Call fold_loop_internal_call.
24307 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24310 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24313 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24314 Require dead FLAGS_REG at the beginning of a peephole.
24316 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24319 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24320 arguments in the call to __builtin_ia32_sbb_u32.
24321 (_subborrow_u64): Swap _X and _Y arguments in the call to
24322 __builtin_ia32_sbb_u64.
24324 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24327 * tree-vrp.c (compare_assert_loc): Turn into a function template
24328 with stable template parameter. Only test if a->e is NULL,
24329 !a->e == !b->e has been verified already. Use e == NULL or
24330 e != NULL instead of e or ! e tests. If stable is true, don't use
24331 iterative_hash_expr, on the other side allow a or b or both NULL
24332 and sort the NULLs last.
24333 (process_assert_insertions): Sort using compare_assert_loc<false>
24334 instead of compare_assert_loc, later sort using
24335 compare_assert_loc<true> before calling process_assert_insertions_for
24336 in a loop. Use break instead of continue once seen NULL pointer.
24338 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24340 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24341 Cortex-R7 and Cortex-R8 processors.
24343 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24345 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24346 uninitialized while src is not.
24348 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24350 * common/config/arm/arm-common.c: Adjust include path for
24352 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24353 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24354 (arm-cpu-data.h): Likewise.
24355 (arm-cpu-cdata.h): Likewise.
24356 * config/arm/arm-cpu.h: Delete.
24357 * config/arm/arm-cpu-cdata.h: Delete.
24358 * config/arm/arm-cpu-data.h: Delete.
24360 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24362 * config/arm/arm-cpus.in (cortex-a55): New.
24363 (cortex-a75): Likewise.
24364 (cortex-a75.cortex-a55): Likewise.
24365 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24367 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24368 * config/arm/arm-cpu-cdata.h: Regenerate.
24369 * config/arm/arm-cpu-data.h: Regenerate.
24370 * config/arm/arm-cpu.h: Regenerate.
24371 * config/arm/arm-tables.opt: Regenerate.
24372 * config/arm/arm-tune.md: Regenerate.
24374 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24376 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24378 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24380 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24383 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24385 PR tree-optimization/81292
24386 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24387 full_string_p, also call adjust_related_strinfos if the adjustment
24388 is simple, otherwise invalidate related strinfos.
24390 2017-07-04 Martin Liska <mliska@suse.cz>
24393 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24394 newly created variable as DECL_IGNORED_P.
24396 2017-07-04 Martin Liska <mliska@suse.cz>
24399 * ipa-inline.c (inline_small_functions):
24400 Use xstrdup_for_dump.
24402 2017-07-04 Tom de Vries <tom@codesourcery.com>
24404 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24406 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24409 * config/darwin.c (darwin_function_switched_text_sections):
24412 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24414 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24416 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24418 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24420 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24422 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24423 min_profitable_iters, and th as inclusive lower bounds.
24424 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24425 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24426 for min_profitable_iters and min_profitable_estimate.
24427 (vect_transform_loop): Treat th as an inclusive lower bound.
24428 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24430 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24433 * config/darwin.c (darwin_function_switched_text_sections):
24434 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24435 in two pieces, and suppress the use of buf.
24437 2017-07-03 Nathan Sidwell <nathan@acm.org>
24439 * hash-table.h (hash_table_mod1): Fix indentation.
24441 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24443 PR middle-end/81290
24444 * predict.c (force_edge_cold): Be more careful about propagation
24446 * profile-count.h (profile_probability::guessed,
24447 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24449 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24451 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24453 * doc/invoke.texi (rcpc architecture extension): Document it.
24455 2017-07-03 Richard Biener <rguenther@suse.de>
24457 PR tree-optimization/60510
24458 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24459 the scalar reduction PHI and use it.
24460 (vectorizable_reduction): Properly guard the single_defuse_cycle
24461 path for non-SLP reduction chains where we cannot use it.
24462 Rework reduc_def/index and vector type deduction. Rework
24463 vector operand gathering during reduction op code-gen.
24464 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24465 chains dissolve the chain and leave it to non-SLP reduction
24468 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24470 * tree-data-ref.h (dr_alignment): Declare.
24471 * tree-data-ref.c (dr_alignment): New function.
24472 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24473 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24475 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24477 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24479 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24480 and base_misalignment fields.
24481 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24482 * tree-data-ref.c: Include builtins.h.
24483 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24484 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24485 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24486 * tree-vect-data-refs.c: Include tree-cfg.h.
24487 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24488 fields instead of calculating an alignment here.
24489 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24490 innermost_loop_behavior fields.
24492 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24494 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24496 (DR_STEP_ALIGNMENT): New macro.
24497 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24498 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24499 (create_data_ref): Print it.
24500 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24501 to tell whether the step preserves vector (mis)alignment.
24502 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24503 Move the check for an integer step and generalise to all INTEGER_CST.
24504 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24505 Print the outer step alignment.
24507 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24509 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24510 with offset_alignment.
24511 (DR_ALIGNED_TO): Delete.
24512 (DR_OFFSET_ALIGNMENT): New macro.
24513 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24514 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24515 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24516 (create_data_ref): Likewise.
24517 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24518 (vect_analyze_data_refs): Likewise.
24519 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24520 creating dummy innermost behavior.
24522 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24524 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24525 with a "innermost_loop_behavior *" and refeence tree.
24526 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24527 (create_data_ref): Update call accordingly.
24528 * tree-predcom.c (find_looparound_phi): Likewise.
24530 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24532 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24533 fields with dr_wrt_vec_loop.
24534 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24535 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24536 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24537 (vect_dr_behavior): New function.
24538 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24539 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24540 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24541 track whether the step preserves the misalignment.
24542 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24543 Use vect_dr_behavior.
24544 (vect_setup_realignment): Update call accordingly.
24545 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24546 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24547 call to vect_create_addr_base_for_vector_ref.
24548 (vect_create_cond_for_align_checks): Likewise.
24549 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24550 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24551 (vect_recog_mask_conversion_pattern): Likewise.
24552 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24553 (new_stmt_vec_info): Remove redundant zeroing.
24555 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24557 * common/config/arm/arm-common.c (arm_be8_option): New function.
24558 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24559 (ISA_ARMv6): Add isa_bit_be8.
24560 * config/arm/arm.h (arm_be8_option): Add prototype.
24561 (BE8_SPEC_FUNCTION): New define.
24562 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24563 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24564 (mlittle-endian): Similarly.
24565 (mbe8, mbe32): New options.
24566 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24567 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24569 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24571 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24573 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24575 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24576 (cleanup_tree_cfg_bb): Use it.
24577 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24579 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24581 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24584 * loop-doloop.c (add_test): Update profile.
24586 2017-07-03 Martin Liska <mliska@suse.cz>
24589 * sanopt.c (rewrite_usage_of_param): New function.
24590 (sanitize_rewrite_addressable_params): Likewise.
24591 (pass_sanopt::execute): Call rewrite_usage_of_param.
24593 2017-07-03 Richard Biener <rguenther@suse.de>
24595 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24596 back to using VIEW_CONVERT_EXPR.
24598 2017-07-03 Martin Liska <mliska@suse.cz>
24601 * doc/extend.texi: Document when a resolver function is
24602 generated for target_clones.
24604 2017-07-03 Martin Liska <mliska@suse.cz>
24606 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24607 and shadow memory of auto variables which are subject of
24608 use-after-scope sanitization.
24609 (asan_expand_mark_ifn): Add do set only when is_poison.
24611 2016-07-03 Richard Biener <rguenther@suse.de>
24613 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24615 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24616 (vectorizable_reduction): Perform reduction PHI creation when
24617 visiting a reduction PHI and adjust and simplify code generation
24618 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24619 (vect_transform_loop): Visit reduction PHIs.
24620 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24621 defs into the SLP tree.
24622 (vect_build_slp_tree): Reduction defs terminate the recursion.
24623 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24625 (vect_get_vec_defs_for_stmt_copy): Export.
24626 (vect_get_vec_defs): Likewise.
24627 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24629 (vect_get_vec_defs_for_stmt_copy): Declare.
24630 (vect_get_vec_defs): Likewise.
24632 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24634 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24635 parameter with a "loop" parameter and use it instead of the
24636 loop containing DR_STMT. Don't check simple_iv when doing
24637 BB analysis. Describe the two analysis modes in the comment.
24639 2017-07-03 Tom de Vries <tom@codesourcery.com>
24641 PR tree-optimization/69468
24642 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24643 (find_same_succ_bb): Handle ignore_edge_flags.
24645 2017-07-03 Tom de Vries <tom@codesourcery.com>
24647 PR tree-optimization/81192
24648 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24650 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24652 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24654 2017-07-03 Tom de Vries <tom@codesourcery.com>
24656 PR tree-optimization/81192
24657 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24658 BB_SAME_SUCC (bb) == NULL.
24660 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24662 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24665 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24667 * dumpfile.c: Include profile-count.h
24668 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24670 (insert_cond_bb): Update profile.
24671 * tree-cfg.h (insert_cond_bb): Update prototype.
24672 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24673 * tree-dump.c: Do not include tree-cfg.
24675 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24677 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24679 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24681 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24684 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24686 * tree-complex.c (expand_complex_div_wide): update profile.
24688 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24689 Alan Hayward <alan.hayward@arm.com>
24690 David Sherwood <david.sherwood@arm.com>
24692 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24693 (CORETYPES_H): New define.
24694 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24695 (insn-modes-inline.h, s-modes-inline-h): New rules.
24696 (generated_files): Add insn-modes-inline.h.
24697 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24698 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24699 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24700 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24701 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24702 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24703 (build/gencodes.o, build/genconditions.o): Likewise.
24704 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24705 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24706 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24707 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24708 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24709 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24710 * coretypes.h: Include everything up to real.h for generators.
24711 Include insn-modes.h first. Include wide-int-print.h after
24712 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24713 * machmode.h: Don't include insn-modes.h here.
24714 * function-tests.c: Remove includes of signop.h, machmode.h,
24715 double-int.h and wide-int.h.
24717 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24719 * optc-save-gen.awk: Likewise.
24720 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24721 * godump.c: Remove include of wide-int-print.h.
24722 * pretty-print.h: Likewise.
24723 * wide-int-print.cc: Likewise.
24724 * wide-int.cc: Likewise.
24725 * hash-map-tests.c: Remove include of signop.h.
24726 * hash-set-tests.c: Likewise.
24727 * rtl-tests.c: Likewise.
24728 * mkconfig.sh: Remove include of machmode.h.
24729 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24731 (emit_insn_modes_inline_h): ...this new function. Emit the code
24732 into an insn-modes-inline.h header file, adding appropriate
24733 include guards and end comments.
24734 (emit_insn_modes_c_header): Remove include of machmode.h.
24735 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24737 (main): Handle -i flag and call emit_insn_modes_inline_h when
24740 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24742 * tree-ssa-strlen.c (strinfo): Rename the length field to
24743 nonzero_chars. Add a full_string_p field.
24744 (compare_nonzero_chars, zero_length_string_p): New functions.
24745 (get_addr_stridx): Add an offset_out parameter.
24746 Use compare_nonzero_chars.
24747 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24748 (new_strinfo): Update after above changes to strinfo.
24749 (set_endptr_and_length): Set full_string_p.
24750 (get_string_length): Update after above changes to strinfo.
24751 (unshare_strinfo): Update call to new_strinfo.
24752 (maybe_invalidate): Likewise.
24753 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24754 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24755 as a uhwi instead of an shwi. Update after above changes to
24756 strinfo and new_strinfo.
24757 (zero_length_string): Assert that chainsi contains full strings.
24758 Use zero_length_string_p. Update call to new_strinfo.
24759 (adjust_related_strinfos): Update after above changes to strinfo.
24760 Copy full_string_p from origsi.
24761 (adjust_last_stmt): Use zero_length_string_p.
24762 (handle_builtin_strlen): Update after above changes to strinfo and
24763 new_strinfo. Install the lhs as the string length if the previous
24764 entry didn't describe a full string.
24765 (handle_builtin_strchr): Update after above changes to strinfo
24767 (handle_builtin_strcpy): Likewise.
24768 (handle_builtin_strcat): Likewise.
24769 (handle_builtin_malloc): Likewise.
24770 (handle_pointer_plus): Likewise.
24771 (handle_builtin_memcpy): Likewise. Track nonzero characters
24772 that aren't necessarily followed by a nul terminator.
24773 (handle_char_store): Likewise.
24775 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24777 PR tree-optimization/80769
24778 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24779 for malloc and calloc. Document the new invariant that all related
24780 strinfos have delayed lengths or none do.
24781 (verify_related_strinfos): Move earlier in file.
24782 (set_endptr_and_length): New function, split out from...
24783 (get_string_length): ...here. Also set the lengths of related
24785 (zero_length_string): Assert that chainsi has known (rather than
24787 (adjust_related_strinfos): Likewise.
24789 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24791 PR tree-optimization/81136
24792 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24793 assert that two references with the same misalignment have the same
24794 compile-time misalignment if those compile-time misalignments
24797 2017-07-01 Andi Kleen <ak@linux.intel.com>
24799 * print-tree.c (print_node): Print all attributes.
24801 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24803 * cfg.c (scale_bbs_frequencies): New function.
24804 * cfg.h (scale_bbs_frequencies): Declare it.
24805 * cfgloopanal.c (single_likely_exit): Cleanup.
24806 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24808 (scale_loop_profile): Likewise.
24809 (loop_version): Likewise.
24810 (create_empty_loop_on_edge): Update.
24811 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24812 scale_loop_frequencies, scale_loop_profile, loopify,
24813 loop_version): Update prototypes.
24814 * modulo-sched.c (sms_schedule): Update.
24815 * predict.c (unlikely_executed_edge_p): Also check probability.
24816 (probably_never_executed_edge_p): Fix typo.
24817 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24818 * tree-parloops.c (gen_parallel_loop): Update.
24819 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24820 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24821 * tree-ssa-loop-split.c (split_loop): Update.
24822 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24823 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24824 (vect_loop_versioning): Update.
24825 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24827 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24829 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24831 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24833 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24834 to keep profile consistent.
24836 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24838 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24839 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24840 * profile-count.h (max_safe_multiplier): Make unsigned.
24841 (profile_count::guessed_zero): New.
24843 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24845 * bb-reorder.c (fix_up_crossing_landing_pad,
24846 fix_crossing_conditional_branches): Use make_single_succ_edge
24847 to keep profile consistent.
24849 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24851 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24854 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24857 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24858 the right scopes, make sure cond_jump isn't preserved between multiple
24859 iterations. Search for fallthru edge whenever there are 3+ edges and
24860 use find_fallthru_edge for it.
24862 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24864 Patch by Alexander Monakov <amonakov@ispras.ru>
24865 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
24866 probabilities consistently.
24868 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24870 * pa.c (pa_expand_compare_and_swap_loop): Update call of
24871 emit_cmp_and_jump_insns.
24873 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24876 * tree-inline.c (expand_call_inline): Combine profile statuses.
24878 2017-06-30 Andrew Pinski <apinski@cavium.com>
24880 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
24881 fold_stmt returned true.
24883 2017-06-30 Nathan Sidwell <nathan@acm.org>
24885 * ggc.h (empty_string): Delete.
24886 * cfgexpand.c (expand_asm_stmt): Use plain "".
24887 * optabs.c (expand_asm_memory_barrier): Likewise.
24888 * stringpool.c (empty_string): Delete.
24889 (digit_vector, digit_string): Delete.
24890 (ggc_alloc_string): Use plain "", don't optimize single digit
24891 strings. Use ggc_alloc_atomic.
24893 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
24895 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
24896 comparison set and one other set, use the cost of the non-comparison
24899 2017-06-30 Nathan Sidwell <nathan@acm.org>
24901 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
24904 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24906 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
24907 loops. Remove now unneeded calls to gimple_switch_set_label() that
24908 just set removed labels to NULL_TREE.
24910 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
24912 * tree-ssanames.c (set_range_info_raw): Abstract from ...
24913 (set_range_info): ...here. Only call set_range_info_raw if domain
24915 (set_nonzero_bits): Call set_range_info_raw.
24916 * tree-ssanames.h (set_range_info_raw): New.
24918 2017-06-30 Jakub Jelinek <jakub@redhat.com>
24921 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
24922 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
24923 of nonimmediate_operand and <store_mask_constraint> instead of m
24924 for the input operand. For V8FI iterator, always split if input
24925 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
24926 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
24927 <store_mask_predicate> instead of register_operand and
24928 <store_mask_constraint> instead of v for the input operand. Make
24929 sure both operands aren't MEMs for if not <mask_applied>.
24931 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
24933 * lto-wrapper.c (copy_file) Close both file descriptors before
24936 2017-06-30 Martin Liska <mliska@suse.cz>
24939 * multiple_target.c (create_dispatcher_calls): Make ifunc
24940 also for function that don't have calls or are not referenced.
24942 2017-06-30 Richard Biener <rguenther@suse.de>
24944 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
24945 analyze the first scalar stmt. Move vector type computation
24946 for the BB case here from ...
24947 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
24948 live operation processing in the SLP case properly.
24950 2017-06-30 Richard Biener <rguenther@suse.de>
24952 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
24954 2017-06-30 Martin Liska <mliska@suse.cz>
24957 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
24958 before BUILT_IN_UNWIND_RESUME when ASAN is used.
24960 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
24962 * doc/invoke.texi (AArch64): Add missing options and remove redundant
24965 2017-06-30 Richard Biener <rguenther@suse.de>
24967 PR tree-optimization/81249
24968 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
24969 condition reduction result to original scalar type.
24971 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24973 * profile-count.h (enum profile_quality): Fix typos and whitespace
24976 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24978 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
24979 type for branch probabilities.
24981 2017-06-29 Julian Brown <julian@codesourcery.com>
24982 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24984 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
24985 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
24986 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
24987 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
24989 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24991 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
24992 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
24993 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
24994 CC usage from generic code to here.
24995 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
24996 CC usage into the target macros.
24998 2017-06-29 Maya Rashish <coypu@sdf.org>
25000 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25003 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25005 * arm/arm-builtins.c: Include profile-count.h
25006 * except.c (sjlj_emit_function_enter): Use
25007 profile_probability::unlikely.
25009 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25011 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25012 and tocrel_offset be pointer args rather than implicitly using
25014 (legitimate_constant_pool_address_p, rs6000_emit_move,
25015 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25016 tocrel_offset and use in toc_relative_expr_p call.
25017 (print_operand, print_operand_address): Use static tocrel_base_oac
25018 and tocrel_offset_oac.
25019 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25022 2017-06-29 Maya Rashish <coypu@sdf.org>
25024 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25026 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25028 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25029 objects, take into account only the alignment of 'op0' and 'mode1' if
25032 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25034 * ccmp.c (ccmp_tree_comparison_p): New function.
25035 (ccmp_candidate_p): Update to use above function.
25036 (get_compare_parts): New function.
25037 (expand_ccmp_next): Update to use new functions.
25038 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25040 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25041 take mode as argument.
25042 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25043 * expr.c (expand_expr_real_1): Pass mode as argument.
25045 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25047 * combine.c (combine_instructions): Print insns to dump_file, together
25050 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25052 * asan.c (asan_emit_stack_protection): Update.
25053 (create_cond_insert_point): Update.
25054 * auto-profile.c (afdo_propagate_circuit): Update.
25055 * basic-block.h (struct edge_def): Turn probability to
25056 profile_probability.
25057 (EDGE_FREQUENCY): Update.
25058 * bb-reorder.c (find_traces_1_round): Update.
25059 (better_edge_p): Update.
25060 (sanitize_hot_paths): Update.
25061 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25062 (make_single_succ_edge): Update.
25063 (check_bb_profile): Update.
25064 (dump_edge_info): Update.
25065 (update_bb_profile_for_threading): Update.
25066 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25067 probabilitycount to 0.
25068 * cfgbuild.c (compute_outgoing_frequencies): Update.
25069 * cfgcleanup.c (try_forward_edges): Update.
25070 (outgoing_edges_match): Update.
25071 (try_crossjump_to_edge): Update.
25072 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25073 (expand_gimple_tailcall): Update.
25074 (construct_init_block): Use make_single_succ_edge.
25075 (construct_exit_block): Use make_single_succ_edge.
25076 * cfghooks.c (verify_flow_info): Update.
25077 (redirect_edge_succ_nodup): Update.
25078 (split_edge): Update.
25079 (account_profile_record): Update.
25080 * cfgloopanal.c (single_likely_exit): Update.
25081 * cfgloopmanip.c (scale_loop_profile): Update.
25082 (set_zero_probability): Remove.
25083 (duplicate_loop_to_header_edge): Update.
25084 * cfgloopmanip.h (loop_version): Update prototype.
25085 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25086 (force_nonfallthru_and_redirect): Update.
25087 (update_br_prob_note): Update.
25088 (rtl_verify_edges): Update.
25089 (purge_dead_edges): Update.
25090 (rtl_lv_add_condition_to_bb): Update.
25091 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25092 * cgraphunit.c (init_lowered_empty_function): Update.
25093 (cgraph_node::expand_thunk): Update.
25094 * cilk-common.c: Include profile-count.h
25095 * dojump.c (inv): Remove.
25096 (jumpifnot): Update.
25097 (jumpifnot_1): Update.
25098 (do_jump_1): Update.
25100 (do_jump_by_parts_greater_rtx): Update.
25101 (do_compare_rtx_and_jump): Update.
25102 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25103 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25104 * dwarf2cfi.c: Include profile-count.h
25105 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25106 (sjlj_emit_dispatch_table): Likewise.
25107 * explow.c: Include profile-count.h
25108 * expmed.c (emit_store_flag_force): Update.
25109 (do_cmp_and_jump): Update.
25110 * expr.c (compare_by_pieces_d::generate): Update.
25111 (compare_by_pieces_d::finish_mode): Update.
25112 (emit_block_move_via_loop): Update.
25113 (store_expr_with_bounds): Update.
25114 (store_constructor): Update.
25115 (expand_expr_real_2): Update.
25116 (expand_expr_real_1): Update.
25117 * expr.h (try_casesi, try_tablejump): Update prototypes.
25118 * gimple-pretty-print.c (dump_probability): Update.
25119 (dump_profile): New.
25120 (dump_gimple_label): Update.
25121 (dump_gimple_bb_header): Update.
25122 * graph.c (draw_cfg_node_succ_edges): Update.
25123 * hsa-gen.c (convert_switch_statements): Update.
25124 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25125 (find_if_case_1): Update.
25126 (find_if_case_2): Update.
25127 * internal-fn.c (expand_arith_overflow_result_store): Update.
25128 (expand_addsub_overflow): Update.
25129 (expand_neg_overflow): Update.
25130 (expand_mul_overflow): Update.
25131 (expand_vector_ubsan_overflow): Update.
25132 * ipa-cp.c (good_cloning_opportunity_p): Update.
25133 * ipa-split.c (split_function): Use make_single_succ_edge.
25134 * ipa-utils.c (ipa_merge_profiles): Update.
25135 * loop-doloop.c (add_test): Update.
25136 (doloop_modify): Update.
25137 * loop-unroll.c (compare_and_jump_seq): Update.
25138 (unroll_loop_runtime_iterations): Update.
25139 * lra-constraints.c (lra_inheritance): Update.
25140 * lto-streamer-in.c (input_cfg): Update.
25141 * lto-streamer-out.c (output_cfg): Update.
25142 * mcf.c (adjust_cfg_counts): Update.
25143 * modulo-sched.c (sms_schedule): Update.
25144 * omp-expand.c (expand_omp_for_init_counts): Update.
25145 (extract_omp_for_update_vars): Update.
25146 (expand_omp_ordered_sink): Update.
25147 (expand_omp_for_ordered_loops): Update.
25148 (expand_omp_for_generic): Update.
25149 (expand_omp_for_static_nochunk): Update.
25150 (expand_omp_for_static_chunk): Update.
25151 (expand_cilk_for): Update.
25152 (expand_omp_simd): Update.
25153 (expand_omp_taskloop_for_outer): Update.
25154 (expand_omp_taskloop_for_inner): Update.
25155 * omp-simd-clone.c (simd_clone_adjust): Update.
25156 * optabs.c (expand_doubleword_shift): Update.
25157 (expand_abs): Update.
25158 (emit_cmp_and_jump_insn_1): Update.
25159 (expand_compare_and_swap_loop): Update.
25160 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25161 * predict.c (predictable_edge_p): Update.
25162 (edge_probability_reliable_p): Update.
25163 (set_even_probabilities): Update.
25164 (combine_predictions_for_insn): Update.
25165 (combine_predictions_for_bb): Update.
25166 (propagate_freq): Update.
25167 (estimate_bb_frequencies): Update.
25168 (force_edge_cold): Update.
25169 * profile-count.c (profile_count::dump): Add missing space into dump.
25170 (profile_count::debug): Add newline.
25171 (profile_count::differs_from_p): Explicitly convert to unsigned.
25172 (profile_count::stream_in): Update.
25173 (profile_probability::dump): New member function.
25174 (profile_probability::debug): New member function.
25175 (profile_probability::differs_from_p): New member function.
25176 (profile_probability::differs_lot_from_p): New member function.
25177 (profile_probability::stream_in): New member function.
25178 (profile_probability::stream_out): New member function.
25179 * profile-count.h (profile_count_quality): Rename to ...
25180 (profile_quality): ... this one.
25181 (profile_probability): New.
25182 (profile_count): Update.
25183 * profile.c (compute_branch_probabilities): Update.
25184 * recog.c (peep2_attempt): Update.
25185 * sched-ebb.c (schedule_ebbs): Update.
25186 * sched-rgn.c (find_single_block_region): Update.
25187 (compute_dom_prob_ps): Update.
25188 (schedule_region): Update.
25189 * sel-sched-ir.c (compute_succs_info): Update.
25190 * stmt.c (struct case_node): Update.
25191 (do_jump_if_equal): Update.
25192 (get_outgoing_edge_probs): Update.
25193 (conditional_probability): Update.
25194 (emit_case_dispatch_table): Update.
25195 (expand_case): Update.
25196 (expand_sjlj_dispatch_table): Update.
25197 (emit_case_nodes): Update.
25198 * targhooks.c: Update.
25199 * tracer.c (better_p): Update.
25200 (find_best_successor): Update.
25201 * trans-mem.c (expand_transaction): Update.
25202 * tree-call-cdce.c: Update.
25203 * tree-cfg.c (gimple_split_edge): Upate.
25204 (move_sese_region_to_fn): Upate.
25205 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25206 * tree-eh.c (lower_resx): Upate.
25207 (cleanup_empty_eh_move_lp): Upate.
25208 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25209 * tree-inline.c (copy_edges_for_bb): Update.
25210 (copy_cfg_body): Update.
25211 * tree-parloops.c (gen_parallel_loop): Update.
25212 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25213 (gimple_gen_time_profiler): Update.
25214 * tree-ssa-dce.c (remove_dead_stmt): Update.
25215 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25216 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25217 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25218 (unloop_loops): Update.
25219 (try_peel_loop): Update.
25220 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25221 * tree-ssa-loop-split.c (connect_loops): Update.
25222 (split_loop): Update.
25223 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25224 (hoist_guard): Update.
25225 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25226 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25227 (value_replacement): Update.
25228 * tree-ssa-reassoc.c (branch_fixup): Update.
25229 * tree-ssa-tail-merge.c (replace_block_by): Update.
25230 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25231 (create_edge_and_update_destination_phis): Update.
25232 (compute_path_counts): Update.
25233 (recompute_probabilities): Update.
25234 (update_joiner_offpath_counts): Update.
25235 (freqs_to_counts_path): Update.
25236 (duplicate_thread_path): Update.
25237 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25238 (struct switch_conv_info): Update.
25239 (gen_inbound_check): Update.
25240 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25241 (vect_do_peeling): Update.
25242 (vect_loop_versioning): Update.
25243 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25244 (optimize_mask_stores): Update.
25245 * ubsan.c (ubsan_expand_null_ifn): Update.
25246 * value-prof.c (gimple_divmod_fixed_value): Update.
25247 (gimple_divmod_fixed_value_transform): Update.
25248 (gimple_mod_pow2): Update.
25249 (gimple_mod_pow2_value_transform): Update.
25250 (gimple_mod_subtract): Update.
25251 (gimple_mod_subtract_transform): Update.
25252 (gimple_ic): Update.
25253 (gimple_stringop_fixed_value): Update.
25254 (gimple_stringops_transform): Update.
25255 * value-prof.h: Update.
25257 2017-06-29 Carl Love <cel@us.ibm.com>
25259 * config/rs6000/rs6000-c.c: Add support for built-in functions
25260 vector signed int vec_signed (vector float);
25261 vector signed long long vec_signed (vector double);
25262 vector signed int vec_signed2 (vector double, vector double);
25263 vector signed int vec_signede (vector double);
25264 vector signed int vec_signedo (vector double);
25265 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25266 instruction generator.
25267 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25268 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25269 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25271 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25272 vunsignede_v2df): Add define_expands.
25273 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25274 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25275 VEC_UNSIGNEDO): Add definitions.
25276 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25277 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25278 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25279 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25280 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25281 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25282 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25283 vec_unsignede, vec_unsignedo): Add builtin defines.
25284 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25286 * doc/extend.texi: Update the built-in documentation file for the
25287 new built-in functions.
25289 2017-06-29 Richard Biener <rguenther@suse.de>
25291 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25292 reduction chains to LOOP_VINFO_REDUCTIONS.
25293 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25294 SLP reductions after processing reduction chains.
25296 2017-06-29 Nathan Sidwell <nathan@acm.org>
25298 * builtins.c (fold_builtin_FUNCTION): Use
25299 lang_hooks.decl_printable_name.
25301 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25303 PR middle-end/81194
25304 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25305 with only one label.
25306 * stmt.c (expand_case): Assert NCASES is greater than one.
25308 2017-06-29 Richard Biener <rguenther@suse.de>
25310 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25312 (group_case_labels): Likewise.
25313 (find_taken_edge): Push sanity checking on val to workers...
25314 (find_taken_edge_cond_expr): ... here
25315 (find_taken_edge_switch_expr): ... and here, handle cases
25316 with just a default label.
25317 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25318 (group_case_labels): Likewise.
25319 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25320 group_case_labels does anything cleanup the CFG again.
25322 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25324 PR tree-optimization/81196
25325 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25326 exit condition comparing two IVs.
25328 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25330 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25331 profile to the dummy entry at the end of the list of architectures.
25332 * config/arm/arm-cpu-cdata.h: Regenerated.
25334 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25335 Michael Collison <michael.collison@arm.com>
25338 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25340 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25341 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25342 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25343 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25344 of shift when the shift amount is masked with constant equal to
25345 the size of the mode.
25346 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25349 2017-06-29 Martin Liska <mliska@suse.cz>
25351 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25353 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25356 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25359 (edge_growth_cache_entry::edge_growth_cache_entry): New
25361 (reset_edge_growth_cache): Update to use constructor.
25363 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25365 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25366 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25367 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25369 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25371 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25372 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25374 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25376 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25377 (*-linux-uclibc*): Add t-uclibc tmake_file.
25378 * config/t-musl: New.
25379 * config/t-uclibc: New.
25381 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25383 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25385 (gen_comm_data): Emit architectural setting of arch_prof.
25386 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25388 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25389 (armv8-m.base, armv8-m.main): Likewise.
25390 * arm-protos.h (arm_build_target): Add profile field.
25391 (arch_option): Likewise.
25392 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25394 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25395 arm_active_target.profile.
25397 2017-06-28 Richard Biener <rguenther@suse.de>
25399 PR middle-end/81227
25400 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25401 TYPE_OVERFLOW_WRAPS.
25402 * match.pd (negate_expr_p): Likewise.
25403 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25404 fold_build2, not fold_binary.
25406 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25408 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25409 Convert memory address to Pmode.
25410 (aarch64_print_operand): Assert MEM operands are always Pmode.
25412 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25415 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25416 Remove redundant if.
25417 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25418 * config/arm/aarch-common-protos.h
25419 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25420 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25422 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25425 * multiple_target.c (create_dispatcher_calls): Set the default
25426 clone to be static, not public.
25428 2017-06-28 Richard Biener <rguenther@suse.de>
25430 * tree-vect-loop.c (vectorizable_reduction): Move special
25431 cond reduction IV var creation ...
25432 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25433 parameter. Use STMT_VINFO_VECTYPE.
25434 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25437 2017-06-28 Martin Liska <mliska@suse.cz>
25440 * ipa-visibility.c (non_local_p): Handle visibility.
25442 2017-06-28 Martin Liska <mliska@suse.cz>
25445 * common.opt: Add IntegerRange to various options.
25446 * opt-functions.awk (integer_range_info): New function.
25447 * optc-gen.awk: Add integer_range_info to cl_options struct.
25448 * opts-common.c (decode_cmdline_option): Handle
25449 CL_ERR_INT_RANGE_ARG.
25450 (cmdline_handle_error): Likewise.
25451 * opts.c (print_filtered_help): Show valid interval in
25452 when --help is provided.
25453 * opts.h (struct cl_option): Add range_min and range_max fields.
25454 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25456 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25458 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25459 (x * C EQ/NE y * C): New transformation.
25461 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25463 * genmultilib (combination_space): Accept '+' in option names.
25465 2017-06-28 Martin Liska <mliska@suse.cz>
25468 * asan.c (instrument_derefs): Bail out inner references
25469 that are hard register variables.
25471 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25474 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25475 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25477 2017-06-28 Richard Biener <rguenther@suse.de>
25479 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25480 (vect_get_slp_defs): Adjust.
25481 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25483 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25485 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25486 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25487 (vectorizable_reduction): Adjust.
25488 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25490 (vect_get_slp_defs): Likewise.
25491 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25492 (vectorizable_bswap): Adjust.
25493 (vectorizable_call): Likewise.
25494 (vectorizable_conversion): Likewise.
25495 (vectorizable_assignment): Likewise.
25496 (vectorizable_shift): Likewise.
25497 (vectorizable_operation): Likewise.
25498 (vectorizable_store): Likewise.
25499 (vectorizable_condition): Likewise.
25500 (vectorizable_comparison): Likewise.
25502 2017-06-28 Michael Collison <michael.collison@arm.com>
25505 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25507 (arm_gen_movmemqi): Removed unused variable 'i'.
25508 Convert 'for' loop into 'while' loop.
25509 (arm_expand_prologue): Remove last unnecessary set of insn.
25510 (thumb_pop): Remove unused variable 'pushed_words'.
25511 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25513 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25515 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25516 * config/s390/s390-protos.h: Add prototype of
25517 s390_rel_address_ok_p.
25518 * config/s390/s390.c (s390_got_symbol): New function.
25519 (s390_rel_address_ok_p): New function.
25520 (legitimize_pic_address): Use s390_rel_address_ok_p.
25521 (s390_load_got): Use s390_got_symbol.
25522 (s390_option_override): Issue error if
25523 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25524 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25526 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25528 2017-06-27 Andrew Pinski <apinski@cavium.com>
25530 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25531 (X * copysign (1.0, X)): New pattern.
25532 (X * copysign (1.0, -X)): New pattern.
25533 (copysign (-1.0, CST)): New pattern.
25535 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25537 * genmultilib (combination_space): Remove variable.
25538 Validate reuse rules against regular expression for any sequence
25539 of multilib options in any order.
25541 2017-06-27 Michael Collison <michael.collison@arm.com>
25543 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25544 call aarch64_split_simd_combine.
25545 * (aarch64_combine_internal<mode>): Delete pattern.
25546 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25547 Allow register and subreg operands.
25549 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25551 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25552 specific need, just fallback on defaults.
25553 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25555 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25556 Olivier Hainque <hainque@adacore.com>
25558 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25560 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25561 targets. Pick a default if no particular attempt applied.
25562 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25565 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25567 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25568 (x86_64-wrs-vxworks7): Likewise.
25570 2017-06-27 Marek Polacek <polacek@redhat.com>
25573 * ubsan.c (instrument_null): Check get_base_address's result for null.
25575 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25577 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25579 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25581 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25582 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25583 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25584 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25585 New function types.
25586 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25587 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25588 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25589 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25590 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25591 BUILT_IN_FEUPDATEENV): New builtins.
25592 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25593 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25594 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25595 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25597 (builtin_structptr_types): Adjust size.
25598 * tree.c (builtin_structptr_types): Add four entries.
25600 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25601 Olivier Hainque <hainque@adacore.com>
25603 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25604 (TLS_SYM): New local macro, forcing reference to __tls__ on
25605 link command lines for VxWorks 7 RTPs, triggering initialization
25607 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25608 OS features TLS support, true for RTPs on VxWorks 7.
25609 * config/vxworks.c (vxworks_override_options): Setup emutls
25612 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25614 * predict.c (test_prediction_value_range): Use -1U instead of -1
25615 to avoid narrowing conversion warning.
25616 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25617 to avoid narrowing conversion warning.
25618 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25620 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25622 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25624 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25625 64bit configurations.
25626 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25627 (SIZE_TYPE): Likewise.
25628 * config/vxworks.c (vxworks_emutls_var_fields): Use
25629 long_unsigned_type_node instead of unsigned_type_node as the offset
25630 field type, which is "pointer" mode in emutls.c.
25632 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25635 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25637 PR middle-end/81207
25638 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25639 gimple_vuse copying separately from gimple_vdef copying.
25641 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25643 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25646 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25647 Olivier Hainque <hainque@adacore.com>
25649 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25650 for all vxworks7 targets.
25651 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25652 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25653 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25654 variations for VX6/VX7 and 32/64bits later on in ...
25655 (VXWORKS_LIB_SPEC): Leverage new macros.
25656 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25657 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25659 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25661 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25662 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25664 2017-06-26 Carl Love <cel@us.ibm.com>
25666 * config/rs6000/rs6000-c.c: Add support for built-in functions
25667 vector bool char vec_reve (vector bool char);
25668 vector signed char vec_reve (vector signed char);
25669 vector unsigned char vec_reve (vector unsigned char);
25670 vector bool int vec_reve (vector bool int);
25671 vector signed int vec_reve (vector signed int);
25672 vector unsigned int vec_reve (vector unsigned int);
25673 vector bool long long vec_reve (vector bool long long);
25674 vector signed long long vec_reve (vector signed long long);
25675 vector unsigned long long vec_reve (vector unsigned long long);
25676 vector bool short vec_reve (vector bool short);
25677 vector signed short vec_reve (vector signed short);
25678 vector double vec_reve (vector double);
25679 vector float vec_reve (vector float);
25680 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25681 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25682 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25683 (altivec_vreve): New pattern.
25684 * config/rs6000/altivec.h (vec_reve): New define.
25685 * doc/extend.texi (vec_rev): Update the built-in documentation file
25686 for the new built-in functions.
25688 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25690 PR tree-optimization/71815
25691 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25693 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25695 (slsr_process_phi): Likewise.
25696 (replace_uncond_cands_and_profitable_phis): Don't replace a
25697 multiply candidate with a stride of 1 (copy or cast).
25698 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25700 (lowest_cost_path): Likewise.
25701 (total_savings): Likewise.
25703 2017-06-26 Richard Biener <rguenther@suse.de>
25706 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25707 Use def_builtin_pure for all gather builtins.
25709 2017-06-26 Richard Biener <rguenther@suse.de>
25711 PR tree-optimization/81203
25712 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25713 non-dominating BBs.
25715 2017-06-26 Marek Polacek <polacek@redhat.com>
25718 * doc/invoke.texi: Document -Wmultistatement-macros.
25720 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25722 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25723 arm_neon_ok_no_float_abi effective target.
25725 2017-06-26 Richard Biener <rguenther@suse.de>
25727 PR tree-optimization/80928
25728 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25729 (copy_bbs): Set BB_DUPLICATED flag early.
25730 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25732 (execute_on_shrinking_pred): Likewise.
25733 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25734 BB_DUPLICATED blocks.
25735 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25736 iterate over all PHIs considering removal of *gsi.
25738 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25740 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25743 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25745 * config/rs6000/rs6000-string.c: (expand_block_clear,
25746 do_load_for_compare, select_block_compare_mode,
25747 compute_current_alignment, expand_block_compare,
25748 expand_strncmp_align_check, expand_strn_compare,
25749 expand_block_move, rs6000_output_load_multiple)
25750 Move functions related to string/block move/compare
25751 to a separate file.
25752 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25753 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25754 for this function which is now used in two files.
25755 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25756 * config.gcc: Add rs6000-string.o to extra_objs for
25757 targets powerpc*-*-* and rs6000*-*-*.
25759 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25762 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25763 32-bit, since indexed is not valid for DImode.
25764 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25765 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25766 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25767 (define_peephole2 for Altivec d-form store): Likewise.
25770 * multiple_target.c (create_dispatcher_calls): Only create the
25771 dispatcher call if the function is the default clone of a
25772 versioned function.
25774 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25776 PR middle-end/80902
25777 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25778 a call, force the call to not be a tail call.
25780 2017-06-23 Jeff Law <law@redhat.com>
25782 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25785 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25787 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25788 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25789 handling for early expansion of vector shifts (sl,sr,sra,rl).
25790 (builtin_function_type): Add vector shift right instructions
25791 to the unsigned argument list.
25793 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25795 rtl-optimizatoin/79286
25796 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25797 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25798 trap. PIC register plus a const unspec without offset can never trap.
25800 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25802 * tree.h (builtin_structptr_type): New type.
25803 (builtin_structptr_types): Declare new array.
25804 * tree.c (builtin_structptr_types): New array.
25805 (free_lang_data, build_common_tree_nodes): Use it.
25807 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25810 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25813 2017-06-22 Matt Turner <mattst88@gmail.com>
25815 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25816 Lake models to skylake case. Assume skylake for unknown
25817 models with clflushopt.
25819 2017-06-22 Jeff Law <law@redhat.com>
25821 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25822 frame sizes that do not satisfy aarch64_uimm12_shift.
25824 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25826 * profile-count.h (apply_probability,
25827 apply_scale, probability_in): Fix checks for zero.
25829 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25831 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25832 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25834 2016-06-22 Richard Biener <rguenther@suse.de>
25836 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25837 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25838 REDUC_MAX_EXPR support.
25839 (vectorizable_reduction): Likewise.
25840 (vect_create_epilog_for_reduction): Likewise.
25842 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25844 * match.pd (A / (1 << B) -> A >> B): New.
25845 * generic-match-head.c: Include optabs-tree.h.
25846 * gimple-match-head.c: Likewise.
25847 * optabs-tree.h (target_supports_op_p): New.
25848 * optabs-tree.c (target_supports_op_p): New.
25850 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25852 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25853 $gcc_cv_ld --help output.
25854 (gcc_cv_ld_demangle): Likewise.
25855 (gcc_cv_ld_eh_frame_hdr): Likewise.
25856 (gcc_cv_ld_pie): Likewise.
25857 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25858 (gcc_cv_ld_buildid): Likewise.
25859 (gcc_cv_ld_sysroot): Likewise.
25860 (ld_bndplt_support): Likewise.
25861 (ld_pushpopstate_support): Likewise.
25862 * configure: Regenerate.
25863 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
25865 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25868 * config/i386/sse.md (round<mode>2): Renumber match_dup and
25869 operands indexes to avoid gap between operands and match_dups.
25871 2017-06-21 Andrew Pinski <apinski@cavium.com>
25873 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
25874 Increment Arith_shift and Arith_shift_reg by 1.
25875 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
25877 * config/aarch64/aarch64.c (thunderx_tunings): Enable
25878 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
25879 (aarch64_strip_extend): Add new argument and test for it.
25880 (aarch64_cheap_mult_shift_p): New function.
25881 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
25882 add a cost if it is true.
25883 Update calls to aarch64_strip_extend.
25884 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
25886 2017-06-21 Andrew Pinski <apinski@cavium.com>
25888 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
25890 (thunderxt88): Likewise.
25891 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
25892 (thunderx_prefetch_tune): New variable.
25893 (thunderx2t99_prefetch_tune): Update for the correct values.
25894 (thunderxt88_tunings): New variable.
25895 (thunderx_tunings): Use thunderx_prefetch_tune instead of
25896 generic_prefetch_tune.
25897 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
25899 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25901 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
25902 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
25903 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
25904 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
25905 (aarch64_atomic_cas<mode>, GPI): Likewise.
25907 2017-06-21 Martin Liska <mliska@suse.cz>
25909 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
25910 statements on cold and hot labels.
25911 * predict.c (tree_estimate_probability_bb): Remove the
25912 prediction from this place.
25914 2017-06-21 Martin Liska <mliska@suse.cz>
25916 PR tree-optimization/79489
25917 * gimplify.c (maybe_add_early_return_predict_stmt): New
25919 (gimplify_return_expr): Call the function.
25920 * predict.c (tree_estimate_probability_bb): Remove handling
25922 * predict.def: Update comment about early return predictor.
25923 * gimple-predict.h (is_gimple_predict): New function.
25924 * predict.def: Change default value of early return to 66.
25925 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
25927 * passes.def: Put pass_strip_predict_hints to the beginning of
25930 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
25932 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
25933 FUNCTION_DECL declarations.
25934 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
25936 (dwaf2out_decl): Likewise.
25937 * godump.c (go_early_global_decl): Skip call to the real debug hook
25938 for FUNCTION_DECL declarations.
25939 * passes.c (rest_of_decl_compilation): Skip call to the
25940 early_global_decl debug hook for FUNCTION_DECL declarations, unless
25941 -fdump-go-spec is passed.
25943 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25945 * config/i386/i386.c (struct builtin_isa): New field pure_p.
25946 Reorder for compactness.
25947 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
25948 (def_builtin_pure, def_builtin_pure2): New functions.
25949 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
25951 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25953 * match.pd (nop_convert): New predicate.
25954 ((A +- CST1) +- CST2): Allow some NOP conversions.
25956 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25959 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
25960 with ctors/dtors if GOVD_SHARED is set.
25962 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25964 * config/aarch64/aarch64.md (movti_aarch64):
25965 Emit mov rather than orr.
25966 (movtf_aarch64): Likewise.
25967 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
25968 Emit mov rather than orr.
25970 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25972 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
25973 Swap alternatives, make integer dup more expensive.
25975 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25977 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25978 Return true for non-tls symbols.
25980 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
25982 * config/aarch64/aarch64-cores.def (cortex-a55): New.
25983 (cortex-a75): Likewise.
25984 (cortex-a75.cortex-a55): Likewise.
25985 * config/aarch64/aarch64-tune.md: Regenerate.
25986 * doc/invoke.texi (-mtune): Document new values for -mtune.
25988 2017-06-21 Tom de Vries <tom@codesourcery.com>
25990 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
25991 stack_size feature.
25992 (Effective-Target Keywords, Other attributes): Suggest using
25993 dg-add-options stack_size feature to get stack limit in stack_size
25994 effective target documentation.
25996 2017-06-21 Julian Brown <julian@codesourcery.com>
25997 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25999 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26000 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26001 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26003 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26004 attribute type list for neon_multiply.
26005 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26006 attribute type list for neon_multiply.
26007 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26008 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26009 attribute type list for neon_multiply.
26010 * config/arm/types.md (crypto_pmull): Add.
26011 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26012 attribute type list.
26014 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26016 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26019 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26021 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26022 to make sure not to dereference a NULL cost_classes_ptr pointer.
26024 2017-06-20 Carl Love <cel@us.ibm.com>
26026 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26027 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26028 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26029 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26030 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26031 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26032 VMULOSW): New enum "unspec" values.
26033 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26034 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26035 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26036 altivec_vmulosw): New patterns.
26037 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26038 VMULOSW): Add definitions.
26040 2017-06-20 Julia Koval <julia.koval@intel.com>
26042 * config/i386/i386.c: Fix rounding expand for new pattern.
26043 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26045 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26047 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26048 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26050 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26052 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26055 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26057 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26058 architecture, then by implementer ID.
26059 * config/aarch64/aarch64-tune.md: Regenerate.
26061 2017-06-20 Richard Biener <rguenther@suse.de>
26063 PR middle-end/81097
26064 * fold-const.c (split_tree): Fold to type before negating.
26066 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26068 * diagnostic-show-locus.c
26069 (selftest::test_fixit_deletion_affecting_newline): New function.
26070 (selftest::diagnostic_show_locus_c_tests): Call it.
26072 2017-06-20 Andreas Schwab <schwab@suse.de>
26075 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26078 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26080 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26081 __ARM_FEATURE_COPROC according to support.
26083 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26085 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26086 Rewritten to avoid overflow for > 32-bit pointers.
26089 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26090 by removing enum keyword.
26091 (ubsan_type_descriptor): Likewise. Formatting fix.
26094 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26095 splitter): Require TARGET_SSE2 in the condition.
26097 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26100 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26101 for doing vector set of SFmode on ISA 3.0.
26102 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26103 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26105 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26106 SFmode value into a V4SF variable that was extracted from another
26107 V4SF variable without converting the element to double precision
26108 and back to single precision vector format.
26109 (vsx_insert_extract_v4sf_p9_2): Likewise.
26111 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26113 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26114 in UWHI to avoid undefined overflow.
26117 * ubsan.h (enum ubsan_encode_value_phase): New.
26118 (ubsan_encode_value): Change second argument to
26119 enum ubsan_encode_value_phase with default value of
26120 UBSAN_ENCODE_VALUE_GENERIC.
26121 * ubsan.c (ubsan_encode_value): Change second argument to
26122 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26123 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26124 create_tmp_var_raw instead of create_tmp_var and use a
26126 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26127 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26128 ubsan_encode_value callers.
26131 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26132 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26133 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26135 2017-06-19 Richard Biener <rguenther@suse.de>
26137 PR middle-end/81118
26138 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26139 estimates if we changed anything.
26141 2017-06-19 Richard Biener <rguenther@suse.de>
26143 PR tree-optimization/80887
26144 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26145 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26146 simplified lookups, then reset mprts_hook.
26147 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26149 (try_to_simplify): Likewise.
26151 2017-06-19 Martin Liska <mliska@suse.cz>
26154 * gimplify.c (gimplify_switch_expr):
26155 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26157 2017-06-19 Martin Liska <mliska@suse.cz>
26159 * doc/install.texi: Document that PGO runs in 4 stages.
26161 2017-06-19 Martin Liska <mliska@suse.cz>
26164 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26165 to dispatcher function name.
26166 * multiple_target.c (replace_function_decl): New function.
26167 (create_dispatcher_calls): Redirect both edges and references.
26169 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26171 * profile-count.c (profile_count::dump): Dump quality.
26172 (profile_count::differs_from_p): Update for unsigned val.
26173 * profile-count.h (profile_count_quality): New enum.
26174 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26176 2017-06-19 Richard Biener <rguenther@suse.de>
26178 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26179 struct function as arg.
26180 (estimate_numbers_of_iterations): Export overload with loop arg.
26181 (free_numbers_of_iterations_estimates_loop): Use an overload of
26182 free_numbers_of_iterations_estimates instead.
26183 * tree-cfg.c (remove_bb): Adjust.
26184 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26185 * tree-parloops.c (gen_parallel_loop): Likewise.
26186 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26188 (tree_unroll_loops_completely): Likewise.
26189 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26190 Use an overload instead and export.
26191 (estimated_loop_iterations): Adjust.
26192 (max_loop_iterations): Likewise.
26193 (likely_max_loop_iterations): Likewise.
26194 (estimate_numbers_of_iterations): Take struct function as arg
26196 (loop_exits_before_overflow): Adjust.
26197 (free_numbers_of_iterations_estimates_loop): Use an overload.
26198 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26199 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26201 2017-06-19 Richard Biener <rguenther@suse.de>
26204 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26205 RANGE_EXPR conservatively.
26207 2017-06-16 Carl Love <cel@us.ibm.com>
26209 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26210 definitions for vec_float, vec_float2, vec_floato,
26211 vec_floate built-ins.
26212 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26213 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26215 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26216 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26217 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26218 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26219 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26220 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26221 vec_floato): Add builtin defines.
26222 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26223 Update the built-in documentation file for the new built-in
26226 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26228 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26229 (mthumb): Mark as the negative of -marm.
26231 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26233 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26235 (ARM Options, -mtune): Document that this accepts the same
26236 extension options as -mcpu.
26237 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26241 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26242 permitted extensions.
26244 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26246 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26247 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26248 (armv8-m.main): Add option +nodsp.
26249 * config/arm/arm-cpu-cdata.h: Regenerated.
26251 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26253 * config/arm/t-fuchsia: New file.
26254 * config.gcc (arm*-*-fuchsia*): Use it.
26256 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26258 * config/arm/t-symbian: Rewrite for new option infrastructure.
26260 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26262 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26263 (MULTILIB_REQUIRED): Likewise.
26265 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26267 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26268 (MULTILIB_RESUE): Likewise.
26269 (MULTILIB_MATCHES): Likewise.
26270 (MULTLIB_REQUIRED): Likewise.
26272 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26274 * config/arm/t-rtems: Rewrite for new option framework.
26276 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26278 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26279 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26280 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26281 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26282 * config/arm/t-multilib: ... here.
26283 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26284 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26285 armv7-a and armv8*-a when A-profile libraries have not been built.
26286 * config/arm/t-rmprofile: Rewrite.
26288 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26290 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26291 with a backslash. Remove the backslash after substituting unescaped
26293 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26295 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26297 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26298 the driver to pass through the default setting of -mfloat-abi.
26299 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26301 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26302 (all_feat_combs): New rule.
26303 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26305 * config/arm/t-aprofile: Rewrite.
26307 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26309 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26310 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26311 fpu is not specified by the user/command-line.
26312 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26313 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26314 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26315 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26316 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26317 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26319 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26321 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26322 * config/arm/t-arm-elf: Rewritten.
26324 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26326 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26327 have some floating-point instructions.
26328 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26329 (TARGET_MAYBE_HARD_FLOAT): New macro.
26330 * config/arm/arm-builtins.c (arm_init_builtins): Use
26331 TARGET_MAYBE_HARD_FLOAT.
26332 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26336 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26337 (configargs.h): Include it.
26338 (arm_print_hint_for_fpu_option): New function.
26339 (arm_parse_fpu_option): New function.
26340 (candidate_extension): New class.
26341 (arm_canon_for_multilib): New function.
26342 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26343 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26344 (ARCH_CANONICAL_SPECS): New macro.
26345 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26347 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26349 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26350 are set after handling multilib fragments. Set target_cpu_default2
26353 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26355 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26357 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26358 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26360 (check_arch): Likewise.
26361 * config/arm/arm.c (arm_configure_build_target): Handle
26362 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26363 options in the default.
26365 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26367 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26368 when an option is an alias of another.
26369 * config/arm/parsecpu.awk (optalias): New parser token.
26370 (gen_comm_data): Mark non-alias options as such. Emit entries
26371 for extension aliases.
26372 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26373 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26374 (armv6kz, armv6zk, armv6t2): Likewise.
26375 (armv7): Make vfpv3-d16 an alias.
26376 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26378 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26379 Sort in canonical order.
26380 (armv8-a): Sort in canonical order.
26381 (armv8.1-a, armv8.2-a): Likewise.
26382 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26384 (cortex-a9): Sort in canonical order.
26385 * config/arm/arm.c (selftests.h): Include it.
26386 (arm_test_cpu_arch_data): New function.
26387 (arm_run_self_tests): New function.
26388 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26389 (targetm): Move declaration to the end of the file.
26390 * arm-cpu-cdata.h: Regenerated.
26392 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26394 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26395 call to target_mode_check describing the type of option passed.
26396 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26397 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26398 arm_parse_cpu_option_name to match parameters against list of
26400 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26401 arm_arch_core_flags data structure.
26402 * config/arm/arm-cpu_cdata.h: Regenerated.
26404 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26406 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26408 (arm_print_hint_for_cpu_option): Likewise.
26409 (arm_print_hint_for_arch_option): Likewise.
26410 (arm_parse_cpu_option_name): Likewise.
26411 (arm_parse_arch_option_name): Likewise.
26412 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26413 of entries in the all_fpus list.
26414 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26415 (arm_parse_cpu_option_name): Declare.
26416 (arm_parse_arch_option_name): Declare.
26417 (arm_parse_option_features): Declare.
26418 (arm_intialize_isa): Declare.
26419 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26421 (gen_comm_data): ... here. Make definitions non-static.
26422 * config/arm/arm-cpu-data.h: Regenerated.
26423 * config/arm/arm-cpu-cdata.h: Regenerated.
26425 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26427 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26428 (cpu_arch_extension): New structure.
26429 (cpu_arch_option, arch_option, cpu_option): New structures.
26430 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26431 architecture types.
26432 (gen_data): Generate new format data tables.
26433 * config/arm/arm.c (cpu_tune): New structure.
26434 (cpu_option, processors): Delete.
26435 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26436 (arm_print_hint_for_cpu_option): ... this and ...
26437 (arm_print_hint_for_arch_option): ... this.
26438 (arm_parse_arch_cpu_name): Delete. Replace with ...
26439 (arm_parse_cpu_option_name): ... this and ...
26440 (arm_parse_arch_option_name): ... this.
26441 (arm_unrecognized_feature): Change type of target parameter to
26443 (arm_parse_arch_cpu_features): Delete. Replace with ...
26444 (arm_parse_option_features): ... this.
26445 (arm_configure_build_target): Rework to use new configuration data
26447 (arm_print_tune_info): Rework for new configuration data tables.
26448 * config/arm/arm-cpu-data.h: Regenerated.
26449 * config/arm/arm-cpu.h: Regenerated.
26451 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26453 * Makefile.in (OBJS): Move sbitmap.o from here ...
26454 (OBJS-libcommon): ... to here.
26456 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26458 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26459 (ISA_ALL_CRYPTO): New macro.
26460 (ISA_ALL_SIMD): New macro
26461 (ISA_ALL_FP): New macro.
26462 * config/arm/arm.c (fpu_bitlist): Update initializer.
26463 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26465 (arm9e): Add fpu. Add option for nofp
26466 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26467 (arm926ej-s, arm1026ej-s): Likewise.
26468 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26469 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26470 neon-fp16, neon-vfpv4, nofp and nosimd.
26471 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26472 (cortex-a8): Add fpu. Add option for nofp.
26473 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26474 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26475 (cortex-r4f): Add fpu.
26476 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26477 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26479 (cortex-r8): Likewise.
26480 (cortex-m4): Add fpu. Add option for nofp.
26481 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26482 (cortex-a17.cortex-a7): Likewise.
26483 (cortex-a32): Add fpu. Add options for crypto and nofp.
26484 (cortex-a35, cortex-a53): Likewise.
26485 (cortex-a57): Add fpu. Add option for crypto.
26486 (cortex-a72, cortex-a73): Likewise.
26487 (exynos-m1): Likewise.
26488 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26489 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26490 (cortex-m33): Add fpu. Add option for nofp.
26491 * config/arm/arm-cpu-cdata.h: Regenerated
26492 * config/arm/arm-cpu-data.h: Regenerated.
26494 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26496 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26497 (armv5te, armv5tej): Likewise.
26498 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26499 (armv7): Add options fp and vfpv3-d16.
26500 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26501 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26503 (armv7ve): Likewise.
26504 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26505 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26506 (armv8-a): Add nocrypto option.
26507 (armv8.1-a, armv8.2-a): Likewise.
26508 (armv8-m.main): add options fp, fp.dp and nofp.
26510 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26512 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26514 (armv8-a+crc): Delete.
26515 (armv8.1-a): Add options simd, crypto and nofp.
26516 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26517 (armv8.2-a+fp16): Delete.
26518 (armv8-m.main): Add option dsp.
26519 (armv8-m.main+dsp): Delete.
26520 (cortex-a8): Add fpu. Add nofp option.
26521 (cortex-a9): Add fpu. Add nofp and nosimd options.
26522 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26523 link to main cpu and architecture data structures.
26524 (gen_comm_data): Only put isa attributes from the main architecture
26526 (option): New statement for architecture and CPU entries.
26527 * arm.c (struct cpu_option): New structure.
26528 (struct processors): Add entry for options.
26529 (arm_unrecognized_feature): New function.
26530 (arm_parse_arch_cpu_name): Ignore any characters after the first
26532 (arm_parse_arch_cpu_feature): New function.
26533 (arm_configure_build_target): Separate out any CPU and architecture
26534 features and parse separately. Don't error out if -mfpu=auto is
26535 used with only an architecture string.
26536 (arm_print_asm_arch_directives): New function.
26537 (arm_file_start): Call it.
26538 * config/arm/arm-cpu-cdata.h: Regenerated.
26539 * config/arm/arm-cpu-data.h: Likewise.
26540 * config/arm/arm-tables.opt: Likewise.
26542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26544 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26545 assembler when it is not -mfpu=auto.
26547 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26549 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26550 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26551 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26552 (ASM_CPU_SPEC): Rewrite.
26553 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26554 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26555 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26556 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26557 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26558 copied string is NUL-terminated. Also strip any characters prefixed
26560 (arm_rewrite_selected_arch): New function.
26561 (arm_rewrite_march): New function.
26563 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26565 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26566 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26567 (march, mcpu, mtune): Convert to string-based options.
26568 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26569 (arm_parse_arch_cpu_name): New function.
26570 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26571 identify selected architecture or CPU.
26572 (arm_option_save): New function.
26573 (TARGET_OPTION_SAVE): Redefine.
26574 (arm_option_restore): Restore string options.
26575 (arm_option_print): Print string options.
26577 2017-06-16 Martin Sebor <msebor@redhat.com>
26579 PR tree-optimization/80933
26580 PR tree-optimization/80934
26581 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26582 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26583 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26584 (gimple_fold_builtin): Call them.
26586 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26588 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26589 as unlikely; update profile.
26591 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26593 * predict.c (force_edge_cold): Handle declaring edges impossible
26596 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26598 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26600 (try_unroll_loop_completely): Fix reporting.
26602 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26604 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26606 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26609 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26610 if given a non-constant argument for an intrinsic which requires a
26613 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26615 * profile.c (compare_freqs): New function.
26616 (branch_prob): Sort edge list.
26617 (find_spanning_tree): Assume that the list is priority sorted.
26619 2017-06-16 Richard Biener <rguenther@suse.de>
26621 PR tree-optimization/81090
26622 * passes.def (pass_record_bounds): Remove.
26623 * tree-pass.h (make_pass_record_bounds): Likewise.
26624 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26625 make_pass_record_bounds): Likewise.
26626 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26627 not free niter estimates at the beginning but at the end.
26628 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26630 2017-06-16 Richard Biener <rguenther@suse.de>
26632 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26633 initializer to workaround ICE in host GCC 4.8.
26635 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26637 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26639 (clone_inlined_nodes): Update.
26641 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26643 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26644 prefetch settings, and enable prefetching by default at -O3.
26646 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26648 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26649 Set flag_prefetch_loop_arrays according to tuning data.
26651 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26653 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26654 New tune structure.
26655 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26656 [Unrelated to main purpose of the patch] Place the pointer field last
26657 to enable type checking errors when tune structure are wrongly merged.
26658 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26659 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26660 (thunderx2t99_prefetch_tune): New tune constants.
26661 (tune_params *_tunings): Update all tunings (no functional change).
26662 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26663 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26664 from tunings structures.
26666 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26669 * ubsan.c (instrument_null): Add T argument, use it instead
26670 of computing it based on IS_LHS.
26671 (instrument_object_size): Likewise.
26672 (pass_ubsan::execute): Adjust instrument_null and
26673 instrument_object_size callers to pass gimple_get_lhs or
26674 gimple_assign_rhs1 result to it. Use instrument_null instead of
26675 calling get_base_address and instrument_mem_ref. Handle
26676 aggregate call arguments for object-size sanitization.
26678 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26680 PR tree-optimization/81089
26681 * tree-vrp.c (is_masked_range_test): Validate operands of
26684 2017-06-15 Martin Sebor <msebor@redhat.com>
26687 * dumpfile.c (dump_register): Avoid calling memset to initialize
26688 a class with a default ctor.
26689 * gcc.c (struct compiler): Remove const qualification.
26690 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26691 * hash-table.h: Ditto.
26692 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26694 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26695 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26697 * params.h (struct param_info): Make struct members non-const.
26698 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26699 with default initialization.
26700 * vec.h (vec_copy_construct, vec_default_construct): New helper
26702 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26703 with vec_copy_construct.
26704 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26705 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26706 * doc/invoke.texi (-Wclass-memaccess): Document.
26708 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26710 * emit-rtl.h (is_leaf): Update comment about local
26711 register allocator.
26713 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26716 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26717 for a variable to have a section before checking if the section has a
26719 Set section to.persistent if persistent attribute is set.
26720 Warn if .persistent attribute is used on an automatic variable.
26722 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26724 PR rtl-optimization/80474
26725 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26727 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26729 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26732 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26734 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26735 (casesi): Emit bounds checking as RTL.
26736 (casesi_internal_mips16_<mode>): Remove bounds checking.
26738 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26740 * config/xtensa/xtensa.c (xtensa_option_override): Append
26741 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26742 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26743 xtensa_doloop_hooks): Define unconditionally.
26744 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26746 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26747 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26748 for it in xtensa_option_override.
26749 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26750 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26752 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26754 * doc/cppopts.texi: Document '-' special value to -MF.
26756 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26758 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26759 (cortex_a53_fconst): Likewise.
26760 (cortex_a53_fpmul): Likewise.
26761 (cortex_a53_f_load_64): Likewise.
26762 (cortex_a53_f_load_many): Likewise.
26763 (cortex_a53_advsimd_alu): Likewise.
26764 (cortex_a53_advsimd_alu_q): Likewise.
26765 (cortex_a53_advsimd_mul): Likewise.
26766 (cortex_a53_advsimd_mul_q): Likewise.
26767 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26768 Add missing fmul, r2f_cvt and fconst cases.
26770 2017-06-14 Richard Biener <rguenther@suse.de>
26772 PR middle-end/81088
26773 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26775 (fold_binary_loc): When associating do not treat pre-existing
26776 TREE_OVERFLOW on literal constants as a reason to allow
26777 TREE_OVERFLOW on associated literal constants.
26779 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26781 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26782 (MASK_FEATURES): New macro.
26783 * config/sparc/sparc.c (sparc_option_override): Remove the special
26784 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26786 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26788 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26789 a division of 0 if non-call exceptions are enabled.
26791 2017-06-14 Andrew Pinski <apinski@cavium.com>
26792 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26795 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26796 Improve vector initialization code gen for only variable case.
26798 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26800 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26802 2017-06-14 Richard Biener <rguenther@suse.de>
26804 PR tree-optimization/81083
26805 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26808 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26810 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26811 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26812 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26813 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26814 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26815 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26817 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26819 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26820 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26822 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26824 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26826 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26828 * config/rs6000/t-rtems: Don't handle SPE.
26830 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26832 * config/rs6000/t-linux: Don't handle SPE.
26834 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26836 * config/rs6000/eabispe.h: Delete file.
26838 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26840 * config/rs6000/t-spe: Delete file.
26842 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26844 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26845 (rs6000_legitimate_offset_address_p): Return false for anything in
26846 V2SImode or V2SFmode.
26848 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26850 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26851 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26852 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26854 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26855 (rs6000_legitimate_offset_address_p): Ditto.
26856 (rs6000_emit_move): Ditto.
26857 (rs6000_init_builtins): Remove V4HI_type_node.
26859 2017-06-13 Martin Liska <mliska@suse.cz>
26862 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
26863 (gate_asan): Likewise.
26864 * asan.h (asan_no_sanitize_address_p): Remove the function.
26865 (sanitize_flags_p): New function.
26866 * builtins.def: Fix coding style.
26867 * common.opt: Use renamed enum value.
26868 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
26869 * doc/extend.texi: Document no_sanitize attribute.
26870 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
26871 to SANITIZE_UNDEFINED_NONDEFAULT.
26872 * gcc.c (sanitize_spec_function): Use the renamed enum value.
26873 * gimple-fold.c (optimize_atomic_compare_exchange_p):
26874 Use sanitize_flags_p.
26875 * gimplify.c (gimplify_function_tree): Likewise.
26876 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
26877 * opts.c (parse_no_sanitize_attribute): New function.
26878 (common_handle_option): Use renamed enum value.
26879 * opts.h (parse_no_sanitize_attribute): Declare.
26880 * tree.c (sanitize_flags_p): New function.
26881 * tree.h: Declared here.
26882 * tsan.c: Use sanitize_flags_p.
26883 * ubsan.c (ubsan_expand_null_ifn): Likewise.
26884 (instrument_mem_ref): Likewise.
26885 (instrument_bool_enum_load): Likewise.
26886 (do_ubsan_in_current_function): Remove the function.
26887 (pass_ubsan::execute): Use sanitize_flags_p.
26888 * ubsan.h: Remove do_ubsan_in_current_function
26889 * tree-cfg.c (print_no_sanitize_attr_value): New function.
26890 (dump_function_to_file): Use it here.
26892 2017-06-13 Martin Jambor <mjambor@suse.cz>
26894 PR tree-optimization/80803
26895 PR tree-optimization/81063
26896 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
26897 (propagate_subaccesses_across_link): Enqueue subtree whenever
26898 necessary instead of relying on the caller.
26900 2017-06-13 Martin Jambor <mjambor@suse.cz>
26902 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
26903 that have a first_link.
26904 (sort_and_splice_var_accesses): Do not check first_link before
26906 (subtree_mark_written_and_enqueue): Likewise.
26907 (propagate_all_subaccesses): Likewise and do not stop at first
26908 parent with a first_link.
26910 2017-06-13 Martin Jambor <mjambor@suse.cz>
26912 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
26915 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26917 * match.pd: New pattern.
26919 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26921 * tree-vrp.c (is_masked_range_test): New function.
26922 (register_edge_assert_for): Determine ranges for
26925 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26927 PR tree-optimization/67328
26928 * fold-const.c (maskable_range_p): New function.
26929 (build_range_check): Generate bittests if possible.
26931 2017-06-13 Martin Liska <mliska@suse.cz>
26933 * gimple-pretty-print.c (dump_probability): Add new argument.
26934 (dump_edge_probability): Dump both probability and count.
26935 (dump_gimple_label): Likewise.
26936 (dump_gimple_bb_header): Likewise.
26938 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
26941 * config/avr/avr-devices.c: Fix indentation.
26942 * config/avr/gen-avr-mmcu-specs.c: Dito.
26944 2017-06-13 Richard Biener <rguenther@suse.de>
26946 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
26947 instead get vector type from stmt_info.
26948 (vectorizable_reduction): Adjust. Remove dead code.
26950 2017-06-13 Richard Biener <rguenther@suse.de>
26952 PR middle-end/81065
26953 * fold-const.c (extract_muldiv_1): Remove bogus distribution
26954 case of C * (x * C2 + C3).
26955 (fold_addr_of_array_ref_difference): Properly fold index difference.
26957 2017-06-12 David S. Miller <davem@davemloft.net>
26960 * config/sparc/sparc.md (return expander): Emit frame blockage if
26961 function uses alloca.
26963 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
26965 * combine.c (make_field_assignment): Check len rather than the mode
26966 precision when calling force_to_mode.
26968 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
26970 Support multilibs and devices that see flash in RAM address range.
26973 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
26974 (avr_mcu_t) <flash_pm_offset>: New field.
26975 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
26976 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
26977 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
26978 (AVR_TINY_PM_OFFSET): Remove macro.
26979 * config/avr/avr.opt (-mshort-calls): New option.
26980 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
26981 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
26982 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
26983 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
26984 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
26985 instead of avr_arch->have_jmp_call.
26986 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
26987 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
26988 avr_arch->flash_pm_offset to define.
26989 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
26990 new field flash_pm_offset. Add entry for avrxmega3.
26991 (avr_texinfo): Add entry for avrxmega3.
26992 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
26993 attiny212, attiny214,
26994 attiny412, attiny414, attiny416, attiny417,
26995 attiny814, attiny816, attiny817,
26996 attiny1614, attiny1616, attiny1617,
26997 attiny3214, attiny3216, attiny3217.
26998 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
26999 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27000 (avr_print_operand_address) [AVR_TINY]: Same.
27001 (avr_asm_init_sections) <readonly_data_section>: Only patch
27002 callback if avr_arch->flash_pm_offset = 0.
27003 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27004 for rodata if avr_arch->flash_pm_offset != 0.
27005 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27006 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27007 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27008 (m_options): Append opt_rcall.
27009 (m_dirnames): Append dir_rcall.
27010 * config/avr/t-multilib: Regenerate.
27012 * configure.ac [target=avr]: Check whether avrxmega3 default
27013 linker description file works as needed.
27014 * configure: Regenerate.
27015 * doc/avr-mmcu.texi: Regenerate.
27016 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27017 <__AVR_ARCH__>: Document avrxmega3 and 103.
27018 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27019 <__AVR_SHORT_CALLS__>: Document it.
27020 <__AVR_PM_BASE_ADDRESS__>: Document it.
27021 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27022 (AVR Variable Attributes) <progmem>: Document this is
27023 not needed for avrxmega3.
27024 (AVR Named Address Spaces) <__flash>: Dito.
27026 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27028 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27030 2017-06-12 Doug Rupp <rupp@adacore.com>
27032 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27033 Append vxworks-stdint.h to the tm_file list.
27034 * config/vxworks-stdint.h: New file.
27036 2017-06-12 Martin Liska <mliska@suse.cz>
27038 PR tree-optimization/81041
27039 * tree-profile.c (gimple_gen_ic_func_profiler):
27040 Create an extra BB in profile-generate
27041 (gimple_gen_time_profiler): Likewise.
27043 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27045 PR tree-optimization/81003
27046 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27047 (update_range_test): Use it instead of force_gimple_operand_gsi.
27049 2017-06-12 Richard Biener <rguenther@suse.de>
27051 PR tree-optimization/81053
27052 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27053 with backedge value not defined in loop. Simplify def stmt
27056 2017-06-11 Tom de Vries <tom@codesourcery.com>
27059 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27061 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27062 nvptx_cannot_force_const_mem.
27064 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27066 * opts.c (finish_options): Move test for flag_split_stack after
27067 it has been initialized.
27069 2017-06-11 Jason Merrill <jason@redhat.com>
27071 * tree.h (id_equal): New.
27072 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27073 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27074 instead of strcmp of IDENTIFIER_POINTER.
27076 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27078 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27079 (mark_all_inlined_calls_cdtor): Fix formating.
27080 (inline_transform): Rescale profile before inlining.
27082 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27084 * cgraph.h (cgraph_edge::clone): Update prototype.
27085 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27086 (cgraph_node::create_clone): Update.
27087 (cgraph_node::create_version_clone): Update.
27088 * tree-inline.c (copy_bb): Update.
27089 (expand_call_inline): Update.
27091 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27093 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27094 factored out from ...
27095 (rs6000_emit_prologue): ... here.
27097 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27099 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27100 factored out from ...
27101 (rs6000_emit_prologue): ... here.
27103 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27105 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27107 (handle_missing_profiles): Fix computation of tp_first_run.
27108 (counts_to_freqs): Do not touch freqs when count is 0.
27110 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27112 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27115 2017-06-10 Tom de Vries <tom@codesourcery.com>
27117 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27118 attributes): Document signal effective target.
27120 2017-06-10 Tom de Vries <tom@codesourcery.com>
27122 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27123 Document effective target stack_size.
27125 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27127 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27128 to the edit_context if they can be auto-applied.
27130 2017-06-9 Ian Lance Taylor <iant@golang.org>
27132 * opts.c (finish_options): If -fsplit-stack, disable implicit
27133 -forder-blocks-and-partition.
27134 * doc/invoke.texi (Optimize Options): Document that when using
27135 -fsplit-stack -forder-blocks-and-partition is not implicitly
27138 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27140 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27141 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27142 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27143 * builtins.def (abort, trap, unreachable): Declare cold.
27144 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27145 * tree-core.h (ECF_COLD): New.
27146 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27147 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27149 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27151 * predict.c (unlikely_executed_stmt_p): Cleanup.
27153 2017-06-09 Richard Biener <rguenther@suse.de>
27155 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27156 model if the ref is always written to.
27158 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27160 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27162 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27164 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27167 2017-06-09 Tom de Vries <tom@codesourcery.com>
27170 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27171 "target cannot support label values" when encountering LABEL_REF.
27173 2017-06-09 Martin Liska <mliska@suse.cz>
27175 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27176 (gimple_gen_ic_func_profiler): Emit direct comparison
27177 of __gcov_indirect_call_callee with NULL.
27178 (gimple_gen_time_profiler): Change probability from
27179 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27181 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27183 * profile.c (edge_gcov_counts): Turn to pointer.
27184 (compute_branch_probabilities, compute_branch_probabilities): Update.
27185 (branch_prob): Do not clear edge_gcov_count.
27186 * profile.h (edge_gcov_counts): Turn to pointer.
27187 (edge_gcov_count): Update.
27189 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27191 * gimple.h (gimple_check_failed): Mark cold.
27193 2017-06-09 Richard Biener <rguenther@suse.de>
27195 PR tree-optimization/66623
27196 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27197 refactor check_reduction into two parts, properly computing
27198 whether we have to check reduction validity for outer loop
27201 2017-06-09 Richard Biener <rguenther@suse.de>
27203 PR tree-optimization/79483
27204 * graphite-scop-detection.c (order): New global.
27205 (get_order): Compute bb to order mapping that satisfies code
27206 generation constraints.
27207 (cmp_pbbs): New helper.
27208 (build_scops): Start domwalk at entry block, sort generated
27211 2017-06-09 Richard Biener <rguenther@suse.de>
27213 PR middle-end/81007
27214 * ipa-polymorphic-call.c
27215 (ipa_polymorphic_call_context::restrict_to_inner_class):
27216 Skip FIELD_DECLs with error_mark_node type.
27217 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27220 2017-06-09 Martin Liska <mliska@suse.cz>
27222 * predict.c (struct branch_predictor): New struct.
27223 (test_prediction_value_range): New test.
27224 (predict_c_tests): New function.
27225 * selftest-run-tests.c (selftest::run_tests): Run the function.
27226 * selftest.h: Declare new tests.
27228 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27231 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27232 gen_add3_insn did not fail.
27233 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27234 r0, construct that number in a temporary reg and add that reg to r0.
27235 If asked to put the result in r0 as well, fail.
27237 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27239 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27240 for early expansion of vec_eqv.
27242 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27244 PR middle-end/81005
27245 * ubsan.c (instrument_null): Avoid pointless code temporary.
27246 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27248 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27252 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27255 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27258 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27260 * predict.c (unlikely_executed_edge_p): Move ahead.
27261 (probably_never_executed_edge_p): Use it.
27263 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27265 PR middle-end/79988
27266 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27267 gimple_call_builtin_p call.
27269 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27271 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27272 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27273 rtl_check_failed_type2, rtl_check_failed_code1,
27274 rtl_check_failed_code2, rtl_check_failed_code_mode,
27275 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27276 rtvec_check_failed_bounds, rtl_check_failed_flag,
27277 _fatal_insn_not_found, _fatal_insn): Likewise.
27278 * tree.h (tree_contains_struct_check_failed,
27279 tree_check_failed, tree_not_check_failed,
27280 tree_class_check_failed, tree_range_check_failed,
27281 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27282 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27283 tree_operand_check_failed, omp_clause_check_failed,
27284 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27287 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27289 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27290 flag_branch_probabilities.
27291 * ipa-inline.c (edge_badness): Likewise.
27292 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27293 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27294 * predict.c (maybe_hot_frequency_p): Likewise.
27295 (probably_never_executed): Likewise.
27296 * sched-ebb.c (schedule_ebbs): Likewise.
27297 * sched-rgn.c (find_single_block_region): Likewise.
27298 * tracer.c (tail_duplicate): Likewise.
27300 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27302 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27303 longer requires x_flag_profile_use.
27305 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27307 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27308 instead of flag_reorder_blocks_and_partition.
27309 * dbxout.c (dbxout_function_end): Likewise.
27310 * dwarf2out.c (gen_subprogram_die): Likewise.
27311 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27312 * hw-doloop.c (reorg_loops): Likewise.
27313 * varasm.c (assemble_start_function,
27314 assemble_end_function): Likewise.
27315 (decide_function_section): Do not check for
27316 flag_reorder_blocks_and_partition.
27318 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27320 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27322 (chkp_get_hard_register_fake_addr_expr): Ditto.
27323 (chkp_build_addr_expr): Add check for hard reg case.
27324 (chkp_parse_array_and_component_ref): Ditto.
27325 (chkp_find_bounds_1): Ditto.
27326 (chkp_process_stmt): Don't generate bounds store for
27329 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27331 * predict.c (maybe_hot_bb_p): Do not check profile status.
27332 (maybe_hot_edge_p): Likewise.
27333 (probably_never_executed): Check for zero counts even if profile
27335 (unlikely_executed_edge_p): New function.
27336 (unlikely_executed_stmt_p): New function.
27337 (unlikely_executed_bb_p): New function.
27338 (set_even_probabilities): Use unlikely predicates.
27339 (combine_predictions_for_bb): Likewise.
27340 (predict_paths_for_bb): Likewise.
27341 (predict_paths_leading_to_edge): Likewise.
27342 (determine_unlikely_bbs): New function.
27343 (estimate_bb_frequencies): Use it.
27344 (compute_function_frequency): Use zero counts even if profile is
27346 * profile-count.h: Fix typo.
27348 2017-08-08 Julia Koval <julia.koval@intel.com>
27350 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27351 _mm512_mask_cvtsepi16_storeu_epi8,
27352 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27353 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27354 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27355 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27356 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27357 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27358 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27359 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27360 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27361 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27362 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27363 __builtin_ia32_pmovuswb256mem_mask,
27364 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27365 __builtin_ia32_pmovwb512mem_mask): New builtins.
27367 2017-08-08 Julia Koval <julia.koval@intel.com>
27369 PR target/73350,80862
27370 * config/i386/subst.md (round): Fix round pattern.
27371 * config/i386/i386.c (ix86_erase_embedded_rounding):
27372 Fix erasing rounding for the fixed pattern.
27374 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27376 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27378 2017-06-08 Martin Liska <mliska@suse.cz>
27380 PR gcov-profile/80911
27381 * gcov.c (block_info::block_info): New constructor.
27383 2017-06-07 Carl Love <cel@us.ibm.com>
27385 * config/rs6000/rs6000-c: The return type of the following
27386 built-in functions was implemented as int not long long. Fix sign
27387 of return value for the unsigned version of vec_mulo and vec_mule.
27388 vector unsigned long long vec_bperm (vector unsigned long long,
27389 vector unsigned char)
27390 vector signed long long vec_mule (vector signed int,
27392 vector unsigned long long vec_mule (vector unsigned int,
27393 vector unsigned int)
27394 vector signed long long vec_mulo (vector signed int,
27396 vector unsigned long long vec_mulo (vector unsigned int,
27397 vector unsigned int)
27398 * doc/extend.texi: Fix the documentation for the built-in
27401 2017-06-07 Carl Love <cel@us.ibm.com>
27404 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27407 2017-06-07 Carl Love <cel@us.ibm.com>
27409 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27410 support, Generate doublehv for signed int/float for BE case only.
27412 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27414 * doc/invoke.texi (mcx16): Rewrite.
27416 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27418 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27419 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27420 *mov<mode>_softfloat, and an anonymous splitter): Use
27421 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27423 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27425 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27427 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27428 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27429 (rs6000_debug_reg_global): Adjust.
27430 (rs6000_init_hard_regno_mode_ok): Adjust.
27431 (rs6000_dbx_register_number): Adjust.
27432 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27433 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27434 Remove SPE_ACC and SPEFSCR.
27435 (REG_ALLOC_ORDER): Ditto.
27436 (FRAME_POINTER_REGNUM): Change to 111.
27437 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27438 (REG_CLASS_NAMES): Ditto.
27439 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27440 (REGISTER_NAMES): Ditto.
27441 (ADDITIONAL_REG_NAMES): Ditto.
27442 (rs6000_reg_names): Ditto.
27443 * config/rs6000/rs6000.md: Renumber some register number
27446 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27448 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27450 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27451 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27453 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27454 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27455 Delete the SPE high registers.
27456 (REG_ALLOC_ORDER): Ditto.
27457 (enum reg_class): Remove SPE_HIGH_REGS.
27458 (REG_CLASS_NAMES): Ditto.
27459 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27460 (REGISTER_NAMES): Ditto.
27461 (rs6000_reg_names): Ditto.
27462 * doc/tm.texi.in: Remove SPE as example.
27463 * doc/tm.texi: Regenerate.
27465 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27467 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27468 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27469 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27470 * config/rs6000/rs6000.md (type): Remove "brinc".
27472 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27474 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27475 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27476 * config/rs6000/linuxspe.h: Delete file.
27477 * config/rs6000/rs6000.md: Don't include spe.md.
27478 * config/rs6000/spe.h: Delete file.
27479 * config/rs6000/spe.md: Delete file.
27480 * config/rs6000/t-rs6000: Remove spe.md.
27482 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27484 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27485 (reg_or_none500mem_operand): Delete.
27486 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27487 instead of reg_or_none500mem_operand.
27489 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27491 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27492 handling of SPE flags.
27493 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27495 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27497 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27499 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27500 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27501 paired_divv2sf3): Similar.
27502 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27503 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27504 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27506 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27507 Rename the paired_* instruction patterns.
27508 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27510 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27511 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27512 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27513 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27514 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27515 PAIRED_VECTOR_MODE.
27516 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27517 (spe_func_has_64bit_regs_p): Delete.
27518 (spe_expand_predicate_builtin): Delete.
27519 (spe_expand_evsel_builtin): Delete.
27520 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27521 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27522 (invalid_e500_subreg): Delete.
27523 (rs6000_legitimize_address): Always force_reg op2 as well, for
27524 paired single memory accesses.
27525 (rs6000_member_type_forces_blk): Delete.
27526 (rs6000_spe_function_arg): Delete.
27527 (rs6000_expand_unop_builtin): Delete SPE handling.
27528 (rs6000_expand_binop_builtin): Ditto.
27529 (spe_expand_stv_builtin): Delete.
27530 (bdesc_2arg_spe): Delete.
27531 (spe_expand_builtin): Delete.
27532 (spe_expand_predicate_builtin): Delete.
27533 (spe_expand_evsel_builtin): Delete.
27534 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27535 (spe_init_builtins): Delete.
27536 (spe_func_has_64bit_regs_p): Delete.
27537 (savres_routine_name): Delete "info" parameter. Adjust callers.
27538 (rs6000_emit_stack_reset): Ditto.
27539 (rs6000_dwarf_register_span): Delete.
27540 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27541 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27542 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27544 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27546 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27547 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27548 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27549 mulv2sf3, divv2sf3): Delete expanders.
27551 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27553 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27555 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27557 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27558 * config/rs6000/rs6000.c: Ditto.
27560 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27562 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27563 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27564 comparison_operator.
27566 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27568 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27569 * config/rs6000/rs6000.opt: Ditto.
27570 * config/rs6000/t-rtems: Ditto.
27572 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27574 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27575 TARGET_E500_SINGLE by 0, simplify.
27576 * config/rs6000/rs6000.c: Ditto.
27577 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27578 (spe_build_register_parallel): Delete.
27579 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27580 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27581 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27582 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27583 (E500_CONVERT): Delete.
27584 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27586 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27588 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27589 * config/rs6000/dfp.md: Ditto.
27590 (negdd2, *negdd2_fpr): Merge.
27591 (absdd2, *absdd2_fpr): Merge.
27592 (negtd2, *negtd2_fpr): Merge.
27593 (abstd2, *abstd2_fpr): Merge.
27594 * config/rs6000/e500.h: Delete file.
27595 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27596 TARGET_FPRS by 1 and simplify.
27597 * config/rs6000/rs6000-c.c: Ditto.
27598 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27599 TARGET_DF_SPE by 0.
27600 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27602 * config/rs6000/rs6000.md: Ditto.
27603 (floatdidf2, *floatdidf2_fpr): Merge.
27604 (move_from_CR_gt_bit): Delete.
27605 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27606 (E500_CR_IOR_COMPARE): Delete.
27607 (All patterns that require !TARGET_FPRS): Delete.
27608 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27610 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27612 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27614 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27616 * graphds.c (add_edge): Intitialize edge's attached data.
27617 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27618 pointer parameter. Call pointed function on each edge during
27619 graph traversing. Skip traversing the edge when the function
27621 (graphds_dfs, graphds_scc): Ditto.
27622 (for_each_edge): New parameter. Pass the new parameter to callback
27624 * graphds.h (skip_edge_callback): New function pointer type.
27625 (graphds_dfs, graphds_scc): New function pointer parameter.
27626 (graphds_edge_callback, for_each_edge): New parameter.
27628 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27630 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27631 out code checking if runtime alias check is possible to below ...
27632 Call the new function.
27633 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27634 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27636 2017-06-07 Marek Polacek <polacek@redhat.com>
27639 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27640 TYPE_OVERFLOW_WRAPS check.
27642 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27644 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27645 if versioning is required.
27646 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27647 peeling with the check for versioning.
27649 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27651 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27652 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27653 Set true to new parameter if new ssa variable is defined.
27654 (vect_gen_vector_loop_niters): Refactor. Set range information
27655 for the new vector loop bound variable.
27656 (vect_do_peeling): Ditto.
27658 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27660 * tree-affine.c (ssa.h): Include header file.
27661 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27662 has wrapping overflow behavior.
27664 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27666 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27668 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27670 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27671 (tree_to_aff_combination): ... here.
27673 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27675 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27676 reg_pressure model function.
27677 (ivopts_global_cost_for_size): Delete.
27678 (determine_set_costs, iv_ca_recount_cost): Call new model function
27679 ivopts_estimate_reg_pressure.
27681 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27683 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27684 expensive than udiv. Remove floating point cases from mod.
27686 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27688 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27689 Increase idiv cost.
27691 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27693 * config/aarch64/aarch64.md
27694 (copysignsf3): Fix mask generation.
27696 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27698 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27700 (class dump_manager): Add register_dumps method.
27701 * dumpfile.c: Include langhooks.h.
27702 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27703 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27704 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27705 (dump_manager::dump_register): Start with 512 entries instead of 32.
27706 (dump_manager::register_dumps): New method.
27707 * toplev.c (general_init): Instead of invoking register_dumps
27708 langhook, invoke register_dumps method on the dump manager.
27709 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27712 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27714 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27715 Say that # requires an associated define_split to exist, and that
27716 the define_split must be suitable for use after register allocation.
27718 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27720 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27721 (compute_outgoing_frequencies): Also initialize zero counts.
27722 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27723 around loops; preserve more of profile when nothing changes.
27725 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27727 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27729 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27730 * config/arm/arm-cpu-cdata.h: Regenerate.
27731 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27732 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27733 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27734 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27735 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27737 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27739 * config/arm/t-rmprofile: Likewise.
27740 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27742 2017-06-06 David S. Miller <davem@davemloft.net>
27745 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27746 blockage if function uses alloca.
27748 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27750 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27751 New "uid" fields to hold pretty-print IDs of group and ref.
27752 Memory references are now identified as <group_id>:<ref_id>
27753 instead of using [random] addresses.
27754 (dump_mem_details): Simplify, no functional change.
27755 (dump_mem_ref): Simplify and make output more concise.
27756 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27757 (find_or_create_group): Initialize group uid.
27758 (record_ref): Initialize ref uid. Improve debug output.
27759 (prune_group_by_reuse, should_issue_prefetch_p,)
27760 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27761 (mark_nontemporal_store, determine_loop_nest_reuse):
27762 Improve debug output.
27764 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27766 * dbgcnt.def (prefetch): New debug counter.
27767 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27768 (schedule_prefetches): Stop issueing prefetches if debug counter
27771 2017-06-06 Tom de Vries <tom@codesourcery.com>
27773 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27774 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27776 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27778 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27779 Use aarch64_reg_or_zero predicate for operand 4.
27780 (aarch64_compare_and_swap<mode> define_insn_and_split):
27781 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27782 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27784 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27786 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27787 (arm_compute_save_core_reg_mask): This.
27788 (thumb1_compute_save_reg_mask): Rename into ...
27789 (thumb1_compute_save_core_reg_mask): This.
27790 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27791 (arm_compute_frame_layout): Likewise.
27793 2017-06-06 Richard Biener <rguenther@suse.de>
27795 PR tree-optimization/80974
27796 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27797 keep or clear leaders SSA info.
27799 2017-06-06 Tom de Vries <tom@codesourcery.com>
27801 * config/nvptx/nvptx.c (split_mode_p): New function.
27802 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27804 2017-06-06 Tom de Vries <tom@codesourcery.com>
27806 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27808 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27811 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27814 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27816 * shrink-wrap.c (handle_simple_exit): Update profile.
27817 (try_shrink_wrapping): Upate profile.
27819 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27821 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27822 (tree_guess_outgoing_edge_probabilities): New.
27823 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27824 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27826 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27828 * ipa-split.c (split_function): Initialize return bb profile.
27830 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27832 * profile.c (compute_branch_probabilities): Also initialize
27833 EXIT_BLOCK profile.
27835 2017-06-06 Richard Biener <rguenther@suse.de>
27837 PR tree-optimization/80928
27838 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27839 (vect_analyze_loop_operations): Properly guard analysis for
27841 (vect_transform_loop): Likewise.
27842 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27843 (vect_model_induction_cost): Do not cost for pure SLP.
27844 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27845 of induction in inner loop vectorization.
27846 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27847 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27848 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27850 (vect_analyze_slp_cost_1): Cost induction.
27851 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27852 (vect_get_slp_vect_defs): Likewise.
27853 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27854 (vect_transform_stmt): Handle SLP reductions.
27855 * tree-vectorizer.h (vectorizable_induction): Adjust.
27857 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27859 * config/rs6000/rs6000.c (make_resolver_func): Update
27860 init_lowered_empty_function call.
27862 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
27864 * doc/invoke.texi: Document the -fprofile-abs-path option.
27865 * common.opt (fprofile-abs-path): New option.
27866 * gcov-io.h (gcov_write_filename): Declare.
27867 * gcov-io.c (gcov_write_filename): New function.
27868 * coverage.c (coverage_begin_function): Use gcov_write_filename.
27869 * profile.c (output_location): Likewise.
27871 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27873 * shring-wrap.c: Revert accidental commit.
27875 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
27877 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
27879 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27881 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
27883 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
27884 profile in callgraph edge.
27885 * profile-count.h (apply_probability): If THIS is 0, then result is 0
27886 (apply_scale): Likewise.
27887 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
27888 Also scale profile when inlining function with zero profile.
27889 (initialize_cfun): Update exit block profile even when it is zero.
27890 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
27891 when profile is read.
27893 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27895 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
27896 (CLONE_*): New constants to define the processors we can generate
27897 code for with the target_clone attribute.
27898 (rs6000_clone_map): New array to identify which clone processors
27899 the current program is running on.
27900 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
27901 target_clone attribute.
27902 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
27903 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
27904 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
27905 (cpu_expand_builtin): Add support for target_clone attribute.
27906 (rs6000_valid_attribute_p): Allow "default" attribute.
27907 (get_decl_name): New debug function to simplify printing the
27908 current function name in debugging statements.
27909 (rs6000_clone_priority): New functions to support the target_clone
27910 attribute, and be able to generate code to switch between ISA 2.05
27911 through ISA 3.0 (power6 through power9).
27912 (rs6000_compare_version_priority): Likewise.
27913 (rs6000_get_function_versions_dispatcher): Likewise.
27914 (make_resolver_func): Likewise.
27915 (add_condition_to_bb): Likewise.
27916 (dispatch_function_versions): Likewise.
27917 (rs6000_generate_version_dispatcher_body): Likewise.
27918 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
27919 (fusion_gpr_load_p): Fix a spacing issue.
27920 * doc/extend.texi (Common Function Attributes): Document that the
27921 PowerPC supports the target_clone attribute.
27923 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27925 * config/arm/arm.h: explain F symbol found in description of ARM
27926 register allocation in its legend.
27928 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27930 * config/mips/frame-header-opt.c: Include profile-count.h.
27931 * config/riscv/riscv.c: Include profile-count.h
27933 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27935 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
27937 (sm_set_flag_if_changed): Add bbs field.
27938 (execute_sm_if_changed_flag_set): Pass BBS.
27939 (execute_sm): Update.
27941 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27943 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
27946 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27948 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
27949 (peephole2): New peephole2 to emit the above.
27950 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
27952 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27954 * config/aarch64/aarch64.c (define_peephole2 above
27955 *sub_<shift>_<mode>): New peephole.
27957 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27959 * config/i386/i386.c (make_resolver_func): Update.
27960 * Makefile.in: Add profile-count.h and profile-count.o
27961 * auto-profile.c (afdo_indirect_call): Update to new API.
27962 (afdo_set_bb_count): Update.
27963 (afdo_propagate_edge): Update.
27964 (afdo_propagate_circuit): Update.
27965 (afdo_calculate_branch_prob): Update.
27966 (afdo_annotate_cfg): Update.
27967 * basic-block.h: Include profile-count.h
27968 (struct edge_def): Turn count to profile_count.
27969 (struct basic_block_def): Likewie.
27970 (REG_BR_PROB_BASE): Move to profile-count.h
27971 (RDIV): Move to profile-count.h
27972 * bb-reorder.c (max_entry_count): Turn to profile_count.
27973 (find_traces): Update.
27974 (rotate_loop):Update.
27975 (connect_traces):Update.
27976 (sanitize_hot_paths):Update.
27977 * bt-load.c (migrate_btr_defs): Update.
27978 * cfg.c (RDIV): Remove.
27979 (init_flow): Use alloc_block.
27980 (alloc_block): Uninitialize count.
27981 (unchecked_make_edge): Uninitialize count.
27982 (check_bb_profile): Update.
27983 (dump_edge_info): Update.
27984 (dump_bb_info): Update.
27985 (update_bb_profile_for_threading): Update.
27986 (scale_bbs_frequencies_int): Update.
27987 (scale_bbs_frequencies_gcov_type): Update.
27988 (scale_bbs_frequencies_profile_count): New.
27989 * cfg.h (update_bb_profile_for_threading): Update.
27990 (scale_bbs_frequencies_profile_count): Declare.
27991 * cfgbuild.c (compute_outgoing_frequencies): Update.
27992 (find_many_sub_basic_blocks): Update.
27993 * cfgcleanup.c (try_forward_edges): Update.
27994 (try_crossjump_to_edge): Update.
27995 * cfgexpand.c (expand_gimple_tailcall): Update.
27996 (construct_exit_block): Update.
27997 * cfghooks.c (verify_flow_info): Update.
27998 (dump_bb_for_graph): Update.
27999 (split_edge): Update.
28000 (make_forwarder_block): Update.
28001 (duplicate_block): Update.
28002 (account_profile_record): Update.
28003 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28004 (get_estimated_loop_iterations): Update.
28005 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28006 (single_likely_exit): Update.
28007 * cfgloopmanip.c (scale_loop_profile): Update.
28009 (set_zero_probability): Update.
28010 (lv_adjust_loop_entry_edge): Update.
28011 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28012 (purge_dead_edges): Update.
28013 (rtl_account_profile_record): Update.
28014 * cgraph.c (cgraph_node::create): Uninitialize count.
28015 (symbol_table::create_edge): Uninitialize count.
28016 (cgraph_update_edges_for_call_stmt_node): Update.
28017 (cgraph_edge::dump_edge_flags): Update.
28018 (cgraph_node::dump): Update.
28019 (cgraph_edge::maybe_hot_p): Update.
28020 * cgraph.h: Include profile-count.h
28021 (create_clone), create_edge, create_indirect_edge): Update.
28022 (cgraph_node): Turn count to profile_count.
28023 (cgraph_edge0: Likewise.
28024 (make_speculative, clone): Update.
28025 (create_edge): Update.
28026 (init_lowered_empty_function): Update.
28027 * cgraphclones.c (cgraph_edge::clone): Update.
28028 (duplicate_thunk_for_node): Update.
28029 (cgraph_node::create_clone): Update.
28030 * cgraphunit.c (cgraph_node::analyze): Update.
28031 (cgraph_node::expand_thunk): Update.
28032 * final.c (dump_basic_block_info): Update.
28033 * gimple-streamer-in.c (input_bb): Update.
28034 * gimple-streamer-out.c (output_bb): Update.
28035 * graphite.c (print_global_statistics): Update.
28036 (print_graphite_scop_statistics): Update.
28037 * hsa-brig.c: Include basic-block.h.
28038 * hsa-dump.c: Include basic-block.h.
28039 * hsa-gen.c (T sum_slice): Update.
28040 (convert_switch_statements):Update.
28041 * hsa-regalloc.c: Include basic-block.h.
28042 * ipa-chkp.c (chkp_produce_thunks): Update.
28043 * ipa-cp.c (struct caller_statistics): Update.
28044 (init_caller_stats): Update.
28045 (gather_caller_stats): Update.
28046 (ipcp_cloning_candidate_p): Update.
28047 (good_cloning_opportunity_p): Update.
28048 (get_info_about_necessary_edges): Update.
28049 (dump_profile_updates): Update.
28050 (update_profiling_info): Update.
28051 (update_specialized_profile): Update.
28052 (perhaps_add_new_callers): Update.
28053 (decide_about_value): Update.
28054 (ipa_cp_c_finalize): Update.
28055 * ipa-devirt.c (struct odr_type_warn_count): Update.
28056 (struct decl_warn_count): Update.
28057 (struct final_warning_record): Update.
28058 (possible_polymorphic_call_targets): Update.
28059 (ipa_devirt): Update.
28060 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28061 * ipa-icf.c (sem_function::merge): Update.
28062 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28063 * ipa-inline.c (compute_uninlined_call_time): Update.
28064 (compute_inlined_call_time): Update.
28065 (want_inline_small_function_p): Update.
28066 (want_inline_self_recursive_call_p): Update.
28067 (edge_badness): Update.
28068 (lookup_recursive_calls): Update.
28069 (recursive_inlining): Update.
28070 (inline_small_functions): Update.
28071 (dump_overall_stats): Update.
28072 (dump_inline_stats): Update.
28073 * ipa-profile.c (ipa_profile_generate_summary): Update.
28074 (ipa_propagate_frequency): Update.
28075 (ipa_profile): Update.
28076 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28077 * ipa-utils.c (ipa_merge_profiles): Update.
28078 * loop-doloop.c (doloop_modify): Update.
28079 * loop-unroll.c (report_unroll): Update.
28080 (unroll_loop_runtime_iterations): Update.
28081 * lto-cgraph.c (lto_output_edge): Update.
28082 (lto_output_node): Update.
28083 (input_node): Update.
28084 (input_edge): Update.
28085 (merge_profile_summaries): Update.
28086 * lto-streamer-in.c (input_cfg): Update.
28087 * lto-streamer-out.c (output_cfg): Update.
28088 * mcf.c (create_fixup_graph): Update.
28089 (adjust_cfg_counts): Update.
28090 (sum_edge_counts): Update.
28091 * modulo-sched.c (sms_schedule): Update.
28092 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28093 * predict.c (maybe_hot_count_p): Update.
28094 (probably_never_executed): Update.
28095 (dump_prediction): Update.
28096 (combine_predictions_for_bb): Update.
28097 (propagate_freq): Update.
28098 (handle_missing_profiles): Update.
28099 (counts_to_freqs): Update.
28100 (rebuild_frequencies): Update.
28101 (force_edge_cold): Update.
28102 * predict.h: Include profile-count.h
28103 (maybe_hot_count_p, counts_to_freqs): UPdate.
28104 * print-rtl-function.c: Do not include cfg.h
28105 * print-rtl.c: Include basic-block.h
28106 * profile-count.c: New file.
28107 * profile-count.h: New file.
28108 * profile.c (is_edge_inconsistent): Update.
28109 (correct_negative_edge_counts): Update.
28110 (is_inconsistent): Update.
28111 (set_bb_counts): Update.
28112 (read_profile_edge_counts): Update.
28113 (compute_frequency_overlap): Update.
28114 (compute_branch_probabilities): Update; Initialize and deinitialize
28116 (branch_prob): Update.
28117 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28118 (edge_gcov_count): New.
28119 (bb_gcov_count): New.
28120 * shrink-wrap.c (try_shrink_wrapping): Update.
28121 * tracer.c (better_p): Update.
28122 * trans-mem.c (expand_transaction): Update.
28123 (ipa_tm_insert_irr_call): Update.
28124 (ipa_tm_insert_gettmclone_call): Update.
28125 * tree-call-cdce.c: Update.
28126 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28127 (gimple_duplicate_sese_tail): Update.
28128 (gimple_account_profile_record): Update.
28129 (execute_fixup_cfg): Update.
28130 * tree-inline.c (copy_bb): Update.
28131 (copy_edges_for_bb): Update.
28132 (initialize_cfun): Update.
28133 (freqs_to_counts): Update.
28134 (copy_cfg_body): Update.
28135 (expand_call_inline): Update.
28136 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28137 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28138 (try_unroll_loop_completely): Update.
28139 (try_peel_loop): Update.
28140 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28141 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28142 * tree-ssa-loop-split.c (connect_loops): Update.
28143 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28144 * tree-ssa-reassoc.c (branch_fixup): Update.
28145 * tree-ssa-tail-merge.c (replace_block_by): Update.
28146 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28147 (compute_path_counts): Update.
28148 (update_profile): Update.
28149 (recompute_probabilities): Update.
28150 (update_joiner_offpath_counts): Update.
28151 (estimated_freqs_path): Update.
28152 (freqs_to_counts_path): Update.
28153 (clear_counts_path): Update.
28154 (ssa_fix_duplicate_block_edges): Update.
28155 (duplicate_thread_path): Update.
28156 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28157 (struct switch_conv_info): Update.
28158 * tree-tailcall.c (decrease_profile): Update.
28159 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28160 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28161 * value-prof.c (check_counter): Update.
28162 (gimple_divmod_fixed_value): Update.
28163 (gimple_mod_pow2): Update.
28164 (gimple_mod_subtract): Update.
28165 (gimple_ic_transform): Update.
28166 (gimple_stringop_fixed_value): Update.
28167 * value-prof.h (gimple_ic): Update.
28169 2017-06-02 Carl Love <cel@us.ibm.com>
28171 * config/rs6000/rs6000-c: Add support for built-in functions
28172 vector double vec_doublee (vector signed int);
28173 vector double vec_doublee (vector unsigned int);
28174 vector double vec_doublee (vector float);
28175 vector double vec_doubleh (vector signed int);
28176 vector double vec_doubleh (vector unsigned int);
28177 vector double vec_doubleh (vector float);
28178 vector double vec_doublel (vector signed int);
28179 vector double vec_doublel (vector unsigned int);
28180 vector double vec_doublel (vector float);
28181 vector double vec_doubleo (vector signed int);
28182 vector double vec_doubleo (vector unsigned int);
28183 vector double vec_doubleo (vector float);.
28184 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28185 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28187 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28188 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28189 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28191 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28192 vec_doublel, vec_doubleh.
28193 * doc/extend.texi: Update the built-in documentation file for the
28194 new built-in functions.
28196 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28199 * ipa-inline-analysis.c (free_growth_caches): Set
28200 edge_removal_hook_holder to NULL after removing it.
28202 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28204 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28205 comparision with zero.
28207 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28208 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28209 for early expansion of vec_min and vec_max builtins.
28210 (builtin_function_type): Add min/max unsigned variants to those
28211 identified as having unsigned arguments.
28213 2017-06-02 Olivier Hainque <hainque@adacore.com>
28215 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28217 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28219 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28220 Use VALL_F16 iterator rather than VALL.
28222 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28224 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28225 Emit CBNZ inside loop when doing a strong exchange and comparing
28226 against zero. Generate the CC flags after the loop.
28228 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28230 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28231 (dl_section_ref): New.
28232 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28233 On AIX, append an expression to subtract the size of the
28234 section length to dl_section_ref.
28236 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28238 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28239 for early expansion of vector absolute builtins.
28241 2017-06-02 Richard Biener <rguenther@suse.de>
28243 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28244 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28246 2017-06-02 Richard Biener <rguenther@suse.de>
28248 PR tree-optimization/80948
28249 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28250 stmt order as well.
28252 2017-06-02 Richard Biener <rguenther@suse.de>
28254 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28256 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28257 for inductions as relevant.
28259 2017-06-02 Richard Biener <rguenther@suse.de>
28261 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28262 (vectorizable_induction): ... this. Remove dead code.
28264 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28266 * builtins. (expand_builtin_alloca): Remove second parameter and
28267 infer its value from the first parameter instead.
28268 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28270 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28272 PR rtl-optimization/80903
28273 * loop-doloop.c (add_test): Unshare sequence.
28275 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28277 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28279 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28281 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28283 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28284 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28285 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28286 (xlogue_layout::get_stub_rtx): Make static.
28287 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28288 (xlogue_layout::compute_stub_managed_regs): Rename to...
28289 (xlogue_layout::count_stub_managed_regs): ...this.
28290 (xlogue_layout::is_stub_managed_reg): New function.
28291 (xlogue_layout::m_stub_names): Rename to...
28292 (xlogue_layout::s_stub_names): ...this, make static.
28293 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28294 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28295 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28296 xlogue_layout::s_stub_names): Instantiate statics.
28297 (stub_managed_regs): Remove.
28298 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28299 (disable_call_ms2sysv_xlogues): Rename to...
28300 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28301 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28303 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28304 change after reload_completed.
28305 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28307 (ix86_expand_prologue): Likewise.
28308 (ix86_expand_epilogue): Likewise.
28309 (ix86_expand_split_stack_prologue): Likewise.
28310 (ix86_compute_frame_layout): Remove frame parameter ...
28311 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28312 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28314 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28315 (ix86_frame): Move from here ...
28316 * config/i386/i386.h (ix86_frame): ... to here.
28317 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28318 complete ix86_frame data structure instead. Remove some_ld_name.
28320 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28322 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28323 symbols that hold a DECL_VALUE_EXPR.
28325 2017-06-01 Martin Jambor <mjambor@suse.cz>
28327 PR tree-optimization/80898
28328 * tree-sra.c (process_subtree_disqualification): Removed.
28329 (disqualify_candidate): Do not acll
28330 process_subtree_disqualification.
28331 (subtree_mark_written_and_enqueue): New function.
28332 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28333 RHS has been disqualified and re-queue LHS if necessary. Apart
28334 from that, ignore disqualified RHS.
28336 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28338 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28339 address fetch for z10 or later.
28341 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28343 * config/arc/arc.md (tst_movb): Add guard when splitting.
28345 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28347 * config/arc/arc.c (arc_can_eliminate): Test against
28348 arc_frame_pointer_needed.
28350 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28352 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28353 to prevent store reordering.
28354 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28355 (type): Add block type.
28356 (stack_tie): Define special instruction to be used in
28359 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28361 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28362 constraint. It is not valid for the pattern.
28363 (noncommutative_binary_comparison): Likewise.
28365 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28367 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28370 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28372 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28373 be used by the reg-alloc.
28375 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28377 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28378 reg-alloc when having mul64 or mul32x16 instructions.
28379 (mulsidi3): Likewise.
28380 (umulsidi3): Likewise.
28381 (mulsi32x16): New pattern.
28382 (mulsi64): Likewise.
28383 (mulsidi64): Likewise.
28384 (umulsidi64): Likewise.
28385 (MUL32x16_REG): Define.
28386 (mul64_600): Use MUL32x16_REG.
28387 (mac64_600): Likewise.
28388 (umul64_600): Likewise.
28389 (umac64_600): Likewise.
28391 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28393 * config/arc/arc.md (mulsi3_700): Make it commutative.
28395 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28397 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28399 (*sign_extendsidi2_insn): Likewise for movstosw.
28401 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28403 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28404 the type of the input discriminant value. Convert the
28405 discriminant value of signedness vary.
28407 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28409 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28410 Add to -Wall section.
28412 2017-06-01 Richard Biener <rguenther@suse.de>
28414 PR middle-end/66313
28415 * fold-const.c (fold_plusminus_mult_expr): If the factored
28416 factor may be zero use a wrapping type for the inner operation.
28417 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28418 and handle moved defs.
28419 (process_assignment): Properly guard the unary op case. Return a
28420 tri-state indicating that moving the stmt before the call may allow
28421 to continue. Pass through to_move.
28422 (find_tail_calls): Handle moving unrelated defs before
28425 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28428 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28429 splitter result in the canonical way.
28431 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28433 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28434 also for 32bit target. Update insn attributes.
28435 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28437 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28439 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28440 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28441 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28442 (_mm_maskz_min_ss): New intrinsics.
28444 2017-05-31 Martin Liska <mliska@suse.cz>
28446 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28447 Change comment style to one we normally use.
28448 (vectorizable_reduction): Likewise.
28449 (vectorizable_induction): Likewise.
28450 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28451 (vectorizable_call): Likewise.
28452 (vectorizable_simd_clone_call): Likewise.
28453 (vectorizable_conversion): Likewise.
28454 (vectorizable_assignment): Likewise.
28455 (vectorizable_shift): Likewise.
28456 (vectorizable_operation): Likewise.
28457 (vectorizable_store): Likewise.
28458 (vectorizable_load): Likewise.
28459 * tree-vectorizer.h: Likewise.
28461 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28463 * passes.c (emergency_dump_function): New.
28464 * tree-pass.h (emergency_dump_function): Declare.
28465 * plugin.c (plugins_internal_error_function): Remove.
28466 * plugin.h (plugins_internal_error_function): Remove declaration.
28467 * toplev.c (internal_error_function): New static function. Use it...
28468 (general_init): ...here.
28470 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28472 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28473 (arc_rtx_costs): Add costs for new patterns.
28474 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28475 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28477 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28479 * tree-ssa-strlen.c (get_next_strinfo): New function.
28480 (get_stridx_plus_constant): Use it.
28481 (zero_length_string): Likewise.
28482 (adjust_related_strinfos): Likewise.
28483 (adjust_last_stmt): Likewise.
28485 2017-05-31 Richard Biener <rguenther@suse.de>
28488 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28489 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28491 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28493 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28494 loop_vinfo argument and use of dependence distance vectors.
28495 Check instead whether the two references differ only in their
28496 initial value and assume that they have the same alignment if the
28497 difference is a multiple of the vector alignment.
28498 (vect_analyze_data_refs_alignment): Update call accordingly.
28500 2017-05-31 Martin Liska <mliska@suse.cz>
28503 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28505 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28507 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28508 (create_intersect_range_checks): Move from ...
28509 * tree-data-ref.c (create_intersect_range_checks_index)
28510 (create_intersect_range_checks): ... to here.
28511 (create_runtime_alias_checks): New function factored from ...
28512 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28513 here. Call above function.
28514 * tree-data-ref.h (create_runtime_alias_checks): New function.
28516 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28518 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28519 segment length for dr_b and compute it in wide_int.
28521 2017-05-31 Richard Biener <rguenther@suse.de>
28523 PR tree-optimization/80906
28524 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28525 and pass through iv_map.
28526 (copy_bb_and_scalar_dependences): Adjust.
28527 (translate_pending_phi_nodes): Likewise.
28528 (copy_loop_close_phi_args): Handle code-generating IVs instead
28531 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28533 * diagnostic-color.c (color_dict): Add "type-diff".
28534 (parse_gcc_colors): Update comment.
28535 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28536 -fdiagnostics-show-template-tree and -fno-elide-type.
28537 (GCC_COLORS): Add type-diff to example.
28539 (-fdiagnostics-show-template-tree): New.
28540 (-fno-elide-type): New.
28541 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28542 the pp_format_decoder callback. Call any m_format_postprocessor's
28544 (pretty_printer::pretty_printer): Initialize
28545 m_format_postprocessor.
28546 (pretty_printer::~pretty_printer): Delete any
28547 m_format_postprocessor.
28548 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28549 (class format_postprocessor): New class.
28550 (struct pretty_printer::format_decoder): Document the new parameters.
28551 (struct pretty_printer::m_format_postprocessor): New field.
28552 * tree-diagnostic.c (default_tree_printer): Update for new
28553 bool and const char ** params.
28554 * tree-diagnostic.h (default_tree_printer): Likewise.
28556 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28558 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28559 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28560 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28561 rs6000_gen_cell_microcode code.
28562 (rs6000_final_prescan_insn): Delete.
28563 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28564 "warn-cell-microcode" entries.
28565 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28566 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28567 throughout. Change cc_reg_not_micro_cr0_operand to
28568 cc_reg_not_cr0_operand throughout.
28569 (*extendhi<mode>2_noload): Delete.
28570 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28571 (mwarn-cell-microcode): Delete.
28572 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28573 -mgen-cell-microcode and -mwarn-cell-microcode.
28575 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28578 * config/i386/constraints.md (Yd): New constraint.
28580 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28581 and (?Yd, r) alternatives. Update insn attributes.
28582 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28583 and (?*Yd, r) alternatives. Update insn attributes.
28584 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28586 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28588 * gimplify.c (gimplify_modify_expr): Don't create a
28589 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28592 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28594 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28596 2017-05-30 Richard Biener <rguenther@suse.de>
28598 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28599 and reduc_def fields.
28600 (STMT_VINFO_REDUC_TYPE): New define.
28601 (STMT_VINFO_REDUC_DEF): Likewise.
28602 (vect_force_simple_reduction): Adjust prototype.
28603 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28604 (vect_is_simple_reduction): Remove check_reduction argument.
28605 (vect_force_simple_reduction): Adjust and set
28606 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28607 (vectorizable_reduction): Do not re-do reduction analysis
28608 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28609 * tree-parloops.c (gather_scalar_reductions): Adjust.
28611 2017-05-30 Richard Biener <rguenther@suse.de>
28613 PR middle-end/80901
28614 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28617 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28619 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28620 Introduce unknown_misalignment parameter and remove vf.
28621 (vect_peeling_hash_get_lowest_cost):
28622 Pass unknown_misalignment parameter.
28623 (vect_enhance_data_refs_alignment):
28624 Fix unsupportable data ref treatment.
28626 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28628 * tree-vect-data-refs.c (vect_get_data_access_cost):
28629 Workaround for SLP handling.
28630 (vect_enhance_data_refs_alignment):
28631 Compute costs for doing no peeling at all, compare to the best
28632 peeling costs so far and avoid peeling if cheaper.
28634 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28636 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28637 Return peeling info and set costs to zero for unlimited cost
28639 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28640 with unknown misalignment. Compute and costs for unknown
28641 misalignment, compare them to the costs for known misalignment
28642 and choose the cheapest for peeling.
28644 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28646 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28647 (vect_get_peeling_costs_all_drs): Create function.
28648 (vect_peeling_hash_get_lowest_cost):
28649 Use vect_get_peeling_costs_all_drs.
28650 (vect_peeling_supportable): Create function.
28651 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28653 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28655 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28656 DR_HAS_NEGATIVE_STEP.
28657 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28658 (vect_enhance_data_refs_alignment): Use.
28659 (vect_duplicate_ssa_name_ptr_info): Use.
28660 * tree-vectorizer.h (dr_misalignment): Use.
28661 (known_alignment_for_access_p): Use.
28663 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28666 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28668 (has_section_name): New function.
28670 2017-05-30 Martin Liska <mliska@suse.cz>
28673 * auto-profile.c (get_function_decl_from_block): Fix
28676 2017-05-30 Richard Biener <rguenther@suse.de>
28678 PR middle-end/80876
28679 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28681 2017-05-30 Martin Liska <mliska@suse.cz>
28683 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28684 * dumpfile.h (struct dump_file_info): Remove ctors.
28686 2017-05-30 Martin Liska <mliska@suse.cz>
28688 * predict.def: Fix GNU coding style.
28690 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28692 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28693 Mark 'to' argument with ATTRIBUTE_UNUSED.
28695 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28697 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28698 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28699 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28702 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28704 * doc/install.texi (Options specification): Restore entry of
28705 --enable-sjlj-exceptions.
28707 2017-05-27 Michael Eager <eager@eagercon.com>
28710 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28712 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28714 * config/microblaze/microblaze.h
28715 (FIXED_REGISTERS): Update in macro.
28716 (CALL_USED_REGISTERS): Update in macro.
28718 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28720 * doc/install.texi: Add links to macOS binary distributions.
28722 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28726 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28728 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28730 2017-05-26 Martin Liska <mliska@suse.cz>
28732 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28734 2017-05-26 Martin Liska <mliska@suse.cz>
28736 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28737 always leading ';; '.
28738 (dump_bb_info): Likewise.
28739 (brief_dump_cfg): Likewise.
28740 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28741 * dumpfile.c: Remove usage of TDF_VERBOSE.
28742 * dumpfile.h (enum dump_kind): Likewise.
28743 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28744 * print-tree.c (debug_verbose): Remove.
28745 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28746 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28747 * tree-diagnostic.c (default_tree_printer): Replace
28748 TDF_DIAGNOSTIC with TDF_SLIM.
28750 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28752 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28753 in parameter loop, rather than loop_vinfo.
28754 (create_intersect_range_checks): Ditto.
28755 (vect_create_cond_for_alias_checks): Update call to above functions.
28757 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28759 PR tree-optimization/80815
28760 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28761 for merging runtime alias checks. Handle negative DR_STEPs.
28763 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28765 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28767 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28768 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28769 out code pruning runtime alias checks.
28770 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28771 factored out from above.
28772 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28774 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28776 (prune_runtime_alias_test_list): New decalaration.
28778 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28780 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28781 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28782 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28783 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28784 (operator==, comp_dr_with_seg_len_pair): Ditto.
28785 (vect_prune_runtime_alias_test_list): Ditto.
28787 2017-05-26 Martin Liska <mliska@suse.cz>
28790 * params.def: Bound partial-inlining-entry-probability param.
28792 2017-05-26 Marek Polacek <polacek@redhat.com>
28795 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28798 2017-05-26 Richard Biener <rguenther@suse.de>
28800 PR tree-optimization/80842
28801 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28804 2017-05-26 Richard Biener <rguenther@suse.de>
28806 PR tree-optimization/80844
28807 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28809 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28811 * doc/md.texi (Machine Constraints): Update x86 family
28812 machine constraints section to match 'config/i386/constraints.md'.
28814 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28816 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28818 2017-05-25 Nathan Sidwell <nathan@acm.org>
28820 * doc/invoke.texi (--enable-languages): Update documentation.
28822 2017-05-25 Martin Liska <mliska@suse.cz>
28824 * dumpfile.c: Add TDF_FOLDING.
28825 * dumpfile.h (enum dump_kind): Likewise.
28826 * genmatch.c (dt_simplify::gen_1): Use it.
28828 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28830 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28832 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28834 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28835 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28837 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28839 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28840 * match.pd (X == C): Rewrite it here.
28841 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28842 with_certain_nonzero_bits2): New predicates.
28843 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28845 2017-05-24 Nathan Sidwell <nathan@acm.org>
28847 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28850 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28853 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28855 * config/powerpcspe: New port. Files are copied from the rs6000
28856 port, with "rs6000" in filenames replaced by "powerpcspe".
28858 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
28860 PR rtl-optimization/80754
28861 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
28863 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
28865 * config/sparc/sparc.md (length): Return the correct value for -mflat
28866 sibcalls to match output_sibcall.
28868 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28872 * config/rs6000/rs6000.c (struct machine_function): Add new field
28874 (rs6000_get_separate_components): Init that field, use it.
28875 (rs6000_components_for_bb): Use the field.
28877 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
28879 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
28881 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
28883 PR middle-end/80823
28884 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
28886 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28889 * config/s390/s390.c (s390_check_qrst_address): Check incoming
28890 address against address_operand predicate.
28891 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
28893 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
28895 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
28896 parameters passed indirectly.
28898 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
28900 * config/i386/i386.md (*movdi_internal): Remove SSE4
28901 alternative 18 (?r, *v). Update insn attributes.
28902 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
28903 Update insn attributes.
28904 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
28905 Update insn attributes.
28906 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
28907 alternative 1 (r, v). Remove isa attribute.
28908 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
28909 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
28910 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
28912 2017-05-23 Tom de Vries <tom@codesourcery.com>
28914 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
28917 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28919 * cgraphunit.c (symbol_table::process_new_functions): Update.
28920 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
28921 (inline_generate_summary): Rename to ...
28922 (ipa_fn_summary_generate): ... this one.
28923 (inline_read_summary): Rename to ...
28924 (ipa_fn_summary_read): ... this one.
28925 (inline_write_summary): Rename to ...
28926 (ipa_fn_summary_write): ... this one.
28927 (inline_free_summary): Rename to ...
28928 (ipa_free_fn_summary): ... this one.
28929 (pass_data_local_fn_summary, pass_local_fn_summary,
28930 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
28931 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
28932 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
28933 make_pass_ipa_fn_summary): New.
28934 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
28935 inline_write_summary, inline_free_summary): Remove.
28936 (ipa_free_fn_summary) : New.
28937 * ipa-inline.c (ipa_inline): Update.
28938 (pass_ipa_inline): Do not generate summaries.
28939 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
28941 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
28942 and add pass_ipa_fn_summary.
28943 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
28945 (make_pass_inline_parameters): Remove.
28947 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
28949 * omp-low.c (struct omp_context): Remove "default_kind" member.
28952 * omp-offload.c (execute_oacc_device_lower): Remove the
28953 parallelism dimensions function attributes for unparallelized
28954 OpenACC kernels constructs.
28956 2017-05-23 Martin Liska <mliska@suse.cz>
28958 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
28960 (cgraph_edge::make_speculative): Likewise.
28961 (cgraph_edge::resolve_speculation): Likewise.
28962 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28963 (cgraph_node::dump): Likewise.
28964 * cgraph.h: Likewise.
28965 * cgraphunit.c (analyze_functions): Likewise.
28966 (symbol_table::compile): Likewise.
28967 * ipa-cp.c (print_all_lattices): Likewise.
28968 (determine_versionability): Likewise.
28969 (initialize_node_lattices): Likewise.
28970 (ipcp_verify_propagated_values): Likewise.
28971 (estimate_local_effects): Likewise.
28972 (update_profiling_info): Likewise.
28973 (create_specialized_node): Likewise.
28974 (perhaps_add_new_callers): Likewise.
28975 (decide_about_value): Likewise.
28976 (decide_whether_version_node): Likewise.
28977 (identify_dead_nodes): Likewise.
28978 (ipcp_store_bits_results): Likewise.
28979 * ipa-devirt.c (dump_targets): Likewise.
28980 (ipa_devirt): Likewise.
28981 * ipa-icf.c (sem_item::dump): Likewise.
28982 (sem_function::equals): Likewise.
28983 (sem_variable::equals): Likewise.
28984 (sem_item_optimizer::read_section): Likewise.
28985 (sem_item_optimizer::execute): Likewise.
28986 (congruence_class::dump): Likewise.
28987 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
28988 (dump_inline_summary): Likewise.
28989 (estimate_node_size_and_time): Likewise.
28990 (inline_analyze_function): Likewise.
28991 * ipa-inline-transform.c (inline_call): Likewise.
28992 * ipa-inline.c (report_inline_failed_reason): Likewise.
28993 (want_early_inline_function_p): Likewise.
28994 (edge_badness): Likewise.
28995 (update_edge_key): Likewise.
28996 (inline_small_functions): Likewise.
28997 * ipa-profile.c (ipa_profile): Likewise.
28998 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
28999 (ipa_make_edge_direct_to_target): Likewise.
29000 (remove_described_reference): Likewise.
29001 (ipa_impossible_devirt_target): Likewise.
29002 (propagate_controlled_uses): Likewise.
29003 (ipa_print_node_params): Likewise.
29004 (ipcp_transform_function): Likewise.
29005 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29006 (propagate_pure_const): Likewise.
29007 * ipa-reference.c (generate_summary): Likewise.
29008 (read_write_all_from_decl): Likewise.
29009 (propagate): Likewise.
29010 (ipa_reference_read_optimization_summary): Likewise.
29011 * ipa-utils.c (ipa_merge_profiles): Likewise.
29012 * ipa.c (walk_polymorphic_call_targets): Likewise.
29013 (symbol_table::remove_unreachable_nodes): Likewise.
29014 (ipa_single_use): Likewise.
29015 * passes.c (execute_todo): Likewise.
29016 * predict.c (drop_profile): Likewise.
29017 * symtab.c (symtab_node::get_dump_name): New function.
29018 (symtab_node::dump_name): Likewise.
29019 (symtab_node::dump_asm_name): Likewise.
29020 (symtab_node::dump_references): Likewise.
29021 (symtab_node::dump_referring): Likewise.
29022 (symtab_node::dump_base): Likewise.
29023 (symtab_node::debug_symtab): Likewise.
29024 * tree-sra.c (convert_callers_for_node): Likewise.
29025 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29026 * value-prof.c (init_node_map): Likewise.
29028 2017-05-23 Martin Liska <mliska@suse.cz>
29030 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29031 and symtab_node::debug_symtab to symbol_table::debug.
29032 * cgraphunit.c (analyze_functions): Use the renamed function.
29033 (symbol_table::compile): Likewise.
29034 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29035 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29036 * passes.c (execute_todo): Likewise.
29037 * symtab.c (symbol_table::dump): New function.
29038 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29040 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29042 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29043 that nonconst implies exec.
29045 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29047 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29048 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29049 (inline_edge_summary_vec): Turn into ...
29050 (ipa_call_summaries): ... this one.
29051 (redirect_to_unreachable, edge_set_predicate,
29052 evaluate_properties_for_edge, inline_summary_alloc,
29053 reset_ipa_call_summary, reset_inline_summary,
29054 inline_summary_t::duplicate): Update.
29055 (inline_edge_duplication_hook): Turn to ...
29056 (ipa_call_summary_t::duplicate): ... this one.
29057 (inline_edge_removal_hook): Turn to ...
29058 (ipa_call_summary_t::remove): ... this one.
29059 (dump_inline_edge_summary): Turn to ...
29060 (dump_ipa_call_summary): ... this one.
29061 (estimate_function_body_sizes): Update.
29062 (inline_update_callee_summaries): Update.
29063 (remap_edge_change_prob): Update.
29064 (remap_edge_summaries): Update.
29065 (inline_merge_summary): Update.
29066 (do_estimate_edge_time): Update.
29067 (inline_generate_summary): Update.
29068 (inline_read_section): Update.
29069 (inline_read_summary): Update.
29070 (inline_free_summary): Update.
29071 * ipa-inline.c (can_inline_edge_p): Update.
29072 (compute_inlined_call_time): Update.
29073 (want_inline_small_function_p): Update.
29074 (edge_badness): Update.
29075 (early_inliner): Update.
29076 * ipa-inline.h (inline_edge_summary): Turn to ...
29077 (ipa_call_summary): ... this one.
29078 (ipa_call_summary_t): New class.
29079 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29080 (ipa_call_summaries): New.
29081 (inline_edge_summary): Remove.
29082 (estimate_edge_growth): Update.
29083 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29084 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29085 * ipa-split.c (execute_split_functions): Update.
29086 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29088 2017-05-23 Tom de Vries <tom@codesourcery.com>
29090 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29091 attributes): Document rdrand effective target.
29093 2017-05-23 Tom de Vries <tom@codesourcery.com>
29095 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29096 attributes): Sort alphabetically.
29098 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29100 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29102 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29105 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29106 V2DF/V2DI splat into two separate patterns, one that handles
29107 registers, and the other that only handles memory. Drop support
29108 for splatting from a GPR on ISA 2.07 and then splitting the
29109 splat into direct move and splat.
29110 (vsx_splat_<mode>_reg): Likewise.
29111 (vsx_splat_<mode>_mem): Likewise.
29113 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29115 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29117 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29119 PR middle-end/80809
29120 * omp-low.c (finish_taskreg_remap): New function.
29121 (finish_taskreg_scan): If unit size of ctx->record_type
29122 is non-constant, unshare the size expression and replace
29123 decls in it with possible outer var refs.
29125 PR middle-end/80809
29126 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29127 GOVD_SHARED rather than GOVD_PRIVATE with it.
29128 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29129 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29131 PR middle-end/80853
29132 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29133 as last argument to build_outer_var_ref for pointer bases of array
29134 section reductions.
29136 2017-05-19 Martin Sebor <msebor@redhat.com>
29138 * print-tree.c (print_node): Print DECL_READ_P flag.
29140 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29142 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29143 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29144 * cgraph.c: Likewise.
29145 * cgraphunit.c: Likewise.
29146 * gengtype.c: Likewise.
29147 * ipa-cp.c: Likewise.
29148 * ipa-devirt.c: Likewise.
29149 * ipa-icf.c: Likewise.
29150 * ipa-predicate.c: Likewise.
29151 * ipa-profile.c: Likewise.
29152 * ipa-prop.c: Likewise.
29153 * ipa-split.c: Likewise.
29155 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29156 edge_predicate_pool, dump_inline_hints,
29157 inline_summary::account_size_time, redirect_to_unreachable,
29158 edge_set_predicate, set_hint_predicate,
29159 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29160 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29161 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29162 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29163 ipa_call_summary_t::remove, initialize_growth_caches,
29164 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29165 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29166 mark_modified, unmodified_parm_1, unmodified_parm,
29167 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29168 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29169 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29170 will_be_nonconstant_predicate, record_modified_bb_info,
29171 get_minimal_bb, record_modified, param_change_prob,
29172 phi_result_unknown_predicate, predicate_for_phi_result,
29173 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29174 estimate_function_body_sizes, compute_inline_parameters,
29175 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29176 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29177 inline_update_callee_summaries, remap_edge_change_prob,
29178 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29179 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29180 inline_analyze_function, inline_summary_t::insert,
29181 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29182 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29183 inline_free_summary): Move to ipa-fnsummary.h
29184 (predicate_t): Remove.
29185 * ipa-fnsummary.c: New file.
29186 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29187 (enum inline_hints_vals, inline_hints, agg_position_info,
29188 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29189 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29190 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29191 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29192 inline_read_summary, inline_write_summary, inline_free_summary,
29193 inline_analyze_function, initialize_inline_failed,
29194 inline_merge_summary, inline_update_overall_summary,
29195 compute_inline_parameters): Move to ipa-fnsummary.h
29196 * ipa-fnsummary.h: New file.
29197 * ipa-inline-transform.h: Include ipa-inline.h.
29198 * ipa-inline.c: LIkewise.
29200 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29202 * ipa-inline.c (edge_badness): Use inlined_time instead of
29203 inline_summaries->get.
29205 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29207 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29209 2017-05-22 Nathan Sidwell <nathan@acm.org>
29211 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29212 (fdump-lang): Document 'raw' option.
29213 * dumpfile.h (TDI_tu): Delete.
29214 * dumpfile.c (dump_files): Remove translation-unit.
29215 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29217 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29219 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29220 command option from $(AWK) call.
29221 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29222 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29223 [FORMAT]: Remove handling of variable.
29224 * config/avr/t-multilib: Regenerate.
29226 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29228 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29230 (dump_inline_summary): Do not print self_time.
29231 (estimate_function_body_sizes): Do not set self_time.
29232 (compute_inline_parameters): Likewise.
29233 (inline_read_section, inline_write_summary): Do not stream self_time.
29234 * ipa-inline.h (inline_summary): Drop self_time.
29236 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29238 * ipa-inline-analysis.c (account_size_time): Rename to ...
29239 (inline_summary::account_size_time): ... this one.
29240 (reset_ipa_call_summary): Turn to ...
29241 (ipa_call_summary::reset): ... this one.
29242 (reset_inline_summary): Turn to ...
29243 (inline_summary::reset): ... this one.
29244 (inline_summary_t::remove): Update.
29245 (inline_summary_t::duplicate): Update.
29246 (ipa_call_summary_t::remove): Update.
29247 (dump_inline_summary): Update.
29248 (estimate_function_body_sizes): Update.
29249 (compute_inline_parameters): Update.
29250 (estimate_node_size_and_time): Update.
29251 (inline_merge_summary): Update.
29252 (inline_update_overall_summary): Update.
29253 (inline_read_section): Update.
29254 (inline_write_summary): Update.
29255 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29256 add account_size_time and reset member functions.
29257 (ipa_call_summary): Add reset function.
29258 * ipa-predicate.h (predicate::operator &): Constify.
29260 2017-05-22 Richard Biener <rguenther@suse.de>
29262 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29264 2017-05-19 Jason Merrill <jason@redhat.com>
29266 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29268 2017-05-19 Marek Polacek <polacek@redhat.com>
29271 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29272 TYPE_OVERFLOW_WRAPS checks.
29274 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29276 * tree-core.h (enum omp_clause_default_kind): Add
29277 "OMP_CLAUSE_DEFAULT_PRESENT".
29278 * tree-pretty-print.c (dump_omp_clause): Handle it.
29279 * gimplify.c (enum gimplify_omp_var_data): Add
29280 "GOVD_MAP_FORCE_PRESENT".
29281 (gimplify_adjust_omp_clauses_1): Map it to
29282 "GOMP_MAP_FORCE_PRESENT".
29283 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29285 * gimplify.c (oacc_default_clause): Clarify.
29287 2017-05-19 Nathan Sidwell <nathan@acm.org>
29289 LANG_HOOK_REGISTER_DUMPS
29290 * toplev.c (general_init): Call register dump lang hook.
29291 * doc/invoke.texi: Document -fdump-lang option family.
29292 * dumpfile.c (dump_files): Remove class dump here.
29293 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29294 * dumpfile.h (tree_dump_index): Remove TDI_class.
29295 * langhooks-def.h (lhd_register_dumps): Declare.
29296 (LANG_HOOKS_REGISTER_DUMPS): Define.
29297 (LANG_HOOKS_INITIALIZER): Add it.
29298 * langhooks.c (lhd_register_dumps): Define.
29299 * langhooks.h (struct lang_hooks): Add register_dumps.
29301 2017-05-19 Nathan Sidwell <nathan@acm.org>
29303 * context.h (context::set_passes): New.
29304 * context.c (context::context): Do not create pass manager.
29305 * toplev.c (general_init): Create pass manager here.
29307 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29309 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29310 use this splitter if two add or or instructions would also work for
29311 the constant we want to generate.
29313 2017-05-19 Richard Biener <rguenther@suse.de>
29316 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29317 predicate evaluation.
29319 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29321 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29323 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29326 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29328 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29329 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29330 (inline_edge_summary_vec): Turn into ...
29331 (ipa_call_summaries): ... this one.
29332 (redirect_to_unreachable, edge_set_predicate,
29333 evaluate_properties_for_edge, inline_summary_alloc,
29334 reset_ipa_call_summary, reset_inline_summary,
29335 inline_summary_t::duplicate): Update.
29336 (inline_edge_duplication_hook): Turn to ...
29337 (ipa_call_summary_t::duplicate): ... this one.
29338 (inline_edge_removal_hook): Turn to ...
29339 (ipa_call_summary_t::remove): ... this one.
29340 (dump_inline_edge_summary): Turn to ...
29341 (dump_ipa_call_summary): ... this one.
29342 (estimate_function_body_sizes): Update.
29343 (inline_update_callee_summaries): Update.
29344 (remap_edge_change_prob): Update.
29345 (remap_edge_summaries): Update.
29346 (inline_merge_summary): Update.
29347 (do_estimate_edge_time): Update.
29348 (inline_generate_summary): Update.
29349 (inline_read_section): Update.
29350 (inline_read_summary): Update.
29351 (inline_free_summary): Update.
29352 * ipa-inline.c (can_inline_edge_p): Update.
29353 (compute_inlined_call_time): Update.
29354 (want_inline_small_function_p): Update.
29355 (edge_badness): Update.
29356 (early_inliner): Update.
29357 * ipa-inline.h (inline_edge_summary): Turn to ...
29358 (ipa_call_summary): ... this one.
29359 (ipa_call_summary_t): New class.
29360 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29361 (ipa_call_summaries): New.
29362 (inline_edge_summary): Remove.
29363 (estimate_edge_growth): Update.
29364 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29365 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29366 * ipa-split.c (execute_split_functions): Update.
29367 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29369 2017-05-19 Richard Biener <rguenther@suse.de>
29371 PR middle-end/80764
29372 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29374 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29376 * config/rs6000/rs6000.c (struct machine_function): Add field
29377 fpr_is_wrapped_separately.
29378 (rs6000_get_separate_components): Use 64 components. Handle the
29379 new FPR components.
29380 (rs6000_components_for_bb): Handle the FPR components.
29381 (rs6000_emit_prologue_components): Handle the FPR components.
29382 (rs6000_emit_epilogue_components): Handle the FPR components.
29383 (rs6000_set_handled_components): Handle the FPR components.
29384 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29385 that are already separately shrink-wrapped.
29386 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29387 that are already separately shrink-wrapped.
29389 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29392 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29395 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29396 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29397 cases where the register allocator uses a move and an offsettable
29398 memory operation to/from a FPR register on ISA 2.06/2.07.
29399 (define_peephole2 for Altivec d-form store): Likewise.
29401 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29404 * config/i386/mmx.md (*mov<mode>_internal): Enable
29405 alternatives 11, 12, 13 and 14 also for 32bit targets.
29406 Remove alternatives 15, 16, 17 and 18.
29407 * config/i386/sse.md (vec_concatv2di): Change
29408 alternative (!x, *y) to (x, ?!*Yn).
29410 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29412 * dumpfile.h (enum dump_kind): Remove stray comma.
29414 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29416 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29417 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29418 predicate::num_conditions
29419 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29420 (CHANGED): turn into predicate::changed.
29421 (agg_position_info): Move to ipa-predicate.h
29422 (add_condition, predicate::add_clause, predicate::operator &=,
29423 predicate::or_with, predicate::evaluate, predicate::probability,
29424 dump_condition, dump_clause, predicate::dump,
29425 predicate::remap_after_duplication, predicate::remap_after_inlining,
29426 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29427 (evaluate_conditions_for_known_args): Update.
29428 (set_cond_stmt_execution_predicate): Update.
29429 * ipa-inline.h: Include ipa-predicate.h
29430 (condition, inline_param_summary, conditions, agg_position_info,
29431 predicate): Move to ipa-predicate.h
29432 * ipa-predicate.c: New file.
29433 * ipa-predicate.h: New file.
29435 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29437 * final.c (leaf_function_p): Check we are not in a sequence.
29439 2017-05-18 Martin Liska <mliska@suse.cz>
29441 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29442 * dumpfile.c (dump_register): Use new enum dump_kind.
29443 (get_dump_file_name): Likewise.
29444 (dump_enable_all): Likewise.
29445 (dump_switch_p_1): Likewise.
29446 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29447 * dumpfile.h (enum dump_kind): New enum type.
29448 (struct dump_file_info): Create constructor and
29449 format fields and comments.
29450 * passes.c (pass_manager::register_one_dump_file):
29452 * statistics.c (statistics_early_init): Likewise.
29453 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29454 TDF_TREE with TDF_SLIM.
29455 (gather_memory_references_ref): Likewise.
29457 2017-05-18 Martin Liska <mliska@suse.cz>
29459 * vec.h (struct vnull): Use it.
29461 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29463 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29464 (true_predicate, false_predicate, true_predicate_p,
29465 false_predicate_p): Remove.
29466 (single_cond_predicate, not_inlined_predicate): Turn to member function
29468 (add_condition): Update.
29469 (add_clause): Turn to...
29470 (predicate::add_clause): ... this one; update; allow passing NULL
29472 (and_predicates): Turn to ...
29473 (predicate::operator &=): ... this one.
29474 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29475 (or_predicates): Turn to ...
29476 (predicate::or_with): ... this one.
29477 (evaluate_predicate): Turn to ...
29478 (predicate::evaluate): ... this one.
29479 (predicate_probability): Turn to ...
29480 (predicate::probability): ... this one.
29481 (dump_condition): Update.
29482 (dump_predicate): Turn to ...
29483 (predicate::dump): ... this one.
29484 (account_size_time): Update.
29485 (edge_set_predicate): Update.
29486 (set_hint_predicate): UPdate.
29487 (evaluate_conditions_for_known_args): Update.
29488 (evaluate_properties_for_edge): Update.
29489 (remap_predicate_after_duplication): Turn to...
29490 (predicate::remap_after_duplication): ... this one.
29491 (remap_hint_predicate_after_duplication): Update.
29492 (inline_summary_t::duplicate): UPdate.
29493 (dump_inline_edge_summary): Update.
29494 (dump_inline_summary): Update.
29495 (set_cond_stmt_execution_predicate): Update.
29496 (set_switch_stmt_execution_predicate): Update.
29497 (compute_bb_predicates): Update.
29498 (will_be_nonconstant_expr_predicate): Update.
29499 (will_be_nonconstant_predicate): Update.
29500 (phi_result_unknown_predicate): Update.
29501 (predicate_for_phi_result): Update.
29502 (array_index_predicate): Update.
29503 (estimate_function_body_sizes): Update.
29504 (estimate_node_size_and_time): Update.
29505 (estimate_ipcp_clone_size_and_time): Update.
29506 (remap_predicate): Rename to ...
29507 (predicate::remap_after_inlining): ... this one.
29508 (remap_hint_predicate): Update.
29509 (inline_merge_summary): Update.
29510 (inline_update_overall_summary): Update.
29511 (estimate_size_after_inlining): Update.
29512 (read_predicate): Rename to ...
29513 (predicate::stream_in): ... this one.
29514 (read_inline_edge_summary): Update.
29515 (write_predicate): Rename to ...
29516 (predicate::stream_out): ... this one.
29517 (write_inline_edge_summary): Update.
29518 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29519 (clause_t): Turn to uint32_t
29520 (predicate): Turn to class; implement constructor and operators
29522 (size_time_entry): Update.
29523 (inline_summary): Update.
29524 (inline_edge_summary): Update.
29526 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29528 * fold-const.c (fold_binary_loc): Move transformation...
29529 * match.pd (C - X CMP X): ... here.
29531 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29533 * config/sparc/sparc.c (sparc_option_override): Set function
29534 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29535 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29537 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29539 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29541 2017-05-18 Marek Polacek <polacek@redhat.com>
29544 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29545 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29546 gimple_assign_load_p.
29548 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29550 PR middle-end/80692
29551 * real.c (do_compare): Give decimal_do_compare preference over
29552 comparing just the signs.
29554 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29556 * doc/md.texi (Canonicalization of Instructions): Describe the
29557 canonical form of instructions that inherently set a condition
29560 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29562 PR middle-end/80775
29563 * tree-cfg.c: Move deletion of unreachable case statements to after
29564 the merging of consecutive case labels.
29566 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29568 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29569 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29570 restoring of callee-saved registers.
29572 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29574 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29575 * config/visium/visium.c (single_set_and_flags): Likewise.
29576 * config/visium/visium.md (Substitutions): Likewise.
29578 2017-05-17 Martin Liska <mliska@suse.cz>
29580 * cfg.c: Introduce dump_flags_t type and
29581 use it instead of int type.
29583 * cfghooks.c: Likewise.
29584 * cfghooks.h (struct cfg_hooks): Likewise.
29585 * cfgrtl.c: Likewise.
29586 * cfgrtl.h: Likewise.
29587 * cgraph.c (cgraph_node::get_body): Likewise.
29588 * coretypes.h: Likewise.
29589 * domwalk.c: Likewise.
29590 * domwalk.h: Likewise.
29591 * dumpfile.c (struct dump_option_value_info): Likewise.
29592 (dump_enable_all): Likewise.
29593 (dump_switch_p_1): Likewise.
29594 (opt_info_switch_p): Likewise.
29595 * dumpfile.h (enum tree_dump_index): Likewise.
29596 (struct dump_file_info): Likewise.
29597 * genemit.c: Likewise.
29598 * generic-match-head.c: Likewise.
29599 * gengtype.c (open_base_files): Likewise.
29600 * gimple-pretty-print.c: Likewise.
29601 * gimple-pretty-print.h: Likewise.
29602 * graph.c (print_graph_cfg): Likewise.
29603 * graphite-scop-detection.c (dot_all_sese): Likewise.
29604 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29605 * loop-unroll.c (report_unroll): Likewise.
29606 * passes.c (pass_manager::register_one_dump_file): Likewise.
29607 * print-tree.c: Likewise.
29608 * statistics.c: Likewise.
29609 * tree-cfg.c: Likewise.
29610 * tree-cfg.h: Likewise.
29611 * tree-dfa.c: Likewise.
29612 * tree-dfa.h: Likewise.
29613 * tree-dump.c (dump_function): Likewise.
29614 * tree-dump.h (struct dump_info): Likewise.
29615 * tree-pretty-print.c: Likewise.
29616 * tree-pretty-print.h: Likewise.
29617 * tree-ssa-live.c: Likewise.
29618 * tree-ssa-live.h: Likewise.
29619 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29620 * tree-vect-loop.c: Likewise.
29621 * tree-vect-slp.c: Likewise.
29623 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29624 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29626 PR tree-optimization/80457
29627 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29628 of all arguments to a statement as scalar_to_vec operations.
29629 (vectorizable_call): Adjust call to vect_model_simple_cost for
29631 (vectorizable_conversion): Likewise.
29632 (vectorizable_assignment): Likewise.
29633 (vectorizable_shift): Likewise.
29634 (vectorizable_operation): Likewise.
29635 (vectorizable_comparison): Likewise.
29636 (vect_is_simple_cond): Record the def types for operands.
29637 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29638 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29639 for statement argument count.
29641 2017-05-16 Carl Love <cel@us.ibm.com>
29643 * config/rs6000/rs6000-c: Add support for built-in functions
29644 vector unsigned long long vec_bperm (vector unsigned long long,
29645 vector unsigned char)
29646 vector signed long long vec_mule (vector signed int,
29648 vector unsigned long long vec_mule (vector unsigned int,
29649 vector unsigned int)
29650 vector signed long long vec_mulo (vector signed int,
29652 vector unsigned long long vec_mulo (vector unsigned int,
29653 vector unsigned int)
29654 vector signed char vec_sldw (vector signed char,
29655 vector signed char,
29657 vector unsigned char vec_sldw (vector unsigned char,
29658 vector unsigned char,
29660 vector signed short vec_sldw (vector signed short,
29661 vector signed short,
29663 vector unsigned short vec_sldw (vector unsigned short,
29664 vector unsigned short,
29666 vector signed int vec_sldw (vector signed int,
29669 vector unsigned int vec_sldw (vector unsigned int,
29670 vector unsigned int,
29672 vector signed long long vec_sldw (vector signed long long,
29673 vector signed long long,
29675 vector unsigned long long vec_sldw (vector unsigned long long,
29676 vector unsigned long long,
29678 * config/rs6000/rs6000-c: Add support for built-in functions
29679 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29680 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29681 * doc/extend.texi: Update the built-in documentation for the
29682 new built-in functions.
29684 2017-05-16 Marek Polacek <polacek@redhat.com>
29688 * tree.c (save_expr): Don't fold the expression.
29690 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29692 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29693 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29694 and (?*y,m). Update insn attributes.
29696 2017-05-16 Martin Liska <mliska@suse.cz>
29698 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29699 flags argument of print_gimple_stmt, print_gimple_expr,
29700 print_generic_stmt and print_generic_expr.
29701 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29702 * coretypes.h: Likewise.
29703 * except.c (dump_eh_tree): Likewise.
29704 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29705 * gimple-pretty-print.h: Likewise.
29706 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29707 (backprop::push_to_worklist): Likewise.
29708 (backprop::pop_from_worklist): Likewise.
29709 (backprop::process_use): Likewise.
29710 (backprop::intersect_uses): Likewise.
29711 (note_replacement): Likewise.
29712 * gimple-ssa-store-merging.c
29713 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29714 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29715 (pass_store_merging::execute): Likewise.
29716 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29717 (ssa_base_cand_dump_callback): Likewise.
29718 (dump_incr_vec): Likewise.
29719 (replace_refs): Likewise.
29720 (replace_mult_candidate): Likewise.
29721 (create_add_on_incoming_edge): Likewise.
29722 (create_phi_basis): Likewise.
29723 (insert_initializers): Likewise.
29724 (all_phi_incrs_profitable): Likewise.
29725 (introduce_cast_before_cand): Likewise.
29726 (replace_one_candidate): Likewise.
29727 * gimplify.c (gimplify_expr): Likewise.
29728 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29729 (set_rename): Likewise.
29730 (rename_uses): Likewise.
29731 (copy_loop_phi_nodes): Likewise.
29732 (add_close_phis_to_merge_points): Likewise.
29733 (copy_loop_close_phi_args): Likewise.
29734 (copy_cond_phi_args): Likewise.
29735 (graphite_copy_stmts_from_block): Likewise.
29736 (translate_pending_phi_nodes): Likewise.
29737 * graphite-poly.c (print_pdr): Likewise.
29738 (dump_gbb_cases): Likewise.
29739 (dump_gbb_conditions): Likewise.
29740 (print_scop_params): Likewise.
29741 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29742 (build_cross_bb_scalars_use): Likewise.
29743 (gather_bbs::before_dom_children): Likewise.
29744 * hsa-dump.c (dump_hsa_immed): Likewise.
29745 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29746 (get_replacement_map): Likewise.
29747 * ipa-inline-analysis.c (dump_condition): Likewise.
29748 (estimate_function_body_sizes): Likewise.
29749 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29750 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29751 * ipa-prop.c (ipa_dump_param): Likewise.
29752 (ipa_print_node_jump_functions_for_edge): Likewise.
29753 (ipa_modify_call_arguments): Likewise.
29754 (ipa_modify_expr): Likewise.
29755 (ipa_dump_param_adjustments): Likewise.
29756 (ipa_dump_agg_replacement_values): Likewise.
29757 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29758 * ipa-pure-const.c (check_stmt): Likewise.
29759 (pass_nothrow::execute): Likewise.
29760 * ipa-split.c (execute_split_functions): Likewise.
29761 * omp-offload.c (dump_oacc_loop_part): Likewise.
29762 (dump_oacc_loop): Likewise.
29763 * trans-mem.c (tm_log_emit): Likewise.
29764 (tm_memopt_accumulate_memops): Likewise.
29765 (dump_tm_memopt_set): Likewise.
29766 (dump_tm_memopt_transform): Likewise.
29767 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29768 (print_loop): Likewise.
29769 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29770 (chkp_gather_checks_info): Likewise.
29771 (chkp_get_check_result): Likewise.
29772 (chkp_remove_check_if_pass): Likewise.
29773 (chkp_use_outer_bounds_if_possible): Likewise.
29774 (chkp_reduce_bounds_lifetime): Likewise.
29775 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29776 (chkp_mark_completed_bounds): Likewise.
29777 (chkp_register_incomplete_bounds): Likewise.
29778 (chkp_mark_invalid_bounds): Likewise.
29779 (chkp_maybe_copy_and_register_bounds): Likewise.
29780 (chkp_build_returned_bound): Likewise.
29781 (chkp_get_bound_for_parm): Likewise.
29782 (chkp_build_bndldx): Likewise.
29783 (chkp_get_bounds_by_definition): Likewise.
29784 (chkp_generate_extern_var_bounds): Likewise.
29785 (chkp_get_bounds_for_decl_addr): Likewise.
29786 * tree-chrec.c (chrec_apply): Likewise.
29787 * tree-data-ref.c (dump_data_reference): Likewise.
29788 (dump_subscript): Likewise.
29789 (dump_data_dependence_relation): Likewise.
29790 (analyze_overlapping_iterations): Likewise.
29791 * tree-inline.c (expand_call_inline): Likewise.
29792 (tree_function_versioning): Likewise.
29793 * tree-into-ssa.c (dump_defs_stack): Likewise.
29794 (dump_currdefs): Likewise.
29795 (dump_names_replaced_by): Likewise.
29796 (dump_update_ssa): Likewise.
29797 (update_ssa): Likewise.
29798 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29799 * tree-parloops.c (build_new_reduction): Likewise.
29800 (try_create_reduction_list): Likewise.
29801 (ref_conflicts_with_region): Likewise.
29802 (oacc_entry_exit_ok_1): Likewise.
29803 (oacc_entry_exit_single_gang): Likewise.
29804 * tree-pretty-print.h: Likewise.
29805 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29806 (get_scalar_evolution): Likewise.
29807 (add_to_evolution): Likewise.
29808 (get_loop_exit_condition): Likewise.
29809 (analyze_evolution_in_loop): Likewise.
29810 (analyze_initial_condition): Likewise.
29811 (analyze_scalar_evolution): Likewise.
29812 (instantiate_scev): Likewise.
29813 (number_of_latch_executions): Likewise.
29814 (gather_chrec_stats): Likewise.
29815 (final_value_replacement_loop): Likewise.
29816 (scev_const_prop): Likewise.
29817 * tree-sra.c (dump_access): Likewise.
29818 (disqualify_candidate): Likewise.
29819 (create_access): Likewise.
29820 (reject): Likewise.
29821 (maybe_add_sra_candidate): Likewise.
29822 (create_access_replacement): Likewise.
29823 (analyze_access_subtree): Likewise.
29824 (analyze_all_variable_accesses): Likewise.
29825 (sra_modify_assign): Likewise.
29826 (initialize_constant_pool_replacements): Likewise.
29827 (find_param_candidates): Likewise.
29828 (decide_one_param_reduction): Likewise.
29829 (replace_removed_params_ssa_names): Likewise.
29830 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29831 * tree-ssa-copy.c (dump_copy_of): Likewise.
29832 (copy_prop_visit_cond_stmt): Likewise.
29833 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29834 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29835 (record_equivalences_from_stmt): Likewise.
29836 * tree-ssa-dse.c (compute_trims): Likewise.
29837 (delete_dead_call): Likewise.
29838 (delete_dead_assignment): Likewise.
29839 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29840 (forward_propagate_into_cond): Likewise.
29841 (pass_forwprop::execute): Likewise.
29842 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29843 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29845 (move_computations_worker): Likewise.
29846 (execute_sm): Likewise.
29847 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29848 (remove_exits_and_undefined_stmts): Likewise.
29849 (remove_redundant_iv_tests): Likewise.
29850 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29851 (adjust_iv_update_pos): Likewise.
29852 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29853 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29854 (value_replacement): Likewise.
29855 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29856 * tree-ssa-pre.c (print_pre_expr): Likewise.
29857 (get_representative_for): Likewise.
29858 (create_expression_by_pieces): Likewise.
29859 (insert_into_preds_of_block): Likewise.
29860 (eliminate_insert): Likewise.
29861 (eliminate_dom_walker::before_dom_children): Likewise.
29862 (eliminate): Likewise.
29863 (remove_dead_inserted_code): Likewise.
29864 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
29865 * tree-ssa-reassoc.c (get_rank): Likewise.
29866 (eliminate_duplicate_pair): Likewise.
29867 (eliminate_plus_minus_pair): Likewise.
29868 (eliminate_not_pairs): Likewise.
29869 (undistribute_ops_list): Likewise.
29870 (eliminate_redundant_comparison): Likewise.
29871 (update_range_test): Likewise.
29872 (optimize_range_tests_var_bound): Likewise.
29873 (optimize_vec_cond_expr): Likewise.
29874 (rewrite_expr_tree): Likewise.
29875 (rewrite_expr_tree_parallel): Likewise.
29876 (linearize_expr): Likewise.
29877 (break_up_subtract): Likewise.
29878 (linearize_expr_tree): Likewise.
29879 (attempt_builtin_powi): Likewise.
29880 (attempt_builtin_copysign): Likewise.
29881 (transform_stmt_to_copy): Likewise.
29882 (transform_stmt_to_multiply): Likewise.
29883 (dump_ops_vector): Likewise.
29884 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
29885 (print_scc): Likewise.
29886 (set_ssa_val_to): Likewise.
29887 (visit_reference_op_store): Likewise.
29888 (visit_use): Likewise.
29889 (sccvn_dom_walker::before_dom_children): Likewise.
29890 (run_scc_vn): Likewise.
29891 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
29893 (expr_hash_elt::print): Likewise.
29894 (const_and_copies::pop_to_marker): Likewise.
29895 (const_and_copies::record_const_or_copy_raw): Likewise.
29896 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
29897 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
29898 (dump_predicates): Likewise.
29899 (find_uninit_use): Likewise.
29900 (warn_uninitialized_phi): Likewise.
29901 (pass_late_warn_uninitialized::execute): Likewise.
29902 * tree-ssa.c (verify_vssa): Likewise.
29903 (verify_ssa): Likewise.
29904 (maybe_optimize_var): Likewise.
29905 * tree-vrp.c (dump_value_range): Likewise.
29906 (dump_all_value_ranges): Likewise.
29907 (dump_asserts_for): Likewise.
29908 (register_edge_assert_for_2): Likewise.
29909 (vrp_visit_cond_stmt): Likewise.
29910 (vrp_visit_switch_stmt): Likewise.
29911 (vrp_visit_stmt): Likewise.
29912 (vrp_visit_phi_node): Likewise.
29913 (simplify_cond_using_ranges_1): Likewise.
29914 (fold_predicate_in): Likewise.
29915 (evrp_dom_walker::before_dom_children): Likewise.
29916 (evrp_dom_walker::push_value_range): Likewise.
29917 (evrp_dom_walker::pop_value_range): Likewise.
29918 (execute_early_vrp): Likewise.
29920 2017-05-16 Richard Biener <rguenther@suse.de>
29922 * dwarf2out.c (loc_list_from_tree_1): Do not create
29923 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
29925 2017-05-16 Richard Biener <rguenther@suse.de>
29927 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
29929 (note_variable_value_in_expr): If we resolved the decl ref
29930 do not push to the stack.
29932 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
29934 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
29935 operations in fast-math mode.
29936 (vaddq_f16): Likewise.
29937 (vmul_f16): Likewise.
29938 (vmulq_f16): Likewise.
29939 (vsub_f16): Likewise.
29940 (vsubq_f16): Likewise.
29941 * config/arm/neon.md (add<mode>3): New.
29943 (fma:<VH:mode>3): New. Also remove outdated comment.
29946 2017-05-16 Martin Liska <mliska@suse.cz>
29950 * ipa-devirt.c (warn_types_mismatch): Fix typo.
29951 (odr_types_equivalent_p): Likewise.
29953 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
29955 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
29957 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
29960 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
29961 non-interunit SSE move alternatives with '?'.
29962 (zero-extendsidi peephole2): New peephole to skip intermediate
29963 general register in SSE zero-extend sequence.
29965 2017-05-15 Jeff Law <law@redhat.com>
29967 * reorg.c (relax_delay_slots): Create a new variable to hold
29968 the temporary target rather than clobbering TARGET_LABEL.
29970 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
29971 missing argument to extract_bit_field call.
29972 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
29974 2017-05-15 Martin Liska <mliska@suse.cz>
29977 * gcc.c (process_command): Do not allow empty argument of -o option.
29979 2017-05-15 Renlin Li <renlin.li@arm.com>
29981 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
29982 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
29983 * config/aarch64/constraints.md (Usf): Add long call check.
29984 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
29985 (call_value): Likewise.
29986 (sibcall): Likewise.
29987 (sibcall_value): Likewise.
29989 (call_value_insn): New.
29990 (sibcall_insn): Update rtx pattern.
29991 (sibcall_value_insn): Likewise.
29992 (call_internal): Remove.
29993 (call_value_internal): Likewise.
29994 (sibcall_internal): Likewise.
29995 (sibcall_value_internal): Likewise.
29996 (call_reg): Likewise.
29997 (call_symbol): Likewise.
29998 (call_value_reg): Likewise.
29999 (call_value_symbol): Likewise.
30001 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30004 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30006 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30008 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30009 compatible with CCGOCmode and with CCZmode.
30011 2017-05-14 Martin Sebor <msebor@redhat.com>
30013 PR middle-end/77671
30014 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30015 (gimple_fold_builtin_snprintf): Same.
30016 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30017 (gimple_fold_builtin_snprintf): Same.
30018 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30019 of character types.
30020 (is_call_safe): New function.
30021 (try_substitute_return_value): Call it.
30022 (try_simplify_call): New function.
30023 (pass_sprintf_length::handle_gimple_call): Call it.
30025 2017-05-14 Martin Sebor <msebor@redhat.com>
30027 PR middle-end/80669
30028 * builtins.c (expand_builtin_stpncpy): Simplify.
30030 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30032 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30033 * config/i386/i386.h
30034 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30035 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30036 (struct machine_function): Add new members call_ms2sysv,
30037 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30038 (struct machine_frame_state): New fields sp_realigned and
30039 sp_realigned_offset.
30040 * config/i386/i386.c
30041 (enum xlogue_stub): New enum.
30042 (enum xlogue_stub_sets): New enum.
30043 (class xlogue_layout): New class.
30044 (struct ix86_frame): New fields stack_realign_allocate_offset,
30045 stack_realign_offset and outlined_save_offset. Modify comments to
30046 detail stack layout when using out-of-line stubs.
30047 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30048 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30049 -mcall-ms2sysv-xlogues.
30050 (stub_managed_regs): New static variable.
30051 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30052 registers managed by out-of-line stub.
30053 (disable_call_ms2sysv_xlogues): New function.
30054 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30055 m->call_ms2sysv when appropriate and compute frame layout for
30057 (sp_valid_at, fp_valid_at): New inline functions.
30058 (choose_basereg): New function.
30059 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30061 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30062 Use align parameter of choose_baseaddr to generated aligned SSE movs
30064 (pro_epilogue_adjust_stack): Modify to track
30065 machine_frame_state::sp_realigned.
30066 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30067 (ix86_nsaved_sseregs): Likewise.
30068 (ix86_emit_save_regs): Likewise.
30069 (ix86_emit_save_regs_using_mov): Likewise.
30070 (ix86_emit_save_sse_regs_using_mov): Likewise.
30071 (get_scratch_register_on_entry): Likewise.
30072 (gen_frame_set): New function.
30073 (gen_frame_load): Likewise.
30074 (gen_frame_store): Likewise.
30075 (emit_outlined_ms2sysv_save): Likewise.
30076 (emit_outlined_ms2sysv_restore): Likewise.
30077 (ix86_expand_prologue): Modify stack re-alignment code and call
30078 emit_outlined_ms2sysv_save when appropriate.
30079 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30080 parameter rtx_insn *insn, which allows the function to be used to only
30081 generate the notes.
30082 (ix86_expand_epilogue): Modify validity checks of frame and stack
30083 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30084 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30085 * config/i386/predicates.md
30086 (save_multiple): New predicate.
30087 (restore_multiple): Likewise.
30088 * config/i386/sse.md
30089 (save_multiple<mode>): New pattern.
30090 (save_multiple_realign<mode>): Likewise.
30091 (restore_multiple<mode>): Likewise.
30092 (restore_multiple_and_return<mode>): Likewise.
30093 (restore_multiple_leave_return<mode>): Likewise.
30094 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30096 2017-05-14 Julia Koval <julia.koval@intel.com>
30098 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30099 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30100 (__builtin_ia32_xsetbv): New builtins.
30101 * config/i386/i386.c (ix86_expand_special_args_builtin):
30103 (ix86_expand_builtin): Special expand for new intrinsics.
30104 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30105 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30106 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30108 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30110 * cfganal.c (inverted_post_order_compute): Change argument type
30112 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30113 * df-core.c (rest_of_handle_df_initialize): Adjust.
30114 (rest_of_handle_df_finish): Likewise.
30115 (df_analyze_1): Likewise.
30116 (df_analyze): Likewise.
30117 (loop_inverted_post_order_compute): Change argument to be a vec *.
30118 (df_analyze_loop): Adjust.
30119 (df_get_n_blocks): Likewise.
30120 (df_get_postorder): Likewise.
30121 * df.h (struct df_d): Change field to be a vec.
30122 * lcm.c (compute_laterin): Adjust.
30123 (compute_available): Likewise.
30124 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30125 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30126 * tree-ssa-pre.c (compute_antic): Likewise.
30128 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30130 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30131 (depth_first_search::depth_first_search): Change structure init
30132 function to this constructor.
30133 (depth_first_search::add_bb): Rename function to this member.
30134 (depth_first_search::execute): Likewise.
30135 (flow_dfs_compute_reverse_finish): Adjust.
30137 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30139 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30140 (longest_simple_path): Likewise.
30141 * shrink-wrap.c (spread_components): Likewise.
30142 (disqualify_problematic_components): Likewise.
30143 (emit_common_heads_for_components): Likewise.
30144 (emit_common_tails_for_components): Likewise.
30145 (insert_prologue_epilogue_for_components): Likewise.
30147 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30149 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30152 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30154 * df-core.c (df_set_blocks): Start using auto_bitmap.
30155 (df_compact_blocks): Likewise.
30156 * df-problems.c (df_rd_confluence_n): Likewise.
30157 * df-scan.c (df_insn_rescan_all): Likewise.
30158 (df_process_deferred_rescans): Likewise.
30159 (df_update_entry_block_defs): Likewise.
30160 (df_update_exit_block_uses): Likewise.
30161 (df_entry_block_bitmap_verify): Likewise.
30162 (df_exit_block_bitmap_verify): Likewise.
30163 (df_scan_verify): Likewise.
30164 * lra-constraints.c (lra_constraints): Likewise.
30165 (undo_optional_reloads): Likewise.
30166 (lra_undo_inheritance): Likewise.
30167 * lra-remat.c (calculate_gen_cands): Likewise.
30168 (do_remat): Likewise.
30169 * lra-spills.c (assign_spill_hard_regs): Likewise.
30170 (spill_pseudos): Likewise.
30171 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30172 (bitmap_set_subtract_values): Likewise.
30174 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30176 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30177 management with auto_bitmap.
30178 (fix_inter_tick): Likewise.
30179 (fix_recovery_deps): Likewise.
30180 * ira.c (add_store_equivs): Likewise.
30181 (find_moveable_pseudos): Likewise.
30182 (split_live_ranges_for_shrink_wrap): Likewise.
30183 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30184 (rtx_reuse_manager::seen_def_p): Likewise.
30185 (rtx_reuse_manager::set_seen_def): Likewise.
30186 * print-rtl.h (class rtx_reuse_manager): Likewise.
30188 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30190 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30192 (migrate_btr_def): Likewise.
30193 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30194 * df-core.c (loop_post_order_compute): Likewise.
30195 (loop_inverted_post_order_compute): Likewise.
30196 * hsa-common.h: Likewise.
30197 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30198 * init-regs.c (initialize_uninitialized_regs): Likewise.
30199 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30200 (inline_small_functions): Likewise.
30201 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30202 * ira.c (combine_and_move_insns): Likewise.
30203 (build_insn_chain): Likewise.
30204 * loop-invariant.c (find_invariants): Likewise.
30205 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30206 * predict.c (tree_predict_by_opcode): Likewise.
30207 (predict_paths_leading_to): Likewise.
30208 (predict_paths_leading_to_edge): Likewise.
30209 (estimate_loops_at_level): Likewise.
30210 (estimate_loops): Likewise.
30211 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30212 (spread_components): Likewise.
30213 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30214 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30215 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30216 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30217 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30218 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30219 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30220 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30221 (mark_threaded_blocks): Likewise.
30222 (thread_through_all_blocks): Likewise.
30223 * tree-ssa.c (verify_ssa): Likewise.
30224 (execute_update_addresses_taken): Likewise.
30225 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30227 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30229 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30231 (post_order_compute): Likewise.
30232 (inverted_post_order_compute): Likewise.
30233 (pre_and_rev_post_order_compute_fn): Likewise.
30235 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30237 * genrecog.c (int_set::int_set): Explicitly construct our
30238 auto_vec base class.
30239 * vec.h (auto_vec::auto_vec): New constructor.
30241 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30243 * bitmap.h (class auto_bitmap): New constructor taking
30244 bitmap_obstack * argument.
30246 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30248 * bitmap.h (class auto_bitmap): Change type of m_bits to
30249 bitmap_head, and adjust ctor / dtor and member operators.
30251 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30253 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30254 when returned register mode doesn't match original mode.
30256 2017-05-12 Jeff Law <law@redhat.com>
30257 Jakub Jelinek <jakub@redhat.com>
30259 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30260 we look for cc setter after the compare-elim changes.
30261 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30262 within the vector to match what compare-elim now expects.
30263 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30264 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30266 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30267 after the compare-elim changes.
30268 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30269 the vector to match what compare-elim now expects.
30270 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30271 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30272 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30273 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30274 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30276 * config/visium/visium.c (single_set_and_flags): Fix where
30277 we look for cc setter after the compare-elim changes.
30278 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30279 with the vector to match what compare-elim now expects.
30280 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30281 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30282 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30283 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30284 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30285 (neg<mode>2_insn_set_overflow): Likewise.
30287 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30289 PR middle-end/79794
30290 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30291 maybe_expand_insn call, set ops[0].target. If still set after call,
30292 set alt_rtl. Add extra arg to recursive calls.
30293 (extract_bit_field): Add alt_rtl argument. Pass to
30295 * expmed.h (extract_bit_field): Fix prototype.
30296 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30297 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30298 to extract_bit_field_calls.
30299 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30300 Pass alt_rtl to extract_bit_field calls.
30301 * calls.c (store_unaligned_arguments_into_psuedos)
30302 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30303 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30305 * optabs.h (struct expand_operand): Add target bitfield.
30307 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30309 * compare-elim.c (try_eliminate_compare): Canonicalize
30310 operation with embedded compare to
30311 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30312 (set (reg) (operation)].
30314 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30316 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30319 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30320 cost of adding a carry flag for ADC instruction.
30321 [case MINUS]: Ignore the cost of subtracting a carry flag
30322 for SBB instruction.
30324 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30326 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30328 * config/rs6000/bmiintrin.h: New file.
30329 * config/rs6000/bmi2intrin.h: New file.
30330 * config/rs6000/x86intrin.h: New file.
30332 2017-05-12 Jeff Law <law@redhat.com>
30334 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30337 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30339 PR middle-end/80707
30340 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30342 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30344 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30345 early expansion of vector divide builtins.
30346 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30347 builtins identified as having unsigned arguments.
30349 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30351 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30352 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30353 expansion of vector logical operations (and, andc, or, xor,
30356 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30358 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30359 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30361 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30363 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30364 early GIMPLE expansion of vector multiplies.
30366 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30368 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30369 TARGET_HAVE_MOVT conditional.
30370 (movt splitter): Likewise.
30372 2017-05-12 Richard Biener <rguenther@suse.de>
30374 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30375 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30376 Fold all stmts not inplace.
30378 2017-05-12 Richard Biener <rguenther@suse.de>
30380 PR tree-optimization/80713
30381 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30382 inserted_exprs bit for not removed stmts.
30384 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30386 PR middle-end/69921
30387 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30388 parallelized" attribute for parallelized OpenACC kernels.
30389 * omp-offload.c (execute_oacc_device_lower): Use it.
30391 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30392 Set "oacc kernels" attribute.
30393 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30394 parameter. Adjust all users.
30395 (oacc_fn_attrib_kernels_p): Remove function.
30396 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30397 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30398 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30399 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30400 assert "oacc kernels" attribute is set.
30402 2017-05-11 Carl Love <cel@us.ibm.com>
30404 * config/rs6000/rs6000-c: Add support for built-in functions
30405 vector unsigned char vec_popcnt (vector signed char)
30406 vector unsigned char vec_popcnt (vector unsigned char)
30407 vector unsigned short vec_popcnt (vector signed short)
30408 vector unsigned short vec_popcnt (vector unsigned short)
30409 vector unsigned int vec_popcnt (vector signed int)
30410 vector unsigned int vec_popcnt (vector unsigned int)
30411 vector unsigned long long vec_popcnt (vector signed long long)
30412 vector unsigned long long vec_popcnt (vector unsigned long long)
30413 vector signed long long vec_slo (vector signed long long,
30414 vector signed char)
30415 vector signed long long vec_slo (vector signed long long,
30416 vector unsigned char)
30417 vector unsigned long long vec_slo (vector unsigned long long,
30418 vector signed char)
30419 vector unsigned long long vec_slo (vector unsigned long long,
30420 vector unsigned char)
30421 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30422 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30423 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30424 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30425 * doc/extend.texi: Update the built-in documentation file for the
30426 new built-in functions.
30428 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30430 * attribs.h (sorted_attr_string): Move machine independent
30431 functions for target clone support from the i386 port to common
30432 code. Rename ix86_function_versions to common_function_versions.
30433 Rename make_name to make_unique_name.
30434 (common_function_versions): Likewise.
30435 (make_unique_name): Likewise.
30436 (make_dispatcher_decl): Likewise.
30437 (is_function_default_version): Likewise.
30438 * attribs.c (attr_strcmp): Likewise.
30439 (sorted_attr_string): Likewise.
30440 (common_function_versions): Likewise.
30441 (make_unique_name): Likewise.
30442 (make_dispatcher_decl): Likewise.
30443 (is_function_default_version): Likewise.
30444 * config/i386/i386.c (attr_strcmp): Likewise.
30445 (sorted_attr_string): Likewise.
30446 (ix86_function_versions): Likewise.
30447 (make_name): Likewise.
30448 (make_dispatcher_decl): Likewise.
30449 (is_function_default_version): Likewise.
30450 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30452 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30455 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30456 Account for direct move costs for vec_construct of integer
30459 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30462 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30463 (UNSPEC_STX_ATOMIC): Ditto.
30464 (loaddi_via_sse): New insn.
30465 (storedi_via_sse): Ditto.
30466 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30467 Update corresponding peephole2 patterns.
30468 (atomic_storedi_fpu): Ditto.
30470 2017-05-11 Julia Koval <julia.koval@intel.com>
30472 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30473 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30475 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30476 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30477 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30479 2017-05-11 Nathan Sidwell <nathan@acm.org>
30481 * graphite-poly.c: Include dumpfile.h.
30483 * dumpfle.h (dump_function): Declare here ...
30484 * tree-dump.h (dump_function): ... not here.
30485 * dumpfile.c: #include tree-cfg.h.
30486 (dump_function): Move here from ...
30487 * tree-dump.c (dump_function): ... here.
30488 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30489 * graphite-poly.c: Don't include tree-dump.h.
30490 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30491 * print-tree.c: Likewise.
30492 * stor-layout.c: Likewise.
30493 * tree-nested.c: Likewise.
30495 * dumpfile.c (dump_start): Use TDF_FLAGS.
30496 (dump_enable_all): Fix TDF_KIND check thinko.
30498 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30500 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30501 array entries to represent two legal parameterizations of the
30502 overloaded __builtin_cmpb function, as represented by the
30503 P6_OV_BUILTIN_CMPB constant.
30504 (altivec_resolve_overloaded_builtin): Add special case handling
30505 for the __builtin_cmpb function, as represented by the
30506 P6_OV_BUILTIN_CMPB constant.
30507 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30508 (BU_P6_64BIT_2): New macro.
30509 (BU_P6_OVERLOAD_2): New macro
30510 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30511 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30512 (CMPB): Add overload support to represent both 32-bit and 64-bit
30513 compare-bytes function.
30514 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30515 support for TARGET_CMPB.
30516 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30517 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30518 documentation of the __builtin_cmpb overloaded built-in function.
30520 2017-05-11 Richard Biener <rguenther@suse.de>
30522 PR tree-optimization/80705
30523 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30524 bases are not vectorizable.
30526 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30528 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30529 when counting register pressure.
30531 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30533 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30534 (iv_ca_more_deps): Renamed to ...
30535 (iv_ca_compare_deps): ... this.
30536 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30538 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30540 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30542 (determine_group_iv_costs): ... here.
30543 (find_inv_vars_cb): Record inv var if it's not recorded before.
30545 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30547 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30548 (get_shiftadd_cost): Ditto.
30550 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30552 * tree-ssa-address.c: Include header file.
30553 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30555 (add_to_parts): Refactor.
30556 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30557 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30560 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30562 PR tree-optimization/53090
30563 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30565 (extract_cond_operands): Detect condition with IV on both sides
30566 and return COMP_IV_EXPR_2.
30567 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30568 (rewrite_use_compare): Simplify by removing call to function
30569 extract_cond_operands.
30571 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30573 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30574 (extract_cond_operands): Detect condition comparing against non-
30575 invariant bound and return appropriate enum value.
30576 (find_interesting_uses_cond): Update use of extract_cond_operands.
30577 Handle its return value accordingly.
30578 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30580 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30582 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30583 nonlinear iv_use computation in loop invariant sensitive way.
30585 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30587 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30588 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30590 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30592 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30593 (dump_cand): Support iv_cand.inv_exprs.
30594 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30596 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30599 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30601 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30603 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30604 as local function. Include necessary header files.
30605 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30607 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30609 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30611 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30613 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30614 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30615 RSHIFT_EXPR and BIT_NOT_EXPR.
30617 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30619 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30620 (adjust_setup_cost): New parameter supporting round up adjustment.
30621 (struct address_cost_data): Delete.
30622 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30623 (split_address_cost, ptr_difference_cost): Delete.
30624 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30625 (struct ainc_cost_data): New struct.
30626 (get_address_cost_ainc): New function.
30627 (get_address_cost, get_computation_cost): Reimplement.
30628 (determine_group_iv_cost_address): Record inv_expr for all uses of
30630 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30631 (iv_ca_has_deps): Reimplemented to ...
30632 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30634 (iv_ca_extend): Call iv_ca_more_deps.
30636 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30638 * tree-ssa-address.c (struct mem_address): Move to header file.
30639 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30640 * tree-ssa-address.h (struct mem_address): Move from C file.
30641 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30643 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30645 * tree-affine.h (aff_combination_type): New interface.
30646 (aff_combination_zero_p): Remove static.
30647 (aff_combination_const_p): New interface.
30648 (aff_combination_singleton_var_p): New interfaces.
30650 2017-05-11 Richard Biener <rguenther@suse.de>
30652 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30653 Skip unreachable blocks and destinations.
30654 (eliminate): Move stmt removal and fixup ...
30655 (fini_eliminate): ... here. Skip inserted exprs.
30656 (pass_pre::execute): Move fini_pre after fini_eliminate.
30657 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30658 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30659 PRE to get rid of dead code that has invalid SSA form and
30660 split critical edges again.
30662 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30664 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30666 2017-05-11 Richard Biener <rguenther@suse.de>
30668 * passes.c (execute_function_todo): Verify loops if they are
30669 said to be up-to-date.
30670 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30671 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30673 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30676 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30677 handle calling assemble_external ourself.
30680 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30681 modes with zero size. Enhance comment.
30683 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30685 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30686 built-ins for vec_xl and vec_xst with short and char pointer
30689 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30691 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30692 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30693 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30694 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30695 (_mm_maskz_min_round_ss): New intrinsics.
30696 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30697 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30698 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30699 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30700 (__builtin_ia32_minss_mask_round): New builtins.
30701 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30702 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30703 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30705 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30706 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30708 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30711 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30713 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30714 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30715 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30716 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30717 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30718 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30719 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30720 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30721 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30722 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30723 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30724 (__builtin_ia32_mulss_mask_round): New builtins.
30725 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30726 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30727 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30729 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30730 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30732 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30735 2017-05-10 Julia Koval <julia.koval@intel.com>
30737 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30738 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30739 (_mm256_setr_m128i): New intrinsics.
30741 2017-05-10 Julia Koval <julia.koval@intel.com>
30743 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30744 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30745 (_mm_maskz_rcp14_ss): New intrinsics.
30746 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30747 (__builtin_ia32_rcp14ss_mask): New builtins.
30748 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30750 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30752 PR tree-optimization/51513
30753 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30754 (assert_unreachable_fallthru_edge_p): Use it.
30755 (group_case_labels_stmt): Likewise.
30756 * tree-cfg.h: Prototype it.
30757 * stmt.c: Include cfghooks.h and tree-cfg.h.
30758 (emit_case_dispatch_table) <gap_label>: New local variable.
30759 Use it to fill dispatch table gaps.
30760 Test for default_label before updating probabilities.
30761 (expand_case) <default_label>: Remove unneeded initialization.
30762 Test for unreachable default case statement and remove its edge.
30763 Set default_label accordingly.
30764 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30766 2017-05-10 Carl Love <cel@us.ibm.com>
30768 * config/rs6000/rs6000-c: Add support for built-in functions
30769 vector signed char vec_neg (vector signed char)
30770 vector signed short int vec_neg (vector short int)
30771 vector signed int vec_neg (vector signed int)
30772 vector signed long long vec_neg (vector signed long long)
30773 vector float vec_neg (vector float)
30774 vector double vec_neg (vector double)
30775 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30777 * config/rs6000/altivec.h: Add define for vec_neg
30778 * doc/extend.texi: Update the built-in documentation for the
30779 new built-in functions.
30781 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30783 PR tree-optimization/77644
30784 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30786 2017-05-10 Nathan Sidwell <nathan@acm.org>
30788 * dumpfile.h (TDI_lang_all): New.
30789 (TDF_KIND): New. Renumber others
30790 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30792 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30794 (get_dump_file_name): Adjust suffix generation.
30795 (dump_enable_all): Use TDF_KIND.
30796 * doc/invoke.texi (-fdump-lang-all): Document.
30798 * dumpfile.h: Tabify.
30800 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30803 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30804 Move member access before delete.
30806 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30808 * tree-inline.c (expand_call_inline): Split block at stmt
30811 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30814 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30815 are now unused after splitting mov{sf,sd}_hardfloat.
30816 (f32_lr2): Likewise.
30817 (f32_lm): Likewise.
30818 (f32_lm2): Likewise.
30819 (f32_li): Likewise.
30820 (f32_li2): Likewise.
30821 (f32_lv): Likewise.
30822 (f32_sr): Likewise.
30823 (f32_sr2): Likewise.
30824 (f32_sm): Likewise.
30825 (f32_sm2): Likewise.
30826 (f32_si): Likewise.
30827 (f32_si2): Likewise.
30828 (f32_sv): Likewise.
30829 (f32_dm): Likewise.
30830 (f32_vsx): Likewise.
30831 (f32_av): Likewise.
30832 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30833 For movsf, order stores so the VSX stores occur before the GPR
30834 store which encourages the register allocator to use a traditional
30835 FPR instead of a GPR. For movsd, order the stores so that the GPR
30836 store comes before the VSX stores to allow the power6 to work.
30837 This is due to the power6 not having a 32-bit integer store
30838 instruction from a FPR.
30839 (movsf_hardfloat): Likewise.
30840 (movsd_hardfloat): Likewise.
30842 2017-05-09 Martin Sebor <msebor@redhat.com>
30844 PR translation/80280
30845 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30848 PR translation/80280
30849 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30850 data member added in r247778.
30851 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30853 2017-05-09 Nathan Sidwell <nathan@acm.org>
30855 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30857 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
30860 2017-05-09 Marek Polacek <polacek@redhat.com>
30862 * doc/invoke.texi: Fix typo.
30864 2017-05-09 Richard Biener <rguenther@suse.de>
30866 * tree-vrp.c (vrp_val_is_max): Adjust comment.
30867 (vrp_val_is_min): Likewise.
30868 (set_value_range_to_value): Likewise.
30869 (set_value_range_to_nonnegative): Likewise.
30870 (gimple_assign_nonzero_p): Likewise.
30871 (gimple_stmt_nonzero_p): Likewise.
30872 (vrp_int_const_binop): Likewise. Remove unreachable case.
30873 (adjust_range_with_scev): Adjust comments.
30874 (compare_range_with_value): Likewise.
30875 (extract_range_from_phi_node): Likewise.
30876 (test_for_singularity): Likewise.
30878 2017-05-09 Richard Biener <rguenther@suse.de>
30880 * tree-vrp.c (get_single_symbol): Add assert that we don't
30881 get overflowed constants as invariant part.
30882 (compare_values_warnv): Add comment before the TREE_NO_WARNING
30883 checks. Use wi::cmp instead of recursing for integer constants.
30884 (compare_values): Just ignore whether we assumed undefined
30885 overflow instead of failing the compare.
30886 (extract_range_for_var_from_comparison_expr): Add comment before the
30887 TREE_NO_WARNING sets.
30888 (test_for_singularity): Likewise.
30889 (extract_range_from_comparison): Do not disable optimization
30890 when we assumed undefined overflow.
30891 (extract_range_basic): Remove init of unused var.
30893 2017-05-09 Richard Biener <rguenther@suse.de>
30895 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
30896 (extract_range_from_multiplicative_op_1): Adjust.
30897 (extract_range_from_binary_expr_1): Use int_const_binop.
30899 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
30902 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
30903 rs6000_store_data_bypass_p in seven define_bypass directives and
30904 in several comments.
30905 * config/rs6000/rs6000-protos.h: Add prototype for
30906 rs6000_store_data_bypass_p function.
30907 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
30908 function implements slightly different (rs6000-specific) semantics
30909 than store_data_bypass_p, returning false rather than aborting
30910 with assertion error when arguments do not satisfy the
30911 requirements of store data bypass.
30912 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
30913 rs6000_store_data_bypass_p.
30915 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
30917 * config/xtensa/xtensa-protos.h
30918 (xtensa_initial_elimination_offset): New declaration.
30919 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30920 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
30921 macro definition, add case for FRAME_POINTER_REGNUM when
30922 FRAME_GROWS_DOWNWARD.
30923 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
30924 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
30925 xtensa_initial_elimination_offset.
30927 2017-05-08 Nathan Sidwell <nathan@acm.org>
30929 * doc/invoke.texi: Alphabetize -fdump options.
30931 2017-05-08 Martin Sebor <msebor@redhat.com>
30933 PR translation/80280
30934 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
30936 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
30938 * target.def (compute_frame_layout): New optional target hook.
30939 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
30940 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
30941 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
30943 * reload1.c (verify_initial_elim_offsets): Likewise.
30944 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
30945 (use_simple_return_p): Call arm_compute_frame_layout if needed.
30946 (arm_get_frame_offsets): Split up into this ...
30947 (arm_compute_frame_layout): ... and this function.
30949 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
30951 * config/aarch64/constraints.md (Usa): New constraint.
30952 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
30954 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
30956 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
30957 with_multilib_list after it has been checked.
30959 2017-05-08 Richard Biener <rguenther@suse.de>
30961 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
30962 (bitmap_set_subtract_values): Likewise.
30964 2017-05-08 Richard Biener <rguenther@suse.de>
30966 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
30967 (gimple_assign_nonzero): ... this and remove strict_overflow_p
30969 (gimple_stmt_nonzero_warnv_p): Rename to ...
30970 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
30972 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
30973 (extract_range_basic): Adjust, do not disable propagation on
30974 strict overflow sensitive simplification.
30975 (vrp_visit_cond_stmt): Likewise.
30977 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
30979 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
30980 body size unconditionally.
30982 2017-05-07 Jeff Law <law@redhat.com>
30985 2017-05-06 Jeff Law <law@redhat.com>
30986 PR tree-optimization/78496
30987 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30990 PR tree-optimization/78496
30991 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30992 (simplify_stmt_using_ranges): Call it.
30993 (vrp_dom_walker::before_dom_children): Extract equivalences
30994 from an ASSERT_EXPR with an equality comparison against a
30997 2017-05-06 Jeff Law <law@redhat.com>
30999 PR tree-optimization/78496
31000 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31003 PR tree-optimization/78496
31004 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31005 (simplify_stmt_using_ranges): Call it.
31006 (vrp_dom_walker::before_dom_children): Extract equivalences
31007 from an ASSERT_EXPR with an equality comparison against a
31010 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31012 * lra-constraints.c (lra_copy_reg_equiv): New function.
31013 (split_reg): Use it to copy equivalence information from the
31014 original register to the spill register.
31016 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31018 PR rtl-optimization/75964
31019 * simplify-rtx.c (simplify_const_relational_operation): Remove
31020 invalid handling of comparisons of integer ABS.
31022 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31024 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31025 initialize to zero.
31026 (init_regs): Remove declaration.
31027 (function_arg_advance_32): Initialize error_p as boolean variable.
31029 2017-05-05 Nathan Sidwell <nathan@acm.org>
31031 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31032 lines. Use for (;;).
31034 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31036 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31037 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31038 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31039 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31040 VF=2 that require versioning.
31042 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31044 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31047 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31049 * diagnostic.h (diagnostic_override_option_index): Convert from
31050 macro to inline function.
31052 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31054 * diagnostic.c (last_module_changed_p): New function.
31055 (set_last_module): New function.
31056 (diagnostic_report_current_module): Convert macro usage to
31057 the above functions.
31058 * diagnostic.h (diagnostic_context::last_module): Strengthen
31059 from const line_map * to const line_map_ordinary *.
31060 (diagnostic_last_module_changed): Delete macro.
31061 (diagnostic_set_last_module): Delete macro.
31063 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31065 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31066 with diagnostic_report_diagnostic.
31067 (diagnostic_n_impl_richloc): Likewise.
31068 * diagnostic.h (report_diagnostic): Delete macro.
31069 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31070 with diagnostic_report_diagnostic.
31071 * substring-locations.c (format_warning_va): Likewise.
31073 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31075 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31076 save/restor of format_spec. Move option-printing code to...
31077 (print_option_information): ...this new function, and
31078 reimplement by simply printing to the pretty_printer,
31079 rather than appending to the format string.
31081 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31083 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31084 handling logic into...
31085 (update_effective_level_from_pragmas): ...this new function.
31087 2017-05-04 Andrew Waterman <andrew@sifive.com>
31089 * config/riscv/riscv.opt (mstrict-align): New option.
31090 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31091 (SLOW_UNALIGNED_ACCESS): Define.
31092 (riscv_slow_unaligned_access): Declare.
31093 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31095 (riscv_slow_unaligned_access): New variable.
31096 (rocket_tune_info): Set slow_unaligned_access to true.
31097 (optimize_size_tune_info): Set slow_unaligned_access to false.
31098 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31099 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31100 (riscv_option_override): Set riscv_slow_unaligned_access.
31101 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31103 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31105 * config/riscv/riscv.md: Unify indentation.
31107 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31112 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31114 (extendsi<mode>2): Add support for doing sign extension via
31115 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31116 don't have ISA 3.0 instructions.
31117 (extendsi<mode>2 splitter): Likewise.
31118 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31119 generate the normal insns since SImode can now go in vector
31120 registers. Disallow the special UNSPECs needed for previous
31121 machines to hide SImode being used. Add new insns
31122 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31123 (fix_trunc<mode>si2_stfiwx): Likewise.
31124 (fix_trunc<mode>si2_internal): Likewise.
31125 (fixuns_trunc<mode>si2): Likewise.
31126 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31127 (fctiw<u>z_<mode>_smallint): Likewise.
31128 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31129 of floating point to 32-bit integer from doing a direct move to
31130 the GPR registers to do a store.
31131 (fctiwz_<mode>): Break long line.
31133 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31135 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31136 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31137 (addr_list, addr_offset_valid_p): New.
31138 (split_address_groups): Check offset validity with above function.
31139 (gt-tree-ssa-loop-ivopts.h): Include header file.
31141 2017-05-05 Nathan Sidwell <nathan@acm.org>
31143 * config.gcc (arm*-*-*): Add missing 'fi'.
31145 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31147 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31148 included in -fopt-info does not matter.
31149 * doc/optinfo.texi (-fopt-info): Fix description of default
31150 behavour. Explicitly say order of options included in -fopt-info
31153 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31155 * config.gcc: Allow combinations of aprofile and rmprofile values for
31156 --with-multilib-list.
31157 * config/arm/t-multilib: New file.
31158 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31159 variables. Remove setting of ISA and floating-point ABI in
31160 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31161 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31162 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31163 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31165 * config/arm/t-rmprofile: Likewise except for the matches changes.
31166 * doc/install.texi (--with-multilib-list): Document the combination of
31167 aprofile and rmprofile values and warn about pitfalls in doing that.
31169 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31171 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31172 (movdi_aarch64): Likewise.
31174 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31176 PR tree-optimization/80632
31177 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31179 (build_arrays): Initialize it for virtual phis.
31180 (fix_phi_nodes): Use it for virtual phis.
31182 PR tree-optimization/80558
31183 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31184 [x, y] op z into [x op, y op z] for op & or | if conditions
31187 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31188 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31191 * config/arm/arm.md (use_literal_pool): Remove.
31192 (64-bit immediate split): No longer takes cost into consideration
31193 if arm_disable_literal_pool is enabled.
31194 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31195 used when arm_disable_literal_pool is enabled.
31196 (arm_max_const_double_inline_cost): Remove use of
31197 arm_disable_literal_pool.
31198 (push_minipool_fix): Add assert.
31199 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31200 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31201 (no_literal_pool_sf_immediate): New.
31203 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31205 PR tree-optimization/80613
31206 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31207 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31209 2017-05-05 Richard Biener <rguenther@suse.de>
31211 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31213 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31215 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31216 of this flag from insn conditions due to removal from r247495.
31218 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31220 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31222 (arm_early_store_addr_dep_ptr): Likewise.
31223 * config/arm/aarch-common-protos.h
31224 (arm_early_load_addr_dep_ptr): Add prototype.
31225 (arm_early_store_addr_dep_ptr): Likewise.
31226 * config/arm/cortex-a53.md: Add new bypasses.
31228 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31230 * tree.c (next_type_uid): Change type to unsigned.
31231 (type_hash_canon): Decrement back next_type_uid if
31232 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31233 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31236 2017-05-04 Martin Sebor <msebor@redhat.com>
31238 * builtins.c: Fix a trivial typo in a comment.
31240 PR middle-end/79234
31241 * builtins.c (check_sizes): Adjust to handle reading past the end.
31242 Avoid printing excessive upper bound of ranges. Use %E to print
31243 tree nodes instead of converting them to %wu.
31244 (expand_builtin_memchr): New function.
31245 (compute_dest_size): Rename...
31246 (compute_objsize): ...to this.
31247 (expand_builtin_memcpy): Adjust.
31248 (expand_builtin_mempcpy): Adjust.
31249 (expand_builtin_strcat): Adjust.
31250 (expand_builtin_strcpy): Adjust.
31251 (check_strncat_sizes): Adjust.
31252 (expand_builtin_strncat): Adjust.
31253 (expand_builtin_strncpy): Adjust and simplify.
31254 (expand_builtin_memset): Adjust.
31255 (expand_builtin_bzero): Adjust.
31256 (expand_builtin_memcmp): Adjust.
31257 (expand_builtin): Handle memcmp.
31258 (maybe_emit_chk_warning): Check strncat just once.
31260 2017-05-04 Martin Sebor <msebor@redhat.com>
31262 PR preprocessor/79214
31263 PR middle-end/79222
31264 PR middle-end/79223
31265 * builtins.c (check_sizes): Add inlining context and issue
31266 warnings even when -Wno-system-headers is set.
31267 (check_strncat_sizes): Same.
31268 (expand_builtin_strncat): Same.
31269 (expand_builtin_memmove): New function.
31270 (expand_builtin_stpncpy): Same.
31271 (expand_builtin): Handle memmove and stpncpy.
31273 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31275 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31276 which is not used any more.
31278 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31280 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31282 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31284 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31285 (cortexa53_tunings): Likewise.
31286 (cortexa57_tunings): Likewise.
31287 (cortexa72_tunings): Likewise.
31288 (cortexa73_tunings): Likewise.
31290 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31292 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31293 Set loop alignment to 8.
31295 2017-05-04 Martin Sebor <msebor@redhat.com>
31297 PR translation/80280
31298 * builtins.c (expand_builtin_object_size): Add missing quoting to
31299 %D and like directives.
31300 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31301 (hsa_type_for_tree_type): Same.
31302 (verify_function_arguments): Same.
31303 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31304 * varasm.c (get_section): Same.
31307 2017-05-04 Martin Sebor <msebor@redhat.com>
31309 PR translation/80280
31310 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31312 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31314 * config/aarch64/aarch64.c (generic_addrcost_table):
31315 Change HI/TI mode setting.
31317 2017-05-04 Martin Jambor <mjambor@suse.cz>
31319 PR tree-optimization/80622
31320 * tree-sra.c (comes_initialized_p): New function.
31321 (build_accesses_from_assign): Only set write lazily when
31322 comes_initialized_p is false.
31323 (analyze_access_subtree): Use comes_initialized_p.
31324 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31325 instead of testing for PARM_DECL.
31327 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31329 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31330 constraint on operand 0 to allow more general addressing modes.
31331 Adjust output template.
31332 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31334 * config/aarch64/aarch64-protos.h
31335 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31336 * config/aarch64/constraints.md (Dp): New address constraint.
31337 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31340 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31342 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31343 update use of estimate_ipcp_clone_size_and_time.
31344 (estimate_local_effects): Update use of
31345 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31346 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31347 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31348 Return nonspecialized time.
31350 2017-05-04 Richard Biener <rguenther@suse.de>
31352 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31353 for the last VUSE which def dominates the PHI. Directly call
31355 (get_continuation_for_phi_1): Remove.
31357 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31359 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31360 to explain the use of truncating division. Cap the number of
31361 iterations to the maximum given by nb_iterations_upper_bound,
31364 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31366 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31367 * configure: Regenerate.
31368 * config.in: Regenerate.
31369 * config/i386/driver-mingw32.c: new file.
31370 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31371 * config.host: Link driver-mingw32.o on MinGW host.
31372 * doc/install.texi: Document new --enable-mingw-wildcard configure
31375 2017-05-04 Marek Polacek <polacek@redhat.com>
31377 PR tree-optimization/80612
31378 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31380 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31381 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31383 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31384 (movt splitter): Likewise.
31385 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31386 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31387 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31388 block for Thumb-1 with MOVT.
31389 (thumb2_legitimate_address_p): Move code block ...
31390 (can_avoid_literal_pool_for_label_p): ... into this new function.
31391 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31393 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31394 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31395 "M-profile targets with the MOVT instruction".
31397 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31399 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31400 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31401 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31403 2017-05-04 Martin Liska <mliska@suse.cz>
31405 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31406 variable cond_code.
31408 2017-05-04 Richard Biener <rguenther@suse.de>
31410 * tree.c (array_at_struct_end_p): Handle arrays at struct
31411 end with flexarrays more conservatively. Refactor and treat
31412 arrays of arrays or aggregates more strict. Fix
31413 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31414 * tree.c (array_at_struct_end_p): Adjust prototype.
31415 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31416 * gimple-fold.c (get_range_strlen): Likewise.
31417 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31419 2017-05-04 Richard Biener <rguenther@suse.de>
31421 PR tree-optimization/31130
31422 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31424 (supports_overflow_infinity): Likewise.
31425 (is_negative_overflow_infinity): Likewise.
31426 (is_positive_overflow_infinity): Likewise.
31427 (is_overflow_infinity): Likewise.
31428 (stmt_overflow_infinity): Likewise.
31429 (overflow_infinity_range_p): Likewise.
31430 (usable_range_p): Remove as always returning true.
31431 (make_overflow_infinity): Remove.
31432 (negative_overflow_infinity): Likewise.
31433 (positive_overflow_infinity): Likewise.
31434 (avoid_overflow_infinity): Likewise.
31435 (set_value_range): Adjust accordingly.
31436 (set_value_range_to_nonnegative): Likewise, remove now unused
31437 overflow_infinity arg.
31438 (vrp_operand_equal_p): Adjust.
31439 (update_value_range): Likewise.
31440 (range_int_cst_singleton_p): Likewise.
31441 (operand_less_p): Likewise.
31442 (compare_values_warnv): Likewise.
31443 (extract_range_for_var_from_comparison_expr): Likewise.
31444 (vrp_int_const_binop): Likewise.
31445 (zero_nonzero_bits_from_vr): Likewise.
31446 (extract_range_from_multiplicative_op_1): Likewise.
31447 (extract_range_from_binary_expr_1): Likewise.
31448 (extract_range_from_unary_expr): Likewise.
31449 (extract_range_from_comparison): Likewise.
31450 (extract_range_basic): Likewise.
31451 (adjust_range_with_scev): Likewise.
31452 (compare_ranges): Likewise.
31453 (compare_range_with_value): Likewise.
31454 (dump_value_range): Likewise.
31455 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31457 (simplify_cond_using_ranges): Adjust.
31459 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31461 * brig-builtins.def: Added a builtin for class_f64.
31462 * builtin-types.def: Added a builtin type needed by class_f64.
31464 2017-05-03 Jason Merrill <jason@redhat.com>
31466 * timevar.def: Add TV_CONSTEXPR.
31468 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31470 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31472 2017-05-03 Martin Jambor <mjambor@suse.cz>
31474 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31475 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31476 * ipa-cp.c (ipcp_propagate_stage): Do not call
31477 ipa_update_after_lto_read.
31478 * ipa-inline.c (ipa_inline): Likewise.
31480 2017-05-03 Martin Jambor <mjambor@suse.cz>
31482 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31483 tag. Added a default constructor and a destructor.
31484 (ipa_edge_args_sum_t): New class;
31485 (ipa_edge_args_sum): Declare.
31486 (ipa_edge_args_vector): Remove declaration.
31487 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31488 (ipa_free_edge_args_substructures): Remove declaration.
31489 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31490 (ipa_edge_args_info_available_for_edge_p): Likewise.
31491 * ipa-prop.c (ipa_edge_args_vector): Removed.
31492 (edge_removal_hook_holder): Likewise.
31493 (edge_duplication_hook_holder): Likewise.
31494 (ipa_edge_args_sum): New variable.
31495 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31496 ipa_edge_args_vector.
31497 (ipa_free_edge_args_substructures): Likewise.
31498 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31499 ipa_edge_args_vector.
31500 (ipa_edge_removal_hook): Turned into method
31501 ipa_edge_args_sum_t::remove.
31502 (ipa_edge_duplication_hook): Turned into method
31503 ipa_edge_args_sum_t::duplicate.
31504 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31505 registering edge hooks.
31506 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31507 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31508 ipa_edge_args_sum instead of ipa_edge_args_vector.
31509 * ipa-profile.c (ipa_profile): Likewise.
31511 2017-05-03 Martin Jambor <mjambor@suse.cz>
31513 * symbol-summary.h (function_summary): New method exists.
31514 (function_summary::symtab_removal): Deallocate through release.
31515 (call_summary): New class.
31516 (gt_ggc_mx): New overload.
31517 (gt_pch_nx): Likewise.
31518 (gt_pch_nx): Likewise.
31520 2017-05-03 Jeff Law <law@redhat.com>
31522 PR tree-optimization/78496
31523 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31524 from simplify_cond_using_ranges. Split off code to walk
31525 backwards through casts into ...
31526 (simplify_cond_using_ranges_2): New function.
31527 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31528 (execute_vrp): After identifying jump threads, call
31529 simplify_cond_using_ranges_2.
31531 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31534 * ipa-inline.h (inline_summary): Add ctor.
31535 (create_ggc): Do not use ggc_cleared_alloc.
31537 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31538 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31540 * gcc.c (handle_braces): Support escaping in switch matching
31542 * doc/invoke.texi (Spec Files): Document it.
31543 Remove superfluous @code markup in items.
31545 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31547 * diagnostic-show-locus.c (struct column_range): New struct.
31548 (get_affected_columns): New function.
31549 (get_printed_columns): New function.
31550 (struct correction): New struct.
31551 (correction::ensure_capacity): New function.
31552 (correction::ensure_terminated): New function.
31553 (struct line_corrections): New struct.
31554 (line_corrections::~line_corrections): New dtor.
31555 (line_corrections::add_hint): New function.
31556 (layout::print_trailing_fixits): Reimplement in terms of the new
31558 (selftest::test_overlapped_fixit_printing): New function.
31559 (selftest::diagnostic_show_locus_c_tests): Call it.
31561 2017-05-03 Nathan Sidwell <nathan@acm.org>
31563 Canonicalize canonical type hashing
31564 * tree.h (type_hash_canon_hash): Declare.
31565 * tree.c (type_hash_list, attribute_hash_list): Move into
31566 type_hash_canon_hash.
31567 (build_type_attribute_qual_variant): Break out hash code calc into
31568 type_hash_canon_hash.
31569 (type_hash_canon_hash): New. Generic type hash computation.
31570 (build_range_type_1, build_array_type_1, build_function_type,
31571 build_method_type_directly, build_offset_type, build_complex_type,
31572 make_vector_type): Call it.
31574 2017-05-03 Richard Biener <rguenther@suse.de>
31576 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31577 When all DRs have unknown misaligned do not always peel
31578 when there is a store but apply the same costing model as if
31579 there were only loads.
31581 2017-05-03 Richard Biener <rguenther@suse.de>
31584 PR tree-optimization/80492
31585 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31586 compare_base_decls returning dont-know properly.
31588 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31590 * config/arm/iterators.md (CCSI): New mode iterator.
31591 (arch): New mode attribute.
31592 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31593 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31594 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31595 code iterator for success result mode.
31596 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31597 the corresponding new insn generators.
31599 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31602 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31603 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31605 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31607 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31608 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31609 (DDR_A): Wrap DDR argument in brackets.
31610 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31611 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31612 (DDR_REVERSED_P): Likewise.
31614 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31616 PR tree-optimization/79472
31617 * tree-switch-conversion.c (struct switch_conv_info): Add
31618 contiguous_range and default_case_nonstandard fields.
31619 (collect_switch_conv_info): Compute contiguous_range and
31620 default_case_nonstandard fields, don't clear final_bb if
31621 contiguous_range and only the default case doesn't have the required
31623 (check_all_empty_except_final): Set default_case_nonstandard instead
31624 of failing if contiguous_range and the default case doesn't have empty
31626 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31627 and only the default case doesn't have the required constants. Skip
31629 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31630 if default_case_nonstandard.
31631 (build_constructors): Build constant 1 just once. Assert that default
31632 values aren't inserted in between cases if contiguous_range. Skip
31634 (build_arrays): Skip virtual phis.
31635 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31636 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31637 Handle virtual phis.
31638 (gen_inbound_check): Handle default_case_nonstandard case.
31639 (process_switch): Adjust check_final_bb caller. Call
31640 gather_default_values with the first non-default case instead of
31641 default case if default_case_nonstandard.
31643 2017-05-02 Nathan Sidwell <nathan@acm.org>
31645 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31646 check. Fix formatting.
31648 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31650 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31651 errors when comparing specialized and unspecialized times.
31653 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31655 * diagnostic-show-locus.c
31656 (layout::should_print_annotation_line_p): Make private.
31657 (layout::print_annotation_line): Make private.
31658 (layout::annotation_line_showed_range_p): Make private.
31659 (layout::show_ruler): Make private.
31660 (layout::print_source_line): Make private. Pass in line and
31661 line_width, rather than calling location_get_source_line. Drop
31663 (layout::print_leading_fixits): New method.
31664 (layout::print_any_fixits): Rename to...
31665 (layout::print_trailing_fixits): ...this, and make private.
31666 Don't print newline fixits.
31667 (diagnostic_show_locus): Move logic for printing one row into...
31668 (layout::print_line): ...this new function. Move the
31669 location_get_source_line call and error-handling from
31670 print_source_line to here. Call print_leading_fixits, and rename
31671 print_any_fixits to print_trailing_fixits.
31672 (selftest::test_fixit_insert_containing_newline): Update now that
31673 newlines are partially supported.
31674 (selftest::test_fixit_insert_containing_newline_2): New test.
31675 (selftest::test_fixit_replace_containing_newline): Update comments.
31676 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31677 * edit-context.c (class added_line): New class.
31678 (class edited_line): Describe newline handling in comment.
31679 (edited_line::actually_edited_p): New method.
31680 (edited_line::print_content): Delete redundant decl.
31681 (edited_line::m_predecessors): New field.
31682 (edited_file::print_content): Call edited_line::print_content.
31683 (edited_file::print_diff): Update to support newlines.
31684 (edited_file::print_diff_hunk): Likewise.
31685 (edited_file::print_run_of_changed_lines): New function.
31686 (edited_file::print_diff_line): Convert to...
31687 (print_diff_line): ...this.
31688 (edited_file::get_effective_line_count): New function.
31689 (edited_line::edited_line): Initialize new field m_predecessors.
31690 (edited_line::~edited_line): Clean up m_predecessors.
31691 (edited_line::apply_fixit): Handle newlines.
31692 (edited_line::get_effective_line_count): New function.
31693 (edited_line::print_content): New function.
31694 (edited_line::print_diff_lines): New function.
31695 (selftest::test_applying_fixits_insert_containing_newline): New
31697 (selftest::test_applying_fixits_replace_containing_newline): New
31699 (selftest::insert_line): New function.
31700 (selftest::test_applying_fixits_multiple_lines): Add example of
31702 (selftest::edit_context_c_tests): Call the new tests.
31704 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31706 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31707 parameter cand. Update dump information.
31708 (get_computation_cost): Update uses.
31710 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31712 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31713 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31714 (get_computation_at, rewrite_use_address): Update use of
31715 get_computation_aff.
31717 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31719 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31720 (get_computation): Delete.
31721 (get_computation_cost): Implement like get_computation_cost_at.
31722 Use get_computation_at.
31723 (get_computation_cost_at): Delete.
31724 (rewrite_use_nonlinear_expr): Use get_computation_at.
31725 (rewrite_use_compare, remove_unused_ivs): Ditto.
31727 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31729 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31731 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31733 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31734 (ivopts_global_cost_for_size): Rename parameter and update uses.
31735 (iv_ca_recount_cost): Update uses.
31736 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31737 candidates seperately in n_invs and n_cands.
31738 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31740 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31742 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31743 (find_inv_vars_cb): New.
31744 (find_depends): Renamed to ...
31745 (find_inv_vars): ... this.
31746 (add_candidate_1, force_var_cost): Call find_inv_vars.
31747 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31749 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31751 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31752 inv_vars. Add inv_exprs.
31753 (struct iv_cand): Rename depends_on to inv_vars.
31754 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31755 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31756 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31757 (dump_cand): Dump inv_vars.
31758 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31759 (record_invariant, find_depends, add_candidate_1): Ditto.
31760 (set_group_iv_cost, force_var_cost): Ditto.
31761 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31762 (get_computation_cost_at, get_computation_cost): Ditto.
31763 (determine_group_iv_cost_generic): Ditto.
31764 (determine_group_iv_cost_address): Ditto.
31765 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31766 (determine_group_iv_costs): Ditto.
31767 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31768 (iv_ca_set_remove_invariants): Renamed to ...
31769 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31770 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31771 (iv_ca_set_add_invariants): Renamed to ...
31772 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31773 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31774 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31775 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31776 (create_new_ivs): Remove useless dump.
31778 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31780 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31782 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31783 (iv_ca_set_no_cp, create_new_iv): Ditto.
31785 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31787 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31789 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31791 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31792 function tree_check2.
31794 2017-05-02 Martin Liska <mliska@suse.cz>
31796 * doc/gcov.texi: Add missing preposition.
31797 * gcov.c (function_info::function_info): Properly fill up
31798 all member variables.
31800 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31802 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31804 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31806 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31808 2017-05-02 Martin Liska <mliska@suse.cz>
31811 * lto-streamer-in.c (lto_read_tree_1): Remove
31812 LTO_STREAMER_DEBUG.
31813 * lto-streamer.c (struct tree_hash_entry): Likewise.
31814 (struct tree_entry_hasher): Likewise.
31815 (tree_entry_hasher::hash): Likewise.
31816 (tree_entry_hasher::equal): Likewise.
31817 (lto_streamer_init): Likewise.
31818 (lto_orig_address_map): Likewise.
31819 (lto_orig_address_get): Likewise.
31820 (lto_orig_address_remove): Likewise.
31821 * lto-streamer.h: Likewise.
31822 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31823 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31825 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31827 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31828 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31829 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31830 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31831 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31832 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31833 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31834 (mm_maskz_sub_ss): New intrinsics.
31835 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31836 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31837 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31838 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31839 (__builtin_ia32_subss_mask_round): New builtins.
31840 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31841 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31842 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31844 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31845 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31847 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31850 2017-05-02 Martin Jambor <mjambor@suse.cz>
31852 PR tree-optimization/78687
31853 * tree-sra.c (access): New field parent.
31854 (process_subtree_disqualification): New function.
31855 (disqualify_candidate): Call it.
31856 (build_accesses_from_assign): Reset write flag if creating an
31858 (build_access_subtree): Fill in parent field and also prpagate
31859 down grp_write flag.
31860 (create_artificial_child_access): New parameter set_grp_write, set
31861 grp_write to its value.
31862 (propagate_subaccesses_across_link): Also propagate grp_write flag
31864 (propagate_all_subaccesses): Push the closest parent back to work
31865 queue if add_access_to_work_queue returned true.
31867 2017-05-02 Richard Biener <rguenther@suse.de>
31869 * common.opt (fstrict-overflow): Alias negative to fwrapv.
31870 * doc/invoke.texi (fstrict-overflow): Remove all traces of
31871 -fstrict-overflow documentation.
31872 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
31873 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
31874 flag_strict_overflow.
31875 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
31876 * lto-opts.c (lto_write_options): Do not stream it.
31877 * lto-wrapper.c (merge_and_complain): Do not handle it.
31878 * opts.c (default_options_table): Do not set -fstrict-overflow.
31879 (finish_options): Likewise do not clear it when sanitizing.
31880 * simplify-rtx.c (simplify_const_relational_operation): Do not
31881 test flag_strict_overflow.
31883 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31885 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
31886 using enabled attribute.
31887 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
31888 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
31889 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
31890 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
31891 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
31892 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
31893 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
31894 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
31895 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
31896 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
31898 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31900 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
31902 2017-05-02 Richard Biener <rguenther@suse.de>
31904 PR tree-optimization/80591
31906 2017-04-10 Richard Biener <rguenther@suse.de>
31908 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31911 2017-05-02 Richard Biener <rguenther@suse.de>
31913 PR tree-optimization/80549
31914 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
31915 (cleanup_tree_cfg_noloop): Create forwarders to known loop
31916 headers if they do not have a preheader.
31918 2017-05-02 Martin Liska <mliska@suse.cz>
31921 * common.opt: Fix typo.
31922 * doc/invoke.texi: Likewise.
31924 2017-05-01 Jan Beulich <jbeulich@suse.com>
31926 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
31927 swapping, add (x,x,m,x,n) alternative.
31929 2017-05-01 Nathan Sidwell <nathan@acm.org>
31931 * calls.c (combine_pending_stack_adjustment_and_call): Remove
31932 unnecessary unadjusted_alignment check.
31934 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31937 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
31939 * gimplify.c (gimplify_cilk_detach): New function.
31940 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
31941 * tree-core.h: Document EXPR_CILK_SPAWN.
31942 * tree.h (EXPR_CILK_SPAWN): Define.
31944 2017-05-01 David Malcolm <dmalcolm@redhat.com>
31946 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
31947 to use new fixit_hint representation, using the "replace" logic.
31948 (get_line_span_for_fixit_hint): Likewise.
31949 (layout::print_any_fixits): Likewise.
31950 (selftest::test_one_liner_many_fixits): Rename to...
31951 (selftest::test_one_liner_many_fixits_1): ...this, and update
31952 comment and expected output to reflect that the multiple fix-it
31953 hints are now consolidated into one insertion.
31954 (selftest::test_one_liner_many_fixits_2): New test.
31955 (selftest::test_diagnostic_show_locus_one_liner): Update for
31957 (selftest::test_fixit_consolidation): Update for fix-it API
31959 * diagnostic.c (print_parseable_fixits): Likewise.
31960 * edit-context.c (edited_line::m_line_events): Convert from
31961 auto_vec <line_event *> to auto_vec <line_event>.
31962 (class line_event): Convert from abstract base class to a concrete
31963 class, taking over the role of replace_event.
31964 (class insert_event): Delete.
31965 (class replace_event): Rename to class line_event. Convert to
31967 (edit_context::add_fixits): Reimplement.
31968 (edit_context::apply_insert): Delete.
31969 (edit_context::apply_replace): Rename to...
31970 (edit_context::apply_fixit): ...this. Convert to half-open range.
31971 (edited_file::apply_insert): Delete.
31972 (edited_file::apply_replace): Rename to...
31973 (edited_file::apply_fixit): ...this.
31974 (edited_line::~edited_line): Drop deletion of events.
31975 (edited_line::apply_insert): Delete.
31976 (edited_line::apply_replace): Rename to...
31977 (edited_line::apply_fixit): ...this. Convert to half-open range.
31978 Update for change to type of m_line_events.
31979 * edit-context.h (edit_context::apply_insert): Delete.
31980 (edit_context::apply_replace): Rename to...
31981 (edit_context::apply_fixit): ...this.
31983 2017-05-01 Martin Sebor <msebor@redhat.com>
31985 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
31988 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
31991 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
31992 __get_cpuid_max returns 0.
31993 (__get_cpuid_count): Ditto.
31995 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
31997 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
31998 replacement expression is another instance of one of its arguments.
32000 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32003 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32004 check for stack push/pop autoinc.
32005 * config/i386/i386.c (ix86_agi_dependent): Return false
32006 if the only reason why modified_in_p returned true is that
32007 addr is SP based and set_insn is a push or pop.
32009 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32011 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32014 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32017 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32018 (account_size_time): Use two predicates - exec_pred and
32020 (evaluate_conditions_for_known_args): Compute both clause and
32022 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32023 (inline_summary_t::duplicate): Update.
32024 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32026 (compute_inline_parameters): Likewise.
32027 (estimate_edge_size_and_time): Update caluclation of time.
32028 (estimate_node_size_and_time): Compute both time and nonspecialized
32030 (estimate_ipcp_clone_size_and_time): Update.
32031 (inline_merge_summary): Update.
32032 (do_estimate_edge_time): Update.
32033 (do_estimate_edge_size): Update.
32034 (do_estimate_edge_hints): Update.
32035 (inline_read_section, inline_write_summary): Stream both new predicates.
32036 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32038 (compute_inlined_call_time): Cleanup.
32039 (big_speedup_p): Update.
32040 (edge_badness): Update.
32041 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32042 (size_time_entry): Replace predicate by exec_predicate and
32043 nonconst_predicate.
32044 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32045 (estimate_edge_time): Return also nonspec_time.
32046 (reset_edge_growth_cache): Update.
32048 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32050 PR rtl-optimization/80491
32051 * ifcvt.c (noce_process_if_block): When looking for x setter
32052 with missing else_bb, don't check only the insn right before
32053 cond_earliest, but look for the last insn that x is modified in
32054 within the same bb.
32056 PR rtl-optimization/80491
32057 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32059 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32061 PR tree-optimization/80487
32062 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32064 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32066 PR tree-optimization/79697
32067 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32068 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32069 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32071 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32072 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32074 2017-04-28 Martin Sebor <msebor@redhat.com>
32076 PR tree-optimization/80523
32077 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32078 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32080 (maybe_warn, format_directive, parse_directive): Use new functions.
32081 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32083 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32085 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32087 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32089 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32090 target_header_dir): Set correctly.
32091 * configure: Regenerated.
32092 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32093 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32094 instead of SYSTEM_HEADER_DIR.
32096 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32098 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32099 (estimate_local_effects): Likewise.
32100 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32101 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32102 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32103 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32104 do_estimate_edge_time, estimate_edge_time): Likewise.
32105 * ipa-inline-analysis.c (estimate_node_size_and_time,
32106 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32107 (estimate_time_after_inlining): Remove.
32109 2017-04-28 Martin Liska <mliska@suse.cz>
32111 * doc/gcov.texi: Enhance documentation of gcov.
32113 2017-04-28 Martin Liska <mliska@suse.cz>
32115 * doc/gcov.texi: Sort options in alphabetic order.
32116 * doc/gcov-dump.texi: Likewise.
32117 * doc/gcov-tool.texi: Likewise.
32118 * gcov.c (print_usage): Likewise.
32119 * gcov-dump.c (print_usage): Likewise.
32120 * gcov-tool.c (print_merge_usage_message): Likewise.
32121 (print_rewrite_usage_message): Likewise.
32122 (print_overlap_usage_message): Likewise.
32124 2017-04-28 Martin Liska <mliska@suse.cz>
32126 PR gcov-profile/53915
32127 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32129 2017-04-28 Martin Liska <mliska@suse.cz>
32131 PR gcov-profile/79891
32132 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32133 is marked by compiler as living on a line.
32134 (get_cycles_count): Remove usage of the union.
32135 (output_intermediate_file): Likewise.
32136 (find_source): Fix GNU coding style.
32137 (accumulate_line_counts): Remove old non-all block mode.
32138 (output_lines): Remove usage of the union.
32139 * profile.c (output_location): Include all BBs, even if
32140 belonging to a same line (and file) as a previous BB.
32142 2017-04-28 Martin Liska <mliska@suse.cz>
32144 * gcov.c (process_args): Handle new argument 'w'.
32145 (read_graph_file): Assign ID to BBs.
32146 (output_branch_count): Display BB # if verbose flag is set.
32147 (output_lines): Likewise for arcs.
32148 (print_usage): Add '--verbose' option help.
32149 * doc/gcov.texi: Document --verbose (-w) option.
32151 2017-04-28 Martin Liska <mliska@suse.cz>
32153 * gcov.c (struct block_location_info): New struct.
32154 (process_file): Fill up the new structure.
32155 (read_graph_file): Replace usage of encoding by the newly added
32157 (add_line_counts): Likewise.
32158 (accumulate_line_counts): Remove usage of the union.
32159 (function_info::function_info): New function.
32160 (function_info::~function_info): Likewise.
32161 (process_file): Call delete instead of release_function.
32162 (release_function): Release the function.
32163 (release_structures): Call delete instead of release_function.
32164 (solve_flow_graph): Replace usage of num_blocks.
32165 (find_exception_blocks): Likewise.
32166 (output_lines): Fix GNU coding style.
32168 2017-04-28 Martin Liska <mliska@suse.cz>
32171 * coverage.c (coverage_remove_note_file): New function.
32172 * coverage.h: Declare the function.
32173 * toplev.c (finalize): Clean if an error has been seen.
32175 2017-04-28 Martin Liska <mliska@suse.cz>
32177 PR gcov-profile/80031
32178 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32179 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32180 * gcov.c (read_graph_file): Read just number of blocks.
32181 * profile.c (branch_prob): Do not stream 0 flags per a basic
32184 2017-04-28 Martin Liska <mliska@suse.cz>
32186 * gcov-dump.c (tag_*): Add new argument to declarations.
32187 (dump_gcov_file): Likewise.
32188 (tag_blocks): Add and use new argument depth.
32189 (tag_arcs): Likewise.
32190 (tag_lines): Likewise.
32191 (tag_counters): Likewise.
32192 (tag_summary): Likewise.
32193 (dump_working_sets): Use depth to do a proper indentation.
32195 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32198 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32199 * symtab.c (symtab_node::debug_symtab): Move definition here.
32201 2017-04-28 Richard Biener <rguenther@suse.de>
32203 * lto-streamer.h (LTO_major_version): Bump to 7.
32205 2017-04-28 Richard Biener <rguenther@suse.de>
32207 * tree-vrp.c (assert_info): New struct.
32208 (add_assert_info): New helper.
32209 (register_edge_assert_for_2): Refactor to add asserts to a vector
32211 (register_edge_assert_for_1): Likewise.
32212 (register_edge_assert_for): Likewise.
32213 (finish_register_edge_assert_for): New helper actually registering
32214 asserts where live on edge.
32215 (find_conditional_asserts): Adjust.
32216 (find_switch_asserts): Likewise.
32217 (evrp_dom_walker::try_find_new_range): Generalize.
32218 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32220 2017-04-27 Marek Polacek <polacek@redhat.com>
32223 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32224 arg10 and arg11 to itype.
32226 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32228 * doc/extend.texi (Object Size Checking): Improve grammar.
32230 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32233 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32234 that the logic for permitting reciprocal estimates matches that
32237 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32240 * tree.c (type_cache_hasher::equal): Only compare
32241 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32242 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32243 non-aggregate element types.
32244 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32245 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32247 2017-04-27 Richard Biener <rguenther@suse.de>
32249 PR middle-end/80533
32250 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32251 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32252 keeping a reference to a trailing array.
32254 2017-04-27 Richard Biener <rguenther@suse.de>
32256 PR middle-end/80539
32257 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32258 being in loop-closed SSA form conservatively.
32259 (chrec_fold_multiply_poly_poly): Likewise.
32261 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32263 PR middle-end/79665
32264 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32265 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32267 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32270 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32271 (aarch64_function_arg_alignment): Return unsigned int again, but still
32272 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32273 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32274 Don't emit -Wpsabi note.
32275 (aarch64_function_arg_boundary): Likewise.
32276 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32279 2017-04-26 Nathan Sidwell <nathan@acm.org>
32281 * tree.h (crc32_unsigned_n): Declare.
32282 (crc32_unsigned, crc32_unsigned): Make inline.
32283 * tree.c (crc32_unsigned_bits): Replace with ...
32284 (crc32_unsigned_n): ... this.
32285 (crc32_unsigned, crc32_byte): Remove.
32286 (crc32_string): Remove unnecessary braces.
32288 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32290 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32291 * ipa-inline-analysis.c (MAX_TIME): Remove.
32292 (account_size_time): Use sreal for time.
32293 (dump_inline_summary): Update.
32294 (estimate_function_body_sizes): Update.
32295 (estimate_edge_size_and_time): Update.
32296 (estimate_calls_size_and_time): Update.
32297 (estimate_node_size_and_time): Update.
32298 (inline_merge_summary): Update.
32299 (inline_update_overall_summary): Update.
32300 (estimate_time_after_inlining): Update.
32301 (inline_read_section): Update.
32302 (inline_write_summary): Update.
32303 * ipa-inline.c (compute_uninlined_call_time): Update.
32304 (compute_inlined_call_time): Update.
32305 (recursive_inlining): Update.
32306 (inline_small_functions): Update.
32307 (dump_overall_stats): Update.
32308 * ipa-inline.h: Include sreal.h.
32309 (size_time_entry): Turn time to sreal.
32310 (inline_summary): Turn self_time nad time to sreal.
32312 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32314 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32316 (sreal::stream_out, sreal::stream_in): New.
32317 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32319 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32321 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32324 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32327 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32329 (dimode_scalar_chain::compute_convert_gain): Ditto.
32330 (dimode_scalar_chain::make_vector_copies): Ditto.
32331 (dimode_scalar_chain::convert_reg): Ditto.
32332 (dimode_scalar_chain::convert_insn): Ditto.
32333 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32334 (VI248_AVX512BW_1): New mode iterator.
32335 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32336 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32339 2017-04-25 Martin Sebor <msebor@redhat.com>
32341 PR tree-optimization/80497
32342 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32343 constants are representable in HOST_WIDE_INT.
32344 (parse_directive): Ditto.
32346 2017-04-25 Martin Sebor <msebor@redhat.com>
32349 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32350 (new_zero_array): Adjust signature.
32351 (dom_info::dom_init): Used unsigned rather that size_t.
32352 (dom_info::dom_info): Same.
32354 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32355 Jakub Jelinek <jakub@redhat.com>
32358 * config/arm/arm.c: Include gimple.h.
32359 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32360 returns negative, increment ncrn only if it returned positive.
32361 (arm_needs_doubleword_align): Return int instead of bool,
32362 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32363 members, but if there is any such non-FIELD_DECL
32364 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32365 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32366 returns negative, increment nregs only if it returned positive.
32367 (arm_setup_incoming_varargs): Likewise.
32368 (arm_function_arg_boundary): Emit -Wpsabi note if
32369 arm_needs_doubleword_align returns negative, return
32370 DOUBLEWORD_ALIGNMENT only if it returned positive.
32372 2017-04-25 Marek Polacek <polacek@redhat.com>
32375 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32376 first argument to type.
32378 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32381 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32382 type checks to test for compatibility instead of equality.
32384 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32385 Jakub Jelinek <jakub@redhat.com>
32388 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32390 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32391 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32392 the alignment computation, but return their maximum in warn_alignment.
32393 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32394 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32396 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32397 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32400 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32402 * config/arc/simdext.md (dmpyh): Fix typo.
32404 2017-04-25 Richard Biener <rguenther@suse.de>
32406 PR tree-optimization/80492
32407 * alias.c (compare_base_decls): Handle registers with asm
32408 specification conservatively.
32409 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32410 compare_base_decls returning dont-know properly.
32412 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32414 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32415 (legitimate_offset_address_p): New function.
32416 (arc_legitimate_address_p): Use above function.
32418 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32420 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32422 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32424 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32425 ACCH registers whenever they are available.
32427 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32429 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32430 double regs fix when not used.
32432 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32434 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32436 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32437 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32439 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32441 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32442 when emitting short ADD instructions.
32444 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32446 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32448 (cmpsi_cc_c_insn): Likewise.
32449 (cbranchsi4_scratch): Compute proper instruction length using
32450 compact_hreg_operand.
32451 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32453 2017-04-25 Richard Biener <rguenther@suse.de>
32455 PR middle-end/80509
32456 * passes.c (pass_manager::pass_manager): Initialize
32457 m_name_to_pass_map.
32459 2017-04-25 Richard Biener <rguenther@suse.de>
32461 PR tree-optimization/79201
32462 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32464 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32467 * config/s390/vector.md: Split MEM->GPR vector moves for
32468 non-s_operand addresses.
32470 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32473 * config/s390/predicates.md (reload_const_wide_int_operand): New
32475 * config/s390/s390.md ("movti"): Remove d/P alternative.
32476 ("movti_bigconst"): New pattern definition.
32478 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32481 * s390-protos.h (s390_expand_cs_hqi): Removed.
32482 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32483 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32484 modes as well as CCZ1mode and CCZmode.
32485 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32486 signature of s390_emit_compare_and_swap.
32487 (s390_expand_cs_hqi): Likewise, make static.
32488 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32489 compare-and-swap, in some cases.
32490 (s390_expand_cs): Wrapper function.
32491 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32493 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32494 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32495 patterns for small and large integers. Forbid symref memory operands.
32496 Move expander to s390.c. Require cc register.
32497 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32498 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32499 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32500 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32501 symref memory operands. Remove CC mode and call s390_match_ccmode
32503 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32505 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32507 * config/s390/s390.md (define_peephole2): New peephole to help
32508 combining the load-and-test pattern with volatile memory.
32510 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32512 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32513 with CCZmode for TARGET_Z196.
32515 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32517 PR rtl-optimization/80501
32518 * combine.c (make_compound_operation_int): Set subreg_code to SET
32519 even for AND with mask of the sign bit of mode.
32521 PR rtl-optimization/80500
32522 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32523 sum's initial value.
32525 2017-04-25 Julian Brown <julian@codesourcery.com>
32526 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32528 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32530 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32532 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32534 2017-04-25 Julian Brown <julian@codesourcery.com>
32535 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32537 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32538 (thunderx2t99_sha): New Reservation.
32540 2017-04-25 Julian Brown <julian@codesourcery.com>
32541 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32543 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32544 type for 1-element load.
32546 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32548 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32550 2017-04-24 Martin Jambor <mjambor@suse.cz>
32552 PR tree-optimization/80293
32553 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32554 char arrays not totally scalarizable if it is false.
32555 (analyze_all_variable_accesses): Pass correct value in the new
32556 parameter. Add a statistics counter.
32558 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32560 PR middle-end/79931
32561 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32563 2017-04-24 Richard Biener <rguenther@suse.de>
32565 PR tree-optimization/80494
32566 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32567 out for complex types.
32569 2017-04-24 Richard Biener <rguenther@suse.de>
32571 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32572 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32573 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32574 (DFS): Adjust and never fail.
32575 (sccvn_dom_walker::fail): Remove.
32576 (sccvn_dom_walker::before_dom_children): Adjust.
32577 (run_scc_vn): Likewise and never fail.
32578 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32579 (pass_fre::execute): Likewise.
32581 2017-04-24 Richard Biener <rguenther@suse.de>
32583 PR tree-optimization/79725
32584 * tree-ssa-sink.c (statement_sink_location): Return whether
32585 failure reason was zero uses. Move that check later.
32586 (sink_code_in_bb): Deal with zero uses by removing the stmt
32589 2017-04-24 Richard Biener <rguenther@suse.de>
32592 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32593 pointer-based references.
32595 2017-04-24 Richard Biener <rguenther@suse.de>
32598 * pass_manager.h (pass_manager::operator new): Remove.
32599 (pass_manager::operator delete): Likewise.
32600 * passes.c (pass_manager::operator new): Remove.
32601 (pass_manager::operator delete): Likewise.
32602 (pass_manager::pass_manager): Zero individual pass members.
32604 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32607 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32608 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32609 Check "XEXP (src, 1)" operand here.
32610 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32611 Check "XEXP (src, 1)" operand here.
32612 (dimode_scalar_chain::make_vector_copies): Detect count register
32613 of a shift instruction. Zero extend count register from QImode
32614 to DImode to satisfy vector shift pattern count operand predicate.
32615 Substitute vector shift count operand with a DImode copy.
32616 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32619 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32621 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32622 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32623 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32624 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32625 (UNSPEC_NOREX_MEM): Remove definition.
32627 2017-04-21 Richard Biener <rguenther@suse.de>
32629 PR tree-optimization/79547
32630 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32631 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32632 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32633 without any constraints.
32635 2017-04-21 Richard Biener <rguenther@suse.de>
32637 PR tree-optimization/78847
32638 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32640 2017-04-21 Richard Biener <rguenther@suse.de>
32642 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32643 (build_distinct_type_copy): Likewise.
32644 (build_variant_type_copy): Likewise.
32645 * tree.c (build_qualified_type): Pass down mem-stat info.
32646 (build_distinct_type_copy): Likewise.
32647 (build_variant_type_copy): Likewise.
32649 2017-04-21 Richard Biener <rguenther@suse.de>
32651 PR tree-optimization/80237
32652 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32654 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32655 for a simplified result.
32657 2016-04-21 Richard Biener <rguenther@suse.de>
32659 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32660 sth as strict as a simple_iv but a chrec without symbols and an
32661 operand defined in the loop we are peeling (and not some subloop).
32662 (propagate_constants_for_unrolling): Propagate all constants.
32664 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32667 * config/i386/i386.c (print_reg): Remove assert for disalowed
32668 regno values, call output_operand_lossage instead.
32670 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32673 * config/i386/constraints.md (Yc): New register constraint.
32674 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32675 Use Yc constraint for alternative 2 of operand 0. Remove
32676 preferred_for_speed attribute.
32678 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32680 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32681 lastprivate clauses in SIMT case.
32683 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32685 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32687 2017-04-20 Richard Biener <rguenther@suse.de>
32689 PR tree-optimization/57796
32690 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32691 as N scalar stores.
32692 (vect_model_load_cost): Cost gathers as N scalar loads.
32694 2017-04-20 Richard Biener <rguenther@suse.de>
32696 * ggc-page.c (ggc_allocated_p): Rename to ...
32697 (safe_lookup_page_table_entry): ... this and return the lookup
32699 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32701 2017-04-20 Richard Biener <rguenther@suse.de>
32703 PR tree-optimization/80453
32704 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32705 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32706 from the conditions.
32707 (vn_phi_eq): Pass them down.
32708 (vn_phi_lookup): Record them.
32709 (vn_phi_insert): Likewise.
32711 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32713 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32714 uninitialized variable warning to avoid buffer overrun.
32716 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32719 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32720 is suppressed for '{ 0 }' in C.
32722 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32724 * BASE-VER: Set to 8.0.0.
32726 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32728 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32729 priority .init_array and .fini_array section with SECTION_NOTYPE
32732 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32734 PR middle-end/80423
32735 * tree.h (build_array_type): Add typeless_storage default argument.
32736 * tree.c (type_cache_hasher::equal): Also compare
32737 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32738 (build_array_type): Add typeless_storage argument, set
32739 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32741 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32742 (build_array_type): Likewise. Add typeless_storage argument.
32744 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32745 Jakub Jelinek <jakub@redhat.com>
32747 PR tree-optimization/80426
32748 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32749 operation on symbolic operands, also compute the overflow for the
32750 invariant part when the operation degenerates into a negation.
32752 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32755 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32756 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32759 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32761 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32764 * config/avr/avr.c (tree.h): Include it.
32765 (cgraph.h): Include it.
32766 (avr_encode_section_info): Don't warn for uninitialized progmem
32767 variable if it's just an alias.
32769 2017-04-19 Richard Biener <rguenther@suse.de>
32772 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32773 when needed by AutoPGO.
32775 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32778 * doc/lto.texi: Remove an extra 'that'.
32780 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32782 PR rtl-optimization/80429
32783 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32784 are only used in debug insns.
32786 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32787 Vladimir Makarov <vmakarov@redhat.com>
32789 * config/sparc/predicates.md (input_operand): Add comment. Return
32790 true for any memory operand when LRA is in progress.
32791 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32793 2017-04-18 Jeff Law <law@redhat.com>
32796 * mips.md ({return,simple_return}_internal): Do not overwrite
32799 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32801 PR tree-optimization/80443
32802 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32803 instead of adding 1, subtract -1 and similarly instead of subtracting
32806 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32808 PR rtl-optimization/80357
32809 * haifa-sched.c (tmp_bitmap): New variable.
32810 (model_recompute): Handle duplicate use records.
32811 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32812 (free_global_sched_pressure_data): Free it.
32814 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32817 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32818 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32819 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32820 instead of SYSTEM_HEADER_DIR.
32822 2017-04-18 Jeff Law <law@redhat.com>
32824 PR middle-end/80422
32825 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32826 predecessors after walking up the insn chain.
32828 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32831 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32832 sizetype type into debug info.
32834 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32837 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32838 unneeded test for TARGET_UPPER_REGS_SF.
32839 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32841 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32844 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32845 instead of gsi_after_labels.
32847 2017-04-18 Jeff Law <law@redhat.com>
32849 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32853 2017-04-13 Jeff Law <law@redhat.com>
32854 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32855 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32857 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
32860 * config/avr/avr.c (intl.h): Include it.
32861 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
32863 2017-04-18 Martin Liska <mliska@suse.cz>
32865 PR gcov-profile/78783
32866 * gcov-tool.c (gcov_output_files): Validate that destination
32867 file is either removed by the tool or by a user.
32869 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32870 Guy Benyei <guybe@mellanox.com>
32872 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
32873 block, and do not negate it, the stored id is already negative.
32875 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32877 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
32879 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32882 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
32883 masks of options that should be turned off if the VSX vector
32884 options are turned off.
32885 (OTHER_P8_VECTOR_MASKS): Likewise.
32886 (OTHER_VSX_VECTOR_MASKS): Likewise.
32887 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
32888 rs6000_disable_incompatible_switches to validate no type switches
32890 (rs6000_incompatible_switch): New function to disallow turning on
32891 other vector options if -mno-vsx, -mno-power8-vector, or
32892 -mno-power9-vector are specified.
32894 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32896 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
32898 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32900 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
32901 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
32902 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
32903 (ARG_POINTER_CFA_OFFSET): Likewise.
32905 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32907 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
32908 conditions to take advantage of various optimizations.
32910 2017-04-13 Jeff Law <law@redhat.com>
32912 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32913 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32914 (zero_extendsidi2_dext): Likewise.
32916 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32919 * fold-const.c (fold_ternary_loc): Revert
32920 use op0 instead of fold_convert_loc (loc, type, arg0) part of
32923 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
32925 PR rtl-optimization/80343
32926 * lra-remat.c (update_scratch_ops): Assign original hard reg to
32927 new scratch pseudo.
32929 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
32932 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
32933 to ubsan_encode_value.
32935 2017-04-13 Jeff Law <law@redhat.com>
32937 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
32938 appearing in DEBUG_INSNs.
32940 2017-04-13 Martin Liska <mliska@suse.cz>
32942 PR gcov-profile/80413
32943 * gcov-io.c (gcov_write_string): Copy to buffer just when
32944 allocated size is greater than zero.
32946 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32949 * dwarf2out.c (decls_for_scope): Ignore declarations of
32950 current_function_decl in BLOCK_NONLOCALIZED_VARS.
32952 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
32955 * ipa-visibility.c (non_local_p): Fix typos.
32956 (localize_node): When localizing symbol in same comdat group,
32957 dissolve the group only when we know external symbols are going
32959 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
32961 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32963 PR tree-optimization/79390
32964 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
32965 order does not result in usable sequence, retry with reversed operand
32971 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
32972 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
32973 op0 instead of fold_convert_loc (loc, type, arg0).
32975 2017-04-12 Jeff Law <law@redhat.com>
32977 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
32978 has a delay slot in the generated code.
32980 * config/cris/cris.md (cris_preferred_reload_class): Return
32981 GENNONACR_REGS rather than GENERAL_REGS.
32983 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32986 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
32987 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
32988 signedness of the result type.
32990 2017-04-12 Richard Biener <rguenther@suse.de>
32991 Jeff Law <law@redhat.com>
32993 PR tree-optimization/80359
32994 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
32995 trim stores to TARGET_MEM_REFs.
32997 2017-04-12 Richard Biener <rguenther@suse.de>
32999 PR tree-optimization/79390
33000 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33001 threading case even more.
33003 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33006 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33007 for quad_address_p for TImode, instead of just not indexed_address.
33009 2017-04-12 Richard Biener <rguenther@suse.de>
33010 Bernd Edlinger <bernd.edlinger@hotmail.de>
33012 PR middle-end/79671
33013 * alias.c (component_uses_parent_alias_set_from): Handle
33014 TYPE_TYPELESS_STORAGE.
33015 (get_alias_set): Likewise.
33016 * tree-core.h (tree_type_common): Add typeless_storage flag.
33017 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33018 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33019 for types containing members with TYPE_TYPELESS_STORAGE.
33020 (place_field): Likewise.
33021 (layout_type): Likewise for ARRAY_TYPE.
33022 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33023 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33024 TYPE_TYPELESS_STORAGE.
33025 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33027 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33030 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33031 first argument to type.
33033 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33037 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33038 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33039 (rs6000_expand_binop_builtin): Likewise.
33040 (rs6000_expand_ternop_builtin): Likewise; also add missing
33041 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33042 vshasigma built-ins.
33043 * doc/extend.texi: Document that vec_xxpermdi's third argument
33044 must be a constant.
33046 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33048 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33049 Use shift_const cost parameter when calculating gain of STV shifts.
33051 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33053 PR rtl-optimization/70478
33054 * lra-constraints.c (process_alt_operands): Check memory for
33055 disfavoring memory insn operand.
33057 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33059 PR middle-end/80100
33060 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33061 left shift in unsigned HOST_WIDE_INT type.
33063 PR rtl-optimization/80385
33064 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33065 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33068 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33069 if they have any depend clauses.
33071 2017-04-11 Martin Liska <mliska@suse.cz>
33074 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33075 * ipa-split.c (split_function): Create a local comdat symbol
33076 if caller is in a comdat group.
33078 2017-04-11 Martin Liska <mliska@suse.cz>
33081 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33084 2017-04-11 Martin Sebor <msebor@redhat.com>
33086 PR middle-end/80364
33087 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33088 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33090 (directive::set_width, directive::set_precision, format_character):
33092 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33095 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33098 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33099 conflict, set target->arch_name instead of target->cpu_name.
33101 2017-04-11 Richard Biener <rguenther@suse.de>
33103 PR tree-optimization/80374
33104 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33105 build_zero_cst, remove fold_convertible_p check again.
33107 2017-04-11 Martin Liska <mliska@suse.cz>
33110 * ubsan.c (instrument_object_size): Do not instrument register
33113 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33116 * config/i386/i386-builtin-types.def
33117 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33118 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33119 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33120 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33121 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33122 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33123 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33124 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33125 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33126 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33127 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33128 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33129 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33130 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33131 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33132 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33133 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33134 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33135 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33136 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33137 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33138 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33139 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33140 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33141 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33142 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33143 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33144 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33145 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33146 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33147 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33148 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33149 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33150 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33151 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33152 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33153 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33154 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33155 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33156 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33157 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33158 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33159 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33160 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33161 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33162 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33164 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33165 flag to second_arg_count, handle 4 argument function type _COUNT
33166 aliases, handle second_arg_count on second argument rather than last.
33168 2017-04-10 Jeff Law <law@redhat.com>
33170 PR tree-optimization/80374
33171 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33172 record anything if we can not convert integer_zero_node to the
33175 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33178 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33179 Enhance special handling given to the TARGET_P9_MINMAX option in
33180 relation to certain other options.
33182 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33184 PR tree-optimization/80153
33185 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33186 remove POINTER_PLUS_EXPR's base part directly, rather than through
33189 2017-04-10 Richard Biener <rguenther@suse.de>
33190 Bin Cheng <bin.cheng@arm.com>
33192 PR tree-optimization/80153
33193 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33194 the first element of pointer type aff_tree. Build result expr in
33196 (add_elt_to_tree): Convert to type unconditionally. Remove other
33197 fold_convert calls.
33198 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33199 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33201 2017-04-10 Richard Biener <rguenther@suse.de>
33203 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33206 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33208 PR rtl-optimization/70478
33209 * lra-constraints.c (curr_small_class_check): New.
33210 (update_and_check_small_class_inputs): New.
33211 (process_alt_operands): Update curr_small_class_check. Disfavor
33212 alternative insn memory operands. Check available regs for small
33215 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33218 * config/mips/mips.opt (-mvirt): Update description.
33219 * doc/invoke.texi (-mvirt): Likewise.
33221 2017-04-10 Richard Biener <rguenther@suse.de>
33223 PR middle-end/80362
33224 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33225 looking for NEGATE_EXPR in -A / -B to A / B folding.
33227 2017-04-10 Martin Liska <mliska@suse.cz>
33229 PR gcov-profile/80224
33230 * gcov.c (print_usage): Fix usage string.
33231 (get_gcov_intermediate_filename): Remove.
33232 (output_gcov_file): Use both for normal and intermediate format.
33233 (generate_results): Do not initialize special file for
33234 intermediate format.
33236 2017-04-10 Richard Biener <rguenther@suse.de>
33238 PR tree-optimization/80304
33239 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33242 2017-04-10 Nathan Sidwell <nathan@acm.org>
33245 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33246 (rs6000_init_builtins): Use it.
33248 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33250 * config/arm/arm.md (<mrc>): Add mode to SET source.
33251 (<mrrc>): Likewise.
33253 2017-04-10 Richard Biener <rguenther@suse.de>
33255 PR middle-end/80344
33256 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33258 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33261 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33262 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33263 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33264 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33265 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33266 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33267 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33268 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33269 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33270 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33271 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33272 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33273 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33274 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33275 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33276 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33277 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33278 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33279 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33280 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33281 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33282 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33283 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33285 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33287 PR rtl-optimization/70478
33288 * lra-constraints.c: Reverse the last patch.
33290 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33292 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33293 Add comment for WCHAR_T.
33295 2017-04-08 Martin Liska <mliska@suse.cz>
33298 2017-04-07 Martin Liska <mliska@suse.cz>
33301 * ipa-split.c (split_function): Add function part to a same comdat
33304 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33307 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33309 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33311 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33312 (vec_pairing): ...to this.
33313 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33314 (rs6000_sched_init): Adjust for name change.
33315 (struct rs6000_sched_context): Likewise.
33316 (rs6000_init_sched_context): Likewise.
33317 (rs6000_set_sched_context): Likewise.
33319 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33325 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33327 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33328 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33329 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33331 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33333 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33335 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33337 PR rtl-optimization/70703
33338 * ira-color.c (update_conflict_hard_regno_costs): Use
33339 int64_t instead of HOST_WIDE_INT.
33341 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33343 PR rtl-optimization/70478
33344 * lra-constraints.c (process_alt_operands): Disfavor alternative
33345 insn memory operands.
33347 2017-04-07 Jeff Law <law@redhat.com>
33349 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33350 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33352 2017-04-07 Martin Liska <mliska@suse.cz>
33355 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33356 Show error message instead of an ICE.
33358 2017-04-07 Martin Liska <mliska@suse.cz>
33361 * ipa-split.c (split_function): Add function part to a same comdat
33364 2017-04-07 Richard Biener <rguenther@suse.de>
33366 PR middle-end/80341
33367 * tree.c (get_unwidened): Also handle ! for_type case for
33369 * convert.c (do_narrow): Split out from ...
33370 (convert_to_integer_1): ... here. Do not pass final truncation
33371 type to get_unwidened for TRUNC_DIV_EXPR.
33373 2017-04-07 Richard Biener <rguenther@suse.de>
33375 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33377 (aff_combination_const): Adjust.
33378 (aff_combination_scale): Likewise.
33379 (aff_combination_add_elt): Likewise.
33380 (aff_combination_add_cst): Likewise.
33381 (aff_combination_convert): Likewise.
33382 (add_elt_to_tree): Likewise. Remove unused argument.
33383 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33385 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33387 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33389 * config/arm/arm.c (arm_default_short_enums): Use
33390 ARM_DEFAULT_SHORT_ENUMS.
33391 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33393 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33396 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33397 members with redundant out-of-class redeclaration.
33399 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33402 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33403 * config/i386/i386.md (*zero_extendsidi2):
33404 Add (?*x,*x) and (?*v,*v) alternatives.
33406 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33409 * config/i386/i386.c (ix86_expand_builtin)
33410 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33411 mode from insn data. Convert operands to insn operand mode.
33412 Copy operands that don't satisfy insn predicate to a register.
33414 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33416 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33419 2017-04-06 Richard Biener <rguenther@suse.de>
33421 PR tree-optimization/80334
33422 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33423 preserve alignment of accesses.
33425 2017-04-06 Richard Biener <rguenther@suse.de>
33427 PR tree-optimization/80262
33428 * tree-sra.c (build_ref_for_offset): Preserve address-space
33430 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33431 Drop useless address-space information on MEM_REF offsets.
33433 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33435 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33437 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33439 PR rtl-optimization/70703
33440 * ira-color.c (update_conflict_hard_regno_costs): Use
33441 HOST_WIDE_INT instead of long.
33443 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33446 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33447 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33449 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33450 for x86_64 target. Handle -m3dnowa option.
33452 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33454 PR rtl-optimization/70703
33455 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33456 (update_conflict_hard_regno_costs): Use long instead of unsigned
33457 arithmetic for cost calculation.
33459 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33460 Bernd Edlinger <bernd.edlinger@hotmail.de>
33463 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33466 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33469 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33470 ptr_mode with Pmode throughout.
33471 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33472 into probe_stack_range and use DImode.
33474 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33477 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33478 call_eh_return is true.
33480 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33482 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33483 Initialize last_match_fntype_index.
33485 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33488 * tree-nvr.c: Include internal-fn.h.
33489 (pass_return_slot::execute): Ignore internal calls without
33492 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33493 Richard Biener <rguenther@suse.de>
33496 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33497 captures used multiple times, except for the last use.
33498 * generic-match-head.c: Include gimplify.h.
33500 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33502 PR tree-optimization/79390
33503 * target.h (struct noce_if_info): Declare.
33504 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33505 * target.def (noce_conversion_profitable_p): New target hook.
33506 * ifcvt.h (struct noce_if_info): New type, moved from ...
33507 * ifcvt.c (struct noce_if_info): ... here.
33508 (noce_conversion_profitable_p): Renamed to ...
33509 (default_noce_conversion_profitable_p): ... this. No longer
33511 (noce_try_store_flag_constants, noce_try_addcc,
33512 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33513 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33514 instead of noce_conversion_profitable_p.
33515 * config/i386/i386.c: Include ifcvt.h.
33516 (ix86_option_override_internal): Don't override
33517 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33518 (ix86_noce_conversion_profitable_p): New function.
33519 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33520 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33521 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33522 * doc/tm.texi: Regenerated.
33524 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33526 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33529 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33532 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33533 instructions for small multiply cores.
33535 2017-04-04 Jeff Law <law@redhat.com>
33537 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33539 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33540 that are not set by the loop over the elements.
33542 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33545 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33546 int mode, convert_modes it to mode as unsigned, otherwise use
33547 lowpart_subreg to mode rather than SImode.
33548 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33549 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33550 Use DImode instead of SImode for the shift count operand.
33551 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33554 2017-04-04 Richard Biener <rguenther@suse.de>
33556 PR middle-end/80281
33557 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33558 arithmetic done for the negate or the plus. Simplify.
33559 (A - (-B) -> A + B): Likewise.
33560 * fold-const.c (split_tree): Make sure to not negate pointers.
33562 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33564 PR rtl-optimization/60818
33565 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33566 a compare of comparisons with the thing compared if this results
33567 in a different machine mode.
33569 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33571 * alias.c (base_alias_check): Fix typo in comment.
33572 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33573 * cgraphunit.c (symbol_table::compile): Likewise.
33574 * collect2.c (maybe_run_lto_and_relink): Likewise.
33575 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33576 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33577 * config/avr/avr.c (avr_map_op_t): Likewise.
33578 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33579 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33580 * config/epiphany/epiphany.md (movcc): Likewise.
33581 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33582 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33584 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33585 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33586 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33587 * config/sh/sh.c (sh_rtx_costs): Likewise.
33588 * fold-const.c (fold_truth_andor): Likewise.
33589 * genautomata.c (collapse_flag): Likewise.
33590 * gengtype.h (struct type::u::s): Likewise.
33591 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33592 * input.c (FORMAT_AMOUNT): Likewise.
33593 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33594 (known_aggs_to_agg_replacement_list): Likewise.
33595 * ipa-inline-analysis.c: Likewise.
33596 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33597 * ipa-polymorphic-call.c
33598 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33599 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33600 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33602 * modulo-sched.c (apply_reg_moves): Likewise.
33603 * omp-expand.c (build_omp_regions_1): Likewise.
33604 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33605 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33606 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33607 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33608 * value-prof.c: Likewise.
33609 * var-tracking.c (val_reset): Likewise.
33611 2017-04-03 Richard Biener <rguenther@suse.de>
33613 PR tree-optimization/80275
33614 * fold-const.c (split_address_to_core_and_offset): Handle
33617 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33619 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33620 descriptors is at least equal to that of functions.
33622 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33624 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33626 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33629 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33630 (mov<IMOD4:mode>): New expander.
33631 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33633 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33635 PR rtl-optimization/79405
33636 * fwprop.c (propagations_left): New variable.
33637 (forward_propagate_into): Decrement it.
33638 (fwprop_init): Initialize it.
33639 (fw_prop): If the variable has reached zero, stop propagating.
33640 (fwprop_addr): Ditto.
33642 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33645 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33646 a FUNCTION_DECL, pass it as decl instead of origin to
33649 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33651 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33654 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33657 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33658 TARGET_VSX_SMALL_INTEGER.
33660 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33662 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33663 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33665 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33667 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33668 extraction from odd-numbered MSA register.
33670 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33672 PR middle-end/80173
33673 * expmed.c (store_bit_field_1): Don't attempt to create
33674 a word subreg out of hard registers wider than word if they
33675 have HARD_REGNO_NREGS of 1 for their mode.
33677 PR middle-end/80163
33678 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33679 conversions to integer types wider than word and pointer.
33682 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33683 (rtx_equal_for_cselib_p): Pass 0 to it.
33684 * cselib.c (cselib_hasher::equal): Likewise.
33685 (rtx_equal_for_cselib_1): Add depth argument. If depth
33686 is 128, don't look up VALUE locs and punt. Increment
33687 depth in recursive calls when walking VALUE locs.
33689 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33691 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33692 (make_gcov_file_name): Use the canonical path name for generating
33694 (read_line): Fix handling of files with ascii null bytes.
33696 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33698 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33699 to initialise a vector register instead
33700 of using a const_int.
33702 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33704 PR translation/80189
33705 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33706 diagnostic messages.
33708 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33711 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33712 (dfp_diex_<mode>): Update mode of operand 1.
33713 * doc/extend.texi (dxex, dxexq): Document change to return type.
33714 (diex, diexq): Document change to argument type.
33716 2017-03-30 Martin Jambor <mjambor@suse.cz>
33719 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33720 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33721 it reflects the signature changes performed at the callee side.
33722 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33723 to cgraph_build_function_type_skip_args.
33724 (build_function_decl_skip_args): Adjust call to the above function.
33726 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33729 * config/i386/sse.md
33730 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33731 register as dest whenever it is a MEM not rtx_equal_p to the
33732 corresponding dup operand, and when forcing into reg move the
33733 reg into the memory afterwards.
33734 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33735 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33736 for the force_reg mode.
33737 (avx512vl_vextractf128<mode>): Use register as dest either
33738 always when a MEM, or when it is a MEM not rtx_equal_p to the
33739 corresponding dup operand, or even not when it is a CONST_VECTOR
33740 depending on the mode and lo vs. hi.
33741 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33743 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33744 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33745 Likewise. Require that operands[2] is even.
33746 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33747 Remove extraneous parens. Require that operands[2] is a multiple
33749 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33750 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33751 disallow memory then.
33753 2017-03-30 Richard Biener <rguenther@suse.de>
33755 PR tree-optimization/77498
33756 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33757 to non-constants over backedges.
33759 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33761 PR rtl-optimization/80233
33762 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33763 as last_combined_insn. Do not test for BARRIER_P separately.
33765 2017-03-29 Andreas Schwab <schwab@suse.de>
33768 * calls.c (prepare_call_address): Convert funexp to Pmode before
33769 copying to temp reg.
33771 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33773 PR tree-optimization/80158
33774 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33775 Handle possible future case of more than one alternate
33777 (replace_rhs_if_not_dup): Likewise.
33778 (replace_one_candidate): Likewise.
33780 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33782 PR rtl-optimization/80193
33783 * ira.c (ira): Do not check allocation for LRA.
33785 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33787 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33788 (nvptx_output_simt_exit): Declare.
33789 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33790 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33791 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33792 (nvptx_declare_function_name): ...here. Emit declaration of local
33793 memory space buffer for omp_simt_enter insn.
33794 (nvptx_output_unisimt_switch): New.
33795 (nvptx_output_softstack_switch): New.
33796 (nvptx_output_simt_enter): New.
33797 (nvptx_output_simt_exit): New.
33798 * config/nvptx/nvptx.h (struct machine_function): New fields
33799 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33800 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33801 (UNSPECV_SIMT_EXIT): Ditto.
33802 (omp_simt_enter_insn): New insn.
33803 (omp_simt_enter): New expansion.
33804 (omp_simt_exit): New insn.
33805 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33807 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33808 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33809 (expand_GOMP_SIMT_EXIT): New.
33810 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33811 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33812 (GOMP_SIMT_EXIT): Ditto.
33813 * target-insns.def (omp_simt_enter): New insn.
33814 (omp_simt_exit): Ditto.
33815 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33817 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33818 (lower_rec_input_clauses): Likewise.
33819 (lower_lastprivate_clauses): Handle SIMT privatization.
33821 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33822 (ompdevlow_adjust_simt_enter): New.
33823 (find_simtpriv_var_op): New.
33824 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33825 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33827 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33828 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33829 (copy_decl_for_dup_finish): Ditto.
33831 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33833 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33836 * config/i386/i386.c (ix86_option_override_internal): Always
33837 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33839 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33841 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33843 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33845 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33846 mark new edge's irreducible flag accordign to it.
33847 (vect_do_peeling): Check loop preheader edge's irreducible flag
33848 and pass it to function slpeel_add_loop_guard.
33850 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33852 PR tree-optimization/80218
33853 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33854 Update block frequencies and counts.
33856 2017-03-28 Richard Biener <rguenther@suse.de>
33858 PR tree-optimization/78644
33859 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
33860 of a simplification result we may not use it at all.
33862 2017-03-28 Richard Biener <rguenther@suse.de>
33865 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
33866 without arguments, generate default definition of a SSA name.
33868 2017-03-28 Richard Biener <rguenther@suse.de>
33870 PR middle-end/80222
33871 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
33872 TYPE_REF_CAN_ALIAS_ALL references.
33873 * fold-const.c (fold_indirect_ref_1): Likewise.
33875 2017-03-28 Martin Liska <mliska@suse.cz>
33878 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
33879 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
33881 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33882 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
33884 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
33885 (EXTRA_SPECS): Define.
33886 (SUBTARGET_EXTRA_SPECS): Likewise.
33887 (SUBTARGET_CPP_SPEC): Likewise.
33888 * config/arc/elf.h (EXTRA_SPECS): Renamed to
33889 SUBTARGET_EXTRA_SPECS.
33890 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
33892 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33894 * config/arc/simdext.md (vst64_insn): Update pattern.
33895 (vld32wh_insn): Likewise.
33896 (vld32wl_insn): Likewise.
33897 (vld64_insn): Likewise.
33898 (vld32_insn): Likewise.
33900 2017-03-28 Marek Polacek <polacek@redhat.com>
33903 * fold-const.c (fold_comparison): Use protected_set_expr_location
33904 instead of SET_EXPR_LOCATION.
33906 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
33908 * tree.c (add_expr): Avoid name lookup warning.
33910 2017-03-27 Jeff Law <law@redhat.com>
33912 PR tree-optimization/80216
33913 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
33914 function name. Limit recursion depth.
33915 (record_temporary_equivalences): Corresponding changes.
33917 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
33919 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
33922 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33925 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
33927 * cfgcleanup.c (reg_note_cfa_p): New array.
33928 (insns_have_identical_cfa_notes): New function.
33929 (old_insns_match_p): Don't cross-jump in between /f
33930 and non-/f instructions. If both i1 and i2 are frame related,
33931 verify all CFA notes, their order and content.
33933 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
33936 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
33937 HImode and SImode with zero extend to DImode to one insn.
33938 (bswap<mode>2_extenddi): Likewise.
33939 (bswapsi2_extenddi): Likewise.
33940 (bswaphi2_extendsi): Likewise.
33941 (bswaphi2): Combine bswap HImode and SImode into one insn.
33942 Separate memory insns from swapping register.
33943 (bswapsi2): Likewise.
33944 (bswap<mode>2): Likewise.
33945 (bswaphi2_internal): Delete, no longer used.
33946 (bswapsi2_internal): Likewise.
33947 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
33948 store, and gpr<-gpr swap insns.
33949 (bswap<mode>2_store): Likewise.
33950 (bswaphi2_reg): Register only splitter, combine with the splitter.
33951 (bswaphi2 splitter): Likewise.
33952 (bswapsi2_reg): Likewise.
33953 (bswapsi2 splitter): Likewise.
33954 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
33955 the insns into load, store, and register/register insns.
33956 (bswapdi2_ldbrx): Likewise.
33957 (bswapdi2_load): Likewise.
33958 (bswapdi2_store): Likewise.
33959 (bswapdi2_reg): Likewise.
33961 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
33963 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
33964 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
33966 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
33969 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
33971 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33972 special handling for target option conflicts between dform
33973 options (-mpower9-dform, -mpower9-dform-vector,
33974 -mpower9-dform-scalar) and -mno-direct-move.
33976 2017-03-27 Richard Biener <rguenther@suse.de>
33978 PR tree-optimization/80181
33979 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
33981 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33983 * config/arc/predicates.md (move_double_src_operand): Replace the
33984 call to move_double_src_operand with a call to address_operand.
33986 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33988 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
33989 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
33990 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
33992 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33994 * config/arc/predicates.md (long_immediate_loadstore_operand):
33995 Consider scaled addresses cases.
33997 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33999 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34000 restored when in interrupt.
34001 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34002 doesn't have delay slot.
34004 2017-03-27 Richard Biener <rguenther@suse.de>
34007 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34008 inlined thunk clones.
34010 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34013 * asan.c (instrument_derefs): Copy over last operand from
34014 original COMPONENT_REF to the new COMPONENT_REF with
34015 DECL_BIT_FIELD_REPRESENTATIVE.
34016 * ubsan.c (instrument_object_size): Likewise.
34018 2017-03-27 Richard Biener <rguenther@suse.de>
34020 PR tree-optimization/80170
34021 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34022 sure DR/SCEV didnt fold in constants we do not see when looking
34023 at the reference base alignment.
34025 2017-03-27 Richard Biener <rguenther@suse.de>
34027 PR middle-end/80171
34028 * gimple-fold.c (fold_ctor_reference): Properly guard against
34029 NULL return value from canonicalize_constructor_val.
34031 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34034 * config/i386/i386.c (ix86_expand_builtin)
34035 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34036 flags reg setting and flags reg using instructions.
34037 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34038 clobbering instructions to zero extend op2.
34040 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34042 * doc/install.texi (Configuration) <--with-aix-soname>:
34043 Update link to AIX ld.
34045 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34047 PR rtl-optimization/80160
34048 PR rtl-optimization/80159
34049 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34050 reg_alternate_class into account.
34052 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34055 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34056 to consider in curr_insn_transform.
34058 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34060 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34061 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34062 and emit_mode_inner.
34064 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34066 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34067 argument to the overloaded builtin variants. Use the new flag to
34068 deprecate certain builtin variants.
34069 * config/s390/s390-builtin-types.def: Add new builtin types.
34070 * config/s390/s390-builtins.h: Support new flags field for
34071 overloaded builtins.
34072 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34073 (s390_macro_to_expand): Enable vector float data type.
34074 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34075 builtins by incrementing the __VEC__ version number.
34076 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34078 (s390_resolve_overloaded_builtin): Emit error messages depending
34079 on the builtin flags.
34080 * config/s390/s390.c (s390_expand_builtin): Support additional
34081 flags argument. Change error message to match the messages
34082 emitted in s390-c.c.
34083 * config/s390/s390.md: New UNSPEC_* constants.
34084 (op_type): Add new instruction types.
34085 * config/s390/vecintrin.h: Add new builtins and test data class
34087 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34088 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34089 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34090 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34091 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34092 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34094 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34095 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34096 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34097 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34099 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34100 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34101 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34102 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34104 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34105 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34106 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34107 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34108 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34109 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34110 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34112 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34113 ("vec_scatter_element<V_HW_4:mode>_DI")
34114 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34115 ("vec_fpint<mode>", "vflls")
34116 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34117 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34118 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34119 ("*vec_cmphe<mode>_cc"): ... these.
34121 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34122 mode constant instead of magic value.
34124 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34126 * config/s390/s390.c (s390_expand_vec_compare): Support other
34127 vector floating point modes than just V2DF.
34128 (s390_expand_vcond): Likewise.
34129 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34130 (s390_cannot_change_mode_class): Prevent mode changes between TF
34131 and V1TF in vector registers.
34132 * config/s390/s390.md (DF, SF): New mode attributes.
34133 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34134 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34135 SFmode support for VRs.
34136 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34138 (VFT, VF_HW): New mode iterators.
34139 (vw, sdx): New mode attributes.
34140 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34141 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34142 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34143 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34144 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34145 also the new vector floating point modes. Renaming to ...
34147 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34148 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34149 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34150 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34151 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34152 ("vec_unordered<mode>"): ... these.
34154 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34155 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34156 ("*vec_extendv2df"): New insn definitions.
34158 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34160 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34161 ("mulditi3_2", "*muldi3_sign"): New patterns.
34162 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34163 rename the pattern definition.
34165 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34167 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34169 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34171 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34173 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34174 instruction if possible.
34175 * config/s390/vector.md (vec_halfnumelts): New mode
34177 ("*vec_vllezlf<mode>"): New pattern.
34179 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34181 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34182 ("popcountv4si2", "popcountv2di2"): Rename to ...
34183 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34184 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34186 ("popcount<mode>2_vxe"): New pattern.
34188 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34190 * common/config/s390/s390-common.c (processor_flags_table): Add
34192 * config.gcc: Add arch12.
34193 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34194 Default to arch12 for unknown CPU model numbers.
34195 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34196 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34197 PROCESSOR_max sanity check.
34198 * config/s390/s390-opts.h (enum processor_type): Add
34200 * config/s390/s390.c (processor_table): Add arch12.
34201 (s390_expand_builtin): Add check for B_VXE flag.
34202 (s390_issue_rate): Add PROCESSOR_ARCH12.
34203 (s390_get_sched_attrmask): Likewise.
34204 (s390_get_unit_mask): Likewise.
34205 (s390_sched_score): Enable z13 scheduling for arch12.
34206 (s390_sched_reorder): Likewise.
34207 (s390_sched_variable_issue): Likewise.
34208 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34210 (s390_tune_attr): Use z13 scheduling also for arch12.
34211 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34212 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34213 (TARGET_VXE_P): New macros.
34214 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34215 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34216 * config/s390/s390.opt: Add arch12 as processor_type.
34218 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34220 * config/s390/s390.md
34221 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34222 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34223 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34225 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34226 Rename expanders to ...
34228 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34229 ("fixuns_truncdddi2_emu"): ... these.
34231 ("fixuns_trunc<mode>si2_emu"): New expander.
34233 ("*fixuns_truncdfdi2_z13"): Rename to ...
34234 ("*fixuns_truncdfdi2_vx"): ... this.
34236 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34238 * config/s390/2964.md: Remove the single element vector compare
34239 instructions which are no longer used.
34240 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34242 (s390_canonicalize_comparison): Remove handling of DFmode
34244 (s390_expand_vec_compare_scalar): Remove function.
34245 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34246 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34248 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34250 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34252 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34254 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34255 will anyway by matched by mov<mode>_64dfp.
34257 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34259 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34260 vlef/vstef. Add missing operand to vleif.
34262 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34264 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34265 pair for all vector types with 64 bit elements.
34266 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34267 * config/s390/vector.md (V_HW_64): ... here.
34268 (V_128_NOSINGLE): New mode iterator.
34269 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34270 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34271 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34272 ("*vec_load_pairv2di"): Change to ...
34273 ("*vec_load_pair<mode>"): ... this one.
34275 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34277 * config/s390/constraints.md: Add comments.
34278 (jKK): Reject element sizes > 8 bytes.
34279 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34281 * config/s390/s390.md: Add the s_operand checks formerly in
34282 s390_split_ok_p to various splitters where they are still
34284 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34285 for 128 bit vectors. Plus two splitters.
34287 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34289 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34292 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34295 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34296 error if the boundary argument is not constant.
34298 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34300 PR rtl-optimization/80112
34301 * loop-doloop.c (doloop_condition_get): Don't check condition
34302 if cmp isn't SET with IF_THEN_ELSE src.
34304 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34306 PR tree-optimization/80158
34307 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34308 replacing a candidate statement, also replace it for the
34309 candidate's alternate interpretation.
34310 (replace_rhs_if_not_dup): Likewise.
34311 (replace_one_candidate): Likewise.
34313 2017-03-24 Richard Biener <rguenther@suse.de>
34315 PR tree-optimization/80167
34316 * graphite-isl-ast-to-gimple.c
34317 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34319 (translate_isl_ast_to_gimple::get_rename): Likewise.
34321 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34323 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34324 handling of certain combinations of target options, including the
34325 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34326 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34328 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34331 * config/arm/arm.md (*load_multiple): Add reload_completed to
34332 matching condition.
34334 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34335 Richard Biener <rguenth@suse.de>
34337 PR tree-optimization/79908
34338 PR tree-optimization/80136
34339 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34340 been cast away, gimplify_and_add suffices.
34342 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34344 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34346 2017-03-23 Richard Biener <rguenther@suse.de>
34348 PR tree-optimization/80032
34349 * gimplify.c (gimple_push_cleanup): Forced unconditional
34350 cleanups still have to go to the conditional_cleanups
34353 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34355 PR tree-optimization/80072
34356 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34358 (next_operand_entry_id): Change type to unsigned int.
34359 (sort_by_operand_rank): Make sure to return the right return value
34360 even if unsigned fields are bigger than INT_MAX.
34361 (struct oecount): Change cnt and id type to unsigned int.
34362 (oecount_hasher::equal): Formatting fix.
34363 (oecount_cmp): Make sure to return the right return value
34364 even if unsigned fields are bigger than INT_MAX.
34365 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34368 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34369 TREE_READONLY on result if writing it more than once.
34372 * doc/invoke.texi (-fsanitize=thread): Document that with
34373 -fnon-call-exceptions atomics are not able to throw
34377 * tsan.c: Include tree-eh.h.
34378 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34379 maybe_clean_or_replace_eh_stmt where needed.
34380 (instrument_memory_accesses): Add cfg_changed argument.
34381 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34382 if it returned true.
34383 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34385 PR rtl-optimization/63191
34386 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34387 wrapper function, moved the whole old content into ...
34388 (ix86_delegitimize_address_1): ... this. New inline function.
34389 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34390 true as last argument instead of ix86_delegitimize_address.
34392 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34394 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34395 cortexa57_branch_cost.
34397 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34399 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34401 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34404 * doc/md.texi (Constraints): Document wA constraint.
34405 * config/rs6000/constraints.md (wA): New.
34406 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34407 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34408 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34409 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34411 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34414 * gimplify.c (is_oacc_declared): New function.
34415 (oacc_default_clause): Use it to set default flags for acc declared
34416 variables inside parallel regions.
34417 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34418 declared variables.
34419 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34420 declare attribute to any decl as necessary.
34422 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34425 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34426 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34427 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34428 (arm_arch_lpae): This.
34429 * config/arm/arm.c (arm_arch7ve): Rename into ...
34430 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34431 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34434 2017-03-22 Martin Liska <mliska@suse.cz>
34437 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34438 error message instead of an ICE.
34440 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34442 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34444 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34446 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34448 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34451 2017-03-21 Martin Sebor <msebor@redhat.com>
34453 * doc/extend.texi: Use "cannot" instead of "can't."
34454 * doc/hostconfig.texi: Same.
34455 * doc/install.texi: Same.
34456 * doc/invoke.texi: Same.
34457 * doc/loop.texi: Same.
34458 * doc/md.texi: Same.
34459 * doc/objc.texi: Same.
34460 * doc/rtl.texi: Same.
34461 * doc/tm.texi: Same.
34462 * doc/tm.texi.in: Same.
34463 * doc/trouble.texi: Same.
34465 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34468 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34469 (collect_checksum_attributes): Set it.
34470 (die_checksum_ordered): Use it.
34472 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34474 PR tree-optimization/79908
34475 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34476 change: For a VA_ARG whose LHS has been cast away, use
34477 force_gimple_operand to construct the side effects.
34479 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34481 PR translation/80001
34482 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34483 more amenable to translation.
34484 (oacc_loop_auto_partitions): Likewise.
34486 2017-03-21 Marek Polacek <polacek@redhat.com>
34487 Martin Sebor <msebor@redhat.com>
34489 PR tree-optimization/80109
34490 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34491 on INTEGRAL_TYPE_P.
34493 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34494 Segher Boessenkool <segher@kernel.crashing.org>
34497 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34498 check reg_used_between_p between insn and one of succ or succ2
34499 depending on if succ is artificial insn not inserted into insn
34502 2017-03-21 Martin Liska <mliska@suse.cz>
34504 PR gcov-profile/80081
34505 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34506 * doc/gcc.texi: Include gcov-dump stuff.
34507 * doc/gcov-dump.texi: New file.
34509 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34511 PR rtl-optimization/79150
34512 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34513 conditional jump, if the jump is the last insn of the loop.
34515 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34516 Richard Biener <rguenth@suse.de>
34518 PR tree-optimization/79908
34519 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34520 been cast away, use force_gimple_operand to construct the side
34523 2017-03-21 Martin Liska <mliska@suse.cz>
34525 PR libfortran/79956
34526 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34529 2017-03-21 Brad Spengler <spender@grsecurity.net>
34532 * plugin.c (htab_hash_plugin): New function.
34533 (add_new_plugin): Use it and adjust.
34534 (parse_plugin_arg_opt): Adjust.
34535 (init_one_plugin): Likewise.
34537 2017-03-21 Richard Biener <rguenther@suse.de>
34539 PR tree-optimization/80032
34540 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34541 if set force the cleanup to happen unconditionally.
34542 (gimplify_target_expr): Push inserted clobbers with force_uncond
34543 to avoid them being removed by control-dependent DCE.
34545 2017-03-21 Richard Biener <rguenther@suse.de>
34547 PR tree-optimization/80122
34548 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34549 va_arg_pack_len when the inlined call stmt requires pack
34551 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34553 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34556 * tsan.c (instrument_builtin_call): If the memory model argument
34557 is not a constant, assume it is valid.
34560 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34563 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34565 PR rtl-optimization/79910
34566 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34567 if its dest is used by an insn before I2 (other than the combined
34568 insns themselves, which are properly handled already).
34570 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34573 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34575 * combine.c (record_used_regs): New static function.
34576 (try_combine): Handle situations where there is an additional
34577 instruction between I2 and I3 which needs to have a LOG_LINK
34581 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34583 * combine.c (try_combine): Delete redundant i1 test. Call
34584 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34586 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34589 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34590 alternatives 13/14.
34592 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34594 PR tree-optimization/80054
34595 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34596 the optimization if a PHI or any of its arguments is not dominated
34597 by the candidate's basis. Use gphi* rather than gimple* as
34599 (replace_profitable_candidates): Clean up a gimple* variable that
34600 should be a gphi* variable.
34602 2017-03-20 Martin Sebor <msebor@redhat.com>
34605 * doc/extend.texi (attribute constructor): Document present limitation.
34607 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34610 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34611 __POWER9_VECTOR__ #ifdef control, change template definition to
34612 use Power9-specific built-in function.
34613 (vec_any_eq): Likewise.
34614 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34615 to control outcomes from this test.
34616 (vector_ae_<mode>p): For VEC_F modes, likewise.
34618 2017-03-20 Ian Lance Taylor <iant@google.com>
34620 * config/i386/i386.c (ix86_function_regparm): Save an extra
34621 register for -fsplit-stack with DECL_STATIC_CHAIN.
34623 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34626 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34627 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34629 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34631 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34633 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34636 2017-03-20 Marek Polacek <polacek@redhat.com>
34639 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34641 2017-03-20 Richard Biener <rguenther@suse.de>
34643 PR tree-optimization/80113
34644 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34645 allocate extra SSA name for PHI def.
34646 (add_close_phis_to_outer_loops): Likewise.
34647 (add_close_phis_to_merge_points): Likewise.
34648 (copy_loop_close_phi_args): Likewise.
34649 (copy_cond_phi_nodes): Likewise.
34651 2017-03-20 Martin Liska <mliska@suse.cz>
34653 PR middle-end/79753
34654 * tree-chkp.c (chkp_build_returned_bound): Do not build
34655 returned bounds for a LHS that's not a BOUNDED_P type.
34657 2017-03-20 Martin Liska <mliska@suse.cz>
34661 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34662 COMPLEX_CST and VECTOR_CST.
34664 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34667 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34668 target operand. A new splitter adds the clobber statement in case
34669 the target operand is dead anyway.
34671 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34673 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34674 to age-old versions of binutils and glibc.
34676 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34678 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34680 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34682 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34684 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34686 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34687 requirement for binutils 2.13.
34689 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34691 * combine.c (try_combine): Delete redundant i1 test. Call
34692 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34694 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34696 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34697 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34699 <riscv64-*-elf>: Re-arrange section
34700 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34701 <riscv32-*-linux>: Likewise.
34702 <riscv64-*-elf>: Likewise
34703 <riscv64-*-linux>: Likewise.
34705 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34708 * aarch64.opt(verbose-cost-dump): Fix typo.
34710 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34713 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34714 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34716 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34718 * reload.c (find_reloads): When reloading a nonoffsettable address,
34719 use RELOAD_OTHER for it and its address reloads.
34721 PR rtl-optimization/79910
34722 * combine.c (record_used_regs): New static function.
34723 (try_combine): Handle situations where there is an additional
34724 instruction between I2 and I3 which needs to have a LOG_LINK
34727 2017-03-17 Jeff Law <law@redhat.com>
34729 PR tree-optimization/71437
34730 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34731 conditional in the hash table first.
34732 (vrp_dom_walker::before_dom_children): Extract condition from
34733 ASSERT_EXPR. Record condition, its inverion and any implied
34734 conditions as well.
34736 2017-03-17 Marek Polacek <polacek@redhat.com>
34737 Markus Trippelsdorf <markus@trippelsdorf.de>
34739 PR tree-optimization/80079
34740 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34743 2017-03-17 Richard Biener <rguenther@suse.de>
34745 PR middle-end/80075
34746 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34747 Properly verify the LHS before the RHS possibly claims to be
34749 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34752 2017-03-17 Martin Jambor <mjambor@suse.cz>
34754 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34755 (List of -O2 options): Likewise.
34756 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34759 2017-03-17 Tom de Vries <tom@codesourcery.com>
34761 * gcov-dump.c (print_usage): Print bug_report_url.
34763 2017-03-17 Richard Biener <rguenther@suse.de>
34765 PR middle-end/80050
34766 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34767 (parser::peek): Likewise.
34769 2017-03-17 Richard Biener <rguenther@suse.de>
34771 PR tree-optimization/80048
34772 * sese.c (free_sese_info): Properly release rename_map and
34773 copied_bb_map elements.
34775 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34777 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34778 Add linked-list forward and backlinks. Insert on
34779 construction, remove on destruction.
34780 (class pass_store_merging): Add m_stores_head field.
34781 (pass_store_merging::terminate_and_process_all_chains):
34782 Iterate over m_stores_head list.
34783 (pass_store_merging::terminate_all_aliasing_chains):
34785 (pass_store_merging::execute): Check for debug stmts first.
34786 Push new chains onto the m_stores_head stack.
34788 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34791 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34792 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34793 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34795 2017-03-16 Jeff Law <law@redhat.com>
34797 PR tree-optimization/71437
34798 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34799 member function. Implementation moved into after_dom_children
34800 member function and into the threader's thread_outgoing_edges
34802 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34803 some code into new thread_outgoing_edges.
34804 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34805 definition. Simplify marker handling (do it here). Assume we always
34806 have the available expression and the const/copies tables.
34807 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34809 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34810 * tree-vrp.c (equiv_stack): No longer file scoped.
34811 (vrp_dom_walker): New class.
34812 (vrp_dom_walker::before_dom_children): New member function.
34813 (vrp_dom_walker::after_dom_children): Likewise.
34814 (identify_jump_threads): Setup domwalker. Use it rather than
34815 walking edges in a random order by hand. Simplify setup/finalization.
34816 (finalize_jump_threads): Remove.
34817 (vrp_finalize): Do not call identify_jump_threads here.
34818 (execute_vrp): Do it here instead and call thread_through_all_blocks
34821 PR tree-optimization/71437
34822 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34824 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34826 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34827 (dom_opt_dom_walker::thread_across_edge): Remove
34828 handle_dominating_asserts argument. All callers changed.
34829 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34830 changes. Remove calls to lhs_of_dominating_assert. Other
34831 uses of handle_dominating_asserts turn into unconditional code
34832 (simplify_control_stmt_condition_1): Likewise.
34833 (simplify_control_stmt_condition): Likewise.
34834 (thread_through_normal_block, thread_across_edge): Likewise.
34835 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34836 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34837 object if it is not an SSA_NAME.
34838 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34839 before calling into the VRP specific simplifiers.
34840 (identify_jump_threads): Remove handle_dominating_asserts
34843 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34846 * tree-diagnostic.c (default_tree_printer): No longer static.
34847 * tree-diagnostic.h (default_tree_printer): New prototype.
34849 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34851 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34852 Change ins into fmov.
34854 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34856 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34857 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34858 Use h_con constraint for operand 1.
34859 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
34860 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
34862 2017-03-15 Jeff Law <law@redhat.com>
34864 PR tree-optimization/71437
34865 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
34866 (record_temporary_equivalences): Use it.
34868 PR tree-optimization/71437
34869 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
34870 tree-ssa-scopedtables.
34871 (lookup_avail_expr, build_and_record_new_cond): Likewise.
34872 (record_conditions, record_cond, vuse_eq): Likewise.
34873 (record_edge_info): Adjust to API tweak of record_conditions.
34874 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
34875 (record_temporary_equivalences, optimize_stmt): Likewise.
34876 (eliminate_redundant_computations): Likewise.
34877 (record_equivalences_from_stmt): Likewise.
34878 * tree-ssa-scopedtables.c: Include options.h and params.h.
34879 (vuse_eq): New function, moved from tree-ssa-dom.c
34880 (build_and_record_new_cond): Likewise.
34881 (record_conditions): Likewise. Accept vector of conditions rather
34882 than edge_equivalence structure for first argument.
34883 for the first argument.
34884 (avail_exprs_stack::lookup_avail_expr): New member function, moved
34885 from tree-ssa-dom.c.
34886 (avail_exprs_stack::record_cond): Likewise.
34887 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
34888 from tree-ssa-dom.c.
34889 (avail_exprs_stack): Add new member functions lookup_avail_expr
34891 (record_conditions): Declare.
34893 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
34896 * lra-constraints.c (process_alt_operands): Increase reject for
34897 reloading an input/output operand.
34899 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
34902 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
34903 insns to convert from signed/unsigned char/short to IEEE 128-bit
34905 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
34907 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
34910 * config/i386/i386.c (ix86_vector_duplicate_value): Create
34911 subreg of inner mode for values already in registers.
34913 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
34915 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
34916 iteration reg is used after the loop.
34918 2017-03-14 Martin Sebor <msebor@redhat.com>
34920 PR tree-optimization/79800
34921 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
34922 precision in negative-positive range.
34923 (format_floating): Call non-const overload with adjusted precision.
34925 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34928 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
34931 2017-03-14 Martin Sebor <msebor@redhat.com>
34933 PR middle-end/80020
34934 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
34935 * builtins.def (aligned_alloc): Use it.
34938 * Makefile.in (GTFILES): Add calls.c.
34939 * calls.c: Include "gt-calls.h".
34941 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
34943 PR rtl-optimization/79728
34944 * regs.h (struct target_regs): New field
34945 x_contains_allocatable_regs_of_mode.
34946 (contains_allocatable_regs_of_mode): New macro.
34947 * reginfo.c (init_reg_sets_1): Initialize it, and change
34948 contains_reg_of_mode so it includes global regs as well.
34949 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
34950 rather than contains_regs_of_mode.
34952 2017-03-14 Martin Liska <mliska@suse.cz>
34954 * doc/invoke.texi: Document options that can't be combined with
34955 -fcheck-pointer-bounds.
34957 2017-03-14 Martin Liska <mliska@suse.cz>
34959 PR middle-end/79831
34960 * doc/invoke.texi (-Wchkp): Document the option.
34962 2017-03-14 Martin Liska <mliska@suse.cz>
34964 * Makefile.in: Install gcov-dump.
34966 2017-03-14 Martin Liska <mliska@suse.cz>
34968 * multiple_target.c (expand_target_clones): Bail out for
34969 an invalid attribute.
34971 2017-03-14 Richard Biener <rguenther@suse.de>
34973 * alias.c (struct alias_set_entry): Pack properly.
34974 * cfgloop.h (struct loop): Likewise.
34975 * cse.c (struct set): Likewise.
34976 * ipa-utils.c (struct searchc_env): Likewise.
34977 * loop-invariant.c (struct invariant): Likewise.
34978 * lra-remat.c (struct cand): Likewise.
34979 * recog.c (struct change_t): Likewise.
34980 * rtl.h (struct address_info): Likewise.
34981 * symbol-summary.h (function_summary): Likewise.
34982 * tree-loop-distribution.c (struct partition): Likewise.
34983 * tree-object-size.c (struct object_size_info): Likewise.
34984 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
34985 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
34986 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
34987 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
34988 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
34989 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
34990 (struct _stmt_vec_info): Likewise.
34992 2017-03-14 Martin Liska <mliska@suse.cz>
34995 * multiple_target.c (create_dispatcher_calls): Check that
34996 a target can create a function dispatcher.
34998 2017-03-14 Martin Liska <mliska@suse.cz>
35001 * multiple_target.c (expand_target_clones): Drop local.local
35002 flag for default implementation.
35004 2017-03-14 Richard Biener <rguenther@suse.de>
35006 PR tree-optimization/80030
35007 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35009 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35011 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35012 gcc_fallthrough() instead of __attribute__((fallthrough));
35014 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35016 * doc/gcc.texi: Remove "up" link to (DIR).
35017 * doc/gccint.texi: Ditto.
35019 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35021 * doc/install.texi (Specific) <avr>: Remove reference to
35024 2017-03-13 Jeff Law <law@redhat.com>
35026 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35027 attribute rather than comments.
35029 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35030 match_scratch operand is highest.
35032 2017-03-13 Martin Liska <mliska@suse.cz>
35034 PR middle-end/78339
35035 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35036 is a CHKP clone, use original declaration.
35038 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35040 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35041 (arc_conditional_register_usage): Use a different allocation order
35042 when optimizing for size.
35043 * common/config/arc/arc-common.c (arc_option_optimization_table):
35044 Section anchors default on when optimizing for size.
35046 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35048 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35050 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35052 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35053 * config/arc/arc.md (cpu_facility): Add cd variant.
35054 (*movqi_insn): Add code density variant.
35055 (*movhi_insn): Likewise.
35056 (*movqi_insn): Likewise.
35057 (*addsi3_mixed): Likewise.
35058 (subsi3_insn): Likewise.
35060 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35062 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35064 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35066 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35067 expressions with MINUS and UNARY ops.
35069 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35072 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35074 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35075 between vec_select and vector argument.
35076 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35077 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35078 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35079 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35080 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35081 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35083 2017-03-13 Richard Biener <rguenther@suse.de>
35086 * params.def (vect-max-peeling-for-alignment): Fix typo.
35088 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35090 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35091 issue that only occurred with binutils below 2.18.
35093 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35095 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35096 refer to binutils 2.11/2.12 minimum.
35098 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35100 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35101 ftp.kernel.org and simplify binutils requirement.
35103 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35105 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35107 (Optimize Options): Ditto. Also remove redundancy.
35109 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35111 PR translation/79848
35112 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35114 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35115 to G_ to avoid double translation.
35117 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35119 PR translation/79923
35120 * auto-profile.c (get_combined_location): Convert leading
35121 character of diagnostics to lower case and remove trailing period.
35122 (read_profile): Likewise for various diagnostics.
35123 * config/arm/arm.c (arm_option_override): Remove trailing period
35124 from various diagnostics.
35125 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35126 (msp430_expand_delay_cycles): Likewise.
35128 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35131 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35132 full command-line argument, rather than just "str".
35133 (aarch64_validate_march): Likewise.
35134 (aarch64_validate_mtune): Likewise.
35136 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35138 PR rtl-optimization/78911
35139 * lra-assigns.c (must_not_spill_p): New function.
35140 (spill_for): Use it.
35142 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35144 PR tree-optimization/79981
35145 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35146 ATOMIC_COMPARE_EXCHANGE ifn result.
35147 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35148 IFN_ATOMIC_COMPARE_EXCHANGE.
35150 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35153 * opts.c (parse_sanitizer_options): Add missing question mark to
35154 "did you mean" message.
35156 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35158 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35160 (VMULEUH_UNS): Likewise.
35161 (VMULOUB_UNS): Likewise.
35162 (VMULOUH_UNS): Likewise.
35163 * config/rs6000/rs6000.c (builtin_function_type): Remove
35164 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35166 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35169 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35170 x with result of extra_parsing_for_operand_code_0.
35171 (function_reader::extra_parsing_for_operand_code_0): Convert
35172 return type from void to rtx, returning x. When reading
35173 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35174 larger size containing struct block_symbol.
35176 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35178 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35179 -mfloat128-hardware without -m64.
35181 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35184 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35185 entries to the case statement that marks unsigned arguments to
35186 overloaded functions.
35188 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35190 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35191 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35193 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35196 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35197 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35199 2017-03-10 Martin Liska <mliska@suse.cz>
35203 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35204 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35207 2017-03-10 Olivier Hainque <hainque@adacore.com>
35209 * tree-switch-conversion (array_value_type): Start by resetting
35210 candidate type to it's main variant.
35212 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35214 PR rtl-optimization/79909
35215 * combine.c (try_combine): Use simplify_replace_rtx on individual
35216 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35217 of the whole CALL_INSN_FUNCTION_USAGE.
35219 PR tree-optimization/79972
35220 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35221 get_range_info on SSA_NAMEs. Formatting fixes.
35223 2017-03-10 Richard Biener <rguenther@suse.de>
35224 Jakub Jelinek <jakub@redhat.com>
35226 PR tree-optimization/77975
35227 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35228 edge to be constant.
35229 (get_val_for): For constant x return it. Formatting fix.
35230 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35231 would use the same bases as the current one.
35233 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35235 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35236 instead of vec_select for V1TImode.
35237 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35239 (VSX_LE_128): Add V1TI to this mode iterator.
35240 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35241 (*vsx_le_perm_store_<mode>): Likewise.
35242 (pre-reload splitter for VSX stores): Likewise.
35243 (post-reload splitter for VSX stores): Likewise.
35244 (*vsx_xxpermdi2_le_<mode>): Likewise.
35245 (*vsx_lxvd2x2_le_<mode>): Likewise.
35246 (*vsx_stxvd2x2_le_<mode>): Likewise.
35248 2017-03-09 Michael Eager <eager@eagercon.com>
35250 Correct failures with --enable-checking=yes,rtl.
35252 * config/microblaze/microblaze.c (microblaze_expand_shift):
35253 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35254 test for const0_rtx.
35255 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35256 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35258 2017-03-09 Richard Biener <rguenther@suse.de>
35260 PR tree-optimization/79977
35261 * graphite-scop-detection.c (scop_detection::merge_sese):
35262 Handle the case of extra exits to blocks dominating the entry.
35264 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35266 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35269 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35271 PR rtl-optimization/79949
35272 * lra-constraints.c (process_alt_operands): Check memory when
35273 trying to predict a cycle. Print about the overall increase.
35275 2017-03-09 Richard Biener <rguenther@suse.de>
35277 PR middle-end/79971
35278 * gimple-expr.c (useless_type_conversion_p): Preserve
35279 TYPE_SATURATING for fixed-point types.
35281 2017-03-09 Richard Biener <rguenther@suse.de>
35284 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35285 alignment of BLKmode params.
35287 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35290 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35291 (VALL_NO_V2Q): Likewise.
35293 * config/aarch64/aarch64-simd.md
35294 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35296 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35297 VALL_NO_V2Q mode iterator.
35298 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35300 2017-03-09 Martin Liska <mliska@suse.cz>
35302 PR tree-optimization/79631
35303 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35304 tree_int_cst_sign_bit just for INTEGER constants.
35306 2017-03-09 Martin Liska <mliska@suse.cz>
35310 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35313 2017-03-09 Marek Polacek <polacek@redhat.com>
35316 * tree.c (inchash::add_expr): Handle TREE_VEC.
35318 2017-03-09 Martin Liska <mliska@suse.cz>
35321 (chkp_narrow_size_and_offset): New function.
35322 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35323 (void chkp_parse_bit_field_ref): New function.
35324 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35325 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35327 2017-03-09 Martin Liska <mliska@suse.cz>
35330 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35331 (chkp_find_bounds_1): Remove gcc_unreachable.
35333 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35336 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35337 BUILT_IN_SYNC*, determine the access type from the size suffix and
35338 always build a MEM_REF with that type. Handle forgotten
35339 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35342 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35343 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35344 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35345 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35346 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35347 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35348 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35349 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35350 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35351 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35352 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35353 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35354 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35355 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35356 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35357 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35358 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35359 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35360 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35361 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35362 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35363 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35364 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35365 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35366 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35367 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35368 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35369 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35370 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35371 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35372 definitions outside of __OPTIMIZE__ guarded section.
35375 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35376 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35377 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35378 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35381 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35383 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35384 ("vfenez<mode>"): Add missing constraints.
35386 2017-03-08 Martin Sebor <msebor@redhat.com>
35389 * config/nds32/nds32.c (nds32_option_override):
35390 Fix misspelled diagnostic.
35392 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35395 * gimplify.c (gimplify_omp_for): Replace index var in outer
35396 taskloop statement with an artificial variable and add
35397 OMP_CLAUSE_PRIVATE clause for it.
35399 2017-03-08 Richard Biener <rguenther@suse.de>
35401 PR tree-optimization/79955
35402 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35403 for accesses that are completely outside of the variable.
35405 2017-03-08 Andrew Haley <aph@redhat.com>
35407 PR tree-optimization/79943
35408 * tree-ssa-loop-split.c (compute_new_first_bound): When
35409 calculating the new upper bound, (END-BEG) should be added, not
35412 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35414 * config/avr/avr.md (setmemhi): Make sure match_dup
35415 operand number comes before match_scratch.
35417 2017-03-08 Richard Biener <rguenther@suse.de>
35419 PR tree-optimization/79920
35420 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35421 with ncopies == 1 to ...
35422 (vect_transform_slp_perm_load): ... here. Properly compute
35423 all element loads by iterating VF times over the group. Do
35424 not handle ncopies (computed in a broken way) in
35425 vect_create_mask_and_perm.
35427 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35430 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35431 is a uniform vector, use uniform_vector_p return value instead of
35432 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35434 2017-03-07 Marek Polacek <polacek@redhat.com>
35436 PR middle-end/79809
35437 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35438 (alloca_call_type): Likewise.
35440 2017-03-07 Martin Liska <mliska@suse.cz>
35442 * gcov.c (process_args): Put comment to correct location.
35444 2017-03-07 Martin Liska <mliska@suse.cz>
35446 PR middle-end/68270
35447 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35448 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35449 (chkp_narrow_bounds_for_field): Likewise.
35450 (chkp_parse_array_and_component_ref): Pass one more argument to
35453 2017-03-07 Richard Biener <rguenther@suse.de>
35455 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35458 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35460 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35461 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35463 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35466 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35467 to end of description.
35468 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35470 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35472 PR rtl-optimization/79901
35473 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35475 (*avx512f_<code><mode>3<mask_name>): ... this.
35476 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35477 iterator instead of VI8_AVX2_AVX512BW.
35479 PR rtl-optimization/79901
35480 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35481 min/max expander, expand it using expand_vec_cond_expr.
35484 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35487 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35490 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35491 to void * for PCH reasons.
35492 * dwarf2out.c (output_loc_operands, output_die): Cast
35493 v.val_vec.array to unsigned char *.
35495 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35498 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35501 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35503 PR rtl-optimization/79571
35504 * lra-constraints.c (process_alt_operands): Calculate static
35505 reject and subtract it from overall when only addresses will be
35508 2017-03-06 Julia Koval <julia.koval@intel.com>
35511 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35512 incoming stack boundary to 128 for 64-bit targets.
35514 2017-03-06 Richard Biener <rguenther@suse.de>
35516 PR tree-optimization/79894
35517 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35518 to NULL after folding it.
35520 2017-03-06 Richard Biener <rguenther@suse.de>
35522 PR tree-optimization/79824
35523 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35524 check disabling peeling for gaps.
35526 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35528 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35529 attributes): Document gettimeofday.
35531 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35533 * config/s390/s390.c (s390_option_override_internal): Set
35534 PARAM_MIN_VECT_LOOP_BOUND
35536 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35538 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35539 * config/s390/s390.md: Likewise.
35541 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35544 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35545 (<avx2_avx512>_perm<mode>): Rename to ...
35546 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35548 (<avx512>_perm<mode>_mask): Rename to ...
35549 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35551 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35552 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35553 instead of VI8F_256_512.
35554 (avx512f_perm<mode>): New define_expand.
35555 (avx512f_perm<mode>_mask): Likewise.
35556 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35557 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35559 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35561 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35562 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35564 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35566 2017-03-06 Martin Liska <mliska@suse.cz>
35569 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35570 when having a SSA NAME w/o VAR_DECL assigned to it.
35572 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35574 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35575 msa_dpsub_<su>_d): Fix MODE for vec_select.
35577 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35579 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35581 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35583 2017-03-06 Richard Biener <rguenther@suse.de>
35585 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35586 * plugin.c (register_plugin_info): Likewise.
35587 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35589 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35591 * config/i386/sse.md (sse_storehps, sse_storelps,
35592 avx_<castmode><avxsizesuffix>_<castmode>,
35593 avx512f_<castmode><avxsizesuffix>_<castmode>,
35594 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35595 in condition that at least one operand is not a MEM.
35597 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35599 PR middle-end/79805
35600 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35601 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35603 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35604 gimple_call_nothrow_p flag based on whether original builtin can throw.
35605 If it can, emit following stmts on the fallthrough edge.
35606 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35607 don't create new bb if inserting just debug stmts on the edge, try to
35608 insert them on the fallthru bb or just reset debug stmts.
35610 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35613 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35614 restore recog_data (including the operand rtxes inside it) around
35615 the call to get_insn_template.
35617 2017-03-03 Martin Sebor <msebor@redhat.com>
35619 PR tree-optimization/79699
35620 * context.c (context::~context): Free MPFR caches to avoid
35621 a memory leak on program exit.
35623 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35625 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35626 Use wide_int::ulow () instead of .elt (0).
35628 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35630 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35631 (*pushxf): Limit oF constraint to 32bit targets and add oC
35632 constraint for 64bit targets.
35633 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35634 (*pushdf): Change rmF constraint to rmC.
35636 2017-03-03 Martin Liska <mliska@suse.cz>
35638 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35639 Remove unused variable.
35641 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35644 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35645 is a memory operand, increase num_memory.
35646 (ix86_expand_args_builtin): Likewise.
35648 2017-03-03 Jan Hubicka <jh@suse.cz>
35651 * ipa-devirt.c (maybe_record_node): Properly handle
35652 __cxa_pure_virtual visibility.
35654 2017-03-03 Martin Liska <mliska@suse.cz>
35656 PR tree-optimization/79803
35657 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35659 (pass_loop_prefetch::execute): Disabled optimization if an
35660 assumption about L1 cache size is not met.
35662 2017-03-03 Martin Liska <mliska@suse.cz>
35664 PR rtl-optimization/79574
35665 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35666 (hash_scan_set): Likewise.
35667 (dump_hash_table): Likewise.
35668 (hoist_code): Likewise.
35670 2017-03-03 Richard Biener <rguenther@suse.de>
35672 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35673 in place of uhigh ().
35674 (fixed_convert_from_real): Likewise.
35676 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35679 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35681 2017-03-03 Richard Biener <rguenther@suse.de>
35683 PR middle-end/79818
35684 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35685 TYPE_OVERFLOW_UNDEFINED check.
35687 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35689 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35691 (vector_ae_<mode>_p): Likewise.
35692 (vector_nez_<mode>_p): Likewise.
35693 (vector_ne_v2di_p): Likewise.
35694 (vector_ae_v2di_p): Likewise.
35695 (vector_ne_<mode>_p): Likewise.
35696 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35698 (vsx_tsqrt<mode>2_fe): Likewise.
35700 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35703 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35705 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35707 PR rtl-optimization/79780
35708 * cprop.c (one_cprop_pass): When second and further conditional trap
35709 in a single basic block is turned into an unconditional trap, turn it
35710 into a deleted note to avoid RTL verification failures.
35712 2017-03-02 Richard Biener <rguenther@suse.de>
35714 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35716 2017-03-02 Richard Biener <rguenther@suse.de>
35718 PR tree-optimization/79345
35720 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35721 param and abort the walk, returning -1 if it is hit.
35722 (walk_aliased_vdefs): Take a limit param and pass it on.
35723 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35724 defaulting to 0 and return a signed int.
35725 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35726 (check_defs): New helper.
35727 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35728 about uninitialized memory.
35729 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35730 bogus uninitialized warning.
35731 (fixed_convert_from_real): Likewise.
35733 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35735 PR tree-optimization/66768
35736 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35737 iv_use if base object can't be determined.
35739 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35741 PR tree-optimization/79345
35742 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35743 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35744 (get_pattern_stats): Initialize it.
35745 * genemit.c (gen_expand): Verify match_scratch numbers come after
35746 match_operand/match_dup numbers.
35747 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35748 match_scratch numbers.
35749 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35751 * config/s390/s390.md (trunctdsd2): Likewise.
35753 2017-03-02 Richard Biener <rguenther@suse.de>
35755 * wide-int.h (wide_int_storage::operator=): Implement in terms
35758 2017-03-02 Richard Biener <rguenther@suse.de>
35760 PR tree-optimization/79777
35761 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35762 the to insert expression to sth existing.
35764 2017-03-01 Martin Sebor <msebor@redhat.com>
35766 PR middle-end/79692
35767 * gimple-ssa-sprintf.c
35768 (directive::known_width_and_precision): New function.
35769 (format_integer): Use it.
35770 (get_mpfr_format_length): Consider the full range of precision
35771 when computing %g output with the # flag. Set the likely byte
35772 count to 3 rather than 1 when precision is indeterminate.
35773 (format_floating): Correct the lower bound of precision.
35775 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35777 * doc/invoke.texi: Document default code model for 64-bit Linux.
35779 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35782 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35783 udiv rather than div since input pattern is unsigned.
35785 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35787 * config/i386/i386.c (print_reg): Warn for values of
35788 unsupported size in integer register.
35790 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35793 * config/rs6000/predicates.md (current_file_function_operand): Do
35794 not allow self calls to be local if the function is replaceable.
35796 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35799 * config/rs6000/altivec.h (vec_ctz and others): Change the
35800 preprocessor macro that controls conditional compilation from
35801 _ARCH_PWR9 to __POWER9_VECTOR__.
35802 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35803 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35804 control (instead of _ARCH_PWR9 control) so that template
35805 definition uses power9-specific function.
35806 (vec_any_eq): Likewise.
35807 (vec_all_ne): Change macro definition to use a power9-specific
35808 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35809 _ARCH_PWR9 control).
35810 (vec_any_eq) Likewise.
35811 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35812 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35813 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35814 support for xvcmpnedp instruction.
35815 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35816 macro expansion so that Power9 implementation of vec_all_ne does
35817 not use the AltiVec predicate framework.
35818 (VCMPNEH_P): Likewise.
35819 (VCMPNEW_P): Likewise.
35820 (VCMPNED_P): Likewise.
35821 (VCMPNEFP_P): Likewise.
35822 (VCMPNEDP_P): Likewise.
35823 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35824 implementation of vec_any_eq to not use AltiVec predicate
35826 (VCMPAEH_P): Likewise.
35827 (VCMPAEW_P): Likewise.
35828 (VCMPAED_P): Likewise.
35829 (VCMPAEFP_P): Likewise.
35830 (VCMPAEDP_P): Likewise.
35831 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35832 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35833 not use the AltiVec predicate framework.
35834 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35835 of vec_any_eq to not use AltiVec predicate framework.
35836 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35837 support for predefined __POWER9_VECTOR__ macro to indicate that
35838 Power9 instruction selection is enabled.
35839 (altivec_overloaded_builtins): Remove extraneous
35840 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35841 function argument types RS6000_BTI_bool_V16QI and
35842 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35843 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35844 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35845 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35846 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35847 Power9 for implementations of vec_cmpne. Change the signature for
35848 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35849 (representing vec_all_ne) to remove the previously described first
35850 argument of type RS6000_BTI_INTSI, as this was an artifact of
35851 reliance on the AltiVec predicate framework, which is no longer
35852 used in the implementation of these functions. Add
35853 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35854 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35855 since, unlike the AltiVec predicate framework implementation, we
35856 do not share function descriptors between vec_alle and vec_anyeq.
35857 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35858 set of modes that receive special treatment even when
35859 TARGET_P9_VECTOR is true. The special treatment emits code that
35860 does not depend on Power9 instructions.
35861 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
35862 define_expand to not rely on AltiVec predicate framework.
35863 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35865 (vector_ne_v2di_p): Change this define_expand to not rely on
35866 AltiVec predicate framework.
35867 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
35869 (vector_ne_<mode>_p): Change this define_expand to not rely on
35870 AltiVec predicate framework.
35871 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35873 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
35874 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
35875 define_insn pattern.
35876 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
35877 define_insn pattern because the xvcmpne<VSs>. instruction is not
35879 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
35880 instruction is not supported.
35882 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35884 * config/nvptx/nvptx.c: Include intl.h.
35886 2017-03-01 Martin Jambor <mjambor@suse.cz>
35889 * ipa-prop.h (ipa_bits): Removed field known.
35890 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
35891 to pointers. Adjusted their comments to warn about their sharing.
35892 (ipcp_transformation_summary): Change bits to a vector of pointers.
35893 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
35894 (ipa_get_ipa_bits_for_value): Declare.
35895 * tree-vrp.h (value_range): Mark as GTY((for_user)).
35896 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
35897 (ipa_bits_hash_table): Likewise.
35898 (ipa_vr_ggc_hash_traits): Likewise.
35899 (ipa_vr_hash_table): Likewise.
35900 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
35901 being pointers and vr_known being removed.
35902 (ipa_set_jf_unknown): Likewise.
35903 (ipa_get_ipa_bits_for_value): New function.
35904 (ipa_set_jfunc_bits): Likewise.
35905 (ipa_get_value_range): New overloaded functions.
35906 (ipa_set_jfunc_vr): Likewise.
35907 (ipa_compute_jump_functions_for_edge): Use the above functions to
35908 construct bits and vr parts of jump functions.
35909 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
35910 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35912 (ipcp_grow_transformations_if_necessary): Also allocate
35913 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35915 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
35916 them. Fix too long lines.
35917 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
35918 vr_known being removed.
35919 (ipa_read_jump_function): Use new setter functions to construct bits
35920 and vr parts of jump functions or set them to NULL.
35921 (write_ipcp_transformation_info): Adjust for bits being pointers.
35922 (read_ipcp_transformation_info): Likewise.
35923 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
35925 Include gt-ipa-prop.h.
35926 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
35928 (ipcp_store_bits_results): Likewise.
35929 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
35930 Do not write to existing jump functions but use a temporary instead.
35932 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35935 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
35936 attempt to use its first operand as BIT_FIELD_REF base.
35938 2017-03-01 Richard Biener <rguenther@suse.de>
35940 PR middle-end/79721
35941 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
35942 interpolating formula in wrapping arithmetic.
35943 (chrec_apply): Convert chrec_evaluate return value to wanted type.
35945 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35947 PR tree-optimization/79734
35948 * tree-vect-generic.c (expand_vector_condition): Optimize
35949 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
35950 Handle VEC_COND_EXPR where comparison has different inner width from
35951 type's inner width.
35953 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
35955 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
35956 markup, and similar issues. Remove @opindex entries for things
35957 that aren't options. Add missing -mmpy-option entries.
35959 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35961 PR tree-optimization/79737
35962 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
35963 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
35964 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
35965 instead of byte_size. Formatting fix.
35966 (shift_bytes_in_array_right): Formatting fix.
35968 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
35971 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
35972 condition on optimize for the leaf function test.
35974 2017-02-28 Martin Liska <mliska@suse.cz>
35977 * read-rtl-function.c (function_reader::handle_unknown_directive):
35978 Bail out when one uses -flto.
35980 2017-02-28 Martin Liska <mliska@suse.cz>
35982 * common.opt: Replace space with tabular for options of <number>
35984 * config/i386/i386.opt: Show <number> value for
35985 -mlarge-data-threshold.
35986 * opts.c (print_filtered_help): Do not display number in hexadecimal
35989 2017-02-28 Martin Liska <mliska@suse.cz>
35991 * common.opt: Fix --help=option -Q for options which are of
35994 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
35996 * config/i386/i386.c (print_reg): Error out for values
35997 of 8-bit size in invalid integer register.
35999 2017-02-28 Martin Sebor <msebor@redhat.com>
36001 PR tree-optimization/79691
36002 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36004 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36007 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36008 gcc_unreachable with output_operand_lossage.
36010 2017-02-28 Richard Biener <rguenther@suse.de>
36012 PR tree-optimization/79740
36013 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36015 (visit_nary_op): Insert the nary into the hashtable if we
36016 pattern-matched sth.
36017 * tree-ssa-pre.c (eliminate_insert): Robustify.
36019 2017-02-28 Richard Biener <rguenther@suse.de>
36021 PR middle-end/79731
36022 * fold-const.c (decode_field_reference): Reject out-of-bound
36025 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36027 * config/i386/i386.c: Include intl.h.
36028 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36029 instead of just cond ? "..." : "...".
36030 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36031 * coverage.c (read_counts_file): Likewise.
36032 * omp-offload.c: Include intl.h.
36033 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36034 of just cond ? "..." : "...".
36035 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36036 of just cond ? "..." : "...".
36038 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36041 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36043 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36044 'tune for' CPU name.
36045 * config/arm/arm-cpu-data.h: Regenerated.
36047 2017-02-28 Richard Biener <rguenther@suse.de>
36049 PR tree-optimization/79732
36050 * tree-inline.c (expand_call_inline): Do not shadow var.
36052 2017-02-28 Richard Biener <rguenther@suse.de>
36054 PR tree-optimization/79723
36055 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36056 address-space properly.
36058 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36060 * doc/optinfo.texi (Optimization groups): Fix option used for
36062 * doc/invoke.texi (-fopt-info): Document "omp".
36063 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36064 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36065 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36066 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36067 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36069 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36071 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36072 OPTGROUP_OMP with "omp".
36074 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36077 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36078 for arithmetic shift of unsigned V2DI.
36080 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36082 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36083 arc/linux.h headers.
36084 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36085 (LINK_SPEC): Likewise.
36086 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36087 (EXTRA_SPECS): Likewise.
36088 (STARTFILE_SPEC): Likewise.
36089 (ENDFILE_SPEC): Likewise.
36090 (LIB_SPEC): Likewise.
36091 (TARGET_SDATA_DEFAULT): Likewise.
36092 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36093 (MULTILIB_DEFAULTS): Likewise.
36094 (DWARF2_UNWIND_INFO): Likewise.
36095 * config/arc/big.h: New file.
36096 * config/arc/elf.h: Likewise.
36097 * config/arc/linux.h: Likewise.
36098 * config/arc/t-uClibc: Remove.
36100 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36102 PR tree-optimization/77536
36103 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36104 (tree_transform_and_unroll_loop): Use above function to compute the
36105 estimated niter of unrolled loop and use it when scaling profile.
36106 Also use count info rather than frequency if it's non-zero.
36107 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36108 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36109 (vect_transform_loop): Call above function.
36111 2017-02-27 Richard Biener <rguenther@suse.de>
36113 PR tree-optimization/45397
36114 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36115 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36116 (visit_nary_op): Add pattern matching for CSEing sign-changed
36117 or truncated operations with wider ones.
36119 2017-02-27 Richard Biener <rguenther@suse.de>
36121 PR tree-optimization/79690
36122 * tree-vect-stmts.c (vectorizable_store): Use vector type
36123 built from the DR with address-space.
36125 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36127 * doc/invoke.texi (Optimize Options): Refine the description
36128 of asan-use-after-return.
36130 2017-02-25 Alan Modra <amodra@gmail.com>
36132 PR rtl-optimization/79584
36133 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36134 base, not ad->base_term, the reg within base. Remove assertion
36135 that ad->base == ad->base_term. Replace gen_int_mode using
36136 bogus mode with const0_rtx.
36138 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36140 PR middle-end/79396
36141 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36142 FMA_EXPR like tcc_binary or tcc_unary.
36144 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36147 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36149 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36150 (output_loc_operands): Handle DW_OP_call_ref and
36151 DW_OP_GNU_variable_value.
36152 (struct variable_value_struct): New type.
36153 (struct variable_value_hasher): Likewise.
36154 (variable_value_hash): New variable.
36155 (string_types): Remove.
36156 (copy_loc_descr): New function.
36157 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36158 (prepend_loc_descr_to_each): New function.
36159 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36160 instead of add_loc_descr_to_each if the first argument is single
36161 location list and the second has multiple.
36162 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36163 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36164 when looking for variable value which doesn't have other location info.
36165 (loc_list_from_tree): Formatting fix.
36166 (gen_array_type_die): Simplify DW_AT_string_length handling.
36167 (adjust_string_types): Remove.
36168 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36169 string_types. Call resolve_variable_values.
36170 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36171 (resolve_addr_in_expr): Likewise. Add A argument.
36172 (copy_deref_exprloc): Remove deref argument. Adjust for the
36173 original expression being DW_OP_GNU_variable_value with optionally
36174 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36175 optionally after it.
36176 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36177 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36178 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36179 (variable_value_hasher::hash, variable_value_hasher::equal): New
36181 (resolve_variable_value_in_expr, resolve_variable_value,
36182 resolve_variable_values, note_variable_value_in_expr,
36183 note_variable_value): New functions.
36184 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36187 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36190 * opts.h (handle_generated_option): Add GENERATED_P argument.
36191 * opts-common.c (handle_option): Adjust function comment.
36192 (handle_generated_option): Add GENERATED_P argument, pass it to
36194 (control_warning_option): Pass false to handle_generated_option
36196 * opts.c (maybe_default_option): Pass true to handle_generated_option
36198 * optc-gen.awk: Likewise.
36200 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36202 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36203 a REG, look at the REG it is a SUBREG of.
36204 (splitter for cmpeqsi_t): Ditto.
36206 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36208 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36209 the special USEs with the pattern of the insn, not the insn itself.
36211 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36214 * doc/invoke.texi: Document -mload-store-pairs.
36216 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36217 Sandra Loosemore <sandra@codesourcery.com>
36219 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36220 argument isn't a CONST_INT.
36221 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36222 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36223 (nios2_validate_compare): Bypass alternate compare logic if *op2
36224 is not a CONST_INT.
36225 (ldstwm_operation_p): Return false if first_base is not a REG or
36226 if first_offset is not a CONST_INT.
36228 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36230 * config/cris/cris.md: Use correct operand in a define_peephole2.
36232 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36234 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36236 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36238 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36239 this_insn if it is an INSN or JUMP_INSN.
36240 (force_offsettable): Look at base, not at addr.
36241 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36242 on things that aren't necessarily CONST_INTs.
36244 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36246 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36247 -mfpmath=sse is the default also for x86-32 targets with SSE2
36248 instruction set when @option{-ffast-math} is enabled
36250 2017-02-24 Jeff Law <law@redhat.com>
36252 PR rtl-optimizatoin/79286
36253 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36256 2017-02-24 Richard Biener <rguenther@suse.de>
36258 PR tree-optimization/79389
36259 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36262 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36264 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36265 function comment to reflect reality.
36266 (loop_exits_before_overflow): Fix typo in function description.
36268 2017-02-24 Richard Biener <rguenther@suse.de>
36270 PR tree-optimization/79389
36271 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36272 properly that a threading opportunity exists. Detect conditional
36273 copy/constant propagation opportunities.
36275 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36277 * config/visium/visium.md (type): Add trap.
36278 (b): New mode attribute.
36279 (*btst): Rename into...
36280 (*btst<mode>): ...this and adjust.
36281 (*cbranchsi4_btst_insn): Rename into...
36282 (*cbranch<mode>4_btst_insn): ...this and adjust.
36283 (trap): New define_insn.
36285 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36287 PR tree-optimization/79389
36288 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36289 (noce_reversed_cond_code): New function.
36290 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36291 reversed_comparison_code. Formatting fix.
36292 (noce_try_store_flag): Test rev_cond != NULL in addition to
36293 reversed_comparison_code.
36294 (noce_try_store_flag_constants): Likewise.
36295 (noce_try_store_flag_mask): Likewise.
36296 (noce_try_addcc): Use rev_cond if non-NULL instead of
36297 reversed_comparison_code.
36298 (noce_try_cmove_arith): Likewise. Formatting fixes.
36299 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36300 (noce_find_if_block): Initialize rev_cond.
36301 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36302 instead of false as last argument never attempt to reverse it
36305 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36307 PR tree-optimization/79663
36308 * tree-predcom.c (combine_chains): Process refs in reverse order
36309 only for ZERO length chains, and add explaining comment.
36311 2017-02-23 Jeff Law <law@redhat.com>
36313 PR tree-optimization/79578
36314 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36315 in call to operand_equal_p.
36317 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36320 * config/i386/cpuid.h: Fix another undefined behavior.
36322 2017-02-23 Richard Biener <rguenther@suse.de>
36324 PR tree-optimization/79683
36325 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36326 vector types for data-refs.
36328 2017-02-23 Martin Liska <mliska@suse.cz>
36330 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36332 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36334 PR middle-end/79665
36335 * internal-fn.c (get_range_pos_neg): Moved to ...
36336 * tree.c (get_range_pos_neg): ... here. No longer static.
36337 * tree.h (get_range_pos_neg): New prototype.
36338 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36339 are known to be in between 0 and signed maximum inclusive, try to
36340 expand both unsigned and signed divmod and use the cheaper one from
36343 2017-02-22 Jeff Law <law@redhat.com>
36345 PR tree-optimization/79578
36346 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36347 to compare base operands.
36349 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36352 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36353 gpc_reg_operand instead of fpr_reg_operand.
36355 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36357 * config/mips/mips.c (mips_return_in_memory): Force FP
36358 vector types to be returned in memory for o32 ABI.
36360 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36362 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36363 instead of DW_TAG_member for static data member declarations and don't
36364 set no_linkage_name for static inline data members.
36365 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36368 2017-02-22 Martin Liska <mliska@suse.cz>
36370 * doc/invoke.texi: Replace inequality signs with square brackets
36373 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36376 * lra-constraints.c (simplify_operand_subreg): Handle
36377 WORD_REGISTER_OPERATIONS targets.
36379 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36382 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36383 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36384 elimination by swapping fld*.
36386 2017-02-22 Richard Biener <rguenther@suse.de>
36388 PR tree-optimization/79673
36389 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36390 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36391 irrelevant address-space qualifiers and avoiding a
36392 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36394 2017-02-22 Richard Biener <rguenther@suse.de>
36396 PR tree-optimization/79666
36397 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36398 to not symbolically negate if that may introduce undefined
36401 2017-02-22 Martin Liska <mliska@suse.cz>
36404 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36405 * data-streamer-out.c (streamer_write_gcov_count_stream):
36407 * value-prof.c (stream_out_histogram_value): Make assert more
36408 precise based on type of counter.
36410 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36413 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36414 Use nonimmediate_operand instead of memory_operand for operand 1.
36415 (float-extend standard_x87sse_constant_load splitter): Ditto.
36417 2017-02-21 Jeff Law <law@redhat.com>
36419 PR tree-optimization/79621
36420 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36421 blocks with edges to themselves.
36423 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36426 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36427 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36428 Use gimple_call_builtin_p.
36431 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36432 on temporarily removed DEBUG_INSNs.
36434 PR tree-optimization/79649
36435 * tree-loop-distribution.c (classify_partition): Give up on
36436 non-generic address space loads/stores.
36438 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36440 * doc/loop.texi (Loop manipulation): Remove nonexistent
36441 tree_ssa_loop_version from the documentation.
36442 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36444 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36447 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36448 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36449 * config/rs6000/rs6000.c: Include except.h.
36450 (rs6000_expand_split_stack_prologue): Call
36451 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36453 2017-02-21 Martin Jambor <mjambor@suse.cz>
36456 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36457 have been analyzed.
36459 2017-02-21 Martin Jambor <mjambor@suse.cz>
36461 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36462 for backward compatibility only.
36463 * doc/invoke.texi (Option Summary): Remove all references to
36464 -fipa-cp-alignment.
36466 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36470 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36472 * lra-constraints.c (curr_insn_transform): Handle
36473 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36475 2017-02-21 Martin Liska <mliska@suse.cz>
36477 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36479 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36481 PR translation/79638
36482 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36484 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36487 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36488 (arm_function_ok_for_sibcall): Return false for an indirect call by
36489 descriptor if all the argument registers are used.
36490 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36491 alignment of the function.
36493 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36495 PR tree-optimization/61441
36496 * simplify-rtx.c (simplify_const_unary_operation): For
36497 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36498 the sNaN unmodified.
36500 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36502 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36503 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36504 instead of SYSTEM_HEADER_DIR.
36506 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36507 Martin Liška <mliska@suse.cz>
36509 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36510 Fix typos and grammar, use active voice, and clarify.
36512 2017-02-20 Marek Polacek <polacek@redhat.com>
36514 PR middle-end/79537
36515 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36518 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36520 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36523 * config/i386/i386.c (ix86_expand_builtin): Handle
36524 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36525 ix86_builtins_isa[fcode].isa as a requirement of those
36526 flags and any other flag in the bitmask.
36527 (ix86_init_mmx_sse_builtins): Use 0 instead of
36528 ~OPTION_MASK_ISA_64BIT as mask.
36529 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36530 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36531 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36532 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36534 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36537 * lra-constraints.c (split_reg): Check requested split mode
36538 is supported by the register.
36540 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36542 * lra-constraints.c (simplify_operand_subreg): Remove early
36545 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36548 * lra-constraints.c (curr_insn_transform): Tighten condition
36549 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36551 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36554 * lra-constraints.c (curr_insn_transform): Handle
36555 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36557 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36560 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36562 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36564 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36567 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36570 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36572 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36575 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36577 * final.c (last_columnnum, override_columnnum): New variables.
36578 (final_start_function): Set last_columnnum, pass it to begin_prologue
36579 hook and pass 0 to dwarf2out_begin_prologue.
36580 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36581 to source_line debug hook.
36582 (notice_source_line): Compute last_columnnum and for debug_column_info
36583 return true on column changes.
36584 * debug.h (struct gcc_debug_hooks): Add column argument to
36585 source_line and begin_prologue hooks.
36586 (debug_nothing_int_charstar_int_bool): Remove prototype.
36587 (debug_nothing_int_int_charstar,
36588 debug_nothing_int_int_charstar_int_bool): New prototypes.
36589 (dwarf2out_begin_prologue): Add column argument.
36590 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36591 begin_prologue hooks.
36592 (debug_nothing_int_charstar_int_bool): Remove.
36593 (debug_nothing_int_int_charstar,
36594 debug_nothing_int_int_charstar_int_bool): New functions.
36595 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36596 through to dwarf2out_source_line.
36597 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36598 (dwarf2out_source_line): Add column argument, emit it if requested.
36599 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36601 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36602 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36603 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36604 through to dwarf2out_begin_prologue.
36605 (vmsdbgout_source_line): Add column argument, pass it through to
36606 dwarf2out_source_line.
36607 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36608 dbxout_source_line caller.
36609 (dbxout_source_line): Add column argument.
36611 * common.opt (gno-column-info, gcolumn-info): New options.
36612 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36613 (check_die): Also test for multiple DW_AT_decl_column attributes.
36614 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36615 DW_AT_decl_column if requested.
36616 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36618 (gen_variable_die): Likewise.
36619 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36620 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36623 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36624 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36625 (ix86_handle_option): Handle OPT_m3dnowa.
36626 * doc/invoke.texi (-m3dnowa): Document.
36627 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36628 -m3dnowa instead of -m3dnow -march=athlon.
36631 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36632 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36634 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36637 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36638 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36639 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36640 generator for vsx_xxpermdi_<mode>_be.
36641 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36642 force big-endian semantics.
36643 (vsx_xxpermdi_<mode>_be): New define_expand with same
36644 implementation as previous version of vsx_xxpermdi_<mode>.
36646 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36648 PR tree-optimization/79327
36649 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36650 variable, its initialization and use.
36652 2017-02-17 Julia Koval <julia.koval@intel.com>
36654 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36655 (OPTION_MASK_ISA_PKU_UNSET): New.
36656 (ix86_handle_option): Handle -mrdpid.
36657 * config/i386/cpuid.h (bit_RDPID): New.
36658 * config/i386/driver-i386.c (host_detect_local_cpu):
36659 Detect RDPID feature.
36660 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36661 * config/i386/i386-c.c (ix86_target_macros_internal):
36663 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36664 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36665 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36666 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36667 * config/i386/i386.md (define_insn "rdpid"): New.
36668 * config/i386/i386.opt Add -mrdpid.
36669 * config/i386/immintrin.h (_rdpid_u32): New.
36671 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36673 PR rtl-optimization/79541
36674 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36675 instead of transforming it into USE.
36677 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36679 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36680 If HONOR_SNANS (SFmode) force the input to a register.
36681 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36682 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36683 an frsp or similar insn.
36685 2017-02-17 Martin Liska <mliska@suse.cz>
36687 PR rtl-optimization/79577
36688 * params.def (selsched-max-sched-times): Increase minimum to 1.
36690 2017-02-17 Martin Liska <mliska@suse.cz>
36692 PR rtl-optimization/79574
36693 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36695 2017-02-17 Martin Liska <mliska@suse.cz>
36697 PR tree-optimization/79529
36698 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36699 ssa_defined_default_def_p to handle cases which are implicitly
36701 * tree-ssa.c (ssa_defined_default_def_p): New function.
36702 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36703 which are implicitly defined.
36704 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36706 2017-02-17 Richard Biener <rguenther@suse.de>
36708 PR middle-end/79576
36709 * params.def (max-ssa-name-query-depth): Limit to 10.
36711 2017-02-17 Richard Biener <rguenther@suse.de>
36713 PR tree-optimization/79552
36714 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36717 2017-02-17 Richard Biener <rguenther@suse.de>
36720 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36722 2017-02-17 Marek Polacek <polacek@redhat.com>
36724 PR middle-end/79536
36725 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36726 (fold_negate_expr): New wrapper.
36728 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36730 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36731 Correct terminology and de-emphasize pre-standard behavior.
36733 2017-02-16 Alan Modra <amodra@gmail.com>
36735 PR rtl-optimization/79286
36736 * ira.c (def_dominates_uses): New function.
36737 (update_equiv_regs): Don't create an equivalence for insns that
36738 may trap where the register def does not dominate the use.
36740 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36742 PR rtl-optimization/78127
36743 * lra.c (lra): Call lra_eliminate before finish the loop after
36746 2017-02-16 Richard Biener <rguenther@suse.de>
36748 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36750 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36751 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36752 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36753 (isl_val_int_from_wi): New function.
36754 (extract_affine_gmp): Rename to ...
36755 (extract_affine_wi): ... this, take a widest_int.
36756 (extract_affine_int): Just wrap extract_affine_wi.
36757 (add_param_constraints): Use isl_val_int_from_wi.
36758 (add_loop_constraints): Likewise, and extract_affine_wi.
36760 2017-02-15 Jeff Law <law@redhat.com>
36762 PR middle-end/79521
36763 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36764 ira_init_register_move_cost_if_necessary.
36766 2017-02-15 Martin Sebor <msebor@redhat.com>
36768 PR middle-end/32003
36769 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36770 removed in a prior commit.
36772 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36774 PR tree-optimization/79347
36775 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36776 counters during peeling.
36778 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36780 * Makefile.in (site.exp): Remove "set ISLVER".
36782 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36785 * real.c (real_from_integer): Call real_convert even for decimal.
36787 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36790 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36792 2017-02-14 Andrew Pinski <apinski@cavium.com>
36794 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36795 cores and change the partno/implementer to be correct.
36796 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36797 the 'B" as the implementer.
36798 * config/aarch64/aarch64-tune.md: Regenerate.
36800 2017-02-14 Carl Love <cel@us.ibm.com>
36802 * config/rs6000/rs6000.c: Add case statement entry to make the
36803 xvcvuxdsp built-in argument unsigned.
36804 * config/rs6000/vsx.md: Fix the source and return operand types so they
36805 match the instruction definitions from the ISA document. Fix typo
36806 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36809 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36812 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36813 member early_clobber_alts.
36814 * lra-lives.c (reg_early_clobber_p): New.
36815 (process_bb_lives): Use it.
36816 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36817 (debug_operand_data): Initialize early_clobber_alts.
36818 (setup_operand_alternative): Set up early_clobber_alts.
36819 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36820 alternatives to new_insn_reg.
36821 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36823 (lra_update_insn_regno_info): Pass the new arg.
36825 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36827 PR middle-end/79505
36828 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36829 (new_oacc_loop_raw): Don't clear already cleared fields.
36832 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36833 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36834 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36836 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36839 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36841 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36844 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36845 the extra instruction to the right place to store 128-bit constant
36848 2017-02-14 Martin Sebor <msebor@redhat.com>
36850 PR middle-end/79448
36851 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36852 warning for strings of unknown length.
36854 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36856 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36858 2017-02-14 Jeff Law <law@redhat.com>
36861 * ira-costs.c (scan_one_insn): Initialize register move costs
36862 for pseudos seen in USE/CLOBBER insns.
36864 PR tree-optimization/79095
36865 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
36866 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
36867 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
36868 if the operands are known to be not equal, then the resulting range
36870 (intersect_ranges): If the new range is ~[0,0] and the old range is
36871 wide, then prefer ~[0,0].
36872 * tree-vrp.c (overflow_comparison_p_1): New function.
36873 (overflow_comparison_p): New function.
36874 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
36875 if NAME is used in an overflow test.
36876 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
36877 overflow check that can be expressed as an equality test, then adjust
36878 ops to be that equality test.
36880 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36882 * config/s390/s390-builtin-types.def: Remove flags argument.
36883 * config/s390/s390.c (s390_init_builtins): Likewise.
36885 2017-02-14 Martin Liska <mliska@suse.cz>
36887 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
36888 vector. Fix trailing white spaces.
36890 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
36892 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
36895 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36897 PR rtl-optimization/68664
36898 * config/arm/arm.c (arm_sched_can_speculate_insn):
36899 New function. Declare prototype.
36900 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36902 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36904 PR rtl-optimization/68664
36905 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
36907 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36909 2017-02-14 Amit Pawar <amit.pawar@amd.com>
36911 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
36912 max skip bytes for function, loop and jump.
36914 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36916 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
36917 ABS_EXPR for gimple dump.
36919 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36922 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
36924 PR tree-optimization/79408
36925 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
36926 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
36927 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
36928 also if rhs1 is INTEGER_CST.
36930 2017-02-14 Richard Biener <rguenther@suse.de>
36932 PR middle-end/79432
36933 * tree-into-ssa.c (insert_phi_nodes): When the function can
36934 have abnormal edges rewrite SSA names with broken use-def
36935 dominance out of SSA and register them for PHI insertion.
36937 2017-02-13 Martin Sebor <msebor@redhat.com>
36939 PR middle-end/79496
36940 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
36941 clearing info.nowrite flag when snprintf size argument is a range.
36943 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36945 * cprop.c (cprop_jump): Add missing space in string literal.
36946 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
36947 (get_constraint_for_component_ref): Likewise.
36948 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
36949 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
36950 * lra-constraints.c (process_alt_operands): Likewise.
36951 * ipa-inline.c (inline_small_functions): Likewise.
36952 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
36953 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
36954 * trans-mem.c (diagnose_tm_1_op): Likewise.
36955 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
36956 (grid_parallel_clauses_gridifiable): Likewise.
36958 * config/nvptx/mkoffload.c (process): Add space in between
36961 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
36962 "MOD4_SSE_REGS" and "ALL_REGS".
36964 * spellcheck.c (test_data): Add , in between "foo" and "food".
36966 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36969 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
36970 boundary crossing check and subsequent code generation agree.
36972 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36974 * config/aarch64/aarch64.c (has_memory_op): Delete.
36975 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
36978 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36980 PR rtl-optimization/79388
36981 PR rtl-optimization/79450
36982 * combine.c (distribute_notes): When removing TEM_INSN for which
36983 corresponding dest has last value recorded, invalidate that last
36986 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36988 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
36989 of explicit '@'. Add missing assembly comment marker on branch costs
36992 2017-02-13 Nathan Sidwell <nathan@acm.org>
36994 * gengtype-lex.l (<in_struct>): Add '/'.
36996 2017-02-13 Martin Liska <mliska@suse.cz>
36999 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37001 2017-02-13 Richard Biener <rguenther@suse.de>
37003 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37005 * configure: Re-generate.
37006 * config.in: Likewise.
37007 * graphite-dependences.c: Simplify as if
37008 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37009 * graphite-isl-ast-to-gimple.c: Likewise.
37010 * graphite-optimize-isl.c: Likewise.
37011 * graphite-poly.c: Likewise.
37012 * graphite-sese-to-poly.c: Likewise.
37013 * graphite.h: Likewise.
37014 * toplev.c: Include isl/version.h and use isl_version () for
37015 printing the ISL version.
37016 * doc/install.texi: Update ISL requirement.
37018 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37020 * doc/standards.texi (Standards): Update reference to
37023 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37025 * doc/extend.texi (Named Address Spaces): sourceware.org now
37027 * doc/install.texi (Binaries): Ditto.
37030 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37032 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37033 standard terminology "stringize"/"stringizing" throughout.
37034 * doc/cppinternals.texi: Likewise.
37036 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37038 * doc/extend.texi: Fix some spelling mistakes and typos.
37039 * doc/invoke.texi: Likewise.
37041 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37044 * params.def (inline-min-speedup) Change from 10 to 8.
37046 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37048 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37051 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37054 * ipa-inline-analysis.c (get_minimal_bb): New function.
37055 (record_modified): Use it.
37056 (remap_edge_change_prob): Handle also ancestor functions.
37058 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37060 * doc/contrib.texi (Contributors): Remove broken link into
37061 the Mauve CVS repository.
37063 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37065 PR middle-end/79454
37066 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37067 result computation whenever lhs doesn't have vector mode, not
37068 just when it has BLKmode.
37070 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37072 * doc/makefile.texi (profiledbootstrap): Refer to the
37073 installation instructions only in textual form.
37075 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37078 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37080 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37082 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37083 (Specific): Update mingw-w64 reference.
37085 (Specific): Remove broken link to Renesas RX processor.
37087 2017-02-10 Richard Biener <rguenther@suse.de>
37089 * toplev.c (process_options): Do not mention obsolete graphite
37090 options when printing sorry message about missing graphite support.
37091 Mention -floop-nest-optimize.
37093 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37095 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37096 (vtst_p16): Likewise.
37097 (vtstq_p8): Likewise.
37098 (vtstq_p16): Likewise.
37100 (vtstq_p64): Likewise.
37101 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37102 (vset_lane_p64): New.
37103 (vsetq_lane_p64): New.
37105 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37107 PR tree-optimization/79411
37108 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37109 stmt operands are SSA_NAMEs used in abnormal phis.
37110 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37113 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37116 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37119 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37121 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37123 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37125 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37129 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37130 not arbitrary TREE_CONSTANT.
37133 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37134 "omp declare target link" attribute unless is_global_var.
37135 * omp-offload.c (find_link_var_op): Likewise.
37137 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37138 Chung-Lin Tang <cltang@codesourcery.com>
37140 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37142 (gimplify_adjust_omp_clauses): Don't delete TILE.
37143 (gimplify_omp_for): Deal with TILE.
37144 * internal-fn.c (expand_GOACC_TILE): New function.
37145 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37147 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37148 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37150 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37151 avoid DIV for outermost collapse var.
37152 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37153 Remove out of date comments, fix whitespace.
37154 * omp-general.c (omp_extract_for_data): Deal with tiling.
37155 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37156 adjust OLF_DIM_BASE value.
37157 (struct omp_for_data): Add tiling field.
37158 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37159 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37160 for auto loops. Remove default auto determining, moved to
37161 oacc_loop_fixed_partitions.
37162 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37163 stmts, add e_mask field.
37164 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37165 (oacc_thread_numbers): Use oacc_dim_call.
37166 (oacc_xform_tile): New.
37167 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37168 (finish_oacc_loop): Adjust for ifns vector.
37169 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37170 add case for GOACC_TILE fns.
37171 (oacc_loop_xform_loop): Delete.
37172 (oacc_loop_process): Iterate over call list directly, and add
37173 handling for GOACC_TILE fns.
37174 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37176 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37177 vector partitioning to outer loops. Assign 2 partitions to loops
37178 when available. Add TILE handling.
37179 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37180 (execite_oacc_device_lower): Process GOACC_TILE fns,
37181 ignore unknown specs.
37182 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37183 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37184 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37186 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37188 * configure.ac (ACX_BUGURL): Update.
37189 * configure: Regenerate.
37191 2017-02-09 Richard Biener <rguenther@suse.de>
37193 PR tree-optimization/69823
37194 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37195 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37197 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37199 * config/arc/arc-c.def: Add __NPS400__ definition.
37200 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37201 (TARGET_NPS400): Define.
37203 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37205 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37207 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37208 pointer, arch_info.
37209 (arc_cpu_types): Fill the arch_info field with a pointer into the
37210 arc_arch_types table.
37211 (arc_selected_cpu): Declare.
37212 * config/arc/arc.c (arc_selected_cpu): Make global.
37213 (arc_selected_arch): Delete.
37214 (arc_base_cpu): Delete.
37215 (arc_override_options): Remove references to deleted variables,
37216 update access to arch information.
37217 (ARC_OPT): Update access to arch information.
37218 (ARC_OPTX): Likewise.
37219 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37220 (TARGET_ARC600): Update access to arch information.
37221 (TARGET_ARC601): Likewise.
37222 (TARGET_ARC700): Likewise.
37223 (TARGET_EM): Likewise.
37224 (TARGET_HS): Likewise.
37225 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37228 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37231 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37232 condition/operands for integer GE/LE/GEU/LEU operations.
37234 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37236 PR translation/79397
37237 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37240 2017-02-08 Martin Jambor <mjambor@suse.cz>
37243 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37244 whether allocation happened.
37245 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37246 nothing was allocated.
37248 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37250 PR tree-optimization/79408
37251 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37252 constant, but SSA_NAME with a known integer range, use the minimum
37253 of that range instead of op1 to determine if modulo can be replaced
37254 with its first operand.
37256 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37258 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37260 2017-02-08 Richard Biener <rguenther@suse.de>
37262 PR tree-optimization/71824
37263 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37264 Check all loops contained in the merged region.
37266 2017-02-07 Andrew Pinski <apinski@cavium.com>
37268 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37270 2017-02-07 Andrew Pinski <apinski@cavium.com>
37272 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37273 (thunderxt88): Likewise.
37274 (thunderxt81): Disable LSE and change v8.1 to v8.
37275 (thunderxt83): Likewise.
37277 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37278 Richard Biener <rguenther@suse.de>
37280 PR middle-end/79399
37281 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37282 type from int to size_t.
37283 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37285 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37287 PR rtl-optimization/79386
37288 * cprop.c (bypass_conditional_jumps): Initialize
37289 bypass_last_basic_block already before splitting bbs after
37290 unconditional traps...
37291 (bypass_conditional_jumps): ... rather than here.
37294 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37295 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37296 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37297 fix -masm=intel patterns.
37299 2017-02-07 Richard Biener <rguenther@suse.de>
37301 PR tree-optimization/79256
37302 PR middle-end/79278
37303 * builtins.c (get_object_alignment_2): Use min_align_of_type
37304 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37305 and ADJUST_FIELD_ALIGN.
37307 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37309 * doc/tm.texi: Regenerate.
37310 * stor-layout.c (layout_decl): Adjust.
37311 (update_alignment_for_field): Likewise.
37312 (place_field): Likewise.
37313 (min_align_of_type): Likewise.
37314 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37315 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37316 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37317 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37318 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37319 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37320 * config/i386/i386.c (x86_field_alignment): Likewise.
37321 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37322 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37323 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37324 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37325 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37326 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37330 2017-01-30 Richard Biener <rguenther@suse.de>
37332 PR tree-optimization/79256
37333 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37334 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37337 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37339 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37340 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37341 builtins to SImode and emit a zero-extend, if necessary.
37343 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37345 * docs/invoke.texi (RISC-V Options): Alphabetize.
37347 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37349 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37352 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37354 * config/riscv/riscv.c: New file.
37355 * common/config/riscv/riscv-common.c: Likewise.
37356 * config.gcc: Likewise.
37357 * config/riscv/constraints.md: Likewise.
37358 * config/riscv/elf.h: Likewise.
37359 * config/riscv/generic.md: Likewise.
37360 * config/riscv/linux.h: Likewise.
37361 * config/riscv/multilib-generator: Likewise.
37362 * config/riscv/peephole.md: Likewise.
37363 * config/riscv/pic.md: Likewise.
37364 * config/riscv/predicates.md: Likewise.
37365 * config/riscv/riscv-builtins.c: Likewise.
37366 * config/riscv/riscv-c.c: Likewise.
37367 * config/riscv/riscv-ftypes.def: Likewise.
37368 * config/riscv/riscv-modes.def: Likewise.
37369 * config/riscv/riscv-opts.h: Likewise.
37370 * config/riscv/riscv-protos.h: Likewise.
37371 * config/riscv/riscv.h: Likewise.
37372 * config/riscv/riscv.md: Likewise.
37373 * config/riscv/riscv.opt: Likewise.
37374 * config/riscv/sync.md: Likewise.
37375 * config/riscv/t-elf-multilib: Likewise.
37376 * config/riscv/t-linux: Likewise.
37377 * config/riscv/t-linux-multilib: Likewise.
37378 * config/riscv/t-riscv: Likewise.
37379 * configure.ac: Likewise.
37380 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37381 Waterman as RISC-V maintainers.
37382 * doc/install.texi: Add RISC-V entries.
37383 * doc/invoke.texi: Add RISC-V options section.
37384 * doc/md.texi: Add RISC-V constraints section.
37385 * configure: Regenerated.
37387 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37390 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37391 false values to be constant vectors with all 0 or all 1 bits set.
37392 (vcondu<mode><mode>): Likewise.
37393 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37395 (fpmask_comparison_operator): Update comment.
37396 (vecint_comparison_operator): New predicate.
37397 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37398 vector conditionals when the true and false values are constant
37399 vectors with all 0 bits or all 1 bits set.
37401 2017-02-06 Martin Sebor <msebor@redhat.com>
37403 PR tree-optimization/79376
37404 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37406 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37408 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37409 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37410 to simplify split condition.
37412 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37414 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37415 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37418 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37420 PR rtl-optimization/68664
37421 * target.def (can_speculate_insn): New hook.
37422 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37423 * doc/tm.texi: Regenerate.
37424 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37425 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37426 (rs6000_sched_can_speculate_insn): New function.
37428 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37430 PR tree-optimization/79284
37431 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37432 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37433 vectorizable_mask_load_store, vectorizable_operation,
37434 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37435 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37436 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37437 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37438 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37439 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37440 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37441 is_gimple_assign (stmt). Replace another such test with
37442 is_gimple_assign (stmt).
37444 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37447 * config/avr/avr.c (rtl-iter.h): Include it.
37448 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37449 (avr_legitimate_combined_insn): ...and implementation.
37451 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37453 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37454 * config/s390/s390.c (s390_const_operand_ok)
37455 (s390_canonicalize_comparison, s390_extract_part)
37456 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37457 (s390_contiguous_bitmask_p, s390_rtx_costs)
37458 (legitimize_pic_address): Likewise.
37459 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37460 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37461 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37462 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37463 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37465 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37467 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37468 REGNO($0) == REGNO($1).
37470 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37472 * config/s390/linux.h(SIZE_TYPE): Add comment.
37474 2017-02-06 Julian Brown <julian@codesourcery.com>
37475 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37476 Virendra Pathak <virendra.pathak@broadcom.com>
37478 * config/aarch64/aarch64-cores.def: Change the scheduler
37480 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37481 * config/aarch64/thunderx2t99.md: New file.
37483 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37485 * doc/standards.texi (Go Language): Update link to language
37488 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37490 * tree-eh.c (lower_resx): Sanitize profile.
37491 (cleanup_empty_eh_move_lp): Likewise.
37493 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37496 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37498 * cfgloopmanip.h (loop_version): Update prototype.
37499 * modulo-sched.c (sms_schedule): Update call of loop_version.
37500 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37501 * tree-parloops.c (gen_parallel_loop): Likewise.
37502 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37503 * tree-ssa-loop-split.c (split_loop): Likewise.
37504 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37505 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37507 2017-02-05 Martin Liska <mliska@suse.cz>
37510 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37512 (print_operand_address): Initialize a struct to zero.
37514 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37516 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37517 garbage collector only in textual form.
37519 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37521 * doc/extend.texi (x86 specific memory model extensions for
37522 transactional memory): Simplify a phrase.
37524 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37527 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37528 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37529 (atomic_storedi_1): Likewise.
37531 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37533 PR tree-optimization/79338
37534 * tree-parloops.c (gather_scalar_reductions): Don't call
37535 vect_analyze_loop_form for loop->inner before destroying loop's
37538 2017-02-03 Martin Sebor <msebor@redhat.com>
37540 PR tree-optimization/79327
37541 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37542 when precision has resulted in leading zeros.
37543 (format_integer): Adjust the likely counter to assume an unknown
37544 argument that may be zero is non-zero.
37546 2017-02-03 Jason Merrill <jason@redhat.com>
37549 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37550 avoid copying non-taken branch.
37552 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37554 PR tree-optimization/79340
37555 * tree-vect-loop.c (vectorizable_reduction): Release
37556 vec_defs elements after safe_splicing them into other vectors.
37559 PR tree-optimization/79327
37560 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37561 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37563 (format_integer): Use wide_int_to_tree instead of build_int_cst
37564 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37565 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37566 of shortest and longest sequence.
37568 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37570 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37571 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37573 2017-02-03 Walter Lee <walt@tilera.com>
37576 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37577 after initial stackframe link reg save.
37578 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37580 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37583 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37584 wu for stxssp alternative.
37586 2017-02-03 Martin Sebor <msebor@redhat.com>
37588 PR tree-optimization/79352
37589 * gimple-fold.c (get_range_strlen): Add argument.
37590 (get_range_strlen): Change return type to bool.
37591 (get_maxval_strlen): Pass in a dummy argument.
37592 * gimple-fold.h (get_range_strlen): Change return type to bool.
37593 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37594 * tree.h (array_at_struct_end_p): Add argument.
37595 * tree.c (array_at_struct_end_p): Handle it.
37597 2017-02-03 Martin Liska <mliska@suse.cz>
37600 * multiple_target.c (create_dispatcher_calls): Redirect edge
37601 from a caller of a dispatcher.
37602 (expand_target_clones): Make the clones local.
37603 (ipa_target_clone): Do both target clones and resolvers.
37604 (ipa_dispatcher_calls): Remove the pass.
37605 (pass_dispatcher_calls::gate): Likewise.
37606 (make_pass_dispatcher_calls): Likewise.
37607 * passes.def (pass_target_clone): Put as very first IPA early
37610 2017-02-03 Martin Liska <mliska@suse.cz>
37612 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37613 in case of a function with ifunc attribute.
37615 2017-02-03 Martin Liska <mliska@suse.cz>
37617 * cgraph.c (cgraph_node::dump): Dump function version info.
37618 * symtab.c (symtab_node::dump_base): Add missing new line.
37620 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37622 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37623 (ifcombine_ifandif): Use it.
37625 2017-02-03 Martin Liska <mliska@suse.cz>
37627 * doc/invoke.texi: Document default value for
37628 use-after-scope-direct-emission-threshold.
37630 2017-02-03 Martin Liska <mliska@suse.cz>
37632 PR tree-optimization/79339
37633 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37634 (format_floating): Likewise.
37636 2017-02-03 Martin Liska <mliska@suse.cz>
37639 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37641 (ipa_node_params_t::remove): Likewise.
37642 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37643 initialization from removed ipa_node_params_t::insert.
37644 (ipa_node_params::~ipa_node_params): Move from removed
37645 ipa_node_params_t::release.
37646 * symbol-summary.h (symbol_summary::m_released): New member.
37647 Do not release a summary twice. Do not allow to call finalizer
37648 for types of a summary that live in GGC memory.
37650 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37652 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37655 2017-02-02 Martin Sebor <msebor@redhat.com>
37657 PR middle-end/79275
37658 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37659 (format_string): Tighten up the range of output for non-constant
37660 strings and correct the expected range for wide non-constant strings.
37662 2017-02-02 Martin Sebor <msebor@redhat.com>
37664 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37666 PR middle-end/32003
37667 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37669 (-fdump-tree-@var): Add to index and document how to come up
37670 with pass-specific option and dump file names.
37671 (-fdump-passes): Clarify where to look for output.
37673 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37675 PR middle-end/77445
37676 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37677 statistics of the analyzed path; allow threading for speed when
37678 any of BBs along the path are optimized for speed.
37680 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37682 PR middle-end/78468
37683 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37684 settings of the virtual registers.
37687 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37689 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37690 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37693 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37695 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37696 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37698 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37700 * config/s390/s390.md: Add missing comments with the expanded
37702 * config/s390/vector.md: Likewise.
37703 * config/s390/vx-builtins.md: Likewise.
37705 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37708 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37709 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37710 conditions on a single line.
37712 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37714 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37715 __S390_VX__ to __VX__.
37717 2017-02-01 Andrew Pinski <apinski@cavium.com>
37719 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37720 stmt_info to record_stmt_cost.
37721 (vect_get_known_peeling_cost): Pass stmt_info if known to
37723 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37724 cpu_vector_cost field into
37725 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37726 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37727 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37728 splitting of scalar_stmt_cost and vec_stmt_cost.
37729 (thunderx_vector_cost): Likewise.
37730 (cortexa57_vector_cost): LIkewise.
37731 (exynosm1_vector_cost): Likewise.
37732 (xgene1_vector_cost): Likewise.
37733 (thunderx2t99_vector_cost): Improve after the splitting of the two
37735 (aarch64_builtin_vectorization_cost): Update for the splitting of
37736 scalar_stmt_cost and vec_stmt_cost.
37738 2017-02-01 Torvald Riegel <triegel@redhat.com>
37739 Richard Henderson <rth@redhat.com>
37741 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37742 conditional on existance of a fast atomic load.
37743 * optabs-query.c (can_atomic_load_p): New function.
37744 * optabs-query.h (can_atomic_load_p): Declare it.
37745 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37746 no fast atomic load is available for the particular size of access.
37747 (expand_atomic_compare_and_swap): Likewise.
37748 (expand_atomic_load): Likewise.
37749 (expand_atomic_store): Likewise.
37750 (expand_atomic_fetch_op): Likewise.
37751 * testsuite/lib/target-supports.exp
37752 (check_effective_target_sync_int_128): Remove x86 because it provides
37753 no fast atomic load.
37754 (check_effective_target_sync_int_128_runtime): Likewise.
37756 2017-02-01 Richard Biener <rguenther@suse.de>
37758 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37759 (graphite_transform_loops): Provide opt-info for optimized nests.
37760 * tree-parloop.c (parallelize_loops): Provide opt-info for
37761 parallelized loops.
37763 2017-02-01 Richard Biener <rguenther@suse.de>
37765 PR middle-end/79315
37766 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37767 was not set before.
37769 2017-02-01 Richard Biener <rguenther@suse.de>
37771 PR tree-optimization/71824
37772 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37773 Verify the loops are valid in the merged SESE region.
37774 (scop_detection::can_represent_loop_1): Check analyzing the
37775 evolution of the number of iterations in the region succeeds.
37777 2017-01-31 Ian Lance Taylor <iant@golang.org>
37779 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37780 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37782 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37784 PR preprocessor/79210
37785 * input.c (get_substring_ranges_for_loc): Replace line_width
37786 assertion with error-handling.
37788 2017-01-31 Richard Biener <rguenther@suse.de>
37790 PR tree-optimization/77318
37791 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37792 (create_pw_aff_from_tree): Take loop parameter.
37793 (add_condition_to_pbb): Pass loop of the condition to
37794 create_pw_aff_from_tree.
37796 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37798 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37799 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37801 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37805 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37807 (convert_int_to_float128): Likewise.
37808 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37809 (convert_int_to_float128): Likewise.
37810 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37811 (UNSPEC_IEEE128_CONVERT): Likewise.
37812 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37813 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37814 Use local variables for IBM extended format.
37815 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37816 (fix_trunc<mode>si2_fprs): Likewise.
37817 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37818 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37819 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37820 to know that we can now have integers of all sizes in vector
37822 (fix<uns>_<mode>di2_hw): Likewise.
37823 (float<uns>_<mode>si2_hw): Likewise.
37824 (fix_<mode>si2_hw): Likewise.
37825 (fixuns_<mode>si2_hw): Likewise.
37826 (float<uns>_<mode>di2_hw): Likewise.
37827 (float_<mode>di2_hw): Likewise.
37828 (float_<mode>si2_hw): Likewise.
37829 (floatuns_<mode>di2_hw): Likewise.
37830 (floatuns_<mode>si2_hw): Likewise.
37831 (xscvqp<su>wz_<mode>): Delete, no longer used.
37832 (xscvqp<su>dz_<mode>): Likewise.
37833 (xscv<su>dqp_<mode>): Likewise.
37834 (ieee128_mfvsrd_64bit): Likewise.
37835 (ieee128_mfvsrd_32bit): Likewise.
37836 (ieee128_mfvsrwz): Likewise.
37837 (ieee128_mtvsrw): Likewise.
37838 (ieee128_mtvsrd_64bit): Likewise.
37839 (ieee128_mtvsrd_32bit): Likewise.
37841 2017-01-31 Martin Liska <mliska@suse.cz>
37844 * ipa-prop.c (ipa_free_all_node_params): Call release method
37845 instead of ~sumbol_summary to not to trigger double times
37848 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37850 PR tree-optimization/71691
37851 * bitmap.h (class auto_bitmap): New.
37852 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37853 is_maybe_undefined instead of ssa_undefined_value_p.
37855 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37857 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37858 __S390_ARCH_LEVEL__ to __ARCH__.
37860 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37862 PR tree-optimization/79267
37863 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
37864 if should_remove_lhs_p is true.
37866 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
37869 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
37870 (add_alignment_attribute): New.
37871 (base_type_die): Add alignment attribute.
37872 (subrange_type_die): Likewise.
37873 (modified_type_die): Likewise.
37874 (gen_array_type_die): Likewise.
37875 (gen_descr_array_type_die: Likewise.
37876 (gen_enumeration_type_die): Likewise.
37877 (gen_subprogram_die): Likewise.
37878 (gen_variable_die): Likewise.
37879 (gen_field_die): Likewise.
37880 (gen_ptr_to_mbr_type_die): Likewise.
37881 (gen_struct_or_union_type_die): Likewise.
37882 (gen_subroutine_type_die): Likewise.
37883 (gen_typedef_die): Likewise.
37884 (base_type_cmp): Compare alignment attribute.
37886 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37889 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
37890 (setb_unsigned) New pattern for setb with CCUNS.
37891 * config/rs6000/rs6000.c (expand_block_compare): Use a different
37892 subfc./subfe sequence to avoid overflow problems. Generate a
37893 shorter sequence with cmpld/setb for power9.
37894 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
37895 for generating subfc. instruction.
37896 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
37897 now uses this instruction.
37899 2017-01-30 Ian Lance Taylor <iant@google.com>
37902 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
37903 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
37905 2017-01-30 Martin Sebor <msebor@redhat.com>
37907 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
37908 Move constant to the right of a relational operator.
37909 (get_mpfr_format_length, format_character, format_string): Ditto.
37910 (should_warn_p, maybe_warn): Same.
37912 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
37914 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
37917 * asan.c (get_translation_unit_decl): Remove function.
37918 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
37920 2017-01-30 Martin Liska <mliska@suse.cz>
37922 PR gcov-profile/79259
37923 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
37924 -fprofile-generate.
37926 2017-01-30 Martin Liska <mliska@suse.cz>
37929 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
37930 Initialize variables with NULL value.
37932 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
37935 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
37937 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
37939 2017-01-30 Richard Biener <rguenther@suse.de>
37941 PR tree-optimization/79276
37942 * tree-vrp.c (process_assert_insertions): Properly adjust common
37943 when removing a duplicate.
37945 2017-01-30 Richard Biener <rguenther@suse.de>
37947 PR tree-optimization/79256
37948 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37949 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37951 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
37953 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
37956 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
37957 ("*r<noxa>sbg_<mode>_sll_bitmask")
37958 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
37959 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
37960 Use contiguous_bitmask_nowrap_operand.
37962 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37965 * config/rs6000/altivec.h (vec_xl): Revise #define.
37966 (vec_xst): Likewise.
37968 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
37970 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
37972 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
37974 PR rtl-optimization/79194
37975 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
37976 traps before call to bypass_conditional_jumps.
37978 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37980 PR tree-optimization/71374
37981 * lra-constraints.c (check_conflict_input_operands): New.
37982 (match_reload): Use it.
37984 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37987 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
37988 account to calculate conflict_set.
37990 2017-01-27 Bin Cheng <bin.cheng@arm.com>
37992 PR rtl-optimization/78559
37993 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
37994 other_insn in combine.
37996 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
37998 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
37999 uint16_type_node for BT_UINT16.
38001 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38003 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38004 "RTL Tests" to menu.
38005 (GIMPLE Tests): New node.
38006 (RTL Tests): New node.
38008 2017-01-27 Richard Biener <rguenther@suse.de>
38010 PR tree-optimization/79245
38011 * tree-loop-distribution.c (distribute_loop): Apply cost
38012 modeling also to detected patterns.
38014 2017-01-27 Richard Biener <rguenther@suse.de>
38016 PR tree-optimization/71433
38017 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38018 (compare_assert_loc): New function.
38019 (process_assert_insertions): Sort and optimize assert locations
38020 to remove duplicates and push down identical assertions on
38021 edges to their destination block.
38023 2017-01-27 Richard Biener <rguenther@suse.de>
38025 PR tree-optimization/79244
38026 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38027 out SSA names even if abnormal.
38029 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38031 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38032 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38033 instead of MPFR_RNDN.
38035 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38038 * arm.c (arm_option_override): Don't call build_target_option_node
38039 until after doing all option overrides.
38040 (arm_valid_target_attribute_tree): Likewise.
38042 2017-01-27 Martin Liska <mliska@suse.cz>
38044 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38045 for {cd}tors and C++ {cd}tors.
38047 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38049 * config/s390/s390.md ("*setmem_long_and")
38050 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38052 2017-01-26 Martin Sebor <msebor@redhat.com>
38054 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38057 2017-01-26 Martin Sebor <msebor@redhat.com>
38059 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38060 HAVE_DFmode before using XFmode or DFmode.
38061 (parse_directive): Avoid using the z length modifier to avoid
38062 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38064 PR middle-end/78703
38065 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38066 to accept adjustment as an array.
38067 (get_int_range): New function.
38068 (struct directive): Make width and prec arrays.
38069 (directive::set_width, directive::set_precision): Call get_int_range.
38070 (format_integer, format_floating): Handle width and precision ranges.
38071 (format_string, parse_directive): Same.
38073 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38076 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38077 comdat_type_p set, just clone them, but keep the children in the
38081 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38082 which have direct callers with -fvar-tracking-assignments enabled
38084 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38085 inside of type units.
38087 2017-01-26 Martin Sebor <msebor@redhat.com>
38089 PR middle-end/78703
38090 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38092 (struct format_result): Replace number_chars, number_chars_min,
38093 and number_chars_max with a single member of struct result_range.
38095 (format_result::operator+=): Adjust.
38096 (struct fmtresult): Remove bounded. Handle likely and unlikely
38098 (fmtresult::adjust_for_width_or_precision): New function.
38099 (fmtresult:type_max_digits): New function.
38100 (bytes_remaining): Handle likely and unlikely counters.
38101 (min_bytes_remaining): Remove.
38102 (format_percent): Simplify.
38103 (format_integer, format_floating): Set likely and unlikely counters.
38104 (get_string_length, format_character, format_string): Same.
38105 (format_plain, should_warn_p): New function.
38106 (maybe_warn): Call should_warn_p. Update diagnostic messages
38107 and handle those for all directives, including plain strings.
38108 (format_directive): Handle likely and unlikely counters.
38109 Remove unnecessary quoting from diagnostics. Add an informational
38111 (add_bytes): Remove.
38112 (pass_sprintf_length::compute_format_length): Simplify.
38113 (try_substitute_return_value): Handle likely and unlikely counters.
38115 2017-01-26 Carl Love <cel@us.ibm.com>
38117 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38118 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38120 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38123 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38124 endianess for subregs into account.
38125 * lra-constraints.c (lra_constraints): Do risky transformations
38126 always on the first iteration.
38127 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38128 last_call_used_reg_set.
38129 (process_bb_lives): Define and use last_call_used_reg_set.
38130 * lra.c (lra): Always continue after lra_constraints on the first
38133 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38135 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38137 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38139 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38141 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38142 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38143 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38144 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38145 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38146 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38147 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38148 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38149 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38151 2017-01-26 Marek Polacek <polacek@redhat.com>
38154 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38155 for the third operand.
38157 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38159 PR middle-end/79236
38160 * omp-low.c (struct omp_context): Add simt_stmt field.
38161 (scan_omp_for): Return omp_context *.
38162 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38163 context to the _simt_ SIMD stmt.
38164 (lower_omp_for): For combined SIMD with sibling _simt_
38165 SIMD, make sure to use the same decls in _looptemp_
38166 clauses as in the sibling.
38168 2017-01-26 David Sherwood <david.sherwood@arm.com>
38170 PR middle-end/79212
38171 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38174 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38177 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38178 emit fld b; fld a; if possible.
38180 * brig-builtins.def: Update copyright years.
38181 * config/arm/arm_acle_builtins.def: Update copyright years.
38183 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38186 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38187 constraint instead of o for the stxsd instruction.
38189 2017-01-25 Carl Love <cel@us.ibm.com>
38191 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38192 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38194 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38196 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38198 2017-01-25 Richard Biener <rguenther@suse.de>
38200 PR tree-optimization/69264
38201 * target.def (vector_alignment_reachable): Improve documentation.
38202 * doc/tm.texi: Regenerate.
38203 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38205 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38206 earlier changes with respect to TYPE_USER_ALIGN.
38207 (vector_alignment_reachable_p): Likewise. Improve dumping.
38209 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38212 * config/arm/arm.md (xordi3): Force constant operand into a register
38215 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38217 * doc/invoke.texi (-fstore-merging): Correct default optimization
38218 levels at which it is enabled.
38219 (-O): Move -fstore-merging from list to...
38222 2017-01-25 Richard Biener <rguenther@suse.de>
38225 * omp-expand.c: Include debug.h.
38226 (expand_omp_taskreg): Make sure to generate early debug before
38227 outlining anything from a function.
38228 (expand_omp_target): Likewise.
38229 (grid_expand_target_grid_body): Likewise.
38231 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38234 * asan.c (get_translation_unit_decl): New function.
38235 (asan_add_global): Extract modules file name from globals
38236 TRANSLATION_UNIT_DECL name.
38238 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38241 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38242 for long calls with APCS frame and VFP.
38244 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38246 * cfg.c (original_copy_tables_initialized_p): New function.
38247 * cfg.h (original_copy_tables_initialized_p): New decl.
38248 * cfgrtl.c (relink_block_chain): Guard the call to
38249 free_original_copy_tables with a call to
38250 original_copy_tables_initialized_p.
38251 * cgraph.h (symtab_node::native_rtl_p): New decl.
38252 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38253 (symtab_node::needed_p): Don't assert for early assembly output
38254 for __RTL functions.
38255 (cgraph_node::finalize_function): Set "force_output" for __RTL
38257 (cgraph_node::analyze): Bail out early for __RTL functions.
38258 (analyze_functions): Update assertion to support __RTL functions.
38259 (cgraph_node::expand): Bail out early for __RTL functions.
38260 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38262 * function.h (struct function): Update comment for field
38264 * gimple-expr.c: Include "tree-pass.h".
38265 (gimple_has_body_p): Return false for __RTL functions.
38266 * Makefile.in (OBJS): Add run-rtl-passes.o.
38267 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38269 (gcc::pass_manager::get_clean_slate): New accessor.
38270 * passes.c: Include "insn-addr.h".
38271 (should_skip_pass_p): Add logging. Update logic for running
38272 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38273 property-provider override so it is only done for gimple passes.
38275 (skip_pass): New function.
38276 (execute_one_pass): Call skip_pass when skipping passes.
38277 * read-md.c (md_reader::read_char): Support filtering
38278 the input to a subset of line numbers.
38279 (md_reader::md_reader): Initialize fields
38280 m_first_line and m_last_line.
38281 (md_reader::read_file_fragment): New function.
38282 * read-md.h (md_reader::read_file_fragment): New decl.
38283 (md_reader::m_first_line): New field.
38284 (md_reader::m_last_line): New field.
38285 * read-rtl-function.c (function_reader::create_function): Only
38286 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38287 curr_properties. Set DECL_INITIAL to a dummy block.
38288 (read_rtl_function_body_from_file_range): New function.
38289 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38291 * run-rtl-passes.c: New file.
38292 * run-rtl-passes.h: New file.
38294 2017-01-24 Jeff Law <law@redhat.com>
38296 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38299 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38301 PR tree-optimization/79159
38302 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38303 (record_nonwrapping_iv): Improve boundary using above function if no
38304 value range information.
38306 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38307 Martin Jambor <mjambor@suse.cz>
38309 * brig-builtins.def: New file.
38310 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38311 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38312 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38313 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38314 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38315 * builtin-types.def (BT_INT8): New.
38316 (BT_INT16): Likewise.
38317 (BT_UINT8): Likewise.
38318 (BT_UINT16): Likewise.
38319 (BT_FN_ULONG): Likewise.
38320 (BT_FN_UINT_INT): Likewise.
38321 (BT_FN_UINT_ULONG): Likewise.
38322 (BT_FN_UINT_LONG): Likewise.
38323 (BT_FN_UINT_PTR): Likewise.
38324 (BT_FN_ULONG_PTR): Likewise.
38325 (BT_FN_INT8_FLOAT): Likewise.
38326 (BT_FN_INT16_FLOAT): Likewise.
38327 (BT_FN_UINT32_FLOAT): Likewise.
38328 (BT_FN_UINT16_FLOAT): Likewise.
38329 (BT_FN_UINT8_FLOAT): Likewise.
38330 (BT_FN_UINT64_FLOAT): Likewise.
38331 (BT_FN_UINT16_UINT32): Likewise.
38332 (BT_FN_UINT32_UINT16): Likewise.
38333 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38334 (BT_FN_INT_PTR_INT): Likewise.
38335 (BT_FN_UINT_PTR_UINT): Likewise.
38336 (BT_FN_LONG_PTR_LONG): Likewise.
38337 (BT_FN_ULONG_PTR_ULONG): Likewise.
38338 (BT_FN_VOID_UINT64_UINT64): Likewise.
38339 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38340 (BT_FN_INT8_INT8_INT8): Likewise.
38341 (BT_FN_INT16_INT16_INT16): Likewise.
38342 (BT_FN_INT_INT_INT): Likewise.
38343 (BT_FN_UINT_FLOAT_UINT): Likewise.
38344 (BT_FN_FLOAT_UINT_UINT): Likewise.
38345 (BT_FN_ULONG_UINT_UINT): Likewise.
38346 (BT_FN_ULONG_UINT_PTR): Likewise.
38347 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38348 (BT_FN_UINT_UINT_UINT): Likewise.
38349 (BT_FN_VOID_UINT_PTR): Likewise.
38350 (BT_FN_UINT_UINT_PTR: Likewise.
38351 (BT_FN_UINT32_UINT64_PTR): Likewise.
38352 (BT_FN_INT_INT_UINT_UINT): Likewise.
38353 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38354 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38355 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38356 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38357 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38358 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38359 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38360 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38361 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38362 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38363 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38364 * doc/frontends.texi: List BRIG FE.
38365 * doc/install.texi (Testing): Add BRIG tesring requirements.
38366 * doc/invoke.texi (Overall Options): Mention BRIG.
38367 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38369 2017-01-24 Richard Biener <rguenther@suse.de>
38371 PR translation/79208
38372 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38374 2017-01-24 Martin Jambor <mjambor@suse.cz>
38377 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38378 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38379 and known_contexts.
38381 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38383 PR middle-end/79123
38384 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38385 casts from signed to unsigned really don't have a range.
38387 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38389 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38390 GMP_RNDx for compatiblity.
38392 2017-01-24 Martin Liska <mliska@suse.cz>
38395 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38396 that would prevent us to call alloca with -1 as argument.
38398 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38400 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38401 Avoid -Wformat-security warning.
38403 2017-01-23 Andrew Pinski <apinski@cavium.com>
38405 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38408 2017-01-23 Martin Sebor <msebor@redhat.com>
38410 PR middle-end/78703
38411 * gimple-ssa-sprintf.c (warn_level): New global.
38412 (format_integer): Use it here and throughout the rest of the file.
38413 Use the same switch to compute sign as base.
38414 (maybe_warn): New function.
38415 (format_directive): Factor out warnings into maybe_warn.
38416 Add debugging output. Use warn_level.
38417 (add_bytes): Use warn_level.
38418 (pass_sprintf_length::compute_format_length): Add debugging output.
38419 (try_substitute_return_value): Same.
38420 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38422 PR middle-end/78703
38423 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38424 (struct fmtresult, format_integer, format_floating): Adjust.
38425 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38426 (get_string_length, format_string,format_directive): Same.
38427 (pass_sprintf_length::compute_format_length): Same.
38428 (try_substitute_return_value): Simplify slightly.
38430 PR middle-end/78703
38431 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38432 (fmtresult::operator+=): Outlined.
38433 (struct fmtresult): Add ctors.
38434 (struct conversion_spec): Rename...
38435 (struct directive): ...to this. Add and remove data members.
38436 (directive::set_width, directive::set_precision): New functions.
38437 (format_percent): Use fmtresult ctor.
38438 (get_width_and_precision): Remove.
38439 (format_integer): Make naming changes. Avoid computing width and
38441 (format_floating): Same. Adjust indentation.
38442 (format_character, format_none): New functions.
38443 (format_string): Moved character handling to format_character.
38444 (format_directive): Remove arguments, change return type.
38445 (parse_directive): New function.
38446 (pass_sprintf_length::compute_format_length): Move directive
38447 parsing to parse_directive.
38449 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38451 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38452 (assign_assembler_name_if_needed): ... this.
38453 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38454 (assign_assembler_name_if_needed): ... this.
38455 (free_lang_data_in_cgraph): Adjust callers.
38456 * cgraphunit.c (cgraph_node::analyze): Likewise.
38457 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38460 2017-01-23 Richard Biener <rguenther@suse.de>
38462 PR tree-optimization/79088
38463 PR tree-optimization/79188
38464 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38465 resetting loop bounds after last path deletion. Reset loop
38466 bounds of the target loop, make code match the comments.
38467 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38468 Make sure loops need no fixups.
38470 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38472 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38473 exponent support with double type for first argument.
38474 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38475 type returned by __builtin_vec_extract_sig,
38476 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38477 functions from "vector int" to "vector unsigned int" or from
38478 "vector long long int" to "vector unsigned long long int".
38479 Changed type returned by __builtin_vec_extract_exp,
38480 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38481 functions from "vector int" to "vector unsigned int" or from
38482 "vector long long int" to "vector unsigned long long int".
38483 Changed return type of __builtin_vec_test_data_class,
38484 __builtin_vec_test_data_class_sp, and
38485 __builtin_vec_test_data_class_dp from "vector int" to
38486 "vector bool int" or from "vector long long int" to "vector bool
38487 long long int" and changed second argument type from "unsigned
38488 int" to "int". Added new overloaded function forms "vector float
38489 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38490 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38491 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38492 double, vector unsigned long long int)" and "vector double
38493 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38494 long int)". Changed return type of
38495 __builtin_scalar_test_data_class and
38496 __builtin_scalar_test_data_class_sp and
38497 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38498 int" and changed second argument from "unsigned int" to "int".
38499 Changed type returned by __builtin_scalar_test_neg,
38500 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38501 from "int" to "bool int". Added new overloaded function form
38502 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38503 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38504 exponent double-precision with floating point first argument.
38505 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38506 documentation of scalar_test_data_class, scalar_test_neg,
38507 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38508 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38509 vec_test_data_class built-in functions to reflect refinements in
38510 their type signatures.
38512 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38514 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38516 (aarch64_elf_asm_destructor): Likewise.
38518 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38520 PR rtl-optimization/78634
38521 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38522 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38523 * ifcvt.c (noce_try_cmove): Add missing cost check.
38525 PR rtl-optimization/71724
38526 * combine.c (if_then_else_cond): Look for situations where it is
38527 beneficial to undo the work of one of the recursive calls.
38529 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38531 PR tree-optimization/70754
38532 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38533 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38534 combined stmt before it if not NULL.
38535 (combine_chains): Process refs reversely and compute dominance point
38538 2017-01-23 Martin Liska <mliska@suse.cz>
38540 PR tree-optimization/79196
38541 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38542 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38544 (strlen_optimize_stmt): Call the renamed function.
38546 2017-01-23 Michael Matz <matz@suse.de>
38548 PR tree-optimization/78384
38549 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38551 2017-01-23 Richard Biener <rguenther@suse.de>
38553 PR tree-optimization/79186
38554 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38555 both incoming edges before moving an assert.
38557 2017-01-23 Martin Jambor <mjambor@suse.cz>
38559 * ipa-prop.c (load_from_param_1): Removed.
38560 (load_from_unmodified_param): Bits from load_from_param_1 put back
38562 (load_from_param): Removed.
38563 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38564 with stmt. Reverted back to use of load_from_unmodified_param.
38566 2017-01-23 Martin Jambor <mjambor@suse.cz>
38569 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38570 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38571 field a pointer to garbage collected vector, mark lattices and
38572 ipcp_orig_node with GTY((skip)).
38573 (ipa_get_param_count): Adjust to descriptors being a pointer.
38574 (ipa_get_param): Likewise.
38575 (ipa_get_type): Likewise.
38576 (ipa_get_param_move_cost): Likewise.
38577 (ipa_set_param_used): Likewise.
38578 (ipa_get_controlled_uses): Likewise.
38579 (ipa_set_controlled_uses): Likewise.
38580 (ipa_is_param_used): Likewise.
38581 (ipa_node_params_t): Move into garbage collector. New methods insert
38583 (ipa_node_params_sum): Annotate wth GTY(()).
38584 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38586 (ipa_load_from_parm_agg): Adjust declaration.
38587 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38588 * ipa-profile.c (ipa_profile): Likewise.
38589 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38590 (ipa_populate_param_decls): Make descriptors parameter garbage
38592 (ipa_dump_param): Adjust to descriptors being a pointer.
38593 (ipa_alloc_node_params): Likewise.
38594 (ipa_initialize_node_params): Likewise.
38595 (load_from_param_1): Make descriptors parameter garbage collected.
38596 (load_from_unmodified_param): Likewise.
38597 (load_from_param): Likewise.
38598 (ipa_load_from_parm_agg): Likewise.
38599 (ipa_node_params::~ipa_node_params): Removed.
38600 (ipa_free_all_node_params): Remove call to delete operator.
38601 (ipa_node_params_t::insert): New.
38602 (ipa_node_params_t::remove): Likewise.
38603 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38604 copy known_csts and known_contexts vectors.
38605 (ipa_read_node_info): Adjust to descriptors being a pointer.
38606 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38608 (ipcp_transform_function): Make descriptors variable garbage
38611 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38613 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38614 * config/i386/avx512dqintrin.h: Ditto.
38615 * config/i386/avx512fintrin.h: Ditto.
38616 * config/i386/i386.c: Handle new builtins.
38617 * config/i386/i386-builtin.def: Add new builtins.
38618 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38619 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38621 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38622 Martin Liska <mliska@suse.cz>
38624 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38625 * asan.c (asan_expand_poison_ifn): Support stores and use
38626 appropriate ASAN report function.
38627 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38628 * internal-fn.def (ASAN_POISON_USE): Declare.
38629 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38630 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38631 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38632 ASAN_POISON calls w/o LHS.
38633 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38634 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38635 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38636 * gimplify.c (asan_poison_variables): Add attribute
38637 use_after_scope_memory to variables that really needs to live
38639 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38640 having the attribute.
38642 2017-01-23 Martin Liska <mliska@suse.cz>
38644 * asan.c (create_asan_shadow_var): New function.
38645 (asan_expand_poison_ifn): Likewise.
38646 * asan.h (asan_expand_poison_ifn): New declaration.
38647 * internal-fn.c (expand_ASAN_POISON): Likewise.
38648 * internal-fn.def (ASAN_POISON): New builtin.
38649 * sanopt.c (pass_sanopt::execute): Expand
38650 asan_expand_poison_ifn.
38651 * tree-inline.c (copy_decl_for_dup_finish): Make function
38653 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38654 * tree-ssa.c (is_asan_mark_p): New function.
38655 (execute_update_addresses_taken): Rewrite local variables
38656 (identified just by use-after-scope as addressable) into SSA.
38658 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38660 * doc/install.texi (Specific): opensource.apple.com uses https
38661 now. Remove trailing slash.
38663 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38665 * README.Portability: Remove note on an Irix compatibility issue.
38667 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38669 * gcov.c (INCLUDE_ALGORITHM): Define.
38670 (INCLUDE_VECTOR): Define.
38671 No longer include <vector> and <algorithm> directly.
38673 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38675 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38677 * doc/invoke.texi (Code Gen Options): Ditto.
38679 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38682 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38684 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38686 rtl-optimization/79125
38687 * cprop.c (local_cprop_pass): Handle cases where we make an
38688 unconditional trap.
38690 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38694 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38695 read from, for big endian.
38697 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38699 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38700 register pauth builtins for LP64 only.
38702 2017-01-20 Marek Polacek <polacek@redhat.com>
38705 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38708 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38710 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38712 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38713 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38714 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38716 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38719 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38720 in big-endian mode when they are not a single duplicated value.
38722 2017-01-20 Richard Biener <rguenther@suse.de>
38724 * BASE-VER: Bump to 7.0.1.
38726 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38728 * omp-low.c (omplow_simd_context): New struct. Use it...
38729 (lower_rec_simd_input_clauses): ...here and...
38730 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38731 references to idx, lane, max_vf, is_simt.
38733 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38735 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38738 2017-01-20 Martin Jambor <mjambor@suse.cz>
38740 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38741 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38743 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38744 (GTFILES): Rename hsa.c to hsa-common.c.
38745 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38746 * hsa-dump.c: Likewise.
38747 * hsa-gen.c: Likewise.
38748 * hsa-regalloc.c: Likewise.
38749 * ipa-hsa.c: Likewise.
38750 * omp-expand.c: Likewise.
38751 * omp-low.c: Likewise.
38752 * toplev.c: Likewise.
38754 2017-01-20 Marek Polacek <polacek@redhat.com>
38757 * doc/invoke.texi: Document -Wduplicated-branches.
38758 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38759 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38760 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38761 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38762 return 0 only when not OEP_LEXICOGRAPHIC.
38763 (fold_build_cleanup_point_expr): Use the expression
38764 location when building CLEANUP_POINT_EXPR.
38765 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38766 * tree.c (add_expr): Handle error_mark_node.
38768 2017-01-20 Martin Liska <mliska@suse.cz>
38771 * tree-profile.c (init_ic_make_global_vars): Do not call
38773 (gimple_init_gcov_profiler): Likewise.
38775 2017-01-20 Martin Liska <mliska@suse.cz>
38778 * cgraph.h (maybe_create_reference): Remove argument and
38780 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38782 * ipa-cp.c (create_specialized_node): Likewise.
38783 * symtab.c (symtab_node::maybe_create_reference): Handle
38784 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38786 2017-01-20 Martin Liska <mliska@suse.cz>
38788 * read-rtl-function.c (function_reader::create_function): Use
38789 build_decl instread of build_decl_stat.
38791 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38793 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38794 * config/i386/avx512dqintrin.h: Ditto.
38795 * config/i386/avx512fintrin.h: Ditto.
38796 * config/i386/i386-builtin-types.def: Add new types.
38797 * config/i386/i386.c: Handle new types.
38798 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38799 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38800 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38801 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38802 (__builtin_ia32_kshiftridi): New.
38803 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38805 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38809 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38810 define to rs6000_init_stack_protect_guard.
38811 (rs6000_init_stack_protect_guard): New function.
38813 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38814 Yunqiang Su <yunqiang.su@imgtec.com>
38816 * config.gcc (supported_defaults): Add madd4.
38817 (with_madd4): Add validation.
38818 (all_defaults): Add madd4.
38819 * config/mips/mips.opt (mmadd4): New option.
38820 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38822 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38824 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38825 (ISA_HAS_FUSED_MADD4): Likewise.
38826 * doc/invoke.texi (-mmadd4): Document the new option.
38827 * doc/install.texi (--with-madd4): Document the new option.
38829 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38831 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38832 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38833 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38834 (aarch64_init_pauth_hint_builtins): New.
38835 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38836 (aarch64_expand_builtin): Expand new builtins.
38838 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38840 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38841 * combine-stack-adj.c (no_unhandled_cfa): Handle
38842 REG_CFA_TOGGLE_RA_MANGLE.
38843 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38844 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38845 info for return address signing.
38846 (aarch64_expand_epilogue): Likewise.
38848 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38850 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38851 * config/aarch64/aarch64-protos.h
38852 (aarch64_return_address_signing_enabled): New declaration.
38853 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38855 (aarch64_expand_prologue): Sign return address before it's pushed onto
38857 (aarch64_expand_epilogue): Authenticate return address fetched from
38859 (aarch64_override_options): Sanity check for ILP32 and ISA level.
38860 (aarch64_attributes): New function attributes for "sign-return-address".
38861 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
38862 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
38863 ("*do_return"): Generate combined instructions according to key index.
38864 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
38865 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
38867 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
38868 * config/aarch64/aarch64.opt (msign-return-address=): New.
38869 * doc/extend.texi (AArch64 Function Attributes): Documents
38870 "sign-return-address=".
38871 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
38873 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38875 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
38876 overall option summary.
38878 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38880 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
38881 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
38882 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
38883 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
38885 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
38887 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
38888 -mpower9-minmax by default for -mcpu=power9.
38889 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
38890 128-bit floating point.
38892 2017-01-20 Alan Modra <amodra@gmail.com>
38894 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
38895 optimizing for size.
38897 2017-01-20 Alan Modra <amodra@gmail.com>
38900 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
38901 for strcmp and strncmp from corresponding builtin decl.
38903 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38905 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
38906 instead of i386/rtems-64.h.
38907 * config/i386/rtems-64.h: Remove.
38909 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38913 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
38915 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
38917 2017-01-19 Tamar Christina <tamar.christina@arm.com>
38919 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
38920 Change int to HOST_WIDE_INT.
38921 * config/aarch64/aarch64-protos.h
38922 (aarch64_simd_gen_const_vector_dup): Likewise.
38923 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
38925 2017-01-19 David Malcolm <dmalcolm@redhat.com>
38927 * langhooks-def.h (lhd_type_for_size): New decl.
38928 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
38929 * langhooks.c (lhd_type_for_size): New function, taken from
38932 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
38934 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
38935 define_bypass for CR latency.
38936 (power9-cracked-alu): Update bypass latency and remove power9-branch.
38937 (power9-alu2): Add define_bypass for CR latency.
38939 (power9-mul): Update insn latency.
38940 (power9-mul-compare): Update insn latency, bypass latency and remove
38943 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38945 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
38947 * config/aarch64/aarch64.md
38948 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
38949 aarch64_nopcrelative_literal_loads.
38950 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
38952 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
38954 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
38955 TARGET_LOONGSON_3A.
38956 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
38958 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
38961 * config.gcc (supported_defaults): Add lxc1-sxc1.
38962 (with_lxc1_sxc1): Add validation.
38963 (all_defaults): Add lxc1-sxc1.
38964 * config/mips/mips.opt (mlxc1-sxc1): New option.
38965 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38967 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38968 __mips_no_lxc1_sxc1.
38969 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
38970 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
38971 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
38973 2017-01-19 Richard Biener <rguenther@suse.de>
38975 PR tree-optimization/72488
38976 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
38977 sure to restore SSA info.
38978 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
38980 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
38982 PR rtl-optimization/79121
38983 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
38984 of the inner type when shifting an extended value.
38986 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38989 * symtab.c (symtab_node::equal_address_to): Fix comparing of
38990 interposable aliases.
38992 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
38995 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
38996 Use the evmergelohi instruction.
38997 (mov_si<mode>_e500_subreg4_2_le): Likewise.
38998 (mov_sitf_e500_subreg8_2_be): Likewise.
38999 (mov_sitf_e500_subreg12_2_le): Likewise.
39000 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39001 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39002 (mov_sitf_e500_subreg8_2_le): Likewise.
39003 (mov_sitf_e500_subreg12_2_be): Likewise.
39005 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39007 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39008 attribute from vecsimple to vecperm.
39009 (altivec_vbpermq2): Likewise.
39011 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39014 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39016 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39017 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39018 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39019 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39020 case where N arg is SIZE_MAX.
39021 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39022 (cmpstrsi): Add pattern.
39024 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39026 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39027 __builtin_vec_revb builtins.
39028 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39029 built-in functions to support generation of the ISA 3.0 XXBR<x>
39030 vector byte reverse instructions.
39031 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39032 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39033 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39034 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39035 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39036 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39037 (P9V_BUILTIN_VEC_REVB): Likewise.
39038 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39039 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39040 (p9_xxbrq_v16qi): Likewise.
39041 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39042 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39043 (p9_xxbrh_v8hi): Likewise.
39044 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39045 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39046 vec_revb built-in functions.
39048 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39050 PR rtl-optimization/78952
39051 * config/i386/i386.md (any_extract): New code iterator.
39052 (*insvqi_2): Use any_extract for source operand.
39053 (*insvqi_3): Use any_shiftrt for source operand.
39055 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39057 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39059 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39061 2017-01-18 Matthias Klose <doko@ubuntu.com>
39063 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39065 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39067 * config/rs6000/altivec.h (vec_bperm): Change #define.
39068 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39069 (altivec_vbpermq2): New define_insn.
39070 (altivec_vbpermd): Likewise.
39071 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39072 function interface.
39073 (VBPERMD): Likewise.
39074 (VBPERM): New polymorphic function interface.
39075 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39076 Add entries for P9V_BUILTIN_VEC_VBPERM.
39077 * doc/extend.texi: Add interfaces for vec_bperm.
39079 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39081 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39082 first letter of error messages.
39083 (s390_resolve_overloaded_builtin): Likewise.
39084 * config/s390/s390.c (s390_expand_builtin): Likewise.
39085 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39086 (s390_valid_target_attribute_inner_p): Likewise.
39087 * config/s390/s390.md ("tabort"): Likewise.
39089 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39091 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39092 (ISA_AVOID_DIV_HILO): New macro.
39093 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39094 (ISA_HAS_DDIV): Likewise.
39096 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39098 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39100 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39102 * doc/invoke.texi (fabi-version): Spelling fix.
39104 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39107 * doc/invoke.texi (fabi-version): Mention mangling fix for
39110 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39113 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39115 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39118 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39119 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39121 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39123 (stack_protect_test): Ditto.
39124 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39125 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39127 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39128 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39129 -mstack-protector-guard-offset=.
39130 (RS/6000 and PowerPC Options): Ditto.
39132 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39134 * config/i386/i386.h (MASK_CLASS_P): New define.
39135 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39136 there are no registers from different register sets also when
39137 mask registers are used. Update function comment.
39138 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39139 to (*k/*r) and (*k/*km) alternatives.
39141 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39143 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39144 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39145 (EH_RETURN_HANDLER_RTX): New define.
39146 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39147 Force frame pointer in EH return functions.
39148 (aarch64_expand_epilogue): Add barrier for eh_return.
39149 (aarch64_final_eh_return_addr): Remove.
39150 (aarch64_eh_return_handler_rtx): New function.
39151 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39153 (aarch64_eh_return_handler_rtx): New prototype.
39155 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39157 * config/rs6000/altivec.h (vec_rlmi): New #define.
39158 (vec_vrlnm): Likewise.
39159 (vec_rlnm): Likewise.
39160 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39161 (UNSPEC_VRLNM): Likewise.
39162 (VIlong): New mode iterator.
39163 (altivec_vrl<VI_char>mi): New define_insn.
39164 (altivec_vrl<VI_char>nm): Likewise.
39165 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39167 (VRLDNM): Likewise.
39168 (RLNM): New polymorphic function entry.
39169 (VRLWMI): New monomorphic function entry.
39170 (VRLDMI): Likewise.
39171 (RLMI): New polymorphic function entry.
39172 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39173 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39174 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39177 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39180 * dwarf2out.c (field_byte_offset): Restore the
39181 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39182 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39183 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39186 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39189 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39191 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39194 * dwarf2out.c (add_data_member_location_attribute): For constant
39195 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39196 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39197 DW_AT_byte_size attributes.
39199 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39201 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39202 after forcing to constant memory when the code model is medium.
39204 2017-01-17 Julia Koval <julia.koval@intel.com>
39207 * config/i386/avx512fintrin.h
39208 (_mm512_i32gather_ps): Change __addr type to void const*.
39209 (_mm512_mask_i32gather_ps): Ditto.
39210 (_mm512_i32gather_pd): Ditto.
39211 (_mm512_mask_i32gather_pd): Ditto.
39212 (_mm512_i64gather_ps): Ditto.
39213 (_mm512_mask_i64gather_ps): Ditto.
39214 (_mm512_i64gather_pd): Ditto.
39215 (_mm512_mask_i64gather_pd): Ditto.
39216 (_mm512_i32gather_epi32): Ditto.
39217 (_mm512_mask_i32gather_epi32): Ditto.
39218 (_mm512_i32gather_epi64): Ditto.
39219 (_mm512_mask_i32gather_epi64): Ditto.
39220 (_mm512_i64gather_epi32): Ditto.
39221 (_mm512_mask_i64gather_epi32): Ditto.
39222 (_mm512_i64gather_epi64): Ditto.
39223 (_mm512_mask_i64gather_epi64): Ditto.
39224 (_mm512_i32scatter_ps): Change __addr type to void*.
39225 (_mm512_mask_i32scatter_ps): Ditto.
39226 (_mm512_i32scatter_pd): Ditto.
39227 (_mm512_mask_i32scatter_pd): Ditto.
39228 (_mm512_i64scatter_ps): Ditto.
39229 (_mm512_mask_i64scatter_ps): Ditto.
39230 (_mm512_i64scatter_pd): Ditto.
39231 (_mm512_mask_i64scatter_pd): Ditto.
39232 (_mm512_i32scatter_epi32): Ditto.
39233 (_mm512_mask_i32scatter_epi32): Ditto.
39234 (_mm512_i32scatter_epi64): Ditto.
39235 (_mm512_mask_i32scatter_epi64): Ditto.
39236 (_mm512_i64scatter_epi32): Ditto.
39237 (_mm512_mask_i64scatter_epi32): Ditto.
39238 (_mm512_i64scatter_epi64): Ditto.
39239 (_mm512_mask_i64scatter_epi64): Ditto.
39240 * config/i386/avx512pfintrin.h
39241 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39242 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39243 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39244 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39245 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39246 (_mm512_prefetch_i32scatter_ps): Ditto.
39247 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39248 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39249 (_mm512_prefetch_i64scatter_pd): Ditto.
39250 (_mm512_prefetch_i64scatter_ps): Ditto.
39251 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39252 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39253 * config/i386/avx512vlintrin.h
39254 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39255 (_mm_mmask_i32gather_ps): Ditto.
39256 (_mm256_mmask_i32gather_pd): Ditto.
39257 (_mm_mmask_i32gather_pd): Ditto.
39258 (_mm256_mmask_i64gather_ps): Ditto.
39259 (_mm_mmask_i64gather_ps): Ditto.
39260 (_mm256_mmask_i64gather_pd): Ditto.
39261 (_mm_mmask_i64gather_pd): Ditto.
39262 (_mm256_mmask_i32gather_epi32): Ditto.
39263 (_mm_mmask_i32gather_epi32): Ditto.
39264 (_mm256_mmask_i32gather_epi64): Ditto.
39265 (_mm_mmask_i32gather_epi64): Ditto.
39266 (_mm256_mmask_i64gather_epi32): Ditto.
39267 (_mm_mmask_i64gather_epi32): Ditto.
39268 (_mm256_mmask_i64gather_epi64): Ditto.
39269 (_mm_mmask_i64gather_epi64): Ditto.
39270 (_mm256_i32scatter_ps): Change __addr type to void*.
39271 (_mm256_mask_i32scatter_ps): Ditto.
39272 (_mm_i32scatter_ps): Ditto.
39273 (_mm_mask_i32scatter_ps): Ditto.
39274 (_mm256_i32scatter_pd): Ditto.
39275 (_mm256_mask_i32scatter_pd): Ditto.
39276 (_mm_i32scatter_pd): Ditto.
39277 (_mm_mask_i32scatter_pd): Ditto.
39278 (_mm256_i64scatter_ps): Ditto.
39279 (_mm256_mask_i64scatter_ps): Ditto.
39280 (_mm_i64scatter_ps): Ditto.
39281 (_mm_mask_i64scatter_ps): Ditto.
39282 (_mm256_i64scatter_pd): Ditto.
39283 (_mm256_mask_i64scatter_pd): Ditto.
39284 (_mm_i64scatter_pd): Ditto.
39285 (_mm_mask_i64scatter_pd): Ditto.
39286 (_mm256_i32scatter_epi32): Ditto.
39287 (_mm256_mask_i32scatter_epi32): Ditto.
39288 (_mm_i32scatter_epi32): Ditto.
39289 (_mm_mask_i32scatter_epi32): Ditto.
39290 (_mm256_i32scatter_epi64): Ditto.
39291 (_mm256_mask_i32scatter_epi64): Ditto.
39292 (_mm_i32scatter_epi64): Ditto.
39293 (_mm_mask_i32scatter_epi64): Ditto.
39294 (_mm256_i64scatter_epi32): Ditto.
39295 (_mm256_mask_i64scatter_epi32): Ditto.
39296 (_mm_i64scatter_epi32): Ditto.
39297 (_mm_mask_i64scatter_epi32): Ditto.
39298 (_mm256_i64scatter_epi64): Ditto.
39299 (_mm256_mask_i64scatter_epi64): Ditto.
39300 (_mm_i64scatter_epi64): Ditto.
39301 (_mm_mask_i64scatter_epi64): Ditto.
39302 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39303 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39304 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39305 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39306 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39307 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39308 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39309 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39310 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39311 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39312 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39313 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39314 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39315 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39316 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39317 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39318 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39319 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39320 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39321 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39322 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39323 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39324 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39325 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39326 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39327 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39328 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39329 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39330 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39331 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39332 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39333 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39334 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39335 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39336 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39337 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39338 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39339 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39340 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39341 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39342 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39343 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39344 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39345 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39346 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39347 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39348 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39349 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39350 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39351 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39352 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39353 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39354 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39355 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39356 definitions accordingly.
39358 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39359 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39362 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39365 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39368 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39369 hard regs for inner regno.
39371 2017-01-17 Martin Liska <mliska@suse.cz>
39374 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39375 assumption and add comment.
39377 2017-01-17 Nathan Sidwell <nathan@acm.org>
39379 * ipa-visibility.c (localize_node): New function, broken out of ...
39380 (function_and_variable_visibility): ... here. Call it.
39382 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39384 PR middle-end/77445
39385 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39386 correctly set frequency of oudgoing edge.
39387 (duplicate_thread_path): Fix profile updating.
39389 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39392 * configure.ac: Add GCC_BASE_VER.
39393 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39394 version from BASE-VER file.
39395 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39396 (gcc.o): Depend on $(BASEVER).
39397 * common.opt (dumpfullversion): New option.
39398 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39399 * doc/invoke.texi: Document -dumpfullversion.
39400 * doc/install.texi: Document --with-gcc-major-version-only.
39401 * configure: Regenerated.
39403 2017-01-17 Richard Biener <rguenther@suse.de>
39405 PR tree-optimization/71433
39406 * tree-vrp.c (register_new_assert_for): Merge same asserts
39407 on all incoming edges.
39408 (process_assert_insertions_for): Handle insertions at the
39411 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39413 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39414 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39416 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39419 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39422 2017-01-17 Alan Modra <amodra@gmail.com>
39425 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39426 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39427 symbolic stack limit when pic.
39429 2017-01-16 Martin Sebor <msebor@redhat.com>
39431 PR tree-optimization/78608
39432 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39434 2017-01-16 Jeff Law <law@redhat.com>
39437 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39438 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39439 for several include directories that may be relative to sysroot.
39440 * config/i386/x-mingw32 (gplus_includedir): Define.
39441 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39442 (native_system_includedir): Likewise.
39443 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39444 override if TARGET_SYSTEM_ROOT is defined.
39445 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39447 PR tree-optimization/79090
39448 PR tree-optimization/33562
39449 PR tree-optimization/61912
39450 PR tree-optimization/77485
39451 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39452 and computed trims into the dump file.
39454 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39456 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39458 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39461 * gimplify.c (gimplify_init_constructor): If want_value and
39462 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39466 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39467 sequence. Formatting fixes.
39468 (doloop_optimize): Formatting fixes.
39471 * gcc.c (debug_level_greater_than_spec_func): New function.
39472 (static_spec_functions): Add debug-level-gt spec function.
39473 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39475 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39476 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39477 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39478 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39479 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39480 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39482 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39484 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39485 QImode fixups to general and mask registers only.
39487 2017-01-16 Carl Love <cel@us.ibm.com>
39489 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39490 for built-in functions
39491 vector signed char vec_nabs (vector signed char)
39492 vector signed short vec_nabs (vector signed short)
39493 vector signed int vec_nabs (vector signed int)
39494 vector signed long long vec_nabs (vector signed long long)
39495 vector float vec_nabs (vector float)
39496 vector double vec_nabs (vector double)
39497 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39499 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39500 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39501 * doc/extend.texi: Update the documentation file for the new built-in
39504 2017-01-16 Martin Sebor <msebor@redhat.com>
39506 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39509 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39511 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39512 UNSPEC_VSX__XXSPLTD to require special splat handling.
39514 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39517 * system.h: Poison strndup.
39519 2017-01-16 Alan Modra <amodra@gmail.com>
39522 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39525 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39527 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39529 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39531 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39532 call recog here. Assert that INSN_CODE (insn) is non-negative.
39534 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39537 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39539 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39540 in the currently scheduled RTL fragment.
39542 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39544 PR rtl-optimization/78751
39545 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39548 2017-01-14 Jeff Law <law@redhat.com>
39550 PR tree-optimization/79090
39551 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39552 variable length stores.
39553 (compute_trims): Delete dead assignment to *trim_tail.
39554 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39557 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39559 PR rtl-optimization/78626
39560 PR rtl-optimization/78727
39561 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39562 of a block, and split such blocks after everything else is finished.
39564 2017-01-14 Alan Modra <amodra@gmail.com>
39567 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39568 target legitimate_combined_insn.
39569 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39570 (rs6000_legitimate_combined_insn): New function.
39571 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39573 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39574 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39575 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39577 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39579 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39581 2017-01-13 Jeff Law <law@redhat.com>
39583 PR tree-optimization/33562
39584 PR tree-optimization/61912
39585 PR tree-optimization/77485
39586 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39588 (delete_dead_assignment): Likewise.
39589 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39590 statement to delete_dead_call and delete_dead_assignment.
39592 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39595 * substring-locations.c (format_warning_va): Strengthen case 1 so
39596 that both endpoints of the substring must be within the format
39597 range for just the substring to be printed.
39599 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39601 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39602 * config/i386/i386.c (ix86_target_string): Add missing options
39603 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39604 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39605 flags_other and ix86_target_other to flags2_other. Display unknown
39607 (ix86_valid_target_attribute_inner_p): Add missing options and
39608 reorder options by implied ISAs, as in ix86_target_string.
39610 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39612 * hash-table.h (hash_table::too_empty_p): New function.
39613 (hash_table::expand): Use it.
39614 (hash_table::traverse): Likewise.
39615 (hash_table::empty_slot): Use sizeof (value_type) instead of
39616 sizeof (PTR) to convert bytes to elements. Shrink the table
39617 if the current size is excessive for the current number of
39620 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39622 * ira-costs.c (record_reg_classes): Break from the inner loop
39623 early once alt_fail is known to be true. Update outer loop
39624 handling accordingly.
39626 2017-01-13 Jeff Law <law@redhat.com>
39628 * tree-ssa-dse.c (decrement_count): New function.
39629 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39630 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39631 when we know the partially dead statement is a mem* function.
39633 PR tree-optimization/61912
39634 PR tree-optimization/77485
39635 * tree-ssa-dse.c: Include expr.h.
39636 (maybe_trim_constructor_store): New function.
39637 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39639 PR tree-optimization/33562
39640 PR tree-optimization/61912
39641 PR tree-optimization/77485
39642 * doc/invoke.texi: Document new dse-max-object-size param.
39643 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39644 * tree-ssa-dse.c: Include params.h.
39645 (dse_store_status): New enum.
39646 (initialize_ao_ref_for_dse): New, partially extracted from
39648 (valid_ao_ref_for_dse, normalize_ref): New.
39649 (setup_live_bytes_from_ref, compute_trims): Likewise.
39650 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39651 (maybe_trim_partially_dead_store): Likewise.
39652 (maybe_trim_complex_store): Likewise.
39653 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39654 Track what bytes live from the original store. Return tri-state
39655 for dead, partially dead or live.
39656 (dse_dom_walker): Add constructor, destructor and new private members.
39657 (delete_dead_call, delete_dead_assignment): New extracted from
39659 (dse_optimize_stmt): Make a member of dse_dom_walker.
39660 Use initialize_ao_ref_for_dse.
39662 PR tree-optimization/33562
39663 PR tree-optimization/61912
39664 PR tree-optimization/77485
39665 * sbitmap.h (bitmap_count_bits): Prototype.
39666 (bitmap_clear_range, bitmap_set_range): Likewise.
39667 * sbitmap.c (bitmap_clear_range): New function.
39668 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39670 2017-01-13 Martin Liska <mliska@suse.cz>
39673 * function.c (set_cfun): Add new argument force.
39674 * function.h (set_cfun): Likewise.
39675 * ipa-inline-transform.c (inline_call): Use the function when
39676 strict alising from is dropped for function we inline to.
39678 2017-01-13 Richard Biener <rguenther@suse.de>
39680 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39681 for dumping GIMPLE INTEGER_CSTs.
39683 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39685 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39686 to 201112L since C++17.
39688 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39691 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39692 if -fsanitize=kernel-address is present.
39694 2017-01-13 Richard Biener <rguenther@suse.de>
39696 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39697 as _Literal ( type ) number in case usual suffixes do not
39698 preserve all information.
39700 2017-01-13 Richard Biener <rguenther@suse.de>
39702 PR tree-optimization/77283
39703 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39704 and ssa-iterators.h.
39705 (is_feasible_trace): Implement a cost model based on joiner
39708 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39711 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39712 char or short to __float128/_Float128 directly.
39714 2017-01-12 Martin Sebor <msebor@redhat.com>
39716 to -Wformat-overflow.
39717 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39718 (min_bytes_remaining): Same.
39719 (get_string_length): Same.
39720 (format_string): Same.
39721 (format_directive): Same.
39723 (pass_sprintf_length::handle_gimple_call): Same.
39725 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39727 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39728 info.nowrite calls with no lhs that can't throw. Return bool
39729 whether gsi_remove has been called or not.
39730 (pass_sprintf_length::handle_gimple_call): Return bool whether
39731 try_substitute_return_value called gsi_remove. Formatting fix.
39732 (pass_sprintf_length::execute): Don't use gsi_remove if
39733 handle_gimple_call returned true.
39736 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39737 be removed due to side-effects, don't remove following barrier nor
39738 turn the successor edge into fallthru edge.
39740 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39743 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39744 element-reversing loads and stores as not swappable.
39746 2017-01-12 Nathan Sidwell <nathan@acm.org>
39747 Nicolai Stange <nicstange@gmail.com>
39749 * combine.c (try_combine): Don't ignore result of overlap checking
39750 loop. Combine overlap & asm check into single loop.
39752 2017-01-12 Richard Biener <rguenther@suse.de>
39754 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39755 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39757 2017-01-12 Richard Biener <rguenther@suse.de>
39759 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39760 and TS_TARGET_OPTION directly derive from TS_BASE.
39761 * tree-core.h (tree_optimization_option): Derive from tree_base.
39762 (tree_target_option): Likewise.
39764 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39766 * config/i386/i386.c (memory_address_length): Increase len
39767 only when rip_relative_addr_p returns false.
39769 2017-01-11 Julia Koval <julia.koval@intel.com>
39771 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39772 (OPTION_MASK_ISA_SGX_SET): New.
39773 (ix86_handle_option): Handle OPT_msgx.
39774 * config.gcc: Added sgxintrin.h.
39775 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39776 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39777 * config/i386/i386.c (ix86_target_string): Add -msgx.
39779 (ix86_option_override_internal): Handle new options.
39780 (ix86_valid_target_attribute_inner_p): Add sgx.
39781 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39782 * config/i386/i386.opt: Add msgx.
39783 * config/i386/sgxintrin.h: New file.
39784 * config/i386/x86intrin.h: Add sgxintrin.h.
39786 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39789 * fold-const.c (maybe_nonzero_address): Return 1 for function
39791 (tree_single_nonzero_warnv_p): Don't handle function local objects
39795 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39798 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39801 * opts.c: Include "spellcheck.h"
39802 (struct string_fragment): New struct.
39803 (struct edit_distance_traits<const string_fragment &>): New
39805 (get_closest_sanitizer_option): New function.
39806 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39808 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39810 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39812 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39813 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39814 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39815 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39816 for initial die_offset if dwarf_split_debug_info.
39817 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39818 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39820 (output_skeleton_debug_sections): Formatting fix. Use
39821 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39822 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39824 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39826 * config/arm/cortex-a53.md: Add bypasses for
39827 cortex_a53_r2f_cvt.
39828 (cortex_a53_r2f): Only use for transfers.
39829 (cortex_a53_f2r): Likewise.
39830 (cortex_a53_r2f_cvt): Add reservation for conversions.
39831 (cortex_a53_f2r_cvt): Likewise.
39833 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39835 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39836 to all inlined functions, change static to extern.
39838 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39841 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39843 (arm_assemble_integer): Likewise.
39845 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39847 * config.gcc: Use new awk script to check CPU, FPU and architecture
39848 parameters for --with-... options.
39849 * config/arm/parsecpu.awk: New file
39850 * config/arm/arm-cpus.in: New file.
39851 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39853 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39855 * config/arm/t-arm: Update dependency rules.
39856 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39857 of processing .def files.
39858 * config/arm/genopt.sh: Deleted.
39859 * config/arm/gentune.sh: Deleted.
39860 * config/arm/arm-cores.def: Deleted.
39861 * config/arm/arm-arches.def: Deleted.
39862 * config/arm/arm-fpus.def: Deleted.
39863 * config/arm/arm-tune.md: Regenerated.
39864 * config/arm/arm-tables.opt: Regenerated.
39865 * config/arm/arm-cpu.h: New generated file.
39866 * config/arm/arm-cpu-data.h: New generated file.
39867 * config/arm/arm-cpu-cdata.h: New generated file.
39869 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
39872 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
39874 (input_varpool_node): Unpack dynamically_initialized bit.
39876 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
39878 PR rtl-optimization/79032
39879 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
39880 the alignment of the adjusted memory reference against that of MODE,
39881 instead of the alignment of the original memory reference.
39883 2017-01-11 Martin Jambor <mjambor@suse.cz>
39885 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
39887 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
39888 decorated functions.
39890 2017-01-11 Richard Biener <rguenther@suse.de>
39892 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
39893 set range/nonnull info for PHI results. Do not set it on
39894 stmts marked for removal.
39896 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
39898 * expr.c (store_field): In the bitfield case, fetch the return value
39899 from the registers before applying a single big-endian adjustment.
39900 Always do a final load for a BLKmode value not larger than a word.
39902 2017-01-10 David Malcolm <dmalcolm@redhat.com>
39905 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39906 that we correctly handle column numbers greater than
39907 LINE_MAP_MAX_COLUMN_NUMBER.
39909 2017-01-10 Martin Sebor <msebor@redhat.com>
39911 PR middle-end/78245
39912 * gimple-ssa-sprintf.c (get_destination_size): Call
39913 {init,fini}object_sizes.
39914 * tree-object-size.c (addr_object_size): Adjust.
39915 (pass_through_call): Adjust.
39916 (pass_object_sizes::execute): Adjust.
39917 * tree-object-size.h (fini_object_sizes): Declare.
39919 2017-01-10 Martin Sebor <msebor@redhat.com>
39921 PR tree-optimization/78775
39922 * builtins.c (get_size_range): Move...
39923 * calls.c: ...to here.
39924 (alloc_max_size): Accept zero argument.
39925 (operand_signed_p): Remove.
39926 (maybe_warn_alloc_args_overflow): Call get_size_range.
39927 * calls.h (get_size_range): Declare.
39929 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
39931 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
39932 from TI's devices.csv file as of September 2016.
39933 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
39935 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
39937 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
39938 * doc/invoke.texi: Likewise.
39939 * doc/md.texi: Likewise.
39940 * doc/objc.texi: Likewise.
39942 2017-01-10 Joshua Conner <joshconner@google.com>
39944 * config/arm/fuchsia-elf.h: New file.
39945 * config/fuchsia.h: New file.
39946 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
39947 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
39949 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
39951 2016-01-10 Richard Biener <rguenther@suse.de>
39953 PR tree-optimization/79034
39954 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
39955 Propagate out degenerate PHIs in the joiner.
39957 2017-01-10 Martin Liska <mliska@suse.cz>
39959 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
39960 (sort_congruence_classes_by_decl_uid): Likewise.
39961 (sort_congruence_class_groups_by_decl_uid): Likewise.
39962 (sem_item_optimizer::merge_classes): Sort class, groups in these
39963 classes and members in the groups by DECL_UID of declarations.
39964 This would make merge operations stable.
39966 2017-01-10 Martin Liska <mliska@suse.cz>
39968 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
39969 usage of m_classes_vec.
39970 (sem_item_optimizer::~sem_item_optimizer): Likewise.
39971 (sem_item_optimizer::get_group_by_hash): Likewise.
39972 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
39973 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
39974 (sem_item_optimizer::verify_classes): Likewise.
39975 (sem_item_optimizer::process_cong_reduction): Likewise.
39976 (sem_item_optimizer::dump_cong_classes): Likewise.
39977 (sem_item_optimizer::merge_classes): Likewise.
39978 * ipa-icf.h (congruence_class_hash): Rename from
39979 congruence_class_group_hash. Remove declaration of m_classes_vec.
39981 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
39983 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
39984 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
39985 * config.gcc: Add avx512vpopcntdqintrin.h.
39986 * config/i386/avx512vpopcntdqintrin.h: New.
39987 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
39988 * config/i386/i386-builtin-types.def: Add new types.
39989 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
39990 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
39991 __builtin_ia32_vpopcountq_v8di_mask): New.
39992 * config/i386/i386-c.c (ix86_target_macros_internal): Define
39993 __AVX512VPOPCNTDQ__.
39994 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
39995 (PTA_AVX512VPOPCNTDQ): Define.
39996 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
39997 TARGET_AVX512VPOPCNTDQ_P): Define.
39998 * config/i386/i386.opt: Add mavx512vpopcntdq.
39999 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40000 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40002 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40004 PR middle-end/77484
40005 * predict.def (PRED_CALL): Set to 67.
40007 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40009 * expr.c (store_field): In the bitfield case, if the value comes from
40010 a function call and is of an aggregate type returned in registers, do
40011 not modify the field mode; extract the value in all cases if the mode
40012 is BLKmode and the size is not larger than a word.
40014 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40017 * config/i386/cpuid.h: Fix undefined behavior.
40019 2017-01-04 Jeff Law <law@redhat.com>
40021 PR tree-optimization/79007
40022 PR tree-optimization/67955
40023 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40024 conservative for pt.null when flag_non_call_exceptions is on.
40026 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40028 PR translation/79019
40029 PR translation/79020
40030 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40031 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40032 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40034 * config/avr/avr.opt (maccumulate-args): Likewise.
40035 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40036 * common.opt (freport-bug): Likewise.
40037 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40038 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40039 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40040 translatable string.
40041 * config/i386/i386.c (function_value_32): Likewise.
40042 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40043 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40045 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40046 * common/config/msp430/msp430-common.c (msp430_handle_option):
40048 * symtab.c (symtab_node::verify_base): Likewise.
40049 * opts.c (set_debug_level): Likewise.
40050 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40051 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40052 missing whitespace to translatable strings.
40053 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40054 * config/sh/superh.h: Likewise.
40055 * config/i386/xopintrin.h: Likewise.
40056 * config/i386/znver1.md: Likewise.
40057 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40058 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40059 * double-int.h (struct double_int): Likewise.
40060 * double-int.c (div_and_round_double): Likewise.
40061 * wide-int.cc: Likewise.
40062 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40063 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40064 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40065 (crossjumps_occurred): ... this.
40066 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40069 PR tree-optimization/78899
40070 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40071 returning bool return struct loop *, NULL for failure and the new
40073 (versionable_outer_loop_p): Don't version outer loop if it has
40074 dont_vectorized bit set.
40075 (tree_if_conversion): When versioning outer loop, ensure
40076 tree_if_conversion is performed also on the inner loop of the
40077 non-vectorizable outer loop copy.
40078 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40079 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40080 prevent vectorization of it.
40081 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40082 the outer loop vectorization of the non-scalar version is attempted
40083 before vectorization of the inner loop in scalar version. If
40084 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40085 vectorization of its inner loop.
40086 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40087 has 2 inner loops, rename also on edges from bb whose single pred
40088 is outer_loop->header. Fix typo in function comment.
40090 2017-01-09 Martin Sebor <msebor@redhat.com>
40093 * asan.c (asan_emit_stack_protection): Increase local buffer size
40094 to avoid snprintf truncation warning.
40096 2017-01-09 Andrew Pinski <apinski@cavium.com>
40098 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40099 to reference thunderx2t99 for the tuning structure
40100 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40102 (thunderx2t99_extra_costs): This.
40103 * config/aarch64/aarch64-tune.md: Regenerate.
40104 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40105 (vulcan_addrcost_table): This.
40106 (vulcan_regmove_cost): Rename to ...
40107 (thunderx2t99_regmove_cost): This.
40108 (vulcan_vector_cost): Rename to ...
40109 (thunderx2t99_vector_cost): this.
40110 (vulcan_branch_cost): Rename to ...
40111 (thunderx2t99_branch_cost): This.
40112 (vulcan_tunings): Rename to ...
40113 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40114 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40116 2017-01-09 Martin Jambor <mjambor@suse.cz>
40120 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40121 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40122 (propagate_vr_accross_jump_function): Use the above function for all
40123 value range computations for pass-through jump functions and type
40124 converasion from explicit value range values.
40125 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40126 parameters from TYPE_ARG_TYPES.
40127 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40128 (ipa_write_node_info): Stream type of the actual argument.
40129 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40131 2017-01-09 Martin Liska <mliska@suse.cz>
40134 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40135 (lookup_compiler): Do not show error message with have_E.
40137 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40139 PR tree-optimization/78938
40140 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40141 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40142 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40143 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40146 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40148 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40151 2017-01-09 Richard Biener <rguenther@suse.de>
40153 PR tree-optimization/78997
40154 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40155 name condition properly.
40157 2017-01-09 Richard Biener <rguenther@suse.de>
40160 * dwarf2out.c (is_cxx): New overload with context.
40161 (is_naming_typedef_decl): Use it.
40163 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40165 * invoke.texi (Option Summary): Correct spacing in option lists
40166 and add line breaks to fix over-long lines.
40168 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40170 PR middle-end/17660
40172 * extend.texi (Common Variable Attributes): Add xref to GCC
40173 Internals manual to explain mode attribute keywords.
40175 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40178 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40179 and Preprocessor Options.
40180 (Options for Linking): Document -pthread here....
40181 (RS/6000 and PowerPC Options): ...not here.
40182 (Solaris 2 Options): ...or here.
40183 * doc/cppopts.texi: Document -pthread.
40185 2017-01-08 Martin Sebor <msebor@redhat.com>
40187 PR middle-end/77708
40188 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40189 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40190 New member functions.
40191 (format_directive): Used them.
40193 (pass_sprintf_length::handle_gimple_call): Same.
40194 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40195 to avoid truncation for any argument.
40196 (extract_affine_mul): Same.
40197 * tree.c (get_file_function_name): Same.
40199 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40201 PR middle-end/77484
40202 * predict.def (PRED_INDIR_CALL): Set to 86.
40204 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40206 PR preprocessor/54124
40207 * doc/cppopts.texi: Reformat -d subtable to list the full name
40208 of the options. Add cross-reference to the docs for the general
40209 compiler -d options.
40210 * doc/invoke.texi (Developer Options): Add cross-reference to the
40211 preprocessor-specific -d option documentation.
40213 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40215 PR preprocessor/13498
40216 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40217 redudant material, and reflect new command-line options.
40218 (System Headers): Likewise.
40220 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40222 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40223 -isystem, and -idirafter. Copy-edit.
40224 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40225 default for -ftrack-macro-expansion. Delete obsolete and
40226 badly-formatted implementation details about -fdebug-cpp output.
40227 * doc/cppwarnopts.texi: Copy-edit.
40229 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40232 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40233 that the transition from a max line width >= 1<<10 to narrower
40234 lines works correctly.
40236 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40238 * doc/options.texi (PerFunction): New.
40239 * opt-functions.awk (switch_flags): Map both Optimization and
40240 PerFunction to CL_OPTIMIZATION.
40241 * opth-gen.awk: Test for PerFunction flag along with
40243 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40244 it only when the latter is present. Skip those that don't in
40245 the hash function generator.
40246 * common.opt (fvar-tracking): Mark as PerFunction instead of
40248 (fvar-tracking-assignments): Likewise.
40249 (fvar-tracking-assignments-toggle): Likewise.
40250 (fvar-tracking-uninit): Likewise.
40252 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40254 PR translation/79018
40255 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40258 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40261 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40262 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40263 TARGET_LONG_JUMP_TABLE_OFFSETS.
40264 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40265 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40266 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40267 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40268 * config/m68k/m68k.md (tablejump expander): Likewise.
40269 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40270 TARGET_LONG_JUMP_TABLE_OFFSETS.
40271 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40272 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40274 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40275 David Holsgrove <david.holsgrove@xilinx.com>
40277 * common/config/microblaze/microblaze-common.c
40278 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40279 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40281 * config/microblaze/microblaze.c (microblaze_must_save_register)
40282 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40284 (microblaze_eh_return): New function.
40285 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40286 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40287 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40288 * config/microblaze/microblaze.md (eh_return): New pattern.
40290 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40292 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40293 GCC_DIAGNOSTIC_STRINGIFY): Define.
40295 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40297 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40299 * config/arm/arm.md (<mcrr>): New.
40301 * config/arm/arm.c (arm_arch5te): New.
40302 (arm_option_override): Set arm_arch5te.
40303 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40305 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40306 (arm_mcrr_qualifiers): ... this. New.
40307 (MRRC_QUALIFIERS): Define to...
40308 (arm_mrrc_qualifiers): ... this. New.
40309 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40311 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40312 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40313 (MRRCI, mrrc, MRRC): New.
40314 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40315 VUNSPEC_MRRC2): New.
40317 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40319 * config/arm/arm.md (<mcr>): New.
40321 * config/arm/arm.c (arm_coproc_builtin_available): Add
40322 support for mcr, mrc, mcr2 and mrc2.
40323 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40324 (arm_mcr_qualifiers): ... this. New.
40325 (MRC_QUALIFIERS): Define to ...
40326 (arm_mrc_qualifiers): ... this. New.
40327 (MCR_QUALIFIERS): Define to ...
40328 (arm_mcr_qualifiers): ... this. New.
40329 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40331 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40332 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40333 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40334 VUNSPEC_MRC2): New.
40336 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40338 * config/arm/arm.md (*ldc): New.
40342 * config/arm/arm.c (arm_coproc_builtin_available): Add
40343 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40344 (arm_coproc_ldc_stc_legitimate_address): New.
40345 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40346 'qualifier_const_pointer'.
40347 (LDC_QUALIFIERS): Define to...
40348 (arm_ldc_qualifiers): ... this. New.
40349 (STC_QUALIFIERS): Define to...
40350 (arm_stc_qualifiers): ... this. New.
40351 * config/arm/arm-protos.h
40352 (arm_coproc_ldc_stc_legitimate_address): New.
40353 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40354 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40355 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40356 stc2, stcl, stc2l): New.
40357 * config/arm/constraints.md (Uz): New.
40358 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40359 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40360 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40361 VUNSPEC_STC2L): New.
40363 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40365 * config/arm/arm.md (<cdp>): New.
40366 * config/arm/arm.c (neon_const_bounds): Rename this ...
40367 (arm_const_bounds): ... this.
40368 (arm_coproc_builtin_available): New.
40369 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40370 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40371 (CDP_QUALIFIERS): Define to...
40372 (arm_cdp_qualifiers): ... this. New.
40374 (arm_expand_builtin_args): Add case for 6 arguments.
40375 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40376 (arm_const_bounds): ... this.
40377 (arm_coproc_builtin_available): New.
40378 * config/arm/arm_acle.h (__arm_cdp): New.
40380 * config/arm/arm_acle_builtins.def (cdp): New.
40382 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40383 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40384 'arm_const_bounds'.
40385 * config/arm/types.md (coproc): New.
40386 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40387 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40388 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40389 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40391 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40393 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40394 (UBINOP_QUALIFIERS): New.
40396 (acle_builtin_data): New. Change comment.
40397 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40398 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40399 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40400 arm_acle_builtins.def.
40401 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40402 (arm_init_acle_builtins): New.
40403 (CRC32_BUILTIN): Remove.
40404 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40405 crc32cb, crc32ch and crc32cw.
40406 (arm_init_crc32_builtins): Remove.
40407 (arm_init_builtins): Use arm_init_acle_builtins rather
40408 than arm_init_crc32_builtins.
40409 (arm_expand_acle_builtin): New.
40410 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40411 * config/arm/arm_acle_builtins.def: New.
40413 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40415 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40416 (arm_builtin_datum): ... this.
40417 (arm_init_neon_builtin): Rename to ...
40418 (arm_init_builtin): ... this. Add a new parameters PREFIX
40419 and USE_SIG_IN_NAME.
40420 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40421 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40422 'arm_builtin_datum'.
40423 (arm_init_vfp_builtins): Likewise.
40424 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40425 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40426 (arm_expand_neon_args): Rename to ...
40427 (arm_expand_builtin_args): ... this. Rename builtin_arg
40428 enum values and differentiate between ARG_BUILTIN_MEMORY
40429 and ARG_BUILTIN_NEON_MEMORY.
40430 (arm_expand_neon_builtin_1): Rename to ...
40431 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40432 values, arm_expand_builtin_args and add bool parameter NEON.
40433 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40434 (arm_expand_vfp_builtin): Likewise.
40435 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40437 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40439 PR middle-end/77484
40440 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40441 * predict.c (tree_estimate_probability_bb): Reverse direction of
40442 polymorphic call predictor.
40444 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40446 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40447 (determine_pass_name_match): ...this new function and...
40448 (should_skip_pass_p): ...this new function.
40450 2017-01-06 Nathan Sidwell <nathan@acm.org>
40452 * ipa-visibility.c (function_and_variable_visibility): Reformat
40453 comments and long lines. Remove extrneous if.
40454 * symtab.c (symtab_node::make_decl_local): Fix code format.
40455 (symtab_node::set_section_for_node): Fix comment typo.
40457 2017-01-06 Martin Liska <mliska@suse.cz>
40460 * lra-constraints.c: Rename invariant to lra_invariant.
40461 * predict.c (set_even_probabilities): Initialize e to NULL.
40463 2017-01-05 Martin Sebor <msebor@redhat.com>
40465 PR tree-optimization/78910
40466 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40467 (format_integer): Correct off-by-one error in the handling
40468 of precision with negative numbers in signed conversions..
40470 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40472 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40474 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40476 PR tree-optimization/71016
40477 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40478 factor_out_conditional_conversion. Formatting fix.
40479 (factor_out_conditional_conversion): Add cond_stmt argument.
40480 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40481 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40484 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40486 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40487 read-rtl-function.o, and selftest-rtl.o.
40488 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40489 (selftest::aarch64_test_loading_full_dump): New function.
40490 (selftest::aarch64_run_selftests): New function.
40491 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40492 selftest::aarch64_run_selftests.
40493 * config/i386/i386.c
40494 (selftest::ix86_test_loading_dump_fragment_1): New function.
40495 (selftest::ix86_test_loading_call_insn): New function.
40496 (selftest::ix86_test_loading_full_dump): New function.
40497 (selftest::ix86_test_loading_unspec): New function.
40498 (selftest::ix86_run_selftests): Call the new functions.
40499 * emit-rtl.c (maybe_set_max_label_num): New function.
40500 * emit-rtl.h (maybe_set_max_label_num): New decl.
40501 * function.c (instantiate_decls): Guard call to
40502 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40503 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40505 * gensupport.c (gen_reader::gen_reader): Pass "false"
40506 for new "compact" param of rtx_reader.
40507 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40508 rather than an empty string for NULL strings.
40509 * read-md.c: Potentially include config.h rather than bconfig.h.
40510 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40511 (have_error): New global, copied from errors.c.
40512 (md_reader::read_name): Rename to...
40513 (md_reader::read_name_1): ...this, adding "out_loc" param,
40514 and converting "missing name or number" to returning false, rather
40516 (md_reader::read_name): Reimplement in terms of read_name_1.
40517 (md_reader::read_name_or_nil): New function.
40518 (md_reader::read_string): Handle "(nil)" by returning NULL.
40519 (md_reader::md_reader): Add new param "compact".
40520 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40521 (md_reader::read_file): New method.
40522 * read-md.h (md_reader::md_reader): Add new param "compact".
40523 (md_reader::read_file): New method.
40524 (md_reader::is_compact): New accessor.
40525 (md_reader::read_name): Convert return type from void to file_location.
40526 (md_reader::read_name_or_nil): New decl.
40527 (md_reader::read_name_1): New decl.
40528 (md_reader::m_compact): New field.
40529 (noop_reader::noop_reader): Pass "false" for new "compact" param
40531 (rtx_reader::rtx_reader): Add new "compact" param.
40532 (rtx_reader::read_rtx_operand): Make virtual and convert return
40533 type from void to rtx.
40534 (rtx_reader::read_until): New decl.
40535 (rtx_reader::handle_any_trailing_information): New virtual function.
40536 (rtx_reader::postprocess): New virtual function.
40537 (rtx_reader::finalize_string): New virtual function.
40538 (rtx_reader::m_in_call_function_usage): New field.
40539 (rtx_reader::m_reuse_rtx_by_id): New field.
40540 * read-rtl-function.c: New file.
40541 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40542 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40543 (selftest::verify_three_block_rtl_cfg): New decl.
40544 * read-rtl-function.h: New file.
40545 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40546 For host, include function.h, memmodel.h, and emit-rtl.h.
40547 (one_time_initialization): New function.
40548 (struct compact_insn_name): New struct.
40549 (compact_insn_names): New array.
40550 (find_code): Handle insn codes in compact dumps.
40551 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40552 (bind_subst_iter_and_attr): Likewise.
40553 (add_condition_to_string): Likewise.
40554 (add_condition_to_rtx): Likewise.
40555 (apply_attribute_uses): Likewise.
40556 (add_current_iterators): Likewise.
40557 (apply_iterators): Likewise.
40558 (initialize_iterators): Guard usage of apply_subst_iterator with
40559 #ifdef GENERATOR_FILE.
40560 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40561 (md_reader::read_mapping): Likewise.
40562 (add_define_attr_for_define_subst): Likewise.
40563 (add_define_subst_attr): Likewise.
40564 (read_subst_mapping): Likewise.
40565 (check_code_iterator): Likewise.
40566 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40568 (one_time_initialization): New function.
40569 (rtx_reader::read_until): New method.
40570 (read_flags): New function.
40571 (parse_reg_note_name): New function.
40572 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40573 Handle reuse_rtx ids.
40574 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40575 Add parsing support for RTL dumps, mirroring the special-cases in
40576 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40577 values, and calling handle_any_trailing_information.
40578 (rtx_reader::read_rtx_operand): Convert return type from void
40579 to rtx, returning return_rtx. Handle case 'e'. Call
40580 finalize_string on XSTR and XTMPL fields.
40581 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40582 "(nil)" values were omitted. Call the postprocess vfunc on the
40584 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40585 class ctor. Initialize m_in_call_function_usage. Call
40586 one_time_initialization.
40587 * rtl-tests.c (selftest::test_uncond_jump): Call
40588 set_new_first_and_last_insn.
40589 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40590 * selftest-rtl.c: New file.
40591 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40592 (selftest::get_insn_by_uid): New decl.
40593 * selftest-run-tests.c (selftest::run_tests): Call
40594 read_rtl_function_c_tests.
40595 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40596 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40599 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40601 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40602 operands in a special way. Assert that pos+len <= mode precision.
40604 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40606 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40607 3 argument Alias with unlimited for the negative form.
40608 (fno-vect-cost-model): Removed.
40610 2017-01-05 Martin Liska <mliska@suse.cz>
40612 * hsa-gen.c (gen_hsa_divmod): New function.
40613 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40615 2017-01-05 Martin Liska <mliska@suse.cz>
40618 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40621 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40623 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40624 small constant length operands.
40626 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40628 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40629 between loop iterations.
40631 2017-01-05 Martin Liska <mliska@suse.cz>
40634 * gimplify.c (gimplify_decl_expr): Compare to
40635 asan_poisoned_variables instread of checking flags.
40636 (gimplify_target_expr): Likewise.
40637 (gimplify_expr): Likewise.
40638 (gimplify_function_tree): Conditionally initialize
40639 asan_poisoned_variables.
40641 2017-01-04 Jeff Law <law@redhat.com>
40643 PR tree-optimizatin/78812
40644 * rtl.h (contains_mem_rtx_p): Prototype.
40645 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40646 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40647 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40648 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40649 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40651 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40653 * input.c (assert_char_at_range): Default-initialize actual_range.
40655 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40657 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40658 to match the caller.
40660 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40662 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40663 insns after final jump in test to emit dummy move.
40665 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40667 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40668 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40670 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40672 * multiple_target.c (create_dispatcher_calls): Init e_next.
40673 * tree-ssa-loop-split.c (split_loop): Init border.
40674 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40677 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40682 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40683 (altivec_register_operand): Do not return true if the operand
40684 contains a SUBREG mixing SImode and SFmode.
40685 (vsx_register_operand): Likewise.
40686 (vsx_reg_sfsubreg_ok): New predicate.
40687 (vfloat_operand): Do not return true if the operand contains a
40688 SUBREG mixing SImode and SFmode.
40689 (vint_operand): Likewise.
40690 (vlogical_operand): Likewise.
40691 (gpc_reg_operand): Likewise.
40692 (int_reg_operand): Likewise.
40693 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40694 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40695 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40697 (rs6000_emit_move_si_sf_subreg): New helper function.
40698 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40699 fixup SUBREGs involving SImode and SFmode.
40700 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40701 numbers for the new peephole2 optimization.
40702 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40703 the GLIBC math library that do AND/IOR/XOR operations on single
40704 precision floating point.
40705 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40706 target macros to say whether we need to avoid SUBREGs mixing
40708 (TARGET_ALLOW_SF_SUBREG): Likewise.
40709 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40710 (UNSPEC_SI_FROM_SF): Likewise.
40711 (iorxor): Change spacing.
40712 (and_ior_xor): New iterator for AND, IOR, and XOR.
40713 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40714 (movdi_from_sf_zero_ext): Likewise.
40715 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40716 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40717 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40718 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40719 (fms<mode>4): Likewise.
40720 (fnma<mode>4): Likewise.
40721 (fnms<mode>4): Likewise.
40722 (nfma<mode>4): Likewise.
40723 (nfms<mode>4): Likewise.
40725 2017-01-04 Marek Polacek <polacek@redhat.com>
40728 * doc/invoke.texi: Document -Wpointer-compare.
40730 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40732 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40735 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40736 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40739 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40742 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40743 documentation of the powerpc_popcntb_ok attribute.
40744 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40745 code to issue warning messages if a requested CPU configuration is
40746 not supported by the binary (assembler and loader) toolchain.
40747 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40748 made to define a built-in function that has been disabled.
40749 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40750 made to define a built-in function that has been disabled.
40751 (altivec_init_builtins): Add comment explaining why definition
40752 of the DST built-in functions is not preceded by an assertion
40753 check. Add assertions to prevent ICE if attempts are made to
40754 define an altivec predicate or an abs* built-in function that has
40756 (htm_init_builtins): Add comment explaining why definition of the
40757 htm built-in functions is not preceded by an assertion check.
40759 2017-01-04 Jeff Law <law@redhat.com>
40761 PR tree-optimizatin/67955
40762 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40763 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40764 the points-to solution does not include pt_null. Use DECL_PT_UID
40767 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40769 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40770 Use gen_int_mode instead of gen_lopwart for const_int operands.
40772 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40774 PR tree-optimization/71563
40775 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40776 out of range value - has low bits known to be zero.
40778 2017-01-04 Alan Modra <amodra@gmail.com>
40780 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40781 * configure: Regenerate.
40782 * config.in: Regenerate.
40784 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40787 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40788 a substring of the message, but strcmp with the whole message. Ifdef
40789 ENABLE_NLS, translate the message first using dgettext.
40791 2017-01-03 Jeff Law <law@redhat.com>
40793 PR tree-optimizatin/78856
40794 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40795 (mark_threaded_blocks): Remove code to truncate thread paths that
40796 cross multiple loop headers. Instead invalidate the cached loop
40797 iteration information and handle case of a thread path walking
40798 into an irreducible region.
40800 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40803 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40804 assertions. Add support for doing the signbit if the IEEE 128-bit
40805 floating point value is in a GPR.
40806 * config/rs6000/rs6000.md (Fsignbit): Delete.
40807 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40808 Update the length attribute if the value is in a GPR.
40809 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40810 the sign or zero extension instruction, since the value is always 0/1.
40811 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40814 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40815 extracting SImode to a GPR register so that we can generate a
40816 store, limit the vector to be in a traditional Altivec register
40817 for the vextuwrx instruction.
40819 2017-01-03 Ian Lance Taylor <iant@google.com>
40821 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40823 2017-01-03 Martin Sebor <msebor@redhat.com>
40825 PR tree-optimization/78696
40826 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40827 precision. Use MPFR for %f for greater fidelity. Correct handling
40829 (pass_sprintf_length::compute_format_length): Set width and precision
40830 specified by asrerisk to void_node for vararg functions.
40831 (try_substitute_return_value): Adjust dump output.
40833 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40835 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40837 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40839 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40840 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40841 -mlra/-mno-lra was passed to the compiler.
40843 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40845 PR rtl-optimization/65618
40846 * emit-rtl.c (try_split): Move initialization of "before" and
40847 "after" to just before the call to emit_insn_after_setloc.
40849 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40851 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40853 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40855 * dwarf2out.c (gen_enumeration_type_die): When
40856 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40858 2017-01-03 Jakub Jelinek <jakub@redhat.com>
40860 PR tree-optimization/78965
40861 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
40862 Change first argument from const call_info & to call_info &. For %n
40863 set info.nowrite to false.
40865 PR middle-end/78901
40866 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
40867 possibly throwing calls.
40869 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
40870 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
40871 and fns handling, rather than in a separate case SSA_NAME.
40873 2017-01-02 Jeff Law <law@redhat.com>
40875 * config/darwin-driver.c (darwin_driver_init): Const-correctness
40876 fixes for first_period and second_period variables.
40878 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
40881 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
40882 (*insvqi_1): New insn pattern.
40883 (*insvqi_1_mem_rex64): Ditto.
40884 (*insvqi_2): Ditto.
40885 (*insvqi_3): Rename from *insvqi.
40887 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
40889 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
40891 * doc/cfg.texi (Edges): Remove reference to Java.
40892 (Maintaining the CFG): Ditto.
40894 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40896 PR middle-end/77674
40897 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
40898 transparent aliases.
40900 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40902 PR middle-end/77484
40903 * predict.def (PRED_CALL): Update hitrate.
40904 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
40905 * predict.c (tree_estimate_probability_bb): Split CALL predictor
40906 into direct/indirect/polymorphic variants.
40908 2017-01-01 Jakub Jelinek <jakub@redhat.com>
40910 Update copyright years.
40912 * gcc.c (process_command): Update copyright notice dates.
40913 * gcov-dump.c (print_version): Ditto.
40914 * gcov.c (print_version): Ditto.
40915 * gcov-tool.c (print_version): Ditto.
40916 * gengtype.c (create_file): Ditto.
40917 * doc/cpp.texi: Bump @copying's copyright year.
40918 * doc/cppinternals.texi: Ditto.
40919 * doc/gcc.texi: Ditto.
40920 * doc/gccint.texi: Ditto.
40921 * doc/gcov.texi: Ditto.
40922 * doc/install.texi: Ditto.
40923 * doc/invoke.texi: Ditto.
40925 Copyright (C) 2017 Free Software Foundation, Inc.
40927 Copying and distribution of this file, with or without modification,
40928 are permitted in any medium without royalty provided the copyright
40929 notice and this notice are preserved.