1 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
5 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
8 * predict.c (maybe_hot_frequency_p): Inline to ...
9 (maybe_hot_count_p): ... here; rewrite to counts.
10 (counts_to_freqs): Rename to ...
11 (update_max_bb_count): ... this one.
12 (expensive_function_p): Use counts.
13 (estimate_bb_frequencies): Update.
14 (rebuild_frequencies): Update.
15 * predict.h (counts_to_freqs): Rename to ...
16 (update_max_bb_count): ... this one.
17 * profile.c (compute_branch_probabilities): Add debug info
18 * tree-inline.c (expand_call_inline): Update debug info.
19 (optimize_inline_calls): Use update_max_bb_count..
20 (tree_function_versioning): Use update_max_bb_count..
21 * value-prof.c (gimple_value_profile_transformations):
22 Do not use update_max_bb_count.
24 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
26 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
27 always use frequencies.
29 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
31 * bb-reorder.c: Remove frequencies from comments.
32 (better_edge_p): Use profile counts.
33 (find_traces): Dump profile counts.
34 (rotate_loop): Use profile counts.
35 (find_traces_1_round): Likewise.
36 (connect_better_edge_p): Use counts instead of probabilities for
38 (copy_bb_p): Drop early check for non-0 frequency.
39 (sanitize_hot_paths): Update comments.
41 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
43 * ipa-split.c (struct split_point): Add count.
44 (consider_split): Do not compute incoming frequency; compute incoming
45 count and store it to split_point.
46 (split_function): Set count of the call to split part correctly.
48 2017-11-13 Carl Love <cel@us.ibm.com>
50 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
52 2017-11-13 Tom Tromey <tom@tromey.com>
54 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
56 2017-11-13 Carl Love <cel@us.ibm.com>
58 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
59 Add support for builtins:
60 unsigned int vec_first_{,miss}_match_{,or_eos}index,
61 vector {un,}signed {char,int,short},
62 vector {un,}signed {char,int,short}) arguments.
63 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
64 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
65 Add BU_P9V_AV_2 expansions for the builtins.
66 * config/rs6000/altivec.h (vec_first_match_index,
67 vec_first_mismatch_index, vec_first_match_or_eos_index,
68 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
69 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
70 new extern declaration.
71 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
72 * config/rs6000/vsx.md (first_match_index_<mode>,
73 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
74 first_mismatch_or_eos_index_<mode>): Add define expand.
75 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
76 * doc/extend.texi: Update the built-in documenation file for the new
79 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
81 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
82 into the min/max operations for _Float<N> and _Float<N>X types.
84 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
87 * dwarf2out.c (do_eh_frame): New static variable.
88 (dwarf2out_begin_prologue): Set it.
89 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
91 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
93 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
95 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
97 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
100 * debug.h (dwarf2out_do_eh_frame): Declare.
101 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
102 (dwarf2out_do_frame): Use it.
103 (dwarf2out_do_cfi_asm): Likewise.
104 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
105 (dwarf2out_assembly_start): Likewise.
106 (dwarf2out_begin_prologue): Fix comment.
107 * toplev.c (compile_file): Always call dwarf2out_frame_finish
108 if the target needs either debug or unwind DWARF2 info.
109 * lto-opts.c (lto_write_options): Do not save -fexceptions,
110 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
111 -ftrapping-math, -ftrapv and -fwrapv.
113 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
115 * cgraph.c (cgraph_edge::sreal_frequency): New function.
116 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
117 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
118 (estimate_edge_size_and_time): Likewise.
119 (ipa_merge_fn_summary_after_inlining): Likewise.
120 * ipa-inline.c (cgraph_freq_base_rec): Remove.
121 (compute_uninlined_call_time): Use sreal_frequency.
122 (compute_inlined_call_time): Likewise.
123 (ipa_inline): Do not initialize cgraph_freq_base_rec.
124 * profile-count.c: Include sreal.h.
125 (profile_count::to_sreal_scale): New.
126 * profile-count.h: Forward declare sreal.
127 (profile_count::to_sreal_scale): Declare.
129 2017-11-13 Nathan Sidwell <nathan@acm.org>
131 * diagnostic.c (maybe_line_and_column): New.
132 (diagnostic_get_location_text): Use it.
133 (diagnostic_report_current_module): Likewise.
134 (test_diagnostic_get_location_text): Add tests.
136 2017-11-13 Luis Machado <luis.machado@linaro.org>
138 * doc/md.texi (Specifying processor pipeline description): Fix
139 incorrect latency for the div instruction example.
141 2017-11-13 Jakub Jelinek <jakub@redhat.com>
143 PR tree-optimization/78821
144 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
145 that bit_not_p is the same.
146 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
147 (split_group): Count precisely bit_not_p bits in each statement.
148 (invert_op): New function.
149 (imm_store_chain_info::output_merged_store): Use invert_op to
150 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
151 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
153 2017-11-13 Martin Liska <mliska@suse.cz>
155 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
156 (struct source_info): Likewise.
157 (add_branch_counts): Likewise.
158 (add_line_counts): Likewise.
159 (function_summary): Likewise.
160 (output_intermediate_line): Likewise.
161 (generate_results): Likewise.
163 2017-11-13 Martin Liska <mliska@suse.cz>
165 * gcov.c (struct block_info): Remove typedef for block_t.
166 (struct line_info): Likewise.
167 (line_info::has_block): Likewise.
168 (EXIT_BLOCK): Likewise.
171 (get_cycles_count): Likewise.
172 (process_file): Likewise.
173 (read_graph_file): Likewise.
174 (solve_flow_graph): Likewise.
175 (find_exception_blocks): Likewise.
176 (add_line_counts): Likewise.
177 (accumulate_line_info): Likewise.
178 (output_line_details): Likewise.
180 2017-11-13 Martin Liska <mliska@suse.cz>
182 * gcov.c (struct arc_info): Remove typedef for arc_t.
183 (struct line_info): Likewise.
184 (add_branch_counts): Likewise.
185 (output_branch_count): Likewise.
186 (function_info::~function_info): Likewise.
188 (output_intermediate_line): Likewise.
189 (read_graph_file): Likewise.
190 (solve_flow_graph): Likewise.
191 (find_exception_blocks): Likewise.
192 (add_line_counts): Likewise.
193 (accumulate_line_info): Likewise.
194 (output_line_details): Likewise.
195 (output_function_details): Likewise.
197 2017-11-13 Martin Liska <mliska@suse.cz>
199 * gcov.c (struct function_info): Remove typedef for function_t.
200 (struct source_info): Likewise.
201 (source_info::get_functions_at_location): Likewise.
202 (solve_flow_graph): Likewise.
203 (find_exception_blocks): Likewise.
204 (add_line_counts): Likewise.
205 (output_intermediate_file): Likewise.
206 (process_file): Likewise.
207 (generate_results): Likewise.
208 (release_structures): Likewise.
209 (read_graph_file): Likewise.
210 (read_count_file): Likewise.
211 (accumulate_line_counts): Likewise.
212 (output_lines): Likewise.
214 2017-11-13 Martin Liska <mliska@suse.cz>
216 * gcov.c (function_info::function_info): Remove num_counts
217 and add vector<gcov_type>.
218 (function_info::~function_info): Use the vector.
219 (process_file): Likewise.
220 (read_graph_file): Likewise.
221 (read_count_file): Likewise.
222 (solve_flow_graph): Likewise.
224 2017-11-13 Martin Liska <mliska@suse.cz>
226 * gcov.c (function_info::is_artificial): New function.
227 (process_file): Erase all artificial early.
228 (generate_results): Skip as all artificial are already
231 2017-11-13 Martin Liska <mliska@suse.cz>
233 * gcov.c (read_graph_file): Store to global vector of functions.
234 (read_count_file): Iterate the vector.
235 (process_file): Likewise.
236 (generate_results): Likewise.
237 (release_structures): Likewise.
239 2017-11-13 Jakub Jelinek <jakub@redhat.com>
241 PR tree-optimization/82954
242 * gimple-ssa-store-merging.c
243 (imm_store_chain_info::coalesce_immediate_stores): If
244 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
246 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
248 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
249 Upddate call to ENDIAN_LANE_N.
250 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
251 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
252 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
253 and use aarch64_endian_lane_rtx.
254 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
256 2017-11-12 Tom de Vries <tom@codesourcery.com>
258 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
260 2017-11-12 Tom de Vries <tom@codesourcery.com>
262 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
264 2017-11-12 Tom de Vries <tom@codesourcery.com>
266 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
268 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
269 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
270 * defaults.h (ASM_OUTPUT_LABELREF): Same.
272 2017-11-11 Martin Sebor <msebor@redhat.com>
275 * doc/extend.texi (attribute nonstring): Remove spurious argument.
278 * prefic.c (translate_name): Replace strncpy with memcpy to
279 avoid -Wstringop-truncation.
281 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
283 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
285 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
287 * predict.c (maybe_hot_frequency_p): Do not use cfun.
289 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
291 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
294 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
296 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
297 enable generating XXBRH if the value is in a vector register.
298 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
299 value is in a vector register.
300 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
301 register to register bswap64's instead of doing the GPR sequence
302 used on previous machines.
303 (bswapdi2_xxbrd): New insn.
304 (bswapdi2_reg): Disallow on ISA 3.0.
305 (register to register bswap64 splitter): Do not split the insn on
306 ISA 3.0 systems that use XXBRD.
308 2017-11-10 Martin Sebor <msebor@redhat.com>
311 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
313 (find_subframework_file): Same.
315 2017-11-10 Jeff Law <law@redhat.com>
317 * vr-values.h (VR_INITIALIZER): Move #define here.
318 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
319 * Makefile.in (OBJS): Add tree-evrp.o
320 * tree-vrp.h (assert_info): Move structure definition here.
321 (set_value_range_to_varying): Prototype.
322 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
323 (infer_value_range, register_edge_assert_for): Likewise.
324 (stmt_interesting_for_vrp): Likewise.
325 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
326 (set_value_range_to_varying): No longer static.
327 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
328 (infer_value_range, register_edge_assert_for): Likewise.
330 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
332 * auto-profile.c (afdo_indirect_call): Drop frequency.
333 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
334 (cgraph_node::create_edge): Drop frequency argument.
335 (cgraph_node::create_indirect_edge): Drop frequency argument.
336 (cgraph_edge::make_speculative): Drop frequency arguments.
337 (cgraph_edge::resolve_speculation): Do not update frequencies
338 (cgraph_edge::dump_edge_flags): Do not dump frequency.
339 (cgraph_node::dump): Check consistency in IPA mode.
340 (cgraph_edge::maybe_hot_p): Use IPA counter.
341 (cgraph_edge::verify_count_and_frequency): Rename to ...
342 (cgraph_edge::verify_count): ... this one; drop frequency checking.
343 (cgraph_node::verify_node): Update.
344 * cgraph.h (struct cgraph_edge): Drop frequency.
345 (cgraph_edge::frequency): New function.
346 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
348 (cgraph_edge::rebuild_edges): Likewise.
349 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
350 (duplicate_thunk_for_node): Do not pass frequency.
351 (cgraph_node::create_clone): Scale only counts.
352 (cgraph_node::create_virtual_clone): Do not pass frequency.
353 (cgraph_node::create_edge_including_clones): Do not pass frequency.
354 (cgraph_node::create_version_clone): Do not pass frequency.
355 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
356 (cgraph_node::expand_thunk): Do not pass frequency.
357 (cgraph_node::create_wrapper): Do not pass frequency.
358 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
360 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
361 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
362 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
363 (gather_caller_stats): Use frequency function.
364 (ipcp_cloning_candidate_p): Use frequency function.
365 (ipcp_propagate_stage): Use frequency function.
366 (get_info_about_necessary_edges): Use frequency function.
367 (update_profiling_info): Update only IPA profile.
368 (update_specialized_profile): Use frequency functoin.
369 (perhaps_add_new_callers): Update only IPA profile.
370 * ipa-devirt.c (ipa_devirt): Use IPA profile.
371 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
372 (dump_ipa_call_summary): Use frequency function.
373 (estimate_edge_size_and_time): Use frequency function.
374 (ipa_merge_fn_summary_after_inlining): Use frequency function.
375 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
376 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
377 (update_noncloned_counts): ... ths one; scale counts only.
378 (clone_inlined_nodes): Do not scale frequency.
379 (inline_call): Do not pass frequency.
380 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
381 (compute_inlined_call_time): Use IPA profile.
382 (want_inline_small_function_p): Use IPA profile.
383 (want_inline_self_recursive_call_p): Use IPA profile.
384 (edge_badness): Use IPA profile.
385 (lookup_recursive_calls): Use IPA profile.
386 (recursive_inlining): Do not pass frequency.
387 (resolve_noninline_speculation): Do not update frequency.
388 (inline_small_functions): Collect max of IPA profile.
389 (dump_overall_stats): Dump IPA porfile.
390 (dump_inline_stats): Dump IPA porfile.
391 (ipa_inline): Collect IPA stats.
392 * ipa-inline.h (clone_inlined_nodes): Update prototype.
393 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
394 (ipa_propagate_frequency): Use frequency function.
395 (ipa_profile): Cleanup.
396 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
397 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
398 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
399 (input_node): Do not stream frequency.
400 (input_edge): Do not stream frequency.
401 (merge_profile_summaries): Scale only IPA profiles.
402 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
403 * predict.c (drop_profile): Do not recompute frequency.
404 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
405 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
406 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
407 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
408 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
409 * tree-inline.c (copy_bb): Do not scale frequency.
410 (expand_call_inline): Do not scale frequency.
411 (tree_function_versioning): Do not scale frequency.
412 * ubsan.c (ubsan_create_edge): Do not pass frequency.
414 2017-11-10 Julia Koval <julia.koval@intel.com>
416 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
417 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
418 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
419 _mm256_maskz_gf2p8affine_epi64_epi8,
420 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
421 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
422 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
423 __builtin_ia32_vgf2p8affineqb_v32qi,
424 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
425 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
427 2017-11-10 Tamar Christina <tamar.christina@arm.com>
431 (arm_option_override): Refactor.
432 (arm_option_reconfigure_globals): New.
433 (arm_options_perform_arch_sanity_checks): New.
434 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
435 (arm_options_perform_arch_sanity_checks): Likewise
437 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
439 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
441 * rs6000/rs6000.md ("type" attr): Add qmul.
442 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
443 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
444 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
446 2017-11-10 Martin Sebor <msebor@redhat.com>
449 * builtins.c (compute_objsize): Handle arrays that
450 compute_builtin_object_size likes to fail for. Make extern.
451 * builtins.h (compute_objsize): Declare.
452 (check_strncpy_sizes): New function.
453 (expand_builtin_strncpy): Call check_strncpy_sizes.
454 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
455 -Wstringop-truncation.
456 (gimple_fold_builtin_strncat): Same.
457 * gimple.c (gimple_build_call_from_tree): Set call location.
458 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
459 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
460 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
461 (handle_builtin_strlen): Use strlen_to_stridx.
462 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
464 Use strlen_to_stridx.
465 (pass_strlen::execute): Release strlen_to_stridx.
466 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
467 (-Wstringop-truncation): Document new option.
469 2017-11-10 Martin Liska <mliska@suse.cz>
471 PR gcov-profile/82702
472 * gcov.c (main): Handle intermediate files in a different
474 (get_gcov_intermediate_filename): New function.
475 (output_gcov_file): Remove support of intermediate files.
476 (generate_results): Allocate intermediate file.
477 (release_structures): Clean-up properly fn_end.
478 (output_intermediate_file): Start iterating with line 1.
480 2017-11-10 Jakub Jelinek <jakub@redhat.com>
482 PR tree-optimization/82929
483 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
484 ops_swapped_p non-static data member.
485 (store_immediate_info::store_immediate_info): Clear it.
486 (imm_store_chain_info::coalesce_immediate_stores): If swapping
487 ops set ops_swapped_p.
488 (count_multiple_uses): Handle ops_swapped_p.
490 2017-11-10 Martin Liska <mliska@suse.cz>
492 * coverage.c (coverage_init): Stream information about
493 support of has_unexecuted_blocks.
494 * doc/gcov.texi: Document that.
495 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
496 * gcov.c (read_graph_file): Likewise.
497 (output_line_beginning): Fix a small issue with
500 2017-11-10 Bin Cheng <bin.cheng@arm.com>
502 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
503 reference of trivial component.
505 2017-11-10 Jakub Jelinek <jakub@redhat.com>
508 * gimple-ssa-store-merging.c
509 (pass_store_merging::terminate_all_aliasing_chains): For
510 gimple_store_p stmts also call refs_output_dependent_p.
512 PR rtl-optimization/82913
513 * compare-elim.c (try_merge_compare): Punt if def_insn is not
516 2017-11-09 Jeff Law <law@redhat.com>
518 * vr-values.h: New file with vr_values class.
519 * tree-vrp.c: Include vr-values.h
520 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
521 data objects into the vr_values class.
522 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
523 (get_value_range): Make it a member function within vr_values class.
524 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
525 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
526 (op_with_constant_singleton_value_range): Likewise.
527 (extract_range_for_var_from_comparison_expr): Likewise.
528 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
529 (extract_range_from_binary_expr): Likewise.
530 (extract_range_from_unary_expr): Likewise.
531 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
532 (check_for_binary_op_overflow, extract_range_basic): Likewise.
533 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
534 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
535 (compare_name_with_value, compare_names): Likewise.
536 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
537 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
538 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
539 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
540 (extract_range_from_phi_node): Likewise.
541 (simplify_truth_ops_using_ranges): Likewise.
542 (simplify_div_or_mod_using_ranges): Likewise.
543 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
544 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
545 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
546 (simplify_float_conversion_using_ranges): Likewise.
547 (simplify_internal_call_using_ranges): Likewise.
548 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
549 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
550 poitner via x_vr_values for calls into gimple folder.
551 (vrp_initialize_lattice): Make this the vr_values ctor.
552 (vrp_free_lattice): Make this the vr_values dtor.
553 (set_vr_value): New function.
554 (class vrp_prop): Add vr_values data member. Add various member
555 functions as well as member functions that delegate to vr_values.
556 (check_array_ref): Make a member function within vrp_prop class.
557 (search_for_addr_array, vrp_initialize): Likewise.
558 (vrp_finalize): Likewise. Revamp to avoid direct access to
559 vr_value, values_propagated, etc.
560 (check_array_bounds): Extract vrp_prop class instance pointer from
561 walk info structure. Use it to call member functions.
562 (check_all_array_refs): Make a member function within vrp_prop class.
563 Smuggle class instance pointer via walk info structure.
564 (x_vr_values): New local static.
565 (vrp_valueize): Use x_vr_values to get class instance.
566 (vr_valueize_1): Likewise.
567 (class vrp_folder): Add vr_values data member. Add various member
568 functions as well as member functions that delegate to vr_values.
569 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
570 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
571 class instance from vr_values. Use it to call member functions.
572 (vrp_dom_walker): Add vr_values data member.
573 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
574 instance via x_vr_values.
575 (identify_jump_threads): Accept vr_values as argument. Store
576 it into the walker structure.
577 (evrp_dom_walker): Add vr_values class data member. Add various
579 (evrp_dom_walker::try_find_new_range): Use vr_values data
580 member to access the memory allocator.
581 (evrp_dom_walker::before_dom_children): Store vr_values class
582 instance into the vrp_folder class.
583 (evrp_dom_walker::push_value_range): Rework to avoid direct
584 access to num_vr_values and vr_value.
585 (evrp_dom_walker::pop_value_range): Likewise.
586 (execute_early_vrp): Remove call to vrp_initialize_lattice.
587 Use vr_values to get to dump_all_value_ranges member function.
588 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
589 and simplify_cond_using_ranges_2 via vrp_prop class instance.
590 Pass vr_values class instance down to identify_jump_threads.
591 Remove call to vrp_free_lattice.
592 (debug_all_value_ranges): Remove.
594 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
595 (vrp_folder): Likewise.
597 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
598 Get it from the existing bitmap instead.
599 (vrp_intersect_ranges_1): Likewise.
601 2017-11-09 Jakub Jelinek <jakub@redhat.com>
603 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
605 (store_immediate_info::store_immediate_info): Add bitnotp argument,
607 (imm_store_chain_info::coalesce_immediate_stores): Break group
608 if bit_not_p is different.
609 (count_multiple_uses, split_group,
610 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
611 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
612 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
613 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
614 to store_immediate_info ctor.
616 2017-11-09 Jim Wilson <jimw@sifive.com>
618 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
619 (scan_prog_file): Likewise.
621 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
623 * bb-reorder.c (max_entry_frequency): Remove.
624 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
625 connect_traces, push_to_next_round_p): Remove prototypes.
626 (find_traces_1_round): Use counts only.
627 (push_to_next_round_p): Likewise.
628 (find_traces): Likewise.
629 (rotate_loop): Likewise.
630 (find_traces_1_round): Likewise.
631 (connect_traces): Likewise.
632 (edge_order): Likewise.
634 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
636 * config/arm/arm.c (output_return_instruction): Add comments to
637 indicate requirement for cmse_nonsecure_entry return to account
638 for the size of clearing instruction output here.
639 (thumb_exit): Likewise.
640 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
641 return in hardfloat mode.
643 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
645 * config/rs6000/rs6000.c (machine_function): Add a bool,
646 "toc_is_wrapped_separately".
647 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
648 if it wasn't explicitly set or unset, we are optimizing for speed, and
649 doing separate shrink-wrapping.
650 (rs6000_get_separate_components): Enable the TOC component if
651 saving the TOC register in the prologue.
652 (rs6000_components_for_bb): Handle the TOC component.
653 (rs6000_emit_prologue_components): Store the TOC register where needed.
654 (rs6000_set_handled_components): Mark TOC as handled, if handled.
655 (rs6000_emit_prologue): Don't save the TOC if that is already done.
657 2017-11-09 Martin Jambor <mjambor@suse.cz>
659 * ipa-param-manipulation.c: New file.
660 * ipa-param-manipulation.h: Likewise.
661 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
662 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
663 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
664 (ipa_parm_adjustment): Likewise.
665 (ipa_parm_adjustment_vec): Likewise.
666 (ipa_get_vector_of_formal_parms): Moved declaration to
667 ipa-param-manipulation.h.
668 (ipa_get_vector_of_formal_parm_types): Likewise.
669 (ipa_modify_formal_parameters): Likewise.
670 (ipa_modify_call_arguments): Likewise.
671 (ipa_combine_adjustments): Likewise.
672 (ipa_dump_param_adjustments): Likewise.
673 (ipa_modify_expr): Likewise.
674 (ipa_get_adjustment_candidate): Likewise.
675 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
676 ipa-param-manipulation.c.
677 (ipa_get_vector_of_formal_parm_types): Likewise.
678 (ipa_modify_formal_parameters): Likewise.
679 (ipa_modify_call_arguments): Likewise.
680 (ipa_modify_expr): Likewise.
681 (get_ssa_base_param): Likewise.
682 (ipa_get_adjustment_candidate): Likewise.
683 (index_in_adjustments_multiple_times_p): Likewise.
684 (ipa_combine_adjustments): Likewise.
685 (ipa_dump_param_adjustments): Likewise.
686 * tree-sra.c: Also include ipa-param-manipulation.h
687 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
690 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
691 Alan Hayward <alan.hayward@arm.com>
692 David Sherwood <david.sherwood@arm.com>
694 * doc/sourcebuild.texi (vect_masked_store): Document.
696 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
697 Alan Hayward <alan.hayward@arm.com>
698 David Sherwood <david.sherwood@arm.com>
700 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
702 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
703 Alan Hayward <alan.hayward@arm.com>
704 David Sherwood <david.sherwood@arm.com>
706 * doc/sourcebuild.texi (vect_variable_length): Document.
708 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
709 Alan Hayward <alan.hayward@arm.com>
710 David Sherwood <david.sherwood@arm.com>
712 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
714 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
715 Alan Hayward <alan.hayward@arm.com>
716 David Sherwood <david.sherwood@arm.com>
718 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
720 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
721 Alan Hayward <alan.hayward@arm.com>
722 David Sherwood <david.sherwood@arm.com>
724 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
725 previously undocumented selectors.
726 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
728 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
730 * doc/rtl.texi (const_vector): Say that elements can be
732 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
733 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
734 (gen_vec_duplicate): Use it instead of CONSTANT_P.
735 * optabs.c (expand_vector_broadcast): Likewise.
737 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
738 Alan Hayward <alan.hayward@arm.com>
739 David Sherwood <david.sherwood@arm.com>
741 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
742 scaled index even if the unscaled address was invalid.
743 Don't increase the complexity of using a scale in that case.
745 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
746 Alan Hayward <alan.hayward@arm.com>
747 David Sherwood <david.sherwood@arm.com>
749 * doc/rtl.texi: Rewrite the subreg rules so that they partition
750 the inner register into REGMODE_NATURAL_SIZE bytes rather than
751 UNITS_PER_WORD bytes.
752 * emit-rtl.c (validate_subreg): Divide subregs into blocks
753 based on REGMODE_NATURAL_SIZE of the inner mode.
754 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
755 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
756 * expmed.c (lowpart_bit_field_p): Divide the value up into
757 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
758 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
759 whether something is likely to occupy more than one register.
761 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
764 * ipa-inline-transform.c (update_noncloned_frequencies): Use
765 profile_count::adjust_for_ipa_scaling.
766 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
767 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
769 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
771 2017-11-09 Jakub Jelinek <jakub@redhat.com>
773 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
774 (split_group): Add total_orig and total_new arguments, estimate the
775 number of statements related to the store group without store merging
776 and with store merging.
777 (imm_store_chain_info::output_merged_store): Adjust split_group
778 callers, punt if estimated number of statements with store merging
779 is not smaller than estimated number of statements without it.
781 (handled_load): Remove has_single_use checks.
782 (pass_store_merging::process_store): Likewise.
784 2017-11-09 Richard Biener <rguenther@suse.de>
786 PR tree-optimization/82902
787 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
789 2017-11-09 Martin Liska <mliska@suse.cz>
792 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
795 2017-11-09 Martin Liska <mliska@suse.cz>
797 PR tree-optimization/82669
798 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
800 2017-11-09 Martin Liska <mliska@suse.cz>
802 PR gcov-profile/48463
803 * coverage.c (coverage_begin_function): Output also end locus
804 of a function and information whether the function is
806 * gcov-dump.c (tag_function): Parse and print the information.
807 * gcov.c (INCLUDE_MAP): Add include.
808 (INCLUDE_SET): Likewise.
809 (struct line_info): Move earlier in the source file because
810 of vector<line_info> in function_info structure.
811 (line_info::line_info): Likewise.
812 (line_info::has_block): Likewise.
813 (struct source_info): Add new member index.
814 (source_info::get_functions_at_location): New function.
815 (function_info::group_line_p): New function.
816 (output_intermediate_line): New function.
817 (output_intermediate_file): Use the mentioned function.
818 (struct function_start): New.
819 (struct function_start_pair_hash): Likewise.
820 (process_file): Add code that identifies group functions.
821 Assign lines either to global or function scope.
822 (generate_results): Skip artificial functions.
823 (find_source): Assign index for each source file.
824 (read_graph_file): Read new flag artificial and end_line.
825 (add_line_counts): Assign it either to global of function scope.
826 (accumulate_line_counts): Isolate core of the function to
827 accumulate_line_info and call it for both function and global
829 (accumulate_line_info): New function.
830 (output_line_beginning): Fix GNU coding style.
831 (print_source_line): New function.
832 (output_line_details): Likewise.
833 (output_function_details): Likewise.
834 (output_lines): Iterate both source (global) scope and function
836 (struct function_line_start_cmp): New class.
837 * doc/gcov.texi: Reflect changes in documentation.
839 2017-11-09 Jakub Jelinek <jakub@redhat.com>
842 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
843 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
844 and similarly for not instead of neg.
846 2017-11-08 Andi Kleen <ak@linux.intel.com>
848 * config/i386/i386.opt: Add -mforce-indirect-call.
849 * config/i386/predicates.md: Check for flag_force_indirect_call.
850 * doc/invoke.texi: Document -mforce-indirect-call
852 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
854 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
856 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
857 config/riscv/riscv.c (predict.h): New include.
858 (riscv_slow_unaligned_access_p): No longer static.
859 (riscv_block_move_straight): Add require.
860 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
862 2017-11-08 Jakub Jelinek <jakub@redhat.com>
865 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
866 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
867 nonimmediate_operand predicate for operand 1 instead of
870 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
872 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
874 * config/aarch64/constraints.md (Uml): New constraint.
875 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
878 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
880 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
881 of two vec_duplicates into a vec_concat.
883 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
885 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
886 Simplify vec_merge of vec_duplicate and vec_concat.
887 * config/aarch64/constraints.md (Utq): New constraint.
888 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
891 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
893 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
894 Simplify vec_merge of vec_duplicate and const_vector.
895 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
897 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
898 mode iterator. Update predicate on operand 1 to
899 handle non-const_vec constants. Delete constraints.
900 (*aarch64_combinez_be<mode>): Likewise for operand 2.
902 2017-11-08 Jakub Jelinek <jakub@redhat.com>
904 PR tree-optimization/78821
905 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
907 (store_operand_info::store_operand_info): Initialize it to false.
908 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
909 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
910 store in the group, and if chain_info is non-NULL, to ignore altogether
912 (compatible_load_p): Fail if bit_not_p does not match.
913 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
914 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
915 (pass_store_merging::process_store): Adjust
916 terminate_all_aliasing_chains calls to pass NULL in all current spots,
917 call terminate_all_aliasing_chains newly when adding a store into
918 a chain with non-NULL chain_info.
920 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
922 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
924 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
926 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
928 (aarch64_layout_frame): Initialise emit_frame_chain.
929 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
930 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
932 2017-11-08 Martin Liska <mliska@suse.cz>
934 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
935 of gimple_call_internal_p.
937 2017-11-07 Tom de Vries <tom@codesourcery.com>
939 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
941 2017-11-07 Tom de Vries <tom@codesourcery.com>
943 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
946 2017-11-08 Martin Liska <mliska@suse.cz>
949 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
951 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
953 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
955 2017-11-07 Jakub Jelinek <jakub@redhat.com>
958 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
959 (*cmp<mode>_ccz_1): New insn with $k alternative.
962 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
963 RTX_COMM_COMPARE as commutative as well.
964 (ix86_binary_operator_ok): Formatting fix.
965 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
966 *<code><mode>3<mask_name><round_saeonly_name>,
967 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
968 *<s>mul<mode>3_highpart<mask_name>,
969 *vec_widen_umult_even_v16si<mask_name>,
970 *vec_widen_umult_even_v8si<mask_name>,
971 *vec_widen_umult_even_v4si<mask_name>,
972 *vec_widen_smult_even_v16si<mask_name>,
973 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
974 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
975 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
976 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
977 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
978 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
979 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
980 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
981 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
982 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
983 ix86_binary_operator_ok. Formatting fixes.
984 (*<plusminus_insn><mode>3<mask_name><round_name>,
985 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
988 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
990 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
991 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
992 eq and ne if TARGET_ISEL.
993 (cmp): New code_iterator.
994 (UNS, UNSU_, UNSIK): New code_attrs.
995 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
996 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
998 ("eq<mode>3"): ... this.
999 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1001 ("ne<mode>3"): ... this.
1003 2017-11-07 Julia Koval <julia.koval@intel.com>
1006 * common/config/i386/i386-common.c
1007 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1008 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1009 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1010 * config/i386/i386.opt: Ditto.
1011 * config/i386/i386.c (ix86_target_string): Ditto.
1012 (ix86_option_override_internal): Ditto.
1013 (ix86_init_mpx_builtins): Move MPX to args2.
1014 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1015 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1016 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1017 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1018 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1019 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1020 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1022 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1025 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1026 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1027 (zero-extendsidi peephole2): Remove peephole.
1029 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1032 * stor-layout.c: Include attribs.h.
1033 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1034 explicit lookup of "aligned" attribute.
1036 2017-11-07 Andrew Waterman <andrew@sifive.com>
1038 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1039 (riscv_expand_block_move): Likewise.
1040 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1042 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1043 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1044 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1045 (riscv_adjust_block_mem): Likewise.
1046 (riscv_block_move_loop): Likewise.
1047 (riscv_expand_block_move): Likewise.
1048 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1050 2017-11-07 Michael Clark <michaeljclark@mac.com>
1052 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1053 (MUSL_DYNAMIC_LINKER): Likewise.
1055 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1057 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1058 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1061 2017-11-07 Richard Biener <rguenther@suse.de>
1063 * match.pd: Fix build.
1065 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1066 Jackson Woodruff <jackson.woodruff@arm.com>
1068 PR tree-optimization/71026
1069 * match.pd: Canonicalize negate in division.
1071 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1074 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1076 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1078 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1079 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1081 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1083 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1084 non-scalar integral types.
1085 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1086 (-(A-B), -(~A)): New transformations.
1088 2017-11-07 Tom de Vries <tom@codesourcery.com>
1090 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1091 semicolon after "do {} while (0)".
1092 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1093 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1094 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1095 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1096 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1098 2017-11-07 Tom de Vries <tom@codesourcery.com>
1100 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1101 after "do {} while (0)".
1102 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1103 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1104 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1105 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1106 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1108 2017-11-07 Tom de Vries <tom@codesourcery.com>
1111 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1113 (arm_rtx_costs_internal): Add missing semicolon after
1114 HANDLE_NARROW_SHIFT_ARITH call.
1116 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1118 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1119 disable isel if it was not set explicitly.
1121 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1123 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1125 (add_type_attribute) likewise.
1127 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1129 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1130 of struct ix86_frame.
1131 (ix86_initial_elimination_offset): Likewise.
1132 (ix86_expand_split_stack_prologue): Likewise.
1134 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1136 * tree-vrp.h (enum value_range_type): Update stale comment.
1138 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1139 Alan Hayward <alan.hayward@arm.com>
1140 David Sherwood <david.sherwood@arm.com>
1142 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1143 (aarch64_expand_vec_perm_const): Take the number of units too.
1144 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1145 (aarch64_expand_vec_perm_const): Likewise.
1146 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1147 (vec_perm<mode>): Update accordingly.
1149 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1150 Alan Hayward <alan.hayward@arm.com>
1151 David Sherwood <david.sherwood@arm.com>
1153 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1154 Take the number of units too.
1155 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1156 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1157 but check for a vector mode before rather than after the call.
1158 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1159 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1160 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1161 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1162 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1163 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1164 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1165 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1166 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1167 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1168 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1169 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1170 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1171 (aarch64_sqdmull2_n<mode>): Update accordingly.
1173 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1174 Alan Hayward <alan.hayward@arm.com>
1175 David Sherwood <david.sherwood@arm.com>
1177 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1178 the number of units too.
1179 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1180 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1181 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1182 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1183 (vec_store_lanesxi<mode>): Update accordingly.
1185 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1186 Alan Hayward <alan.hayward@arm.com>
1187 David Sherwood <david.sherwood@arm.com>
1189 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1190 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1191 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1192 of units rather than the mode.
1193 * config/aarch64/iterators.md (nunits): New mode attribute.
1194 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1195 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1196 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1197 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1198 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1199 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1200 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1201 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1202 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1203 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1204 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1205 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1206 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1207 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1208 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1209 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1210 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1211 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1212 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1213 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1214 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1215 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1216 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1217 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1218 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1219 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1220 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1221 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1222 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1223 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1224 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1225 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1226 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1227 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1228 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1229 (aarch64_simd_vec_setv2di): Likewise.
1231 2017-11-06 Carl Love <cel@us.ibm.com>
1233 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1235 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1236 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1238 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1240 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1241 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1242 Add power 8 macro expansions.
1243 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1244 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1245 power 8 instructions. (VSX_XXBR): Add iterator.
1247 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1249 * config/arm/arm.md (predicable_short_it): Change default to "no",
1250 improve documentation, remove uses that are identical to the default.
1251 (enabled_for_depr_it): Rename to enabled_for_short_it.
1252 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
1253 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1254 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1255 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1256 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1258 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1261 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1262 float128 helper macros, which are no longer used after deleting
1263 the old 'q' built-in functions, and moving the round to odd
1264 built-in functions to being special built-in functions.
1265 (BU_FLOAT128_2): Likewise.
1266 (BU_FLOAT128_1_HW): Likewise.
1267 (BU_FLOAT128_2_HW): Likewise.
1268 (BU_FLOAT128_3_HW): Likewise.
1269 (FABSQ): Delete old 'q' built-in functions.
1270 (COPYSIGNQ): Likewise.
1271 (SQRTF128_ODD): Move round to odd built-in functions to be
1272 special built-in functions, so that we can handle
1273 -mabi=ieeelongdouble.
1274 (TRUNCF128_ODD): Likewise.
1275 (ADDF128_ODD): Likewise.
1276 (SUBF128_ODD): Likewise.
1277 (MULF128_ODD): Likewise.
1278 (DIVF128_ODD): Likewise.
1279 (FMAF128_ODD): Likewise.
1280 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1281 built-in names to 'f128'.
1282 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1283 old 'q' built-in functions, as the machine independent code for
1284 'f128' built-in functions handles this.
1285 (rs6000_expand_builtin): Add expansion for float128 round to odd
1286 functions, keying off on -mabi=ieeelongdouble of whether to use
1287 the KFmode or TFmode variant.
1288 (rs6000_init_builtins): Initialize the _Float128 round to odd
1290 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1291 _Float128 'q' built-in functions are now mapped into the new
1292 'f128' built-in functions.
1294 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1296 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1297 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1299 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1302 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1303 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1304 * toplev.c: Include "ipa-fnsummary.h".
1305 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1307 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1309 PR tree-optimization/82838
1310 * gimple-ssa-store-merging.c
1311 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1312 on a separate gimple_seq which is then appended to seq.
1314 2017-11-06 Jeff Law <law@redhat.com>
1317 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1318 (get_probe_interval): New functions.
1319 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1320 (ix86_adjust_stack_and_probe): Likewise.
1321 (output_adjust_stack_and_probe): Likewise.
1322 (ix86_emit_probe_stack_range): Likewise.
1323 (ix86_expand_prologue): Likewise.
1325 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1327 PR tree-optimization/82816
1328 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1329 if the modes of the two types are the same.
1330 (convert_plusminus_to_widen): Likewise.
1332 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1334 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1336 (usadv16qi): New define_expand.
1337 (usadv8hi): New define_expand.
1339 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1342 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1344 (inline_transform): Likewise.
1345 * predict.c (counts_to_freqs): Remove useless conditional.
1346 * profile-count.h (profile_count::apply_scale): Move sanity check.
1347 * tree-inline.c (copy_bb): Always scale.
1348 (copy_cfg_body): Likewise.
1350 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1353 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1355 (*cmp_ite0): Add enabled_for_depr_it attribute.
1356 (*cmp_ite1): Likewise.
1358 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1360 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1363 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1365 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1366 return the value by pointer.
1367 (extract_range_from_multiplicative_op_1): Update accordingly.
1368 Return as soon as an operation fails.
1370 2017-11-05 Tom de Vries <tom@codesourcery.com>
1373 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1374 (DEF_SANITIZER_BUILTIN): ... here.
1375 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1376 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1378 2017-11-05 Tom de Vries <tom@codesourcery.com>
1381 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1383 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1384 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1385 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1387 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1388 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1389 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1391 2017-11-05 Tom de Vries <tom@codesourcery.com>
1394 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1397 2017-11-04 Michael Clark <michaeljclark@mac.com>
1399 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1400 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1402 (*addsi3_extended): Likewise.
1403 (*addsi3_extended2): Likewise.
1404 (<optab>si3): Likewise.
1405 (<optab>di3): Likewise.
1406 (<optab><mode>3): Likewise.
1407 (<*optabe>si3_internal): Likewise.
1408 (zero_extendqi<SUPERQI:mode>2): Likewise.
1409 (*add<mode>hi3): Likewise.
1410 (*xor<mode>hi3): Likewise.
1411 (<optab>di3): Likewise.
1412 (*<optab>si3_extend): Likewise.
1413 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1414 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1415 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1417 2017-11-04 Andrew Waterman <andrew@sifive.com>
1419 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1420 TARGET_STRICT_ALIGN based upon -mtune argument.
1422 2017-11-04 Andrew Waterman <andrew@sifive.com>
1424 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1426 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1428 config/i386/i386.c (choose_basereg): Use optional scratch
1429 register and add assertion.
1430 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1431 needed, and don't allocate stack.
1432 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1433 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1434 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1436 2017-11-03 Jeff Law <law@redhat.com>
1438 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1439 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1440 to probe at the start of a noreturn function.
1442 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1444 PR tree-optimization/78821
1445 * gimple-ssa-store-merging.c: Update the file comment.
1446 (MAX_STORE_ALIAS_CHECKS): Define.
1447 (struct store_operand_info): New type.
1448 (store_operand_info::store_operand_info): New constructor.
1449 (struct store_immediate_info): Add rhs_code and ops data members.
1450 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1451 and op1r arguments to the ctor, initialize corresponding data members.
1452 (struct merged_store_group): Add load_align_base and load_align
1454 (merged_store_group::merged_store_group): Initialize them.
1455 (merged_store_group::do_merge): Update them.
1456 (merged_store_group::apply_stores): Pick the constant for
1457 encode_tree_to_bitpos from one of the two operands, or skip
1458 encode_tree_to_bitpos if neither operand is a constant.
1459 (class pass_store_merging): Add process_store method decl. Remove
1460 bool argument from terminate_all_aliasing_chains method decl.
1461 (pass_store_merging::terminate_all_aliasing_chains): Remove
1462 var_offset_p argument and corresponding handling.
1463 (stmts_may_clobber_ref_p): New function.
1464 (compatible_load_p): New function.
1465 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1466 if there is overlap and rhs_code is not INTEGER_CST. For
1467 non-overlapping stores terminate group if rhs is not mergeable.
1468 (get_alias_type_for_stmts): Change first argument from
1469 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1470 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1471 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1473 (get_location_for_stmts): Change first argument from
1474 auto_vec<gimple *> & to vec<gimple *> &.
1475 (struct split_store): Remove orig_stmts data member, add orig_stores.
1476 (split_store::split_store): Create orig_stores rather than orig_stmts.
1477 (find_constituent_stmts): Renamed to ...
1478 (find_constituent_stores): ... this. Change second argument from
1479 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1480 to info structures rather than the statements.
1481 (split_group): Rename ALLOW_UNALIGNED argument to
1482 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1483 it. Adjust find_constituent_stores caller.
1484 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1485 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1486 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1487 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1488 (mem_valid_for_store_merging): New function.
1489 (handled_load): New function.
1490 (pass_store_merging::process_store): New method.
1491 (pass_store_merging::execute): Use process_store method. Adjust
1492 terminate_all_aliasing_chains caller.
1494 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1496 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1497 Return true for more constants, symbols and label references.
1498 (aarch64_valid_floating_const): Remove unused function.
1500 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1503 * config/aarch64/aarch64.c (aarch64_layout_frame):
1504 Undo forcing of LR at bottom of frame.
1506 2017-11-03 Jeff Law <law@redhat.com>
1509 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1510 for int_registers_saved.
1512 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1513 extracted from tree-ssa-dom.c.
1514 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1515 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1516 (record_equivalences_from_incoming_edge): Add additional argument
1517 to single_pred_edge_ignoring_loop_edges call.
1518 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1519 (uncprop_dom_walker::before_dom_children): Add additional argument
1520 to single_pred_edge_ignoring_loop_edges call.
1521 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1522 single_pred_edge_ignoring_loop_edges rather than open coding.
1523 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1525 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1527 * match.pd (-(-A)): Rewrite.
1529 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1531 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1532 (rs6000_emit_int_cmove): New declaration.
1533 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1534 (rs6000_emit_sISEL): Delete.
1535 (rs6000_emit_int_cmove): Make non-static.
1536 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1537 instead of rs6000_emit_sISEL.
1539 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1541 * asan.c (create_cond_insert_point): Maintain profile.
1542 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1544 * basic-block.h (struct basic_block_def): Remove frequency.
1545 (EDGE_FREQUENCY): Use to_frequency
1546 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1548 (find_traces): Update to use to_frequency.
1549 (find_traces_1_round): Likewise; use only IPA counts.
1550 (bb_to_key): Likewise.
1551 (connect_traces): Use IPA counts only.
1552 (copy_bb_p): Update to use to_frequency.
1553 (fix_up_crossing_landing_pad): Likewise.
1554 (sanitize_hot_paths): Likewise.
1555 * bt-load.c (basic_block_freq): Likewise.
1556 * cfg.c (init_flow): Set count_max to uninitialized.
1557 (check_bb_profile): Remove frequencies; check counts.
1558 (dump_bb_info): Do not dump frequencies.
1559 (update_bb_profile_for_threading): Update counts only.
1560 (scale_bbs_frequencies_int): Likewise.
1561 (MAX_SAFE_MULTIPLIER): Remove.
1562 (scale_bbs_frequencies_gcov_type): Update counts only.
1563 (scale_bbs_frequencies_profile_count): Update counts only.
1564 (scale_bbs_frequencies): Update counts only.
1565 * cfg.h (struct control_flow_graph): Add count-max.
1566 (update_bb_profile_for_threading): Update prototype.
1567 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1568 (find_many_sub_basic_blocks): Likewise.
1569 * cfgcleanup.c (try_forward_edges): Likewise.
1570 (try_crossjump_to_edge): Likewise.
1571 * cfgexpand.c (expand_gimple_cond): Likewise.
1572 (expand_gimple_tailcall): Likewise.
1573 (construct_init_block): Likewise.
1574 (construct_exit_block): Likewise.
1575 * cfghooks.c (verify_flow_info): Check consistency of counts.
1576 (dump_bb_for_graph): Do not dump frequencies.
1577 (split_block_1): Do not update frequencies.
1578 (split_edge): Do not update frequencies.
1579 (make_forwarder_block): Do not update frequencies.
1580 (duplicate_block): Do not update frequencies.
1581 (account_profile_record): Do not update frequencies.
1582 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1583 for global heuristics.
1584 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1585 (expected_loop_iterations_unbounded): Use counts only.
1586 * cfgloopmanip.c (scale_loop_profile): Simplify.
1587 (create_empty_loop_on_edge): Simplify
1589 (duplicate_loop_to_header_edge): Simplify
1590 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1591 (update_br_prob_note): Take care of removing note when profile
1593 (relink_block_chain): Do not dump frequency.
1594 (rtl_account_profile_record): Use to_frequency.
1595 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1596 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1597 (cgraph_update_edges_for_call_stmt_node): Likewise.
1598 (cgraph_edge::verify_count_and_frequency): Update.
1599 (cgraph_node::verify_node): Temporarily disable frequency verification.
1600 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1601 to_cgraph_frequency.
1602 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1603 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1605 (cgraph_node::expand_thunk): Update profile.
1606 * except.c (dw2_build_landing_pads): Do not update frequency.
1607 * final.c (compute_alignments): Use to_frequency.
1608 (dump_basic_block_info): Do not dump frequency.
1609 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1610 (dump_gimple_bb_header): Do not dump frequency.
1611 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1613 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1614 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1615 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1616 (init_before_recovery): Do not update frequency.
1617 (sched_create_recovery_edges): Do not update frequency.
1618 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1619 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1620 (ipa_cp_c_finalize): Set max_count to uninitialized.
1621 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1622 (param_change_prob): Use counts.
1623 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1625 * ipa-split.c (consider_split): Use to_frequency.
1626 (split_function): Use to_frequency.
1627 * ira-build.c (loop_compare_func): Likewise.
1628 (mark_loops_for_removal): Likewise.
1629 (mark_all_loops_for_removal): Likewise.
1630 * loop-doloop.c (doloop_modify): Do not update frequency.
1631 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1633 * lto-streamer-in.c (input_function): Update count_max.
1634 * omp-expand.c (expand_omp_taskreg): Update count_max.
1635 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1636 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1637 (maybe_hot_count_p): Use ipa counts only.
1638 (maybe_hot_bb_p): Simplify.
1639 (maybe_hot_edge_p): Simplify.
1640 (probably_never_executed): Do not take frequency argument.
1641 (probably_never_executed_bb_p): Do not pass frequency.
1642 (probably_never_executed_edge_p): Likewise.
1643 (combine_predictions_for_bb): Check that profile is nonzero.
1644 (propagate_freq): Do not set frequency.
1645 (drop_profile): Simplify.
1646 (counts_to_freqs): Simplify.
1647 (expensive_function_p): Use to_frequency.
1648 (propagate_unlikely_bbs_forward): Simplify.
1649 (determine_unlikely_bbs): Simplify.
1650 (estimate_bb_frequencies): Add hack to silence graphite issues.
1651 (compute_function_frequency): Use ipa counts.
1652 (pass_profile::execute): Update.
1653 (rebuild_frequencies): Use counts only.
1654 (force_edge_cold): Use counts only.
1655 * profile-count.c (profile_count::dump): Dump new count types.
1656 (profile_count::differs_from_p): Check compatiblity.
1657 (profile_count::to_frequency): New function.
1658 (profile_count::to_cgraph_frequency): New function.
1659 * profile-count.h (struct function): Declare.
1660 (enum profile_quality): Add profile_guessed_local and
1661 profile_guessed_global0.
1662 (class profile_proability): Decrease number of bits to 29;
1663 update from_reg_br_prob_note and to_reg_br_prob_note.
1664 (class profile_count: Update comment; decrease number of bits
1665 to 61. Check compatibility.
1666 (profile_count::compatible_p): New private member function.
1667 (profile_count::ipa_p): New member function.
1668 (profile_count::operator<): Handle global zero correctly.
1669 (profile_count::operator>): Handle global zero correctly.
1670 (profile_count::operator<=): Handle global zero correctly.
1671 (profile_count::operator>=): Handle global zero correctly.
1672 (profile_count::nonzero_p): New member function.
1673 (profile_count::force_nonzero): New member function.
1674 (profile_count::max): New member function.
1675 (profile_count::apply_scale): Handle IPA scalling.
1676 (profile_count::guessed_local): New member function.
1677 (profile_count::global0): New member function.
1678 (profile_count::ipa): New member function.
1679 (profile_count::to_frequency): Declare.
1680 (profile_count::to_cgraph_frequency): Declare.
1681 * profile.c (OVERLAP_BASE): Delete.
1682 (compute_frequency_overlap): Delete.
1683 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1684 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1685 * sched-ebb.c (rank): Use counts only.
1686 * shrink-wrap.c (handle_simple_exit): Use counts only.
1687 (try_shrink_wrapping): Use counts only.
1688 (place_prologue_for_one_component): Use counts only.
1689 * tracer.c (find_best_predecessor): Use to_frequency.
1690 (find_trace): Use to_frequency.
1691 (tail_duplicate): Use to_frequency.
1692 * trans-mem.c (expand_transaction): Do not update frequency.
1693 * tree-call-cdce.c: Do not update frequency.
1694 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1695 (gimple_merge_blocks): Likewise.
1696 (gimple_split_edge): Likewise.
1697 (gimple_duplicate_sese_region): Likewise.
1698 (gimple_duplicate_sese_tail): Likewise.
1699 (move_sese_region_to_fn): Likewise.
1700 (gimple_account_profile_record): Likewise.
1701 (insert_cond_bb): Likewise.
1702 * tree-complex.c (expand_complex_div_wide): Likewise.
1703 * tree-eh.c (lower_resx): Update profile.
1704 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1706 (initialize_cfun): Do not initialize frequencies
1707 (freqs_to_counts): Delete.
1708 (copy_cfg_body): Ignore count parameter.
1709 (copy_body): Update.
1710 (expand_call_inline): Update count_max.
1711 (optimize_inline_calls): Update count_max.
1712 (tree_function_versioning): Update count_max.
1713 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1714 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1716 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1717 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1718 (try_peel_loop): Likewise.
1719 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1721 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1722 (tree_transform_and_unroll_loop): Do not use frequencies
1723 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1724 Use reliable prediction only.
1725 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1726 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1727 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1728 probability scaling.
1729 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1730 not update frequency
1731 (any_remaining_duplicated_blocks): Likewise.
1732 (update_profile): Likewise.
1733 (estimated_freqs_path): Delete.
1734 (freqs_to_counts_path): Delete.
1735 (clear_counts_path): Delete.
1736 (ssa_fix_duplicate_block_edges): Likewise.
1737 (duplicate_thread_path): Likewise.
1738 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1739 * tree-tailcall.c (decrease_profile): Do not update frequency.
1740 (eliminate_tail_call): Likewise.
1741 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1742 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1743 (optimize_mask_stores): Likewise.
1744 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1745 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1746 (ubsan_expand_ptr_ifn): Update profile.
1747 * value-prof.c (gimple_ic): Simplify.
1748 * value-prof.h (gimple_ic): Update prototype.
1749 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1750 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1752 (want_inline_self_recursive_call_p): Likewise.
1753 (resolve_noninline_speculation): Only cummulate defined counts.
1754 (inline_small_functions): Use nonzero_p.
1755 (ipa_inline): Do not access freed node.
1757 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1759 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1760 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1762 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1764 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1765 non-legitimate address.
1767 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1769 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1770 (*lt0_<mode>di, *lt0_<mode>si): New.
1772 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1774 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1775 TARGET_PAIRED_FLOAT.
1777 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1778 Jim Wilson <jim.wilson@linaro.org>
1780 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1781 * config/aarch64/aarch64-tune.md: Regenerated.
1782 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1783 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1785 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1787 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1788 (arc_expand_prologue): Restore blink for millicode.
1789 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1791 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1794 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1795 gen_vec_duplicate after forcing the scalar into a register.
1797 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1799 * combine (try_combine): Print the insns input to try_combine to the
1802 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1805 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1806 Remove second argument from aarch64_process_target_attr call.
1807 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1809 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1811 (aarch64_handle_attr_arch): Remove second argument.
1812 (aarch64_handle_attr_cpu): Ditto.
1813 (aarch64_handle_attr_tune): Ditto.
1814 (aarch64_handle_attr_isa_flags): Ditto.
1815 (aarch64_process_one_target_attr): Ditto.
1816 (aarch64_process_target_attr): Ditto.
1817 (aarch64_option_valid_attribute_p): Remove second argument.
1818 on aarch64_process_target_attr call.
1820 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1822 * diagnostic.c: Include "selftest-diagnostic.h".
1823 (selftest::assert_location_text): New function.
1824 (selftest::test_diagnostic_get_location_text): New function.
1825 (selftest::diagnostic_c_tests): Call it.
1827 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1829 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1830 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1831 (class selftest::test_diagnostic_context): Move to...
1832 * selftest-diagnostic.c: New file.
1833 * selftest-diagnostic.h: New file.
1835 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1837 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1838 offset range for FT32B.
1839 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1840 * config/ft32/ft32.md: Add TARGET_NOPM.
1841 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1842 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1844 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1846 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1848 2017-11-02 Jeff Law <law@redhat.com>
1850 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1851 virtual keyword on FINAL OVERRIDE members.
1853 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1854 virtuals together. Add virtual destructor.
1855 (substitute_and_fold_engine): Similarly.
1857 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1859 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1861 2017-11-02 Richard Biener <rguenther@suse.de>
1863 PR tree-optimization/82795
1864 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1866 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1868 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1870 (gcc_GAS_CHECK_FEATURE): Remove.
1871 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1872 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1873 * configure: Regenerate.
1875 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1877 * config/arc/arc.c (hwloop_optimize): Account for empty
1880 2017-11-02 Richard Biener <rguenther@suse.de>
1883 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1884 Truncate ARRAY_REF index and element size.
1886 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1888 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1890 2017-11-01 Jeff Law <law@redhat.com>
1892 * tree-ssa-ccp.c (ccp_folder): New class derived from
1893 substitute_and_fold_engine.
1894 (ccp_folder::get_value): New member function.
1895 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1896 (ccp_fold_stmt): Remove prototype.
1897 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1898 * tree-ssa-copy.c (copy_folder): New class derived from
1899 substitute_and_fold_engine.
1900 (copy_folder::get_value): Renamed from get_value.
1901 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1902 * tree-vrp.c (vrp_folder): New class derived from
1903 substitute_and_fold_engine.
1904 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1905 (vrp_folder::get_value): New member function.
1906 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1907 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1908 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1909 provide a class interface to folder/substitute routines.
1910 (ssa_prop_fold_stmt_fn): Remove typedef.
1911 (ssa_prop_get_value_fn): Likewise.
1912 (subsitute_and_fold): Remove prototype.
1913 (replace_uses_in): Likewise.
1914 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1915 Renamed from replace_uses_in. Call the virtual member function
1916 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1917 (substitute_and_fold_dom_walker): Remove initialization of
1918 data member entries for calbacks. Add substitute_and_fold_engine
1919 member and initialize it.
1920 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1921 member functions for get_value, replace_phi_args_in c
1922 replace_uses_in, and fold_stmt calls.
1923 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1924 substitute_and_fold. Remove assert. Update ctor call.
1926 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1927 (ssa_prop_visit_phi_fn): Likewise.
1928 (class ssa_propagation_engine): New class to provide an interface
1930 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1932 (ssa_prop_visit_phi): Likewise.
1933 (ssa_propagation_engine::simulate_stmt): Moved into class.
1934 Call visit_phi/visit_stmt from the class rather than via
1935 file scoped static variables.
1936 (ssa_propagation_engine::simulate_block): Moved into class.
1937 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1938 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1939 set file scoped statics for the visit_stmt/visit_phi callbacks.
1940 * tree-complex.c (complex_propagate): New class derived from
1941 ssa_propagation_engine.
1942 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1943 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1944 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1946 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1947 ssa_propagation_engine.
1948 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1949 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1950 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1951 * tree-ssa-copy.c (copy_prop): New class derived from
1952 ssa_propagation_engine.
1953 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1954 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1955 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1956 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1957 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1958 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1959 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1961 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1963 PR rtl-optimization/82778
1964 PR rtl-optimization/82597
1965 * compare-elim.c (struct comparison): Add in_a_setter field.
1966 (find_comparison_dom_walker::before_dom_children): Remove killed
1967 bitmap and df_simulate_find_defs call, instead walk the defs.
1968 Compute last_setter and initialize in_a_setter. Merge definitions
1969 with first initialization for a few variables.
1970 (try_validate_parallel): Use insn_invalid_p instead of
1971 recog_memoized. Return insn rather than just the pattern.
1972 (try_merge_compare): Fix up comment. Don't uselessly test if
1973 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1975 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1978 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1979 Alan Hayward <alan.hayward@arm.com>
1980 David Sherwood <david.sherwood@arm.com>
1982 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1983 aarch64_hard_regno_nregs to get the number of registers
1986 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1987 Alan Hayward <alan.hayward@arm.com>
1988 David Sherwood <david.sherwood@arm.com>
1990 * config/aarch64/constraints.md (Upl): Rename to...
1992 * config/aarch64/aarch64.md
1993 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1996 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1997 Alan Hayward <alan.hayward@arm.com>
1998 David Sherwood <david.sherwood@arm.com>
2000 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2001 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2004 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2005 Alan Hayward <alan.hayward@arm.com>
2006 David Sherwood <david.sherwood@arm.com>
2008 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2009 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2010 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2012 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2013 of a single element.
2014 * config/aarch64/iterators.md: Add a comment above the permute
2015 unspecs to say that they are generated directly by
2016 aarch64_expand_vec_perm_const.
2017 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2019 2017-11-01 Nathan Sidwell <nathan@acm.org>
2021 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2023 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2025 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2026 and medany code models, and describe what they do.
2028 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2030 Revert accidental duplicate:
2032 * combine.c (can_change_dest_mode): Reject changes in
2033 REGMODE_NATURAL_SIZE.
2035 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2037 PR rtl-optimization/64682
2038 PR rtl-optimization/69567
2039 PR rtl-optimization/69737
2040 PR rtl-optimization/82683
2041 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2042 register mentioned in the note, drop the note, unless it came from I3,
2043 in which case it should go to I3 again.
2045 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2047 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2048 and return false if not.
2049 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2051 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2053 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2054 range is known to be empty.
2056 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2057 Alan Hayward <alan.hayward@arm.com>
2058 David Sherwood <david.sherwood@arm.com>
2060 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2061 and CONST_VECTOR_NUNITS instead of computing the number of units from
2062 the byte sizes of the vector and element.
2063 (simplify_binary_operation_1): Likewise.
2064 (simplify_const_binary_operation): Likewise.
2065 (simplify_ternary_operation): Likewise.
2067 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2068 Alan Hayward <alan.hayward@arm.com>
2069 David Sherwood <david.sherwood@arm.com>
2071 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2072 (int_mem_offset): ...this new function.
2073 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2074 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2075 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2078 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2079 Alan Hayward <alan.hayward@arm.com>
2080 David Sherwood <david.sherwood@arm.com>
2082 * lower-subreg.c (interesting_mode_p): New function.
2083 (compute_costs, find_decomposable_subregs, decompose_register)
2084 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2085 (resolve_clobber, dump_choices): Use it.
2087 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2088 Alan Hayward <alan.hayward@arm.com>
2089 David Sherwood <david.sherwood@arm.com>
2091 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2093 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2094 Alan Hayward <alan.hayward@arm.com>
2095 David Sherwood <david.sherwood@arm.com>
2097 * alias.c (find_base_value, find_base_term): Only process integer
2098 truncations. Check the precision rather than the size.
2100 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2101 Alan Hayward <alan.hayward@arm.com>
2102 David Sherwood <david.sherwood@arm.com>
2104 * machmode.h (is_narrower_int_mode): New function
2105 * optabs.c (expand_float, expand_fix): Use it.
2106 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2108 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2109 Alan Hayward <alan.hayward@arm.com>
2110 David Sherwood <david.sherwood@arm.com>
2112 * rtl.h (narrower_subreg_mode): New function.
2113 * ira-color.c (update_costs_from_allocno): Use it.
2115 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2116 Alan Hayward <alan.hayward@arm.com>
2117 David Sherwood <david.sherwood@arm.com>
2119 * optabs-query.h (convert_optab_p): New function, split out from...
2120 (convert_optab_handler): ...here.
2121 (widening_optab_handler): Delete.
2122 (find_widening_optab_handler): Remove permit_non_widening parameter.
2123 (find_widening_optab_handler_and_mode): Likewise. Provide an
2124 override that operates on mode class wrappers.
2125 * optabs-query.c (widening_optab_handler): Delete.
2126 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2127 parameter. Assert that the two modes are the same class and that
2128 the "from" mode is narrower than the "to" mode. Use
2129 convert_optab_handler instead of widening_optab_handler.
2130 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2131 instead of widening_optab_handler.
2132 * expr.c (expand_expr_real_2): Update calls to
2133 find_widening_optab_handler.
2134 * optabs.c (expand_widen_pattern_expr): Likewise.
2135 (expand_binop_directly): Take the insn_code as a parameter.
2136 (expand_binop): Only call find_widening_optab_handler for
2137 conversion optabs; use optab_handler otherwise. Update calls
2138 to find_widening_optab_handler and expand_binop_directly.
2139 Use convert_optab_handler instead of widening_optab_handler.
2140 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2141 find_widening_optab_handler and use scalar_mode rather than
2143 (convert_plusminus_to_widen): Likewise.
2145 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2146 Alan Hayward <alan.hayward@arm.com>
2147 David Sherwood <david.sherwood@arm.com>
2149 * machmode.h (fixed_size_mode): New class.
2150 * rtl.h (get_pool_mode): Return fixed_size_mode.
2151 * gengtype.c (main): Add fixed_size_mode.
2152 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2153 (get_raw_arg_mode): Likewise.
2154 * doc/tm.texi: Regenerate.
2155 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2156 * targhooks.c (default_get_reg_raw_mode): Likewise.
2157 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2158 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2159 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2160 (msp430_get_raw_result_mode): Likewise.
2161 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2162 * dbxout.c (dbxout_parms): Require fixed-size modes.
2163 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2164 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2165 * omp-low.c (lower_oacc_reductions): Likewise.
2166 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2167 (simplify_subreg): Update accordingly.
2168 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2169 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2170 that aren't fixed-size.
2171 (get_pool_mode): Return a fixed_size_mode.
2172 (output_constant_pool_2): Take a fixed_size_mode.
2174 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2175 Alan Hayward <alan.hayward@arm.com>
2176 David Sherwood <david.sherwood@arm.com>
2178 * doc/rtl.texi (vec_series): Document.
2179 (const): Say that the operand can be a vec_series.
2180 * rtl.def (VEC_SERIES): New rtx code.
2181 * rtl.h (const_vec_series_p_1): Declare.
2182 (const_vec_series_p): New function.
2183 * emit-rtl.h (gen_const_vec_series): Declare.
2184 (gen_vec_series): Likewise.
2185 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2186 (gen_vec_series): Likewise.
2187 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2188 * simplify-rtx.c (simplify_unary_operation): Handle negations
2190 (simplify_binary_operation_series): New function.
2191 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2192 (test_vector_ops_series): New function.
2193 (test_vector_ops): Call it.
2194 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2195 gen_const_vec_series.
2196 (altivec_lvsr): Likewise.
2197 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2199 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2200 Alan Hayward <alan.hayward@arm.com>
2201 David Sherwood <david.sherwood@arm.com>
2203 * doc/rtl.texi (const): Update description of address constants.
2204 Say that vector constants are allowed too.
2205 * common.md (E, F): Use CONSTANT_P instead of checking for
2207 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2208 checking for CONST_VECTOR.
2209 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2211 * expr.c (expand_expr_real_2): Check for vector modes instead
2212 of checking for CONST_VECTOR.
2213 * rtl.h (const_vec_p): New function.
2214 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2215 (unwrap_const_vec_duplicate): Handle them here too.
2217 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2218 David Malcolm <dmalcolm@redhat.com>
2219 Alan Hayward <alan.hayward@arm.com>
2220 David Sherwood <david.sherwood@arm.com>
2222 * rtl.h (vec_duplicate_p): New function.
2223 * selftest-rtl.c (assert_rtx_eq_at): New function.
2224 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2225 (assert_rtx_eq_at): Declare.
2226 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2227 * selftest-run-tests.c (selftest::run_tests): Call it.
2228 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2229 (simplify_unary_operation_1): Recursively handle vector duplicates.
2230 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2232 (simplify_subreg): Handle subregs of vector duplicates.
2233 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2234 (selftest::simplify_rtx_c_tests): New functions.
2236 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2237 Alan Hayward <alan.hayward@arm.com>
2238 David Sherwood <david.sherwood@arm.com>
2240 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2241 (gen_vec_duplicate): Likewise.
2242 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2244 (gen_const_vector): ...here.
2245 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2246 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2247 whose elements are all equal.
2248 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2249 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2250 (simplify_relational_operation): Likewise.
2251 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2253 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2254 (aarch64_expand_vector_init): Likewise.
2255 * config/arm/arm.c (neon_vdup_constant): Likewise.
2256 (neon_expand_vector_init): Likewise.
2257 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2258 (arm_block_set_unaligned_vect): Likewise.
2259 (arm_block_set_aligned_vect): Likewise.
2260 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2261 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2262 (expand_vec_perm_even_odd_pack): Likewise.
2263 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2264 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2265 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2266 gen_const_vec_duplicate.
2267 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2268 * config/mips/mips.c (mips_gen_const_int_vector): Use
2269 gen_const_vec_duplicate.
2270 (mips_expand_vector_init): Use CONST0_RTX.
2271 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2272 (define_split): Use gen_const_vec_duplicate.
2273 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2274 (define_split): Use gen_const_vec_duplicate.
2275 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2276 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2277 * config/spu/spu.c (spu_const): Likewise.
2279 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2280 Alan Hayward <alan.hayward@arm.com>
2281 David Sherwood <david.sherwood@arm.com>
2283 * combine.c (can_change_dest_mode): Reject changes in
2284 REGMODE_NATURAL_SIZE.
2286 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2288 * configure.ac (--enable-libssp): New.
2289 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2290 trying to determine target-specific default. Adjust indentation.
2291 * configure: Regenerated.
2292 * doc/install.texi (Configuration): Expand --disable-libssp
2295 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2297 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2300 2017-10-31 Martin Jambor <mjambor@suse.cz>
2303 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2305 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2307 * auto-profile.c (autofdo_source_profile::read): Use
2308 UNKNOWN_LOCATION rather than 0.
2309 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2310 (warning_at): ...this overload.
2311 (warning_at_rich_loc_n): Rename to...
2312 (warning_n): ...this overload.
2313 (error_at_rich_loc): Rename to...
2314 (error_at): ...this overload.
2315 (pedwarn_at_rich_loc): Rename to...
2316 (pedwarn): ...this overload.
2317 (permerror_at_rich_loc): Rename to...
2318 (permerror): ...this overload.
2319 (inform_at_rich_loc): Rename to...
2320 (inform): ...this overload.
2321 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2322 (diagnostic_n_impl_richloc): Rename to...
2323 (diagnostic_n_impl): ...this rich_location *-based decl.
2324 (inform_at_rich_loc): Rename to...
2325 (inform): ...this, and add an assertion.
2326 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2327 (warning_at_rich_loc): Rename to...
2328 (warning_at): ...this, and add an assertion.
2329 (warning_at_rich_loc_n): Rename to...
2330 (warning_n): ...this, and add an assertion.
2331 (warning_n): Update location_t-based implementation for removal of
2332 location_t-based diagnostic_n_impl.
2333 (pedwarn_at_rich_loc): Rename to...
2334 (pedwarn): ...this, and add an assertion.
2335 (permerror_at_rich_loc): Rename to...
2336 (permerror): ...this, and add an assertion.
2337 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2338 (error_at_rich_loc): Rename to...
2339 (error_at): ...this, and add an assertion.
2340 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2341 (driver::do_spec_on_infiles): Likewise.
2342 * substring-locations.c (format_warning_va): Update for renaming
2343 of inform_at_rich_loc.
2345 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2347 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2348 _Float<N>X built-in functions so that the variant without the
2349 "__builtin_" prefix is only enabled for the GNU C and Objective C
2350 languages when they are in non-strict ANSI/ISO mode.
2351 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2352 * target.def (floatn_builtin_p): Add a target hook to control
2353 whether _Float<N> and _Float<N>X built-in functions without the
2354 "__builtin_" prefix are enabled, and return true for C and
2355 Objective C in the default hook. Include langhooks.h in
2357 * targhooks.h (default_floatn_builtin_p): Likewise.
2358 * targhooks.c (default_floatn_builtin_p): Likewise.
2359 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2360 floatn_builtin_p target hook.
2361 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2363 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2364 Eric Botcazou <ebotcazou@adacore.com>
2366 PR rtl-optimization/81803
2367 * lra-constraints.c (curr_insn_transform): Also reload the whole
2368 register for a strict subreg no wider than a word if this is for
2369 a WORD_REGISTER_OPERATIONS target.
2371 2017-10-31 Jason Merrill <jason@redhat.com>
2373 * gdbinit.in: Skip over inlines from timevar.h.
2375 2017-10-31 Martin Liska <mliska@suse.cz>
2377 * doc/gcov.texi: Document new option.
2378 * gcov.c (print_usage): Likewise print it.
2379 (process_args): Support the argument.
2380 (format_count): New function.
2381 (format_gcov): Use the function.
2383 2017-10-31 Martin Liska <mliska@suse.cz>
2385 * gcov.c (struct name_map): do not use typedef.
2386 Define operator== and operator<.
2387 (name_search): Remove.
2388 (name_sort): Remove.
2389 (main): Do not allocate names.
2390 (process_file): Add vertical space.
2391 (generate_results): Use std::find.
2392 (release_structures): Do not release memory.
2393 (find_source): Use std::find.
2395 2017-10-31 Martin Liska <mliska@suse.cz>
2397 * gcov.c (struct line_info): Remove it's typedef.
2398 (line_info::line_info): Add proper ctor.
2399 (line_info::has_block): Do not use a typedef.
2400 (struct source_info): Do not use typedef.
2401 (circuit): Likewise.
2402 (get_cycles_count): Likewise.
2403 (output_intermediate_file): Iterate via vector iterator.
2404 (add_line_counts): Use std::vector methods.
2405 (accumulate_line_counts): Likewise.
2406 (output_lines): Likewise.
2408 2017-10-31 Martin Liska <mliska@suse.cz>
2410 * gcov.c (struct source_info): Remove typedef.
2411 (source_info::source_info): Add proper ctor.
2412 (accumulate_line_counts): Use struct, not it's typedef.
2413 (output_gcov_file): Likewise.
2414 (output_lines): Likewise.
2415 (main): Do not allocate an array.
2416 (output_intermediate_file): Use size of vector container.
2417 (process_file): Resize the vector.
2418 (generate_results): Do not preallocate, use newly added vector
2420 (release_structures): Do not release sources.
2421 (find_source): Use vector methods.
2422 (add_line_counts): Do not use typedef.
2424 2017-10-31 Martin Liska <mliska@suse.cz>
2426 * doc/gcov.texi: Document that.
2427 * gcov.c (add_line_counts): Mark lines with a non-executed
2429 (output_line_beginning): Handle such lines.
2430 (output_lines): Pass new argument.
2431 (output_intermediate_file): Print it in intermediate format.
2433 2017-10-31 Martin Liska <mliska@suse.cz>
2435 * color-macros.h: New file.
2436 * diagnostic-color.c: Factor out color related to macros to
2438 * doc/gcov.texi: Document -k option.
2439 * gcov.c (INCLUDE_STRING): Include string.h.
2440 (print_usage): Add -k option.
2441 (process_args): Parse it.
2442 (pad_count_string): New function.
2443 (output_line_beginning): Likewise.
2444 (DEFAULT_LINE_START): New macro.
2445 (output_lines): Support color output.
2447 2017-10-31 Martin Liska <mliska@suse.cz>
2449 PR gcov-profile/82633
2450 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2451 their interaction with GCOV infrastructure.
2452 * configure.ac: Add -fkeep-{inline,static}-functions to
2454 * configure: Regenerate.
2456 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2459 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2461 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2464 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2465 into a register if it does not fit into an immediate offset field.
2467 2017-10-31 Olivier Hainque <hainque@adacore.com>
2469 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2471 2017-10-31 Julia Koval <julia.koval@intel.com>
2473 * config.gcc: Add gfniintrin.h.
2474 * config/i386/gfniintrin.h: New.
2475 * config/i386/i386-builtin-types.def
2476 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2477 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2478 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2479 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2480 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2481 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2482 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2483 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2484 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2485 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2486 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2487 * config/i386/immintrin.h: Include gfniintrin.h.
2488 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2490 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2492 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2494 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2496 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2497 (arm_ashldi3_1bit): Remove pattern.
2498 (ashrdi3): Remove shift by 1 expansion.
2499 (arm_ashrdi3_1bit): Remove pattern.
2500 (lshrdi3): Remove shift by 1 expansion.
2501 (arm_lshrdi3_1bit): Remove pattern.
2502 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2503 cost of ashldi3 by 1.
2504 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2505 (<shift>di3_neon): Likewise.
2507 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2509 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2510 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2511 and (*aarch64_simd_mov<VQ:mode>).
2512 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2513 pattern alternative.
2514 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2515 attributes to match pattern alternative.
2517 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2519 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2520 * config/rs6000/emmintrin.h: New file.
2521 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2523 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2525 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2526 * (movdi_vfp_cortexa8): Remove pattern.
2528 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2530 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2532 (Specific, alpha*-dec-osf5.1): Remove.
2533 (Specific, mips-sgi-irix5): Remove.
2534 (Specific, mips-sgi-irix6): Remove.
2536 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2539 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2540 arguments to clear_bit_region_be.
2542 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2544 * gimplify.c: Include memmodel.h.
2546 2017-10-30 Martin Jambor <mjambor@suse.cz>
2548 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2549 condition whether loop should be executed at all.
2551 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2553 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2554 gimple folding of vec_madd() intrinsics.
2555 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2556 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2557 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2559 2017-10-30 Richard Biener <rguenther@suse.de>
2561 PR tree-optimization/82762
2563 2017-10-23 Richard Biener <rguenther@suse.de>
2565 PR tree-optimization/82129
2567 2017-08-01 Richard Biener <rguenther@suse.de>
2569 PR tree-optimization/81181
2570 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2571 (compute_antic): ... end of iteration here.
2573 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2575 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2576 -std=iso9899:2017 and -std=gnu17.
2577 * doc/standards.texi (C Language): Document C17 support.
2578 * doc/cpp.texi (Overview): Mention -std=c17.
2579 (Standard Predefined Macros): Document C11 and C17 values of
2580 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2581 * doc/extend.texi (Inline): Do not list individual options for
2582 standards newer than C99.
2583 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2585 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2588 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2590 * asan.c (asan_finish_file): Align asan globals array by shadow
2593 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2596 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2597 (struct store_immediate_info): Add bitregion_start and bitregion_end
2599 (store_immediate_info::store_immediate_info): Add brs and bre
2600 arguments and initialize bitregion_{start,end} from those.
2601 (struct merged_store_group): Add bitregion_start, bitregion_end,
2602 align_base and mask fields. Drop unnecessary struct keyword from
2603 struct store_immediate_info. Add do_merge method.
2604 (clear_bit_region_be): Use memset instead of loop storing zeros.
2605 (merged_store_group::do_merge): New method.
2606 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2607 stores as long as the surrounding bitregions have no gaps.
2608 (merged_store_group::merge_overlapping): Use do_merge.
2609 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2610 is byte aligned, rather than requiring that start and width are
2611 byte aligned. Drop unnecessary struct keyword from
2612 struct store_immediate_info. Allocate and populate also mask array.
2613 Make start of the arrays relative to bitregion_start rather than
2614 start and size them according to bitregion_{end,start} difference.
2615 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2616 struct store_immediate_info.
2617 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2618 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2619 struct keyword from struct store_immediate_info.
2620 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2621 between stores as long as the surrounding bitregions have no gaps.
2623 (struct split_store): Add orig non-static data member.
2624 (split_store::split_store): Initialize orig to false.
2625 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2626 if there is exactly a single original stmt. Change stmts argument
2627 to pointer from reference, if NULL, don't push anything to it. Add
2628 first argument, use it to optimize skipping over orig stmts that
2629 are known to be before bitpos already. Simplify.
2630 (split_group): Return unsigned int count how many stores are or
2631 would be needed rather than a bool. Add allow_unaligned argument.
2632 Change split_stores argument from reference to pointer, if NULL,
2633 only do a dry run computing how many stores would be produced.
2634 Rewritten algorithm to use both alignment and misalign if
2635 !allow_unaligned and handle bitfield stores with gaps.
2636 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2637 from bitregion_start instead of start. Compute allow_unaligned
2638 here, if true, do 2 split_group dry runs to compute which one
2639 produces fewer stores and prefer aligned if equal. Punt if
2640 new count is bigger or equal than original before emitting any
2641 statements, rather than during that. Remove no longer needed
2642 new_ssa_names tracking. Replace num_stmts with
2643 split_stores.length (). Use 32-bit stack allocated entries
2644 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2645 if possible. Handle bitfields with gaps.
2646 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2647 Compute bitregion_{start,end} for the stores and construct
2648 store_immediate_info with that. Formatting fixes.
2650 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2653 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2654 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2656 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2658 * gimplify.c: Include tm_p.h.
2660 * common.opt (gcoff): Re-add as ignored option.
2661 (gcoff1, gcoff2, gcoff3): Likewise.
2663 * Makefile.in (OBJS): Delete sdbout.o.
2664 (GTFILES): Delete $(srcdir)/sdbout.c.
2665 * debug.h: Delete sdb_debug_hooks.
2666 * final.c: Delete sdbout.h include.
2667 (final_scan_insn): Delete SDB_DEBUG check.
2668 (rest_of_clean_state): Likewise.
2669 * output.h: Delete sdb_begin_function_line.
2672 * toplev.c: Delete sdbout.h include.
2673 (process_options): Delete SDB_DEBUG check.
2674 * tree-core.h (tree_type_common): Delete pointer field of
2676 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2677 TYPE_SYMTAB_POINTER.
2678 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2679 (TYPE_SYMTAB_IS_POINTER): Delete.
2680 (TYPE_SYMTAB_IS_DIE): Renumber.
2681 * xcoffout.c: Refer to former sdbout.c file.
2682 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2684 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2685 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2686 (Debugging Options): Delete -gcoff.
2687 (-gstabs): Delete SDB reference.
2689 (-gcoff@var{level}): Delete.
2690 * doc/passes.texi (Debugging information output): Delete SDB and
2691 sdbout.c references.
2692 * doc/tm.texi: Regenerate.
2693 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2694 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2696 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2697 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2698 (SDB_DEBUGGING_INFO): Delete.
2699 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2700 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2701 * target.def (output_source_filename): Delete COFF reference.
2703 * common.opt (gcoff): Delete.
2704 (gxcoff+): Update Negative chain.
2705 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2707 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2708 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2709 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2711 (expand_function_start): Change sdb reference to past tense.
2712 (expand_function_end): Change sdb reference to past tense.
2713 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2714 * opts.c (debug_type_names): Delete coff entry.
2715 (common_handle_option): Delete OPT_gcoff case.
2716 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2718 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2719 * config/cris/cris.h: Delete SDB reference in comment.
2720 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2721 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2722 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2723 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2725 (ix86_expand_prologue): Likewise.
2726 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2727 * config/ia64/ia64.h: Likewise.
2728 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2729 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2731 * config/mmix/mmix.h: Likewise.
2732 * config/nds32/nds32.c: Likewise.
2733 * config/stormy/storym16.h: Likewise.
2734 * config/visium/visium.h: Likewise.
2735 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2737 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2739 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2740 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2741 FRAME_POINTER_REGNUM point at high end of local var area.
2743 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2745 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2746 Move comment around. Do not reset best_edge for a copiable
2747 destination if the copy would cause a partition change.
2748 (better_edge_p): Remove redundant check.
2750 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2752 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2754 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2756 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2757 for math functions that have _Float<N> and _Float<N>X variants.
2758 (mathfn_built_in_2): Add support for math functions that have
2759 _Float<N> and _Float<N>X variants.
2760 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2761 (expand_builtin_mathfn_ternary): Add support for fma with
2762 _Float<N> and _Float<N>X variants.
2763 (expand_builtin): Likewise.
2764 (fold_builtin_3): Likewise.
2765 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2766 create math function _Float<N> and _Float<N>X variants as external
2768 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2769 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2770 the __builtin_ prefix and if not strict ansi, without the prefix.
2771 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2772 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2773 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2774 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2775 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2776 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2777 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2778 function signatures for fma _Float<N> and _Float<N>X variants.
2779 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2780 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2781 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2782 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2783 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2784 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2785 * gencfn-macros.c (print_case_cfn): Add support for math functions
2786 that have _Float<N> and _Float<N>X variants.
2787 (print_define_operator_list): Likewise.
2788 (fltfn_suffixes): Likewise.
2790 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2791 for math functions that have _Float<N> and _Float<N>X variants.
2792 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2793 and _Float<N>X variants.
2794 (COPYSIGN): Likewise.
2797 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2798 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2800 (integer_valued_read_call_p): Likewise.
2801 * fold-const-call.c (fold_const_call_ss): Likewise.
2802 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2803 _Float<N> and _Float<N>X variants.
2804 (fold_const_call_ssss): Add support for fma _Float<N> and
2805 _Float<N>X variants.
2806 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2807 support for copysign and fma _Float<N> and _Float<N>X variants.
2808 (backprop::process_builtin_call_use): Likewise.
2809 * tree-call-cdce.c (can_test_argument_range); Add support for
2810 sqrt _Float<N> and _Float<N>X variants.
2811 (edom_only_function): Likewise.
2812 (get_no_error_domain): Likewise.
2813 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2814 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2815 copysign _Float<N> and _Float<N>X variants.
2816 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2817 handled by machine independent code.
2818 (FMAF128): Likewise.
2819 * doc/cpp.texi (Common Predefined Macros): Document defining
2820 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2821 fma _Float<N> and _Float<N>X variants.
2823 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2826 * config/i386/i386-modes.def (CCFPU): Remove definition.
2827 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2828 (ix86_cc_modes_compatible): Ditto.
2829 (ix86_expand_carry_flag_compare): Ditto.
2830 (ix86_expand_int_movcc): Ditto.
2831 (ix86_expand_int_addcc): Ditto.
2832 (ix86_reverse_condition): Ditto.
2833 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2834 Return true/false for unordered/ordered fp comparisons.
2835 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2836 (ix86_prepare_fp_compare_args): Update for rename.
2837 (ix86_expand_fp_compare): Update for rename. Generate unordered
2838 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2839 (ix86_expand_sse_compare_and_jump): Ditto.
2840 * config/i386/predicates.md (fcmov_comparison_operator):
2841 Remove CCFPU mode handling.
2842 (ix86_comparison_operator): Ditto.
2843 (ix86_carry_flag_operator): Ditto.
2844 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2845 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2846 (*cmpu<mode>_cc_i387): Ditto.
2847 (FPCMP): Remove mode iterator.
2848 (unord): Remove mode attribute.
2849 (unord_subst): New define_subst transformation
2850 (unord): New define_subst attribute.
2852 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2853 (*cmpi<unord>xf_i387): Ditto.
2854 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2855 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2856 using unord_subst transformation.
2857 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2858 (round_saeonly): Also handle CCFP mode.
2859 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2860 Remove UNSPEC_SAHF unspec handling.
2862 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2864 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2866 2017-10-27 Jeff Law <law@redhat.com>
2868 * gimple-ssa-sprintf.c: Include domwalk.h.
2869 (class sprintf_dom_walker): New class, derived from dom_walker.
2870 (sprintf_dom_walker::before_dom_children): New function.
2871 (struct call_info): Moved into sprintf_dom_walker class
2872 (compute_formath_length, handle_gimple_call): Likewise.
2873 (sprintf_length::execute): Call the dominator walker rather
2874 than walking the statements.
2876 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2877 gimple statement locations.
2878 (check_array_bounds): Corresponding changes. Get the statement's
2879 location directly from wi->stmt.
2881 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2884 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2886 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2888 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2889 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2891 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2894 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2895 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2896 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2897 maybe_get_pool_constant.
2898 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2901 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2903 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2904 2.26 caveat. Update gas and gld versions.
2905 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2908 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2910 * cgraph.h (set_malloc_flag): Declare.
2911 * cgraph.c (set_malloc_flag_1): New function.
2912 (set_malloc_flag): Likewise.
2913 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2914 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2916 (read_ipa_call_summary): Add support for reading is_return_callee.
2917 (write_ipa_call_summary): Stream is_return_callee.
2918 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2919 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2920 ipa-prop.h, ipa-fnsummary.h.
2921 (pure_const_names): Change to static.
2922 (malloc_state_e): Define.
2923 (malloc_state_names): Define.
2924 (funct_state_d): Add field malloc_state.
2925 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2926 (check_retval_uses): New function.
2927 (malloc_candidate_p): Likewise.
2928 (analyze_function): Add support for malloc attribute.
2929 (pure_const_write_summary): Stream malloc_state.
2930 (pure_const_read_summary): Add support for reading malloc_state.
2931 (dump_malloc_lattice): New function.
2932 (propagate_malloc): New function.
2933 (warn_function_malloc): New function.
2934 (ipa_pure_const::execute): Call propagate_malloc and
2935 ipa_free_fn_summary.
2936 (pass_local_pure_const::execute): Add support for malloc attribute.
2937 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2938 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2940 2017-10-27 Martin Liska <mliska@suse.cz>
2942 PR gcov-profile/82457
2943 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2944 for fork-like functions to be properly instrumented.
2946 2017-10-27 Richard Biener <rguenther@suse.de>
2949 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2950 info when we redirected EH.
2952 2017-10-26 Michael Collison <michael.collison@arm.com>
2954 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2956 (<optab>_trunchf<GPI:mode>2: New pattern.
2957 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2958 * config/aarch64/iterators.md (wv): New mode attribute.
2959 (vf, VF): New mode attributes.
2960 (vgp, VGP): New mode attributes.
2961 (s): Update attribute with SImode and DImode prefixes.
2963 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2965 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2966 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2967 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2968 (nios2_option_overide): Initialize it. Don't allow R0-relative
2969 addressing with PIC.
2970 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2971 (nios2_symbolic_constant_p): Likewise.
2972 (nios2_legitimate_address_p): Likewise.
2973 (nios2_r0rel_section_name_p): New.
2974 (nios2_symbol_ref_in_r0rel_data_p): New.
2975 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2976 (r0rel_constant_p): New.
2977 (nios2_print_operand_address): Handle r0rel_constant_p.
2978 (nios2_cdx_narrow_form_p): Likewise.
2979 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2980 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2981 (Nios II Options): Document -mr0rel-sec.
2983 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2985 * config/nios2/nios2.c: Include xregex.h.
2986 (nios2_gprel_sec_regex): New.
2987 (nios2_option_overide): Initialize it. Don't allow GP-relative
2988 addressing with PIC.
2989 (nios2_small_section_name_p): Check for regex match.
2990 * config/nios2/nios2.opt (mgprel-sec=): New option.
2991 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2992 (Nios II Options): Document -mgprel-sec.
2994 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2996 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2998 2017-10-26 Tom de Vries <tom@codesourcery.com>
3000 PR tree-optimization/82707
3001 * gimple.c (gimple_copy): Fix unsharing of
3002 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3004 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3006 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3007 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3008 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3009 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3010 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3011 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3012 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3013 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3014 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3015 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3016 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3017 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3018 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3019 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3020 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3021 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3022 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3024 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3026 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3028 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3029 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3030 warning to rs6000.c. Remove the Undocumented flag, since it has
3032 (-mabi=ibmlongdouble): Likewise.
3033 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3034 already set, set the default format for long double.
3035 (rs6000_debug_reg_global): Print whether long double is IBM or
3037 (rs6000_option_override_internal): Rework setting long double
3038 format. Only warn if the user is changing the long double default
3039 and they did not use -Wno-psabi.
3040 * doc/invoke.texi (PowerPC options): Update the documentation for
3041 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3043 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3044 Alan Hayward <alan.hayward@arm.com>
3045 David Sherwood <david.sherwood@arm.com>
3047 * rtl.h (wider_subreg_mode): New function.
3048 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3049 rather than an unsigned int *.
3050 * ira-color.c (regno_max_ref_width): Replace with...
3051 (regno_max_ref_mode): ...this new variable.
3052 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3053 Use wider_subreg_mode.
3054 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3055 rather than an unsigned int *.
3056 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3057 (process_alt_operands): Likewise.
3058 (invariant_p): Likewise.
3059 * lra-spills.c (assign_mem_slot): Likewise.
3060 (add_pseudo_to_slot): Likewise.
3061 * lra.c (collect_non_operand_hard_regs): Likewise.
3062 (add_regs_to_insn_regno_info): Likewise.
3063 * reload1.c (regno_max_ref_width): Replace with...
3064 (regno_max_ref_mode): ...this new variable.
3065 (reload): Update accordingly. Update call to
3066 ira_sort_regnos_for_alter_reg.
3067 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3068 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3069 (scan_paradoxical_subregs): Likewise.
3071 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3073 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3074 (aarch64_frame): Add emit_frame_chain boolean.
3075 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3076 Move eh_return case to aarch64_layout_frame.
3077 (aarch64_layout_frame): Initialize emit_frame_chain.
3078 (aarch64_expand_prologue): Use emit_frame_chain.
3080 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3082 * config/aarch64/aarch64.c (aarch64_layout_frame):
3083 Ensure LR is always stored at the bottom of the callee-saves.
3084 Remove rarely used frame layout which saves callee-saves at top of
3085 frame, so the store of LR can be used as a valid probe in all cases.
3087 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3089 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3090 Improve unaligned TImode/TFmode base/offset split.
3092 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3093 Alan Hayward <alan.hayward@arm.com>
3094 David Sherwood <david.sherwood@arm.com>
3096 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3097 * combine.c (find_single_use_1): Likewise.
3098 (expand_field_assignment): Likewise.
3099 (move_deaths): Likewise.
3100 * lra-constraints.c (simplify_operand_subreg): Likewise.
3101 (curr_insn_transform): Likewise.
3102 * lra.c (collect_non_operand_hard_regs): Likewise.
3103 (add_regs_to_insn_regno_info): Likewise.
3104 * rtlanal.c (reg_referenced_p): Likewise.
3105 (covers_regno_no_parallel_p): Likewise.
3107 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3109 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3110 Don't print any bits outside the precision of the value.
3111 * wide-int.cc (test_printing): Add some new tests.
3113 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3115 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3116 supports -xbrace_comment option.
3117 * configure: Regenerate.
3118 * config.in: Regenerate.
3119 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3120 (ASM_CPU_SPEC): Use it.
3122 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3124 * target.def (static_rtx_alignment): New hook.
3125 * targhooks.h (default_static_rtx_alignment): Declare.
3126 * targhooks.c (default_static_rtx_alignment): New function.
3127 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3128 * doc/tm.texi: Regenerate.
3129 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3130 instead of targetm.constant_alignment. Remove call to
3132 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3133 (cris_preferred_mininum_alignment): New function, split out from...
3134 (cris_constant_alignment): ...here.
3135 (cris_static_rtx_alignment): New function.
3136 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3138 (ix86_constant_alignment): ...here.
3139 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3140 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3141 (mmix_static_rtx_alignment): New function.
3142 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3143 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3145 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3148 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3149 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3151 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3153 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3154 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3155 mark as a sign-extending load.
3156 (local_pic_load_u): Define.
3158 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3161 * fold-const.c (operand_equal_for_comparison_p): Also return true
3162 if ARG0 is a simple variant of ARG1 with narrower precision.
3163 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3165 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3167 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3169 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3170 scatter_static, scatter_per_elt.
3171 * x86-tune-costs.h: Add new cost entries.
3173 2017-10-25 Richard Biener <rguenther@suse.de>
3175 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3176 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3177 class pass_fre): Move to ...
3178 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3179 class pass_fre): ... here and adjust for statistics.
3181 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3184 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3185 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3186 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3187 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3190 2017-10-25 Richard Biener <rguenther@suse.de>
3192 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3193 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3194 eliminate_push_avail, eliminate_insert): Move inside...
3195 (class eliminate_dom_walker): ... this class in preparation
3197 (fini_eliminate): Remove by merging with ...
3198 (eliminate): ... this function. Adjust for class changes.
3199 (pass_pre::execute): Remove fini_eliminate call.
3200 (pass_fre::execute): Likewise.
3202 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3205 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3206 (VPERMI2, VPERMI2I): New mode iterators.
3207 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3208 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3210 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3211 mode iterator. Remove 3 old define_insn patterns.
3212 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3213 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3214 VPERMI2 mode iterator, remove the other two expanders.
3215 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3216 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3217 instructions, remove the other two patterns.
3218 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3219 mode iterator, remove the other two patterns.
3220 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3221 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3222 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3223 and adjust argument order accordingly.
3224 (ix86_expand_vec_perm): Adjust caller.
3225 (expand_vec_perm_1): Likewise.
3226 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3227 (expand_vec_perm_vpermt2_vpshub2): ... this.
3228 (ix86_expand_vec_perm_const_1): Adjust caller.
3229 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3232 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3233 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3234 (vec_shl_<mode>): Remove unused expander.
3235 (avx512bw_<shift_insn><mode>3): New define_insn.
3236 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3237 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3239 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3242 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3245 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3247 PR rtl-optimization/82396
3248 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3249 (autopref_multipass_init): Simplify initialization.
3250 (autopref_rank_data): Simplify sort order.
3251 * gcc/sched-int.h (autopref_multipass_data_): Remove
3252 multi_mem_insn_p, min_offset and max_offset.
3254 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3257 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3258 Check special value of flag_omit_frame_pointer.
3259 (aarch64_can_eliminate): Likewise.
3260 (aarch64_override_options_after_change_1): Simplify handling of
3261 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3263 2017-10-24 Richard Biener <rguenther@suse.de>
3265 PR tree-optimization/82697
3266 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3267 zero for conditional load and unconditional store.
3269 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3271 * doc/install.texi: Document bootstrap-cet.
3273 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3276 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3277 ENDBR instruction at function entrance if function is only
3280 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3283 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3284 patterns to better describe from which operation the CF is computed.
3285 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3286 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3287 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3288 is 0, use _0 suffixed expanders instead of emitting a comparison
3291 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3293 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3294 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3296 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3299 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3300 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3301 * loop-iv.c (iv_get_reaching_def): Likewise.
3302 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3303 variable is promoted and the partition contains undefined values.
3305 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3307 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3309 (nios2_address_cost): Define.
3310 (nios2_legitimize_address): Recognize (exp + constant) directly.
3311 (TARGET_ADDRESS_COST): Define.
3313 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3315 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3316 (nios2_symbolic_memory_operand_p): Declare.
3317 (nios2_split_large_constant): Declare.
3318 (nios2_split_symbolic_memory_operand): Declare.
3319 * config/nios2/nios2.c: Adjust includes.
3320 (nios2_symbolic_constant_allowed): New.
3321 (nios2_symbolic_constant_p): New.
3322 (nios2_plus_symbolic_constant_p): New.
3323 (nios2_valid_addr_expr_p): Recognize addresses involving
3325 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3326 (nios2_symbolic_memory_operand_p): New.
3327 (nios2_large_constant_p): New.
3328 (nios2_split_large_constant): New.
3329 (nios2_split_plus_large_constant): New.
3330 (nios2_split_symbolic_memory_operand): New.
3331 (nios2_legitimize_address): Code refactoring. Handle addresses
3332 involving symbolic constants.
3333 (nios2_emit_move_sequence): Likewise.
3334 (nios2_print_operand): Improve error output.
3335 (nios2_print_operand_address): Handle LO_SUM.
3336 (nios2_cdx_narrow_form_p): Likewise.
3337 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3338 operands involving symbolic constants.
3339 (movhi_internal, movsi_internal): Likewise.
3340 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3341 (extendhisi2, extendqi<mode>2): Likewise.
3343 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3345 * tree-pass.h (PROP_rtl_split_insns): Define.
3346 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3348 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3350 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3352 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3355 * target.def (const_not_ok_for_debug_p): Default to
3356 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3357 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3358 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3359 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3360 which targetm.const_not_ok_for_debug_p returned true.
3361 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3363 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3365 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3366 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3367 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3369 (ix86_const_not_ok_for_debug_p): New function.
3370 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3371 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3373 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3376 * system.h: Conditionally include "unique-ptr.h" if
3377 INCLUDE_UNIQUE_PTR is defined.
3378 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3379 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3381 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3383 * config/rl78/rl78.md: New define_expand "subdi3".
3385 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3388 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3389 DF_REF_INSN if DF_REF_INSN_INFO is false.
3391 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3393 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3394 xmm_move instead of sse_move.
3395 (sse_store_index): New function.
3396 (ix86_register_move_cost): Be more sensible about mismatch stall;
3397 model AVX moves correctly; make difference between sse->integer and
3399 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3400 moves; make difference between SSE and AVX.
3401 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3402 and zmm_move. Increase size of sse load and store tables;
3403 add unaligned load and store tables; add ssemmx_to_integer.
3404 * x86-tune-costs.h: Update all entries according to real
3405 move latencies from Agner Fog's manual and chip documentation.
3407 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3410 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3411 * config/i386/constraints.md (Wf): New constraint.
3412 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3413 (cmp<dwi>_doubleword): Removed.
3414 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3415 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3416 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3417 expand with cmp<dwi>_doubleword. For LTU and GEU use
3418 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3420 * common.opt (gcolumn-info): Enable by default.
3421 * doc/invoke.texi (gcolumn-info): Document new default.
3423 2017-10-23 Richard Biener <rguenther@suse.de>
3425 PR tree-optimization/82672
3426 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3427 Fold the stmt if we propagated into it.
3429 2017-10-23 Richard Biener <rguenther@suse.de>
3431 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3432 (bitmap_remove_expr_from_set): ... this. All callers call this
3433 for non-constant values.
3434 (bitmap_set_subtract): Rename to...
3435 (bitmap_set_subtract_expressions): ... this. Adjust and
3437 (bitmap_set_contains_value): Remove superfluous check.
3438 (bitmap_set_replace_value): Inline into single caller ...
3439 (bitmap_value_replace_in_set): ... here and simplify.
3440 (dependent_clean): Merge into ...
3441 (clean): ... this using an overload. Adjust.
3442 (prune_clobbered_mems): Adjust.
3443 (compute_antic_aux): Likewise.
3444 (compute_partial_antic_aux): Likewise.
3446 2017-10-23 Richard Biener <rguenther@suse.de>
3448 PR tree-optimization/82129
3450 2017-08-01 Richard Biener <rguenther@suse.de>
3452 PR tree-optimization/81181
3453 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3454 (compute_antic): ... end of iteration here.
3456 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3458 * target.def (starting_frame_offset): New hook.
3459 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3460 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3461 * doc/tm.texi.in: Regenerate.
3462 * hooks.h (hook_hwi_void_0): Declare.
3463 * hooks.c (hook_hwi_void_0): New function.
3464 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3465 STARTING_FRAME_OFFSET.
3466 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3467 * reload1.c (reload): Likewise.
3468 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3469 instead of STARTING_FRAME_OFFSET.
3470 * function.c (try_fit_stack_local): Likewise.
3471 (assign_stack_local_1): Likewise
3472 (instantiate_virtual_regs): Likewise.
3473 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3474 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3475 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3476 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3477 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3478 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3479 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3480 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3481 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3482 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3483 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3484 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3485 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3486 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3487 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3488 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3489 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3490 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3491 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3492 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3493 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3494 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3495 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3496 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3497 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3498 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3499 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3500 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3501 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3502 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3503 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3504 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3505 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3506 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3507 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3508 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3509 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3510 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3511 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3512 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3513 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3514 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3515 return a HOST_WIDE_INT.
3516 (avr_builtin_setjmp_frame_value): Use it instead of
3517 STARTING_FRAME_OFFSET.
3518 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3519 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3520 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3522 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3523 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3524 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3525 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3526 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3527 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3528 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3529 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3530 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3531 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3532 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3533 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3535 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3536 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3537 * config/mips/mips.c (mips_compute_frame_info): Refer to
3538 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3539 (mips_starting_frame_offset): New function.
3540 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3541 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3542 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3543 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3544 and return a HOST_WIDE_INT.
3545 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3546 (mmix_initial_elimination_offset): Refer to
3547 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3548 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3549 * config/pa/pa.c (pa_starting_frame_offset): New function.
3550 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3551 (pa_expand_prologue): Likewise.
3552 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3553 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3554 !FRAME_GROWS_DOWNWARD handling to...
3555 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3556 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3557 !FRAME_GROWS_DOWNWARD handling to...
3558 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3559 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3560 !FRAME_GROWS_DOWNWARD handling to...
3561 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3562 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3564 (rs6000_starting_frame_offset): New function.
3565 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3566 !FRAME_GROWS_DOWNWARD handling to...
3567 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3568 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3569 !FRAME_GROWS_DOWNWARD handling to...
3570 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3571 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3572 !FRAME_GROWS_DOWNWARD handling to...
3573 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3574 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3575 (rs6000_starting_frame_offset): New function.
3576 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3577 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3578 * config/vax/vax.c (vax_starting_frame_offset): New function.
3579 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3580 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3581 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3582 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3583 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3584 * system.h (STARTING_FRAME_OFFSET): Poison.
3586 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3588 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3589 SCALAR_TYPE_MODE instead of TYPE_MODE.
3591 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3592 Alan Hayward <alan.hayward@arm.com>
3593 David Sherwood <david.sherwood@arm.com>
3595 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3597 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3598 Alan Hayward <alan.hayward@arm.com>
3599 David Sherwood <david.sherwood@arm.com>
3601 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3603 2017-10-23 Richard Biener <rguenther@suse.de>
3605 PR tree-optimization/82129
3606 * tree-ssa-pre.c (bitmap_set_and): Remove.
3607 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3608 canonicalizing expressions in the set to those with lowest
3609 ID rather than taking that from the first edge.
3611 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3613 * combine.c (rtx_equal_for_field_assignment_p): Use
3614 byte_lowpart_offset.
3616 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3617 Alan Hayward <alan.hayward@arm.com>
3618 David Sherwood <david.sherwood@arm.com>
3620 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3621 to a SUBREG_PROMOTED_VAR.
3623 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3624 Alan Hayward <alan.hayward@arm.com>
3625 David Sherwood <david.sherwood@arm.com>
3627 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3628 (expand_debug_source_expr): Likewise.
3629 * combine.c (combine_simplify_rtx): Likewise.
3630 * cse.c (fold_rtx): Likewise.
3631 * optabs.c (expand_float): Likewise.
3632 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3633 (simplify_binary_operation_1): Likewise.
3635 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3636 Alan Hayward <alan.hayward@arm.com>
3637 David Sherwood <david.sherwood@arm.com>
3639 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3640 (record_promoted_value): Likewise.
3641 * expr.c (expand_expr_real_2): Likewise.
3642 * ree.c (update_reg_equal_equiv_notes): Likewise.
3643 (combine_set_extension): Likewise.
3644 * rtlanal.c (low_bitmask_len): Likewise.
3645 * simplify-rtx.c (neg_const_int): Likewise.
3646 (simplify_binary_operation_1): Likewise.
3648 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3649 Alan Hayward <alan.hayward@arm.com>
3650 David Sherwood <david.sherwood@arm.com>
3652 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3653 * regcprop.c (maybe_mode_change): Likewise.
3654 * reload1.c (alter_reg): Likewise.
3656 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3658 * inchash.h (inchash::hash::add_wide_int): New function.
3659 * lto-streamer-out.c (hash_tree): Use it.
3661 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3663 * inchash.h (inchash::hash::add_wide_int): Rename to...
3664 (inchash::hash::add_hwi): ...this.
3665 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3666 (polymorphic_call_target_hasher::hash): Likewise.
3667 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3668 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3669 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3670 * lto-streamer-out.c (hash_tree): Likewise.
3671 * optc-save-gen.awk: Likewise.
3672 * tree.c (add_expr): Likewise.
3674 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3677 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3678 for ordered inequality comparisons even with TARGET_IEEE_FP.
3680 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3683 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3684 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3685 Expand with cmp<dwi>_doubleword.
3687 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3689 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3690 List CET intrinsics.
3691 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3692 specific to -fcf-protection option.
3694 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3696 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3697 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3698 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3699 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3700 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3701 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3702 (extra_objs): Add cet.o for Linux/x86 targets.
3703 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3704 * config/i386/cet.c: New file.
3705 * config/i386/cetintrin.h: Likewise.
3706 * config/i386/t-cet: Likewise.
3707 * config/i386/cpuid.h (bit_SHSTK): New.
3708 (bit_IBT): Likewise.
3709 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3710 pass IBT and SHSTK bits.
3711 * config/i386/i386-builtin-types.def
3712 (VOID_FTYPE_UNSIGNED_PVOID): New.
3713 (VOID_FTYPE_UINT64_PVOID): Likewise.
3714 * config/i386/i386-builtin.def: Add CET intrinsics.
3715 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3716 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3717 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3718 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3720 * config/i386/i386.c (rest_of_insert_endbranch): New.
3721 (pass_data_insert_endbranch): Likewise.
3722 (pass_insert_endbranch): Likewise.
3723 (make_pass_insert_endbranch): Likewise.
3724 (ix86_notrack_prefixed_insn_p): Likewise.
3725 (ix86_target_string): Add -mibt, -mshstk flags.
3726 (ix86_option_override_internal): Add flag_cf_protection
3728 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3729 (ix86_print_operand): Add 'notrack' prefix output.
3730 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3731 (ix86_expand_builtin): Expand CET intrinsics.
3732 (x86_output_mi_thunk): Add 'endbranch' instruction.
3733 * config/i386/i386.h (TARGET_IBT): New.
3734 (TARGET_IBT_P): Likewise.
3735 (TARGET_SHSTK): Likewise.
3736 (TARGET_SHSTK_P): Likewise.
3737 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3738 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3739 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3740 (builtin_setjmp_setup): New pattern.
3741 (builtin_longjmp): Likewise.
3742 (rdssp<mode>): Likewise.
3743 (incssp<mode>): Likewise.
3744 (saveprevssp): Likewise.
3745 (rstorssp): Likewise.
3746 (wrss<mode>): Likewise.
3747 (wruss<mode>): Likewise.
3748 (setssbsy): Likewise.
3749 (clrssbsy): Likewise.
3750 (nop_endbr): Likewise.
3751 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3753 * config/i386/immintrin.h: Include <cetintrin.h>.
3754 * config/i386/linux-common.h
3755 (file_end_indicate_exec_stack_and_cet): New prototype.
3756 (TARGET_ASM_FILE_END): New.
3758 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3760 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3761 latencies instead of having separate table; make difference between
3762 integer and float costs.
3763 * i386.h (processor_costs): Remove scalar_stmt_cost,
3764 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3765 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3767 * x86-tune-costs.h: Remove entries which has been removed in
3768 procesor_costs from all tables; make cond_taken_branch_cost
3769 and cond_not_taken_branch_cost COST_N_INSNS based.
3771 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3773 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3775 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3777 * config/i386/i386.md (isa): Remove fma_avx512f.
3778 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3779 <avx512>_fmadd_<mode>_mask3<round_name>,
3780 <avx512>_fmsub_<mode>_mask<round_name>,
3781 <avx512>_fmsub_<mode>_mask3<round_name>,
3782 <avx512>_fnmadd_<mode>_mask<round_name>,
3783 <avx512>_fnmadd_<mode>_mask3<round_name>,
3784 <avx512>_fnmsub_<mode>_mask<round_name>,
3785 <avx512>_fnmsub_<mode>_mask3<round_name>,
3786 <avx512>_fmaddsub_<mode>_mask<round_name>,
3787 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3788 <avx512>_fmsubadd_<mode>_mask<round_name>,
3789 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3790 (*vec_widen_umult_even_v16si<mask_name>,
3791 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3792 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3794 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3796 * extend.texi: Add 'nocf_check' documentation.
3797 * gimple.texi: Add second parameter to
3798 gimple_build_call_from_tree.
3799 * invoke.texi: Add -fcf-protection documentation.
3800 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3802 2017-10-20 Richard Biener <rguenther@suse.de>
3804 PR tree-optimization/82473
3805 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3806 the largest input type.
3808 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3810 * c-attribs.c (handle_nocf_check_attribute): New function.
3811 (c_common_attribute_table): Add 'nocf_check' handling.
3812 * gimple-parser.c: Add second argument NULL to
3813 gimple_build_call_from_tree.
3814 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3815 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3817 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3818 * common.opt: Add fcf-protection flag.
3819 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3820 * flag-types.h: Add enum cf_protection_level.
3821 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3822 Add 'nocf_check' attribute propagation to gimple call.
3823 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3824 (gimple_build_call_from_tree): Update prototype.
3825 (gimple_call_nocf_check_p): New function.
3826 (gimple_call_set_nocf_check): Likewise.
3827 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3828 * ipa-icf.c: Add nocf_check attribute in statement hash.
3829 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3830 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3831 * toplev.c (process_options): Add flag_cf_protection handling.
3833 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3835 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3837 2017-10-20 Richard Biener <rguenther@suse.de>
3839 PR tree-optimization/82603
3840 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3841 remove false predicated stores.
3843 2017-10-20 Richard Biener <rguenther@suse.de>
3845 * graphite-isl-ast-to-gimple.c
3846 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3847 Remove return value and simplify, dump copied stmt after lhs
3849 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3850 Reduce dump verbosity.
3851 (gsi_insert_earliest): Likewise.
3852 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3853 * graphite.c (print_global_statistics): Adjust dumping.
3854 (print_graphite_scop_statistics): Likewise.
3855 (print_graphite_statistics): Do not dump loops here.
3856 (graphite_transform_loops): But here.
3858 2017-10-20 Nicolas Roche <roche@adacore.com>
3860 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3861 * configure: Regenerate.
3863 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3866 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3867 functions when optimizing replace GIMPLE_RETURN stmts with
3868 calls to __builtin_unreachable ().
3871 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3872 for -fsanitize=thread link of executables.
3873 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3874 link of executables.
3877 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3879 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3880 define_insns for logical vector shifts to use VI248_AVX512BW
3881 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3882 condition, useless isa and prefix attributes. Change the first
3883 2 of these define_insns to ...
3884 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3885 define_insn for avx512vl.
3886 (<shift_insn><mode>3): ... and this, new define_insn without
3887 masking for non-avx512vl.
3890 * config/i386/sse.md (*andnot<mode>3,
3891 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3892 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3893 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3894 not applied use empty suffix even for TARGET_AVX512VL.
3895 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3896 is applied, supply evex,evex or evex,evex,evex instead of just
3899 2017-10-20 Julia Koval <julia.koval@intel.com>
3901 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3902 (OPTION_MASK_ISA_GFNI_UNSET): New.
3903 (ix86_handle_option): Handle OPT_mgfni.
3904 * config/i386/cpuid.h (bit_GFNI): New.
3905 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3906 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3907 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3908 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3909 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3910 * config/i386/i386.opt: Add mgfni.
3912 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3914 * config/msp430/msp430.c (msp430_option_override): Disable
3915 -fdelete-null-pointer-checks.
3916 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3918 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3920 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3921 of x87 and SSE instructions.
3923 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3925 * asan.c (create_cond_insert_point): Do not update edge count.
3926 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3927 (afdo_propagate_circuit): Likewise.
3928 (afdo_calculate_branch_prob): Likewise.
3929 (afdo_annotate_cfg): Likewise.
3930 * basic-block.h (struct edge_def): Remove count.
3931 (edge_def::count): New accessor.
3932 * bb-reorder.c (rotate_loop): Update.
3933 (find_traces_1_round): Update.
3934 (connect_traces): Update.
3935 (sanitize_hot_paths): Update.
3936 * cfg.c (unchecked_make_edge): Update.
3937 (make_single_succ_edge): Update.
3938 (check_bb_profile): Update.
3939 (dump_edge_info): Update.
3940 (update_bb_profile_for_threading): Update.
3941 (scale_bbs_frequencies_int): Update.
3942 (scale_bbs_frequencies_gcov_type): Update.
3943 (scale_bbs_frequencies_profile_count): Update.
3944 (scale_bbs_frequencies): Update.
3945 * cfganal.c (connect_infinite_loops_to_exit): Update.
3946 * cfgbuild.c (compute_outgoing_frequencies): Update.
3947 (find_many_sub_basic_blocks): Update.
3948 * cfgcleanup.c (try_forward_edges): Update.
3949 (try_crossjump_to_edge): Update
3950 * cfgexpand.c (expand_gimple_cond): Update
3951 (expand_gimple_tailcall): Update
3952 (construct_exit_block): Update
3953 * cfghooks.c (verify_flow_info): Update
3954 (redirect_edge_succ_nodup): Update
3955 (split_edge): Update
3956 (make_forwarder_block): Update
3957 (duplicate_block): Update
3958 (account_profile_record): Update
3959 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3960 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3961 * cfgloopmanip.c (scale_loop_profile): Update.
3963 (lv_adjust_loop_entry_edge): Update.
3964 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3965 (force_nonfallthru_and_redirect): Update.
3966 (purge_dead_edges): Update.
3967 (rtl_flow_call_edges_add): Update.
3968 * cgraphunit.c (init_lowered_empty_function): Update.
3969 (cgraph_node::expand_thunk): Update.
3970 * gimple-pretty-print.c (dump_probability): Update.
3971 (dump_edge_probability): Update.
3972 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3973 * haifa-sched.c (sched_create_recovery_edges): Update.
3974 * hsa-gen.c (convert_switch_statements): Update.
3975 * ifcvt.c (dead_or_predicable): Update.
3976 * ipa-inline-transform.c (inline_transform): Update.
3977 * ipa-split.c (split_function): Update.
3978 * ipa-utils.c (ipa_merge_profiles): Update.
3979 * loop-doloop.c (add_test): Update.
3980 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3981 * lto-streamer-in.c (input_cfg): Update.
3982 (input_function): Update.
3983 * lto-streamer-out.c (output_cfg): Update.
3984 * modulo-sched.c (sms_schedule): Update.
3985 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3986 * predict.c (maybe_hot_edge_p): Update.
3987 (unlikely_executed_edge_p): Update.
3988 (probably_never_executed_edge_p): Update.
3989 (dump_prediction): Update.
3990 (drop_profile): Update.
3991 (propagate_unlikely_bbs_forward): Update.
3992 (determine_unlikely_bbs): Update.
3993 (force_edge_cold): Update.
3994 * profile.c (compute_branch_probabilities): Update.
3995 * reg-stack.c (better_edge): Update.
3996 * shrink-wrap.c (handle_simple_exit): Update.
3997 * tracer.c (better_p): Update.
3998 * trans-mem.c (expand_transaction): Update.
3999 (split_bb_make_tm_edge): Update.
4000 * tree-call-cdce.c: Update.
4001 * tree-cfg.c (gimple_find_sub_bbs): Update.
4002 (gimple_split_edge): Update.
4003 (gimple_duplicate_sese_region): Update.
4004 (gimple_duplicate_sese_tail): Update.
4005 (gimple_flow_call_edges_add): Update.
4006 (insert_cond_bb): Update.
4007 (execute_fixup_cfg): Update.
4008 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4009 * tree-complex.c (expand_complex_div_wide): Update.
4010 * tree-eh.c (lower_resx): Update.
4011 (unsplit_eh): Update.
4012 (cleanup_empty_eh_move_lp): Update.
4013 * tree-inline.c (copy_edges_for_bb): Update.
4014 (freqs_to_counts): Update.
4015 (copy_cfg_body): Update.
4016 * tree-ssa-dce.c (remove_dead_stmt): Update.
4017 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4018 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4019 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4020 (unloop_loops): Update.
4021 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4022 * tree-ssa-loop-split.c (connect_loops): Update.
4023 (split_loop): Update.
4024 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4025 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4026 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4027 * tree-ssa-reassoc.c (branch_fixup): Update.
4028 * tree-ssa-tail-merge.c (replace_block_by): Update.
4029 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4030 (compute_path_counts): Update.
4031 (update_profile): Update.
4032 (recompute_probabilities): Update.
4033 (update_joiner_offpath_counts): Update.
4034 (estimated_freqs_path): Update.
4035 (freqs_to_counts_path): Update.
4036 (clear_counts_path): Update.
4037 (ssa_fix_duplicate_block_edges): Update.
4038 (duplicate_thread_path): Update.
4039 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4040 (case_bit_test_cmp): Update.
4041 (collect_switch_conv_info): Update.
4042 (gen_inbound_check): Update.
4043 (do_jump_if_equal): Update.
4044 (emit_cmp_and_jump_insns): Update.
4045 * tree-tailcall.c (decrease_profile): Update.
4046 (eliminate_tail_call): Update.
4047 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4048 (vect_do_peeling): Update.
4049 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4050 * ubsan.c (ubsan_expand_null_ifn): Update.
4051 (ubsan_expand_ptr_ifn): Update.
4052 * value-prof.c (gimple_divmod_fixed_value): Update.
4053 (gimple_mod_pow2): Update.
4054 (gimple_mod_subtract): Update.
4055 (gimple_ic): Update.
4056 (gimple_stringop_fixed_value): Update.
4058 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4061 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4063 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4065 PR rtl-optimization/82395
4066 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4067 based on non_spilled_static_chain_regno_p.
4069 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4071 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4072 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4073 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4075 2017-10-19 Martin Sebor <msebor@redhat.com>
4077 PR tree-optimization/82596
4078 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4080 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4082 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4083 (get_mem_refs_of_builtin_call): Likewise.
4084 * builtins.c (expand_builtin_apply): Adjust call to
4085 allocate_dynamic_stack_space.
4086 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4087 the third argument to allocate_dynamic_stack_space, otherwise -1.
4088 (expand_builtin): Deal with all alloca variants.
4089 (is_inexpensive_builtin): Likewise.
4090 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4091 * calls.c (special_function_p): Deal with all alloca variants.
4092 (initialize_argument_information): Adjust call to
4093 allocate_dynamic_stack_space.
4094 (expand_call): Likewise.
4095 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4096 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4097 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4098 use it for the stack usage computation.
4099 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4100 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4101 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4102 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4103 (in_loop_p): Remove first argument and useless check.
4104 (pass_walloca::execute): Remove useless test and adjust call to above.
4105 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4106 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4107 (gimplify_call_expr): Deal with all alloca variants.
4108 * hsa-gen.c (gen_hsa_alloca): Likewise.
4109 (gen_hsa_insns_for_call): Likewise.
4110 * ipa-pure-const.c (special_builtin_state): Likewise.
4111 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4112 * tree-object-size.c (alloc_object_size): Likewise.
4113 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4114 (call_may_clobber_ref_p_1): Likewise.
4115 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4116 (ccp_fold_stmt): Likewise.
4117 (optimize_stack_restore): Likewise.
4118 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4119 (mark_all_reaching_defs_necessary_1): Likewise.
4120 (propagate_necessity): Likewise.
4121 (eliminate_unnecessary_stmts): Likewise.
4122 * tree.c (build_common_builtin_nodes): Build
4123 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4124 (build_alloca_call_expr): New function.
4125 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4126 (CASE_BUILT_IN_ALLOCA): Likewise.
4127 (build_alloca_call_expr): Declare.
4128 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4130 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4133 * dwarf2out.c (new_die_raw): New static inline function.
4134 (new_die): Use it to create the DIE.
4135 (add_AT_external_die_ref): Likewise.
4136 (clone_die): Likewise.
4137 (clone_as_declaration): Likewise.
4138 (dwarf2out_vms_debug_main_pointer): Likewise.
4139 (base_type_die): Likewise. Remove early return for corner cases.
4140 Do not call add_pubtype on the DIE here.
4141 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4142 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4143 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4144 native order exists for base types, attach the DIE manually and call
4145 add_pubtype on it. Do not equate a reverse order DIE to the type.
4147 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4149 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4150 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4152 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4154 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4156 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4157 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4158 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4159 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4160 * opts.c (sanitizer_opts): Add builtin.
4161 * ubsan.c (instrument_builtin): New function.
4162 (pass_ubsan::execute): Call it.
4163 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4164 * doc/invoke.texi: Document -fsanitize=builtin.
4166 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4167 builtins, store max (log2 (align), 0) into uchar field instead of
4168 align into uptr field.
4169 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4170 store uchar 0 field instead of uptr 0 field.
4171 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4172 instead of passing one address of struct with 2 locations pass
4173 two addresses of structs with 1 location each.
4174 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4175 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4176 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4177 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4178 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4179 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4180 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4181 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4183 2017-10-19 Martin Liska <mliska@suse.cz>
4186 * file-find.c (remove_prefix): Remove.
4187 * file-find.h (remove_prefix): Likewise.
4188 * gcc-ar.c: Remove smartness of lookup.
4190 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4192 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4193 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4194 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4196 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4199 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4200 (setcc + and to xor + setcc): New peephole2.
4202 2017-10-19 Tom de Vries <tom@codesourcery.com>
4204 * doc/sourcebuild.texi (Test Directives, Variants of
4205 dg-require-support): Add dg-require-stack-size.
4207 2017-10-19 Martin Liska <mliska@suse.cz>
4210 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4211 that have a large alignment.
4212 (gimplify_target_expr): Likewise.
4214 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4216 PR rtl-optimization/82602
4217 * ira.c (rtx_moveable_p): Return false for volatile asm.
4219 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4222 * config/i386/i386-modes.def (CCGZ): New CC mode.
4223 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4224 * config/i386/predicates.md (ix86_comparison_operator):
4226 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4227 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4228 with double-word subtraction.
4229 (put_condition_code): Handle CCGZmode.
4231 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4233 * wide-int.cc (debug (const wide_int &)): New.
4234 (debug (const wide_int *)): New.
4235 (debug (const widest_int &)): New.
4236 (debug (const widest_int *)): New.
4238 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4241 * lra-constraints.c (curr_insn_transform): Use non-input operand
4242 instead of output one for matched reload.
4244 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4246 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4247 (tree-ssa-loop-ivopts.h): New header file.
4248 (struct builtin_info): New fields.
4249 (classify_builtin_1): Compute and record base and offset parts for
4250 memset builtin partition by calling strip_offset.
4251 (offset_cmp, fuse_memset_builtins): New functions.
4252 (finalize_partitions): Fuse adjacent memset partitions by calling
4254 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4255 Expose the interface.
4256 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4258 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4260 PR tree-optimization/82574
4261 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4262 that data reference must be executed exactly once per iteration
4263 against the outermost loop in nest.
4264 (classify_partition): Update call to above function.
4266 2017-10-18 Richard Biener <rguenther@suse.de>
4268 PR tree-optimization/82591
4269 * graphite.c (graphite_transform_loops): Move code gen message
4271 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4272 Here. Handle scop_to_isl_ast failing.
4273 (scop_to_isl_ast): Limit the number of ISL operations.
4275 2017-10-18 Richard Biener <rguenther@suse.de>
4277 * graphite-isl-ast-to-gimple.c
4278 (translate_isl_ast_to_gimple::set_rename): Simplify.
4279 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4280 (graphite_copy_stmts_from_block): ... here.
4281 (copy_bb_and_scalar_dependences): Simplify.
4282 (add_parameters_to_ivs_params): Canonicalize.
4283 (generate_entry_out_of_ssa_copies): Simplify.
4284 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4285 by passing in ISL dimension.
4286 (parameter_index_in_region_1): Rename to ...
4287 (parameter_index_in_region): ... this.
4288 (extract_affine): Adjust assert, pass down parameter index.
4289 (add_param_constraints): Use range-info when available.
4290 (build_scop_context): Adjust.
4291 * sese.c (new_sese_info): Adjust.
4292 (free_sese_info): Likewise.
4293 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4294 Remove unused typedefs.
4295 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4297 2017-10-18 Martin Liska <mliska@suse.cz>
4299 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4301 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4303 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4304 (s390_sched_init): Do not reset s390_sched_state if we entered the
4305 current basic block via a fallthru edge and all others are unlikely.
4307 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4309 * config/s390/s390.c (NUM_SIDES): New variable.
4310 (LONGRUNNING_THRESHOLD): New variable.
4311 (LATENCY_FACTOR): New variable.
4312 (s390_sched_score): Decrease score for long-running instructions on
4314 (s390_sched_variable_issue): Perform bookkeeping for long-running
4317 2017-10-18 Richard Biener <rguenther@suse.de>
4319 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4320 Simplify with removal of the parameter rename map.
4321 (set_rename): Likewise.
4322 (should_copy_to_new_region): Likewise.
4323 (graphite_copy_stmts_from_block): Likewise.
4324 (copy_bb_and_scalar_dependences): Remove initialization of
4325 unused copied_bb_map.
4327 (copy_internal_parameters): Likewise.
4328 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4329 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4330 Use INTEGRAL_TYPE_P.
4331 (parameter_index_in_region_1): Rename to ...
4332 (assign_parameter_index_in_region): ... this. Assert we have
4333 a parameter we handle.
4334 (scan_tree_for_params): Adjust.
4335 * sese.h (parameter_rename_map_t): Remove.
4336 (struct sese_info_t): Remove unused parameter_rename_map and
4337 copied_bb_map members.
4338 * sese.c (new_sese_info): Adjust.
4339 (free_sese_info): Likewise.
4341 2017-10-18 Martin Liska <mliska@suse.cz>
4344 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4345 on an abnormal edge.
4347 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4349 * doc/invoke.texi (ffunction-sections and fdata-sections):
4352 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4354 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4355 the use statement can throw internally.
4357 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4359 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4360 any RTX present on the RHS of a SET.
4361 * compare-elim.c (try_eliminate_compare): Restore comment.
4363 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4365 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4366 may be also called on tcc_type nodes.
4367 * langhooks.c (lhd_tree_size): Likewise.
4369 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4371 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4372 format_warning_at_substring.
4373 (maybe_warn): Convert source_range * param to a location_t. Pass
4374 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4375 (format_directive): Remove code to extract source_ranges and
4376 source_range * in favor of just a location_t.
4377 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4379 * substring-locations.c (format_warning_va): Convert
4380 source_range * param to a location_t.
4381 (format_warning_at_substring): Likewise.
4382 * substring-locations.h (format_warning_va): Likewise.
4383 (format_warning_at_substring): Likewise.
4385 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4387 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4389 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4390 and scatter/gather ops.
4392 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4393 vec_gather_load and vec_scatter_store.
4394 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4395 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4396 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4397 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4398 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4399 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4401 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4403 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4404 Detect FTST instruction and handle its register pops. Only pop
4405 second operand if can_pop_second_op is true.
4406 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4407 set can_pop_second_op to false in the compare_for_stack_reg call.
4409 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4410 output_fp_compare for stack register operands.
4411 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4412 instructions here. Do not emit stack register pops here. Assert
4413 that FCOMPP pops next to top stack register. Rewrite function.
4415 2017-10-17 Nathan Sidwell <nathan@acm.org>
4418 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4419 use DECL_ASSEMBLER_NAME_RAW.
4422 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4425 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4426 Wilco Dijkstra <wilco.dijkstra@arm.com>
4428 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4429 converstion to Pmode from the buf_addr.
4431 2017-10-17 Richard Biener <rguenther@suse.de>
4433 * graphite-dependences.c (scop_get_reads_and_writes): Change
4434 output parameters to references.
4436 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4438 PR 71026/tree-optimization
4439 * fold-const.c (distribute_real_division): Removed.
4440 (fold_binary_loc): Remove calls to distribute_real_divison.
4442 2017-10-17 Richard Biener <rguenther@suse.de>
4444 * graphite-scop-detection.c
4445 (scop_detection::stmt_has_simple_data_refs_p): Always use
4446 the full nest as region.
4447 (try_generate_gimple_bb): Likewise.
4448 * sese.c (scalar_evolution_in_region): Simplify now that
4449 SCEV can handle instantiation in regions.
4450 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4451 in the non-loop part of a function if requested.
4453 2017-10-17 Richard Biener <rguenther@suse.de>
4455 PR tree-optimization/82563
4456 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4458 (graphite_regenerate_ast_isl): Call it.
4459 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4461 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4463 PR tree-optimization/82549
4464 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4465 Formatting fixes. Instead of calling make_bit_field_ref with negative
4468 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4470 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4471 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4472 _mm_maskz_reduce_ss): New.
4473 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4474 __builtin_ia32_reducess_mask): Ditto..
4475 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4476 * config/i386/sse.md (reduces<mode>): Renamed to ...
4477 (reduces<mode><mask_scalar_name>): ... this.
4478 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4480 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4481 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4483 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4485 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4486 * selftest-run-tests.c (selftest::run_tests): Call
4487 selftest::unique_ptr_tests_cc_tests.
4488 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4489 * unique-ptr-tests.cc: New file.
4491 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4494 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4496 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4497 (make_hard_regno_born, make_hard_regno_dead): Update
4498 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4500 2017-10-16 Jeff Law <law@redhat.com>
4502 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4504 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4506 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4508 2017-10-16 Olivier Hainque <hainque@adacore.com>
4510 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4511 with_cpu if we were configured for an e500v2 target cpu name.
4513 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4515 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4516 * doc/invoke.texi: Document +nodsp as a valid extension for
4519 2017-10-16 Martin Liska <mliska@suse.cz>
4521 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4522 (test_set_range): Likewise.
4523 (test_range_functions): Rename to ...
4524 (test_bit_in_range): ... this.
4525 (sbitmap_c_tests): Add new test.
4527 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4529 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4531 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4532 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4534 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4536 * config/aarch64/aarch64-builtins.c
4537 (aarch64_types_quadopu_lane_qualifiers): New.
4538 (TYPES_QUADOPU_LANE): New.
4539 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4540 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4541 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4542 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4543 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4544 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4545 (Vdottype, DOTPROD): New.
4546 (sur): Add SDOT and UDOT.
4548 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4550 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4551 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4552 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4554 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4555 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4556 Enable TARGET_DOTPROD.
4557 (cortex-a75.cortex-a55): Likewise.
4558 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4560 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4562 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4563 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4565 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4567 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4568 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4569 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4570 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4571 * config/arm/types.md (neon_dot, neon_dot_q): New.
4572 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4574 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4576 * config/arm/arm.h (TARGET_DOTPROD): New.
4577 * config/arm/arm.c (arm_arch_dotprod): New.
4578 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4579 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4580 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4581 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4582 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4583 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4584 * doc/invoke.texi (armv8.2-a): Document dotprod
4586 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4588 * i386.c (ix86_vec_cost): New function.
4589 (ix86_rtx_costs): Handle vector operations better.
4590 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4591 * x86-tune-costs.h: Add new costs to all tables.
4593 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4595 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4597 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4598 divsd, sqrtss and sqrtsd
4599 * x86-tune-costs.h: Add new entries to all costs.
4600 (znver1_cost): Fix to match real instruction latencies.
4602 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4603 Michael Collison <michael.collison@arm.com>
4605 * compare-elim.c: Include emit-rtl.h.
4606 (can_merge_compare_into_arith): New function.
4607 (try_validate_parallel): Likewise.
4608 (try_merge_compare): Likewise.
4609 (try_eliminate_compare): Call the above when no previous clobber
4611 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4614 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4618 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4619 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4623 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4624 to be any operand_equal_p operands. For & (B - 1) require
4625 B to be power of 2. Recognize
4626 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4628 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4631 * optabs.c (expand_memory_blockage): Fix call of
4632 targetm.have_memory_blockage.
4634 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4637 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4638 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4639 objects to extra_objs instead of overwriting it.
4641 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4643 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4644 Use any_fp_register_operand as operand[3] predicate. Simplify
4645 equality test for operands[2] and operands[4] memory location.
4646 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4647 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4648 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4649 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4650 any_fp_register_operand as operand[1] predicate. Simplify
4651 equality test for operands[0] and operands[3] memory location.
4652 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4653 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4654 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4656 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4658 * target-insns.def: Add memory_blockage.
4659 * optabs.c (expand_memory_blockage): New function.
4660 (expand_asm_memory_barrier): Rename ...
4661 (expand_asm_memory_blockage): ... to this.
4662 (expand_mem_thread_fence): Call expand_memory_blockage
4663 instead of expand_asm_memory_barrier.
4664 (expand_mem_singnal_fence): Ditto.
4665 (expand_atomic_load): Ditto.
4666 (expand_atomic_store): Ditto.
4667 * doc/md.texi (Standard Pattern Names For Generation):
4668 Document memory_blockage instruction pattern.
4670 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4672 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4673 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4674 * config/rl78/rl78.md: New define_expand "adddi3".
4676 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4678 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4681 2017-10-13 Jeff Law <law@redhat.com>
4683 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4685 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4688 * internal-fn.c (expand_mul_overflow): If both operands have
4689 the same highpart of -1 or 0 and the topmost bit of lowpart
4690 is different, overflow is if res <= 0 rather than res < 0.
4692 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4694 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4695 TARGET_P9_VECTOR code for unaligned_load case.
4697 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4699 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4701 2017-10-13 Nathan Sidwell <nathan@acm.org>
4703 * tree-core.h (tree_contains_struct): Make bool.
4704 * tree.c (tree_contains_struct): Likewise.
4705 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4706 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4707 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4708 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4709 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4711 2017-10-13 Richard Biener <rguenther@suse.de>
4713 * graphite-isl-ast-to-gimple.c
4714 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4715 parameters and dominance check.
4716 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4717 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4718 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4719 Do not update SSA form here or do intermediate IL verification.
4720 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4721 (graphite_initialize): Remove check on the number of loops in
4722 the function and inline into graphite_transform_loops.
4723 (graphite_finalize): Inline into graphite_transform_loops.
4724 (graphite_transform_loops): Perform SSA update and IL verification
4726 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4728 2017-10-13 Richard Biener <rguenther@suse.de>
4730 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4731 graphite_expression_type_precision): Avoid global constructor
4733 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4734 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4735 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4736 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4737 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4738 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4740 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4743 * config/i386/i386.h (ix86_red_zone_size): New.
4744 * config/i386/i386.md (push peephole2s): Replace
4745 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4747 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4748 Alan Hayward <alan.hayward@arm.com>
4749 David Sherwood <david.sherwood@arm.com>
4751 * combine.c (can_change_dest_mode): Reject changes in
4752 REGMODE_NATURAL_SIZE.
4754 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4755 Alan Hayward <alan.hayward@arm.com>
4756 David Sherwood <david.sherwood@arm.com>
4758 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4759 (expand_debug_source_expr): Likewise.
4760 * combine.c (combine_simplify_rtx): Likewise.
4761 * cse.c (fold_rtx): Likewise.
4762 * fwprop.c (canonicalize_address): Likewise.
4763 * targhooks.c (default_shift_truncation_mask): Likewise.
4765 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4766 Alan Hayward <alan.hayward@arm.com>
4767 David Sherwood <david.sherwood@arm.com>
4769 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4770 (widened_mode): Likewise.
4771 (expand_unop): Likewise.
4772 * ree.c (transform_ifelse): Likewise.
4773 (merge_def_and_ext): Likewise.
4774 (combine_reaching_defs): Likewise.
4775 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4777 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4778 Alan Hayward <alan.hayward@arm.com>
4779 David Sherwood <david.sherwood@arm.com>
4781 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4782 * combine.c (gen_lowpart_for_combine): Likewise.
4783 * dwarf2out.c (rtl_for_decl_location): Likewise.
4784 * final.c (alter_subreg): Likewise.
4785 * rtlhooks.c (gen_lowpart_general): Likewise.
4786 (gen_lowpart_if_possible): Likewise.
4788 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4789 Alan Hayward <alan.hayward@arm.com>
4790 David Sherwood <david.sherwood@arm.com>
4792 * calls.c (expand_call): Use subreg_lowpart_offset.
4793 * cse.c (cse_insn): Likewise.
4794 * regcprop.c (copy_value): Likewise.
4795 (copyprop_hardreg_forward_1): Likewise.
4797 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4800 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4801 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4802 =Q constraints to +Q and into insn condition add check
4803 that operands[0] and operands[1] are equal.
4804 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4805 =Q constraints to +Q and into insn condition add check
4806 that operands[0] is equal to either operands[1] or operands[2].
4809 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4810 instead of handling MINUS_EXPR twice (once for each argument),
4811 canonicalize operand order and handle just once, use rtype where
4812 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4815 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4816 any values of __C while still being pattern recognizable as a simple
4819 2017-10-13 Richard Biener <rguenther@suse.de>
4821 PR tree-optimization/82451
4823 2017-10-02 Richard Biener <rguenther@suse.de>
4825 PR tree-optimization/82355
4826 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4827 a mapping for the enclosing loop but avoid generating one for
4829 (copy_bb_and_scalar_dependences): Remove premature codegen
4830 error on PHIs in blocks duplicated into multiple places.
4831 * graphite-scop-detection.c
4832 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4833 in the region use it as loop and nest to analyze the DR in.
4834 (try_generate_gimple_bb): Likewise.
4835 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4836 (add_loop_constraints): For blocks in a loop not in the region
4837 create a dimension with a single iteration.
4838 * sese.h (gbb_loop_at_index): Remove assert.
4840 * cfgloop.c (loop_preheader_edge): For the loop tree root
4841 return the single successor of the entry block.
4842 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4843 Reset the SCEV hashtable and niters.
4844 * graphite-scop-detection.c
4845 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4846 assert that we only have POLYNOMIAL_CHREC that vary in loops
4847 contained in the region.
4848 (scop_detection::graphite_can_represent_expr): Adjust.
4849 (scop_detection::stmt_has_simple_data_refs_p): For loops
4850 not in the region set loop to NULL. The nest is now the
4851 entry edge to the region.
4852 (try_generate_gimple_bb): Likewise.
4853 * sese.c (scalar_evolution_in_region): Adjust for
4854 instantiate_scev change.
4855 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4856 Make nest parameter the edge into the region.
4857 (create_data_ref): Likewise.
4858 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4859 entry edge into a region and adjust instantiate_scev calls.
4860 (create_data_ref): Likewise.
4861 (graphite_find_data_references_in_stmt): Likewise.
4862 (find_data_references_in_stmt): Pass the loop preheader edge
4863 from the nest argument.
4864 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4865 parameter the edge into the region.
4866 (instantiate_parameters): Use the loop preheader edge as entry.
4867 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4869 (get_instantiated_value_entry): Make instantiate_below parameter
4870 the edge into the region.
4871 (instantiate_scev_name): Likewise. Adjust dominance checks,
4872 when we cannot use loop-based instantiation instantiate by
4873 walking use-def chains.
4874 (instantiate_scev_poly): Adjust.
4875 (instantiate_scev_binary): Likewise.
4876 (instantiate_scev_convert): Likewise.
4877 (instantiate_scev_not): Likewise.
4878 (instantiate_array_ref): Remove.
4879 (instantiate_scev_3): Likewise.
4880 (instantiate_scev_2): Likewise.
4881 (instantiate_scev_1): Likewise.
4882 (instantiate_scev_r): Do not blindly handle N-operand trees.
4883 Do not instantiate array-refs. Handle all constants and invariants.
4884 (instantiate_scev): Make instantiate_below parameter
4885 the edge into the region.
4886 (resolve_mixers): Use the loop preheader edge for the region
4887 parameter to instantiate_scev_r.
4888 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4890 2017-10-13 Richard Biener <rguenther@suse.de>
4892 PR tree-optimization/82525
4893 * graphite-isl-ast-to-gimple.c
4894 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4896 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4897 Fail code generation when we cannot represent the isl integer.
4898 (binary_op_to_tree): Elide modulo operations that are no-ops
4899 in the type we code generate. Remove now superfluous code
4902 2017-10-13 Richard Biener <rguenther@suse.de>
4904 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4905 (scop_detection::harmful_loop_in_region): Remove premature
4906 IV type restriction.
4907 (scop_detection::graphite_can_represent_scev): We can handle
4908 pointer IVs just fine.
4910 2017-10-13 Alan Modra <amodra@gmail.com>
4912 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4913 "Clobbers and Scratch Registers". Add paragraph on
4914 alternative to clobbers for scratch registers and OpenBLAS
4917 2017-10-13 Alan Modra <amodra@gmail.com>
4919 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4920 example of a memory input for a string of known length. Move
4921 commentary out of table. Add a number of new examples
4922 covering array memory inputs.
4924 2017-10-12 Martin Liska <mliska@suse.cz>
4926 PR tree-optimization/82493
4927 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4928 (test_range_functions): New function.
4929 (sbitmap_c_tests): Likewise.
4930 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4931 * selftest.h (sbitmap_c_tests): New function.
4933 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4935 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4937 * config/rs6000/amo.h: Fix spacing issue.
4939 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4942 * config/i386/i386.md (*ashl<mode>3_mask_1,
4943 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4944 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4947 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4949 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4950 (profile_probability): Set max_probability
4951 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4953 * profile-count.c (profile_probability::differs_from_p): Do not
4954 rely on max_probaiblity == 10000
4956 2017-10-12 Jeff Law <law@redhat.com>
4958 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4961 2017-10-12 Martin Sebor <msebor@redhat.com>
4965 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4966 (handle_alias_pairs): Call it.
4967 * common.opt (-Wattribute-alias): New option.
4968 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4969 * doc/invoke.texi (-Wattribute-alias): Document.
4971 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4974 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4976 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4978 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4979 (make_hard_regno_born, make_hard_regno_dead): Update
4980 bb_killed_pseudos and bb_gen_pseudos.
4982 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4984 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4986 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4988 * config/alpha/alpha.c (alpha_split_conditional_move):
4989 Use std::swap instead of manually swapping.
4990 (alpha_stdarg_optimize_hook): Ditto.
4991 (alpha_canonicalize_comparison): Ditto.
4993 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4995 * tree-loop-distribution.c (struct builtin_info): New struct.
4996 (struct partition): Refactor fields into struct builtin_info.
4997 (partition_free): Free struct builtin_info.
4998 (build_size_arg_loc, build_addr_arg_loc): Delete.
4999 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5000 information from struct builtin_info.
5001 (find_single_drs): New function refactored from classify_partition.
5002 Also moved builtin validity checks to this function.
5003 (compute_access_range, alloc_builtin): New functions.
5004 (classify_builtin_st, classify_builtin_ldst): New functions.
5005 (classify_partition): Refactor code into functions find_single_drs,
5006 classify_builtin_st and classify_builtin_ldst.
5007 (distribute_loop): Don't do runtime alias check when distributing
5009 (find_seed_stmts_for_distribution): New function.
5010 (pass_loop_distribution::execute): Refactor code finding seed
5011 stmts into above function. Support distribution for the innermost
5012 two-level loop nest. Adjust dump information.
5014 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5016 * tree-loop-distribution.c: Adjust the general comment.
5017 (NUM_PARTITION_THRESHOLD): New macro.
5018 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5019 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5020 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5021 in call to build_partition_graph.
5022 (finalize_partitions): New parameter. Make loop distribution more
5023 conservative by fusing more partitions.
5024 (distribute_loop): Don't do runtime alias check in case of loop nest
5026 (find_seed_stmts_for_distribution): New function.
5027 (prepare_perfect_loop_nest): New function.
5028 (pass_loop_distribution::execute): Refactor code finding seed stmts
5029 and loop nest into above functions. Support loop nest distribution.
5030 Adjust dump information accordingly.
5032 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5034 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5035 and set PTYPE_SEQUENTIAL for merged partition.
5037 2017-10-12 Richard Biener <rguenther@suse.de>
5039 PR tree-optimization/69728
5041 2017-09-19 Richard Biener <rguenther@suse.de>
5043 PR tree-optimization/69728
5044 * graphite-sese-to-poly.c (schedule_error): New global.
5045 (add_loop_schedule): Handle empty domain by failing the
5047 (build_original_schedule): Handle schedule_error.
5049 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5050 domain by returning an unchanged schedule.
5052 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5054 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5055 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5057 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5059 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5062 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5065 * expr.c (store_field): Don't optimize away bitsize == 0 store
5066 from CALL_EXPR with addressable return type.
5068 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5070 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5071 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5072 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5073 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5074 TARGET_ISEL instead of TARGET_ISEL<sel>.
5076 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5078 * config/rs6000/rs6000.c
5079 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5081 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5083 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5085 (reg_or_cint_operand): Fix comment.
5086 (reg_or_zero_operand): New predicate.
5087 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5088 * config/rs6000/rs6000.c (output_isel): Delete.
5089 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5090 instead of reg_or_cint_operand. Output instruction directly (not via
5092 (isel_unsigned_<mode>): Ditto.
5093 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5094 gpc_reg_operand. Add an instruction alternative for this. Output
5095 instruction directly.
5096 (*isel_reversed_unsigned_<mode>): Ditto.
5098 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5100 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5101 (TARGET_CANONICALIZE_COMPARISON): Define.
5103 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5106 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5107 Check whether the dest is REG before adding REG_EQUIV note.
5109 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5112 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5114 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5115 (make_hard_regno_born, make_hard_regno_dead): Update
5116 bb_killed_pseudos and bb_gen_pseudos.
5118 2017-10-11 Nathan Sidwell <nathan@acm.org>
5120 * incpath.h (enum incpath_kind): Name enum, prefix values.
5121 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5122 * incpath.c (heads, tails): Use INC_MAX.
5123 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5124 (merge_include_chains, split_quote_chain,
5125 register_include_chains): Update incpath_kind names.
5126 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5127 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5129 (add_framework_path, darwin_register_objc_includes): Likewise.
5130 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5132 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5134 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5135 Do not use float_operator operator predicate.
5136 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5137 * config/i386/predicates.md (float_operator): Remove predicate.
5139 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5141 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5142 (*jccxf_i387): Ditto.
5143 (*jcc<mode>_i387): Ditto.
5144 (*jccu<mode>_i387): Ditto.
5145 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5146 (*jcc_*_i387 splitters): Remove.
5147 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5148 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5149 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5152 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5154 * profile-count.h (slow_safe_scale_64bit): New function.
5155 (safe_scale_64bit): New inline.
5156 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5157 * profile-count.c: Include wide-int.h
5158 (slow_safe_scale_64bit): New.
5160 2017-10-11 Nathan Sidwell <nathan@acm.org>
5162 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5163 HAS_DECL_ASSEMBLER_NAME_P.
5164 * gimple-expr.c (gimple_decl_printable_name: Check
5165 HAS_DECL_ASSEMBLER_NAME_P too.
5166 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5167 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5168 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5169 * passes.c (rest_of_decl_compilation): Check
5170 HAS_DECL_ASSEMBLER_NAME_P too.
5171 * recog.c (verify_changes): Likewise.
5172 * tree-pretty-print.c (dump_decl_name): Likewise.
5173 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5175 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5176 (SET_DECL_ASSEMBLER_NAME): Use it.
5177 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5178 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5179 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5181 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5183 * config.gcc (i386, x86_64): Add extra objects.
5184 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5185 (ix86_min_insn_size): Declare.
5186 (ix86_issue_rate): Declare.
5187 (ix86_adjust_cost): Declare.
5188 (ia32_multipass_dfa_lookahead): Declare.
5189 (ix86_macro_fusion_p): Declare.
5190 (ix86_macro_fusion_pair_p): Declare.
5191 (ix86_bd_has_dispatch): Declare.
5192 (ix86_bd_do_dispatch): Declare.
5193 (ix86_core2i7_init_hooks): Declare.
5194 (ix86_atom_sched_reorder): Declare.
5195 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5196 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5197 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5198 (rip_relative_addr_p): Rename to ...
5199 (ix86_rip_relative_addr_p): ... this one; export.
5200 (memory_address_length): Update.
5201 (ix86_issue_rate): Move to x86-tune-sched.c.
5202 (ix86_flags_dependent): Move to x86-tune-sched.c.
5203 (ix86_agi_dependent): Move to x86-tune-sched.c.
5204 (exact_dependency_1): Move to x86-tune-sched.c.
5205 (exact_store_load_dependency): Move to x86-tune-sched.c.
5206 (ix86_adjust_cost): Move to x86-tune-sched.c.
5207 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5208 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5209 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5210 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5211 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5212 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5213 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5214 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5215 (min_insn_size): Rename to ...
5216 (ix86_min_insn_size): ... this one; export.
5217 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5218 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5219 (core2i7_first_cycle_multipass_backtrack): Move to
5220 x86-tune-sched-core.c.
5221 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5222 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5223 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5224 (ix86_avoid_jump_mispredicts): Update.
5225 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5226 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5227 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5228 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5229 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5230 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5231 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5232 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5233 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5234 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5235 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5236 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5237 (BIG): Move to ix86-tune-sched-bd.c.
5238 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5239 (enum insn_path): Move to ix86-tune-sched-bd.c.
5240 (get_mem_group): Move to ix86-tune-sched-bd.c.
5241 (is_cmp): Move to ix86-tune-sched-bd.c.
5242 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5243 (is_branch): Move to ix86-tune-sched-bd.c.
5244 (is_prefetch): Move to ix86-tune-sched-bd.c.
5245 (init_window): Move to ix86-tune-sched-bd.c.
5246 (allocate_window): Move to ix86-tune-sched-bd.c.
5247 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5248 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5249 (process_end_window): Move to ix86-tune-sched-bd.c.
5250 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5251 (find_constant): Move to ix86-tune-sched-bd.c.
5252 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5253 (has_immediate): Move to ix86-tune-sched-bd.c.
5254 (get_insn_path): Move to ix86-tune-sched-bd.c.
5255 (get_insn_group): Move to ix86-tune-sched-bd.c.
5256 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5257 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5258 (add_insn_window): Move to ix86-tune-sched-bd.c.
5259 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5260 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5261 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5262 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5263 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5264 (do_dispatch): Move to ix86-tune-sched-bd.c.
5265 (has_dispatch): Move to ix86-tune-sched-bd.c.
5266 * config/i386/t-i386: Add new object files.
5267 * config/i386/x86-tune-costs.h: New file.
5268 * config/i386/x86-tune-sched-atom.c: New file.
5269 * config/i386/x86-tune-sched-bd.c: New file.
5270 * config/i386/x86-tune-sched-core.c: New file.
5271 * config/i386/x86-tune-sched.c: New file.
5273 2017-10-11 Liu Hao <lh_mouse@126.com>
5275 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5276 the generic version below instead.
5277 (should_colorize): Recognize Windows consoles as terminals
5279 * pretty-print.c [__MINGW32__] (write_all): New function.
5280 [__MINGW32__] (find_esc_head): Likewise.
5281 [__MINGW32__] (find_esc_terminator): Likewise.
5282 [__MINGW32__] (eat_esc_sequence): Likewise.
5283 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5285 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5288 2017-10-11 Richard Biener <rguenther@suse.de>
5290 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5291 Properly call analyze_scalar_evolution with the loop of the stmt.
5293 2017-10-11 Richard Biener <rguenther@suse.de>
5295 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5296 * tree-core.h (tree_base): Add chrec_var union member.
5297 * tree.h (CHREC_VAR): Remove.
5298 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5299 * tree-chrec.h (build_polynomial_chrec): Adjust.
5300 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5301 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5303 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5305 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5306 * match.pd: ... here.
5307 ((T) X == (T) Y): Relax condition.
5309 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5311 PR tree-optimization/82472
5312 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5314 (break_alias_scc_partitions): Update postorder number.
5316 2017-10-11 Martin Liska <mliska@suse.cz>
5319 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5321 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5323 2017-10-11 Martin Liska <mliska@suse.cz>
5325 * print-rtl.c (print_insn): Move declaration of idbuf
5326 to same scope as name.
5328 2017-10-11 Martin Liska <mliska@suse.cz>
5333 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5336 2017-10-11 Martin Liska <mliska@suse.cz>
5339 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5342 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5344 * config.gcc (default_gnu_indirect_function): Default to yes for
5345 arm*-*-linux* with glibc.
5347 2017-10-11 Richard Biener <rguenther@suse.de>
5349 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5350 default-defs and types we do not want to analyze.
5351 (interpret_loop_phi): Replace unreachable code with an assert.
5352 (compute_scalar_evolution_in_loop): Remove and inline ...
5353 (analyze_scalar_evolution_1): ... here, replacing condition with
5354 what makes the intent clearer. Remove handling of cases
5355 get_scalar_evolution now handles.
5357 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5359 PR rtl-optimization/81434
5360 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5361 comment for main loop. In sched_group_found if, also add checks for
5362 pass and min_cost_group.
5364 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5366 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5367 (rs6000_insn_cost): New function.
5368 * config/rs6000/rs6000.md (cost): New attribute.
5370 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5371 H.J. Lu <hongjiu.lu@intel.com>
5375 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5376 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5377 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5378 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5379 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5380 OPTION_MASK_ISA_MMX and some other bit set require both
5381 mmx and the ISAs without the mmx bit.
5382 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5383 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5384 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5385 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5386 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5387 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5388 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5389 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5390 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5391 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5392 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5393 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5394 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5395 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5396 Add OPTION_MASK_ISA_MMX.
5398 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5400 * config.gcc (armv7*-*-freebsd*): New target.
5401 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5403 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5405 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5406 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5409 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5411 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5412 a template parameter.
5413 (WIDE_INT_REF_FOR): Update accordingly.
5414 * tree.h (wi::int_traits <const_tree>): Delete.
5415 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5416 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5417 (wi::tree_to_wide_ref): New typedef.
5418 (wi::to_wide): New function.
5419 * calls.c (get_size_range): Use wi::to_wide when operating on
5421 * cgraph.c (cgraph_node::create_thunk): Likewise.
5422 * config/i386/i386.c (ix86_data_alignment): Likewise.
5423 (ix86_local_alignment): Likewise.
5424 * dbxout.c (stabstr_O): Likewise.
5425 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5426 * expr.c (const_vector_from_tree): Likewise.
5427 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5428 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5429 (fold_negate_expr_1, int_const_binop_1, const_binop)
5430 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5431 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5432 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5433 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5434 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5435 (fold_not_const, round_up_loc): Likewise.
5436 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5437 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5438 (alloca_call_type): Likewise.
5439 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5440 * godump.c (go_output_typedef): Likewise.
5441 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5442 * internal-fn.c (get_min_precision): Likewise.
5443 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5444 * ipa-polymorphic-call.c
5445 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5446 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5447 (ipa_modify_call_arguments): Likewise.
5448 * match.pd: Likewise.
5449 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5450 * print-tree.c (print_node_brief, print_node): Likewise.
5451 * stmt.c (expand_case): Likewise.
5452 * stor-layout.c (layout_type): Likewise.
5453 * tree-affine.c (tree_to_aff_combination): Likewise.
5454 * tree-cfg.c (group_case_labels_stmt): Likewise.
5455 * tree-data-ref.c (dr_analyze_indices): Likewise.
5456 (prune_runtime_alias_test_list): Likewise.
5457 * tree-dump.c (dequeue_and_dump): Likewise.
5458 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5459 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5460 * tree-pretty-print.c (dump_generic_node): Likewise.
5461 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5462 (simple_iv_with_niters): Likewise.
5463 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5464 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5465 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5466 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5467 (refine_value_range_using_guard, number_of_iterations_ne_max)
5468 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5469 (get_cst_init_from_scev, record_nonwrapping_iv)
5470 (scev_var_range_cant_overflow): Likewise.
5471 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5472 * tree-ssa-pre.c (compute_avail): Likewise.
5473 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5474 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5475 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5476 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5477 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5478 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5479 (dump_case_nodes, try_switch_expansion): Likewise.
5480 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5481 (vect_do_peeling): Likewise.
5482 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5483 * tree-vect-stmts.c (vectorizable_load): Likewise.
5484 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5485 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5486 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5487 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5488 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5489 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5490 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5491 (evrp_dom_walker::before_dom_children): Likewise.
5492 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5493 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5494 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5495 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5496 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5497 (get_range_pos_neg): Likewise.
5498 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5499 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5500 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5501 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5502 * config/avr/avr.c (avr_fold_builtin): Likewise.
5503 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5504 * config/msp430/msp430.c (msp430_attr): Likewise.
5505 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5506 * config/powerpcspe/powerpcspe-c.c
5507 (altivec_resolve_overloaded_builtin): Likewise.
5508 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5509 (rs6000_expand_ternop_builtin): Likewise.
5510 * config/rs6000/rs6000-c.c
5511 (altivec_resolve_overloaded_builtin): Likewise.
5512 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5513 (rs6000_expand_ternop_builtin): Likewise.
5514 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5516 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5518 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5519 when copying loop nest with only one inner loop.
5521 2017-10-10 Richard Biener <rguenther@suse.de>
5523 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5524 blocks if SCEV is active.
5525 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5527 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5528 (scev_initialize): Assert we are not yet initialized.
5530 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5532 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5533 inner loop's exit stmt by making it always exit the loop, otherwise
5534 we would generate an infinite empty loop.
5536 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5538 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5539 renaming variables in new preheader if it's deleted.
5541 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5543 * tree-loop-distribution.c (struct partition): Remove unused field
5544 loops of the structure.
5545 (partition_alloc, partition_free): Ditto.
5546 (build_rdg_partition_for_vertex): Ditto.
5548 2017-10-09 Jeff Law <law@redhat.com>
5550 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5551 return type to match prototype and documentation.
5553 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5555 * config/rs6000/rs6000.c (processor_costs): Move to ...
5556 * config/rs6000/rs6000.h: ... here.
5557 (rs6000_cost): Declare.
5559 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5561 * except.c (setjmp_fn): New global variable.
5562 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5563 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5564 if DONT_USE_BUILTIN_SETJMP is defined.
5566 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5568 * target.def (insn_cost): New hook.
5569 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5570 * doc/tm.texi: Regenerate.
5571 * rtlanal.c (insn_cost): Use the new hook.
5573 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5575 * combine.c (combine_validate_cost): Compute the new insn_cost,
5576 not just pattern_cost.
5577 (try_combine): Adjust comment.
5579 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5581 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5583 * combine.c (uid_insn_cost): Adjust comment.
5584 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5586 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5587 * dse.c (find_shift_sequence): Ditto.
5588 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5589 (bb_valid_for_noce_process_p): Use pattern_cost.
5590 * rtl.h (insn_rtx_cost): Delete.
5591 (pattern_cost): New prototype.
5592 (insn_cost): New prototype.
5593 * rtlanal.c (insn_rtx_cost): Rename to...
5594 (pattern_cost): ... this.
5597 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5599 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5600 (*jcc<mode>_0_r_i387): Ditto.
5601 (*jccxf_r_i387): Ditto.
5602 (*jcc<mode>_r_i387): Ditto.
5603 (*jccu<mode>_r_i387): Ditto.
5604 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5605 (*jcc): Rename from *jcc_1.
5607 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5609 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5610 deferred rescans after the lvx/stvx recombination pre-pass.
5612 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5614 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5615 memory operation instruction support.
5616 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5617 (rs6000-ibm-aix[789]*): Likewise.
5618 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5619 Document new functions.
5621 2017-10-09 Richard Biener <rguenther@suse.de>
5623 PR tree-optimization/82397
5624 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5625 equality only for semantically equal trees.
5627 2017-10-09 Richard Biener <rguenther@suse.de>
5629 PR tree-optimization/82449
5630 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5631 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5632 allow constant addresses.
5633 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5636 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5638 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5641 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5644 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5647 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5650 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5652 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5655 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5656 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5658 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5660 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5661 (WI_BINARY_PREDICATE_RESULT): Likewise.
5662 (wi::binary_traits::operator_result): New type.
5663 (wi::binary_traits::predicate_result): Likewise.
5664 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5665 (generic_wide_int::operator==, generic_wide_int::operator!=)
5666 (generic_wide_int::operator&, generic_wide_int::and_not)
5667 (generic_wide_int::operator|, generic_wide_int::or_not)
5668 (generic_wide_int::operator^, generic_wide_int::operator+
5669 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5671 (operator~, unary operator-, operator==, operator!=, operator&)
5672 (operator|, operator^, operator+, binary operator-, operator*): New
5674 * expr.c (get_inner_reference): Use wi::bit_and_not.
5675 * fold-const.c (fold_binary_loc): Likewise.
5676 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5677 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5678 (bit_value_binop): Likewise.
5679 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5680 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5681 (extract_range_from_binary_expr_1): Likewise.
5682 (masked_increment): Likewise.
5683 (simplify_bit_ops_using_ranges): Likewise.
5685 2017-10-09 Martin Jambor <mjambor@suse.cz>
5688 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5689 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5690 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5692 (hsa_fixup_mov_insn_type): New function.
5693 (hsa_op_with_type::get_in_type): Use it.
5694 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5695 immediates in an assert.
5696 (hsa_op_with_type::extend_int_to_32bit): New method.
5697 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5698 types. Convert to dest type if necessary.
5699 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5700 (reg_for_gimple_ssa): Pass false as min32int to
5701 hsa_type_for_scalar_tree_type.
5702 (gen_hsa_addr): Fixup type when creating addresable temporary.
5703 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5704 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5705 necessary. Call hsa_fixup_mov_insn_type.
5706 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5707 extend operands and convert to dest type if necessary.
5708 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5709 to dest type if necessary.
5710 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5711 if conversion nt necessary and size matches.
5712 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5713 to dest type if necessary.
5714 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5715 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5717 (gen_hsa_clrsb): Likewise.
5718 (gen_hsa_ffs): Likewise.
5719 (gen_hsa_divmod): Extend operands and convert to dest type if
5721 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5723 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5725 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5726 Remove empty default arguments. Use a brace block as output
5728 (conditional return): Ditto.
5730 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5731 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5733 (group_ending_nop): Ditto.
5734 (doloop_end): Ditto.
5735 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5736 (splitters for those): Ditto.
5738 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5740 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5741 a conditional jump (and the compare for it) so that pc_rtx is the
5743 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5744 for the deleted and renamed ctr<mode>_internal[234] patterns.
5745 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5746 Delete second conditional return pattern.
5747 (ctr<mode>_internal2): Delete this second bdnz pattern.
5748 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5749 (ctr<mode>_internal4): Delete this second bdz pattern.
5751 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5753 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5754 (always_initialized_rtx_for_ssa_name_p): New predicate.
5755 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5756 (finish_out_of_ssa): Free new field of SA.
5757 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5758 * tree-ssa-coalesce.c: Include tree-ssa.h.
5759 (get_parm_default_def_partitions): Remove extern keyword.
5760 (get_undefined_value_partitions): New function.
5761 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5762 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5764 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5766 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5768 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5770 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5772 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5773 for targets that preffer 128bit.
5775 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5777 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5779 2017-10-08 Olivier Hainque <hainque@adacore.com>
5781 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5782 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5784 (thumb_set_return_address): Likewise.
5786 2017-10-08 Olivier Hainque <hainque@adacore.com>
5788 * common/config/arm/arm-common.c (arm_except_unwind_info):
5789 Handle DWARF2_UNWIND_INFO.
5791 2017-10-07 Michael Collison <michael.collison@arm.com>
5793 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5796 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5798 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5799 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5800 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5801 defined, force the creation of a new block for a dispatch label.
5803 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5805 * invoke.texi (Wsuggest-attribute=cold): Document.
5806 * common.opt (Wsuggest-attribute=cold): New
5807 * ipa-pure-const.c (warn_function_cold): New function.
5808 * predict.c (compute_function_frequency): Use it.
5809 * predict.h (warn_function_cold): Declare.
5811 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5813 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5816 2017-10-06 Martin Liska <mliska@suse.cz>
5818 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5819 keyword for member functions.
5820 (struct sanopt_tree_couple): New struct.
5821 (struct sanopt_tree_couple_hash): New function.
5822 (struct sanopt_ctx): Add new hash_map.
5823 (has_dominating_ubsan_ptr_check): New function.
5824 (record_ubsan_ptr_check_stmt): Likewise.
5825 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5826 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5827 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5829 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5832 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5833 aarch64_simd_valid_immediate on CONST_VECTORs.
5834 (aarch64_reg_or_bic_imm): Likewise.
5836 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5838 PR rtl-optimization/82396
5839 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5841 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5843 * graphite-dependences.c (scop_get_reads): Move code to...
5844 (scop_get_must_writes): Move code to...
5845 (scop_get_may_writes): Move code to...
5846 (scop_get_reads_and_writes): ... here.
5847 (scop_get_dependences): Call scop_get_reads_and_writes.
5849 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5851 PR tree-optimization/82434
5852 * fold-const.h (can_native_encode_type_p,
5853 can_native_encode_string_p): Remove.
5854 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5855 don't encode anything, just return what would be otherwise returned.
5856 (native_encode_fixed, native_encode_complex, native_encode_vector):
5858 (native_encode_string): Likewise. Inline by hand
5859 can_native_encode_string_p.
5860 (can_native_encode_type_p): Remove.
5861 (can_native_encode_string_p): Remove.
5862 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5863 STRING_CSTs using can_native_encode_string_p, test all
5864 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5865 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5866 argument from native_encode_expr.
5867 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5868 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5871 2017-10-06 Richard Biener <rguenther@suse.de>
5873 PR tree-optimization/82397
5874 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5875 operand_equal_p but rely on data_ref_compare_tree for detecting
5877 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5878 to match up with dr_group_sort_cmp.
5880 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5883 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5885 * config/s390/s390-builtin-types.def: Regenerate.
5887 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5890 * config/s390/s390-builtin-types.def: Regenerate.
5891 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5892 Change flag from B_VXE to B_VX.
5893 (s390_vec_min_dbl): Remove B_VXE flag.
5895 2017-10-06 Richard Biener <rguenther@suse.de>
5897 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5898 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5899 translate_isl_ast_to_gimple::is_valid_rename,
5900 translate_isl_ast_to_gimple::get_rename,
5901 translate_isl_ast_to_gimple::get_def_bb_for_const,
5902 translate_isl_ast_to_gimple::get_new_name,
5903 translate_isl_ast_to_gimple::collect_all_ssa_names,
5904 translate_isl_ast_to_gimple::copy_loop_phi_args,
5905 translate_isl_ast_to_gimple::collect_all_ssa_names,
5906 translate_isl_ast_to_gimple::copy_loop_phi_args,
5907 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5908 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5909 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5910 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5911 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5912 translate_isl_ast_to_gimple::copy_cond_phi_args,
5913 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5914 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5915 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5916 translate_isl_ast_to_gimple::rename_uses,
5917 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5918 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5919 (set_rename_for_each_def): Likewise.
5920 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5921 here. Handle rewriting SCEV analyzable uses here.
5922 (copy_bb_and_scalar_dependences): Generate code for PHI
5924 (graphite_regenerate_ast_isl): Adjust.
5925 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5926 (add_write, add_read): New functions.
5927 (build_cross_bb_scalars_def): Use it and simplify.
5928 (build_cross_bb_scalars_use): Likewise.
5929 (graphite_find_cross_bb_scalar_vars): Inline into...
5930 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5931 simulating out-of-SSA. Compute liveout and add dependencies.
5932 (build_scops): Force an empty entry block.
5933 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5935 (sese_build_liveouts): Declare.
5936 (sese_trivially_empty_bb_p): Likewise.
5937 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5938 compute liveout and debug_liveout.
5939 (sese_bad_liveouts_use): Remove.
5940 (sese_reset_debug_liveouts_bb): Likewise.
5941 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5942 (sese_build_liveouts): Build liveout and debug_liveout and store
5944 (new_sese_info): Adjust.
5945 (free_sese_info): Likewise.
5946 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5947 do not build liveout here.
5948 (move_sese_in_condition): Adjust region entry.
5949 (scev_analyzable_p): Match up with chrec_apply requirements.
5950 (sese_trivially_empty_bb_p): New.
5951 * tree-into-ssa.c (get_reaching_def): Properly support generating
5952 default-defs for incremental rewrite of anonymous names.
5954 2017-10-06 Richard Biener <rguenther@suse.de>
5956 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5957 precision do not perform modulo reduction.
5959 2017-10-06 Richard Biener <rguenther@suse.de>
5961 PR tree-optimization/82436
5962 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5963 conservatively choose the vectorization factor when checking
5964 whether we can perform the required load permutation.
5965 (vect_transform_slp_perm_load): Assert when we may not fail.
5967 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5969 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5970 message for incompatible -msdata=* and -mcall-* options.
5972 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5974 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5975 rate for post-reload scheduling.
5977 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5979 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5981 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5983 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5984 to improve monte carlo in scimark.
5986 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5988 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5989 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5990 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5991 pentium4_cost, nocona_cost): Set reassociation width to 1.
5992 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5993 width to 2 for fp operations and 1 otherwise.
5994 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5996 (atom_cost): Set reassociation width to 2.
5997 (slm_cost, generic_cost): Set fp reassociation width
5998 to 2 and 1 otherwise.
5999 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6000 (core_cost): Set fp reassociation width to 4 and vector to 2.
6001 (ix86_reassociation_width): Rewrite using cost table; special case
6002 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6003 and TARGET_AVX128_OPTIMAL.
6004 * config/i386/i386.h (processor_costs): Add
6005 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6006 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6007 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6008 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6009 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6010 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6012 2017-10-05 Nathan Sidwell <nathan@acm.org>
6014 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6016 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6018 * config/arm/arm.c (arm_test_fpu_data): New.
6019 (arm_run_selftests): Call arm_test_fpu_data.
6021 2017-10-04 Nathan Sidwell <nathan@acm.org>
6023 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6025 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6027 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6029 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6030 check type for aarch64_simd_valid_immediate.
6031 (aarch64_output_simd_mov_immediate): Update prototype.
6032 (aarch64_simd_valid_immediate): Update prototype.
6033 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6034 support for ORR-immediate.
6035 (and<mode>3): modified pattern to add support for BIC-immediate.
6036 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6037 now checks for valid immediate for BIC and ORR based on new enum
6039 (aarch64_output_simd_mov_immediate): Function now used to output
6040 BIC/ORR imm as well based on new enum argument.
6041 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6043 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6044 (aarch64_reg_or_bic_imm): Likewise.
6046 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6048 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6049 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6051 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6055 PR rtl-optimization/82396
6056 * haifa-sched.c (autopref_multipass_init): Simplify
6058 (autopref_rank_data): Simplify sort order.
6059 * sched-int.h (autopref_multipass_data_): Remove
6060 multi_mem_insn_p, min_offset and max_offset.
6062 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6064 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6066 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6068 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6071 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6073 * doc/sourcebuild.texi: Document vect_long_mult.
6075 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6077 PR tree-optimization/82413
6078 * fold-const.c (build_range_check): Use widest_int when comparing
6079 the maximum ETYPE value with HIGH.
6081 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6083 PR rtl-optimization/82396
6084 * haifa-sched.c (autopref_multipass_init): Simplify
6086 (autopref_rank_data): Simplify sort order.
6087 * sched-int.h (autopref_multipass_data_): Remove
6088 multi_mem_insn_p, min_offset and max_offset.
6090 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6092 PR tree-optimization/82381
6093 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6094 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6097 PR tree-optimization/82374
6098 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6099 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6100 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6101 current_function_decl to the new decl.
6103 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6105 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6106 helper macro for IEEE float128 hardware built-in functions.
6107 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6109 (TRUNCF128_ODD): Likewise.
6110 (ADDF128_ODD): Likewise.
6111 (SUBF128_ODD): Likewise.
6112 (MULF128_ODD): Likewise.
6113 (DIVF128_ODD): Likewise.
6114 (FMAF128_ODD): Likewise.
6115 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6116 UNSPEC_TRUNC_ROUND_TO_ODD.
6117 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6118 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6119 floating point round to odd instructions.
6120 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6121 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6122 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6123 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6124 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6125 (trunc<mode>sf2_hw): Change the truncate with round to odd
6126 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6127 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6128 to odd hardware instructions.
6129 (sub<mode>3_odd): Likewise.
6130 (mul<mode>3_odd): Likewise.
6131 (div<mode>3_odd): Likewise.
6132 (sqrt<mode>2_odd): Likewise.
6133 (fma<mode>4_odd): Likewise.
6134 (fms<mode>4_odd): Likewise.
6135 (nfma<mode>4_odd): Likewise.
6136 (nfms<mode>4_odd): Likewise.
6137 (trunc<mode>df2_odd): Change the truncate with round to odd
6138 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6140 * doc/extend.texi (PowerPC built-in functions): Update documentation
6141 for existing IEEE float128-bit built-in functions. Add built-in
6142 functions that generate the IEEE 128-bit floating point round to
6145 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6147 PR rtl-optimization/77729
6148 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6149 to (X&(C1&~C2))|C2 transformations.
6151 2017-10-03 Martin Jambor <mjambor@suse.cz>
6153 PR tree-optimization/82363
6154 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6155 mismatch, mark lacc written regardless of racc.
6157 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6159 PR tree-optimization/82381
6160 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6161 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6162 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6163 fallthrough into reassoc_stmt_dominates_stmt_p.
6166 * combine.c (combine_instructions): Don't combine in unreachable
6169 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6172 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6173 function to not use the have_cpu variable. Do not set cpu_index,
6174 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6176 (rs6000_valid_attribute_p): Remove duplicate initializations of
6177 old_optimize and func_optimize.
6178 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6179 (rs6000_activate_target_options): Make global.
6180 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6183 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6185 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6186 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6187 Set *poffset to 0 and *psize and *pmax_size to -1 if
6188 *poffset + *psize overflows in HOST_WIDE_INT.
6190 PR tree-optimization/82387
6191 PR tree-optimization/82388
6192 PR tree-optimization/82389
6193 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6194 instead of live_bytes non-NULL.
6196 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6199 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6202 2017-10-02 Richard Biener <rguenther@suse.de>
6204 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6205 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6206 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6208 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6210 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6211 requested precision matches the type's.
6212 * calls.c (alloc_max_size): Calculate the new candidate size as
6213 a widest_int and use wi::to_widest when comparing it with the
6214 current candidate size.
6215 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6216 zero rather than integer_zero_node.
6217 * match.pd: Check for a no-op conversion before using wi::add
6218 rather than after. Use tree_to_uhwi when summing small shift
6219 counts into an unsigned int.
6221 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6222 Alan Hayward <alan.hayward@arm.com>
6223 David Sherwood <david.sherwood@arm.com>
6226 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6227 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6228 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6229 POINTER_AND_FP_REGS.
6231 2017-10-02 Richard Biener <rguenther@suse.de>
6233 PR tree-optimization/82355
6234 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6235 a mapping for the enclosing loop but avoid generating one for
6237 (copy_bb_and_scalar_dependences): Remove premature codegen
6238 error on PHIs in blocks duplicated into multiple places.
6239 * graphite-scop-detection.c
6240 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6241 in the region use it as loop and nest to analyze the DR in.
6242 (try_generate_gimple_bb): Likewise.
6243 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6244 (add_loop_constraints): For blocks in a loop not in the region
6245 create a dimension with a single iteration.
6246 * sese.h (gbb_loop_at_index): Remove assert.
6248 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6250 * omp-expand.c (adjust_context_scope): New function.
6251 (expand_parallel_call): Call adjust_context_scope.
6253 2017-10-01 Jeff Law <law@redhat.com>
6255 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6256 dom_opt_dom_walker class with direct access to private members.
6257 Add comments. Call test_for_singularity.
6258 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6259 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6260 m_dummy_cond anymore.
6261 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6263 (pass_dominator:execute): Build the dummy_cond here and pass it
6264 to the dom_opt_dom_walker ctor.
6265 (test_for_singularity): New function.
6267 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6268 Maya Rashish <coypu@sdf.org>
6270 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6271 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6272 (alpha*-*-netbsd*) Use nbsd_tm_file.
6273 (arm*-*-netbsdelf*) Likewise.
6274 (i[34567]86-*-netbsdelf*) Likewise.
6275 (x86_64-*-netbsd*) Likewise.
6276 (mips*-*-netbsd*) Likewise.
6277 (powerpc-*-netbsd*) Likewise.
6278 (sh*-*-netbsd*) Likewise.
6279 (sparc-*-netbsdelf*) Likewise.
6280 (sparc64-*-netbsd*) Likewise.
6281 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6283 (vax-*-netbsdelf*) Likewise.
6284 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6285 (UINT_FAST8_TYPE) Likewise.
6286 (INT_FAST16_TYPE) Check CHAR_FAST16.
6287 (UINT_FAST16_TYPE) Likewise.
6289 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6292 * config/i386/i386.md
6293 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6294 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6295 *divmodsi4_zext_2): New define_insn_and_split.
6296 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6297 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6298 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6299 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6300 New define_insn_and_split.
6301 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6302 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6303 operands[1] having DImode when mode is SImode.
6305 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6306 always SImode for DIV and MOD in REG_EQUAL notes.
6308 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6311 * match.pd: Fix handling of NaNs in pattern.
6313 2017-09-29 Jeff Law <law@redhat.com>
6315 * sbitmap.c (bitmap_bit_in_range_p): New function.
6316 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6317 * tree-ssa-dse.c (live_bytes_read): New function.
6318 (dse_classify_store): Ignore reads of dead bytes.
6320 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6321 typos and whitespace errors.
6322 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6323 * config/s390/s390.c (s390_emit_prologue): Likewise.
6325 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6328 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6329 with a symbol for LRA.
6331 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6333 PR rtl-optimization/82338
6334 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6336 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6338 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6339 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6340 (qsort_chk): Declare.
6341 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6342 (qsort_chk): New function.
6344 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6346 PR tree-optimization/82337
6347 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6348 phi definition if the PHI result appears in an abnormal PHI.
6349 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6350 basis appears in an abnormal PHI.
6352 2017-09-29 Richard Biener <rguenther@suse.de>
6354 * graphite-isl-ast-to-gimple.c
6355 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6356 (binary_op_to_tree): Use it.
6357 (get_rename_from_scev): Likewise.
6358 (copy_loop_phi_nodes): Likewise.
6359 (copy_bb_and_scalar_dependences): Likewise.
6360 (translate_pending_phi_nodes): Likewise.
6362 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6365 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6366 for movabsq $(i32 << shift), r64.
6368 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6370 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6371 index when encoding %esp as %rsp to avoid 0x67 prefix.
6373 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6375 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6376 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6378 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6380 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6381 Extensions with more than 16 double VFP registers.
6382 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6383 to_clear_mask and all code related to it. Replace the remaining
6384 entry by a sbitmap and adapt code accordingly.
6386 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6388 * brig-builtins.def: Change pure attributes to const.
6390 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6392 * config.gcc (default_gnu_indirect_function): Default to yes for
6393 sparc*-*-linux* with glibc.
6395 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6397 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6398 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6399 when creating .init_array and .fini_array sections with priority
6402 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6405 * config/aarch64/aarch64.c
6406 (aarch64_builtin_support_vector_misalignment): Always return false
6407 when misalignment is unknown.
6409 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6411 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6412 this function to return false if the definition used by the swap
6413 instruction is artificial, or if the memory address from which the
6414 constant value is loaded is not represented by a base address held
6415 in a register or if the base address register is a frame or stack
6416 pointer. Additionally, return false if the base address of the
6417 loaded constant is a SYMBOL_REF but is not considered to be a
6419 (replace_swapped_load_constant): New function.
6420 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6421 loaded constant vector with a load of a swapped constant vector.
6423 2017-09-27 Carl Love <cel@us.ibm.com>
6425 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6426 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6427 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6430 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6432 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6433 first, always call autopref_rank_data otherwise.
6435 2017-09-27 Richard Biener <rguenther@suse.de>
6437 * graphite-scop-detection.c (find_scop_parameters): Move
6438 loop bound handling ...
6439 (gather_bbs::before_dom_children): ... here, avoiding the need
6440 to build scop_info->loop_nest.
6441 (record_loop_in_sese): Remove.
6442 * sese.h (sese_info_t::loop_nest): Remove.
6443 * sese.c (new_sese_info): Do not allocate loop_nest.
6444 (free_sese_info): Do not free loop_nest.
6446 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6449 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6450 lhs from calls if the lhs has addressable type.
6452 2017-09-27 Richard Biener <rguenther@suse.de>
6454 * graphite.h (scop::max_alias_set): New member.
6455 * graphite-scop-detection.c: Remove references to non-existing
6456 --param in comments.
6457 (build_alias_sets): Record the maximum alias set used for drs.
6458 (build_scops): Support zero as unlimited for
6459 --param graphite-max-arrays-per-scop.
6460 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6462 (build_poly_sr_1): ... here. Compute alias set based on the
6463 maximum alias set used for drs rather than
6464 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6466 2017-09-27 Richard Biener <rguenther@suse.de>
6468 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6469 --param loop-block-tile-size=0 to disable tiling.
6471 2017-09-27 Richard Biener <rguenther@suse.de>
6473 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6474 (graphite-max-nb-scop-params): Document special value zero.
6475 * domwalk.h (dom_walker::STOP): New symbolical constant.
6476 (dom_walker::dom_walker): Add optional parameter for bb to
6478 (dom_walker::~dom_walker): Declare.
6479 (dom_walker::before_dom_children): Document STOP return value.
6480 (dom_walker::m_user_bb_to_rpo): New member.
6481 (dom_walker::m_bb_to_rpo): Likewise.
6482 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6483 mapping here if not provided by the user.
6484 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6485 provided by the user.
6486 (dom_walker::STOP): Define.
6487 (dom_walker::walk): Do not compute bb to RPO mapping here.
6488 Support STOP return value from before_dom_children to stop
6490 * graphite-optimize-isl.c (optimize_isl): If the schedule
6491 is the same still generate code if -fgraphite-identity
6492 or -floop-parallelize-all are given.
6493 * graphite-scop-detection.c: Include cfganal.h.
6494 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6496 (gather_bbs::before_dom_children): Return STOP for BBs
6498 (build_scops): Compute bb to RPO mapping and pass it to
6499 the domwalk. Treat --param graphite-max-nb-scop-params=0
6500 as not limiting the number of params.
6501 * graphite.c (graphite_initialize): Remove limit on the
6502 number of basic-blocks in a function.
6503 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6504 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6505 default value of 10.
6507 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6509 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6510 Adjust code to eliminate needing to do the shift right 32-bits
6511 operation after XSCVDPSPN.
6513 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6515 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6516 ((X / Y) != 0 -> X >= Y): Likewise.
6518 2017-09-26 Carl Love <cel@us.ibm.com>
6520 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6521 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6522 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6523 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6524 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6525 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6526 definitions and overloading.
6527 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6528 statement for P9V_BUILTIN_XST_LEN_R.
6529 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6530 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6531 define_expand and define_insn for the instructions and builtins.
6532 * doc/extend.texi: Update the built-in documentation file for the new
6534 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6535 define_insn for the instructions
6537 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6540 * gcc/config/netbsd-protos.h: New file.
6541 * gcc/config/netbsd.c: New file.
6542 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6543 * gcc/config/t-netbsd: New file.
6544 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6545 (tmake_file) Add t-netbsd.
6546 (extra_objs) Add netbsd.o.
6548 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6552 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6554 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6556 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6557 sign extension from a vector register to a GPR by doing a 32-bit
6558 direct move and then an EXTSW.
6559 (extendsi<mode>2 splitter): Likewise.
6560 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6561 right or vector extract after doing XSCVDPSPN. Use
6562 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6564 (movdi_from_sf_zero_ext): Likewise.
6565 (reload_gpr_from_vsxsf): Likewise.
6566 (p8_mfvsrd_4_disf): Delete, no longer used.
6567 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6568 then needing to move the SFmode to a GPR to use the XSCVDPSP
6569 instruction instead of FRSP and XSCVDPSPN.
6570 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6571 it is adjacent to the other XSCVSPDP insns.
6572 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6573 SFmode to be in traditional Altivec registers.
6574 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6575 (vsx_xscvspdpn): Likewise.
6576 (vsx_xscvspdpn_scalar): Likewise.
6578 2017-09-26 Martin Jambor <mjambor@suse.cz>
6580 * tree-sra.c (compare_access_positions): Put integral types first,
6581 stabilize sorting of integral types, remove conditions putting
6582 non-full-precision integers last.
6583 (sort_and_splice_var_accesses): Disable scalarization if a
6584 non-integert would be represented by a non-full-precision integer.
6586 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6588 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6589 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6590 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6591 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6592 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6593 conditionals inside the function instead of around it. Call
6594 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6595 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6597 2017-09-26 Richard Biener <rguenther@suse.de>
6599 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6601 (scop_detection::build_scop_breadth): ... this. Removed.
6602 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6603 (scop_detection::harmful_stmt_in_bb): Likewise.
6604 (scop_detection::graphite_can_represent_stmt): Likewise.
6605 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6606 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6607 (scop_detection::can_represent_loop_1): ... this. Removed.
6608 (scop_detection::harmful_loop_in_region): Simplify after inlining
6609 the above and remove more quadraticness.
6610 (build_scops): Adjust.
6611 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6614 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6617 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6618 REGNO (base) == SP_REG if base is a REG.
6621 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6622 if it is different SSA_NAME.
6623 (optimize_range_tests_cmp_bitwise): New function.
6624 (optimize_range_tests): Call it.
6626 2017-09-26 Richard Biener <rguenther@suse.de>
6628 PR tree-optimization/82321
6629 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6630 for the def being inside the loop.
6632 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6634 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6636 * config/s390/s390-builtins.def: Fix constraint on op4.
6638 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6640 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6641 independent expanders.
6642 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6643 ("vec_ordered", "vec_unordered"): New expanders.
6645 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6647 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6650 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6652 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6653 vec_unpacks_lo_v16qi.
6654 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6656 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6658 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6659 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6660 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6662 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6664 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6666 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6668 ("*vec_slb<mode>"): New insn pattern.
6669 ("vec_shr_<mode>"): New expander.
6670 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6671 and force the shift count operand to V16QImode.
6672 ("vec_srb<mode>"): Set shift count mode to V16QI.
6674 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6676 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6677 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6678 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6680 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6683 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6685 2017-09-26 Richard Biener <rguenther@suse.de>
6687 PR tree-optimization/82320
6688 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6691 2017-09-25 Jeff Law <law@redhat.com>
6693 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6694 prototype for new argument.
6695 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6696 mostly extracted from rs6000_emit_allocate_stack.
6697 (rs6000_emit_probe_stack_range_stack_clash): New function.
6698 (rs6000_emit_allocate_stack): Call
6699 rs6000_emit_probe_stack_range_stack_clash as needed.
6700 (rs6000_emit_probe_stack_range): Add additional argument
6701 to call to gen_probe_stack_range{si,di}.
6702 (output_probe_stack_range): New.
6703 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6704 (output_probe_stack_range_stack_clash): New.
6705 (rs6000_emit_prologue): Emit notes into dump file as requested.
6706 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6707 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6708 Add additional operand and pass it to output_probe_stack_range.
6710 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6712 PR tree-optimization/82163
6713 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6714 (checking_verify_loop_closed_ssa): New parameter.
6715 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6716 (check_loop_closed_ssa_stmt): Delete.
6717 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6718 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6719 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6722 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6724 * brig-builtins.def: Treat HSAIL barrier builtins as
6725 setjmp/longjump style functions.
6727 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6729 * target.def (constant_alignment): New hook.
6730 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6731 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6732 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6733 * doc/tm.texi: Regenerate.
6734 * targhooks.h (default_constant_alignment): Declare.
6735 (constant_alignment_word_strings): Likewise.
6736 * targhooks.c (default_constant_alignment): New function.
6737 (constant_alignment_word_strings): Likewise.
6738 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6739 instead of CONSTANT_ALIGNMENT.
6740 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6741 (force_const_mem): Likewise.
6742 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6743 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6744 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6745 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6746 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6748 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6749 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6750 constant_alignment_word_strings.
6751 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6752 (CONSTANT_ALIGNMENT): Likewise.
6753 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6754 (arm_constant_alignment): New function.
6755 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6756 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6757 constant_alignment_word_strings.
6758 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6759 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6760 constant_alignment_word_strings.
6761 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6762 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6763 (cris_constant_alignment): New function.
6764 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6765 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6766 (epiphany_constant_alignment): New function.
6767 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6768 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6769 constant_alignment_word_strings.
6770 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6771 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6772 constant_alignment_word_strings.
6773 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6774 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6775 constant_alignment_word_strings.
6776 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6777 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6778 * config/i386/i386.c (ix86_constant_alignment): Make static.
6779 Use the same interface as the target hook.
6780 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6781 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6782 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6783 constant_alignment_word_strings.
6784 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6785 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6786 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6787 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6788 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6789 constant_alignment_word_strings.
6790 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6791 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6792 constant_alignment_word_strings.
6793 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6794 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6795 constant_alignment_word_strings.
6796 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6797 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6799 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6800 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6801 * config/mips/mips.c (mips_constant_alignment): New function.
6802 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6803 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6804 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6805 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6806 (mmix_constant_alignment): Make static. Use the same interface
6808 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6809 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6810 constant_alignment_word_strings.
6811 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6812 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6813 constant_alignment_word_strings.
6814 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6815 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6816 constant_alignment_word_strings.
6817 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6818 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6819 (rs6000_constant_alignment): New function.
6820 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6821 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6822 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6823 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6824 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6825 (rs6000_constant_alignment): New function.
6826 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6827 * config/s390/s390.c (s390_constant_alignment): New function.
6828 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6829 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6830 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6831 constant_alignment_word_strings.
6832 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6833 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6834 (sparc_constant_alignment): New function.
6835 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6836 * config/spu/spu.c (spu_constant_alignment): New function.
6837 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6838 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6839 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6840 constant_alignment_word_strings.
6841 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6842 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6843 constant_alignment_word_strings.
6844 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6845 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6846 constant_alignment_word_strings.
6847 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6848 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6849 (visium_constant_alignment): New function.
6850 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6851 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6852 (xtensa_constant_alignment): New function.
6853 * system.h (CONSTANT_ALIGNMENT): Poison.
6855 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6857 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6858 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6859 (rs6000_builtin_valid_without_lhs): New helper function.
6860 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6861 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6863 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6865 * target.h (vec_perm_indices): Use unsigned short rather than
6867 (auto_vec_perm_indices): Likewise.
6868 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6869 Use unsigned int rather than unsigned char.
6870 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6872 2017-09-25 Richard Biener <rguenther@suse.de>
6874 * cfgloop.h (sort_sibling_loops): Declare.
6875 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6876 (sort_sibling_loops): New function sorting the sibling loop list
6878 * graphite.c (graphite_transform_loops): Sort sibling loops.
6880 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6882 * target.def (vec_perm_const_ok): Change sel parameter to
6884 * optabs-query.c (can_vec_perm_p): Update accordingly.
6885 * doc/tm.texi: Regenerate.
6886 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6887 auto_vec_perm_indices and remove separate nelt field.
6888 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6889 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6890 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6891 (aarch64_expand_vec_perm_const): Update accordingly.
6892 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6893 to vec_perm_indices.
6894 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6895 auto_vec_perm_indices and remove separate nelt field.
6896 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6897 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6898 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6900 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6901 to vec_perm_indices.
6902 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6903 sel to vec_perm_indices.
6904 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6905 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6906 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6908 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6911 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6914 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6915 on the FUNCTION_DECL function context if it has a DIE that is a
6918 2017-09-25 Richard Biener <rguenther@suse.de>
6920 PR tree-optimization/82285
6921 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6924 2017-09-25 Tom de Vries <tom@codesourcery.com>
6928 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6931 2017-09-25 Richard Biener <rguenther@suse.de>
6933 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6934 ISL errors other than isl_error_quota happen. Dump if the
6935 schedule is the same.
6936 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6937 errors instead of aborting inside ISL.
6939 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6942 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6943 of libgcc_eh for m64.
6944 * config/i386/darwin64.h: Likewise.
6946 2017-09-25 Richard Biener <rguenther@suse.de>
6949 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6950 attribute for incomplete types nor twice for complete ones.
6952 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6955 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6956 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6959 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6962 * config/i386/i386.opt (mprefer-avx256): Use
6963 ix86_target_flags variable.
6964 * config/i386/i386.c (ix86_target_string): Move
6965 -mprefer-avx256 to flag2_opts.
6967 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6970 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6971 and x != -1 | y != -1 into (x & y) != -1.
6973 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6975 * config.gcc: Add new case statement to set
6976 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6977 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6978 s390x-*-linux* case statements. Added aarch64 to the list of
6979 supported architectures.
6981 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6983 PR tree-optimization/82289
6984 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6985 STMT_VINFO_RELEVANT_P.
6987 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6988 Alan Hayward <alan.hayward@arm.com>
6989 David Sherwood <david.sherwood@arm.com>
6991 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6992 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6993 (extract_range_from_binary_expr_1): Don't call
6994 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6996 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6997 Alan Hayward <alan.hayward@arm.com>
6998 David Sherwood <david.sherwood@arm.com>
7000 * target.def (preferred_vector_alignment): New hook.
7001 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7003 * doc/tm.texi: Regenerate.
7004 * targhooks.h (default_preferred_vector_alignment): Declare.
7005 * targhooks.c (default_preferred_vector_alignment): New function.
7006 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7008 (DR_TARGET_ALIGNMENT): New macro.
7009 (aligned_access_p): Update commentary.
7010 (vect_known_alignment_in_bytes): New function.
7011 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7013 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7014 Calculate the misalignment based on the target alignment rather than
7016 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7017 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7018 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7019 the target alignment, rather than masking the element misalignment
7020 with the number of elements in a vector. Also use the target
7021 alignment when calculating the maximum number of peels.
7022 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7023 instead of TYPE_ALIGN_UNIT.
7024 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7025 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7026 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7027 (vect_create_data_ref_ptr): Likewise.
7028 (vect_setup_realignment): Realign by ANDing with
7029 -DR_TARGET_MISALIGNMENT.
7030 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7031 the number of peels based on DR_TARGET_ALIGNMENT.
7032 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7033 with the guaranteed alignment boundary when deciding whether
7035 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7036 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7037 (ensure_base_align): Remove stmt_info parameter. Get the
7038 target base alignment from DR_TARGET_ALIGNMENT.
7039 (vectorizable_store): Update call accordingly. Interpret
7040 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7042 (vectorizable_load): Likewise.
7044 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7045 Alan Hayward <alan.hayward@arm.com>
7046 David Sherwood <david.sherwood@arm.com>
7048 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7049 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7050 (vect_enhance_data_refs_alignment): Likewise.
7052 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7054 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7055 error. Only quit immediately if parsing is complete.
7056 (BEGIN): Initialize fatal_err and parse_done.
7057 (begin fpu, end fpu): Check number of arguments.
7058 (begin arch, end arch): Likewise.
7059 (begin cpu, end cpu): Likewise.
7060 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7061 (optalias): Likewise.
7063 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7065 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7066 * config/arm/arm-isa.h: Delete. Move definitions to ...
7067 * arm-cpus.in: ... here. Use new feature and fgroup values.
7068 * config/arm/arm.c (arm_option_override): Use lower case for feature
7070 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7071 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7072 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7074 (print_isa_bits_for): New function.
7075 (gen_isa): New function.
7076 (gen_comm_data): Use print_isa_bits_for.
7077 (define feature): New keyword.
7078 (define fgroup): New keyword.
7079 * config/arm/t-arm (TM_H): Remove.
7080 (GTM_H): Add arm-isa.h.
7081 (arm-isa.h): Add rule to generate file.
7082 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7083 case for feature bit names.
7085 2017-09-22 Richard Biener <rguenther@suse.de>
7087 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7089 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7090 print of no dependency loops ...
7091 * graphite.c (graphite_transform_loops): ... here.
7092 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7094 (same_close_phi_node, remove_duplicate_close_phi,
7095 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7096 (canonicalize_loop_closed_ssa): ... here and simplify.
7097 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7098 (optimize_isl): Use dump_printf_loc to tell when we stopped
7099 optimizing because of an ISL timeout.
7101 2017-09-22 Richard Biener <rguenther@suse.de>
7103 PR tree-optimization/82291
7104 * tree-if-conv.c (predicate_mem_writes): Make sure to
7105 remove writes in blocks predicated with false.
7107 2017-09-22 Richard Biener <rguenther@suse.de>
7109 * sese.c: Include cfganal.h.
7110 (if_region_set_false_region): Remove.
7111 (create_if_region_on_edge): Likewise.
7112 (move_sese_in_condition): Re-implement without destroying
7115 2017-09-22 Richard Biener <rguenther@suse.de>
7117 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7118 Verify both BBs contain loop PHI nodes before dispatching to
7120 (graphite_regenerate_ast_isl): Do not recompute dominators,
7121 do not verify three times. Restructure for clarity.
7122 * graphite-scop-detection.c (same_close_phi_node,
7123 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7124 defined_in_loop_p, canonicalize_loop_closed_ssa,
7125 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7126 checking and SSA rewrite, move to ...
7127 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7128 (graphite_initialize): Do not pass in ctx, do not reset the
7129 SCEV cache, compute only dominators.
7130 (graphite_transform_loops): Allocate ISL ctx after
7131 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7132 Maintain post-dominators only around build_scops.
7133 * sese.c (if_region_set_false_region): Make static. Free
7134 and recompute dominators.
7135 (move_sese_in_condition): Assert we don't get called with
7136 post-dominators computed.
7137 * sese.h (if_region_set_false_region): Remove.
7139 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7141 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7142 mode attribute for TARGET_AVX512VL.
7144 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7146 * config/i386/i386.opt (mprefer-avx256): New option.
7147 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7149 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7150 for TARGET_PREFER_AVX256.
7151 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7153 2017-09-21 Jeff Law <law@redhat.com>
7155 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7156 Fix dump output if the only stack space is for pushed registers.
7158 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7160 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7163 2017-09-21 Martin Sebor <msebor@redhat.com>
7166 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7167 code (in C++) or code that triggers warnings.
7169 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7171 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7173 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7175 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7176 * sched-rgn.c: Ditto.
7177 * sel-sched-ir.c: Ditto.
7179 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7181 * toplev.h (set_random_seed): Adjust return type.
7182 * toplev.c (init_local_tick): Move eager initialization of random_seed
7183 to get_random_seed. Adjust comment.
7184 (init_random_seed): Inline to get_random_seed, delete.
7185 (get_random_seed): Initialize random_seed lazily.
7186 (set_random_seed): Do not return previous value.
7187 (print_switch_value): Do not call get_random_seed.
7189 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7191 * cgraph.c (delete_function_version): New, broken out from...
7192 (cgraph_node::delete_function_version): ...here. Rename to
7193 cgraph_node::delete_function_version_by_decl. Update all uses.
7194 (cgraph_node::remove): Call delete_function_version.
7196 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7199 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7200 VAR_DECLs to which addressable non-volatile parameters are mapped
7201 and for id->retvar after the return value assignment. Clear
7202 id->retval and id->retbnd after inlining.
7204 2017-09-21 Richard Biener <rguenther@suse.de>
7206 PR tree-optimization/82276
7207 PR tree-optimization/82244
7208 * tree-vrp.c (build_assert_expr_for): Set
7209 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7211 (remove_range_assertions): Revert earlier change.
7213 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7216 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7218 2017-09-21 Richard Biener <rguenther@suse.de>
7220 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7221 Restore valid IL after code generation errors.
7222 * graphite.c (graphite_transform_loops): Diagnose code
7223 generation issues as MSG_MISSED_OPTIMIZATION and continue
7224 with processing SCOPs.
7226 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7227 Alan Hayward <alan.hayward@arm.com>
7228 David Sherwood <david.sherwood@arm.com>
7230 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7231 rather than choosing between plus_constant and gen_rtx_<CODE>.
7232 * expr.c (emit_push_insn): Likewise.
7233 (expand_expr_real_2): Likewise.
7235 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7236 Alan Hayward <alan.hayward@arm.com>
7237 David Sherwood <david.sherwood@arm.com>
7239 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7241 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7242 Alan Hayward <alan.hayward@arm.com>
7243 David Sherwood <david.sherwood@arm.com>
7245 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7246 calling tree_to_uhwi.
7248 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7249 Alan Hayward <alan.hayward@arm.com>
7250 David Sherwood <david.sherwood@arm.com>
7252 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7253 INTEGER_CST rather than a negative test for ADDR_EXPR.
7255 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7256 Alan Hayward <alan.hayward@arm.com>
7257 David Sherwood <david.sherwood@arm.com>
7259 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7260 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7262 2017-09-21 Richard Biener <rguenther@suse.de>
7264 PR tree-optimization/71351
7265 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7266 graphite_create_new_loop_guard): Remove, fold remaining parts
7268 (translate_isl_ast_node_for): ... here and simplify.
7270 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7273 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7274 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7275 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7276 alternative always use QI mode, for -Os imov (=R,R) alternative
7277 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7280 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7281 Jeff Law <law@redhat.com>
7283 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7284 (allocate_stack_space): New function, partially extracted from
7286 (s390_emit_prologue): Track offset to most recent stack probe.
7287 Code to allocate space moved into allocate_stack_space.
7288 Dump actions when no stack is allocated.
7289 (s390_prologue_plus_offset): New function.
7290 (s390_emit_stack_probe): Likewise.
7292 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7294 * common.opt (Wa, Wl, Wp, g, gz=): Add
7296 (gno-column-info): Remove.
7297 (gcolumn-info): Drop RejectNegative.
7299 (gno-record-gcc-switches): Remove.
7300 (grecord-gcc-switches): Drop RejectNegative.
7301 (gno-split-dwarf): Remove.
7302 (gsplit-dwarf): Drop RejectNegative.
7303 (gno-strict-dwarf): Remove.
7304 (gstrict-dwarf): Drop RejectNegative.
7305 * config/darwin.opt (gfull, gused): Add RejectNegative.
7306 * dwarf2out.c (gen_producer_string): Drop
7307 gno-record-gcc-switches handler.
7308 * optc-gen.awk: Add g to prefixes with negative forms.
7309 * opts-common.c (remapping_prefix_p): New.
7310 (find_opt): Check it.
7311 (generate_canonical_option): Test g prefix.
7312 (option_map): Add -gno- mapping.
7313 (add_misspelling_candidates): Check remapping_prefix_p.
7315 2017-09-20 Jeff Law <law@redhat.com>
7317 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7318 thinko in stack clash protection support.
7320 * explow.c (compute_stack_clash_protection_loop_data): Use
7321 CONST_INT_P instead of explicit test. Verify object is a
7322 CONST_INT_P before looking at INTVAL.
7323 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7324 instead of explicit test.
7326 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7329 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7330 address instead of to r1 and r11.
7332 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7334 * config.gcc: Support "knm".
7335 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7336 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7338 * config/i386/i386.c (m_KNM): Define.
7339 (processor_target_table): Add "knm".
7341 (ix86_option_override_internal): Add "knm".
7342 (ix86_issue_rate): Add PROCESSOR_KNM.
7343 (ix86_adjust_cost): Ditto.
7344 (ia32_multipass_dfa_lookahead): Ditto.
7345 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7346 (fold_builtin_cpu): Add M_INTEL_KNM.
7347 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7348 (processor_type): Add PROCESSOR_KNM.
7349 * config/i386/x86-tune.def: Add m_KNM.
7350 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7352 2017-09-20 Richard Biener <rguenther@suse.de>
7354 PR tree-optimization/80213
7355 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7356 are allowed in empty BBs as well.
7357 (canonicalize_loop_closed_ssa): Also look for other complex
7359 (scop_detection::get_sese): Include the loop-closed PHI block
7361 (scop_detection::merge_sese): Remove code adding extra blocks.
7362 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7363 (build_scops): Assert the final returned scop is invalid.
7365 2017-09-20 Richard Biener <rguenther@suse.de>
7367 PR tree-optimization/82264
7368 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7370 (vn_phi_lookup): Likewise.
7371 (vn_phi_insert): Likewise.
7373 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7375 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7376 that fits into uhwi or shwi, add DW_AT_const_value regardless
7377 of early_dwarf without going through RTL, using add_AT_unsigned
7380 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7381 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7382 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7383 (DEBUG_MACRO_SECTION): Likewise.
7384 (DEBUG_LTO_MACRO_SECTION): Likewise.
7385 (DEBUG_STR_DWO_SECTION): Likewise.
7386 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7387 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7388 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7389 (DEBUG_LTO_LINE_STR_SECTION): Define.
7390 (init_sections_and_labels): Initialize debug_line_str_section
7391 variable. Initialize debug_loc_section for -gdwarf-5 to
7392 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7394 2017-09-20 Richard Biener <rguenther@suse.de>
7396 * graphite-sese-to-poly.c (extract_affine): Properly handle
7397 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7399 2017-09-20 Richard Biener <rguenther@suse.de>
7401 PR tree-optimization/81373
7402 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7403 Force SESE live-out defs to be handled even if they are
7406 2017-09-19 Jeff Law <law@redhat.com>
7408 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7409 nothing for stack adjustments with REG_STACK_CHECK.
7410 * sched-deps.c (parse_add_or_inc): Reject insns with
7411 REG_STACK_CHECK from dependency breaking.
7412 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7413 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7414 * reg-notes.def (STACK_CHECK): New note.
7416 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7417 (ix86_expand_prologue): Dump stack clash info as needed.
7418 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7420 * function.c (dump_stack_clash_frame_info): New function.
7421 * function.h (dump_stack_clash_frame_info): Prototype.
7422 (enum stack_clash_probes): New enum.
7424 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7425 flag_stack_clash_protection.
7426 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7427 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7428 (arm_frame_pointer_required): Likewise.
7429 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7430 (ia64_expand_prologue): Likewise.
7431 * config/mips/mips.c (mips_expand_prologue): Likewise.
7432 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7433 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7434 (sparc_flat_expand_prologue): Likewise.
7435 * config/spu/spu.c (spu_expand_prologue): Likewise.
7437 * explow.c: Include "params.h".
7438 (anti_adjust_stack_and_probe_stack_clash): New function.
7439 (get_stack_check_protect): Likewise.
7440 (compute_stack_clash_protection_loop_data): Likewise.
7441 (emit_stack_clash_protection_loop_start): Likewise.
7442 (emit_stack_clash_protection_loop_end): Likewise.
7443 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7444 Use anti_adjust_stack_and_probe_stack_clash.
7445 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7446 (emit_stack_clash_protection_loop_start): Likewise.
7447 (emit_stack_clash_protection_loop_end): Likewise.
7448 * rtl.h (get_stack_check_protect): Prototype.
7449 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7450 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7451 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7453 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7455 * doc/tm.texi: Rebuilt.
7456 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7457 get_stack_check_protect.
7458 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7459 * config/arm/arm.c (arm_expand_prologue): Likewise.
7460 (arm_frame_pointer_required): Likewise.
7461 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7462 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7463 * config/mips/mips.c (mips_expand_prologue): Likewise.
7464 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7465 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7466 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7467 (sparc_flat_expand_prologue): Likewise.
7469 * common.opt (-fstack-clash-protection): New option.
7470 * flag-types.h (enum stack_check_type): Note difference between
7471 -fstack-check= and -fstack-clash-protection.
7472 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7473 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7474 * toplev.c (process_options): Issue warnings/errors for cases
7475 not handled with -fstack-clash-protection.
7476 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7477 (-fstack-check): Note additional problem with -fstack-check=generic.
7478 Note that -fstack-check is primarily for Ada and refer users
7479 to -fstack-clash-protection for stack-clash-protection.
7480 Document new params for stack clash protection.
7482 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7484 * config/i386/i386.c (ix86_split_long_move): Do not handle
7485 address used for LEA in a special way.
7487 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7489 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7491 2017-09-19 Martin Sebor <msebor@redhat.com>
7494 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7495 of incompatible types.
7497 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7499 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7500 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7501 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7502 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7504 2017-09-19 Richard Biener <rguenther@suse.de>
7506 PR tree-optimization/82244
7507 * tree-vrp.c (remove_range_assertions): Do not propagate
7508 a constant to abnormals but replace the assert with a copy.
7510 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7512 PR rtl-optimization/57878
7513 PR rtl-optimization/68988
7514 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7515 avoidance test involving non_reload_pseudos. Move frequency test
7516 below the general fragmentation avoidance test.
7518 2017-09-19 Richard Biener <rguenther@suse.de>
7520 PR tree-optimization/69728
7521 * graphite-sese-to-poly.c (schedule_error): New global.
7522 (add_loop_schedule): Handle empty domain by failing the
7524 (build_original_schedule): Handle schedule_error.
7526 2017-09-19 Richard Biener <rguenther@suse.de>
7528 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7529 Do not iterate to sibling loops but only to siblings of inner
7532 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7535 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7536 that the registers are different.
7538 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7540 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7541 to processor_model and "amdfam17h" to arch_names_table.
7542 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7544 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7547 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7549 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7550 Alan Hayward <alan.hayward@arm.com>
7551 David Sherwood <david.sherwood@arm.com>
7553 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7555 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7557 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7558 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7559 vect_schedule_slp_instance.
7560 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7561 Update call to vect_slp_analyze_node_operations. Simplify return
7563 (vect_slp_analyze_bb_1): Update call accordingly.
7564 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7565 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7566 (vect_schedule_slp): Update call accordingly.
7568 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7569 Alan Hayward <alan.hayward@arm.com>
7570 David Sherwood <david.sherwood@arm.com>
7572 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7573 with types that aren't in fact scalar.
7575 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7577 * tree-vect-slp.c (vect_record_max_nunits): New function,
7579 (vect_build_slp_tree_1): ...here.
7580 (vect_build_slp_tree_2): Call it for phis too.
7582 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7584 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7585 to vect_get_vec_def_for_operand when getting the mask operand.
7587 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7588 Alan Hayward <alan.hayward@arm.com>
7589 David Sherwood <david.sherwood@arm.com>
7591 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7594 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7595 Alan Hayward <alan.hayward@arm.com>
7596 David Sherwood <david.sherwood@arm.com>
7598 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7599 calculation for vector booleans.
7601 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7602 Alan Hayward <alan.hayward@arm.com>
7603 David Sherwood <david.sherwood@arm.com>
7605 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7607 (vect_transform_stmt): ...here.
7608 (vect_analyze_stmt): Use it instead of calling
7609 vectorizable_live_operation directly.
7611 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7613 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7614 non-SIMT targets in acc vector loops.
7616 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7618 * configure.ac: Add arc and check if assembler supports gdwarf2.
7619 * configure: Regenerate.
7621 2017-09-18 Richard Biener <rguenther@suse.de>
7623 PR tree-optimization/82220
7624 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7625 epilogue niters from the min_profitable_iters compute.
7627 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7630 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7631 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7632 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7634 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7637 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7638 switching to a new text section.
7640 2017-09-18 Richard Biener <rguenther@suse.de>
7642 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7644 (build_alias_set): Reject aliases with no access function.
7646 2017-09-18 Richard Biener <rguenther@suse.de>
7648 PR tree-optimization/79622
7649 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7651 (build_cross_bb_scalars_use): Likewise.
7653 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7655 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7657 2017-09-18 Alan Modra <amodra@gmail.com>
7660 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7661 stack_pointer_rtx for count 0. Update comments. Break up
7662 large rtl expression.
7664 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7666 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7667 Increase to 20 bytes.
7668 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7669 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7670 or avx version of the stub.
7672 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7675 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7676 compute the minimum stack alignment. Also update preferred stack
7677 boundary for leaf functions.
7679 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7681 PR tree-optimization/82228
7682 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7685 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7687 * common/config/nds32/nds32-common.c
7688 (nds32_option_optimization_table): Refine formatting.
7689 (nds32_option_optimization_table): Use -fsched-pressure and
7690 -fomit-frame-pointer for specific optimization level.
7692 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7694 * config/nds32/nds32.c: Refine formatting and comments.
7695 * config/nds32/nds32.h: Likewise.
7696 * config/nds32/nds32.md: Likewise.
7697 * config/nds32/nds32-cost.c: Likewise.
7698 * config/nds32/nds32-isr.c: Likewise.
7699 * config/nds32/nds32-md-auxiliary.c: Likewise.
7700 * config/nds32/nds32-multiple.md: Likewise.
7701 * config/nds32/nds32-predicates.c: Likewise.
7703 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7704 Jakub Jelinek <jakub@redhat.com>
7706 Add support for -std=c++2a.
7707 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7709 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7711 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7714 * doc/extend.texi (Common Function Attributes): Add
7715 references to ARM, AArch64, and S/390 specific attributes.
7716 (Function Specific Option Pragmas): Add AArch64 and S/390
7717 to list of back ends that support the target pragma.
7719 2017-09-15 Nathan Sidwell <nathan@acm.org>
7721 * doc/standards.texi: Fix C++17 description. Update URLs for
7724 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7726 * common.opt (Wcast-align=strict): New warning option.
7727 * doc/invoke.texi: Document -Wcast-align=strict.
7729 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7731 * cgraph.h (cgraph_thunk_info): Add comments.
7732 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7733 assert for VIRTUAL_* arguments stricter.
7735 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7737 PR tree-optimization/71026
7738 * match.pd: Move RDIV patterns from fold-const.c
7739 * fold-const.c (distribute_real_division): Removed.
7740 (fold_binary_loc): Remove calls to distribute_real_divison.
7742 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7744 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7745 c++1z and gnu++1z as deprecated. Change other references to
7746 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7747 Change -Wc++1z-compat to -Wc++17-compat.
7748 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7749 * dwarf2out.c (highest_c_language): Handle C++17.
7750 (gen_compile_unit_die): Likewise.
7752 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7754 PR rtl-optimization/82192
7755 * combine.c (make_extraction): Don't look through non-paradoxical
7756 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7759 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7760 Alan Hayward <alan.hayward@arm.com>
7761 David Sherwood <david.sherwood@arm.com>
7763 * target.def (function_arg_offset): New hook.
7764 * targhooks.h (default_function_arg_offset): Declare.
7765 * targhooks.c (default_function_arg_offset): New function.
7766 * function.c (locate_and_pad_parm): Use
7767 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7768 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7769 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7770 * doc/tm.texi: Regenerate.
7771 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7772 * config/spu/spu.c (spu_function_arg_offset): New function.
7773 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7774 * system.h (FUNCTION_ARG_OFFSET): Poison.
7776 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7777 Alan Hayard <alan.hayward@arm.com>
7778 David Sherwood <david.sherwood@arm.com>
7780 * target.def (truly_noop_truncation): New hook.
7781 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7782 than TRULY_NOOP_TRUNCATION.
7783 * hooks.h (hook_bool_uint_uint_true): Declare.
7784 * hooks.c (hook_bool_uint_uint_true): New function.
7785 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7786 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7787 * doc/tm.texi: Regenerate.
7788 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7789 rather than TRULY_NOOP_TRUNCATION in comments.
7790 (simplify_comparison): Likewise.
7791 (record_truncated_value): Likewise.
7792 * expmed.c (extract_bit_field_1): Likewise.
7793 (extract_split_bit_field): Likewise.
7794 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7795 instead of TRULY_NOOP_TRUNCATION.
7796 * function.c (assign_parm_setup_block): Likewise.
7797 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7798 * rtlhooks.c: Include target.h.
7799 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7800 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7801 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7802 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7803 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7804 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7805 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7806 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7807 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7808 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7809 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7810 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7811 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7812 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7813 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7814 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7815 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7816 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7817 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7818 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7819 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7820 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7821 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7822 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7823 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7824 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7825 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7826 rather than TRULY_NOOP_TRUNCATION in comments.
7827 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7828 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7829 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7830 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7831 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7832 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7833 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7834 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7835 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7836 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7837 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7838 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7839 rather than TRULY_NOOP_TRUNCATION in comments.
7840 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7841 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7842 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7843 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7844 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7845 TRULY_NOOP_TRUNCATION condition.
7846 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7847 (TRULY_NOOP_TRUNCATION): Delete.
7848 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7849 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7850 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7851 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7852 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7853 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7854 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7855 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7856 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7857 rather than TRULY_NOOP_TRUNCATION in comments.
7858 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7859 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7860 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7861 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7862 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7863 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7865 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7868 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7869 (*cmp_ior): Likewise.
7870 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7871 (*ior_scc_scc_cmp): Likewise.
7872 (*and_scc_scc): Likewise.
7873 (*and_scc_scc_cmp): Likewise.
7875 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7876 Alan Hayard <alan.hayward@arm.com>
7877 David Sherwood <david.sherwood@arm.com>
7879 * target.def (can_change_mode_class): New hook.
7880 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7881 (hard_regno_nregs): Likewise.
7882 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7883 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7884 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7885 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7886 (LOAD_EXTEND_OP): Update accordingly.
7887 * doc/tm.texi: Regenerate.
7888 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7889 CANNOT_CHANGE_MODE_CLASS.
7890 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7891 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7892 * combine.c (simplify_set): Update accordingly.
7893 * emit-rtl.c (validate_subreg): Likewise.
7894 * recog.c (general_operand): Likewise.
7895 * regcprop.c (mode_change_ok): Likewise.
7896 * reload1.c (choose_reload_regs): Likewise.
7897 (inherit_piecemeal_p): Likewise.
7898 * rtlanal.c (simplify_subreg_regno): Likewise.
7899 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7900 instead of CANNOT_CHANGE_MODE_CLASS.
7901 (reload_cse_simplify_operands): Likewise.
7902 * reload.c (push_reload): Use targetm.can_change_mode_class
7903 instead of CANNOT_CHANGE_MODE_CLASS.
7904 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7905 REG_CANNOT_CHANGE_MODE_P.
7906 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7907 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7908 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7909 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7910 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7911 (arm_can_change_mode_class): New function.
7912 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7913 than CANNOT_CHANGE_MODE_CLASS in comments.
7914 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7915 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7916 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7917 (ix86_can_change_mode_class): ...this new function, inverting the
7918 sense of the return value.
7919 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7920 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7921 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7922 (ia64_can_change_mode_class): New function.
7923 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7924 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7925 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7926 (m32c_can_change_mode_class): ...this new function, inverting the
7927 sense of the return value.
7928 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7929 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7930 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7931 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7932 (mips_can_change_mode_class): ...this new function, inverting the
7933 sense of the return value.
7934 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7935 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7936 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7937 (msp430_can_change_mode_class): New function.
7938 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7939 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7940 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7941 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7942 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7943 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7944 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7945 (pa_cannot_change_mode_class): Replace with...
7946 (pa_can_change_mode_class): ...this new function, inverting the
7947 sense of the return value.
7948 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7949 than CANNOT_CHANGE_MODE_CLASS in comments.
7950 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7951 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7952 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7953 (pdp11_cannot_change_mode_class): Replace with...
7954 (pdp11_can_change_mode_class): ...this new function, inverting the
7955 sense of the return value.
7956 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7957 * config/powerpcspe/powerpcspe-protos.h
7958 (rs6000_cannot_change_mode_class_ptr): Delete.
7959 * config/powerpcspe/powerpcspe.c
7960 (rs6000_cannot_change_mode_class_ptr): Delete.
7961 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7962 (rs6000_option_override_internal): Assign to
7963 targetm.can_change_mode_class instead of
7964 rs6000_cannot_change_mode_class_ptr.
7965 (rs6000_cannot_change_mode_class): Replace with...
7966 (rs6000_can_change_mode_class): ...this new function, inverting the
7967 sense of the return value.
7968 (rs6000_debug_cannot_change_mode_class): Replace with...
7969 (rs6000_debug_can_change_mode_class): ...this new function.
7970 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7971 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7972 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7973 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7974 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7976 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7977 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7978 (rs6000_option_override_internal): Assign to
7979 targetm.can_change_mode_class instead of
7980 rs6000_cannot_change_mode_class_ptr.
7981 (rs6000_cannot_change_mode_class): Replace with...
7982 (rs6000_can_change_mode_class): ...this new function, inverting the
7983 sense of the return value.
7984 (rs6000_debug_cannot_change_mode_class): Replace with...
7985 (rs6000_debug_can_change_mode_class): ...this new function.
7986 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7987 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7988 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7989 (s390_can_change_mode_class): ...this new function, inverting the
7990 sense of the return value.
7991 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7992 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7993 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7994 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7995 (sh_cannot_change_mode_class): Replace with...
7996 (sh_can_change_mode_class): ...this new function, inverting the
7997 sense of the return value.
7998 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7999 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8000 (sparc_can_change_mode_class): New function.
8001 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8002 * config/spu/spu.c (spu_can_change_mode_class): New function.
8003 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8004 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8005 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8006 (visium_can_change_mode_class): New function.
8007 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8009 2017-09-15 Richard Biener <rguenther@suse.de>
8011 PR tree-optimization/82217
8012 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8013 but not undefined case.
8015 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8018 * postreload.c (reload_cse_simplify_operands): Skip
8019 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8021 2017-09-15 Richard Biener <rguenther@suse.de>
8023 PR tree-optimization/68823
8024 * graphite-scop-detection.c (build_alias_set): If we have a
8025 possible dependence check whether we can handle them by just
8026 looking at the DRs DR_ACCESS_FNs.
8027 (build_scops): If build_alias_set fails, fail the SCOP.
8029 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8031 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8032 to support float128 built-in functions that require the ISA 3.0
8034 (BU_FLOAT128_3_HW): Likewise.
8035 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8037 (FMAF128): Likewise.
8039 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8040 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8041 floating point instructions.
8042 (rs6000_invalid_builtin): Likewise.
8043 (rs6000_builtin_mask_names): Likewise.
8044 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8045 (RS6000_BTM_FLOAT128_HW): Likewise.
8046 (RS6000_BTM_COMMON): Likewise.
8047 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8049 * doc/extend.texi (RS/6000 built-in functions): Document the
8050 IEEE 128-bit floating point square root and fused multiply-add
8053 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8055 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8056 reg (r2) isn't in the set of registers defined in the prologue.
8058 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8059 Alan Hayward <alan.hayward@arm.com>
8060 David Sherwood <david.sherwood@arm.com>
8062 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8063 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8064 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8065 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8066 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8068 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8069 max_vectorization_factor.
8070 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8072 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8073 Alan Hayward <alan.hayward@arm.com>
8074 David Sherwood <david.sherwood@arm.com>
8076 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8077 (vect_worthwhile_without_simd_p): Declare.
8078 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8079 (vectorizable_reduction): Use it.
8080 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8081 (vectorizable_operation): Likewise.
8083 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8084 Alan Hayward <alan.hayward@arm.com>
8085 David Sherwood <david.sherwood@arm.com>
8087 * tree-vectorizer.h (vect_get_num_copies): New function.
8088 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8089 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8090 (vectorizable_induction): Likewise.
8091 (vectorizable_live_operation): Likewise.
8092 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8093 (vectorizable_bswap): Likewise.
8094 (vectorizable_call): Likewise.
8095 (vectorizable_conversion): Likewise.
8096 (vectorizable_assignment): Likewise.
8097 (vectorizable_shift): Likewise.
8098 (vectorizable_operation): Likewise.
8099 (vectorizable_store): Likewise.
8100 (vectorizable_load): Likewise.
8101 (vectorizable_condition): Likewise.
8102 (vectorizable_comparison): Likewise.
8103 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8105 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8106 Alan Hayward <alan.hayward@arm.com>
8107 David Sherwood <david.sherwood@arm.com>
8109 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8110 of vect_init_vector.
8112 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8113 Alan Hayward <alan.hayward@arm.com>
8114 David Sherwood <david.sherwood@arm.com>
8116 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8117 an inline wrapper that provides a location.
8118 (gimple_build_vector): Likewise.
8119 * gimple-fold.c (gimple_build_vector_from_val): New function.
8120 (gimple_build_vector): Likewise.
8121 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8122 functions to build the initial value. Always return a gimple value.
8123 (get_initial_defs_for_reduction): Likewise. Only compute
8125 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8126 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8127 (vectorizable_induction): Use gimple_build_vector rather than
8130 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8131 Alan Hayward <alan.hayward@arm.com>
8132 David Sherwood <david.sherwood@arm.com>
8134 * target.h (vec_perm_indices): New typedef.
8135 (auto_vec_perm_indices): Likewise.
8136 * optabs-query.h: Include target.h
8137 (can_vec_perm_p): Take a vec_perm_indices *.
8138 * optabs-query.c (can_vec_perm_p): Likewise.
8139 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8140 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8141 * tree-vect-generic.c (lower_vec_perm): Likewise.
8142 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8143 (vect_grouped_load_supported): Likewise.
8144 (vect_shift_permute_load_chain): Likewise.
8145 (vect_permute_store_chain): Use auto_vec_perm_indices.
8146 (vect_permute_load_chain): Likewise.
8147 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8148 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8149 Update uses of can_vec_perm_p.
8150 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8151 mode with a number of elements. Take a vec_perm_indices *.
8152 (vect_create_epilog_for_reduction): Update accordingly.
8153 Use auto_vec_perm_indices.
8154 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8155 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8156 (vect_transform_slp_perm_load): Likewise.
8157 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8158 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8159 (vect_gen_perm_mask_checked): Likewise.
8160 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8161 (vect_gen_perm_mask_checked): Likewise.
8162 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8163 (vectorizable_store): Likewise.
8164 (vectorizable_load): Likewise.
8165 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8166 (vectorizable_bswap): Likewise.
8168 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8169 Alan Hayward <alan.hayward@arm.com>
8170 David Sherwood <david.sherwood@arm.com>
8172 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8173 * tree.c (build_vector): Likewise.
8174 (build_vector_from_ctor): Update accordingly.
8175 (build_vector_from_val): Likewise.
8176 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8177 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8178 * tree-vect-generic.c (add_rshift): Likewise.
8179 (expand_vector_divmod): Likewise.
8180 (optimize_vector_constructor): Likewise.
8181 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8182 (vect_transform_slp_perm_load): Likewise.
8183 (vect_schedule_slp_instance): Likewise.
8184 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8185 (vectorizable_call): Likewise.
8186 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8187 * expmed.c (make_tree): Likewise.
8188 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8189 a vector passed to build_vector.
8190 (fold_convert_const): Likewise.
8191 (exact_inverse): Likewise.
8192 (fold_ternary_loc): Likewise.
8193 (fold_relational_const): Likewise.
8194 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8195 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8196 (const_unop): Likewise. Store the reduction accumulator in a
8197 variable rather than an array.
8198 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8199 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8200 the new vector, rather than constructing it after the input arrays.
8201 (native_interpret_vector): Use auto_vec<tree> when building
8202 a vector passed to build_vector. Add elements in order.
8203 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8204 auto_vec<tree> when building a vector passed to build_vector.
8205 (vect_create_epilog_for_reduction): Likewise.
8206 (vectorizable_induction): Likewise.
8207 (get_initial_def_for_reduction): Likewise. Fix indentation of
8209 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8211 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8212 passed to build_vector.
8214 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8215 Alan Hayward <alan.hayward@arm.com>
8216 David Sherwood <david.sherwood@arm.com>
8218 * tree-core.h (tree_base::u): Add an "nelts" field.
8219 (tree_vector): Use VECTOR_CST_NELTS as the length.
8220 * tree.c (tree_size): Likewise.
8221 (make_vector): Initialize VECTOR_CST_NELTS.
8222 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8223 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8224 TYPE_VECTOR_SUBPARTS.
8225 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8226 as the number of units, setting it from VECTOR_CST_NELTS.
8227 (const_vector_from_tree): Likewise.
8228 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8229 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8230 (fold_negate_expr_1): Likewise.
8231 (fold_convert_const): Likewise.
8232 (const_binop): Likewise. Differentiate the number of output and
8234 (const_unop): Likewise.
8235 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8236 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8237 in cases that did the opposite.
8239 2017-09-14 Richard Biener <rguenther@suse.de>
8241 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8244 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8246 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8248 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8251 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8252 if and where to split a bb, except for splitting before debug insn
8253 sequences followed by non-label real insn. Delete debug insns
8254 in between basic blocks.
8256 * combine.c (make_compound_operation_int): Formatting fixes.
8258 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8259 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8260 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8261 * config/sol2.h (LINK_EH_SPEC): Likewise.
8262 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8263 * config/s390/linux.h (LINK_SPEC): Likewise.
8264 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8265 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8266 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8267 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8268 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8269 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8270 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8271 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8273 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8275 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8277 (ENDFILE_LINUX_SPEC): Likewise.
8278 (LINK_EH_SPEC): Likewise.
8279 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8280 (LINK_OS_LINUX_SPEC32): Likewise.
8281 (LINK_OS_LINUX_SPEC64): Likewise.
8282 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8283 (LINK_OS_LINUX_SPEC): Likewise.
8285 2017-09-13 Martin Liska <mliska@suse.cz>
8288 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8289 CASE_HIGH is NULL_TREE.
8291 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8292 Alan Hayward <alan.hayward@arm.com>
8293 David Sherwood <david.sherwood@arm.com>
8295 * target.def (secondary_memory_needed): New hook.
8296 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8297 instead of SECONDARY_MEMORY_NEEDED.
8298 (secondary_memory_needed_mode): Likewise.
8299 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8300 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8301 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8302 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8303 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8304 * doc/tm.texi: Regenerate.
8305 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8306 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8307 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8308 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8309 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8310 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8311 mode argument first and change the reg_class arguments to reg_class_t.
8312 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8313 Make static. Update the call to inline_secondary_memory_needed.
8314 (ix86_register_move_cost): Update the call to
8315 inline_secondary_memory_needed.
8316 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8317 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8319 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8320 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8322 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8323 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8324 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8325 and match hook interface. Add comment from mips.h.
8326 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8327 * config/mmix/mmix.md (truncdfsf2): Refer to
8328 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8330 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8331 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8332 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8333 (pa_secondary_memory_needed): New function.
8334 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8335 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8336 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8337 (pdp11_secondary_memory_needed): Make static and match hook interface.
8338 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8339 * config/powerpcspe/powerpcspe-protos.h
8340 (rs6000_secondary_memory_needed_ptr): Delete.
8341 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8343 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8344 (rs6000_option_override_internal): Assign to
8345 targetm.secondary_memory_needed rather than
8346 rs6000_secondary_memory_needed_ptr.
8347 (rs6000_secondary_memory_needed): Match hook interface.
8348 (rs6000_debug_secondary_memory_needed): Likewise.
8349 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8350 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8351 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8352 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8353 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8354 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8356 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8357 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8358 (rs6000_option_override_internal): Assign to
8359 targetm.secondary_memory_needed rather than
8360 rs6000_secondary_memory_needed_ptr.
8361 (rs6000_secondary_memory_needed): Match hook interface.
8362 (rs6000_debug_secondary_memory_needed): Likewise.
8363 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8364 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8365 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8366 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8367 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8368 (sparc_secondary_memory_needed): New function.
8369 * lra-constraints.c (check_and_process_move): Refer to
8370 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8372 (curr_insn_transform): Likewise.
8373 (process_alt_operands): Use targetm.secondary_memory_needed
8374 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8375 (check_secondary_memory_needed_p): Likewise.
8376 (choose_split_class): Likewise.
8377 * reload.c: Unconditionally include code that was previously
8378 conditional on SECONDARY_MEMORY_NEEDED.
8379 (push_secondary_reload): Use targetm.secondary_memory_needed
8380 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8381 (push_reload): Likewise.
8382 * reload1.c: Unconditionally include code that was previously
8383 conditional on SECONDARY_MEMORY_NEEDED.
8384 (choose_reload_regs): Use targetm.secondary_memory_needed
8385 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8386 (gen_reload): Likewise.
8387 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8389 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8390 Alan Hayward <alan.hayward@arm.com>
8391 David Sherwood <david.sherwood@arm.com>
8393 * target.def (secondary_memory_needed_mode): New hook:
8394 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8395 * targhooks.h (default_secondary_memory_needed_mode): New function.
8396 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8397 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8398 * doc/tm.texi: Regenerate.
8399 * lra-constraints.c (check_and_process_move): Use
8400 targetm.secondary_memory_needed_mode instead of
8401 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8402 (curr_insn_transform): Likewise.
8403 * reload.c (get_secondary_mem): Likewise.
8404 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8405 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8407 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8408 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8409 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8410 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8411 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8413 * config/powerpcspe/powerpcspe-protos.h
8414 (rs6000_secondary_memory_needed_mode): Delete.
8415 * config/powerpcspe/powerpcspe.c
8416 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8417 (rs6000_secondary_memory_needed_mode): Make static.
8418 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8419 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8421 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8423 (rs6000_secondary_memory_needed_mode): Make static.
8424 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8425 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8426 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8427 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8428 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8430 (sparc_secondary_memory_needed_mode): New function.
8431 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8433 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8435 * config/aarch64/constraints.md (Umq): New constraint.
8436 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8438 (mov<mode>): Update condition.
8440 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8442 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8443 when bitposition is the same.
8445 2017-09-13 Richard Biener <rguenther@suse.de>
8447 * dwarf2out.c (output_die_symbol): Remove.
8448 (output_die): Do not output a DIEs symbol.
8450 2017-09-13 Richard Biener <rguenther@suse.de>
8453 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8454 default-def to avoid breaking iterator update with the weird
8455 interaction with cgraph_update_edges_for_call_stmt_node.
8457 2017-09-13 Richard Biener <rguenther@suse.de>
8459 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8460 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8461 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8462 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8464 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8466 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8467 Disable pc relative literal load irrespective of
8468 TARGET_FIX_ERR_A53_84341 for default.
8470 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8472 * config/sparc/sparc.c (output_return): Output the source location of
8473 the insn in the delay slot, if any.
8474 (output_sibcall): Likewise.
8476 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8479 * common.opt (-static-pie): New alias.
8480 (shared): Negate static-pie.
8481 (-no-pie): Update help text.
8483 (static-pie): New option.
8484 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8485 -static-pie support.
8486 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8487 (LINK_EH_SPEC): Likewise.
8488 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8489 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8490 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8491 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8492 (init_gcc_specs): Likewise.
8493 (init_spec): Likewise.
8494 (display_help): Update help message for -pie.
8495 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8498 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8500 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8501 (movdi_aarch64): Likewise.
8502 (movti_aarch64): Likewise.
8504 2017-09-12 Simon Wright <simon@pushface.org>
8507 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8508 calculation of the minor version, always output as 0.
8510 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8513 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8514 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8515 on it early, rather than manual conversion late. For
8516 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8517 instead of performing manual conversion.
8519 2017-09-12 Carl Love <cel@us.ibm.com>
8521 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8522 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8524 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8525 VMULOSW): Add definitions.
8526 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8527 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8528 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8529 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8530 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8532 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8534 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8536 (movti_aarch64): Likewise.
8537 (movdf_aarch64): Likewise.
8538 (movtf_aarch64): Likewise.
8539 (load_pairdi): Likewise.
8540 (store_pairdi): Likewise.
8541 (load_pairdf): Likewise.
8542 (store_pairdf): Likewise.
8543 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8544 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8545 (ldr_got_small_<mode>): Likewise.
8546 (ldr_got_small_28k_<mode>): Likewise.
8547 (ldr_got_tiny): Likewise.
8548 * config/aarch64/iterators.md (ldst_sz): New.
8549 (ldpstp_sz): Likewise.
8550 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8552 (thunderx_load): Split load_8 to load_16.
8553 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8555 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8556 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8557 (xgene1_store_pair): Split store_8 to store_16.
8558 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8559 (falkor_st_0_st_sd): Split store_8 to store_16.
8561 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8563 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8564 and store1/2/3/4 to store_4/8/12/16.
8565 * config/aarch64/aarch64.md: Update for rename.
8566 * config/arm/arm.md: Likewise.: Likewise.
8567 * config/arm/arm.c: Likewise.
8568 * config/arm/thumb1.md: Likewise.
8569 * config/arm/thumb2.md: Likewise.
8570 * config/arm/vfp.md: Likewise.
8571 * config/arm/arm-generic.md: Likewise.
8572 * config/arm/arm1020e.md: Likewise.
8573 * config/arm/arm1026ejs.md: Likewise.
8574 * config/arm/arm1136jfs.md: Likewise.
8575 * config/arm/arm926ejs.md: Likewise.
8576 * config/arm/cortex-a15.md: Likewise.
8577 * config/arm/cortex-a17.md: Likewise.
8578 * config/arm/cortex-a5.md: Likewise.
8579 * config/arm/cortex-a53.md: Likewise.
8580 * config/arm/cortex-a57.md: Likewise.
8581 * config/arm/cortex-a7.md: Likewise.
8582 * config/arm/cortex-a8.md: Likewise.
8583 * config/arm/cortex-a9.md: Likewise.
8584 * config/arm/cortex-m4.md: Likewise.
8585 * config/arm/cortex-m7.md: Likewise.
8586 * config/arm/cortex-r4.md: Likewise.
8587 * config/arm/exynos-m1.md: Likewise.
8588 * config/arm/fa526.md: Likewise.
8589 * config/arm/fa606te.md: Likewise.
8590 * config/arm/fa626te.md: Likewise.
8591 * config/arm/fa726te.md: Likewise.
8592 * config/arm/fmp626.md: Likewise.
8593 * config/arm/iwmmxt.md: Likewise.
8594 * config/arm/ldmstm.md: Likewise.
8595 * config/arm/marvell-pj4.md: Likewise.
8596 * config/arm/xgene1.md: Likewise.
8597 * config/aarch64/thunderx.md: Likewise.
8598 * config/aarch64/thunderx2t99.md: Likewise.
8599 * config/aarch64/falkor.md: Likewise.
8601 2017-09-12 Martin Liska <mliska@suse.cz>
8603 * attribs.c (private_lookup_attribute): New function.
8604 * attribs.h (private_lookup_attribute): Declared here.
8605 (lookup_attribute): Called from this place.
8607 2017-09-12 Richard Biener <rguenther@suse.de>
8609 PR tree-optimization/82157
8610 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8611 stmts with side-effects.
8613 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8614 Alan Hayward <alan.hayward@arm.com>
8615 David Sherwood <david.sherwood@arm.com>
8617 * target.def (hard_regno_nregs): New hook.
8618 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8619 * targhooks.h (default_hard_regno_nregs): Declare.
8620 * targhooks.c (default_hard_regno_nregs): New function.
8621 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8622 (TARGET_HARD_REGNO_NREGS): ...this hook.
8623 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8624 (CLASS_MAX_NREGS): Likewise.
8625 * doc/tm.texi: Regenerate.
8626 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8627 instead of HARD_REGNO_NREGS.
8628 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8629 HARD_REGNO_NREGS in the comment.
8630 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8631 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8632 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8633 Return an unsigned int.
8634 (TARGET_HARD_REGNO_NREGS): Redefine.
8635 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8636 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8637 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8638 (arc_hard_regno_nregs): New function.
8639 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8640 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8641 (arm_hard_regno_nregs): New function.
8642 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8643 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8644 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8645 (TARGET_HARD_REGNO_NREGS): Redefine.
8646 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8647 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8648 (HARD_REGNO_NREGS): Delete.
8649 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8650 (cr16_hard_regno_nregs): New function.
8651 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8652 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8653 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8654 (cris_hard_regno_nregs): New function.
8655 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8656 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8657 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8658 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8659 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8660 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8661 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8662 (frv_hard_regno_nregs): Make static. Take and return an
8664 (frv_class_max_nregs): Remove outdated copy of documentation.
8665 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8666 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8667 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8668 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8669 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8670 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8671 (TARGET_HARD_REGNO_NREGS): Redefine.
8672 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8673 (CLASS_MAX_NREGS): Update comment.
8674 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8675 (ia64_hard_regno_nregs): New function.
8676 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8677 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8678 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8679 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8680 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8682 (m32c_hard_regno_nregs): Likewise. Make static.
8683 (TARGET_HARD_REGNO_NREGS): Redefine.
8684 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8685 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8686 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8687 (m68k_hard_regno_nregs): New function.
8688 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8689 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8690 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8691 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8692 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8693 Take and return an unsigned int.
8694 (TARGET_HARD_REGNO_NREGS): Redefine.
8695 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8696 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8697 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8698 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8699 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8700 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8701 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8702 (msp430_hard_regno_nregs): Make static. Take and return an
8704 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8705 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8706 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8707 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8708 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8709 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8710 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8711 (TARGET_HARD_REGNO_NREGS): Redefine.
8712 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8713 (PA_HARD_REGNO_NREGS): ...this.
8714 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8715 (PA_HARD_REGNO_NREGS): ...this.
8716 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8717 (pa_hard_regno_nregs): New function.
8718 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8719 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8720 (pdp11_hard_regno_nregs): New function.
8721 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8722 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8723 (rs6000_hard_regno_nregs_hook): New function.
8724 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8725 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8726 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8727 Take and return an unsigned int. Move earlier in file.
8728 (TARGET_HARD_REGNO_NREGS): Redefine.
8729 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8730 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8731 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8732 (rl78_hard_regno_nregs): Make static. Take and return an
8734 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8735 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8736 (rs6000_hard_regno_nregs_hook): New function.
8737 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8738 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8739 (TARGET_HARD_REGNO_NREGS): Redefine.
8740 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8741 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8742 instead of HARD_REGNO_NREGS.
8743 (s390_hard_regno_nregs): New function.
8744 (s390_hard_regno_mode_ok): Add comment from s390.h.
8745 (TARGET_HARD_REGNO_NREGS): Redefine.
8746 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8747 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8748 (sh_hard_regno_nregs): New function.
8749 (sh_pass_in_reg_p): Use it.
8750 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8751 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8752 (sparc_hard_regno_nregs): New function.
8753 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8754 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8755 (spu_function_arg_advance): Use it, supplying a valid register number.
8756 (TARGET_HARD_REGNO_NREGS): Redefine.
8757 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8758 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8759 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8760 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8761 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8762 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8763 (CLASS_MAX_NREGS): Remove copy of old documentation.
8764 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8765 (visium_hard_regno_nregs): New function.
8766 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8767 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8768 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8769 xtensa_hard_regno_nregs): New function.
8770 * system.h (HARD_REGNO_NREGS): Poison.
8772 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8774 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8775 hard_regno_nregs instead of HARD_REGNO_NREGS.
8776 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8777 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8778 (c6x_expand_epilogue): Likewise.
8779 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8780 (frv_read_iacc_argument): Likewise.
8781 * config/sh/sh.c: Include regs.h.
8782 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8783 (regs_used): Likewise.
8784 (output_stack_adjust): Likewise.
8785 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8786 * expmed.c: Include regs.h.
8787 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8788 * ree.c: Include regs.h.
8789 (combine_reaching_defs): Use hard_regno_nregs instead of
8791 (add_removable_extension): Likewise.
8793 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8795 * regs.h (hard_regno_nregs): Turn into a function.
8796 (end_hard_regno): Update accordingly.
8797 * caller-save.c (setup_save_areas): Likewise.
8798 (save_call_clobbered_regs): Likewise.
8799 (replace_reg_with_saved_mem): Likewise.
8800 (insert_restore): Likewise.
8801 (insert_save): Likewise.
8802 * combine.c (can_change_dest_mode): Likewise.
8803 (move_deaths): Likewise.
8804 (distribute_notes): Likewise.
8805 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8806 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8807 (rs6000_split_multireg_move): Likewise.
8808 (rs6000_register_move_cost): Likewise.
8809 (rs6000_memory_move_cost): Likewise.
8810 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8811 (rs6000_split_multireg_move): Likewise.
8812 (rs6000_register_move_cost): Likewise.
8813 (rs6000_memory_move_cost): Likewise.
8814 * cselib.c (cselib_reset_table): Likewise.
8815 (cselib_lookup_1): Likewise.
8816 * emit-rtl.c (set_mode_and_regno): Likewise.
8817 * function.c (aggregate_value_p): Likewise.
8818 * ira-color.c (setup_profitable_hard_regs): Likewise.
8819 (check_hard_reg_p): Likewise.
8820 (calculate_saved_nregs): Likewise.
8821 (assign_hard_reg): Likewise.
8822 (improve_allocation): Likewise.
8823 (calculate_spill_cost): Likewise.
8824 * ira-emit.c (modify_move_list): Likewise.
8825 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8826 (ira_hard_reg_in_set_p): Likewise.
8827 * ira.c (setup_reg_mode_hard_regset): Likewise.
8828 (clarify_prohibited_class_mode_regs): Likewise.
8829 (check_allocation): Likewise.
8830 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8831 (lra_setup_reg_renumber): Likewise.
8832 (setup_try_hard_regno_pseudos): Likewise.
8833 (spill_for): Likewise.
8834 (assign_hard_regno): Likewise.
8835 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8836 * lra-constraints.c (in_class_p): Likewise.
8837 (lra_constraint_offset): Likewise.
8838 (simplify_operand_subreg): Likewise.
8839 (lra_constraints): Likewise.
8840 (split_reg): Likewise.
8841 (split_if_necessary): Likewise.
8842 (invariant_p): Likewise.
8843 (inherit_in_ebb): Likewise.
8844 * lra-lives.c (process_bb_lives): Likewise.
8845 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8846 (get_hard_regs): Likewise.
8847 (do_remat): Likewise.
8848 * lra-spills.c (assign_spill_hard_regs): Likewise.
8849 * mode-switching.c (create_pre_exit): Likewise.
8850 * postreload.c (reload_combine_recognize_pattern): Likewise.
8851 * recog.c (peep2_find_free_register): Likewise.
8852 * regcprop.c (kill_value_regno): Likewise.
8853 (set_value_regno): Likewise.
8854 (copy_value): Likewise.
8855 (maybe_mode_change): Likewise.
8856 (find_oldest_value_reg): Likewise.
8857 (copyprop_hardreg_forward_1): Likewise.
8858 * regrename.c (check_new_reg_p): Likewise.
8859 (regrename_do_replace): Likewise.
8860 * reload.c (push_reload): Likewise.
8861 (combine_reloads): Likewise.
8862 (find_dummy_reload): Likewise.
8863 (operands_match_p): Likewise.
8864 (find_reloads): Likewise.
8865 (find_equiv_reg): Likewise.
8866 (reload_adjust_reg_for_mode): Likewise.
8867 * reload1.c (count_pseudo): Likewise.
8868 (count_spilled_pseudo): Likewise.
8869 (find_reg): Likewise.
8870 (clear_reload_reg_in_use): Likewise.
8871 (free_for_value_p): Likewise.
8872 (allocate_reload_reg): Likewise.
8873 (choose_reload_regs): Likewise.
8874 (reload_adjust_reg_for_temp): Likewise.
8875 (emit_reload_insns): Likewise.
8876 (delete_output_reload): Likewise.
8877 * rtlanal.c (subreg_get_info): Likewise.
8878 * sched-deps.c (sched_analyze_reg): Likewise.
8879 * sel-sched.c (init_regs_for_mode): Likewise.
8880 (mark_unavailable_hard_regs): Likewise.
8881 (choose_best_reg_1): Likewise.
8882 (verify_target_availability): Likewise.
8883 * valtrack.c (dead_debug_insert_temp): Likewise.
8884 * var-tracking.c (track_loc_p): Likewise.
8885 (emit_note_insn_var_location): Likewise.
8886 * varasm.c (make_decl_rtl): Likewise.
8887 * reginfo.c (choose_hard_reg_mode): Likewise.
8888 (init_reg_modes_target): Refer directly to
8889 this_target_regs->x_hard_regno_nregs.
8891 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8893 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8894 instead of hard_regno_nregs.
8896 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8898 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8899 end_hard_regno instead of hard_regno_nregs.
8900 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8901 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8902 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8903 * ira-color.c (improve_allocation): Likewise.
8904 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8905 * lra-lives.c (mark_regno_live): Likewise.
8906 (mark_regno_dead): Likewise.
8907 * lra-remat.c (operand_to_remat): Likewise.
8908 * lra.c (collect_non_operand_hard_regs): Likewise.
8909 * postreload.c (reload_combine_note_store): Likewise.
8910 (move2add_valid_value_p): Likewise.
8911 * reload.c (regno_clobbered_p): Likewise.
8913 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8915 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8917 * config/v850/v850.c (v850_reorg): Likewise.
8918 * reload.c (refers_to_regno_for_reload_p): Likewise.
8919 (find_equiv_reg): Likewise.
8920 * reload1.c (reload_reg_reaches_end_p): Likewise.
8922 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8924 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8926 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8927 * config/arm/arm.c (output_move_neon): Likewise.
8928 (arm_attr_length_move_neon): Likewise.
8929 (neon_split_vcombine): Likewise.
8930 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8931 (c6x_mark_reg_written): Likewise.
8932 (c6x_dwarf_register_span): Likewise.
8933 * config/i386/i386.c (ix86_save_reg): Likewise.
8934 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8935 (rws_access_reg): Likewise.
8936 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8937 * mode-switching.c (create_pre_exit): Likewise.
8938 * ree.c (combine_reaching_defs): Likewise.
8939 (add_removable_extension): Likewise.
8940 * regcprop.c (find_oldest_value_reg): Likewise.
8941 (copyprop_hardreg_forward_1): Likewise.
8942 * reload.c (reload_inner_reg_of_subreg): Likewise.
8943 (push_reload): Likewise.
8944 (combine_reloads): Likewise.
8945 (find_dummy_reload): Likewise.
8946 (reload_adjust_reg_for_mode): Likewise.
8947 * reload1.c (find_reload_regs): Likewise.
8948 (forget_old_reloads_1): Likewise.
8949 (reload_reg_free_for_value_p): Likewise.
8950 (reload_adjust_reg_for_temp): Likewise.
8951 (emit_reload_insns): Likewise.
8952 (delete_output_reload): Likewise.
8953 * sel-sched.c (choose_best_reg_1): Likewise.
8954 (choose_best_pseudo_reg): Likewise.
8956 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8957 Alan Hayward <alan.hayward@arm.com>
8958 David Sherwood <david.sherwood@arm.com>
8960 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8961 * target.def (slow_unaligned_access): New hook.
8962 * targhooks.h (default_slow_unaligned_access): Declare.
8963 * targhooks.c (default_slow_unaligned_access): New function.
8964 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8965 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8966 * doc/tm.texi: Regenerate.
8967 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8968 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8969 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8971 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8972 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8974 (rs6000_slow_unaligned_access): New function.
8975 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8976 (expand_block_compare): Likewise.
8977 (expand_strn_compare): Likewise.
8978 (rs6000_rtx_costs): Likewise.
8979 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8980 (riscv_slow_unaligned_access): Likewise.
8981 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8982 (riscv_slow_unaligned_access_p): ...this and make static.
8983 (riscv_option_override): Update accordingly.
8984 (riscv_slow_unaligned_access): New function.
8985 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8986 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8987 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8988 (rs6000_slow_unaligned_access): New function.
8989 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8990 (rs6000_rtx_costs): Likewise.
8991 * config/rs6000/rs6000-string.c (expand_block_compare)
8992 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8993 of SLOW_UNALIGNED_ACCESS.
8994 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8995 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8996 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8997 of SLOW_UNALIGNED_ACCESS.
8998 * expmed.c (simple_mem_bitfield_p): Likewise.
8999 * expr.c (alignment_for_piecewise_move): Likewise.
9000 (emit_group_load_1): Likewise.
9001 (emit_group_store): Likewise.
9002 (copy_blkmode_from_reg): Likewise.
9003 (emit_push_insn): Likewise.
9004 (expand_assignment): Likewise.
9005 (store_field): Likewise.
9006 (expand_expr_real_1): Likewise.
9007 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9008 * lra-constraints.c (simplify_operand_subreg): Likewise.
9009 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9010 * gimple-ssa-store-merging.c: Likewise in block comment at start
9012 * tree-ssa-strlen.c: Include target.h.
9013 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9014 of SLOW_UNALIGNED_ACCESS.
9015 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9017 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9019 PR rtl-optimization/82185
9020 * expmed.c (emit_store_flag_int): Only test tem if it has been
9023 2017-09-12 Richard Biener <rguenther@suse.de>
9026 * match.pd ((FTYPE) N CMP CST): Fix typo.
9028 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9030 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9032 (mips_near_type_p): Add 'short_call' attribute as a synonym
9034 * doc/extend.texi (short_call): Document new function attribute.
9036 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9039 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9040 assertion check that in the condition.
9041 (get_atomic_generic_size): Likewise. Before testing if parameter
9042 has pointer type, if it has array type, call for C++
9043 default_conversion to perform array-to-pointer conversion.
9045 2017-09-12 Richard Biener <rguenther@suse.de>
9047 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9048 for operations we cannot scalarize.
9050 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9052 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9053 vectors heap vectors. Clean up comments.
9054 Make visited_bbs a reference.
9055 (profitable_jump_thread_path): Make GC
9056 vectors heap vectors. Clean up comments.
9058 (convert_and_register_jump_thread_path): Make GC vectors heap
9060 (check_subpath_and_update_thread_path): Same. Clean up comments.
9061 Make visited_bbs a reference.
9062 (handle_phi): Abstract common code to to
9063 register_jump_thread_path_if_profitable.
9064 Rename VAR_BB to DEF_BB.
9066 Make GC vectors heap vectors.
9067 Make visited_bbs a reference.
9068 (handle_assignment): Same.
9069 (register_jump_thread_path_if_profitable): New.
9070 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9072 Make GC vectors heap vectors. Clean up comments.
9073 Make visited_bbs a reference.
9074 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9075 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9078 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9081 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9082 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9084 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9086 Revert r251800 and r251799.
9088 2017-09-11 Martin Jambor <mjambor@suse.cz>
9091 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9092 arguments in advance.
9093 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9094 use it to find predecessor edges.
9095 (naive_outof_ssa): Collect vector of predecessors.
9097 2017-09-08 Jason Merrill <jason@redhat.com>
9099 PR c++/70029 - ICE with ref-qualifier and -flto
9100 * langhooks.h (struct lang_hooks_for_types): Add
9101 copy_lang_qualifiers.
9102 * attribs.c (build_type_attribute_qual_variant): Use it.
9103 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9105 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9106 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9108 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9111 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9112 (*mulsi3_sp64): New instruction.
9113 (mulsi3): New expander.
9115 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9117 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9119 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9121 * sancov.c: Include memmodel.h.
9123 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9126 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9129 2017-09-07 Carl Love <cel@us.ibm.com>
9131 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9132 the sldi instruction.
9134 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9136 * sancov.c: Include tm_p.h.
9138 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9141 * output.h (switch_to_other_text_partition): New declaration.
9142 * varasm.c (switch_to_other_text_partition): New function.
9143 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9144 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9145 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9146 to the other text partition before emitting LCL label and switch back
9147 after emitting the word after it.
9149 2017-09-07 Richard Biener <rguenther@suse.de>
9151 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9152 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9153 critical edges here, after loop init.
9154 (pass_data_pre): Remove PROP_no_crit_edges flags.
9155 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9156 for valueization of call args to avoid leaking VN_TOP.
9157 (visit_use): Assert we do not visit default defs.
9158 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9159 Use error_mark_node to more easily detect leaking VN_TOP.
9160 All default-defs are varying, not VN_TOP. Mark them visited.
9161 (run_scc_vn): Make code match comment.
9163 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9165 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9166 OPTION_MASK_FLOAT128_KEYWORD.
9167 (POWERPC_MASKS): Likewise.
9168 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9169 support for the -mfloat128-type option, and make -mfloat128
9170 default on PowerPC Linux systems. Define or undefine
9171 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9172 Define __float128 to be __ieee128 if IEEE 128-bit support is
9173 enabled, or undefine it.
9174 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9175 Delete defining __FLOAT128_TYPE__.
9176 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9177 -mfloat128-type option and make -mfloat128 default on PowerPC
9179 (TARGET_FLOAT128_TYPE): Likewise.
9180 (-mfloat128-type): Likewise.
9181 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9182 Delete the -mfloat128-type option and make -mfloat128 default on
9183 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9184 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9185 128-bit floating point is enabled. Change tests from using
9186 -mfloat128-type to -mfloat128.
9187 (rs6000_mangle_type): Use the correct mangling for the __float128
9188 type even if normal long double is restricted to 64-bits.
9189 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9191 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9192 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9193 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9195 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9196 documentation for -mfloat128.
9198 2017-09-06 Olivier Hainque <hainque@adacore.com>
9200 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9202 2017-09-06 Wish Wu <wishwu007@gmail.com>
9203 Jakub Jelinek <jakub@redhat.com>
9205 * asan.c (initialize_sanitizer_builtins): Add
9206 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9207 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9208 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9209 BT_FN_VOID_UINT64_PTR variables.
9210 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9211 (BT_FN_VOID_UINT16_UINT16): Likewise.
9212 (BT_FN_VOID_UINT32_UINT32): Likewise.
9213 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9214 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9215 (BT_FN_VOID_UINT64_PTR): Likewise.
9216 * common.opt (flag_sanitize_coverage): New variable.
9217 (fsanitize-coverage=trace-pc): Remove.
9218 (fsanitize-coverage=): Add.
9219 * flag-types.h (enum sanitize_coverage_code): New enum.
9220 * fold-const.c (fold_range_test): Disable non-short-circuit
9221 optimization if flag_sanitize_coverage.
9222 (fold_truth_andor): Likewise.
9223 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9224 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9225 (coverage_sanitizer_opts): New array.
9226 (get_closest_sanitizer_option): Add OPTS argument, handle also
9227 OPT_fsanitize_coverage_.
9228 (parse_sanitizer_options): Adjusted to also handle
9229 OPT_fsanitize_coverage_.
9230 (common_handle_option): Add OPT_fsanitize_coverage_.
9231 * sancov.c (instrument_comparison, instrument_switch): New function.
9232 (sancov_pass): Add trace-cmp support.
9233 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9234 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9235 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9236 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9237 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9238 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9239 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9240 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9241 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9242 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9244 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9246 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9247 error. Only quit immediately if parsing is complete.
9248 (BEGIN): Initialize fatal_err and parse_done.
9249 (begin fpu, end fpu): Check number of arguments.
9250 (begin arch, end arch): Likewise.
9251 (begin cpu, end cpu): Likewise.
9252 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9253 (optalias): Likewise.
9255 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9257 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9258 * config/arm/arm-isa.h: Delete. Move definitions to ...
9259 * arm-cpus.in: ... here. Use new feature and fgroup values.
9260 * config/arm/arm.c (arm_option_override): Use lower case for feature
9262 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9263 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9264 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9266 (print_isa_bits_for): New function.
9267 (gen_isa): New function.
9268 (gen_comm_data): Use print_isa_bits_for.
9269 (define feature): New keyword.
9270 (define fgroup): New keyword.
9271 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9272 (arm-isa.h): Add rule to generate file.
9273 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9274 case for feature bit names.
9276 2017-09-06 Richard Biener <rguenther@suse.de>
9278 * tree-ssa-pre.c (NECESSARY): Remove.
9279 (create_expression_by_pieces): Do not touch pass-local flags.
9280 (insert_into_preds_of_block): Likewise.
9281 (do_pre_regular_insertion): Likewise.
9282 (eliminate_insert): Likewise.
9283 (eliminate_dom_walker::before_dom_children): Likewise.
9284 (fini_eliminate): Do not look at inserted_exprs.
9285 (mark_operand_necessary): Remove.
9286 (remove_dead_inserted_code): Replace with simple work-list
9287 algorithm based on inserted_exprs and SSA uses.
9288 (pass_pre::execute): Re-order fini_eliminate and
9289 remove_dead_inserted_code.
9291 2017-09-06 Olivier Hainque <hainque@adacore.com>
9293 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9294 for VxWorks 7. Adjust surrounding comments.
9296 2017-09-06 Richard Biener <rguenther@suse.de>
9298 * gimple-ssa-strength-reduction.c
9299 (find_candidates_dom_walker::before_dom_children): Also allow
9302 2017-09-06 Richard Biener <rguenther@suse.de>
9304 PR tree-optimization/82108
9305 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9306 for gap in the non-permutation SLP case.
9308 2017-09-06 Martin Jambor <mjambor@suse.cz>
9310 PR tree-optimization/82078
9311 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9312 add_access_to_work_queue...
9313 (build_accesses_from_assign): ...here.
9314 (propagate_all_subaccesses): Make sure racc is the group
9315 representative, if there is one.
9317 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9320 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9323 2017-09-06 Richard Biener <rguenther@suse.de>
9325 * gimple-ssa-strength-reduction.c
9326 (find_candidates_dom_walker::before_doom_children): Use a
9327 type and not a mode check.
9329 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9332 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9334 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9336 2017-09-05 Jeff Law <law@redhat.com>
9338 PR tree-optimization/64910
9339 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9340 cases where we have 3 or more operands.
9342 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9345 * omp-low.c (lower_omp_for): Recompute tree invariant if
9346 gimple_omp_for_initial/final is ADDR_EXPR.
9349 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9350 into gimple val before gimplification fo the COND_EXPR.
9352 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9354 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9355 REGION_COPY argument.
9356 (thread_through_all_blocks): Remove unused argument to
9357 duplicate_thread_path.
9359 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9360 Alan Hayward <alan.hayward@arm.com>
9361 David Sherwood <david.sherwood@arm.com>
9363 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9364 Take a scalar_mode rather than a machine_mode.
9365 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9366 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9367 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9368 (aarch64_gen_adjusted_ldpstp): Likewise.
9369 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9371 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9372 Alan Hayward <alan.hayward@arm.com>
9373 David Sherwood <david.sherwood@arm.com>
9375 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9376 Take a scalar_int_mode instead of a machine_mode.
9377 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9378 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9379 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9380 (aarch64_simd_attr_length_rglist): Delete.
9381 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9382 a scalar_int_mode instead of a machine_mode.
9383 (aarch64_add_offset): Likewise.
9384 (aarch64_internal_mov_immediate): Likewise
9385 (aarch64_add_constant_internal): Likewise.
9386 (aarch64_add_constant): Likewise.
9387 (aarch64_movw_imm): Likewise.
9388 (aarch64_rtx_arith_op_extract_p): Likewise.
9389 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9390 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9391 Remove assert that the mode isn't a vector.
9392 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9393 (aarch64_expand_mov_immediate): Update calls after above changes.
9394 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9395 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9396 (aarch64_move_imm): Likewise.
9397 (aarch64_can_const_movi_rtx_p): Likewise.
9398 (aarch64_strip_extend): Likewise.
9399 (aarch64_extr_rtx_p): Likewise.
9400 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9401 a CONST_INT when the mode parameter is VOIDmode.
9402 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9404 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9406 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9407 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9408 (bitwise_type_for_mode): Update accordingly.
9410 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9412 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9413 * stor-layout.c (mode_for_size_tree): Likewise.
9414 (mode_for_array): Update accordingly.
9415 (layout_decl): Likewise.
9416 (compute_record_mode): Likewise. Only set the mode once.
9418 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9420 * target.def (get_mask_mode): Change return type to opt_mode.
9422 * doc/tm.texi: Regenerate.
9423 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9424 * targhooks.c (default_get_mask_mode): Likewise.
9425 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9426 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9427 targetm.get_mask_mode.
9428 * tree.c (build_truth_vector_type): Likewise.
9430 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9432 * machmode.h (mode_for_vector): Return an opt_mode.
9433 * stor-layout.c (mode_for_vector): Likewise.
9434 (mode_for_int_vector): Update accordingly.
9435 (layout_type): Likewise.
9436 * config/i386/i386.c (emit_memmov): Likewise.
9437 (ix86_expand_set_or_movmem): Likewise.
9438 (ix86_expand_vector_init): Likewise.
9439 (ix86_get_mask_mode): Likewise.
9440 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9442 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9443 * expmed.c (extract_bit_field_1): Likewise.
9444 * expr.c (expand_expr_real_2): Likewise.
9445 * optabs-query.c (can_vec_perm_p): Likewise.
9446 (can_vec_mask_load_store_p): Likewise.
9447 * optabs.c (expand_vec_perm): Likewise.
9448 * targhooks.c (default_get_mask_mode): Likewise.
9449 * tree-vect-stmts.c (vectorizable_store): Likewise.
9450 (vectorizable_load): Likewise.
9451 (get_vectype_for_scalar_type_and_size): Likewise.
9453 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9455 * machmode.h (mode_for_int_vector): New function.
9456 * stor-layout.c (mode_for_int_vector): Likewise.
9457 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9458 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9459 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9460 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9461 (s390_expand_vcond): Likewise.
9463 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9465 * machmode.h (opt_machine_mode): New type.
9466 (opt_mode<T>): Allow construction from anything that can be
9468 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9469 (mode_for_size): Return an opt_machine_mode.
9470 * stor-layout.c (mode_for_size): Likewise.
9471 (mode_for_size_tree): Update call accordingly.
9472 (bitwise_mode_for_mode): Likewise.
9473 (make_fract_type): Likewise.
9474 (make_accum_type): Likewise.
9475 * caller-save.c (replace_reg_with_saved_mem): Update call
9477 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9478 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9479 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9480 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9481 * expmed.c (extract_bit_field_1): Likewise.
9482 * reload.c (get_secondary_mem): Likewise.
9483 * varasm.c (assemble_integer): Likewise.
9484 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9487 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9489 * machmode.h (decimal_float_mode_for_size): New function.
9490 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9491 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9492 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9493 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9494 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9495 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9497 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9499 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9500 (get_builtin_sync_mode): Likewise.
9501 (expand_ifn_atomic_compare_exchange): Likewise.
9502 (expand_builtin_atomic_clear): Likewise.
9503 (expand_builtin_atomic_test_and_set): Likewise.
9504 (fold_builtin_atomic_always_lock_free): Likewise.
9505 * calls.c (compute_argument_addresses): Likewise.
9506 (emit_library_call_value_1): Likewise.
9507 (store_one_arg): Likewise.
9508 * combine.c (combine_instructions): Likewise.
9509 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9510 * config/arm/arm.c (arm_function_value): Likewise.
9511 (aapcs_allocate_return_reg): Likewise.
9512 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9513 * config/i386/i386.c (construct_container): Likewise.
9514 (ix86_gimplify_va_arg): Likewise.
9515 (ix86_expand_sse_cmp): Likewise.
9516 (emit_memmov): Likewise.
9517 (emit_memset): Likewise.
9518 (expand_small_movmem_or_setmem): Likewise.
9519 (ix86_expand_pextr): Likewise.
9520 (ix86_expand_pinsr): Likewise.
9521 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9522 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9524 * config/mips/mips.c (mips_function_value_1) Likewise.
9525 (mips_block_move_straight): Likewise.
9526 (mips_expand_ins_as_unaligned_store): Likewise.
9527 * config/powerpcspe/powerpcspe.c
9528 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9529 (rs6000_darwin64_record_arg_flush): Likewise.
9530 * config/rs6000/rs6000.c
9531 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9532 (rs6000_darwin64_record_arg_flush): Likewise.
9533 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9534 (sparc_function_value_1): Likewise.
9535 * config/spu/spu.c (adjust_operand): Likewise.
9536 (spu_emit_branch_or_set): Likewise.
9537 (arith_immediate_p): Likewise.
9538 * emit-rtl.c (gen_lowpart_common): Likewise.
9539 * expr.c (expand_expr_real_1): Likewise.
9540 * function.c (assign_parm_setup_block): Likewise.
9541 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9542 * reload1.c (alter_reg): Likewise.
9543 * stor-layout.c (mode_for_vector): Likewise.
9544 (layout_type): Likewise.
9546 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9548 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9549 (spu_convert_move): Likewise.
9550 * lower-subreg.c (resolve_simple_move): Likewise.
9552 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9555 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9556 define_insn to a define_expand.
9557 (altivec_vsum2sws_direct): New define_insn.
9558 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9560 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9562 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9565 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9570 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9571 realign stack if stack alignment needed is less than incoming
9574 2017-09-05 Marek Polacek <polacek@redhat.com>
9577 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9580 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9582 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9584 2017-09-05 Richard Biener <rguenther@suse.de>
9586 PR tree-optimization/82084
9587 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9589 2017-09-05 Richard Biener <rguenther@suse.de>
9591 PR tree-optimization/82102
9592 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9594 2017-09-05 Martin Liska <mliska@suse.cz>
9596 PR tree-optimization/82032
9597 * tree-cfg.c (generate_range_test): New function.
9598 * tree-cfg.h (generate_range_test): Declared here.
9599 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9600 (cleanup_control_expr_graph): Use it.
9601 * tree-switch-conversion.c (try_switch_expansion): Remove
9603 (emit_case_nodes): Use generate_range_test.
9605 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9608 * config/i386/i386.md (*<btsc><mode>_mask): Add
9609 TARGET_USE_BT to insn constraint.
9610 (*btr<mode>_mask): Ditto.
9612 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9614 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9615 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9617 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9620 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9621 TARGET_NEON and TARGET_IWMMXT.
9622 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9623 TARGET_NEON and TARGET_IWMMXT.
9624 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9626 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9628 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9629 (ix86_rewrite_tls_address): Ditto.
9630 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9631 (ix86_rewrite_tls_address_1): Ditto.
9632 (ix86_rewrite_tls_address): Ditto.
9633 * config/i386/predicates.md (tls_address_pattern): New predicate.
9634 * config/i386/i386.md (TLS address splitter): New splitter.
9636 2017-09-04 Richard Biener <rguenther@suse.de>
9638 PR tree-optimization/82084
9639 * fold-const.h (can_native_encode_string_p): Declare.
9640 * fold-const.c (can_native_encode_string_p): Factor out from ...
9641 (native_encode_string): ... here.
9642 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9643 vectorizing stores from constants we later cannot handle.
9645 2017-09-04 Marek Polacek <polacek@redhat.com>
9648 * doc/invoke.texi: Update -Wtautological-compare documentation.
9650 2017-09-04 Jeff Law <law@redhat.com>
9652 PR tree-optimization/64910
9653 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9654 swap the first and last operand if the last is a constant.
9656 2017-09-04 Marek Polacek <polacek@redhat.com>
9659 * convert.c (do_narrow): When sanitizing signed integer overflows,
9660 bail out for signed types.
9661 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9663 2017-09-04 Richard Biener <rguenther@suse.de>
9665 PR tree-optimization/82060
9666 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9667 Move devirtualization after stmt folding and before EH/AB/noreturn
9668 cleanup to get the stmt refs canonicalized. Use a bool instead
9669 of gimple_modified_p since that doesn't work for NOPs. Schedule
9670 NOPs generated by folding for removal.
9672 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9673 Alan Hayward <alan.hayward@arm.com>
9674 David Sherwood <david.sherwood@arm.com>
9676 * coretypes.h (pad_direction): New enum.
9677 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9678 (FUNCTION_ARG_PADDING): Likewise.
9679 * target.def (function_arg_padding): New hook.
9680 * targhooks.h (default_function_arg_padding): Declare.
9681 * targhooks.c (default_function_arg_padding): New function.
9682 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9683 (TARGET_FUNCTION_ARG_PADDING): ...this.
9684 * doc/tm.texi: Regenerate.
9685 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9686 instead of direction.
9687 (compute_argument_addresses): Likewise.
9688 (load_register_parameters): Likewise.
9689 (emit_library_call_value_1): Likewise.
9690 (store_one_arg): Use targetm.calls.function_arg_padding instead
9691 of FUNCTION_ARG_PADDING.
9692 (must_pass_in_stack_var_size_or_pad): Likewise.
9693 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9694 (emit_group_store): Likewise.
9695 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9696 instead of FUNCTION_ARG_PADDING.
9697 (emit_push_insn): Likewise, and propagate enum change throughout
9699 * function.h (direction): Delete.
9700 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9702 * function.c (assign_parm_find_stack_rtl): Likewise.
9703 (assign_parm_setup_block_p): Likewise.
9704 (assign_parm_setup_block): Likewise.
9705 (gimplify_parameters): Likewise.
9706 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9707 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9709 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9710 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9711 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9712 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9713 (aarch64_function_arg_padding): ...this new function.
9714 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9715 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9716 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9717 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9718 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9719 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9720 (arm_pad_arg_upward): Replace with...
9721 (arm_function_arg_padding): ...this new function.
9722 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9724 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9725 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9726 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9727 (ia64_hpux_function_arg_padding): Replace with...
9728 (ia64_function_arg_padding): ...this new function. Use pad_direction
9729 instead of direction. Check for TARGET_HPUX.
9730 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9731 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9732 (iq2000_function_arg_padding): New function.
9733 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9734 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9735 (mips_function_arg_padding): ...this new function.
9736 (mips_pad_reg_upward): Update accordingly.
9737 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9738 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9739 targetm.calls.function_arg_padding.
9740 (FUNCTION_ARG_PADDING): Delete.
9741 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9742 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9743 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9744 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9745 (nios2_block_reg_padding): Return pad_direction instead of direction.
9746 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9747 instead of direction.
9748 (nios2_function_arg_padding): Likewise. Make static.
9749 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9750 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9751 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9752 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9753 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9754 (pa_function_arg_padding): Make static. Return pad_direction instead
9756 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9757 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9758 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9759 instead of direction. Use targetm.calls.function_arg_padding.
9760 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9761 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9762 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9763 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9764 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9766 (function_arg_padding): Rename to...
9767 (rs6000_function_arg_padding): ...this. Make static. Return
9768 pad_direction instead of direction.
9769 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9770 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9771 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9772 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9773 instead of direction. Use targetm.calls.function_arg_padding.
9774 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9775 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9776 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9777 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9778 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9779 (function_arg_padding): Rename to...
9780 (rs6000_function_arg_padding): ...this. Make static. Return
9781 pad_direction instead of direction.
9782 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9783 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9784 * config/s390/s390.c (s390_function_arg_padding): New function.
9785 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9786 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9787 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9788 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9789 (function_arg_padding): Rename to...
9790 (sparc_function_arg_padding): ...this. Make static. Return
9791 pad_direction instead of direction.
9792 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9793 * config/spu/spu.c (spu_function_arg_padding): New function.
9794 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9795 * system.h (FUNCTION_ARG_PADDING): Poison.
9797 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9798 Alan Hayward <alan.hayward@arm.com>
9799 David Sherwood <david.sherwood@arm.com>
9801 * target.def (modes_tieable_p): New hook.
9802 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9803 (TARGET_MODES_TIEABLE_P): ...this.
9804 * doc/tm.texi.in: Regenerate.
9805 * hooks.h (hook_bool_mode_mode_true): Declare.
9806 * hooks.c (hook_bool_mode_mode_true): New function.
9807 * combine.c (subst): Use targetm.modes_tieable_p instead of
9809 * dse.c (find_shift_sequence): Likewise.
9810 * expmed.c (extract_low_bits): Likewise.
9811 * lower-subreg.c: Include target.h.
9812 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9814 * rtlanal.c (rtx_cost): Likewise.
9815 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9816 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9817 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9818 (TARGET_MODES_TIEABLE_P): Redefine.
9819 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9820 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9821 (TARGET_MODES_TIEABLE_P): Redefine.
9822 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9823 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9824 (arc_modes_tieable_p): New function.
9825 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9826 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9827 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9828 (arm_modes_tieable_p): Make static.
9829 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9830 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9831 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9832 (TARGET_MODES_TIEABLE_P): Redefine.
9833 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9834 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9835 (TARGET_MODES_TIEABLE_P): Redefine.
9836 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9837 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9838 (cr16_modes_tieable_p): New function.
9839 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9840 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9841 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9842 (TRULY_NOOP_TRUNCATION): Update comment.
9843 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9844 (TRULY_NOOP_TRUNCATION): Update comment.
9845 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9846 (frv_modes_tieable_p): New function.
9847 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9848 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9849 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9850 (TARGET_MODES_TIEABLE_P): Redefine.
9851 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9852 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9853 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9854 (TARGET_MODES_TIEABLE_P): Redefine.
9855 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9856 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9857 (ia64_modes_tieable_p): New function.
9858 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9859 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9860 (iq2000_modes_tieable_p): New function.
9861 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9862 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9863 (lm32_modes_tieable_p): New function.
9864 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9865 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9866 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9867 (TARGET_MODES_TIEABLE_P): Redefine.
9868 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9869 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9870 (m32r_modes_tieable_p): New function.
9871 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9872 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9873 (m68k_modes_tieable_p): New function.
9874 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9875 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9876 (mcore_modes_tieable_p): New function.
9877 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9878 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9880 (TARGET_MODES_TIEABLE_P): Redefine.
9881 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9882 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9883 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9884 (TARGET_MODES_TIEABLE_P): Redefine.
9885 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9886 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9887 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9888 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9889 (mn10300_modes_tieable_p): ...this and make static.
9890 (TARGET_MODES_TIEABLE_P): Redefine.
9891 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9892 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9893 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9894 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9895 (msp430_modes_tieable_p): Make static.
9896 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9897 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9898 (TARGET_MODES_TIEABLE_P): Redefine.
9899 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9900 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9901 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9902 (TARGET_MODES_TIEABLE_P): Redefine.
9903 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9904 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9905 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9906 (TARGET_MODES_TIEABLE_P): Redefine.
9907 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9908 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9909 (pdp11_modes_tieable_p): New function.
9910 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9911 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9912 (rs6000_modes_tieable_p): New function.
9913 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9914 * config/powerpcspe/powerpcspe.md: Update comment.
9915 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9916 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9917 (TARGET_MODES_TIEABLE_P): Redefine.
9918 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9919 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9920 (rl78_modes_tieable_p): New function.
9921 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9922 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9923 (rs6000_modes_tieable_p): New function.
9924 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9925 * config/rs6000/rs6000.md: Update comment.
9926 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9927 * config/rx/rx.c (rx_modes_tieable_p): New function.
9928 (TARGET_MODES_TIEABLE_P): Redefine.
9929 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9930 * config/s390/s390.c (s390_modes_tieable_p): New function.
9931 (TARGET_MODES_TIEABLE_P): Redefine.
9932 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9933 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9934 (sh_modes_tieable_p): New function.
9935 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9936 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9937 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9938 (sparc_modes_tieable_p): Make static.
9939 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9940 * config/spu/spu.c (spu_modes_tieable_p): New function.
9941 (TARGET_MODES_TIEABLE_P): Redefine.
9942 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9943 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9944 (TARGET_MODES_TIEABLE_P): Redefine.
9945 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9946 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9947 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9948 * config/v850/v850.c (v850_modes_tieable_p): New function.
9949 (TARGET_MODES_TIEABLE_P): Redefine.
9950 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9951 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9952 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9953 (visium_modes_tieable_p): New function.
9954 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9955 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9956 (xtensa_modes_tieable_p): New function.
9957 * system.h (MODES_TIEABLE_P): Poison.
9959 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9960 Alan Hayward <alan.hayward@arm.com>
9961 David Sherwood <david.sherwood@arm.com>
9963 * target.def (hard_regno_mode_ok): New hook.
9964 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9965 (TARGET_HARD_REGNO_MODE_OK): ...this.
9966 * doc/tm.texi.in: Regenerate.
9967 * hooks.h (hook_bool_uint_mode_true): Declare.
9968 * hooks.c (hook_bool_uint_mode_true): New function.
9969 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9971 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9972 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9973 instead of HARD_REGNO_MODE_OK.
9974 * caller-save.c: Include target.h.
9975 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9977 * combine.c (can_combine_p): Likewise.
9978 (combinable_i3pat): Likewise.
9979 (can_change_dest_mode): Likewise.
9980 * expr.c (init_expr_target): Likewise.
9981 (convert_move): Likewise.
9982 (convert_modes): Likewise.
9983 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9984 (setup_prohibited_mode_move_regs): Likewise.
9985 * ira.h (target_ira): Likewise.
9986 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9987 * lra-constraints.c (process_alt_operands): Likewise.
9988 (split_reg): Likewise.
9989 * recog.c (peep2_find_free_register): Likewise.
9990 * ree.c (combine_reaching_defs): Likewise.
9991 * regcprop.c (maybe_mode_change): Likewise.
9992 * reginfo.c (init_reg_sets_1): Likewise.
9993 (choose_hard_reg_mode): Likewise.
9994 (simplifiable_subregs): Likewise.
9995 * regrename.c (check_new_reg_p): Likewise.
9996 * reload.c (find_valid_class): Likewise.
9997 (find_valid_class_1): Likewise.
9998 (reload_inner_reg_of_subreg): Likewise.
9999 (push_reload): Likewise.
10000 (combine_reloads): Likewise.
10001 (find_dummy_reload): Likewise.
10002 (find_reloads): Likewise.
10003 * reload1.c (find_reg): Likewise.
10004 (set_reload_reg): Likewise.
10005 (allocate_reload_reg): Likewise.
10006 (choose_reload_regs): Likewise.
10007 (reload_adjust_reg_for_temp): Likewise.
10008 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10009 (simplify_subreg_regno): Likewise.
10010 * sel-sched.c (init_regs_for_mode): Likewise.
10011 * varasm.c (make_decl_rtl): Likewise.
10012 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10013 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10014 HARD_REGNO_MODE_OK.
10015 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10016 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10017 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10018 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10019 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10020 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10021 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10022 (arc_mode_class): Delete.
10023 (HARD_REGNO_MODE_OK): Delete.
10024 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10025 (arc_hard_regno_mode_ok): Rename old array to...
10026 (arc_hard_regno_mode_ok_modes): ...this.
10027 (arc_conditional_register_usage): Update accordingly.
10028 (arc_mode_class): Make static.
10029 (arc_hard_regno_mode_ok): New function.
10030 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10031 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10032 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10033 (arm_hard_regno_mode_ok): Make static.
10034 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10035 HARD_REGNO_MODE_OK.
10036 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10037 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10038 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10040 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10041 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10042 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10043 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10044 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10045 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10046 * config/bfin/predicates.md (valid_reg_operand): Use
10047 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10048 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10049 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10050 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10051 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10052 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10053 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10054 (cr16_hard_regno_mode_ok): Make static and return a bool.
10055 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10056 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10057 (cris_hard_regno_mode_ok): New function.
10058 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10059 (epiphany_mode_class): Delete.
10060 (HARD_REGNO_MODE_OK): Delete.
10061 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10062 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10063 (hard_regno_mode_ok): Rename to...
10064 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10065 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10066 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10067 HARD_REGNO_MODE_OK.
10068 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10069 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10070 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10071 (frv_hard_regno_mode_ok): Make static and return a bool.
10072 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10073 HARD_REGNO_MODE_OK.
10074 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10075 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10076 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10077 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10079 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10080 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10081 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10082 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10084 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10085 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10086 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10087 (ia64_hard_regno_mode_ok): New function.
10088 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10089 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10090 (iq2000_hard_regno_mode_ok): New function.
10091 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10092 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10093 (lm32_hard_regno_mode_ok): New function.
10094 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10095 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10096 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10097 instead of HARD_REGNO_MODE_OK.
10098 (m32c_hard_regno_ok): Rename to...
10099 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10100 (m32c_cannot_change_mode_class): Update accordingly.
10101 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10102 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10103 (m32r_mode_class): Delete.
10104 (HARD_REGNO_MODE_OK): Delete.
10105 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10106 (m32r_hard_regno_mode_ok): Rename to...
10107 (m32r_hard_regno_modes): ...this.
10108 (m32r_mode_class): Make static.
10109 (m32r_hard_regno_mode_ok): New function.
10110 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10111 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10112 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10113 (m68k_hard_regno_mode_ok): Make static.
10114 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10115 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10116 (mcore_hard_regno_mode_ok): New function.
10117 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10118 (HARD_REGNO_MODE_OK): Delete.
10119 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10121 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10122 (microblaze_hard_regno_mode_ok): New function.
10123 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10124 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10125 (mips_hard_regno_mode_ok): Delete.
10126 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10127 (mips_hard_regno_mode_ok_p): ...this and make static.
10128 (mips_hard_regno_mode_ok_p): Rename to...
10129 (mips_hard_regno_mode_ok_uncached): ...this.
10130 (mips_hard_regno_mode_ok): New function.
10131 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10132 of HARD_REGNO_MODE_OK.
10133 (mips_option_override): Update after above name changes.
10134 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10135 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10136 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10137 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10138 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10139 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10140 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10141 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10142 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10143 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10144 (msp430_hard_regno_mode_ok): Make static and return a bool.
10145 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10146 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10147 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10149 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10150 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10151 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10152 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10153 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10154 (PA_HARD_REGNO_MODE_OK): ...this
10155 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10156 (PA_HARD_REGNO_MODE_OK): ...this.
10157 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10158 (pa_hard_regno_mode_ok): New function.
10159 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10160 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10161 (pdp11_hard_regno_mode_ok): New function.
10162 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10163 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10165 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10167 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10168 (rs6000_hard_regno_mode_ok): Rename to...
10169 (rs6000_hard_regno_mode_ok_uncached): ...this.
10170 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10171 (rs6000_hard_regno_mode_ok): New function.
10172 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10173 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10174 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10175 (riscv_hard_regno_mode_ok): ...this and make static.
10176 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10177 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10178 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10179 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10180 (rl78_hard_regno_mode_ok): Make static and return bool.
10181 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10182 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10184 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10185 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10186 (rs6000_hard_regno_mode_ok): Rename to...
10187 (rs6000_hard_regno_mode_ok_uncached): ...this.
10188 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10189 (rs6000_hard_regno_mode_ok): New function.
10190 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10191 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10192 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10193 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10194 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10195 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10196 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10197 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10198 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10199 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10200 (sh_hard_regno_mode_ok): Make static.
10201 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10202 instead of HARD_REGNO_MODE_OK.
10203 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10204 (sparc_mode_class): Delete.
10205 (HARD_REGNO_MODE_OK): Delete.
10206 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10207 (hard_regno_mode_classes): Make static.
10208 (sparc_mode_class): Likewise.
10209 (sparc_hard_regno_mode_ok): New function.
10210 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10211 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10212 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10214 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10215 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10216 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10217 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10218 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10219 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10220 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10221 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10222 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10223 (visium_hard_regno_mode_ok): New function.
10224 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10225 instead of HARD_REGNO_MODE_OK.
10226 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10227 (HARD_REGNO_MODE_OK): Delete.
10228 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10229 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10230 (xtensa_option_override): Update accordingly.
10231 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10232 (xtensa_hard_regno_mode_ok): New function.
10233 * system.h (HARD_REGNO_MODE_OK): Poison.
10235 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10236 Alan Hayward <alan.hayward@arm.com>
10237 David Sherwood <david.sherwood@arm.com>
10239 * target.def (hard_regno_call_part_clobbered): New hook.
10240 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10241 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10242 * doc/tm.texi: Regenerate.
10243 * hooks.h (hook_bool_uint_mode_false): Declare.
10244 * hooks.c (hook_bool_uint_mode_false): New function.
10245 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10246 * cselib.c (cselib_process_insn): Use
10247 targetm.hard_regno_call_part_clobbered instead of
10248 HARD_REGNO_CALL_PART_CLOBBERED.
10249 * ira-conflicts.c (ira_build_conflicts): Likewise.
10250 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10251 * lra-constraints.c (need_for_call_save_p): Likewise.
10252 * lra-lives.c: Include target.h.
10253 (check_pseudos_live_through_calls): Use
10254 targetm.hard_regno_call_part_clobbered instead of
10255 HARD_REGNO_CALL_PART_CLOBBERED.
10256 * regcprop.c: Include target.h.
10257 (copyprop_hardreg_forward_1): Use
10258 targetm.hard_regno_call_part_clobbered instead of
10259 HARD_REGNO_CALL_PART_CLOBBERED.
10260 * reginfo.c (choose_hard_reg_mode): Likewise.
10261 * regrename.c (check_new_reg_p): Likewise.
10262 * reload.c (find_equiv_reg): Likewise.
10263 * reload1.c (emit_reload_insns): Likewise.
10264 * sched-deps.c (deps_analyze_insn): Likewise.
10265 * sel-sched.c (init_regs_for_mode): Likewise.
10266 (mark_unavailable_hard_regs): Likewise.
10267 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10268 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10269 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10271 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10272 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10273 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10275 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10277 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10278 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10279 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10281 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10282 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10283 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10285 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10286 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10288 * config/powerpcspe/powerpcspe.c
10289 (rs6000_hard_regno_call_part_clobbered): New function.
10290 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10291 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10292 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10294 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10295 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10296 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10298 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10299 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10300 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10302 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10303 Alan Hayward <alan.hayward@arm.com>
10304 David Sherwood <david.sherwood@arm.com>
10306 * rtl.h (subreg_memory_offset): Declare.
10307 * emit-rtl.c (subreg_memory_offset): New function.
10308 * expmed.c (store_bit_field_1): Use it.
10309 * expr.c (undefined_operand_subword_p): Likewise.
10310 * simplify-rtx.c (simplify_subreg): Likewise.
10312 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10314 PR rtl-optimization/57448
10317 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10318 using atomic_load pattern.
10319 (expand_atomic_store): Likewise.
10321 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10324 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10325 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10326 handling. Use replace_call_with_value with NULL instead of
10327 gsi_replace, unlink_stmt_vdef and release_defs.
10329 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10332 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10334 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10337 * rtl.h (emit_library_call_value_1): Declare.
10338 (emit_library_call): Replace declaration with a series of overloads.
10339 Remove the parameter count argument.
10340 (emit_library_call_value): Likewise.
10341 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10342 with an "rtx_mode_t *".
10343 (emit_library_call_value): Delete.
10344 (emit_library_call): Likewise.
10345 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10346 (asan_emit_allocas_unpoison): Likewise.
10347 * builtins.c (expand_builtin_powi): Likewise.
10348 (expand_asan_emit_allocas_unpoison): Likewise.
10349 * cfgexpand.c (expand_main_function): Likewise.
10350 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10351 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10352 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10353 * config/arm/arm.c (arm_trampoline_init): Likewise.
10354 (arm_call_tls_get_addr): Likewise.
10355 (arm_expand_divmod_libfunc): Likewise.
10356 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10357 (smulsi3_highpart): Likewise.
10358 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10359 (c6x_expand_compare): Likewise.
10360 (c6x_expand_movmem): Likewise.
10361 * config/frv/frv.c (frv_trampoline_init): Likewise.
10362 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10363 (ix86_expand_divmod_libfunc): Likewise.
10364 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10365 (ia64_expand_compare): Likewise.
10366 (ia64_profile_hook): Likewise.
10367 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10368 (nonlocal_goto): Likewise.
10369 (restore_stack_nonlocal): Likewise.
10370 * config/m32r/m32r.c (block_move_call): Likewise.
10371 (m32r_trampoline_init): Likewise.
10372 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10373 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10374 (m68k_call_m68k_read_tp): Likewise.
10375 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10376 (microblaze_expand_divide): Likewise.
10377 * config/mips/mips.h (mips_args): Likewise.
10378 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10379 (MIPS_ICACHE_SYNC): Likewise.
10380 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10381 (nios2_trampoline_init): Likewise.
10382 * config/pa/pa.c (hppa_tls_call): Likewise.
10383 (pa_trampoline_init): Likewise.
10384 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10385 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10386 (expand_strn_compare): Likewise.
10387 (rs6000_generate_compare): Likewise.
10388 (rs6000_expand_float128_convert): Likewise.
10389 (output_profile_hook): Likewise.
10390 (rs6000_trampoline_init): Likewise.
10391 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10392 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10393 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10394 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10395 (rs6000_generate_compare): Likewise.
10396 (rs6000_expand_float128_convert): Likewise.
10397 (output_profile_hook): Likewise.
10398 (rs6000_trampoline_init): Likewise.
10399 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10400 * config/sh/sh.c (sh_trampoline_init): Likewise.
10401 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10402 (sparc_emit_float_lib_cmp): Likewise.
10403 (sparc32_initialize_trampoline): Likewise.
10404 (sparc64_initialize_trampoline): Likewise.
10405 (sparc_profile_hook): Likewise.
10406 * config/spu/spu.c (ea_load_store): Likewise.
10407 * config/spu/spu.md (floatunssidf2): Likewise.
10408 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10409 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10410 * config/visium/visium.c (expand_block_move_4): Likewise.
10411 (expand_block_move_2): Likewise.
10412 (expand_block_move_1): Likewise.
10413 (expand_block_set_4): Likewise.
10414 (expand_block_set_2): Likewise.
10415 (expand_block_set_1): Likewise.
10416 (visium_trampoline_init): Likewise.
10417 (visium_profile_hook): Likewise.
10418 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10419 (xtensa_setup_frame_addresses): Likewise.
10420 (xtensa_trampoline_init): Likewise.
10421 * except.c (sjlj_emit_function_enter): Likewise.
10422 (sjlj_emit_function_exit): Likewise.
10423 * explow.c (allocate_dynamic_stack_space): Likewise.
10424 (probe_stack_range): Likewise.
10425 * expr.c (convert_mode_scalar): Likewise.
10426 * optabs.c (expand_binop): Likewise.
10427 (expand_twoval_binop_libfunc): Likewise.
10428 (expand_unop): Likewise.
10429 (prepare_cmp_insn): Likewise.
10430 (prepare_float_lib_cmp): Likewise.
10431 (expand_float): Likewise.
10432 (expand_fix): Likewise.
10433 (expand_fixed_convert): Likewise.
10434 (maybe_emit_sync_lock_test_and_set): Likewise.
10435 (expand_atomic_compare_and_swap): Likewise.
10436 (expand_mem_thread_fence): Likewise.
10437 (expand_atomic_fetch_op): Likewise.
10439 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10441 * doc/generic.texi (OpenACC): Adjust URL.
10442 * doc/invoke.texi (C Dialect Options): Ditto.
10444 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10446 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10447 predicate for operand 1. Add (m,<S>) constraint.
10448 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10449 Prevent memory operand 1 with register operand 2.
10451 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10453 PR rtl-optimization/82024
10454 * combine.c (try_combine): If the combination result is a PARALLEL,
10455 and we only need to retain the SET in there that would be placed
10456 at I2, check that we can place that at I3 instead, before doing so.
10458 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10461 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10463 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10464 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10467 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10468 Jeff Law <law@redhat.com>
10470 * varasm.c (bss_initializer_p): Do not put constants into .bss
10471 (categorize_decl_for_section): Handle bss_initializer_p returning
10472 false when DECL_INITIAL is NULL.
10474 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10477 * config/s390/s390.c (s390_can_inline_p): New function.
10479 2017-09-01 Jeff Law <law@redhat.com>
10481 PR tree-optimization/82052
10482 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10483 Always initialize the returned slot after a hash table miss
10484 when INSERT is true.
10486 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10488 * config/s390/s390.md (mem_signal_fence): Remove.
10489 * doc/md.texi (mem_signal_fence): Remove.
10490 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10492 * target-insns.def (mem_signal_fence): Remove.
10494 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10497 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10500 * asan.c (create_odr_indicator): Strip name encoding from assembler
10501 name before appending it after __odr_asan_.
10503 2017-09-01 Martin Liska <mliska@suse.cz>
10505 PR tree-optimization/82059
10506 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10507 frequency only when an edge is redirected.
10509 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10511 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10512 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10513 (arc_conditional_register_usage): Remove ARC600 lp_count
10515 (arc_file_start): Emit Tag_ARC_CPU_variation.
10516 (arc_can_use_doloop_p): New conditions to use ZOLs.
10517 (hwloop_fail): New function.
10518 (hwloop_optimize): Likewise.
10519 (hwloop_pattern_reg): Likewise.
10520 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10521 (arc_reorg_loops): New function, calls reorg_loops.
10522 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10523 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10525 (arc_loop_hazard): Remove function, functionality moved into
10527 (arc_hazard): Remove arc_loop_hazard call.
10528 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10529 into hwloop_optimize.
10530 (arc_label_align): Remove ZOL handling.
10531 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10532 * config/arc/arc.md (doloop_begin): Remove pattern.
10533 (doloop_begin_i): Likewise.
10534 (doloop_end_i): Likewise.
10535 (doloop_fallback): Likewise.
10536 (doloop_fallback_m): Likewise.
10537 (doloop_end): Reimplement expand.
10538 (arc_lp): New pattern for LP instruction.
10539 (loop_end): New pattern.
10540 (loop_fail): Likewise.
10541 (decrement_and_branch_until_zero): Likewise.
10542 * config/arc/arc.opt (mlpc-width): New option.
10543 * doc/invoke.texi (mlpc-width): Document option.
10545 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10547 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10548 (arc_ccfsm_advance): Fix checking for delay slots.
10549 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10551 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10553 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10555 (movhi_insn): Update store instruction constraint which are saving
10556 6-bit short immediates.
10557 (movsi_insn): Consider also short scaled load operations.
10558 (zero_extendhisi2_i): Use Usd constraint instead of T.
10559 (extendhisi2_i): Add q constraint.
10560 (arc_clzsi2): Add type and length attributes.
10561 (arc_ctzsi2): Likewise.
10562 * config/arc/constraints.md (Usc): Update constraint, the
10563 assembler can parse two relocations for a single instruction.
10565 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10567 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10568 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10570 2017-08-31 Olivier Hainque <hainque@adacore.com>
10572 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10573 match as powerpc-wrs-vxworks*.
10575 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10577 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10578 register constraint for by-element operand.
10579 (aarch64_mls_elt_merge<mode>): Likewise.
10581 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10583 * config/arc/arc.c (arc_can_follow_jump): Check for short
10586 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10588 * config.gcc: Use g.opt for arc.
10589 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10590 functionality moved to ...
10591 (legitimate_scaled_address_p): New function, ...here.
10592 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10593 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10594 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10596 (arc_override_options): Handle G option.
10597 (arc_output_pic_addr_const): Correct function definition.
10598 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10599 (arc_decl_anon_ns_mem_p): Delete.
10600 (arc_in_small_data_p): Overhaul this function to take into
10601 consideration the value given via G option.
10602 (arc_rewrite_small_data_1): Renamed and corrected old
10603 arc_rewrite_small_data function.
10604 (arc_rewrite_small_data): New function.
10605 (small_data_pattern): Don't use pic_offset_table_rtx.
10606 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10607 * config/arc/simdext.md (movmisalignv2hi): Use
10608 prepare_move_operands function.
10610 (movmisalign*): Likewise.
10611 * doc/invoke.texi (ARC options): Document -G option.
10613 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10615 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10617 * config/arc/arc.c (arc_print_operand): Output scalled address for
10618 sdata whenever is possible.
10619 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10620 load/stores are available.
10621 (compact_sda_memory_operand): Check for the alignment required by
10622 code density instructions.
10623 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10625 * config/arc/constraints.md (Usd): Update constraint.
10626 (Us0): New constraint.
10627 (Usc): Update constraint.
10629 2017-08-31 Richard Biener <rguenther@suse.de>
10631 PR middle-end/82054
10632 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10633 function only once.
10635 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10637 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10638 Resize type_signature.
10640 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10641 Alan Hayward <alan.hayward@arm.com>
10642 David Sherwood <david.sherwood@arm.com>
10644 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10645 subregs whose inner modes can be stored in GPRs.
10646 (aarch64_classify_index): Likewise.
10648 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10649 Alan Hayward <alan.hayward@arm.com>
10650 David Sherwood <david.sherwood@arm.com>
10652 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10653 (V_INT_EQUIV): ...this.
10654 (v_cmp_result): Rename to...
10655 (v_int_equiv): ...this.
10656 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10657 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10658 (copysign<mode>3): Likewise.
10659 (aarch64_simd_bsl<mode>_internal): Likewise.
10660 (aarch64_simd_bsl<mode>): Likewise.
10661 (vec_cmp<mode><mode>): Likewise.
10662 (vcond<mode><mode>): Likewise.
10663 (vcond<v_cmp_mixed><mode>): Likewise.
10664 (vcondu<mode><v_cmp_mixed>): Likewise.
10665 (aarch64_cm<optab><mode>): Likewise.
10666 (aarch64_cmtst<mode>): Likewise.
10667 (aarch64_fac<optab><mode>): Likewise.
10668 (vec_perm_const<mode>): Likewise.
10669 (vcond_mask_<mode><v_cmp_result>): Rename to...
10670 (vcond_mask_<mode><v_int_equiv>): ...this.
10671 (vec_cmp<mode><v_cmp_result>): Rename to...
10672 (vec_cmp<mode><v_int_equiv>): ...this.
10674 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10675 Alan Hayward <alan.hayward@arm.com>
10676 David Sherwood <david.sherwood@arm.com>
10678 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10680 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10681 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10682 (UNSPEC_LD4_DREG): New unspecs.
10683 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10684 (aarch64_ld2<mode>_dreg_be): Replace with...
10685 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10687 (aarch64_ld3<mode>_dreg_le)
10688 (aarch64_ld3<mode>_dreg_be): Replace with...
10689 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10691 (aarch64_ld4<mode>_dreg_le)
10692 (aarch64_ld4<mode>_dreg_be): Replace with...
10693 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10696 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10698 PR tree-optimization/81987
10699 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10700 insert an initializer in a location not dominated by the stride
10703 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10705 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10706 on the entire header of the finally block in the fallthru case.
10708 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10710 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10712 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10714 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10715 rs6000_emit_move_from_cr and call renamed function.
10716 (rs6000_emit_prologue): Call renamed functions.
10717 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10718 movesi_from_cr, remove volatile CRs.
10720 2017-08-30 Jon Beniston <jon@beniston.com>
10721 Richard Biener <rguenther@suse.de>
10723 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10724 of VECTOR_MODE_P check.
10725 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10726 element vector types.
10728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10730 * df.h (df_read_modify_subreg_p): Remove in favor of...
10731 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10732 const_rtx instead of an rtx.
10733 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10734 * df-problems.c (df_word_lr_mark_ref): Likewise.
10735 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10736 (mark_pseudo_reg_dead): Likewise.
10737 (mark_ref_dead): Likewise.
10738 * reginfo.c (init_subregs_of_mode): Likewise.
10739 * sched-deps.c (sched_analyze_1): Likewise.
10740 * df-scan.c (df_def_record_1): Likewise.
10741 (df_uses_record): Likewise.
10742 (df_read_modify_subreg_p): Remove in favor of...
10743 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10744 const_rtx instead of an rtx.
10746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10747 Alan Hayward <alan.hayward@arm.com>
10748 David Sherwood <david.sherwood@arm.com>
10750 * rtl.h (partial_subreg_p): New function.
10751 * caller-save.c (save_call_clobbered_regs): Use it.
10752 * calls.c (expand_call): Likewise.
10753 * combine.c (combinable_i3pat): Likewise.
10754 (simplify_set): Likewise.
10755 (make_extraction): Likewise.
10756 (make_compound_operation_int): Likewise.
10757 (gen_lowpart_or_truncate): Likewise.
10758 (force_to_mode): Likewise.
10759 (make_field_assignment): Likewise.
10760 (reg_truncated_to_mode): Likewise.
10761 (record_truncated_value): Likewise.
10762 (move_deaths): Likewise.
10763 * cse.c (record_jump_cond): Likewise.
10764 (cse_insn): Likewise.
10765 * cselib.c (cselib_lookup_1): Likewise.
10766 * expmed.c (extract_bit_field_using_extv): Likewise.
10767 * function.c (assign_parm_setup_reg): Likewise.
10768 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10769 * ira-build.c (create_insn_allocnos): Likewise.
10770 * lra-coalesce.c (merge_pseudos): Likewise.
10771 * lra-constraints.c (match_reload): Likewise.
10772 (simplify_operand_subreg): Likewise.
10773 (curr_insn_transform): Likewise.
10774 * lra-lives.c (process_bb_lives): Likewise.
10775 * lra.c (new_insn_reg): Likewise.
10776 (lra_substitute_pseudo): Likewise.
10777 * regcprop.c (mode_change_ok): Likewise.
10778 (maybe_mode_change): Likewise.
10779 (copyprop_hardreg_forward_1): Likewise.
10780 * reload.c (push_reload): Likewise.
10781 (find_reloads): Likewise.
10782 (find_reloads_subreg_address): Likewise.
10783 * reload1.c (alter_reg): Likewise.
10784 (eliminate_regs_1): Likewise.
10785 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10787 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10789 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10790 back to if statements, including unpack.
10792 2017-08-30 Martin Liska <mliska@suse.cz>
10794 PR inline-asm/82001
10795 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10797 (func_checker::compare_asm_inputs_outputs): ... this function.
10798 (func_checker::compare_gimple_asm): Use the function to compare
10799 also ASM constrains.
10800 * ipa-icf-gimple.h: Rename the function.
10802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10803 Alan Hayward <alan.hayward@arm.com>
10804 David Sherwood <david.sherwood@arm.com>
10806 * coretypes.h (complex_mode): New type.
10807 * gdbhooks.py (build_pretty_printer): Handle it.
10808 * machmode.h (complex_mode): New class.
10809 (complex_mode::includes_p): New function.
10810 (is_complex_int_mode): Likewise.
10811 (is_complex_float_mode): Likewise.
10812 * genmodes.c (get_mode_class): Handle complex mode classes.
10813 * function.c (expand_function_end): Use is_complex_int_mode.
10815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10816 Alan Hayward <alan.hayward@arm.com>
10817 David Sherwood <david.sherwood@arm.com>
10819 * coretypes.h (scalar_mode_pod): New typedef.
10820 * gdbhooks.py (build_pretty_printer): Handle it.
10821 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10822 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10823 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10824 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10825 as_a <scalar_mode>.
10827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10828 Alan Hayward <alan.hayward@arm.com>
10829 David Sherwood <david.sherwood@arm.com>
10831 * machmode.h (mode_for_vector): Take a scalar_mode instead
10833 * stor-layout.c (mode_for_vector): Likewise.
10834 * explow.c (promote_mode): Use as_a <scalar_mode>.
10835 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10838 Alan Hayward <alan.hayward@arm.com>
10839 David Sherwood <david.sherwood@arm.com>
10841 * target.def (preferred_simd_mode): Take a scalar_mode
10842 instead of a machine_mode.
10843 * targhooks.h (default_preferred_simd_mode): Likewise.
10844 * targhooks.c (default_preferred_simd_mode): Likewise.
10845 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10846 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10847 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10848 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10849 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10850 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10851 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10852 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10854 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10855 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10856 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10857 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10858 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10859 * doc/tm.texi: Regenerate.
10860 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10864 Alan Hayward <alan.hayward@arm.com>
10865 David Sherwood <david.sherwood@arm.com>
10867 * target.def (scalar_mode_supported_p): Take a scalar_mode
10868 instead of a machine_mode.
10869 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10870 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10871 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10872 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10873 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10874 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10875 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10876 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10877 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10878 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10879 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10880 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10881 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10882 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10884 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10885 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10886 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10887 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10888 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10890 * doc/tm.texi: Regenerate.
10892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10893 Alan Hayward <alan.hayward@arm.com>
10894 David Sherwood <david.sherwood@arm.com>
10896 * coretypes.h (opt_scalar_mode): New typedef.
10897 * gdbhooks.py (build_pretty_printers): Handle it.
10898 * machmode.h (mode_iterator::get_2xwider): Add overload for
10900 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10902 * expr.c (convert_mode_scalar): Likewise.
10903 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10904 * optabs.c (expand_float): Likewise.
10905 (expand_fix): Likewise.
10906 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10909 Alan Hayward <alan.hayward@arm.com>
10910 David Sherwood <david.sherwood@arm.com>
10912 * optabs.c (expand_float): Explicitly check for scalars before
10913 using a branching expansion.
10914 (expand_fix): Likewise.
10916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10917 Alan Hayward <alan.hayward@arm.com>
10918 David Sherwood <david.sherwood@arm.com>
10920 * expr.c (convert_mode): Split scalar handling out into...
10921 (convert_mode_scalar): ...this new function. Treat the modes
10924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10925 Alan Hayward <alan.hayward@arm.com>
10926 David Sherwood <david.sherwood@arm.com>
10928 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10930 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10933 Alan Hayward <alan.hayward@arm.com>
10934 David Sherwood <david.sherwood@arm.com>
10936 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10937 rather than a machine_mode.
10938 (fixed_from_string): Likewise.
10939 (fixed_convert): Likewise.
10940 (fixed_convert_from_int): Likewise.
10941 (fixed_convert_from_real): Likewise.
10942 (real_convert_from_fixed): Likewise.
10943 * fixed-value.c (fixed_from_double_int): Likewise.
10944 (fixed_from_string): Likewise.
10945 (fixed_convert): Likewise.
10946 (fixed_convert_from_int): Likewise.
10947 (fixed_convert_from_real): Likewise.
10948 (real_convert_from_fixed): Likewise.
10949 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10952 Alan Hayward <alan.hayward@arm.com>
10953 David Sherwood <david.sherwood@arm.com>
10955 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10956 of separate mode class checks. Do not allow vector modes here.
10957 (immed_wide_int_const): Use as_a <scalar_mode>.
10958 * explow.c (trunc_int_for_mode): Likewise.
10959 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10960 (wi::shwi): Likewise.
10961 (wi::min_value): Likewise.
10962 (wi::max_value): Likewise.
10963 * dwarf2out.c (loc_descriptor): Likewise.
10964 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10965 for CONST_WIDE_INT.
10967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10968 Alan Hayward <alan.hayward@arm.com>
10969 David Sherwood <david.sherwood@arm.com>
10971 * tree.h (SCALAR_TYPE_MODE): New macro.
10972 * expr.c (expand_expr_addr_expr_1): Use it.
10973 (expand_expr_real_2): Likewise.
10974 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10975 (fold_convert_const_fixed_from_int): Likewise.
10976 (fold_convert_const_fixed_from_real): Likewise.
10977 (native_encode_fixed): Likewise
10978 (native_encode_complex): Likewise
10979 (native_encode_vector): Likewise.
10980 (native_interpret_fixed): Likewise.
10981 (native_interpret_real): Likewise.
10982 (native_interpret_complex): Likewise.
10983 (native_interpret_vector): Likewise.
10984 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10985 (simd_clone_adjust_argument_types): Likewise.
10986 (simd_clone_init_simd_arrays): Likewise.
10987 (simd_clone_adjust): Likewise.
10988 * stor-layout.c (layout_type): Likewise.
10989 * tree.c (build_minus_one_cst): Likewise.
10990 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10991 * tree-inline.c (estimate_move_cost): Likewise.
10992 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10993 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10994 (vectorizable_reduction): Likewise.
10995 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10996 (vect_recog_mixed_size_cond_pattern): Likewise.
10997 (check_bool_pattern): Likewise.
10998 (adjust_bool_pattern): Likewise.
10999 (search_type_for_mask_1): Likewise.
11000 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11001 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11002 (vectorizable_load): Likewise.
11003 (vectorizable_store): Likewise.
11004 * ubsan.c (ubsan_encode_value): Likewise.
11005 * varasm.c (output_constant): Likewise.
11007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11008 Alan Hayward <alan.hayward@arm.com>
11009 David Sherwood <david.sherwood@arm.com>
11011 * coretypes.h (scalar_mode): New class.
11012 * machmode.h (scalar_mode): Likewise.
11013 (scalar_mode::includes_p): New function.
11014 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11015 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11016 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11017 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11018 * expmed.c (store_bit_field_1): Likewise.
11019 (extract_bit_field_1): Likewise.
11020 * expr.c (write_complex_part): Likewise.
11021 (read_complex_part): Likewise.
11022 (emit_move_complex_push): Likewise.
11023 (expand_expr_real_2): Likewise.
11024 * function.c (assign_parm_setup_reg): Likewise.
11025 (assign_parms_unsplit_complex): Likewise.
11026 * optabs.c (expand_binop): Likewise.
11027 * rtlanal.c (subreg_get_info): Likewise.
11028 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11029 * varasm.c (output_constant_pool_2): Likewise.
11031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11032 Alan Hayward <alan.hayward@arm.com>
11033 David Sherwood <david.sherwood@arm.com>
11035 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11037 (expmed_mult_highpart_optab): Likewise.
11038 (expmed_mult_highpart): Likewise.
11040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11041 Alan Hayward <alan.hayward@arm.com>
11042 David Sherwood <david.sherwood@arm.com>
11044 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11045 instead of a machine_mode.
11046 (builtin_memset_read_str): Likewise.
11047 * builtins.c (c_readstr): Likewise.
11048 (builtin_memcpy_read_str): Likewise.
11049 (builtin_strncpy_read_str): Likewise.
11050 (builtin_memset_read_str): Likewise.
11051 (builtin_memset_gen_str): Likewise.
11052 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11053 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11054 instead of a machine_mode.
11055 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11057 (make_extraction): Likewise.
11058 (try_widen_shift_mode): Take and return scalar_int_modes instead
11060 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11061 a scalar_int_mode instead of a machine_mode.
11062 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11063 (avr_addr_space_pointer_mode): Likewise.
11064 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11065 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11066 (msp430_unwind_word_mode): Likewise.
11067 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11068 (spu_addr_space_pointer_mode): Likewise.
11069 (spu_addr_space_address_mode): Likewise.
11070 (spu_libgcc_cmp_return_mode): Likewise.
11071 (spu_libgcc_shift_count_mode): Likewise.
11072 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11073 (rl78_addr_space_pointer_mode): Likewise.
11074 (fl78_unwind_word_mode): Likewise.
11075 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11077 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11078 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11079 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11080 (mips_valid_pointer_mode): Likewise.
11081 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11082 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11083 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11085 (ft32_addr_space_address_mode): Likewise.
11086 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11087 scalar_int_mode instead of a machine_mode.
11088 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11090 (m32c_addr_space_address_mode): Likewise.
11091 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11092 (rs6000_eh_return_filter_mode): Likewise.
11093 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11094 (rs6000_eh_return_filter_mode): Likewise.
11095 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11096 (s390_libgcc_shift_count_mode): Likewise.
11097 (s390_unwind_word_mode): Likewise.
11098 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11100 * target.def (mode_rep_extended): Likewise.
11101 (valid_pointer_mode): Likewise.
11102 (addr_space.valid_pointer_mode): Likewise.
11103 (eh_return_filter_mode): Return a scalar_int_mode rather than
11105 (libgcc_cmp_return_mode): Likewise.
11106 (libgcc_shift_count_mode): Likewise.
11107 (unwind_word_mode): Likewise.
11108 (addr_space.pointer_mode): Likewise.
11109 (addr_space.address_mode): Likewise.
11110 * doc/tm.texi: Regenerate.
11111 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11113 (do_jump): Use scalar_int_mode for local variables.
11114 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11115 rather than a machine_mode.
11116 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11117 (scompare_loc_descriptor_wide): Likewise.
11118 (scompare_loc_descriptor_narrow): Likewise.
11119 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11121 * except.c (sjlj_emit_dispatch_table): Likewise.
11122 (expand_builtin_eh_copy_values): Likewise.
11123 * explow.c (convert_memory_address_addr_space_1): Likewise.
11124 Take a scalar_int_mode rather than a machine_mode.
11125 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11126 than a machine_mode.
11127 (memory_address_addr_space): Use scalar_int_mode for local variables.
11128 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11129 rather than a machine_mode.
11130 * expmed.c (mask_rtx): Likewise.
11131 (init_expmed_one_conv): Likewise.
11132 (expand_mult_highpart_adjust): Likewise.
11133 (extract_high_half): Likewise.
11134 (expmed_mult_highpart_optab): Likewise.
11135 (expmed_mult_highpart): Likewise.
11136 (expand_smod_pow2): Likewise.
11137 (expand_sdiv_pow2): Likewise.
11138 (emit_store_flag_int): Likewise.
11139 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11141 (extract_low_bits): Likewise.
11142 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11144 * expr.c (pieces_addr::adjust): Likewise.
11145 (can_store_by_pieces): Likewise.
11146 (store_by_pieces): Likewise.
11147 (clear_by_pieces_1): Likewise.
11148 (expand_expr_addr_expr_1): Likewise.
11149 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11150 (expand_expr_real_1): Likewise.
11151 (try_casesi): Likewise.
11152 * final.c (shorten_branches): Likewise.
11153 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11154 type of "mode" to machine_mode.
11155 * internal-fn.c (expand_arith_overflow_result_store): Take a
11156 scalar_int_mode rather than a machine_mode.
11157 (expand_mul_overflow): Use scalar_int_mode for local variables.
11158 * loop-doloop.c (doloop_modify): Likewise.
11159 (doloop_optimize): Likewise.
11160 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11161 than a machine_mode.
11162 (expand_doubleword_shift_condmove): Likewise.
11163 (expand_doubleword_shift): Likewise.
11164 (expand_doubleword_clz): Likewise.
11165 (expand_doubleword_popcount): Likewise.
11166 (expand_doubleword_parity): Likewise.
11167 (expand_absneg_bit): Use scalar_int_mode for local variables.
11168 (prepare_float_lib_cmp): Likewise.
11169 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11170 rather than a machine_mode.
11171 (convert_memory_address_addr_space): Likewise.
11172 (get_mode_bounds): Likewise.
11173 (get_address_mode): Return a scalar_int_mode rather than a
11175 * rtlanal.c (get_address_mode): Likewise.
11176 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11177 than a machine_mode.
11178 * targhooks.c (default_mode_rep_extended): Likewise.
11179 (default_valid_pointer_mode): Likewise.
11180 (default_addr_space_valid_pointer_mode): Likewise.
11181 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11182 than a machine_mode.
11183 (default_libgcc_cmp_return_mode): Likewise.
11184 (default_libgcc_shift_count_mode): Likewise.
11185 (default_unwind_word_mode): Likewise.
11186 (default_addr_space_pointer_mode): Likewise.
11187 (default_addr_space_address_mode): Likewise.
11188 * targhooks.h (default_eh_return_filter_mode): Likewise.
11189 (default_libgcc_cmp_return_mode): Likewise.
11190 (default_libgcc_shift_count_mode): Likewise.
11191 (default_unwind_word_mode): Likewise.
11192 (default_addr_space_pointer_mode): Likewise.
11193 (default_addr_space_address_mode): Likewise.
11194 (default_mode_rep_extended): Take a scalar_int_mode rather than
11196 (default_valid_pointer_mode): Likewise.
11197 (default_addr_space_valid_pointer_mode): Likewise.
11198 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11200 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11201 rather than a machine_mode.
11202 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11203 for local variables.
11204 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11205 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11206 than a machine_mode.
11208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11209 Alan Hayward <alan.hayward@arm.com>
11210 David Sherwood <david.sherwood@arm.com>
11212 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11213 the mode argument to scalar_int_mode.
11214 (do_jump_by_parts_zero_rtx): Likewise.
11215 (do_jump_by_parts_equality_rtx): Likewise.
11216 (do_jump_by_parts_greater): Take a mode argument.
11217 (do_jump_by_parts_equality): Likewise.
11218 (do_jump_1): Update calls accordingly.
11220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11221 Alan Hayward <alan.hayward@arm.com>
11222 David Sherwood <david.sherwood@arm.com>
11224 * is-a.h (safe_dyn_cast): New function.
11225 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11226 (jump_table_for_label): Likewise.
11227 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11228 instead of an rtx_insn *.
11229 (shorten_branches): Use dyn_cast instead of LABEL_P and
11230 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11231 rtx_jump_table_data::get_data_mode.
11232 (final_scan_insn): Likewise.
11234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11235 Alan Hayward <alan.hayward@arm.com>
11236 David Sherwood <david.sherwood@arm.com>
11238 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11239 trying to combine a full-register integer set with a subreg
11242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11243 Alan Hayward <alan.hayward@arm.com>
11244 David Sherwood <david.sherwood@arm.com>
11246 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11247 that is always either address_mode or pointer_mode.
11249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11250 Alan Hayward <alan.hayward@arm.com>
11251 David Sherwood <david.sherwood@arm.com>
11253 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11254 when the two are known to be equal.
11256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11257 Alan Hayward <alan.hayward@arm.com>
11258 David Sherwood <david.sherwood@arm.com>
11260 * simplify-rtx.c (simplify_const_unary_operation): Use
11261 is_a <scalar_int_mode> instead of checking for a nonzero
11262 precision. Forcibly convert op_mode to a scalar_int_mode
11263 in that case. More clearly differentiate the operand and
11264 result modes and use the former when deciding what the value
11265 of a count-bits operation should be. Use is_int_mode instead
11266 of checking for a MODE_INT. Remove redundant check for whether
11267 this mode has a zero precision.
11269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11270 Alan Hayward <alan.hayward@arm.com>
11271 David Sherwood <david.sherwood@arm.com>
11273 * optabs.c (widen_leading): Change the type of the mode argument
11274 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11275 (widen_bswap): Likewise.
11276 (expand_parity): Likewise.
11277 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11278 (expand_ffs): Likewise.
11279 (epand_unop): Check for scalar integer modes before calling the
11282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11283 Alan Hayward <alan.hayward@arm.com>
11284 David Sherwood <david.sherwood@arm.com>
11286 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11288 (expand_expr_real_1): Update call accordingly.
11290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11291 Alan Hayward <alan.hayward@arm.com>
11292 David Sherwood <david.sherwood@arm.com>
11294 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11295 value_mode arguments. Use scalar_int_mode internally.
11296 (store_bit_field_1): Rename the new integer mode from imode
11297 to op0_mode and use it instead of GET_MODE (op0). Update calls
11298 to store_split_bit_field, store_bit_field_using_insv and
11299 store_fixed_bit_field.
11300 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11301 Use scalar_int_mode internally. Use a bit count rather than a mode
11302 when calculating the largest bit size for get_best_mode.
11303 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11304 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11305 Remove assertion that OP0 has a scalar integer mode.
11306 (store_split_bit_field): Add op0_mode and value_mode arguments.
11307 Update calls to extract_fixed_bit_field.
11308 (extract_bit_field_using_extv): Add an op0_mode argument.
11309 Use scalar_int_mode internally.
11310 (extract_bit_field_1): Rename the new integer mode from imode to
11311 op0_mode and use it instead of GET_MODE (op0). Update calls to
11312 extract_split_bit_field, extract_bit_field_using_extv and
11313 extract_fixed_bit_field.
11314 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11315 to extract_split_bit_field and extract_fixed_bit_field_1.
11316 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11317 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11318 on the target mode.
11319 (extract_split_bit_field): Add an op0_mode argument. Update call
11320 to extract_fixed_bit_field.
11322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11323 Alan Hayward <alan.hayward@arm.com>
11324 David Sherwood <david.sherwood@arm.com>
11326 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11327 * explow.c (hard_function_value): Likewise.
11328 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11329 convert_to_mode call outside the loop.
11330 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11331 for the mode iterator. Require the mode specified by max_pieces
11333 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11335 (copy_blkmode_to_reg): Likewise.
11336 (set_storage_via_setmem): Likewise.
11337 * optabs.c (prepare_cmp_insn): Likewise.
11338 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11339 * stor-layout.c (finish_bitfield_representative): Likewise.
11341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11342 Alan Hayward <alan.hayward@arm.com>
11343 David Sherwood <david.sherwood@arm.com>
11345 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11346 * expr.c (convert_move): Use them.
11347 (convert_modes): Likewise.
11348 (store_expr_with_bounds): Likewise.
11350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11351 Alan Hayward <alan.hayward@arm.com>
11352 David Sherwood <david.sherwood@arm.com>
11354 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11355 parameter for the mode of "x". Remove the "known_x", "known_mode"
11356 and "known_ret" arguments. Change the type of the mode argument
11357 to scalar_int_mode.
11358 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11359 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11360 (reg_num_sign_bit_copies_for_combine): Likewise.
11361 * rtlanal.c (nonzero_bits1): Likewise.
11362 (num_sign_bit_copies1): Likewise.
11363 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11364 (reg_num_sign_bit_copies_general): Likewise.
11365 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11366 (reg_nonzero_bits_general): Likewise.
11368 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11369 Alan Hayward <alan.hayward@arm.com>
11370 David Sherwood <david.sherwood@arm.com>
11372 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11373 than in subroutines. Return 1 for non-integer modes.
11374 (cached_num_sign_bit_copies): Change the type of the mode parameter
11375 to scalar_int_mode.
11376 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11377 classes. Handle CONST_INT_P first and then check whether X also
11378 has a scalar integer mode. Check the same thing for inner registers
11379 of a SUBREG and for values that are being extended or truncated.
11381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11382 Alan Hayward <alan.hayward@arm.com>
11383 David Sherwood <david.sherwood@arm.com>
11385 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11386 in subroutines. Return the mode mask for non-integer modes.
11387 (cached_nonzero_bits): Change the type of the mode parameter
11388 to scalar_int_mode.
11389 (nonzero_bits1): Likewise. Remove early exit for other mode
11390 classes. Handle CONST_INT_P first and then check whether X
11391 also has a scalar integer mode.
11393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11394 Alan Hayward <alan.hayward@arm.com>
11395 David Sherwood <david.sherwood@arm.com>
11397 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11398 Return a scalar_int_mode and assert that the size is greater than
11400 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11401 (op_by_pieces_d::op_by_pieces_d): Likewise.
11402 (op_by_pieces_d::run): Likewise.
11403 (can_store_by_pieces): Likewise.
11405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11406 Alan Hayward <alan.hayward@arm.com>
11407 David Sherwood <david.sherwood@arm.com>
11409 * combine.c (extract_left_shift): Add a mode argument and update
11411 (make_compound_operation_int): Change the type of the mode parameter
11412 to scalar_int_mode and update the call to extract_left_shift.
11414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11415 Alan Hayward <alan.hayward@arm.com>
11416 David Sherwood <david.sherwood@arm.com>
11418 * combine.c (simplify_and_const_int): Change the type of the mode
11419 parameter to scalar_int_mode.
11420 (simplify_and_const_int_1): Likewise. Update recursive call.
11422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11423 Alan Hayward <alan.hayward@arm.com>
11424 David Sherwood <david.sherwood@arm.com>
11426 * combine.c (simplify_compare_const): Check that the mode is a
11427 scalar_int_mode (rather than VOIDmode) before testing its
11429 (simplify_comparison): Move COMPARISON_P handling out of the
11430 loop and restrict the latter part of the loop to scalar_int_modes.
11431 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11432 and when considering SUBREG_REGs. Use is_int_mode instead of
11433 checking GET_MODE_CLASS against MODE_INT.
11435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11436 Alan Hayward <alan.hayward@arm.com>
11437 David Sherwood <david.sherwood@arm.com>
11439 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11440 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11441 shift_unit_mode and for modes involved in scalar shifts.
11443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11444 Alan Hayward <alan.hayward@arm.com>
11445 David Sherwood <david.sherwood@arm.com>
11447 * combine.c (force_int_to_mode): New function, split out from...
11448 (force_to_mode): ...here. Keep xmode up-to-date and use it
11449 instead of GET_MODE (x).
11451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11452 Alan Hayward <alan.hayward@arm.com>
11453 David Sherwood <david.sherwood@arm.com>
11455 * optabs-query.h (extraction_insn::struct_mode): Change type to
11456 opt_scalar_int_mode and update comment.
11457 (extraction_insn::field_mode): Change type to scalar_int_mode.
11458 (extraction_insn::pos_mode): Likewise.
11459 * combine.c (make_extraction): Update accordingly.
11460 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11461 (get_optab_extraction_insn): Likewise.
11462 * recog.c (simplify_while_replacing): Likewise.
11463 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11464 parameter to opt_scalar_int_mode.
11466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11467 Alan Hayward <alan.hayward@arm.com>
11468 David Sherwood <david.sherwood@arm.com>
11470 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11471 to a scalar_int_mode instead of a machine_mode.
11472 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11473 (get_best_mode): Return a boolean and use a pointer argument to store
11474 the selected mode. Replace the limit mode parameter with a bit limit.
11475 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11476 for the values returned by bit_field_mode_iterator::next_mode.
11477 (store_bit_field): Update call to get_best_mode.
11478 (store_fixed_bit_field): Likewise.
11479 (extract_fixed_bit_field): Likewise.
11480 * expr.c (optimize_bitfield_assignment_op): Likewise.
11481 * fold-const.c (optimize_bit_field_compare): Likewise.
11482 (fold_truth_andor_1): Likewise.
11483 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11484 Update for new type of m_mode.
11485 (get_best_mode): As above.
11487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11488 Alan Hayward <alan.hayward@arm.com>
11489 David Sherwood <david.sherwood@arm.com>
11491 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11492 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11493 (store_bit_field): Check is_a <scalar_int_mode> before calling
11494 strict_volatile_bitfield_p.
11495 (extract_bit_field): Likewise.
11497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11498 Alan Hayward <alan.hayward@arm.com>
11499 David Sherwood <david.sherwood@arm.com>
11501 * target.def (cstore_mode): Return a scalar_int_mode.
11502 * doc/tm.texi: Regenerate.
11503 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11504 * targhooks.h (default_cstore_mode): Likewise.
11505 * targhooks.c (default_cstore_mode): Likewise, using a forced
11507 * expmed.c (emit_cstore): Expect the target of the cstore to be
11510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11511 Alan Hayward <alan.hayward@arm.com>
11512 David Sherwood <david.sherwood@arm.com>
11514 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11516 (niter_desc): Likewise mode.
11517 (iv_analyze): Add a mode parameter.
11519 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11520 and change its type to scalar_int_mode.
11521 * loop-iv.c: Update commentary at head of file.
11522 (iv_constant): Pass the mode paraeter before the rtx it describes
11523 and change its type to scalar_int_mode. Remove VOIDmode handling.
11524 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11525 (iv_extend): Likewise.
11526 (shorten_into_mode): Likewise.
11527 (iv_add): Use scalar_int_mode.
11528 (iv_mult): Likewise.
11529 (iv_shift): Likewise.
11530 (canonicalize_iv_subregs): Likewise.
11531 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11532 it describes and change its mode to scalar_int_mode. Also change
11533 the type of the returned inner_mode to scalar_int_mode.
11534 (get_biv_step): Likewise, turning outer_mode from a pointer
11535 into a direct parameter. Update call to get_biv_step_1.
11536 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11537 iv_constant and get_biv_step.
11538 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11539 and change its type to scalar_int_mode. Don't initialise iv->mode
11540 to VOIDmode and remove later checks for its still being VOIDmode.
11541 Update calls to iv_analyze_op and iv_analyze_expr. Check
11542 is_a <scalar_int_mode> when changing the mode under consideration.
11543 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11544 Update call to iv_analyze_expr.
11545 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11546 inner register is not also a scalar_int_mode. Update call to
11548 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11549 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11550 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11551 separate mode class checks. Update calls to iv_analyze. Remove
11552 fix-up of VOIDmodes after iv_analyze_biv.
11553 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11554 don't have a scalar_int_mode. Update call to biv_p.
11556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11557 Alan Hayward <alan.hayward@arm.com>
11558 David Sherwood <david.sherwood@arm.com>
11560 * cfgexpand.c (convert_debug_memory_address): Use
11561 as_a <scalar_int_mode>.
11562 * combine.c (expand_compound_operation): Likewise.
11563 (make_extraction): Likewise.
11564 (change_zero_ext): Likewise.
11565 (simplify_comparison): Likewise.
11566 * cse.c (cse_insn): Likewise.
11567 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11568 (mem_loc_descriptor): Likewise.
11569 (loc_descriptor): Likewise.
11570 * expmed.c (init_expmed_one_mode): Likewise.
11571 (synth_mult): Likewise.
11572 (emit_store_flag_1): Likewise.
11573 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11574 of a comparison with size.
11575 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11576 (reduce_to_bit_field_precision): Likewise.
11577 * function.c (expand_function_end): Likewise.
11578 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11579 * loop-doloop.c (doloop_modify): Likewise.
11580 * optabs.c (expand_binop): Likewise.
11581 (expand_unop): Likewise.
11582 (expand_copysign_absneg): Likewise.
11583 (prepare_cmp_insn): Likewise.
11584 (maybe_legitimize_operand): Likewise.
11585 * recog.c (const_scalar_int_operand): Likewise.
11586 * rtlanal.c (get_address_mode): Likewise.
11587 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11588 (simplify_cond_clz_ctz): Likewise.
11589 * tree-nested.c (get_nl_goto_field): Likewise.
11590 * tree.c (build_vector_type_for_mode): Likewise.
11591 * var-tracking.c (use_narrower_mode): Likewise.
11593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11594 Alan Hayward <alan.hayward@arm.com>
11595 David Sherwood <david.sherwood@arm.com>
11597 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11598 * builtins.c (expand_builtin_signbit): Use it.
11599 * cfgexpand.c (expand_debug_expr): Likewise.
11600 * dojump.c (do_jump): Likewise.
11601 (do_compare_and_jump): Likewise.
11602 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11603 * expmed.c (make_tree): Likewise.
11604 * expr.c (expand_expr_real_2): Likewise.
11605 (expand_expr_real_1): Likewise.
11606 (try_casesi): Likewise.
11607 * fold-const-call.c (fold_const_call_ss): Likewise.
11608 * fold-const.c (unextend): Likewise.
11609 (extract_muldiv_1): Likewise.
11610 (fold_single_bit_test): Likewise.
11611 (native_encode_int): Likewise.
11612 (native_encode_string): Likewise.
11613 (native_interpret_int): Likewise.
11614 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11615 * internal-fn.c (expand_addsub_overflow): Likewise.
11616 (expand_neg_overflow): Likewise.
11617 (expand_mul_overflow): Likewise.
11618 (expand_arith_overflow): Likewise.
11619 * match.pd: Likewise.
11620 * stor-layout.c (layout_type): Likewise.
11621 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11622 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11623 * tree-ssanames.c (get_range_info): Likewise.
11624 * tree-switch-conversion.c (array_value_type) Likewise.
11625 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11626 (vect_recog_divmod_pattern): Likewise.
11627 (vect_recog_mixed_size_cond_pattern): Likewise.
11628 * tree-vrp.c (extract_range_basic): Likewise.
11629 (simplify_float_conversion_using_ranges): Likewise.
11630 * tree.c (int_fits_type_p): Likewise.
11631 * ubsan.c (instrument_bool_enum_load): Likewise.
11632 * varasm.c (mergeable_string_section): Likewise.
11633 (narrowing_initializer_constant_valid_p): Likewise.
11634 (output_constant): Likewise.
11636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11637 Alan Hayward <alan.hayward@arm.com>
11638 David Sherwood <david.sherwood@arm.com>
11640 * machmode.h (NARROWEST_INT_MODE): New macro.
11641 * expr.c (alignment_for_piecewise_move): Use it instead of
11642 GET_CLASS_NARROWEST_MODE (MODE_INT).
11643 (push_block): Likewise.
11644 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11646 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11649 Alan Hayward <alan.hayward@arm.com>
11650 David Sherwood <david.sherwood@arm.com>
11652 * postreload.c (move2add_valid_value_p): Change the type of the
11653 mode parameter to scalar_int_mode.
11654 (move2add_use_add2_insn): Add a mode parameter and use it instead
11656 (move2add_use_add3_insn): Likewise.
11657 (reload_cse_move2add): Update accordingly.
11659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11660 Alan Hayward <alan.hayward@arm.com>
11661 David Sherwood <david.sherwood@arm.com>
11663 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11665 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11666 * optabs.c (expand_unop): Likewise.
11668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11669 Alan Hayward <alan.hayward@arm.com>
11670 David Sherwood <david.sherwood@arm.com>
11672 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11673 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11674 (popcount_loc_descriptor): Likewise.
11675 (bswap_loc_descriptor): Likewise.
11676 (rotate_loc_descriptor): Likewise.
11677 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11678 calling the functions above.
11680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11681 Alan Hayward <alan.hayward@arm.com>
11682 David Sherwood <david.sherwood@arm.com>
11684 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11686 (try_combine): Likewise.
11687 (simplify_if_then_else): Likewise.
11688 * cse.c (cse_insn): Likewise.
11689 * dwarf2out.c (mem_loc_descriptor): Likewise.
11690 * emit-rtl.c (gen_lowpart_common): Likewise.
11691 * simplify-rtx.c (simplify_truncation): Likewise.
11692 (simplify_binary_operation_1): Likewise.
11693 (simplify_const_relational_operation): Likewise.
11694 (simplify_ternary_operation): Likewise.
11695 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11698 Alan Hayward <alan.hayward@arm.com>
11699 David Sherwood <david.sherwood@arm.com>
11701 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11702 * reload.c (push_reload): Likewise.
11703 (find_reloads): Likewise.
11705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11706 Alan Hayward <alan.hayward@arm.com>
11707 David Sherwood <david.sherwood@arm.com>
11709 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11710 (make_compound_operation_int): Likewise.
11711 (change_zero_ext): Likewise.
11712 * expr.c (convert_move): Likewise.
11713 (convert_modes): Likewise.
11714 * fwprop.c (forward_propagate_subreg): Likewise.
11715 * loop-iv.c (get_biv_step_1): Likewise.
11716 * optabs.c (widen_operand): Likewise.
11717 * postreload.c (move2add_valid_value_p): Likewise.
11718 * recog.c (simplify_while_replacing): Likewise.
11719 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11720 (simplify_binary_operation_1): Likewise. Remove redundant
11721 mode equality check.
11723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11724 Alan Hayward <alan.hayward@arm.com>
11725 David Sherwood <david.sherwood@arm.com>
11727 * combine.c (combine_simplify_rtx): Add checks for
11728 is_a <scalar_int_mode>.
11729 (simplify_if_then_else): Likewise.
11730 (make_field_assignment): Likewise.
11731 (simplify_comparison): Likewise.
11732 * ifcvt.c (noce_try_bitop): Likewise.
11733 * loop-invariant.c (canonicalize_address_mult): Likewise.
11734 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11737 Alan Hayward <alan.hayward@arm.com>
11738 David Sherwood <david.sherwood@arm.com>
11740 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11741 is_a <scalar_int_mode> instead of != BLKmode.
11743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11744 Alan Hayward <alan.hayward@arm.com>
11745 David Sherwood <david.sherwood@arm.com>
11747 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11748 instead of != VOIDmode.
11749 * combine.c (if_then_else_cond): Likewise.
11750 (change_zero_ext): Likewise.
11751 * dwarf2out.c (mem_loc_descriptor): Likewise.
11752 (loc_descriptor): Likewise.
11753 * rtlanal.c (canonicalize_condition): Likewise.
11754 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11757 Alan Hayward <alan.hayward@arm.com>
11758 David Sherwood <david.sherwood@arm.com>
11760 * simplify-rtx.c (simplify_binary_operation_1): Use
11761 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11764 Alan Hayward <alan.hayward@arm.com>
11765 David Sherwood <david.sherwood@arm.com>
11767 * wide-int.h (int_traits<unsigned char>) New class.
11768 (int_traits<unsigned short>) Likewise.
11769 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11770 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11772 * combine.c (set_nonzero_bits_and_sign_copies): Use
11773 is_a <scalar_int_mode>.
11774 (find_split_point): Likewise.
11775 (combine_simplify_rtx): Likewise.
11776 (simplify_logical): Likewise.
11777 (expand_compound_operation): Likewise.
11778 (expand_field_assignment): Likewise.
11779 (make_compound_operation): Likewise.
11780 (extended_count): Likewise.
11781 (change_zero_ext): Likewise.
11782 (simplify_comparison): Likewise.
11783 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11784 (ucompare_loc_descriptor): Likewise.
11785 (minmax_loc_descriptor): Likewise.
11786 (mem_loc_descriptor): Likewise.
11787 (loc_descriptor): Likewise.
11788 * expmed.c (init_expmed_one_mode): Likewise.
11789 * lra-constraints.c (lra_constraint_offset): Likewise.
11790 * optabs.c (prepare_libcall_arg): Likewise.
11791 * postreload.c (move2add_note_store): Likewise.
11792 * reload.c (operands_match_p): Likewise.
11793 * rtl.h (load_extend_op): Likewise.
11794 * rtlhooks.c (gen_lowpart_general): Likewise.
11795 * simplify-rtx.c (simplify_truncation): Likewise.
11796 (simplify_unary_operation_1): Likewise.
11797 (simplify_binary_operation_1): Likewise.
11798 (simplify_const_binary_operation): Likewise.
11799 (simplify_const_relational_operation): Likewise.
11800 (simplify_subreg): Likewise.
11801 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11802 * var-tracking.c (adjust_mems): Likewise.
11803 (prepare_call_arguments): Likewise.
11805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11806 Alan Hayward <alan.hayward@arm.com>
11807 David Sherwood <david.sherwood@arm.com>
11809 * machmode.h (is_int_mode): New fuction.
11810 * combine.c (find_split_point): Use it.
11811 (combine_simplify_rtx): Likewise.
11812 (simplify_if_then_else): Likewise.
11813 (simplify_set): Likewise.
11814 (simplify_shift_const_1): Likewise.
11815 (simplify_comparison): Likewise.
11816 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11817 * cse.c (notreg_cost): Likewise.
11818 (cse_insn): Likewise.
11819 * cselib.c (cselib_lookup_1): Likewise.
11820 * dojump.c (do_jump_1): Likewise.
11821 (do_compare_rtx_and_jump): Likewise.
11822 * dse.c (get_call_args): Likewise.
11823 * dwarf2out.c (rtl_for_decl_init): Likewise.
11824 (native_encode_initializer): Likewise.
11825 * expmed.c (emit_store_flag_1): Likewise.
11826 (emit_store_flag): Likewise.
11827 * expr.c (convert_modes): Likewise.
11828 (store_field): Likewise.
11829 (expand_expr_real_1): Likewise.
11830 * fold-const.c (fold_read_from_constant_string): Likewise.
11831 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11832 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11833 * optabs.c (expand_binop): Likewise.
11834 (expand_unop): Likewise.
11835 (expand_abs_nojump): Likewise.
11836 (expand_one_cmpl_abs_nojump): Likewise.
11837 * simplify-rtx.c (mode_signbit_p): Likewise.
11838 (val_signbit_p): Likewise.
11839 (val_signbit_known_set_p): Likewise.
11840 (val_signbit_known_clear_p): Likewise.
11841 (simplify_relational_operation_1): Likewise.
11842 * tree.c (vector_type_mode): Likewise.
11844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11845 Alan Hayward <alan.hayward@arm.com>
11846 David Sherwood <david.sherwood@arm.com>
11848 * machmode.h (smallest_mode_for_size): Fix formatting.
11849 (smallest_int_mode_for_size): New function.
11850 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11851 instead of smallest_mode_for_size.
11852 * combine.c (make_extraction): Likewise.
11853 * config/arc/arc.c (arc_expand_movmem): Likewise.
11854 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11855 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11856 * config/s390/s390.c (s390_expand_insv): Likewise.
11857 * config/sparc/sparc.c (assign_int_registers): Likewise.
11858 * config/spu/spu.c (spu_function_value): Likewise.
11859 (spu_function_arg): Likewise.
11860 * coverage.c (get_gcov_type): Likewise.
11861 (get_gcov_unsigned_t): Likewise.
11862 * dse.c (find_shift_sequence): Likewise.
11863 * expmed.c (store_bit_field_1): Likewise.
11864 * expr.c (convert_move): Likewise.
11865 (store_field): Likewise.
11866 * internal-fn.c (expand_arith_overflow): Likewise.
11867 * optabs-query.c (get_best_extraction_insn): Likewise.
11868 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11869 * stor-layout.c (layout_type): Likewise.
11870 (initialize_sizetypes): Likewise.
11871 * targhooks.c (default_get_mask_mode): Likewise.
11872 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11875 Alan Hayward <alan.hayward@arm.com>
11876 David Sherwood <david.sherwood@arm.com>
11878 * machmode.h (opt_mode::else_blk): New function.
11879 (int_mode_for_mode): Declare.
11880 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11881 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11883 * cfgexpand.c (expand_debug_expr): Likewise.
11884 * combine.c (gen_lowpart_or_truncate): Likewise.
11885 (gen_lowpart_for_combine): Likewise.
11886 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11887 * config/avr/avr.c (avr_to_int_mode): Likewise.
11888 (avr_out_plus_1): Likewise.
11889 (avr_out_plus): Likewise.
11890 (avr_out_round): Likewise.
11891 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11892 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11893 (s390_expand_vcond): Likewise.
11894 * config/spu/spu.c (spu_split_immediate): Likewise.
11895 (spu_expand_mov): Likewise.
11896 * dse.c (get_stored_val): Likewise.
11897 * expmed.c (store_bit_field_1): Likewise.
11898 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11900 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11901 (extract_low_bits): Likewise.
11902 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11903 handling rather than repeating the check.
11904 (emit_group_store): Likewise.
11905 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11906 * optabs.c (expand_absneg_bit): Likewise.
11907 (expand_copysign_absneg): Likewise.
11908 (expand_copysign_bit): Likewise.
11909 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11910 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11911 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11912 * var-tracking.c (prepare_call_arguments): Likewise.
11913 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11914 int_mode_for_mode instead of mode_for_size.
11915 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11918 Alan Hayward <alan.hayward@arm.com>
11919 David Sherwood <david.sherwood@arm.com>
11921 * machmode.h (int_mode_for_size): New function.
11922 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11923 instead of mode_for_size.
11924 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11926 * combine.c (expand_field_assignment): Use int_mode_for_size
11927 instead of mode_for_size.
11928 (make_extraction): Likewise.
11929 (simplify_shift_const_1): Likewise.
11930 (simplify_comparison): Likewise.
11931 * dojump.c (do_jump): Likewise.
11932 * dwarf2out.c (mem_loc_descriptor): Likewise.
11933 * emit-rtl.c (init_derived_machine_modes): Likewise.
11934 * expmed.c (flip_storage_order): Likewise.
11935 (convert_extracted_bit_field): Likewise.
11936 * expr.c (copy_blkmode_from_reg): Likewise.
11937 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11938 * internal-fn.c (expand_mul_overflow): Likewise.
11939 * lower-subreg.c (simple_move): Likewise.
11940 * optabs-libfuncs.c (init_optabs): Likewise.
11941 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11942 * tree.c (vector_type_mode): Likewise.
11943 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11944 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11945 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11946 * tree-vect-stmts.c (vectorizable_load): Likewise.
11947 (vectorizable_store): Likewise.
11949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11950 Alan Hayward <alan.hayward@arm.com>
11951 David Sherwood <david.sherwood@arm.com>
11953 * coretypes.h (pod_mode): New type.
11954 (scalar_int_mode_pod): New typedef.
11955 * machmode.h (pod_mode): New class.
11956 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11957 * genmodes.c (emit_mode_int_n): Update accordingly.
11958 * lower-subreg.h (target_lower_subreg): Change type to
11959 scalar_int_mode_pod.
11960 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11961 scalar_int_mode_pod.
11963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11964 Alan Hayward <alan.hayward@arm.com>
11965 David Sherwood <david.sherwood@arm.com>
11967 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11968 machine_mode to scalar_int_mode.
11969 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11970 (rs6000_option_override_internal): Remove cast to int.
11971 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11972 machine_mode to scalar_int_mode.
11973 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11974 (rs6000_option_override_internal): Remove cast to int.
11975 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11976 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11978 * config/s390/s390.c (s390_expand_builtin): Likewise.
11979 * coretypes.h (scalar_int_mode): New type.
11980 (opt_scalar_int_mode): New typedef.
11981 * machmode.h (scalar_int_mode): New class.
11982 (scalar_int_mode::includes_p): New function.
11983 (byte_mode): Change type to scalar_int_mode.
11984 (word_mode): Likewise.
11985 (ptr_mode): Likewise.
11986 * emit-rtl.c (byte_mode): Likewise.
11987 (word_mode): Likewise.
11988 (ptr_mode): Likewise.
11989 (init_derived_machine_modes): Update accordingly.
11990 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11991 and MODE_PARTIAL_INT.
11992 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11993 opt_scalar_int_mode.
11995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11996 Alan Hayward <alan.hayward@arm.com>
11997 David Sherwood <david.sherwood@arm.com>
11999 * target.def (libgcc_floating_mode_supported_p): Take a
12001 * doc/tm.texi: Regenerate.
12002 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12004 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12005 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12009 Alan Hayward <alan.hayward@arm.com>
12010 David Sherwood <david.sherwood@arm.com>
12012 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12013 * doc/tm.texi: Regenerate.
12014 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12015 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12016 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12017 * targhooks.h (default_floatn_mode): Likewise.
12018 * targhooks.c (default_floatn_mode): Likewise.
12019 * tree.c (build_common_tree_nodes): Update accordingly.
12021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12022 Alan Hayward <alan.hayward@arm.com>
12023 David Sherwood <david.sherwood@arm.com>
12025 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12026 (mode_iterator::iterate_p): Likewise.
12027 (mode_iterator::get_wider): Likewise.
12028 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12031 Alan Hayward <alan.hayward@arm.com>
12032 David Sherwood <david.sherwood@arm.com>
12034 * coretypes.h (opt_scalar_float_mode): New typedef.
12035 * machmode.h (float_mode_for_size): New function.
12036 * emit-rtl.c (double_mode): Delete.
12037 (init_emit_once): Use float_mode_for_size.
12038 * stor-layout.c (layout_type): Likewise.
12039 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12042 Alan Hayward <alan.hayward@arm.com>
12043 David Sherwood <david.sherwood@arm.com>
12045 * output.h (assemble_real): Take a scalar_float_mode.
12046 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12047 * config/arm/arm.md (consttable_4): Likewise.
12048 (consttable_8): Likewise.
12049 (consttable_16): Likewise.
12050 * config/mips/mips.md (consttable_float): Likewise.
12051 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12052 * varasm.c (assemble_real): Take a scalar_float_mode.
12053 (output_constant_pool_2): Update accordingly.
12054 (output_constant): Likewise.
12056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12057 Alan Hayward <alan.hayward@arm.com>
12058 David Sherwood <david.sherwood@arm.com>
12060 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12061 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12062 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12063 (native_encode_real): Likewise.
12064 (native_interpret_real): Likewise.
12065 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12066 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12069 Alan Hayward <alan.hayward@arm.com>
12070 David Sherwood <david.sherwood@arm.com>
12072 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12073 <scalar_float_mode>. Simplify.
12074 (gen_extend_conv_libfunc): Likewise.
12076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12077 Alan Hayward <alan.hayward@arm.com>
12078 David Sherwood <david.sherwood@arm.com>
12080 * coretypes.h (scalar_float_mode): New type.
12081 * machmode.h (mode_traits::from_int): Use machine_mode if
12082 USE_ENUM_MODES is defined.
12083 (is_a): New function.
12085 (dyn_cast): Likewise.
12086 (scalar_float_mode): New class.
12087 (scalar_float_mode::includes_p): New function.
12088 (is_float_mode): Likewise.
12089 * gdbhooks.py (MachineModePrinter): New class.
12090 (build_pretty_printer): Use it for scalar_float_mode.
12091 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12092 (format_helper::format_helper): Turn into a template.
12093 * genmodes.c (get_mode_class): New function.
12094 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12095 or machine_mode if none.
12096 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12097 as_a <scalar_float_mode>.
12098 * dwarf2out.c (mem_loc_descriptor): Likewise.
12099 (insert_float): Likewise.
12100 (add_const_value_attribute): Likewise.
12101 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12102 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12103 (expand_unop): Update accordingly.
12104 (expand_abs_nojump): Likewise.
12105 (expand_copysign_absneg): Take a scalar_float_mode.
12106 (expand_copysign_bit): Likewise.
12107 (expand_copysign): Update accordingly.
12109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12110 Alan Hayward <alan.hayward@arm.com>
12111 David Sherwood <david.sherwood@arm.com>
12113 * coretypes.h (opt_mode): New class.
12114 * machmode.h (opt_mode): Likewise.
12115 (opt_mode::else_void): New function.
12116 (opt_mode::require): Likewise.
12117 (opt_mode::exists): Likewise.
12118 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12119 (GET_MODE_2XWIDER_MODE): Likewise.
12120 (mode_iterator::get_wider): Update accordingly.
12121 (mode_iterator::get_2xwider): Likewise.
12122 (mode_iterator::get_known_wider): Likewise, turning into a template.
12123 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12124 forcing a wider mode to exist.
12125 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12126 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12127 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12128 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12129 * lower-subreg.c (init_lower_subreg): Likewise.
12130 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12131 on the final iteration.
12132 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12133 a wider mode exists before asking for a move pattern.
12134 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12135 forcing a wider mode to exist.
12136 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12137 returning false if no such mode exists.
12138 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12139 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12140 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12141 Avoid checking for a MODE_INT if we already know the mode is not a
12143 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12144 forcing a wider mode to exist.
12145 (expmed_mult_highpart_optab): Likewise.
12146 (expmed_mult_highpart): Likewise.
12147 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12149 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12150 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12151 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12152 * internal-fn.c (expand_mul_overflow): Update use of
12153 GET_MODE_2XWIDER_MODE.
12154 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12155 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12156 GET_MODE_WIDER_MODE.
12157 (convert_plusminus_to_widen): Likewise.
12158 * tree-switch-conversion.c (array_value_type): Likewise.
12159 * var-tracking.c (emit_note_insn_var_location): Likewise.
12160 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12161 Return false inside rather than outside the loop if no wider mode
12163 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12164 and GET_MODE_2XWIDER_MODE
12165 (can_compare_p): Use else_void.
12166 * gdbhooks.py (OptMachineModePrinter): New class.
12167 (build_pretty_printer): Use it for opt_mode.
12169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12170 Alan Hayward <alan.hayward@arm.com>
12171 David Sherwood <david.sherwood@arm.com>
12173 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12174 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12177 Alan Hayward <alan.hayward@arm.com>
12178 David Sherwood <david.sherwood@arm.com>
12180 * machmode.h (mode_traits): New structure.
12181 (get_narrowest_mode): New function.
12182 (mode_iterator::start): Likewise.
12183 (mode_iterator::iterate_p): Likewise.
12184 (mode_iterator::get_wider): Likewise.
12185 (mode_iterator::get_known_wider): Likewise.
12186 (mode_iterator::get_2xwider): Likewise.
12187 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12188 (FOR_EACH_MODE): Likewise.
12189 (FOR_EACH_MODE_FROM): Likewise.
12190 (FOR_EACH_MODE_UNTIL): Likewise.
12191 (FOR_EACH_WIDER_MODE): Likewise.
12192 (FOR_EACH_2XWIDER_MODE): Likewise.
12193 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12194 * combine.c (simplify_comparison): Likewise
12195 * config/i386/i386.c (type_natural_mode): Likewise.
12196 * cse.c (cse_insn): Likewise.
12197 * dse.c (find_shift_sequence): Likewise.
12198 * emit-rtl.c (init_derived_machine_modes): Likewise.
12199 (init_emit_once): Likewise.
12200 * explow.c (hard_function_value): Likewise.
12201 * expmed.c (extract_fixed_bit_field_1): Likewise.
12202 (extract_bit_field_1): Likewise.
12203 (expand_divmod): Likewise.
12204 (emit_store_flag_1): Likewise.
12205 * expr.c (init_expr_target): Likewise.
12206 (convert_move): Likewise.
12207 (alignment_for_piecewise_move): Likewise.
12208 (widest_int_mode_for_size): Likewise.
12209 (emit_block_move_via_movmem): Likewise.
12210 (copy_blkmode_to_reg): Likewise.
12211 (set_storage_via_setmem): Likewise.
12212 (compress_float_constant): Likewise.
12213 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12214 * optabs-query.c (get_best_extraction_insn): Likewise.
12215 * optabs.c (expand_binop): Likewise.
12216 (expand_twoval_unop): Likewise.
12217 (expand_twoval_binop): Likewise.
12218 (widen_leading): Likewise.
12219 (widen_bswap): Likewise.
12220 (expand_parity): Likewise.
12221 (expand_unop): Likewise.
12222 (prepare_cmp_insn): Likewise.
12223 (prepare_float_lib_cmp): Likewise.
12224 (expand_float): Likewise.
12225 (expand_fix): Likewise.
12226 (expand_sfix_optab): Likewise.
12227 * postreload.c (move2add_use_add2_insn): Likewise.
12228 * reg-stack.c (reg_to_stack): Likewise.
12229 * reginfo.c (choose_hard_reg_mode): Likewise.
12230 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12231 * stor-layout.c (mode_for_size): Likewise.
12232 (smallest_mode_for_size): Likewise.
12233 (mode_for_vector): Likewise.
12234 (finish_bitfield_representative): Likewise.
12235 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12236 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12237 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12238 * var-tracking.c (prepare_call_arguments): Likewise.
12240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12241 Alan Hayward <alan.hayward@arm.com>
12242 David Sherwood <david.sherwood@arm.com>
12244 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12245 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12246 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12247 * machmode.h (mode_size): Move earlier in file.
12248 (mode_precision): Likewise.
12249 (mode_inner): Likewise.
12250 (mode_nunits): Likewise.
12251 (mode_unit_size): Likewise.
12252 (unit_unit_precision): Likewise.
12253 (mode_wider): Likewise.
12254 (mode_2xwider): Likewise.
12255 (machine_mode): New class.
12256 (mode_to_bytes): New function.
12257 (mode_to_bits): Likewise.
12258 (mode_to_precision): Likewise.
12259 (mode_to_inner): Likewise.
12260 (mode_to_unit_size): Likewise.
12261 (mode_to_unit_precision): Likewise.
12262 (mode_to_nunits): Likewise.
12263 (GET_MODE_SIZE): Use mode_to_bytes.
12264 (GET_MODE_BITSIZE): Use mode_to_bits.
12265 (GET_MODE_PRECISION): Use mode_to_precision.
12266 (GET_MODE_INNER): Use mode_to_inner.
12267 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12268 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12269 (GET_MODE_NUNITS): Use mode_to_nunits.
12270 * system.h (ALWAYS_INLINE): New macro.
12271 * config/powerpcspe/powerpcspe-c.c
12272 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12273 int for arg1_mode and arg2_mode.
12275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12276 Alan Hayward <alan.hayward@arm.com>
12277 David Sherwood <david.sherwood@arm.com>
12279 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12280 Prefix mode names with E_ in case statements.
12281 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12282 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12283 (aarch64_split_simd_move): Likewise.
12284 (aarch64_gen_storewb_pair): Likewise.
12285 (aarch64_gen_loadwb_pair): Likewise.
12286 (aarch64_gen_store_pair): Likewise.
12287 (aarch64_gen_load_pair): Likewise.
12288 (aarch64_get_condition_code_1): Likewise.
12289 (aarch64_constant_pool_reload_icode): Likewise.
12290 (get_rsqrte_type): Likewise.
12291 (get_rsqrts_type): Likewise.
12292 (get_recpe_type): Likewise.
12293 (get_recps_type): Likewise.
12294 (aarch64_gimplify_va_arg_expr): Likewise.
12295 (aarch64_simd_container_mode): Likewise.
12296 (aarch64_emit_load_exclusive): Likewise.
12297 (aarch64_emit_store_exclusive): Likewise.
12298 (aarch64_expand_compare_and_swap): Likewise.
12299 (aarch64_gen_atomic_cas): Likewise.
12300 (aarch64_emit_bic): Likewise.
12301 (aarch64_emit_atomic_swap): Likewise.
12302 (aarch64_emit_atomic_load_op): Likewise.
12303 (aarch64_evpc_trn): Likewise.
12304 (aarch64_evpc_uzp): Likewise.
12305 (aarch64_evpc_zip): Likewise.
12306 (aarch64_evpc_ext): Likewise.
12307 (aarch64_evpc_rev): Likewise.
12308 (aarch64_evpc_dup): Likewise.
12309 (aarch64_gen_ccmp_first): Likewise.
12310 (aarch64_gen_ccmp_next): Likewise.
12311 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12312 (alpha_emit_xfloating_libcall): Likewise.
12313 (emit_insxl): Likewise.
12314 (alpha_arg_type): Likewise.
12315 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12316 (arc_preferred_simd_mode): Likewise.
12317 (arc_secondary_reload): Likewise.
12318 (get_arc_condition_code): Likewise.
12319 (arc_print_operand): Likewise.
12320 (arc_legitimate_constant_p): Likewise.
12321 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12322 * config/arc/arc.md (casesi_load): Likewise.
12323 (casesi_compact_jump): Likewise.
12324 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12325 (cc_use_register): Likewise.
12326 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12327 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12328 (arm_init_iwmmxt_builtins): Likewise.
12329 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12330 (neon_expand_vector_init): Likewise.
12331 (arm_attr_length_move_neon): Likewise.
12332 (maybe_get_arm_condition_code): Likewise.
12333 (arm_emit_vector_const): Likewise.
12334 (arm_preferred_simd_mode): Likewise.
12335 (arm_output_iwmmxt_tinsr): Likewise.
12336 (thumb1_output_casesi): Likewise.
12337 (thumb2_output_casesi): Likewise.
12338 (arm_emit_load_exclusive): Likewise.
12339 (arm_emit_store_exclusive): Likewise.
12340 (arm_expand_compare_and_swap): Likewise.
12341 (arm_evpc_neon_vuzp): Likewise.
12342 (arm_evpc_neon_vzip): Likewise.
12343 (arm_evpc_neon_vrev): Likewise.
12344 (arm_evpc_neon_vtrn): Likewise.
12345 (arm_evpc_neon_vext): Likewise.
12346 (arm_validize_comparison): Likewise.
12347 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12348 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12349 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12350 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12351 (c6x_preferred_simd_mode): Likewise.
12352 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12353 (epiphany_rtx_costs): Likewise.
12354 * config/epiphany/predicates.md (proper_comparison_operator):
12356 * config/frv/frv.c (condexec_memory_operand): Likewise.
12357 (frv_emit_move): Likewise.
12358 (output_move_single): Likewise.
12359 (output_condmove_single): Likewise.
12360 (frv_hard_regno_mode_ok): Likewise.
12361 (frv_matching_accg_mode): Likewise.
12362 * config/h8300/h8300.c (split_adds_subs): Likewise.
12363 (h8300_rtx_costs): Likewise.
12364 (h8300_print_operand): Likewise.
12365 (compute_mov_length): Likewise.
12366 (output_logical_op): Likewise.
12367 (compute_logical_op_length): Likewise.
12368 (compute_logical_op_cc): Likewise.
12369 (h8300_shift_needs_scratch_p): Likewise.
12370 (output_a_shift): Likewise.
12371 (compute_a_shift_length): Likewise.
12372 (compute_a_shift_cc): Likewise.
12373 (expand_a_rotate): Likewise.
12374 (output_a_rotate): Likewise.
12375 * config/i386/i386.c (classify_argument): Likewise.
12376 (function_arg_advance_32): Likewise.
12377 (function_arg_32): Likewise.
12378 (function_arg_64): Likewise.
12379 (function_value_64): Likewise.
12380 (ix86_gimplify_va_arg): Likewise.
12381 (ix86_legitimate_constant_p): Likewise.
12382 (put_condition_code): Likewise.
12383 (split_double_mode): Likewise.
12384 (ix86_avx256_split_vector_move_misalign): Likewise.
12385 (ix86_expand_vector_logical_operator): Likewise.
12386 (ix86_split_idivmod): Likewise.
12387 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12388 (ix86_build_const_vector): Likewise.
12389 (ix86_build_signbit_mask): Likewise.
12390 (ix86_match_ccmode): Likewise.
12391 (ix86_cc_modes_compatible): Likewise.
12392 (ix86_expand_branch): Likewise.
12393 (ix86_expand_sse_cmp): Likewise.
12394 (ix86_expand_sse_movcc): Likewise.
12395 (ix86_expand_int_sse_cmp): Likewise.
12396 (ix86_expand_vec_perm_vpermi2): Likewise.
12397 (ix86_expand_vec_perm): Likewise.
12398 (ix86_expand_sse_unpack): Likewise.
12399 (ix86_expand_int_addcc): Likewise.
12400 (ix86_split_to_parts): Likewise.
12401 (ix86_vectorize_builtin_gather): Likewise.
12402 (ix86_vectorize_builtin_scatter): Likewise.
12403 (avx_vpermilp_parallel): Likewise.
12404 (inline_memory_move_cost): Likewise.
12405 (ix86_tieable_integer_mode_p): Likewise.
12406 (x86_maybe_negate_const_int): Likewise.
12407 (ix86_expand_vector_init_duplicate): Likewise.
12408 (ix86_expand_vector_init_one_nonzero): Likewise.
12409 (ix86_expand_vector_init_one_var): Likewise.
12410 (ix86_expand_vector_init_concat): Likewise.
12411 (ix86_expand_vector_init_interleave): Likewise.
12412 (ix86_expand_vector_init_general): Likewise.
12413 (ix86_expand_vector_set): Likewise.
12414 (ix86_expand_vector_extract): Likewise.
12415 (emit_reduc_half): Likewise.
12416 (ix86_emit_i387_round): Likewise.
12417 (ix86_mangle_type): Likewise.
12418 (ix86_expand_round_sse4): Likewise.
12419 (expand_vec_perm_blend): Likewise.
12420 (canonicalize_vector_int_perm): Likewise.
12421 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12422 (expand_vec_perm_1): Likewise.
12423 (expand_vec_perm_interleave3): Likewise.
12424 (expand_vec_perm_even_odd_pack): Likewise.
12425 (expand_vec_perm_even_odd_1): Likewise.
12426 (expand_vec_perm_broadcast_1): Likewise.
12427 (ix86_vectorize_vec_perm_const_ok): Likewise.
12428 (ix86_expand_vecop_qihi): Likewise.
12429 (ix86_expand_mul_widen_hilo): Likewise.
12430 (ix86_expand_sse2_abs): Likewise.
12431 (ix86_expand_pextr): Likewise.
12432 (ix86_expand_pinsr): Likewise.
12433 (ix86_preferred_simd_mode): Likewise.
12434 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12435 * config/i386/sse.md (*andnot<mode>3): Likewise.
12436 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12437 (*<code><mode>3): Likewise.
12438 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12439 (ia64_expand_atomic_op): Likewise.
12440 (ia64_arg_type): Likewise.
12441 (ia64_mode_to_int): Likewise.
12442 (ia64_scalar_mode_supported_p): Likewise.
12443 (ia64_vector_mode_supported_p): Likewise.
12444 (expand_vec_perm_broadcast): Likewise.
12445 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12446 (iq2000_function_arg_advance): Likewise.
12447 (iq2000_function_arg): Likewise.
12448 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12449 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12450 (m68k_libcall_value): Likewise.
12451 (m68k_function_value): Likewise.
12452 (sched_attr_op_type): Likewise.
12453 * config/mcore/mcore.c (mcore_output_move): Likewise.
12454 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12456 (microblaze_function_arg): Likewise.
12457 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12458 (mips_print_operand): Likewise.
12459 (mips_mode_ok_for_mov_fmt_p): Likewise.
12460 (mips_vector_mode_supported_p): Likewise.
12461 (mips_preferred_simd_mode): Likewise.
12462 (mips_expand_vpc_loongson_even_odd): Likewise.
12463 (mips_expand_vec_unpack): Likewise.
12464 (mips_expand_vi_broadcast): Likewise.
12465 (mips_expand_vector_init): Likewise.
12466 (mips_expand_vec_reduc): Likewise.
12467 (mips_expand_msa_cmp): Likewise.
12468 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12469 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12470 (cc_flags_for_mode): Likewise.
12471 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12472 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12473 (nds32_output_casesi_pc_relative): Likewise.
12474 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12475 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12476 (nvptx_gen_unpack): Likewise.
12477 (nvptx_gen_pack): Likewise.
12478 (nvptx_gen_shuffle): Likewise.
12479 (nvptx_gen_wcast): Likewise.
12480 (nvptx_preferred_simd_mode): Likewise.
12481 * config/pa/pa.c (pa_secondary_reload): Likewise.
12482 * config/pa/predicates.md (base14_operand): Likewise.
12483 * config/powerpcspe/powerpcspe-c.c
12484 (altivec_resolve_overloaded_builtin): Likewise.
12485 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12487 (rs6000_preferred_simd_mode): Likewise.
12488 (output_vec_const_move): Likewise.
12489 (rs6000_expand_vector_extract): Likewise.
12490 (rs6000_split_vec_extract_var): Likewise.
12491 (reg_offset_addressing_ok_p): Likewise.
12492 (rs6000_legitimate_offset_address_p): Likewise.
12493 (rs6000_legitimize_address): Likewise.
12494 (rs6000_emit_set_const): Likewise.
12495 (rs6000_const_vec): Likewise.
12496 (rs6000_emit_move): Likewise.
12497 (spe_build_register_parallel): Likewise.
12498 (rs6000_darwin64_record_arg_recurse): Likewise.
12499 (swap_selector_for_mode): Likewise.
12500 (spe_init_builtins): Likewise.
12501 (paired_init_builtins): Likewise.
12502 (altivec_init_builtins): Likewise.
12503 (do_load_for_compare): Likewise.
12504 (rs6000_generate_compare): Likewise.
12505 (rs6000_expand_float128_convert): Likewise.
12506 (emit_load_locked): Likewise.
12507 (emit_store_conditional): Likewise.
12508 (rs6000_output_function_epilogue): Likewise.
12509 (rs6000_handle_altivec_attribute): Likewise.
12510 (rs6000_function_value): Likewise.
12511 (emit_fusion_gpr_load): Likewise.
12512 (emit_fusion_p9_load): Likewise.
12513 (emit_fusion_p9_store): Likewise.
12514 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12515 (fusion_gpr_mem_load): Likewise.
12516 (fusion_addis_mem_combo_load): Likewise.
12517 (fusion_addis_mem_combo_store): Likewise.
12518 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12519 (fusion_gpr_mem_load): Likewise.
12520 (fusion_addis_mem_combo_load): Likewise.
12521 (fusion_addis_mem_combo_store): Likewise.
12522 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12524 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12525 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12526 (rs6000_preferred_simd_mode): Likewise.
12527 (output_vec_const_move): Likewise.
12528 (rs6000_expand_vector_extract): Likewise.
12529 (rs6000_split_vec_extract_var): Likewise.
12530 (reg_offset_addressing_ok_p): Likewise.
12531 (rs6000_legitimate_offset_address_p): Likewise.
12532 (rs6000_legitimize_address): Likewise.
12533 (rs6000_emit_set_const): Likewise.
12534 (rs6000_const_vec): Likewise.
12535 (rs6000_emit_move): Likewise.
12536 (rs6000_darwin64_record_arg_recurse): Likewise.
12537 (swap_selector_for_mode): Likewise.
12538 (paired_init_builtins): Likewise.
12539 (altivec_init_builtins): Likewise.
12540 (rs6000_expand_float128_convert): Likewise.
12541 (emit_load_locked): Likewise.
12542 (emit_store_conditional): Likewise.
12543 (rs6000_output_function_epilogue): Likewise.
12544 (rs6000_handle_altivec_attribute): Likewise.
12545 (rs6000_function_value): Likewise.
12546 (emit_fusion_gpr_load): Likewise.
12547 (emit_fusion_p9_load): Likewise.
12548 (emit_fusion_p9_store): Likewise.
12549 * config/rx/rx.c (rx_gen_move_template): Likewise.
12550 (flags_from_mode): Likewise.
12551 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12552 (s390_slb_comparison): Likewise.
12553 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12554 (s390_vector_mode_supported_p): Likewise.
12555 (s390_cc_modes_compatible): Likewise.
12556 (s390_match_ccmode_set): Likewise.
12557 (s390_canonicalize_comparison): Likewise.
12558 (s390_emit_compare_and_swap): Likewise.
12559 (s390_branch_condition_mask): Likewise.
12560 (s390_rtx_costs): Likewise.
12561 (s390_secondary_reload): Likewise.
12562 (__SECONDARY_RELOAD_CASE): Likewise.
12563 (s390_expand_cs): Likewise.
12564 (s390_preferred_simd_mode): Likewise.
12565 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12566 * config/sh/sh.c (sh_print_operand): Likewise.
12567 (dump_table): Likewise.
12568 (sh_secondary_reload): Likewise.
12569 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12570 * config/sh/sh.md (casesi_worker_1): Likewise.
12571 (casesi_worker_2): Likewise.
12572 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12573 (fcc_comparison_operator): Likewise.
12574 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12575 (emit_soft_tfmode_cvt): Likewise.
12576 (sparc_preferred_simd_mode): Likewise.
12577 (output_cbranch): Likewise.
12578 (sparc_print_operand): Likewise.
12579 (sparc_expand_vec_perm_bmask): Likewise.
12580 (vector_init_bshuffle): Likewise.
12581 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12582 (spu_vector_mode_supported_p): Likewise.
12583 (spu_expand_insv): Likewise.
12584 (spu_emit_branch_or_set): Likewise.
12585 (spu_handle_vector_attribute): Likewise.
12586 (spu_builtin_splats): Likewise.
12587 (spu_builtin_extract): Likewise.
12588 (spu_builtin_promote): Likewise.
12589 (spu_expand_sign_extend): Likewise.
12590 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12591 (tilegx_simd_int): Likewise.
12592 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12593 (tilepro_simd_int): Likewise.
12594 * config/v850/v850.c (const_double_split): Likewise.
12595 (v850_print_operand): Likewise.
12596 (ep_memory_offset): Likewise.
12597 * config/vax/vax.c (vax_rtx_costs): Likewise.
12598 (vax_output_int_move): Likewise.
12599 (vax_output_int_add): Likewise.
12600 (vax_output_int_subtract): Likewise.
12601 * config/visium/predicates.md (visium_branch_operator): Likewise.
12602 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12603 (visium_print_operand_address): Likewise.
12604 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12605 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12606 (xtensa_expand_conditional_branch): Likewise.
12607 (xtensa_copy_incoming_a7): Likewise.
12608 (xtensa_output_literal): Likewise.
12609 * dfp.c (decimal_real_maxval): Likewise.
12610 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12613 Alan Hayward <alan.hayward@arm.com>
12614 David Sherwood <david.sherwood@arm.com>
12616 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12617 (mode_nunits_inline): Likewise.
12618 (mode_inner_inline): Likewise.
12619 (mode_unit_size_inline): Likewise.
12620 (mode_unit_precision_inline): Likewise.
12621 (emit_insn_modes_h): Likewise. Also emit a #define of the
12623 (emit_mode_wider): Add an E_ prefix to mode names.
12624 (emit_mode_complex): Likewise.
12625 (emit_mode_inner): Likewise.
12626 (emit_mode_adjustments): Likewise.
12627 (emit_mode_int_n): Likewise.
12628 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12629 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12630 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12631 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12632 (CRC32_BUILTIN, ENTRY): Likewise.
12633 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12634 (aarch64_pop_regs): Likewise.
12635 (aarch64_process_components): Likewise.
12636 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12637 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12638 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12639 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12640 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12641 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12643 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12644 (pcmpistr): Likewise.
12645 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12646 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12647 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12649 * config/rl78/rl78.c (mduc_regs): Likewise.
12650 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12651 (htm_expand_builtin): Likewise.
12652 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12653 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12654 * config/xtensa/xtensa.c (print_operand): Likewise.
12655 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12656 (NUM_MODE_VECTOR_INT): Likewise.
12657 * genoutput.c (null_operand): Likewise.
12658 (output_operand_data): Likewise.
12659 * genrecog.c (print_parameter_value): Likewise.
12660 * lra.c (debug_operand_data): Likewise.
12662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12664 * dwarf2out.c (scompare_loc_descriptor_wide)
12665 (scompare_loc_descriptor_narrow): New functions, split out from...
12666 (scompare_loc_descriptor): ...here.
12667 * expmed.c (emit_store_flag_int): New function, split out from...
12668 (emit_store_flag): ...here.
12670 2017-08-30 Richard Biener <rguenther@suse.de>
12672 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12673 (dwarf2out_early_finish): Move setting of AT_pubnames from
12674 early debug output to early finish.
12676 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12678 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12679 and -mdata-region to the assembler.
12681 2017-08-30 Richard Biener <rguenther@suse.de>
12683 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12685 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12686 (dwarf2out_early_global_decl): Only generate a DIE for the
12687 abstract origin if it doesn't already exist or is a declaration DIE.
12688 (resolve_addr): Do not add the linkage name twice when
12689 generating a stub DIE for the DW_TAG_GNU_call_site target.
12691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12693 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12694 Use machine_mode rather than int for arg1_mode.
12696 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12699 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12700 that the second argument of the built-in functions to unpack
12701 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12702 switch statement instead a lot of if statements.
12703 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12704 Allow 64-bit values to be in Altivec registers as well as
12705 traditional floating point registers.
12706 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12708 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12710 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12711 MAX_REGS_PER_ADDRESS == 1.
12713 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12715 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12716 * config/i386/i386.c (ix86_option_override_internal): Simplify
12717 setting of opts->x_flag_entry.
12719 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12720 Jakub Jelinek <jakub@redhat.com>
12721 Richard Biener <rguenther@suse.de>
12723 PR tree-optimization/81503
12724 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12725 folded constant fits in the target type; reorder tests for clarity.
12727 2017-08-29 Martin Liska <mliska@suse.cz>
12729 * passes.def: Include pass_lower_switch.
12730 * stmt.c (dump_case_nodes): Remove and move to
12731 tree-switch-conversion.
12732 (case_values_threshold): Likewise.
12733 (expand_switch_as_decision_tree_p): Likewise.
12734 (emit_case_decision_tree): Likewise.
12735 (expand_case): Likewise.
12736 (balance_case_nodes): Likewise.
12737 (node_has_low_bound): Likewise.
12738 (node_has_high_bound): Likewise.
12739 (node_is_bounded): Likewise.
12740 (emit_case_nodes): Likewise.
12741 (struct simple_case_node): New struct.
12742 (add_case_node): Remove.
12743 (emit_case_dispatch_table): Use vector instead of case_list.
12744 (reset_out_edges_aux): Remove.
12745 (compute_cases_per_edge): Likewise.
12746 (expand_case): Build list of simple_case_node.
12747 (expand_sjlj_dispatch_table): Use it.
12748 * tree-switch-conversion.c (struct case_node): Moved from
12749 stmt.c and adjusted.
12750 (emit_case_nodes): Likewise.
12751 (node_has_low_bound): Likewise.
12752 (node_has_high_bound): Likewise.
12753 (node_is_bounded): Likewise.
12754 (case_values_threshold): Likewise.
12755 (reset_out_edges_aux): Likewise.
12756 (compute_cases_per_edge): Likewise.
12757 (add_case_node): Likewise.
12758 (dump_case_nodes): Likewise.
12759 (balance_case_nodes): Likewise.
12760 (expand_switch_as_decision_tree_p): Likewise.
12761 (emit_jump): Likewise.
12762 (emit_case_decision_tree): Likewise.
12763 (try_switch_expansion): Likewise.
12764 (do_jump_if_equal): Likewise.
12765 (emit_cmp_and_jump_insns): Likewise.
12766 (fix_phi_operands_for_edge): New function.
12767 (record_phi_operand_mapping): Likewise.
12768 (class pass_lower_switch): New pass.
12769 (pass_lower_switch::execute): New function.
12770 (make_pass_lower_switch): Likewise.
12771 (conditional_probability):
12772 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12773 * tree-pass.h: Add make_pass_lower_switch.
12775 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12778 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12781 2017-08-29 Richard Biener <rguenther@suse.de>
12783 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12784 we do not add a DW_AT_inline attribute twice.
12785 (gen_subprogram_die): Remove code setting DW_AT_inline on
12786 DECL_ABSTRACT_P nodes.
12788 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12790 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12791 calls to internal functions.
12792 (gimplify_modify_expr): Likewise.
12793 * tree-call-cdce.c (use_internal_fn): Likewise.
12794 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12795 (convert_to_divmod): Set the nothrow flag.
12796 * tree-if-conv.c (predicate_mem_writes): Likewise.
12797 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12798 (vectorizable_call): Likewise.
12799 (vectorizable_store): Likewise.
12800 (vectorizable_load): Likewise.
12801 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12802 (vect_recog_mask_conversion_pattern): Likewise.
12804 2017-08-29 Martin Liska <mliska@suse.cz>
12807 * gcc.c (driver_handle_option): Add new argument.
12808 * opts-common.c (handle_option): Pass
12809 target_option_override_hook.
12810 * opts-global.c (lang_handle_option): Add new option.
12811 (set_default_handlers): Add new argument.
12812 (decode_options): Likewise.
12813 * opts.c (target_handle_option): Likewise.
12814 (common_handle_option): Call target_option_override_hook.
12815 * opts.h (struct cl_option_handler_func): Add hook for
12816 target option override.
12817 (struct cl_option_handlers): Likewise.
12818 (set_default_handlers): Add new argument.
12819 (decode_options): Likewise.
12820 (common_handle_option): Likewise.
12821 (target_handle_option): Likewise.
12822 * toplev.c (toplev::main): Pass targetm.target_option.override
12825 2017-08-29 Richard Biener <rguenther@suse.de>
12826 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12828 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12829 life to the active subtree.
12831 2017-08-28 Jeff Law <law@redhat.com>
12833 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12834 derive_equivalences.
12835 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12836 Code moved into....
12837 (edge_info::derive_equivalences): New private member function
12839 * tree-ssa-dom.c (class edge_info): Changed from a struct
12840 to a class. Add ctor/dtor, methods and data members.
12841 (edge_info::edge_info): Renamed from allocate_edge_info.
12842 Initialize additional members.
12843 (edge_info::~edge_info): New.
12844 (free_dom_edge_info): Delete the edge info.
12845 (record_edge_info): Use new class & associated member functions.
12846 Tighten forms for testing for edge equivalences.
12847 (record_temporary_equivalences): Iterate over the simple
12848 equivalences rather than assuming there's only one per edge.
12849 (cprop_into_successor_phis): Iterate over the simple
12850 equivalences rather than assuming there's only one per edge.
12851 (optimize_stmt): Use operand_equal_p rather than pointer
12852 equality for mini-DSE code.
12854 2017-08-28 Nathan Sidwell <nathan@acm.org>
12856 * gcc.c (execute): Fold SIGPIPE handling into switch
12857 statement. Adjust internal error message.
12859 2017-08-28 Richard Biener <rguenther@suse.de>
12862 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12863 Do nothing for removed DIEs.
12865 2017-08-28 Richard Biener <rguenther@suse.de>
12867 PR tree-optimization/81977
12868 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12871 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12874 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12875 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12876 using targetm.gen_mem_thread_fence.
12878 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12881 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12882 predicate to register_operand. Reorder operands.
12883 (*btr<mode>): Ditto.
12884 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12885 (*btr<mode>_mask): Ditto.
12887 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12889 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12890 * config/rs6000/xmmintrin.h: New file.
12891 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12893 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12896 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12897 parameter and_insn and return it.
12898 (recombine_lvx_pattern): Insert a copy to ensure availability of
12899 the base register of the copied masking operation at the point of
12900 the instruction replacement.
12901 (recombine_stvx_pattern): Likewise.
12903 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12905 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12906 undocumented switches.
12907 (-mpower9-dform-vector): Likewise.
12908 (-mpower9-dform): Likewise.
12909 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12910 comments to delete references to -mpower9-dform* switches.
12911 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12912 Delete reference to -mpower9-dform* switches, test for
12913 -mpower9-vector instead.
12914 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12915 (OTHER_P9_VECTOR_MASKS): Likewise.
12916 (POWERPC_MASKS): Likewise.
12917 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12918 tests against -mpower9-dform* to -mpower9-vector. Delete code
12919 that checked for -mpower9-dform* consistancy with other options.
12920 Add test for -mpower9-misc to enable other power9 switches.
12921 (rs6000_init_hard_regno_mode_ok): Likewise.
12922 (rs6000_option_override_internal): Likewise.
12923 (rs6000_emit_prologue): Likewise.
12924 (rs6000_emit_epilogue): Likewise.
12925 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12926 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12927 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12929 (emit_fusion_p9_store): Likewise.
12930 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12931 resetting these macros if the assembler does not support ISA 3.0
12933 (TARGET_P9_DFORM_VECTOR): Likewise.
12934 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12935 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12937 2017-08-25 Alan Modra <amodra@gmail.com>
12939 PR rtl-optimization/81747
12940 * cse.c (cse_extended_basic_block): Don't attempt to record
12941 equivalences for degenerate conditional branches that branch
12942 to their fall-through.
12944 2017-08-24 Martin Sebor <msebor@redhat.com>
12946 PR middle-end/81908
12947 * gimple-fold.c (size_must_be_zero_p): New function.
12948 (gimple_fold_builtin_memory_op): Call it.
12950 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12952 * config/rs6000/mm_malloc.h: New file.
12954 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12956 PR tree-optimization/81913
12957 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12958 analysis when either IVs in condition can wrap.
12960 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12962 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12963 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12965 2017-08-24 Richard Biener <rguenther@suse.de>
12968 * targhooks.c (default_target_can_inline_p): Properly
12969 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12970 is present and always compare.
12971 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12972 infer -mfpmath=sse from TARGET_SSE_P.
12973 (ix86_can_inline_p): Properly use target_option_default_node when
12974 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12976 2017-08-24 Richard Biener <rguenther@suse.de>
12979 * dwarf2out.c (output_die): Handle flag_generate_offload like
12981 (output_comp_unit): Likewise.
12982 (gen_array_type_die): Likewise.
12983 (dwarf2out_early_finish): Likewise.
12984 (note_variable_value_in_expr): Likewise.
12985 (dwarf2out_finish): Likewise. Adjust assert.
12986 * cgraphunit.c (symbol_table::compile): Move setting of
12987 flag_generate_offload earlier ...
12988 (symbol_table::finalize_compilation_unit): ... here, before
12989 early debug finalization.
12991 2017-08-24 Richard Biener <rguenther@suse.de>
12993 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12994 and ipa-fnsummary.h.
12995 (ix86_can_inline_p): When ix86_fpmath flags do not match
12996 check whether the callee uses FP math at all.
12998 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13000 PR middle-end/81931
13001 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13002 instead of TYPE_PRECISION.
13004 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13005 Alan Hayward <alan.hayward@arm.com>
13006 David Sherwood <david.sherwood@arm.com>
13008 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13010 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13012 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13014 (ix86_frame::stack_realign_allocate): New field.
13015 (struct machine_frame_state): Modify comments.
13016 (machine_frame_state::sp_realigned_fp_end): New field.
13017 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13018 layout calculation.
13019 (sp_valid_at): Add assertion to assure no attempt to access invalid
13020 offset of a realigned stack.
13021 (fp_valid_at): Likewise.
13022 (choose_baseaddr): Modify comments.
13023 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13024 ix86_expand_prologue.
13025 (ix86_expand_prologue): Modify stack realignment and allocation.
13026 (ix86_expand_epilogue): Modify comments.
13027 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13028 avx2_runtime, avx512f, and avx512f_runtime.
13030 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13032 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13033 (mstackrealign): Do not init to -1.
13034 * config/i386/i386.c (ix86_option_override_internal):
13035 Check opts_set, not opts when setting default value of
13036 opts->x_ix86_force_align_arg_pointer.
13038 2017-08-23 Richard Biener <rguenther@suse.de>
13040 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13041 lang_hooks.decl_printable_name.
13042 * print-rtl-function.c (print_rtx_function): Likewise.
13043 * tree-pretty-print.c (dump_function_header): Likewise.
13045 2017-08-23 Richard Biener <rguenther@suse.de>
13048 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13049 -g0 at compile-time.
13051 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13053 PR middle-end/19706
13054 * doc/sourcebuild.texi (Other hardware attributes):
13057 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13059 PR middle-end/19706
13060 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13061 Fix single-use check.
13063 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13065 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13067 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13069 * doc/install.texi: Modify to add more details on running selected
13072 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13074 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13075 is combined with -mabi=ms.
13076 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13079 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13081 PR tree-optimization/81488
13082 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13083 and cached_basis fields.
13084 (MAX_SPREAD): New constant.
13085 (alloc_cand_and_find_basis): Initialize new fields.
13086 (clear_visited): New function.
13087 (create_phi_basis_1): Rename from create_phi_basis, set visited
13088 and cached_basis fields.
13089 (create_phi_basis): New wrapper function.
13090 (phi_add_costs_1): Rename from phi_add_costs, add spread
13091 parameter, set visited field, short-circuit when limits reached.
13092 (phi_add_costs): New wrapper function.
13093 (record_phi_increments_1): Rename from record_phi_increments, set
13095 (record_phi_increments): New wrapper function.
13096 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13097 (phi_incr_cost): New wrapper function.
13098 (all_phi_incrs_profitable_1): Rename from
13099 all_phi_incrs_profitable, set visited field.
13100 (all_phi_incrs_profitable): New wrapper function.
13102 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13103 Alan Hayward <alan.hayward@arm.com>
13104 David Sherwood <david.sherwood@arm.com>
13106 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13107 that takes the outer and inner modes.
13108 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13109 comparison as the canonical test for a paradoxical subreg.
13110 * combine.c (simplify_set): Use paradoxical_subreg_p.
13111 (make_extraction): Likewise.
13112 (force_to_mode): Likewise.
13113 (rtx_equal_for_field_assignment_p): Likewise.
13114 (gen_lowpart_for_combine): Likewise.
13115 (simplify_comparison): Likewise.
13116 * cse.c (equiv_constant): Likewise.
13117 * expmed.c (store_bit_field_1): Likewise.
13118 * final.c (alter_subreg): Likewise.
13119 * fwprop.c (propagate_rtx): Likewise.
13120 (forward_propagate_subreg): Likewise.
13121 * ira-conflicts.c (ira_build_conflicts): Likewise.
13122 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13123 * lra-constraints.c (curr_insn_transform): Likewise.
13124 (split_reg): Likewise.
13125 * lra-eliminations.c (move_plus_up): Likewise.
13126 (lra_eliminate_regs_1): Likewise.
13127 * recog.c (general_operand): Likewise.
13128 * ree.c (combine_reaching_defs): Likewise.
13129 * reload.c (push_reload): Likewise.
13130 (find_reloads): Likewise.
13131 * reload1.c (elimination_effects): Likewise.
13132 (compute_reload_subreg_offset): Likewise.
13133 (choose_reload_regs): Likewise.
13134 * rtlanal.c (subreg_lsb_1): Likewise.
13135 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13136 (simplify_subreg): Likewise.
13137 * var-tracking.c (track_loc_p): Likewise.
13138 * emit-rtl.c (byte_lowpart_offset): Likewise.
13139 (paradoxical_subreg_p): Delete out-of-line definition.
13141 2017-08-22 Jeff Law <law@redhat.com>
13143 PR tree-optimization/81741
13144 PR tree-optimization/71947
13145 * tree-ssa-dom.c: Include tree-inline.h.
13146 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13147 equivalences if one is more expensive to compute than the other.
13148 * tree-ssa-scopedtables.h (class const_or_copies): Make
13149 record_const_or_copy_raw method private.
13150 (class avail_exprs_stack): New method simplify_binary_operation.
13151 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13152 avail_exprs_stack::simplify_binary_operation as needed.
13153 (avail_exprs_stack::simplify_binary_operation): New function.
13155 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13157 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13158 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13159 (DOT_SYMBOLS): Likewise.
13160 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13161 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13162 (RS6000_ABI_NAME): Likewise.
13163 (TARGET_CMODEL): Likewise.
13164 (TOC_SECTION_ASM_OP): Likewise.
13165 (SET_CMODEL): New macro.
13166 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13168 2017-08-22 Richard Biener <rguenther@suse.de>
13170 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13171 to allow for free-lang-data replacements similar to verify_type_variant.
13173 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13174 Alan Hayward <alan.hayward@arm.com>
13175 David Sherwood <david.sherwood@arm.com>
13177 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13178 VOIDmode for the LABEL_REF.
13180 2017-08-22 Richard Biener <rguenther@suse.de>
13182 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13184 2017-08-22 Richard Biener <rguenther@suse.de>
13186 * common.opt (feliminate-dwarf2-dups): Ignore.
13187 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13188 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13189 same_die_p_wrap, compute_section_prefix,
13190 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13191 (comdat_symbol_id, comdat_symbol_number): Likewise.
13192 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13194 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13195 (output_die): Mark unreachable path unreachable.
13196 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13197 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13198 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13199 (dwarf2out_early_finish): Likewise.
13201 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13203 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13205 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13208 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13209 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13210 (avr_attribute_table) <io, io_low, address>: Initialize
13211 .decl_required with true.
13213 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13215 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13216 undocumented debugging options.
13217 (-mvsx-scalar-double): Likewise.
13218 (-mallow-df-permute): Likewise.
13219 (-mvectorize-builtins): Likewise.
13220 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13221 (rs6000_builtin_vectorized_function): Likewise.
13222 (rs6000_builtin_md_vectorized_function): Likewise.
13223 (rs6000_opt_vars): Likewise.
13225 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13228 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13229 (*btrq_imm): Rename from *btrq.
13230 (*btcq_imm): Rename from *btcq.
13231 (btsc): New code attribute.
13232 (*<btsc><mode>): New insn pattern.
13233 (*btr<mode>): Ditto.
13234 (*<btsc><mode>_mask): New insn_and_split pattern.
13235 (*btr<mode>_mask): Ditto.
13237 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13238 Alan Hayward <alan.hayward@arm.com>
13239 David Sherwood <david.sherwood@arm.com>
13241 * function.c (pad_below): Simplify padding calculation.
13243 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13244 Alan Hayward <alan.hayward@arm.com>
13245 David Sherwood <david.sherwood@arm.com>
13247 * target.def (function_prologue): Remove frame size argument.
13248 (function_epilogue): Likewise.
13249 * doc/tm.texi: Regenerate.
13250 * final.c (final_start_function): Update call to function_prologue.
13251 (final_end_function): Update call to function_epilogue.
13252 (default_function_pro_epilogue): Remove frame size argument.
13253 * output.h (default_function_pro_epilogue): Likewise.
13254 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13255 (arm_output_function_prologue): Likewise.
13256 * config/frv/frv.c (frv_function_prologue): Likewise.
13257 (frv_function_epilogue): Likewise.
13258 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13259 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13260 (ia64_output_function_epilogue): Likewise.
13261 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13262 (m32r_output_function_epilogue): Likewise.
13263 * config/microblaze/microblaze.c (microblaze_function_prologue)
13264 (microblaze_function_epilogue): Likewise.
13265 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13266 (mips_output_function_epilogue): Likewise.
13267 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13268 (mmix_target_asm_function_epilogue): Likewise.
13269 * config/msp430/msp430.c (msp430_start_function): Likewise.
13270 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13271 (nds32_asm_function_epilogue): Likewise.
13272 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13273 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13274 (pa_output_function_epilogue): Likewise.
13275 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13276 (rs6000_output_function_epilogue): Likewise.
13277 * config/rl78/rl78.c (rl78_start_function): Likewise.
13278 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13279 (rs6000_output_function_epilogue): Likewise.
13280 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13281 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13282 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13283 (sparc_asm_function_epilogue): Likewise.
13285 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13287 * tree.h (type_has_mode_precision_p): New function.
13288 * convert.c (convert_to_integer_1): Use it.
13289 * expr.c (expand_expr_real_2): Likewise.
13290 (expand_expr_real_1): Likewise.
13291 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13292 * match.pd: Likewise.
13293 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13294 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13295 * tree-tailcall.c (process_assignment): Likewise.
13296 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13297 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13298 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13299 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13300 (vectorizable_assignment): Likewise.
13301 (vectorizable_shift): Likewise.
13302 (vectorizable_operation): Likewise.
13303 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13305 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13307 * match.pd: Add pow (C, x) simplification.
13309 2017-08-21 Richard Biener <rguenther@suse.de>
13311 PR tree-optimization/81900
13312 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13313 for blocks with abnormal predecessors.
13314 (compute_antic): Do not set visited flag prematurely.
13316 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13319 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13321 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13323 * stor-layout.h (vector_type_mode): Move to...
13324 * tree.h (vector_type_mode): ...here.
13325 * stor-layout.c (vector_type_mode): Move to...
13326 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13328 2017-08-21 Richard Biener <rguenther@suse.de>
13330 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13331 register_external_die hooks.
13332 (debug_false_tree_charstarstar_uhwistar): Declare.
13333 (debug_nothing_tree_charstar_uhwi): Likewise.
13334 * debug.c (do_nothing_debug_hooks): Adjust.
13335 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13336 (debug_nothing_tree_charstar_uhwi): Likewise.
13337 * dbxout.c (dbx_debug_hooks): Adjust.
13338 (xcoff_debug_hooks): Likewise.
13339 * sdbout.c (sdb_debug_hooks): Likewise.
13340 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13341 * dwarf2out.c (macinfo_label_base): New global.
13342 (dwarf2out_register_external_die): New function for the
13343 register_external_die hook.
13344 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13345 (dwarf2_debug_hooks): Use them.
13346 (dwarf2_lineno_debug_hooks): Adjust.
13347 (struct die_struct): Add with_offset flag.
13348 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13349 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13350 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13351 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13352 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13353 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13354 defining section names for the early LTO debug variants.
13355 (reset_indirect_string): New helper.
13356 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13357 (print_dw_val): Add support for offsetted symbol references.
13358 (get_ultimate_context): Split out from is_cxx.
13359 (is_cxx): Use get_ultimate_context.
13360 (is_fortran): Add decl overload.
13361 (compute_comp_unit_symbol): Split out worker from
13362 compute_section_prefix.
13363 (compute_section_prefix): Call compute_comp_unit_symbol and
13364 set comdat_type_p here.
13365 (output_die): Skip DIE symbol output for the LTO added one.
13366 Handle DIE symbol references with offset.
13367 (output_comp_unit): Guard section name mangling properly.
13368 For LTO debug sections emit a symbol at the section beginning
13369 which we use to refer to its DIEs.
13370 (add_abstract_origin_attribute): For DIEs registered via
13371 dwarf2out_register_external_die directly refer to the early
13372 DIE rather than indirectly through the shadow one we created.
13373 Remove obsolete call to dwarf2out_abstract_function for
13374 non-function/block origins.
13375 (gen_array_type_die): When generating early LTO debug do
13376 not emit DW_AT_string_length.
13377 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13378 late when in LTO. As suggested place a gcc_unreachable for
13379 the DECL_ABSTRACT_P case.
13380 (gen_subprogram_die): Avoid another specification DIE
13381 for early built declarations/definitions for the late LTO case.
13382 (gen_variable_die): Add type references for late duplicated VLA dies
13384 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13385 we have the abstract instance already.
13386 (process_scope_var): Adjust decl DIE contexts in LTO which
13387 first puts them in limbo.
13388 (gen_decl_die): Do not generate type DIEs late apart from
13389 types for VLAs or for decls we do not yet have a DIE. Do not
13390 call dwarf2out_abstract_function late.
13391 (dwarf2out_early_global_decl): Make sure to create DIEs
13392 for abstract instances of a decl first.
13393 (dwarf2out_late_global_decl): Adjust comment.
13394 (output_macinfo_op): With multiple macro sections use
13395 macinfo_label_base to distinguish labels.
13396 (output_macinfo): Likewise. Update macinfo_label_base.
13397 Pass in the line info label.
13398 (note_variable_value_in_expr): When generating LTO resolve
13399 all variable values here by generating DIEs as needed.
13400 (init_sections_and_labels): Add early LTO debug flag parameter
13401 and generate different sections and names if set. Add generation
13402 counter for the labels so we can have multiple of them.
13403 (reset_dies): Helper to allow DIEs to be output multiple times.
13404 (dwarf2out_finish): When outputting DIEs to the fat part of an
13405 LTO object first reset DIEs.
13406 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13407 (modified_type_die): Check for decl_ultimate_origin being self
13409 (gen_type_die_with_usage): Likewise.
13410 (gen_typedef_die): Allow decl_ultimate_origin being self.
13411 (set_decl_abstract_flags): Remove.
13412 (set_block_abstract_flags): Likewise.
13413 (dwarf2out_abstract_function): Treat the early generated DIEs
13414 as the abstract copy and only add DW_AT_inline and
13415 DW_AT_artificial here and call set_decl_origin_self.
13416 If the DIE has an abstract origin don't do anything.
13417 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13418 if we have none yet (Go fails to build one, PR78628).
13419 (variably_modified_type_p): Prevent endless recursion for Ada
13420 cyclic pointer types.
13421 * lto-streamer-in.c: Include debug.h.
13422 (dref_queue): New global.
13423 (lto_read_tree_1): Stream in DIE references.
13424 (lto_input_tree): Register DIE references.
13425 (input_function): Stream DECL_DEBUG_ARGS.
13426 * lto-streamer-out.c: Include debug.h.
13427 (lto_write_tree_1): Output DIE references.
13428 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13429 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13430 (output_function): Stream DECL_DEBUG_ARGS.
13431 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13432 Stream DECL_ABSTRACT_ORIGIN.
13433 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13434 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13435 DECL_CONTEXT for file-scope decls.
13436 * lto-streamer.h (struct dref_entry): Declare.
13437 (dref_queue): Likewise.
13438 * cfgexpand.c (pass_expand::execute): Do not call the
13439 outlining_inline_function hook here.
13440 * lto-wrapper.c (debug_obj): New global.
13441 (tool_cleanup): Unlink it if required.
13442 (debug_objcopy): New function.
13443 (run_gcc): Handle early debug sections in the IL files by
13444 extracting them to separate files, partially linkin them and
13445 feeding the result back as result to the linker.
13446 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13447 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13448 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13449 sections into a separate segment.
13450 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13452 (darwin_asm_dwarf_section): Likewise.
13453 (darwin_asm_output_dwarf_offset): Likewise.
13454 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13456 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13457 Alan Hayward <alan.hayward@arm.com>
13458 David Sherwood <david.sherwood@arm.com>
13460 * read-md.h (md_reader::record_potential_iterator_use): Replace
13461 pointer argument with an rtx and an index.
13462 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13463 (apply_mode_iterator): Likewise.
13464 (apply_code_iterator): Likewise.
13465 (apply_int_iterator): Likewise.
13466 (apply_subst_iterator): Likewise.
13467 (record_iterator_use): Likewise.
13468 (record_attribute_use): Likewise.
13469 (md_reader::record_potential_iterator_use): Likewise. Update calls
13470 to record_iterator_use and apply_iterator.
13471 (iterator_use): Replace ptr with x and index.
13472 (attribute_use): Likewise.
13473 (apply_attribute_uses): Update calls to apply_iterator.
13474 (apply_iterators): Likewise. Update initialization of iterator_use.
13475 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13476 and record_potential_iterator_use.
13477 (rtx_reader::read_rtx_operand): Likewise.
13479 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13480 Alan Hayward <alan.hayward@arm.com>
13481 David Sherwood <david.sherwood@arm.com>
13483 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13486 2017-08-21 Richard Biener <rguenther@suse.de>
13488 PR middle-end/81884
13489 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13490 at struct end conservatively when comparing common bases.
13492 2017-08-21 Richard Biener <rguenther@suse.de>
13494 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13495 (mem_ref_in_stmt): Remove.
13496 (determine_max_movement): Use ref index to get at the reference.
13497 (invariantness_dom_walker::before_dom_children): Deal with
13498 lim data already initialized.
13499 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13501 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13503 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13504 (TARGET_ISA_ROUND): Ditto.
13505 (TARGET_ROUND): Ditto.
13506 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13507 * config/i386/i386.md: Ditto.
13508 * config/i386/sse.md: Ditto.
13509 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13510 with OPTION_MASK_ISA_SSE4_1.
13512 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13515 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13516 __builtin_ia32_lzcnt_u16.
13518 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13521 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13522 (rs6000_set_current_function): Rewrite function to use it.
13524 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13527 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13528 and TYPE_WARN_IF_NOT_ALIGN.
13529 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13530 (handle_warn_if_not_align): New.
13531 (place_union_field): Call handle_warn_if_not_align.
13532 (place_field): Call handle_warn_if_not_align.
13533 Copy TYPE_WARN_IF_NOT_ALIGN.
13534 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13535 (layout_type): Likewise.
13536 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13538 (tree_decl_common): Add warn_if_not_align.
13539 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13540 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13541 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13542 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13543 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13544 * doc/extend.texi: Document warn_if_not_aligned attribute.
13545 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13547 2017-08-17 Martin Liska <mliska@suse.cz>
13550 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13551 (get_data_dependence): Use it as pointer type.
13552 (distribute_loop): Likewise.
13554 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13556 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13557 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13558 (p8_vmrgow_<mode>_direct): New define_insn.
13559 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13560 handle endianness for vmrgew and vmrgow permute patterns.
13562 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13564 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13565 * config/rs6000/rs6000-cpus.def: Remove comment.
13566 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13567 (POWERPC_MASKS): Likewise.
13568 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13569 use of TARGET_VSX_TIMODE.
13570 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13571 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13572 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13573 (rs6000_option_override_internal): Remove dead code.
13574 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13575 (rs6000_legitimize_reload_address): Likewise.
13576 (rs6000_legitimate_address_p): Likewise.
13577 (rs6000_opt_masks): Delete "vsx-timode".
13578 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13579 from function comment.
13580 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13581 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13582 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13584 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13585 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13586 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13587 (VSX_TI): Likewise.
13589 (define_peephole2): Likewise.
13591 2017-08-17 Martin Sebor <msebor@redhat.com>
13594 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13595 past the end of an array.
13596 (test_pp_format): Add test cases.
13598 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13600 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13601 missing ECF_NOTHROW flags.
13603 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13606 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13607 operands residing in integer registers.
13608 (*vsx_le_perm_load_<mode>): Likewise.
13609 (*vsx_le_perm_store_<mode>): Likewise.
13610 (define_peephole2): Add peepholes to optimize the above.
13612 2017-08-17 Marek Polacek <polacek@redhat.com>
13614 PR middle-end/81814
13615 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13616 to mimic what shorten_compare did. Change the return type to bool.
13617 (fold_cond_expr_with_comparison): Update call to
13618 operand_equal_for_comparison_p.
13619 (fold_ternary_loc): Likewise.
13621 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13623 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13625 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13627 2017-08-17 Richard Biener <rguenther@suse.de>
13629 * tree-ssa-structalias.c (solve_graph): When propagating
13630 to successors update the graphs succ edges and avoid duplicate work.
13632 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13635 * config/i386/i386.c (ix86_option_override_internal): Save target
13636 specific options after ix86_stack_protector_guard_reg was changed.
13638 2017-08-17 Richard Biener <rguenther@suse.de>
13640 PR tree-optimization/81827
13641 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13642 (new_var_info): Initialize it conservatively.
13643 (get_call_vi): Mark register vars.
13644 (new_scalar_tmp_constraint_exp): Likewise.
13645 (handle_rhs_call): Likewise.
13646 (handle_const_call): Likewise.
13647 (create_function_info_for): Likewise.
13648 (solve_constraints): Sort varinfos to separate register from
13649 non-register vars to pack points-to solution bitmaps during
13652 2017-08-17 Marek Polacek <polacek@redhat.com>
13654 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13656 2017-08-17 Richard Biener <rguenther@suse.de>
13658 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13659 to true when overflow is undefined and we saturated the result.
13661 2017-08-17 Alan Modra <amodra@gmail.com>
13664 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13665 Don't use store multiple if only one reg needs saving.
13666 (interesting_frame_related_regno): New function.
13667 (rs6000_frame_related): Don't emit frame info for regs that
13669 (rs6000_emit_epilogue): Likewise.
13671 2017-08-16 Nathan Sidwell <nathan@acm.org>
13673 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13674 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13675 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13676 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13677 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13678 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13679 (verify_type): Adjust for TYPE_BINFO move.
13680 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13681 process TYPE_BINFO directly.
13682 (hash_tree): Likewise.
13683 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13685 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13688 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13690 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13692 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13695 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13696 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13697 Add ix86_binary_operator_ok to insn constraint.
13698 (*iordi_1_bts): Ditto.
13699 (*xordi_1_btc): Ditto.
13700 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13701 Update corresponding peephole2 pattern.
13705 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13707 PR tree-optimization/81832
13708 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13709 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13711 2017-08-16 Marek Polacek <polacek@redhat.com>
13714 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13715 regarding size_zero_node.
13717 2017-08-16 Martin Liska <mliska@suse.cz>
13720 * config.gcc: Respect previously set extra_objs in case
13723 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13725 PR tree-optimization/81835
13726 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13727 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13728 not depend on the phi.
13730 2017-08-16 Alan Modra <amodra@gmail.com>
13732 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13735 2017-08-16 Alan Modra <amodra@gmail.com>
13737 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13738 (save_reg_p): ..into this. Update all callers.
13739 (first_reg_to_save): Simplify.
13741 2017-08-16 Alan Modra <amodra@gmail.com>
13743 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13746 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13750 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13751 constant count if that count is less than 32.
13753 2017-08-15 Nathan Sidwell <nathan@acm.org>
13755 * gcc.c (execute): Emit friendlier message if inferior is killed
13756 by an external cause.
13758 2017-08-15 Richard Biener <rguenther@suse.de>
13760 PR tree-optimization/81790
13761 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13762 CONSTRUCTORs from simplifying and VN.
13764 2017-08-14 Martin Sebor <msebor@redhat.com>
13766 * builtin-attrs.def: Add comments.
13768 2017-08-14 Martin Sebor <msebor@redhat.com>
13771 * doc/extend.texi (attribute nonstring): Document new attribute.
13773 2017-08-14 Martin Sebor <msebor@redhat.com>
13776 * tree-diagnostic.c (default_tree_printer): Handle %G.
13777 * gimple-pretty-print.h (percent_G_format): Declare new function.
13778 * gimple-pretty-print.c (percent_G_format): Define.
13779 * tree-pretty-print.c (percent_K_format): Add argument.
13781 2017-08-14 Martin Sebor <msebor@redhat.com>
13783 PR translation/79998
13784 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13785 Remove a stray space.
13787 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13790 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13791 (*iordi_1_bts): Ditto.
13792 (*xordi_1_btc): Ditto.
13794 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13797 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13798 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13800 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13801 quoted strings, and make more translator-friendly.
13802 (darwin_rs6000_override_options): Likewise.
13803 (rs6000_option_override_internal): Likewise.
13804 (rs6000_return_in_memory): Fix overlong line.
13805 (init_cmulative_args): Use quoted strings, and make more
13806 translator-friendly.
13807 (rs6000_pass_by_reference): Fix overlong line.
13808 (def_builtin): Use quoted strings.
13809 (altivec_expand_predicate_builtin): Use quoted strings, and make
13810 more translator-friendly.
13811 (htm_expand_builtin): Use quoted strings.
13812 (cpu_expand_builtin): Use quoted strings, and make more
13813 translator-friendly.
13814 (altivec_expand_builtin): Likewise.
13815 (paired_expand_predicate_builtin): Likewise.
13816 (rs6000_invalid_builtin): Likewise.
13817 (builtin_function_type): Use quoted strings.
13818 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13819 more translator-friendly.
13820 (rs6000_trampoline_init): Likewise.
13821 (rs6000_handle_altivec_attribute): Likewise.
13822 (rs6000_inner_target_options): Use quoted strings.
13823 (rs6000_disable_incompatible_switches): Likewise.
13824 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13825 strings, and make more translator-friendly.
13826 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13828 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13830 PR tree-optimization/81799
13831 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13832 cond_expr to simple gimple operand.
13834 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13836 PR middle-end/46932
13837 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13839 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13843 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13844 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13845 TARGET_GASISR_PROLOGUES.
13846 * config/avr/avr.c (avr_option_override): Same.
13847 (avr_pass_pre_proep::execute): Same.
13849 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13852 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13853 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13855 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13857 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13858 define_insn_and_split. Split to a memory load from 0 in
13859 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13860 using PTR mode iterator.
13861 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13862 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13863 (*add_tp_<mode>): Redefine as define_insn_and_split.
13864 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13865 address space. Merge with *add_tp_x32 using PTR mode iterator.
13866 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13867 Split to an add with a memory load from 0 in
13868 DEFAULT_TLS_SEG_REG address space.
13870 2017-08-12 Andrew Pinski <apinski@cavium.com>
13872 * config/aarch64/aarch64-option-extensions.def (rdma):
13873 Fix feature string to what Linux prints out in /proc/cpuinfo.
13875 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13878 * dwarf2out.c (modified_type_die): For C typedef types that have
13879 an ultimate origin, process the ultimate origin instead of the
13881 (gen_typedef_die): Assert that input DECLs have no ultimate
13883 (gen_type_die_with_usage): For typedef variants that have an
13884 ultimate origin, just call gen_decl_die on the original DECL.
13885 (process_scope_var): Avoid creating DIEs for local typedefs and
13886 concrete static variables.
13888 2017-08-12 Alan Modra <amodra@gmail.com>
13892 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13893 match gnu-user.h startfile.
13894 (ENDFILE_LINUX_SPEC): Similarly.
13896 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13899 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13901 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13903 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13904 * config/aarch64/aarch64.md (mov<mode>): Change.
13905 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13906 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13907 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13909 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13911 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13912 for storage order barriers.
13914 2017-08-11 Martin Liska <mliska@suse.cz>
13917 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13918 variables of void type.
13920 2017-08-11 Martin Liska <mliska@suse.cz>
13922 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13923 TARGET_SUPPORTS_ALIASES.
13924 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13925 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13926 (optimize_weakref): Likewise.
13927 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13928 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13929 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13931 2017-08-11 Martin Liska <mliska@suse.cz>
13934 * config/i386/i386.c (make_resolver_func): Do complete
13935 refactoring of the function.
13937 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13940 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13941 * config/i386/i386.c (ix86_stack_protect_guard): Use
13942 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13943 * doc/invoke.texi (x86 Options): Document
13944 -mstack-protector-guard-symbol= option.
13946 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13948 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13949 * config/i386/i386.c (ix86_split_stack_guard): New function.
13950 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13951 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13952 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13953 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13954 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13955 (split_stack_space_check): Call ix86_split_stack_guard.
13957 2017-08-10 Martin Sebor <msebor@redhat.com>
13959 * print-tree.c (print_node): Print location using the established
13961 Replace spaces with colons.
13962 (debug_raw, debug): Ditto.
13964 2017-08-10 Martin Sebor <msebor@redhat.com>
13967 * pretty-print.c (pp_format): Correct the handling of %s precision.
13969 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13972 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13974 (ix86_finalize_stack_frame_flags): This. Also clear
13975 frame_pointer_needed if -fno-omit-frame-pointer is used without
13977 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13978 with ix86_finalize_stack_frame_flags.
13979 (ix86_expand_epilogue): Likewise.
13980 (ix86_expand_split_stack_prologue): Likewise.
13981 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13983 2017-08-10 Martin Liska <mliska@suse.cz>
13986 * c-attribs.c (handle_target_attribute):
13987 Report warning for an empty string argument of target attribute.
13989 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13992 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13994 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13995 or DECL_NONLOCAL labels.
13996 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13997 or DECL_NONLOCAL labels here.
13999 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14001 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14002 to indicate when early gimple folding has been disabled.
14003 (rs6000_gimple_fold_builtin): Add debug content.
14004 (rs6000_invalid_builtin): Fix whitespace.
14005 (rs6000_expand_builtin): Fix whitespace.
14006 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14008 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14011 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14012 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14014 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14016 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14017 (qdf24xx): Likewise.
14018 * config/aarch64/aarch64.md: Include falkor.md.
14019 * config/aarch64/falkor.md: New.
14021 2017-08-09 Marek Polacek <polacek@redhat.com>
14024 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14025 * diagnostic.c (emit_diagnostic): Add a comment.
14026 (emit_diagnostic_valist): New function.
14028 2017-08-09 Marek Polacek <polacek@redhat.com>
14031 * input.c (make_location): New overload.
14032 * input.h (make_location): Declare.
14034 2017-08-08 Alan Modra <amodra@gmail.com>
14035 H.J. Lu <hongjiu.lu@intel.com>
14038 * gcc.c (NO_PIE_SPEC): Delete.
14039 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14041 (LINK_PIE_SPEC): ..to here.
14042 (LINK_COMMAND_SPEC): Support -no-pie.
14043 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14044 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14046 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14047 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14048 (ENDFILE_CRTEND_SPEC): Similarly.
14050 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14053 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14054 (mstack-protector-guard-offset=): Ditto.
14055 * config/i386/i386.c (ix86_option_override): Handle
14056 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14058 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14059 ix86_stack_protect_guard_offset variables.
14060 (TARGET_STACK_PROTECT_GUARD): Always define.
14061 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14062 and -mstack-protector-guard-offset= options.
14064 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14066 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14067 boundary case for the last candidate.
14069 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14071 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14072 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14074 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14076 PR middle-end/19706
14077 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14078 * config/aarch64/aarch64-builtins.c
14079 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14080 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14081 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14083 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14084 Andrew Pinski <pinskia@gmail.com>
14086 PR middle-end/19706
14087 * internal-fn.def (XORSIGN): New.
14088 * optabs.def (xorsign_optab): New.
14089 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14090 (convert_expand_mult_copysign): New.
14091 (pass_optimize_widening_mul::execute): Call
14092 convert_expand_mult_copysign.
14094 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14096 PR tree-optimization/81354
14097 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14098 Insert on edges rather than explicitly creating landing pads.
14099 (analyze_candidates_and_replace): Commit edge inserts.
14101 2017-08-08 Richard Biener <rguenther@suse.de>
14103 PR middle-end/81719
14104 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14105 (expand_simple_operations): Also look through ADDR_EXPRs with
14106 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14108 2017-08-08 Richard Biener <rguenther@suse.de>
14110 PR tree-optimization/81723
14111 * tree-vect-slp.c (struct bst_traits): New hash traits.
14112 (bst_fail): New global.
14113 (vect_build_slp_tree_2): New worker, split out from ...
14114 (vect_build_slp_tree): ... this now wrapping it with using
14115 bst_fail set to cache SLP tree build fails. Properly handle
14117 (vect_analyze_slp_instance): Allocate and free bst_fail.
14119 2017-08-08 Martin Liska <mliska@suse.cz>
14122 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14123 LABEL_DECLs that can be from a different function.
14125 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14127 PR tree-optimization/81744
14128 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14129 loop's number of iterations.
14131 2017-08-08 Martin Liska <mliska@suse.cz>
14133 * asan.c: Include header files.
14134 * attribs.c (build_decl_attribute_variant): New function moved
14136 (build_type_attribute_qual_variant): Likewise.
14137 (cmp_attrib_identifiers): Likewise.
14138 (simple_cst_list_equal): Likewise.
14139 (omp_declare_simd_clauses_equal): Likewise.
14140 (attribute_value_equal): Likewise.
14141 (comp_type_attributes): Likewise.
14142 (build_type_attribute_variant): Likewise.
14143 (lookup_ident_attribute): Likewise.
14144 (remove_attribute): Likewise.
14145 (merge_attributes): Likewise.
14146 (merge_type_attributes): Likewise.
14147 (merge_decl_attributes): Likewise.
14148 (merge_dllimport_decl_attributes): Likewise.
14149 (handle_dll_attribute): Likewise.
14150 (attribute_list_equal): Likewise.
14151 (attribute_list_contained): Likewise.
14152 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14153 (lookup_attribute_by_prefix): Likewise.
14154 * bb-reorder.c: Include header files.
14155 * builtins.c: Likewise.
14156 * calls.c: Likewise.
14157 * cfgexpand.c: Likewise.
14158 * cgraph.c: Likewise.
14159 * cgraphunit.c: Likewise.
14160 * convert.c: Likewise.
14161 * dwarf2out.c: Likewise.
14162 * final.c: Likewise.
14163 * fold-const.c: Likewise.
14164 * function.c: Likewise.
14165 * gimple-expr.c: Likewise.
14166 * gimple-fold.c: Likewise.
14167 * gimple-pretty-print.c: Likewise.
14168 * gimple.c: Likewise.
14169 * gimplify.c: Likewise.
14170 * hsa-common.c: Likewise.
14171 * hsa-gen.c: Likewise.
14172 * internal-fn.c: Likewise.
14173 * ipa-chkp.c: Likewise.
14174 * ipa-cp.c: Likewise.
14175 * ipa-devirt.c: Likewise.
14176 * ipa-fnsummary.c: Likewise.
14177 * ipa-inline.c: Likewise.
14178 * ipa-visibility.c: Likewise.
14180 * lto-cgraph.c: Likewise.
14181 * omp-expand.c: Likewise.
14182 * omp-general.c: Likewise.
14183 * omp-low.c: Likewise.
14184 * omp-offload.c: Likewise.
14185 * omp-simd-clone.c: Likewise.
14186 * opts-global.c: Likewise.
14187 * passes.c: Likewise.
14188 * predict.c: Likewise.
14189 * sancov.c: Likewise.
14190 * sanopt.c: Likewise.
14191 * symtab.c: Likewise.
14192 * toplev.c: Likewise.
14193 * trans-mem.c: Likewise.
14194 * tree-chkp.c: Likewise.
14195 * tree-eh.c: Likewise.
14196 * tree-into-ssa.c: Likewise.
14197 * tree-object-size.c: Likewise.
14198 * tree-parloops.c: Likewise.
14199 * tree-profile.c: Likewise.
14200 * tree-ssa-ccp.c: Likewise.
14201 * tree-ssa-live.c: Likewise.
14202 * tree-ssa-loop.c: Likewise.
14203 * tree-ssa-sccvn.c: Likewise.
14204 * tree-ssa-structalias.c: Likewise.
14205 * tree-ssa.c: Likewise.
14206 * tree-streamer-in.c: Likewise.
14207 * tree-vectorizer.c: Likewise.
14208 * tree-vrp.c: Likewise.
14209 * tsan.c: Likewise.
14210 * ubsan.c: Likewise.
14211 * varasm.c: Likewise.
14212 * varpool.c: Likewise.
14213 * tree.c: Remove functions moved to attribs.[ch].
14214 * tree.h: Likewise.
14215 * config/aarch64/aarch64.c: Add attrs.h header file.
14216 * config/alpha/alpha.c: Likewise.
14217 * config/arc/arc.c: Likewise.
14218 * config/arm/arm.c: Likewise.
14219 * config/avr/avr.c: Likewise.
14220 * config/bfin/bfin.c: Likewise.
14221 * config/c6x/c6x.c: Likewise.
14222 * config/cr16/cr16.c: Likewise.
14223 * config/cris/cris.c: Likewise.
14224 * config/darwin.c: Likewise.
14225 * config/epiphany/epiphany.c: Likewise.
14226 * config/fr30/fr30.c: Likewise.
14227 * config/frv/frv.c: Likewise.
14228 * config/ft32/ft32.c: Likewise.
14229 * config/h8300/h8300.c: Likewise.
14230 * config/i386/winnt.c: Likewise.
14231 * config/ia64/ia64.c: Likewise.
14232 * config/iq2000/iq2000.c: Likewise.
14233 * config/lm32/lm32.c: Likewise.
14234 * config/m32c/m32c.c: Likewise.
14235 * config/m32r/m32r.c: Likewise.
14236 * config/m68k/m68k.c: Likewise.
14237 * config/mcore/mcore.c: Likewise.
14238 * config/microblaze/microblaze.c: Likewise.
14239 * config/mips/mips.c: Likewise.
14240 * config/mmix/mmix.c: Likewise.
14241 * config/mn10300/mn10300.c: Likewise.
14242 * config/moxie/moxie.c: Likewise.
14243 * config/msp430/msp430.c: Likewise.
14244 * config/nds32/nds32-isr.c: Likewise.
14245 * config/nds32/nds32.c: Likewise.
14246 * config/nios2/nios2.c: Likewise.
14247 * config/nvptx/nvptx.c: Likewise.
14248 * config/pa/pa.c: Likewise.
14249 * config/pdp11/pdp11.c: Likewise.
14250 * config/powerpcspe/powerpcspe.c: Likewise.
14251 * config/riscv/riscv.c: Likewise.
14252 * config/rl78/rl78.c: Likewise.
14253 * config/rx/rx.c: Likewise.
14254 * config/s390/s390.c: Likewise.
14255 * config/sh/sh.c: Likewise.
14256 * config/sol2.c: Likewise.
14257 * config/sparc/sparc.c: Likewise.
14258 * config/spu/spu.c: Likewise.
14259 * config/stormy16/stormy16.c: Likewise.
14260 * config/tilegx/tilegx.c: Likewise.
14261 * config/tilepro/tilepro.c: Likewise.
14262 * config/v850/v850.c: Likewise.
14263 * config/vax/vax.c: Likewise.
14264 * config/visium/visium.c: Likewise.
14265 * config/xtensa/xtensa.c: Likewise.
14267 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14270 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14271 constraints since the -mupper-regs-* switches have been
14273 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14274 into a vector from a double word element that was extracted from
14275 another vector, and eliminate extra XXPERMDI instructions.
14276 (vsx_concat_<mode>_2): Likewise.
14277 (vsx_concat_<mode>_3): Likewise.
14278 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14279 concat to allow optimizing inserts from previous extracts.
14281 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14283 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14284 memory reference to a SSP offset in TLS address space.
14285 (ix86_print_operand) <case '@'>: Remove.
14286 (ix86_print_operand_punct_valid_p): Remove '@' code.
14287 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14288 UNSPEC_SP_TLS_TEST.
14289 (stack_tls_protect_set_<mode>): Remove.
14290 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14291 (stack_tls_protect_test_<mode>): Remove.
14292 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14294 2017-08-07 Olivier Hainque <hainque@adacore.com>
14297 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14299 2017-08-07 Douglas Rupp <rupp@adacore.com>
14301 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14302 variable was referenced as multidir in command.
14304 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14307 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14310 2017-08-07 Martin Liska <mliska@suse.cz>
14312 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14313 * config/rl78/rl78.c: Add include of attribs.h.
14314 * config/sh/sh.c: Likewise.
14315 * config/v850/v850.c: Likewise.
14317 2017-08-07 Tom de Vries <tom@codesourcery.com>
14319 PR middle-end/78266
14320 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14322 2017-08-07 Martin Liska <mliska@suse.cz>
14324 * config/mips/mips.c: Include attribs.h.
14326 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14329 * doc/invoke.texi: Document change in behvaior for -Ofast for
14332 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14334 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14336 (aarch64_pop_regs): Likewise.
14337 (aarch64_gen_load_pair): Likewise.
14338 (aarch64_save_callee_saves): Likewise.
14339 (aarch64_restore_callee_saves): Likewise.
14341 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14343 * config/i386/i386.c: Revert the last change.
14345 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14348 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14350 (ix86_finalize_stack_frame_flags): This. Also clear
14351 frame_pointer_needed if -fno-omit-frame-pointer is used without
14353 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14354 with ix86_finalize_stack_frame_flags.
14355 (ix86_expand_epilogue): Likewise.
14356 (ix86_expand_split_stack_prologue): Likewise.
14358 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14361 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14362 to P_AES for Westmere.
14364 2017-08-07 Jonathan Yong <10walls@gmail.com>
14366 * config/i386/mingw.opt (fset-stack-executable): Removed.
14367 * config/i386/cygming.opt (fset-stack-executable): Moved
14369 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14371 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14373 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14375 2017-08-07 Marek Polacek <polacek@redhat.com>
14377 PR middle-end/81737
14378 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14380 2017-08-07 Martin Liska <mliska@suse.cz>
14382 * attribs.h (canonicalize_attr_name): New function.
14383 (cmp_attribs): Move from c-format.c and adjusted.
14384 (is_attribute_p): Moved from tree.h.
14385 * tree-inline.c: Add new includes.
14386 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14387 (private_is_attribute_p): Remove.
14388 (private_lookup_attribute): Likewise.
14389 (private_lookup_attribute_by_prefix): Simplify.
14390 (remove_attribute): Use is_attribute_p.
14391 * tree.h: Remove removed declarations.
14393 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14395 PR middle-end/81698
14396 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14397 instead of computing it in the function. Formatting fix.
14398 (expand_case): Don't rely on default_edge being the first edge,
14399 clear it if removing it, pass default_edge to
14400 emit_case_dispatch_table.
14401 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14404 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14406 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14407 insn in the function, emit NOP after the insn.
14409 2017-08-06 Tom de Vries <tom@codesourcery.com>
14411 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14414 2017-08-06 Tom de Vries <tom@codesourcery.com>
14416 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14419 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14421 PR tree-optimization/57371
14422 * match.pd: New pattern.
14424 2017-08-04 Marek Polacek <polacek@redhat.com>
14426 PR middle-end/81695
14427 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14428 perform the computation in offset_int.
14430 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14432 PR tree-optimization/81136
14433 * tree-vectorizer.h: Include tree-hash-traits.h.
14434 (vec_base_alignments): New typedef.
14435 (vec_info): Add a base_alignments field.
14436 (vect_record_base_alignments): Declare.
14437 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14439 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14440 (create_data_ref): Add an is_conditional_in_stmt argument.
14441 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14442 the is_conditional_in_stmt field.
14443 (data_ref_loc): Add an is_conditional_in_stmt field.
14444 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14445 (find_data_references_in_stmt): Update call to create_data_ref.
14446 (graphite_find_data_references_in_stmt): Likewise.
14447 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14448 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14449 (vect_record_base_alignment): New function.
14450 (vect_record_base_alignments): Likewise.
14451 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14452 for nested statements even if we fail to compute a misalignment.
14453 Use pooled base alignments for unconditional references.
14454 (vect_find_same_alignment_drs): Compare base addresses instead
14456 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14457 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14459 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14461 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14462 Add an explicit name for the enum. Use auto_vec for slp_instances
14463 and grouped_stores.
14464 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14466 (_bb_vec_info): Add a constructor and destructor.
14467 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14468 (destroy_loop_vec_info): Delete.
14469 (vect_destroy_datarefs): Likewise.
14470 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14471 (vec_info::vec_info): New function.
14472 (vec_info::~vec_info): Likewise.
14473 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14474 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14475 destroy_loop_vec_info.
14476 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14477 (_loop_vec_info::_loop_vec_info): ...this.
14478 (destroy_loop_vec_info): Replace with...
14479 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14480 the stmt_vec_infos. Leave handling of vec_info information to its
14481 destructor. Remove explicit vector releases.
14482 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14483 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14484 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14485 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14486 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14487 (destroy_bb_vec_info): Replace with...
14488 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14489 information to its destructor.
14490 (vect_slp_analyze_bb_1): Use new and delete instead of
14491 new_bb_vec_info and destroy_bb_vec_info.
14492 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14495 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14497 * tree-data-ref.h (subscript): Add access_fn field.
14498 (data_dependence_relation): Add could_be_independent_p.
14499 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14500 (same_access_functions): Move to tree-data-ref.c.
14501 * tree-data-ref.c (ref_contains_union_access_p): New function.
14502 (access_fn_component_p): Likewise.
14503 (access_fn_components_comparable_p): Likewise.
14504 (dr_analyze_indices): Add a reference to access_fn_component_p.
14505 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14507 (constant_access_functions): Likewise.
14508 (add_other_self_distances): Likewise.
14509 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14510 (initialize_data_dependence_relation): Use XCNEW and remove
14511 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14512 of access functions that have the same type. Allow the
14513 subsequence to end with different bases in some circumstances.
14514 Record the chosen access functions in SUB_ACCESS_FN.
14515 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14516 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14517 (subscript_dependence_tester_1): Likewise dra and drb.
14518 (build_classic_dist_vector): Update calls accordingly.
14519 (subscript_dependence_tester): Likewise.
14520 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14521 DDR_COULD_BE_INDEPENDENT_P.
14522 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14523 comp_alias_ddrs instead of may_alias_ddrs.
14524 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14526 (vect_analyze_data_ref_dependence): Use it if
14527 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14528 distance vectors if that fails.
14529 (dependence_distance_ge_vf): New function.
14530 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14531 LOOP_VINFO_MAY_ALIAS_DDRS.
14533 2017-08-04 Richard Biener <rguenther@suse.de>
14535 PR middle-end/81705
14536 * fold-const.c (fold_binary_loc): Properly restrict
14537 minus_var0 && minus_var1 case when associating undefined overflow
14540 2017-08-04 Tom de Vries <tom@codesourcery.com>
14542 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14544 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14546 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14547 Don't start diagnostic messages with a capital letter.
14548 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14550 (rs6000_invalid_builtin): Likewise.
14551 (rs6000_trampoline_init): Likewise.
14553 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14556 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14557 after setting changeable df flags.
14559 2017-08-03 Richard Biener <rguenther@suse.de>
14561 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14562 up if the use is in USE - X.
14564 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14566 * toplev.c (dumpfile.h): New include.
14567 (internal_error_reentered): New static function. Use it...
14568 (internal_error_function): ...here to handle reentered internal_error.
14570 2017-08-03 Richard Biener <rguenther@suse.de>
14572 PR middle-end/81148
14573 * fold-const.c (split_tree): Add minus_var and minus_con
14574 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14575 here but always use minus_*.
14576 (associate_trees): Assert we never associate with MINUS_EXPR
14577 and NULL first operand. Do not recurse for PLUS_EXPR operands
14578 when associating as MINUS_EXPR either.
14579 (fold_binary_loc): Track minus_var and minus_con.
14581 2017-08-03 Tom de Vries <tom@codesourcery.com>
14584 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14585 ACCEL_COMPILER, apply finish_options on
14586 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14588 2017-08-03 Tom de Vries <tom@codesourcery.com>
14591 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14592 function_entry_patch_area_size > 0.
14594 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14597 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14598 instead of 10??LU, perform unit multiplication in wide_int,
14599 don't change alloc_object_size_limit if the limit is larger
14602 PR tree-optimization/81655
14603 PR tree-optimization/81588
14604 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14605 the case when ranges[i].low and high are 1 for unsigned type with
14608 PR middle-end/81052
14609 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14610 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14612 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14614 * tree-vrp.h: Add include guard.
14616 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14619 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14620 (ud2): New insn pattern.
14621 * config/i386/i386.c (ix86_expand_epilogue):
14622 For naked functions, generate ud2 instead of trap insn.
14624 2017-08-02 Marek Polacek <polacek@redhat.com>
14627 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14629 2017-08-02 Tom de Vries <tom@codesourcery.com>
14630 Cesar Philippidis <cesar@codesourcery.com>
14632 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14633 Add missing edge probabilities.
14635 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14637 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14638 Correct endianness.
14640 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14642 PR middle-end/79499
14643 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14644 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14645 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14646 prologue_seq sequences - if any.
14648 2017-08-02 Richard Biener <rguenther@suse.de>
14650 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14651 via vectors if supported, integer extracts via punning if supported
14652 or otherwise vector extracts.
14654 2017-08-02 Richard Biener <rguenther@suse.de>
14656 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14658 (bitmap_insert_into_set): ... this.
14660 2017-08-02 Richard Biener <rguenther@suse.de>
14662 PR tree-optimization/81633
14664 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14666 PR tree-optimization/71752
14667 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14669 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14671 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14672 (machine_function::call_ms2sysv_pad_out): Remove field.
14673 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14674 (ix86_compute_frame_layout): Likewise.
14676 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14679 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14680 attribute with interrupt attribute.
14682 2017-08-01 Andrew Pinski <apinski@cavium.com>
14684 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14685 BIT_INSERT_EXPR's operand 1
14686 to see if the types precision matches.
14688 2017-08-01 Martin Liska <mliska@suse.cz>
14690 PR middle-end/70140
14691 * builtins.c (expand_builtin_memcpy_args): Remove.
14692 (expand_builtin_memcpy): Call newly added function
14693 expand_builtin_memory_copy_args.
14694 (expand_builtin_memcpy_with_bounds): Likewise.
14695 (expand_builtin_mempcpy): Remove last argument.
14696 (expand_builtin_mempcpy_with_bounds): Likewise.
14697 (expand_builtin_memory_copy_args): New function created from
14698 expand_builtin_mempcpy_args with small modifications.
14699 (expand_builtin_mempcpy_args): Remove.
14700 (expand_builtin_stpcpy): Remove unused argument.
14701 (expand_builtin): Likewise.
14702 (expand_builtin_with_bounds): Likewise.
14704 2017-08-01 Martin Liska <mliska@suse.cz>
14707 Make mempcpy more optimal (PR middle-end/70140).
14709 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14712 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14713 __builtin_vec_cmpne verify both arguments are compatible vectors
14714 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14715 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14716 move computation of aligned to after checking the argument types.
14720 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14721 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14724 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14725 Doug Rupp <rupp@adacore.com>
14726 Olivier Hainque <hainque@adacore.com>
14728 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14729 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14731 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14732 for TARGET_OS_CPP_BUILTIN.
14733 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14734 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14736 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14737 passing required abi options to the assembler for EABI configurations.
14738 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14739 of .text.hot and .text.unlikely sections for kernel modules when
14740 using ARM style exceptions.
14741 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14742 options. Add EXTRA_CC1_SPEC.
14743 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14745 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14747 (ARM_TARGET2_DWARF_FORMAT): Define.
14748 * config/arm/t-vxworks: Adjust multilib control to removal of the
14749 Diab command line options.
14751 2017-08-01 Martin Liska <mliska@suse.cz>
14753 PR gcov-profile/81561
14754 * gcov.c (unblock): Make unblocking safe as we need to preserve
14755 index correspondence of blocks and block_lists.
14757 2017-08-01 Richard Biener <rguenther@suse.de>
14759 PR tree-optimization/81181
14760 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14761 (compute_antic): ... end of iteration here.
14763 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14765 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14766 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14767 (ftree-slp-vectorize): Likewise.
14768 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14769 can no longer be set independent of flag_tree_loop_vectorize.
14770 * omp-general.c (emp_max_vf): Likewise.
14771 * opts.c (enable_fdo_optimizations): Remove references to
14772 flag_tree_vectorize, these are now implicit.
14773 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14774 and leave it for the options machinery.
14776 2017-08-01 Martin Liska <mliska@suse.cz>
14778 PR middle-end/70140
14779 * builtins.c (expand_builtin_memcpy_args): Remove.
14780 (expand_builtin_memcpy): Call newly added function
14781 expand_builtin_memory_copy_args.
14782 (expand_builtin_memcpy_with_bounds): Likewise.
14783 (expand_builtin_mempcpy): Remove last argument.
14784 (expand_builtin_mempcpy_with_bounds): Likewise.
14785 (expand_builtin_memory_copy_args): New function created from
14786 expand_builtin_mempcpy_args with small modifications.
14787 (expand_builtin_mempcpy_args): Remove.
14788 (expand_builtin_stpcpy): Remove unused argument.
14789 (expand_builtin): Likewise.
14790 (expand_builtin_with_bounds): Likewise.
14792 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14795 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14796 print "ds:" only for immediates in generic address space.
14798 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14801 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14802 (ix86_function_ok_for_sibcall): Return false for naked functions.
14804 2017-08-01 Richard Biener <rguenther@suse.de>
14806 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14807 (compute_antic): Seed worklist with exit block predecessors.
14808 * cfganal.c (dfs_find_deadend): For a cycle return the source
14809 of the edge closing it.
14811 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14813 * config/aarch64/aarch64.c
14814 (aarch64_can_const_movi_rtx_p): Move 0 check.
14816 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14818 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14819 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14821 * match.pd: Ditto in address comparison pattern.
14823 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14825 PR tree-optimization/81627
14826 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14827 closed ssa form for store-store chain.
14829 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14831 PR tree-optimization/81620
14832 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14833 for store-store chain.
14835 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14837 PR tree-optimization/81588
14838 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14839 ranges[i].in_p, invert comparison code ccode. For >/>=,
14840 swap rhs1 and rhs2 and comparison code unconditionally,
14841 for </<= don't do that. Don't swap rhs1/rhs2 again if
14842 ranges[i].in_p, instead invert comparison code ccode if
14843 opcode or oe->rank is BIT_IOR_EXPR.
14846 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14847 a direct optab to conversion optab.
14848 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14849 with GET_MODE_INNER as last argument instead of optab_handler.
14850 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14851 vector extraction if possible and optab is available.
14852 * expr.c (store_constructor): Use convert_optab_handler instead
14853 of optab_handler. Use vector initialization from smaller
14854 vectors if possible and optab is available.
14855 * tree-vect-stmts.c (vectorizable_load): Likewise.
14856 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14857 now have two modes.
14858 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14859 of vec_init from half-sized vectors with the same element mode.
14860 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14861 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14862 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14863 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14864 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14865 after mode in gen_vec_extract* calls.
14866 (vec_extract<mode>): Renamed to ...
14867 (vec_extract<mode><ssescalarmodelower>): ... this.
14868 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14869 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14870 element mode after mode in gen_vec_init* calls.
14871 (VEC_INIT_HALF_MODE): New mode iterator.
14872 (vec_init<mode>): Renamed to ...
14873 (vec_init<mode><ssescalarmodelower>): ... this.
14874 (vec_init<mode><ssehalfvecmodelower>): New expander.
14875 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14876 (vec_extractv2sfsf): ... this.
14877 (vec_initv2sf): Renamed to ...
14878 (vec_initv2sfsf): ... this.
14879 (vec_extractv2si): Renamed to ...
14880 (vec_extractv2sisi): ... this.
14881 (vec_initv2si): Renamed to ...
14882 (vec_initv2sisi): ... this.
14883 (vec_extractv4hi): Renamed to ...
14884 (vec_extractv4hihi): ... this.
14885 (vec_initv4hi): Renamed to ...
14886 (vec_initv4hihi): ... this.
14887 (vec_extractv8qi): Renamed to ...
14888 (vec_extractv8qiqi): ... this.
14889 (vec_initv8qi): Renamed to ...
14890 (vec_initv8qiqi): ... this.
14891 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14892 (vec_init<mode>): Renamed to ...
14893 (vec_init<mode><VEC_base_l>): ... this.
14894 (vec_extract<mode>): Renamed to ...
14895 (vec_extract<mode><VEC_base_l>): ... this.
14896 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14897 (vec_initv2sfsf): ... this.
14898 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14899 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14900 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14901 element mode after mode in gen_vec_init* calls.
14902 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14903 (vec_init<mode><Vel>): ... this.
14904 (vec_extract<mode>): Renamed to ...
14905 (vec_extract<mode><Vel>): ... this.
14906 * config/aarch64/iterators.md (Vel): New mode attribute.
14907 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14908 Add element mode after mode in gen_vec_extract* calls.
14909 * config/s390/vector.md (non_vec_l): New mode attribute.
14910 (vec_extract<mode>): Renamed to ...
14911 (vec_extract<mode><non_vec_l>): ... this.
14912 (vec_init<mode>): Renamed to ...
14913 (vec_init<mode><non_vec_l>): ... this.
14914 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14915 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14917 * config/arm/iterators.md (V_elem_l): New mode attribute.
14918 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14919 (vec_extract<mode><V_elem_l>): ... this.
14920 (vec_extractv2di): Renamed to ...
14921 (vec_extractv2didi): ... this.
14922 (vec_init<mode>): Renamed to ...
14923 (vec_init<mode><V_elem_l>): ... this.
14924 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14925 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14926 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14927 Add element mode after gen_vec_extract* calls.
14928 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14929 (vec_init<mode><unitmode>): ... this.
14930 (vec_extract<mode>): Renamed to ...
14931 (vec_extract<mode><unitmode>): ... this.
14932 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14933 (vec_init<mode><unitmode>): ... this.
14934 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14935 (vec_initv2sfsf): ... this.
14936 (vec_extractv2sf): Renamed to ...
14937 (vec_extractv2sfsf): ... this.
14938 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14939 Add element mode after gen_vec_extract* calls.
14940 * config/mips/mips.md (unitmode): New mode iterator.
14941 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14942 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14943 * config/spu/spu.md (inner_l): New mode attribute.
14944 (vec_init<mode>): Renamed to ...
14945 (vec_init<mode><inner_l>): ... this.
14946 (vec_extract<mode>): Renamed to ...
14947 (vec_extract<mode><inner_l>): ... this.
14948 * config/sparc/sparc.md (veltmode): New mode iterator.
14949 (vec_init<VMALL:mode>): Renamed to ...
14950 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14951 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14952 (vec_initv2sisi): ... this.
14953 (vec_initv2sf): Renamed to ...
14954 (vec_initv2sfsf): ... this.
14955 (vec_extractv2sf): Renamed to ...
14956 (vec_extractv2sfsf): ... this.
14957 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14958 (vec_init<mode>): Renamed to ...
14959 (vec_init<mode><VEC_base_l>): ... this.
14960 (vec_extract<mode>): Renamed to ...
14961 (vec_extract<mode><VEC_base_l>): ... this.
14962 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14963 (vec_initv2sfsf): ... this.
14964 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14965 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14966 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14967 gen_vec_init* calls.
14969 2017-08-01 Richard Biener <rguenther@suse.de>
14971 PR tree-optimization/81297
14972 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14973 TREE_OVERFLOW from INTEGER_CSTs.
14975 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14977 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14979 2017-07-31 Carl Love <cel@us.ibm.com>
14981 * config/rs6000/rs6000-c: Add support for built-in functions
14982 vector signed char vec_xl_be (signed long long, signed char *);
14983 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14984 vector signed int vec_xl_be (signed long long, signed int *);
14985 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14986 vector signed long long vec_xl_be (signed long long, signed long long *);
14987 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14988 vector signed short vec_xl_be (signed long long, signed short *);
14989 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14990 vector double vec_xl_be (signed long long, double *);
14991 vector float vec_xl_be (signed long long, float *);
14992 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14993 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14994 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14996 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14997 (altivec_expand_builtin): Add switch statement to call
14998 altivec_expand_xl_be for each builtin.
14999 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15000 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15001 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15002 __builtin_vsx_le_be_v16qi.
15003 * doc/extend.texi: Update the built-in documentation file for the
15004 new built-in functions.
15006 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15009 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15011 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15013 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15015 * config.gcc: Add z14.
15016 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15017 CPU model numbers for z13s and z14.
15018 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15020 * config/s390/s390-opts.h (enum processor_type): Rename
15021 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15022 * config/s390/s390.c (processor_table): Add field for CPU name to
15023 be passed to Binutils.
15024 (s390_asm_output_machine_for_arch): Use the new field in
15025 processor_table for Binutils.
15026 (s390_expand_builtin): Replace arch12 with z14.
15027 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15028 (s390_get_sched_attrmask): Likewise.
15029 (s390_get_unit_mask): Likewise.
15030 * config/s390/s390.opt: Add z14 to processor_type enum.
15032 2017-07-31 Martin Jambor <mjambor@suse.cz>
15035 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15036 regardless of optimization level.
15038 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15039 Martin Liska <mliska@suse.cz>
15041 * predict.def: Remove old comment and adjust probability.
15042 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15043 PREDICT statements.
15045 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15048 * config/i386/i386.c (ix86_function_naked): New function.
15049 (ix86_can_use_return_insn_p): Return false for naked functions.
15050 (ix86_expand_prologue): Skip prologue for naked functions.
15051 (ix86_expand_epilogue): Skip epilogue for naked functions
15052 and emit trap instruction.
15053 (ix86_warn_func_return): New function.
15054 (ix86_attribute_table): Add "naked" attribute specification.
15055 (TARGET_WARN_FUNC_RETURN): Define.
15056 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15058 2017-07-31 Martin Liska <mliska@suse.cz>
15060 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15061 (dump_gimple_bb_header): Always dump BB info.
15062 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15064 2017-07-31 Martin Liska <mliska@suse.cz>
15067 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15068 also with current_function_decl non-null equality.
15070 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15073 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15074 change type to the element type, instead add eltype variable and
15075 use it where we are interested in the element type.
15077 PR tree-optimization/81603
15078 * ipa-polymorphic-call.c
15079 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15080 offset arithmetic in offset_int, bail out if the resulting bit offset
15081 doesn't fit into shwi.
15083 2017-07-31 Martin Liska <mliska@suse.cz>
15085 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15086 (gimplify_save_expr): Fix comment.
15088 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15091 * config/i386/i386.c (ix86_function_arg): Update arguments for
15093 (ix86_compute_frame_layout): Set the initial stack offset to
15094 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15095 INCOMING_FRAME_SP_OFFSET.
15096 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15097 stack before exception handler returns.
15098 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15099 the 'ERROR_CODE' for exception handler.
15101 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15103 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15104 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15105 (ASM_OUTPUT_REG_POP): Ditto.
15106 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15107 instead of asm_fprintf to output pure string.
15109 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15111 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15112 to imported_module_or_decl hook.
15113 (debug_nothing_tree_tree_tree_bool): Remove.
15114 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15115 * debug.c (do_nothing_debug_hooks): Use
15116 debug_nothing_tree_tree_tree_bool_bool instead of
15117 debug_nothing_tree_tree_tree_bool.
15118 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15119 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15120 * sdbout.c (sdb_debug_hooks): Likewise.
15121 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15122 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15124 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15125 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15126 attribute, don't add anything.
15128 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15130 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15131 (fold_build2_stat_loc): Likewise.
15132 (fold_build3_stat_loc): Likewise.
15133 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15134 (fold_build1_loc): Remove macro.
15135 (fold_build2_loc): Likewise.
15136 (fold_build3_loc): Likewise.
15138 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15140 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15141 (gimple_build_debug_bind_source_stat): Likewise.
15142 * gimple.h (gimple_build_debug_bind): Remove macro.
15143 (gimple_build_debug_bind_source): Likewise.
15145 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15147 * bitmap.c (bitmap_alloc): Adjust.
15148 (bitmap_gc_alloc): Likewise.
15149 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15151 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15153 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15154 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15155 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15156 (bitmap_gc_alloc_stat): Likewise.
15157 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15161 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15162 * rtl.h (shallow_copy_rtx): Remove macro.
15164 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15166 * emit-rtl.c (gen_raw_REG): Adjust.
15167 * gengenrtl.c (gendef): Likewise.
15168 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15169 * rtl.h (rtx_alloc): Remove macro.
15171 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15173 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15174 (build_tree_list_stat): Likewise.
15175 * tree.h (build_tree_list): Remove macro.
15176 (build_tree_list_vec): Likewise.
15178 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15180 * tree.c (make_vector_stat): Remove _stat from name.
15181 (build_vector_stat): Likewise.
15182 * tree.h (make_vector_stat): Remove macro.
15183 (build_vector_stat): Likewise.
15185 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15187 * tree.h (build_var_debug_value): Remove prototype.
15189 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15191 * tree.c (tree_cons_stat): Remove _stat from name.
15192 * tree.h (tree_cons): Remove macro.
15194 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15196 * tree.c (build_vl_exp_stat): Remove _stat from name.
15197 * tree.h (build_vl_exp): Remove macro.
15199 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15201 * tree.c (build_decl_stat): Remove _stat from name.
15202 * tree.h (build_decl): Remove macro.
15204 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15206 * gimple.c (gimple_build_with_ops_stat): Adjust.
15207 (gimple_alloc_stat): Remove _stat from name.
15208 * gimple.h (gimple_alloc): Remove macro.
15210 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15212 * tree.c (make_tree_vec_stat): Remove _stat from name.
15213 (grow_tree_vec_stat): Likewise.
15214 * tree.h (make_tree_vec_stat): Adjust prototype.
15215 (grow_tree_vec_stat): Likewise.
15216 (make_tree_vec): Remove macro.
15217 (grow_tree_vec): Likewise.
15219 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15221 * fold-const.c (fold_build1_stat_loc): Adjust.
15222 (fold_build2_stat_loc): Likewise.
15223 (fold_build3_stat_loc): Likewise.
15224 * tree.c (build0_stat): Remove _stat from name.
15225 (build1_stat): Likewise.
15226 (build2_stat): Likewise.
15227 (build3_stat): Likewise.
15228 (build4_stat): Likewise.
15229 (build5_stat): Likewise.
15230 * tree.h (build1_loc): Remove macro, and rename _stat function
15232 (build2_loc): Likewise.
15233 (build3_loc): Likewise.
15234 (build4_loc): Likewise.
15235 (build5_loc): Likewise.
15237 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15239 * tree.c (make_int_cst_stat): Remove _stat from name.
15240 * tree.h (make_int_cst_stat): Adjust prototype.
15241 (make_int_cst): Remove macro.
15243 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15245 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15246 * tree.h (make_tree_binfo_stat): Adjust prototype.
15247 (make_tree_binfo): Remove.
15249 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15251 * tree.c (copy_node_stat): Rename to copy_node.
15252 (build_distinct_type_copy): Adjust.
15253 * tree.h (copy_node_stat): Adjust prototype.
15254 (copy_node): Remove macro.
15256 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15258 * tree.c (make_node_stat): rename to make_node.
15259 (build_tree_list_stat): Adjust.
15260 (build0_stat): Likewise.
15261 (build2_stat): Likewise.
15262 (build3_stat): Likewise.
15263 (build4_stat): Likewise.
15264 (build5_stat): Likewise.
15265 (build_decl_stat): Likewise.
15266 * tree.h (make_node_stat): Adjust prototype.
15267 (make_node): remove macro.
15269 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15271 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15272 (PPC_FEATURE2_SCV): Likewise.
15273 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15275 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15277 * config/aarch64/aarch64.c
15278 (aarch64_internal_mov_immediate): Add new special pattern.
15279 * config/aarch64/aarch64.md (*movdi_aarch64):
15280 Add reg/32bit const mov case.
15282 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15283 Richard Sandiford <richard.sandiford@linaro.org>
15285 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15286 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15287 Add integer and movi cases.
15288 (movi-split-hf-df-sf split, fp16): New.
15289 (enabled): Added TARGET_FP_F16INST.
15290 * config/aarch64/iterators.md (GPF_HF): New.
15291 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15293 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15295 * config/aarch64/aarch64.c
15296 (aarch64_simd_container_mode): Add prototype.
15297 (aarch64_expand_mov_immediate): Add HI support.
15298 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15299 (aarch64_can_const_movi_rtx_p): New.
15300 (aarch64_preferred_reload_class):
15301 Remove restrictions of using FP registers for certain SIMD operations.
15302 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15303 (aarch64_valid_floating_const): Add integer move validation.
15304 (aarch64_simd_imm_scalar_p): Remove.
15305 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15306 (aarch64_legitimate_constant_p): Expand list of supported cases.
15307 * config/aarch64/aarch64-protos.h
15308 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15309 (aarch64_reinterpret_float_as_int): New.
15310 (aarch64_simd_imm_scalar_p): Remove.
15311 * config/aarch64/constraints.md (Uvi): New.
15312 (Dd): Split into Ds and new Dd.
15313 * config/aarch64/aarch64.md (*movsi_aarch64):
15315 (*movdi_aarch64): Add SIMD mov case.
15317 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15319 * tree-predcom.c: (struct chain): Handle store-store chain in which
15320 stores for elimination only store loop invariant values.
15321 (execute_pred_commoning_chain): Ditto.
15322 (prepare_initializers_chain_store_elim): Ditto.
15323 (prepare_finalizers): Ditto.
15324 (is_inv_store_elimination_chain): New function.
15325 (initialize_root_vars_store_elim_1): New function.
15327 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15329 * tree-predcom.c: Revise general description of the pass.
15330 (enum chain_type): New enum type for store elimination.
15331 (struct chain): New field supporting store elimination.
15332 (struct component): Ditto.
15333 (dump_chain): Dump store-stores chain.
15334 (release_chain): Release resources.
15335 (split_data_refs_to_components): Compute and create component
15336 contains only stores for elimination.
15337 (get_chain_last_ref_at): New function.
15338 (make_invariant_chain): Initialization.
15339 (make_rooted_chain): Specify chain type in parameter and record it.
15340 (add_looparound_copies): Skip for store-stores chain.
15341 (determine_roots_comp): Compute type of chain and pass it to
15343 (initialize_root_vars_store_elim_2): New function.
15344 (finalize_eliminated_stores): New function.
15345 (remove_stmt): Handle store for elimination.
15346 (execute_pred_commoning_chain): Execute predictive commoning on
15347 store-store chains.
15348 (determine_unroll_factor): Skip unroll for store-stores chain.
15349 (prepare_initializers_chain_store_elim): New function.
15350 (prepare_initializers_chain): Hanlde store-store chain.
15351 (prepare_finalizers_chain, prepare_finalizers): New function.
15352 (tree_predictive_commoning_loop): Return integer value indicating
15353 if loop is unrolled or lcssa form is corrupted.
15354 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15356 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15358 * tree-predcom.c (initialize_root): Delete.
15359 (execute_pred_commoning_chain): Initialize root vars and replace
15360 reference of non-combined chain directly, rather than call above
15363 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15365 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15366 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15368 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15370 * tree-predcom.c (struct chain): New field init_seq.
15371 (release_chain): Release init_seq.
15372 (prepare_initializers_chain): Record intialization stmts in above
15374 (insert_init_seqs): New function.
15375 (tree_predictive_commoning_loop): Call insert_init_seqs.
15377 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15379 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15381 2017-07-28 Richard Biener <rguenther@suse.de>
15383 * match.pd: Remove superfluous :c.
15384 * genmatch.c (simplify::id): Add member.
15385 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15387 (current_id): New global.
15388 (dt_node::parent): Move from ...
15389 (dt_operand::parent): ... here. Add for_id member.
15390 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15391 (decision_tree::find_node): Relax order requirement when
15392 merging DT_TRUE nodes to ones inbetween the current simplify
15393 and the one we try to merge with. Add diagnostic whenever
15394 we need to enforce pattern order by not merging.
15395 (decision_tree::insert): Set current_id.
15396 (decision_tree::print_node): Dump parent node and for_id.
15397 (parser::last_id): Add member.
15398 (parser::push_simplify): Assign unique id.
15399 (parser::parser): Initialize last_id.
15401 2017-07-28 Martin Liska <mliska@suse.cz>
15404 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15405 gimple_build_debug_bind.
15407 2017-07-28 Richard Biener <rguenther@suse.de>
15409 PR tree-optimization/81502
15410 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15412 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15414 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15415 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15416 for BIT_FIELD_REF args.
15417 * fold-const.c (make_bit_field_ref): Likewise.
15418 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15420 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15423 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15424 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15425 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15426 Or it into SANITIZER_UNDEFINED.
15427 * ubsan.c: Include gimple-fold.h and varasm.h.
15428 (ubsan_expand_ptr_ifn): New function.
15429 (instrument_pointer_overflow): New function.
15430 (maybe_instrument_pointer_overflow): New function.
15431 (instrument_object_size): Formatting fix.
15432 (pass_ubsan::execute): Call instrument_pointer_overflow
15433 and maybe_instrument_pointer_overflow.
15434 * internal-fn.c (expand_UBSAN_PTR): New function.
15435 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15436 * sanitizer.def (__ubsan_handle_pointer_overflow,
15437 __ubsan_handle_pointer_overflow_abort): New builtins.
15438 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15439 * internal-fn.def (UBSAN_PTR): New internal function.
15440 * opts.c (sanitizer_opts): Add pointer-overflow.
15441 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15442 * fold-const.c (build_range_check): Compute pointer range check in
15443 integral type if pointer arithmetics would be needed. Formatting
15446 2017-07-28 Martin Liska <mliska@suse.cz>
15449 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15450 parameters that are of a variable-length.
15452 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15454 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15456 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15457 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15458 (CRT_CALL_STATIC_FUNCTION): Likewise.
15459 (ASM_DEFAULT_SPEC): New define.
15460 (ASM_SPEC32): Likewise.
15461 (ASM_SPEC64): Likewise.
15462 (ASM_SPEC_COMMON): Likewise.
15463 (ASM_SPEC): Likewise.
15464 (INVALID_64BIT): Likewise.
15465 (LINK_OS_DEFAULT_SPEC): Likewise.
15466 (LINK_OS_SPEC32): Likewise.
15467 (LINK_OS_SPEC64): Likewise.
15468 (POWERPC_LINUX): Likewise.
15469 (PTRDIFF_TYPE): Likewise.
15470 (RESTORE_FP_PREFIX): Likewise.
15471 (RESTORE_FP_SUFFIX): Likewise.
15472 (SAVE_FP_PREFIX): Likewise.
15473 (SAVE_FP_SUFFIX): Likewise.
15474 (SIZE_TYPE): Likewise.
15475 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15476 (TARGET_64BIT): Likewise.
15477 (TARGET_64BIT): Likewise.
15478 (TARGET_AIX): Likewise.
15479 (WCHAR_TYPE_SIZE): Likewise.
15480 (WCHAR_TYPE): Undefine.
15481 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15482 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15483 (CPP_OS_RTEMS_SPEC): Delete.
15484 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15485 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15487 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15489 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15491 PR tree-optimization/81578
15492 * tree-parloops.c (build_new_reduction): Bail out if
15493 reduction_code isn't one of the standard OpenMP reductions.
15494 Move the details printing after that decision.
15496 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15498 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15499 related to reload_in_progress.
15500 (splat_input_operand): Likewise.
15501 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15503 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15505 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15506 (TARGET_INSTANTIATE_DECLS): Likewise.
15507 (legitimate_indexed_address_p): Delete reload_in_progress code.
15508 (rs6000_debug_legitimate_address_p): Likewise.
15509 (rs6000_eliminate_indexed_memrefs): Likewise.
15510 (rs6000_emit_le_vsx_store): Likewise.
15511 (rs6000_emit_move_si_sf_subreg): Likewise.
15512 (rs6000_emit_move): Likewise.
15513 (register_to_reg_type): Likewise.
15514 (rs6000_pre_atomic_barrier): Likewise.
15515 (rs6000_machopic_legitimize_pic_address): Likewise.
15516 (rs6000_allocate_stack_temp): Likewise.
15517 (rs6000_address_for_fpconvert): Likewise.
15518 (rs6000_address_for_altivec): Likewise.
15519 (rs6000_secondary_memory_needed_rtx): Delete function.
15520 (rs6000_check_sdmode): Likewise.
15521 (rs6000_alloc_sdmode_stack_slot): Likewise.
15522 (rs6000_instantiate_decls): Likewise.
15523 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15524 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15525 Delete reload_in_progress.
15526 (*vec_reload_and_plus_<mptrsize>): Likewise.
15527 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15528 (vsx_div_v2di): Likewise.
15529 (vsx_udiv_v2di): Likewise.
15531 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15533 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15534 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15535 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15536 (rs6000_debug_reg_global): Delete print of LRA status.
15537 (rs6000_option_override_internal): Delete dead LRA related code.
15538 (rs6000_lra_p): Delete function.
15539 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15541 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15543 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15544 * config/riscv/rtems.h: New file.
15546 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15547 Sudakshina Das <sudi.das@arm.com>
15549 * config/aarch64/aarch64.md
15550 (define_split for and<mode>3nr_compare): Move
15551 non aarch64_logical_operand to a register.
15552 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15553 register immediate operand to a register.
15554 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15556 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15558 PR middle-end/81564
15559 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15561 2017-07-27 Richard Biener <rguenther@suse.de>
15563 PR tree-optimization/81573
15564 PR tree-optimization/81494
15565 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15566 multi defuse cycle case.
15568 2017-07-27 Richard Biener <rguenther@suse.de>
15570 PR tree-optimization/81571
15571 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15574 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15576 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15577 earlier and only if MASK_FPU is set. Adjust formatting.
15579 2017-07-27 Martin Liska <mliska@suse.cz>
15581 * opt-functions.awk: Add validation of value of Init.
15582 * optc-gen.awk: Pass new argument.
15584 2017-07-27 Martin Liska <mliska@suse.cz>
15586 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15587 Fix wrong condition.
15589 2017-07-27 Martin Liska <mliska@suse.cz>
15591 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15592 BBs and edges seen by autoFDO.
15594 2017-07-27 Richard Biener <rguenther@suse.de>
15596 PR tree-optimization/81502
15597 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15598 with incompatible but same sized type.
15599 (execute_update_addresses_taken): Likewise.
15601 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15603 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15604 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15606 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15609 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15610 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15611 Change s_operand to memory_operand.
15613 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15615 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15616 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15617 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15618 Emit instructions rather than returning an expression. Handle TFmode
15619 and KFmode by casting to TImode.
15620 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15621 (rs6000_emit_le_vsx_store): Likewise.
15622 * config/rs6000/vsx.md (VSX_TI): New iterator.
15623 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15624 (*vsx_le_undo_permute_<mode>): Likewise.
15625 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15626 emit the split sequence.
15627 (*vsx_le_perm_store_<mode>): Likewise.
15629 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15631 PR tree-optimization/81555
15632 PR tree-optimization/81556
15633 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15634 if true, force CHANGED for the recursive invocation.
15635 (reassociate_bb): Remember original length of ops array, pass
15636 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15638 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15639 attributes for noipa attribute. For naked attribute use
15640 lookup_attribute first before lookup_attribute_spec.
15641 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15643 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15644 for functions with noipa attribute.
15645 (cgraph_externally_visible_p): Return true for functions with noipa
15647 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15648 for functions with noipa attribute.
15649 * doc/extend.texi: Document noipa function attribute.
15650 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15651 also for functions with noipa attribute.
15652 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15654 2017-07-26 Andrew Pinski <apinski@cavium.com>
15656 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15657 vec_unalign_load_cost and vec_unalign_store_cost.
15659 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15661 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15662 -mvsx-small-integer option.
15663 (ISA_3_0_MASKS_IEEE): Likewise.
15664 (OTHER_VSX_VECTOR_MASKS): Likewise.
15665 (POWERPC_MASKS): Likewise.
15666 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15667 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15668 code, only testing for DImode being allowed in non-VSX floating
15670 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15671 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15673 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15674 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15675 TARGET_P8_VECTOR test.
15676 (rs6000_secondary_reload_simple_move): Likewise.
15677 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15678 since TARGET_P9_VECTOR was already tested.
15679 (rs6000_opt_masks): Remove -mvsx-small-integer.
15680 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15681 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15683 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15684 test for TARGET_VEXTRACTUB was used, and that uses
15686 (p9 extract splitter): Likewise.
15687 (vsx_extract_<mode>_di_p9): Likewise.
15688 (vsx_extract_<mode>_store_p9): Likewise.
15689 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15690 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15691 the elimination of TARGET_VSX_SMALL_INTEGER.
15692 (vsx_extract_<mode>_p8): Likewise.
15693 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15694 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15695 (vsx_set_<mode>_p9): Likewise.
15696 (vsx_set_v4sf_p9): Likewise.
15697 (vsx_set_v4sf_p9_zero): Likewise.
15698 (vsx_insert_extract_v4sf_p9): Likewise.
15699 (vsx_insert_extract_v4sf_p9_2): Likewise.
15700 * config/rs6000/rs6000.md (sign extend splitter): Change
15701 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15702 (floatsi<mode>2_lfiwax_mem): Likewise.
15703 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15704 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15705 since a test for TARGET_P9_VECTOR was used.
15706 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15707 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15708 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15709 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15710 TARGET_P8_VECTOR test.
15711 (fix_trunc<mode>si2_stfiwx): Likewise.
15712 (fix_trunc<mode>si2_internal): Likewise.
15713 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15714 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15716 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15717 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15718 TARGET_P8_VECTOR test.
15719 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15720 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15721 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15723 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15724 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15725 since a test for TARGET_P9_VECTOR was used.
15726 (splitter for loading small constants): Likewise.
15728 2017-07-26 Andrew Pinski <apinski@cavium.com>
15730 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15733 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15736 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15737 (fp_valid_at): Likewise.
15739 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15741 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15742 (qdf24xx_addrcost_table): Likewise.
15743 (cortexa57_tunings): Update to use generic_branch_cost.
15744 (cortexa72_tunings): Likewise.
15745 (cortexa73_tunings): Likewise.
15746 (qdf24xx_tunings): Likewise.
15748 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15750 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15751 (thunderx2t99_branch_cost): Likewise.
15752 (cortexa35_tunings): Update to use generic_branch_cost.
15753 (cortexa53_tunings): Likewise.
15754 (cortexa57_tunings): Likewise.
15755 (cortexa72_tunings): Likewise.
15756 (cortexa73_tunings): Likewise.
15757 (thunderx2t99_tunings): Likewise.
15759 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15761 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15762 (sparc_option_override): Honour MASK_FSMULD.
15763 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15764 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15765 * config/sparc/sparc.opt (mfsmuld): New option.
15766 * doc/invoke.texi (mfsmuld): Document option.
15768 2017-07-26 Marek Polacek <polacek@redhat.com>
15770 PR middle-end/70992
15771 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15773 2017-07-26 Richard Biener <rguenther@suse.de>
15775 * gimple-match-head.c (do_valueize): Return OP if valueize
15777 (get_def): New helper to get at the def stmt of a SSA name
15778 if valueize allows.
15779 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15780 do_valueize to get at the def stmt.
15781 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15783 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15785 PR middle-end/46932
15786 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15788 2017-07-26 Martin Liska <mliska@suse.cz>
15791 * function.c (expand_function_start): Make expansion of
15792 nonlocal_goto_save_area after parm_birth_insn.
15794 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15796 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15797 from all CPU target flags enable members.
15799 2017-07-26 Richard Biener <rguenther@suse.de>
15801 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15802 (decision_tree::gen): Make 'type' const.
15803 (write_predicate): Likewise.
15805 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15807 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15808 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15809 (rs6000_option_override_internal): Likewise.
15810 (rs6000_expand_vector_set): Likewise.
15811 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15812 (TARGET_UPPER_REGS_SF): Likewise.
15813 (TARGET_UPPER_REGS_DI): Likewise.
15814 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15815 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15816 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15817 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15818 (Splitters for DI constants in Altivec registers): Likewise.
15819 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15820 (vsx_set_v4sf_p9): Likewise.
15821 (vsx_set_v4sf_p9_zero): Likewise.
15822 (vsx_insert_extract_v4sf_p9): Likewise.
15823 (vsx_insert_extract_v4sf_p9_2): Likewise.
15825 2017-07-25 Carl Love <cel@us.ibm.com>
15827 * doc/extend.texi: Update the built-in documentation file for the
15828 existing built-in functions
15829 vector signed char vec_cnttz (vector signed char);
15830 vector unsigned char vec_cnttz (vector unsigned char);
15831 vector signed short vec_cnttz (vector signed short);
15832 vector unsigned short vec_cnttz (vector unsigned short);
15833 vector signed int vec_cnttz (vector signed int);
15834 vector unsigned int vec_cnttz (vector unsigned int);
15835 vector signed long long vec_cnttz (vector signed long long);
15836 vector unsigned long long vec_cnttz (vector unsigned long long);
15838 2017-07-25 Andrew Pinski <apinski@cavium.com>
15840 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15841 accesses where the use is for the first operand of a BIT_INSERT.
15843 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15846 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15847 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15849 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15851 * config/i386/gstabs.h: Delete.
15852 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15854 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15856 * config/i386/i386.c (ix86_decompose_address): Do not check for
15857 register RTX when looking at index_reg or base_reg.
15858 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15860 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15862 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15863 to update EH info here.
15865 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15867 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15869 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15871 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15873 2017-07-25 Torsten Duwe <duwe@suse.de>
15875 * common.opt: Introduce -fpatchable-function-entry
15876 command line option, and its variables function_entry_patch_area_size
15877 and function_entry_patch_area_start.
15878 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15879 including a two-value parser.
15880 * target.def (print_patchable_function_entry): New target hook.
15881 * targhooks.h (default_print_patchable_function_entry): New function.
15882 * targhooks.c (default_print_patchable_function_entry): Likewise.
15883 * toplev.c (process_options): Switch off IPA-RA if
15884 patchable function entries are being generated.
15885 * varasm.c (assemble_start_function): Look at the
15886 patchable-function-entry command line switch and current
15887 function attributes and maybe generate NOP instructions by
15888 calling the print_patchable_function_entry hook.
15889 * doc/extend.texi: Document patchable_function_entry attribute.
15890 * doc/invoke.texi: Document -fpatchable_function_entry
15891 command line option.
15892 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15894 * doc/tm.texi: Re-generate.
15896 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15899 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15900 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15902 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15904 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15906 2017-07-25 Richard Biener <rguenther@suse.de>
15908 PR tree-optimization/81455
15909 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15910 not walk in cycles when looking for guards.
15912 2017-07-25 Richard Biener <rguenther@suse.de>
15914 PR tree-optimization/81529
15915 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15916 when optimizing backedge uses.
15918 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15920 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15922 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15923 to dl_section_ref. On AIX, append an expression to subtract
15924 the size of the section length to dl_section_ref.
15926 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15928 * configure.ac: If any of the config.* scripts fail, exit 1.
15929 * configure: Regenerate.
15931 2017-07-25 Richard Biener <rguenther@suse.de>
15933 PR middle-end/81546
15934 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15935 of immediate uses, be more verbose on errors.
15937 2017-07-25 Richard Biener <rguenther@suse.de>
15939 PR tree-optimization/81510
15940 * tree-vect-loop.c (vect_is_simple_reduction): When the
15941 reduction stmt is not inside the loop bail out.
15943 2017-07-25 Richard Biener <rguenther@suse.de>
15945 PR tree-optimization/81303
15946 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15947 profitability check against LOOP_VINFO_NITERSM1.
15949 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15951 * domwalk.c (cmp_bb_postorder): Simplify.
15952 (sort_bbs_postorder): New function. Use it...
15953 (dom_walker::walk): ...here to optimize common cases.
15955 2017-07-25 Martin Liska <mliska@suse.cz>
15958 * ipa-visibility.c (function_and_variable_visibility): Make the
15959 redirection just on target that supports aliasing.
15960 Fix GNU coding style.
15962 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15965 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15967 * config/arm/rtems.h: Likewise.
15968 * config/bfin/rtems.h: Likewise.
15969 * config/i386/rtemself.h: Likewise.
15970 * config/lm32/rtems.h: Likewise.
15971 * config/m32c/rtems.h: Likewise.
15972 * config/m68k/rtemself.h: Likewise.
15973 * config/microblaze/rtems.h: Likewise.
15974 * config/mips/rtems.h: Likewise.
15975 * config/moxie/rtems.h: Likewise.
15976 * config/nios2/rtems.h: Likewise.
15977 * config/powerpcspe/rtems.h: Likewise.
15978 * config/rs6000/rtems.h: Likewise.
15979 * config/rtems.h: Likewise.
15980 * config/sh/rtems.h: Likewise.
15981 * config/sh/rtemself.h: Likewise.
15982 * config/sparc/rtemself.h: Likewise.
15984 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15987 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15988 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15989 * tree-ssa-structalias.c (alias_get_name): Same.
15991 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15994 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15995 instructions if no du chain is found.
15997 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15999 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16001 2017-07-25 Richard Biener <rguenther@suse.de>
16003 PR middle-end/81505
16004 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16007 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16009 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16010 upper-regs options.
16011 (ISA_2_7_MASKS_SERVER): Likewise.
16012 (ISA_3_0_MASKS_IEEE): Likewise.
16013 (OTHER_P8_VECTOR_MASKS): Likewise.
16014 (OTHER_VSX_VECTOR_MASKS): Likewise.
16015 (POWERPC_MASKS): Likewise.
16016 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16017 duplicate list of options.
16018 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16019 explicit -mupper-regs options.
16020 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16021 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16022 alias for -mupper-regs-df.
16023 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16024 (rs6000_init_hard_regno_mode_ok): Likewise.
16025 (rs6000_option_override_internal): Likewise.
16026 (rs6000_opt_masks): Likewise.
16027 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16028 options in terms of whether -mvsx or -mpower8-vector was used.
16029 (TARGET_UPPER_REGS_DI): Likewise.
16030 (TARGET_UPPER_REGS_SF): Likewise.
16031 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16032 -mupper-regs-* options.
16034 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16036 * passes.c (emergency_dump_function): Print some empty lines and a
16037 header before the RTL dump.
16039 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16041 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16043 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16046 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16047 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16049 2017-07-24 Carl Love <cel@us.ibm.com>
16051 * config/rs6000/rs6000-c.c: Add support for built-in functions
16052 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16053 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16054 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16055 vec_extract_fp_from_shortl): Add defines for the two builtins.
16056 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16057 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16059 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16060 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16061 * doc/extend.texi: Update the built-in documentation file for the
16062 new built-in function.
16064 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16067 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16069 * doc/generic.texi: Likewise.
16070 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16071 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16073 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16075 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16076 (aarch64_mls_elt_merge<mode>): Likewise.
16078 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16080 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16081 having __cxa_atexit.
16083 2017-07-23 Michael Collison <michael.collison@arm.com>
16085 * config/arm/arm.c (arm_option_override): Deprecate
16086 use of -mstructure-size-boundary.
16087 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16088 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16090 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16093 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16094 Reduce cost estimate for direct moves.
16096 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16099 * config/i386/i386.c (ix86_option_override_internal): Disable
16100 BMI, BMI2 and TBM instructions for -m16.
16102 2017-07-21 Carl Love <cel@us.ibm.com>
16104 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16105 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16106 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16107 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16108 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16109 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16110 VMULOSW): New enum "unspec" values.
16111 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16112 altivec_vmulosw): New patterns.
16113 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16114 VMULOSW): Add definitions.
16116 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16118 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16119 (qdf24xx): Likewise.
16120 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16121 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16122 (AARCH64_FL_V8_1): Renumber.
16123 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16124 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16125 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16126 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16127 rdma to feature modifiers list.
16129 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16131 PR middle-end/56727
16132 * ipa-visibility (function_and_variable_visibility): Convert
16133 recursive PLT call to direct call if appropriate.
16135 2017-07-21 Andrew Pinski <apinski@cavium.com>
16137 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16138 operand 1 to see if the types precision matches.
16139 * fold-const.c (operand_equal_p): Likewise.
16141 2017-07-21 Richard Biener <rguenther@suse.de>
16143 PR tree-optimization/81303
16144 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16145 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16146 (vect_peeling_hash_get_lowest_cost): Adjust.
16147 (vect_enhance_data_refs_alignment): Likewise. Use
16148 vect_get_peeling_costs_all_drs to compute the penalty for no
16149 peeling to match up costs.
16151 2017-07-21 Richard Biener <rguenther@suse.de>
16153 PR tree-optimization/81500
16154 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16155 we didn't identify a reduction path.
16157 2017-07-21 Tom de Vries <tom@codesourcery.com>
16158 Cesar Philippidis <cesar@codesourcery.com>
16160 PR gcov-profile/81442
16161 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16164 2017-07-21 Tom de Vries <tom@codesourcery.com>
16167 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16169 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16170 nvptx_override_options_after_change.
16172 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16174 * dwarf2out.c (output_file_names): Avoid double testing for
16175 dwarf_version >= 5.
16177 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16179 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16181 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16183 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16185 (try_crossjump_to_edge): Do not punt on partitioned functions.
16187 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16189 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16190 Put all BBs reachable only via paths crossing cold region to cold
16192 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16194 2016-07-21 Richard Biener <rguenther@suse.de>
16196 PR tree-optimization/81303
16197 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16198 into account prologue and epilogue iterations when raising
16199 min_profitable_iters to sth at least covering one vector iteration.
16201 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16203 * config/arm/arm.c (arm_test_cpu_arch_dat):
16206 2017-07-20 Nathan Sidwell <nathan@acm.org>
16208 Remove TYPE_METHODS.
16209 * tree.h (TYPE_METHODS): Delete.
16210 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16211 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16212 (dbxout_type_methods): Scan TYPE_FIELDS.
16213 (dbxout_type): Don't check TYPE_METHODS here.
16214 * function.c (use_register_for_decl): Always ignore register for
16215 class types when not optimizing.
16216 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16217 * tree.c (free_lang_data_in_type): Stitch out member functions and
16218 templates from TYPE_FIELDS.
16219 (build_distinct_type_copy, verify_type_variant,
16220 verify_type): Member fns are on TYPE_FIELDS.
16221 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16222 * tree-pretty-print.c (dump_generic_node): Likewise.
16224 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16227 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16228 V2TImode and V4TImode.
16229 (ix86_expand_vector_extract): Likewise.
16230 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16231 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16232 (ssescalarmode): Handle V4TImode and V2TImode.
16233 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16234 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16235 (VEXTRACTI128_MODE): New mode iterator.
16236 (splitter for *vec_extractv?ti first element): New.
16237 (VEC_INIT_MODE): New mode iterator.
16238 (vec_init<mode>): Consolidate 3 expanders into one using
16239 VEC_INIT_MODE mode iterator.
16241 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16243 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16244 non_spilled_static_chain_regno_p.
16246 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16248 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16250 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16252 * bb-reorder.c (connect_traces): Allow copying of blocks within
16255 2017-07-20 Richard Biener <rguenther@suse.de>
16257 * gimple.h (gimple_phi_result): Add gphi * overload.
16258 (gimple_phi_result_ptr): Likewise.
16259 (gimple_phi_arg): Likewise. Adjust index assert to only
16260 allow actual argument accesses rather than all slots available
16262 (gimple_phi_arg_def): Add gphi * overload.
16263 * tree-phinodes.c (make_phi_node): Initialize only actual
16265 (resize_phi_node): Clear memory not covered by old node,
16266 do not initialize excess argument slots.
16267 (reserve_phi_args_for_new_edge): Initialize new argument slot
16270 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16272 PR tree-optimization/81388
16274 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16276 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16277 by removing computation of may_be_zero.
16279 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16280 Tom de Vries <tom@codesourcery.com>
16282 PR middle-end/81030
16283 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16284 when gimple level profile disagrees with what RTL expander did.
16286 2017-07-20 Richard Biener <rguenther@suse.de>
16288 PR tree-optimization/61171
16289 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16290 (vect_analyze_stmt): Add slp instance parameter.
16291 (vectorizable_reduction): Likewise.
16292 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16293 (vect_is_simple_reduction): Deal with chains not detected
16294 as SLP reduction chain, specifically not properly associated
16295 chains containing a mix of plus/minus.
16296 (get_reduction_op): Remove.
16297 (get_initial_defs_for_reduction): Simplify, pass in whether
16298 this is a reduction chain, pass in the SLP node for the PHIs.
16299 (vect_create_epilog_for_reduction): Get the SLP instance as
16301 (vectorizable_reduction): Get the SLP instance as arg.
16302 During analysis remember the SLP node with the PHIs in the
16303 instance. Simplify getting at the vectorized reduction PHIs.
16304 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16305 through SLP instance.
16306 (vect_slp_analyze_operations): Likewise.
16307 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16308 (vect_transform_stmt): Likewise.
16310 2017-07-20 Tom de Vries <tom@codesourcery.com>
16312 PR tree-optimization/81489
16313 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16314 read of phi arg location to before loop that modifies phi.
16316 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16318 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16321 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16323 PR middle-end/81331
16324 * except.c (execute): Fix ordering issue.
16326 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16328 PR rtl-optimization/81423
16329 * combine.c (make_compound_operation_int): Don't try to optimize
16330 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16332 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16334 PR rtl-optimization/81423
16335 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16336 with a constant that is -1 in the truncated to mode.
16338 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16340 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16341 (determine_unlikely_bbs): ... here.
16342 * predict.h (propagate_unlikely_bbs_forward): Declare.
16343 * cfgexpand.c (pass_expand::execute): Use it.
16344 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16346 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16347 propagate_unlikely_bbs_forward.
16349 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16351 PR middle-end/81331
16352 * except.c (maybe_add_nop_after_section_switch): New function.
16355 2017-07-19 Tom de Vries <tom@codesourcery.com>
16357 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16359 2017-07-19 Tom de Vries <tom@codesourcery.com>
16361 * gimple.h (gimple_phi_arg): Make assert more strict.
16363 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16365 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16366 * config/rs6000/mmintrin.h: New file.
16367 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16369 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16371 PR tree-optimization/81346
16372 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16374 2017-07-19 Tom de Vries <tom@codesourcery.com>
16376 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16378 2017-07-19 Tom de Vries <tom@codesourcery.com>
16380 * config/nvptx/nvptx-modes.def: Add V2DImode.
16381 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16382 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16383 (nvptx_output_mov_insn): Handle lack of mov.b128.
16384 (nvptx_print_operand): Handle 'H' and 'L' codes.
16385 (nvptx_vector_mode_supported): Allow V2DImode.
16386 (nvptx_preferred_simd_mode): New function.
16387 (nvptx_data_alignment): New function.
16388 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16389 nvptx_preferred_simd_mode.
16390 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16392 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16394 2017-07-19 Tom de Vries <tom@codesourcery.com>
16396 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16397 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16398 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16399 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16400 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16401 (mov<VECIM>_insn): New define_insn.
16402 (define_expand "mov<VECIM>): New define_expand.
16404 2017-07-19 Tom de Vries <tom@codesourcery.com>
16406 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16408 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16410 PR tree-optimization/81346
16411 * fold-const.h (fold_div_compare, range_check_type): Declare.
16412 * fold-const.c (range_check_type): New function.
16413 (build_range_check): Use range_check_type.
16414 (fold_div_compare): No longer static, rewritten into
16415 a match.pd helper function.
16416 (fold_comparison): Don't call fold_div_compare here.
16417 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16418 as helper function.
16420 2017-07-19 Nathan Sidwell <nathan@acm.org>
16422 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16423 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16424 * tree.c (find_decls_types_r, verify_type): Use
16425 TYPE_{MIN,MAX}_VALUE_RAW.
16426 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16427 (hash_tree): Likewise.
16428 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16430 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16433 2017-07-18 Tom de Vries <tom@codesourcery.com>
16435 PR middle-end/81464
16436 * omp-expand.c (expand_omp_for_static_chunk): Handle
16437 equal-argument loop exit phi.
16439 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16442 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16443 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16444 operand 2 predicate.
16445 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16446 operand 2 predicate.
16447 (ror,rol -> rorx splitters): Use const_int_operand as
16448 operand 2 predicate.
16450 2017-06-18 Richard Biener <rguenther@suse.de>
16452 PR tree-optimization/81410
16453 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16454 the gap in the ! slp_perm SLP case after each group.
16456 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16458 PR middle-end/81463
16459 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16462 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16464 PR middle-end/81462
16465 * predict.c (set_even_probabilities): Cleanup; do not affect
16466 probabilities that are already known.
16467 (combine_predictions_for_bb): Call even when count is set.
16469 2017-07-18 Nathan Sidwell <nathan@acm.org>
16471 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16474 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16477 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16478 optimization for loop niter analysis.
16480 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16483 * config/avr/avr.c (avr_optimize_casesi): Don't use
16484 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16486 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16488 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16489 body_cost_vec from _vect_peel_extended_info.
16490 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16491 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16494 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16496 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16498 2017-07-18 Richard Biener <rguenther@suse.de>
16500 PR tree-optimization/80620
16501 PR tree-optimization/81403
16502 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16503 info when re-using a VN table entry.
16505 2017-07-18 Richard Biener <rguenther@suse.de>
16507 PR tree-optimization/81418
16508 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16509 vectype_in. Verify that with lane-reducing reduction operations
16510 we have a single def-use cycle.
16512 2017-07-17 Carl Love <cel@us.ibm.com>
16514 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16516 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16517 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16518 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16519 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16520 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16521 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16522 VMULOSW): New enum "unspec" values.
16523 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16524 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16525 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16526 altivec_vmulosw): New patterns.
16527 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16528 VMULOSW): Add definitions.
16529 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16531 * config/alpha/alpha.c: Include predict.h.
16533 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16535 * tree-vrp.c (compare_assert_loc): Fix comparison function
16536 to return predictable results.
16538 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16540 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16542 (subdi3): Likewise.
16543 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16544 * doc/invoke.texi (mexpand-adddi): Update text.
16546 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16548 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16549 that also clobbers the CC register. The old expand code is moved
16551 (*arc_clzsi2): ... here.
16552 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16553 the CC register. The old expand code is moved to ...
16554 (arc_ctzsi2): ... here.
16556 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16558 * config/arc/arc.opt (mindexed-loads): Use initial value
16559 TARGET_INDEXED_LOADS_DEFAULT.
16560 (mauto-modify-reg): Use initial value
16561 TARGET_AUTO_MODIFY_REG_DEFAULT.
16562 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16563 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16564 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16565 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16567 2017-07-17 Martin Liska <mliska@suse.cz>
16570 * opts.c (finish_options): Do not allow -fgnu-tm
16571 w/ -fsanitize={kernel-,}address. Say sorry.
16573 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16576 * tree-loop-distribution.c (classify_partition): Only assert on
16577 numer of iterations.
16578 (merge_dep_scc_partitions): Delete prameter. Update function call.
16579 (distribute_loop): Remove code handling loop with unknown niters.
16580 (pass_loop_distribution::execute): Skip loop with unknown niters.
16582 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16585 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16586 function sort_partitions_by_post_order.
16588 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16590 PR tree-optimization/81374
16591 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16592 the max index of basic blocks, rather than number of basic blocks.
16594 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16596 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16598 (arc_legitimate_pic_operand_p): Likewise.
16599 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16601 (arc_needs_pcl_p): Likewise.
16602 (arc_legitimate_pc_offset_p): Likewise.
16603 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16604 function is also used in constrains.md.
16605 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16606 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16607 PLUS. Only return true/false in known cases, otherwise assert.
16608 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16609 is already called in arc_legitimate_constant_p.
16610 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16612 (LEGITIMATE_PIC_OPERAND_P): Use
16613 arc_raw_symbolic_reference_mentioned_p function.
16614 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16619 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16620 Andrew Burgess <andrew.burgess@embecosm.com>
16622 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16623 (arc_return_address_register): New function.
16624 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16625 (arc_handle_fndecl_attribute): Add naked attribute.
16626 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16627 (TARGET_WARN_FUNC_RETURN): Likewise.
16628 (arc_allocate_stack_slots_for_args): New function.
16629 (arc_warn_func_return): Likewise.
16630 (machine_function): Change type fn_type.
16631 (arc_compute_function_type): Consider new naked function type,
16632 change function return type.
16633 (arc_must_save_register): Adapt to handle new
16634 arc_compute_function_type's return type.
16635 (arc_expand_prologue): Likewise.
16636 (arc_expand_epilogue): Likewise.
16637 (arc_return_address_regs): Delete.
16638 (arc_return_address_register): New function.
16639 (arc_epilogue_uses): Use above function.
16640 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16641 (arc_function_type): Change encoding, add naked type.
16642 (ARC_INTERRUPT_P): Change to handle the new encoding.
16643 (ARC_FAST_INTERRUPT_P): Likewise.
16644 (ARC_NORMAL_P): Define.
16645 (ARC_NAKED_P): Likewise.
16646 (arc_compute_function_type): Delete prototype.
16647 * config/arc/arc.md (in_ret_delay_slot): Use
16648 arc_return_address_register function.
16649 (simple_return): Likewise.
16650 (p_return_i): Likewise.
16652 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16654 PR tree-optimization/81428
16655 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16656 can't be built for those types.
16658 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16660 Remove stuff dead since r239246.
16662 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16663 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16664 (avr_inform_devices): Remove dead stuff.
16666 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16668 * config/arm/arm_neon.h: Fix softp typo.
16670 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16672 PR tree-optimization/81365
16673 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16674 aggregate moves onto bb predecessor edges, make sure there are no
16675 loads that could alias the lhs in between the start of bb and the
16678 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16681 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16682 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16683 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16685 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16687 PR tree-optimization/81396
16688 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16689 (init_symbolic_number): Initialize it to 1.
16690 (perform_symbolic_merge): Add n_ops from both operands into the new
16692 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16693 without base_addr as useless if they need more than one operation.
16694 (bswap_replace): Handle !bswap case for NULL base_addr.
16696 2017-07-17 Tom de Vries <tom@codesourcery.com>
16699 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16702 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16704 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16705 conditional builtin define __FIX_LEON3FT_B2BST.
16707 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16709 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16710 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16713 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16715 PR rtl-optimization/81424
16716 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16717 to remove potential trapping from operands if -fnon-call-exceptions.
16719 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16721 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16722 profile_proability for scalling.
16723 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16725 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16727 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16729 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16731 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16732 fixpoint arithmetics.
16734 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16736 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16737 fixpoint arithmetics.
16739 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16741 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16742 fixpoint arithmetics.
16744 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16746 * profile-count.h (profile_probability::from_reg_br_prob_note,
16747 profile_probability::to_reg_br_prob_note): New functions.
16748 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16749 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16750 * predict.c (probability_reliable_p): Update.
16751 (edge_probability_reliable_p): Update.
16752 (br_prob_note_reliable_p): Update.
16753 (invert_br_probabilities): Update.
16754 (add_reg_br_prob_note): New function.
16755 (combine_predictions_for_insn): Update.
16756 * asan.c (asan_clear_shadow): Update.
16757 * cfgbuild.c (compute_outgoing_frequencies): Update.
16758 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16759 (update_br_prob_note): Update.
16760 (rtl_verify_edges): Update.
16761 (purge_dead_edges): Update.
16762 (fixup_reorder_chain): Update.
16763 * emit-rtl.c (try_split): Update.
16764 * ifcvt.c (cond_exec_process_insns): Update.
16765 (cond_exec_process_if_block): Update.
16766 (dead_or_predicable): Update.
16767 * internal-fn.c (expand_addsub_overflow): Update.
16768 (expand_neg_overflow): Update.
16769 (expand_mul_overflow): Update.
16770 * loop-doloop.c (doloop_modify): Update.
16771 * loop-unroll.c (compare_and_jump_seq): Update.
16772 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16773 * predict.h: Update.
16774 * reorg.c (mostly_true_jump): Update.
16776 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16777 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16778 * config/arc/arc.c: (emit_unlikely_jump): Update.
16779 * config/arm/arm.c: (emit_unlikely_jump): Update.
16780 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16781 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16782 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16783 (ix86_print_operand): Update.
16784 (ix86_split_fp_branch): Update.
16785 (predict_jump): Update.
16786 * config/ia64/ia64.c (ia64_print_operand): Update.
16787 * config/mmix/mmix.c (mmix_print_operand): Update.
16788 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16789 (rs6000_expand_split_stack_prologue): Update.
16790 * config/rs6000/rs6000.c: Update.
16791 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16792 (s390_expand_vec_movstr): Update.
16793 (s390_expand_cs_tdsi): Update.
16794 (s390_expand_split_stack_prologue): Update.
16795 * config/sh/sh.c (sh_print_operand): Update.
16796 (expand_cbranchsi4): Update.
16797 (expand_cbranchdi4): Update.
16798 * config/sparc/sparc.c (output_v9branch): Update.
16799 * config/spu/spu.c (get_branch_target): Update.
16800 (ea_load_store_inline): Update.
16801 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16802 * config/tilepro/tilepro.c: Update.
16804 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16806 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16807 (gimplify_save_expr): Likewise.
16809 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16811 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16813 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16815 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16817 * timevar.def (TV_IPA_FNSUMMARY): Define.
16819 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16821 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16822 to back store errata sensitive sequence from being generated.
16823 (sqrtdf2_fix): Likewise.
16825 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16827 * tree-ssa-threadupdate.c (compute_path_counts,
16828 update_joiner_offpath_counts): Use profile_probability.
16830 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16833 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16835 * config/arm/arm-c.c (arm_cpu_builtins): Define
16836 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16838 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16840 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16841 array entries to represent __ieee128 versions of the
16842 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16843 scalar_extract_sig, and scalar_insert_exp built-in functions.
16844 (altivec_resolve_overloaded_builtin): Add special case handling
16845 for the __builtin_scalar_insert_exp function, as represented by
16846 the P9V_BUILTIN_VEC_VSIEDP constant.
16847 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16848 exponent support for __ieee128 argument.
16849 (VSESQP): Add scalar extract signature support for __ieee128
16851 (VSTDCNQP): Add scalar test negative support for __ieee128
16853 (VSIEQP): Add scalar insert exponent support for __int128 argument
16854 with __ieee128 result.
16855 (VSIEQPF): Add scalar insert exponent support for __ieee128
16856 argument with __ieee128 result.
16857 (VSTDCQP): Add scalar test data class support for __ieee128
16859 (VSTDCNQP): Add overload support for scalar test negative with
16860 __ieee128 argument.
16861 (VSTDCQP): Add overload support for scalar test data class
16862 __ieee128 argument.
16863 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16864 UNSPEC_VSX_SXSIGDP.
16865 (UNSPEC_VSX_SIEXPQP): New constant.
16866 (xsxexpqp): New insn for VSX scalar extract exponent quad
16868 (xsxsigqp): New insn for VSX scalar extract significand quad
16870 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16871 precision with floating point argument.
16872 (xststdcqp): New expand for VSX scalar test data class quad
16874 (xststdcnegqp): New expand for VSX scalar test negative quad
16876 (xststdcqp): New insn to match expansions for VSX scalar test data
16877 class quad precision and VSX scalar test negative quad precision.
16878 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16879 special case operand checking to enforce that second operand of
16880 VSX scalar test data class with quad precision argument is a 7-bit
16882 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16883 prototypes and descriptions of __ieee128 versions of
16884 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16885 scalar_test_data_class, and scalar_test_neg built-in functions.
16887 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16889 PR tree-optimization/81162
16890 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16891 replace a negate with an add.
16893 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16895 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16897 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16899 * config/arm/arm-c.c (arm_cpu_builtins): Define
16900 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16902 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16904 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16905 (armv8-r): Set ARM Cortex-R52 as default CPU.
16906 * config/arm/arm-tables.opt: Regenerate.
16907 * config/arm/arm-tune.md: Regenerate.
16908 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16910 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16911 extension for -mcpu=cortex-r52.
16913 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16915 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16916 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16917 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16918 (fp-armv8): Define it as FP_ARMv8 only.
16919 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16920 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16922 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16923 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16924 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16925 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16926 than TARGET_FPU_ARMV8.
16927 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16928 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16929 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16931 * config/arm/neon.md (neon_vrint): Likewise.
16932 (neon_vcvt): Likewise.
16933 (neon_<fmaxmin_op><mode>): Likewise.
16934 (<fmaxmin><mode>3): Likewise.
16935 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16936 * config/arm/predicates.md (arm_cond_move_operator): Check against
16937 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16939 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16941 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16942 to top of function.
16944 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16946 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16947 loop in comment with memset.
16949 2017-07-14 Martin Liska <mliska@suse.cz>
16951 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16952 * dwarf2out.c (is_java): Remove the function.
16953 (output_pubname): Remove usage of the function.
16954 (lower_bound_default): Remove usage of DW_LANG_Java.
16955 (gen_compile_unit_die): Likewise.
16956 * gcc.c: Remove compiler defaults for .java and .zip files.
16957 * gimple-expr.c (remove_suffix): Change as there's no longer
16958 extension than 4-letter one.
16959 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16960 (gimplify_save_expr): Likewise.
16961 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16962 as it's possible even for other languages than Java.
16963 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16964 * lto-opts.c (lto_write_options): Remove reference to Java.
16965 * opts.c (strip_off_ending): Update file extension handling.
16966 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16967 * tree-eh.c (lower_resx): Likewise.
16968 * tree.c (free_lang_data_in_type): Remove dead code.
16969 (find_decls_types_r): Likewise.
16970 (build_common_builtin_nodes): Remove Java from a comment.
16971 (verify_type): Remove dead code.
16972 * varasm.c (assemble_external): Remove Java from a comment.
16974 2017-07-14 Martin Liska <mliska@suse.cz>
16976 * opts.c (finish_options): Add quotes.
16977 (common_handle_option): Likewise.
16979 2017-07-14 Martin Liska <mliska@suse.cz>
16981 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16982 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16983 Remove N_SO_PASCAL.
16984 * dwarf2out.c (lower_bound_default): Do not handle
16986 (gen_compile_unit_die): Likewise.
16987 * gcc.c: Remove default extension binding for GNU Pascal.
16988 * stmt.c: Remove Pascal language from a comment.
16989 * xcoffout.c: Likewise.
16991 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16994 * diagnostic-show-locus.c (fixit_cmp): New function.
16995 (layout::layout): Sort m_fixit_hints.
16996 (column_range::column_range): Assert that the values are valid.
16997 (struct char_span): New struct.
16998 (correction::overwrite): New method.
16999 (struct source_line): New struct.
17000 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17001 calls in terms of classes source_line and char_span, and
17002 correction::overwrite.
17003 (selftest::test_overlapped_fixit_printing_2): New function.
17004 (selftest::diagnostic_show_locus_c_tests): Call it.
17006 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17008 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17009 early if there is no lhs.
17011 2017-07-13 Martin Liska <mliska@suse.cz>
17013 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17014 (gen_reference_type_die): Likewise.
17015 * stor-layout.c: Remove Pascal-related comment.
17017 2017-07-13 Martin Liska <mliska@suse.cz>
17019 * opts.c (finish_options): Add quotes to error messages.
17020 (parse_sanitizer_options): Likewise.
17022 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17024 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17026 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17028 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17030 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17032 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17034 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17036 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17039 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17040 provides the hardware capability bits, define the macro
17041 __BUILTIN_CPU_SUPPORTS__.
17042 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17043 if GLIBC does not provide the hardware capability bits. Add a
17044 gcc_unreachable call if the built-in cpu function is neither
17045 __builtin_cpu_is nor __builtin_cpu_supports.
17046 (rs6000_get_function_versions_dispatcher): Change the warning
17047 that an old GLIBC is used which does not export the capability
17048 bits to be an error.
17049 * doc/extend.texi (target_clones attribute): Document the
17050 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17051 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17052 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17053 the macros defined by GCC if the newer GLIBC is available.
17055 2017-07-12 Jeff Law <law@redhat.com>
17057 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17058 remaining includes slightly.
17059 * config/riscv/riscv-builtins.c: Include profile-count.h.
17061 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17064 * config/avr/avr.c (avr_set_current_function): In diagnostic
17065 messages: Quote keywords and (parts of) identifiers.
17066 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17069 2017-07-12 Carl Love <cel@us.ibm.com>
17071 * config/rs6000/rs6000-c.c: Add support for built-in functions
17072 vector bool char vec_revb (vector bool char);
17073 vector bool short vec_revb (vector short char);
17074 vector bool int vec_revb (vector bool int);
17075 vector bool long long vec_revb (vector bool long long);
17076 * doc/extend.texi: Update the built-in documentation file for the
17077 new built-in functions.
17079 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17081 * config/s390/s390.md: Remove movcc splitter.
17083 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17085 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17086 load/store on condition.
17088 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17091 * config/avr/avr.c (avr_encode_section_info)
17092 [progmem && !TREE_READONLY]: Error if progmem object needs
17095 2017-07-11 Michael Collison <michael.collison@arm.com>
17097 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17100 2017-07-11 Carl Love <cel@us.ibm.com>
17102 * config/rs6000/rs6000-c.c: Add support for builtins
17103 vector unsigned int vec_parity_lsbb (vector signed int);
17104 vector unsigned int vec_parity_lsbb (vector unsigned int);
17105 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17106 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17107 vector unsigned long long vec_parity_lsbb (vector signed long long);
17108 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17109 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17110 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17111 * doc/extend.texi: Update the built-in documentation file for the
17112 new built-in functions.
17114 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17116 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17117 (layout::m_primary_loc): New field.
17118 (layout::layout): Initialize new field. Move location filtering
17119 logic from here to...
17120 (layout::maybe_add_location_range): ...this new method. Add
17121 support for filtering to just the lines already specified by other
17123 (layout::will_show_line_p): New method.
17124 (gcc_rich_location::add_location_if_nearby): New method.
17125 (selftest::test_add_location_if_nearby): New test function.
17126 (selftest::diagnostic_show_locus_c_tests): Call it.
17127 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17130 2017-07-11 Tom de Vries <tom@codesourcery.com>
17132 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17133 (bb_first_real_insn): New function.
17134 (nvptx_single): Add extra initialization of broadcasted condition
17137 2017-07-11 Nathan Sidwell <nathan@acm.org>
17139 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17141 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17143 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17144 Binutils doc as TEXI will mess them up.
17145 * doc/invoke.texi (AVR Options): Same here.
17147 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17149 * config/sparc/sparc.opt (mfix-ut700): New option.
17150 (mfix-gr712rc): Likewise.
17151 (sparc_fix_b2bst): New variable.
17152 * doc/invoke.texi (SPARC options): Document them.
17153 (ARM options): Fix warnings.
17154 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17155 instructions to prevent sequences that can trigger the store-store
17156 errata for certain LEON3FT processors.
17157 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17158 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17159 * config/sparc/sparc.md (fix_b2bst): New attribute.
17160 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17162 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17165 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17167 (*rsqrtsf2_sse): Ditto.
17169 (div<mode>3): Macroize insn from divdf3 and divsf3
17170 using MODEF mode iterator.
17172 2017-07-10 Martin Sebor <msebor@redhat.com>
17174 PR tree-optimization/80397
17175 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17176 instead of testing for equality to INTEGER_TYPE.
17178 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17180 * config.gcc: Remove uclibc from arc target spec.
17182 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17184 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17186 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17189 * lto-wrapper.c (remove_option): New function.
17190 (merge_and_complain): Merge PIC/PIE options more realistically.
17192 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17194 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17198 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17199 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17200 * config.in: Regenerate.
17201 * configure: Regenerate.
17202 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17203 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17204 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17205 (TARGET_GASISR_PROLOGUES): ...target mask.
17206 * common/config/avr/avr-common.c
17207 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17208 Set -mgas-isr-prologues.
17209 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17210 INSERT_PASS_BEFORE for it.
17211 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17212 * config/avr/avr.c (avr_option_override)
17213 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17214 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17215 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17216 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17217 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17219 (make_avr_pass_pre_proep): New function.
17220 (emit_push_sfr) <treg>: Add argument to function and use it
17221 instead of TMP_REG.
17222 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17223 and set machine->gasisr.yes.
17224 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17225 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17226 __gcc_isr.n_pushed to .L__stack_usage.
17227 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17228 (avr_asm_final_postscan_insn): ...this new static function.
17229 * config/avr/avr.h (machine_function)
17230 <is_no_gccisr, use_L__stack_usage>: New fields.
17231 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17232 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17233 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17234 (gasisr, *gasisr): New expander and insn.
17235 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17236 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17237 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17239 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17241 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17244 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17246 Move jump-tables out of .text again.
17249 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17250 (ASM_OUTPUT_ADDR_VEC): New function.
17251 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17252 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17253 INSN_ADDRESSes as asm comment.
17254 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17255 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17256 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17257 * config/avr/avr.md (*tablejump): Adjust comment.
17258 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17259 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17261 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17262 (avr_output_addr_vec): New proto.
17263 (avr_log_t) <insn_addresses>: New field.
17265 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17268 * config/i386/i386.c (ix86_function_arg_advance): Set
17269 outgoing_args_on_stack to true if there are outgoing arguments
17271 (ix86_function_arg): Likewise.
17272 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17273 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17274 * config/i386/i386.h (machine_function): Add
17275 outgoing_args_on_stack.
17277 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17279 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17280 supporting pthreds.
17281 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17283 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17285 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17286 (REAL_H): Remove $(MACHMODE_H).
17287 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17289 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17290 $(MACHMODE_H) and double-int.h.
17291 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17293 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17296 2017-07-07 Andrew Pinski <apinski@cavium.com>
17298 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17299 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17301 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17303 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17304 Add warning if GCC was not configured to link against a GLIBC that
17305 exports the hardware capability bits.
17306 (make_resolver_func): Make resolver function private and not a
17307 COMDAT function. Create the name with clone_function_name instead
17308 of make_unique_name.
17311 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17312 correct operand in doing the split.
17314 2017-07-07 Carl Love <cel@us.ibm.com>
17316 * config/rs6000/rs6000-c: Add support for built-in function
17317 vector unsigned short vec_pack_to_short_fp32 (vector float,
17319 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17320 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17321 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17322 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17323 (convert_4f32_8i16): Add define_expand.
17324 * doc/extend.texi: Update the built-in documentation file for the
17325 new built-in function.
17327 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17329 * config/sparc/m8.md: New file.
17330 * config/sparc/sparc.md: Include m8.md.
17332 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17334 * config/sparc/sparc.opt: New option -mvis4b.
17335 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17336 (sparc_option_override): Handle VIS4B.
17337 (enum sparc_builtins): Define
17338 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17339 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17340 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17341 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17342 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17343 (check_constant_argument): New function.
17344 (sparc_vis_init_builtins): Define builtins
17345 __builtin_vis_dictunpack{8,16,32},
17346 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17347 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17348 __builtin_vis_fpcmpde{8,16,32}shl and
17349 __builtin_vis_fpcmpur{8,16,32}shl.
17350 (sparc_expand_builtin): Check that the constant operands to
17351 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17352 constant and in range.
17353 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17355 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17356 (SPARC_IMM5_P): Likewise.
17357 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17358 (enabled): Handle vis4b.
17359 (UNSPEC_DICTUNPACK): New unspec.
17360 (UNSPEC_FPCMPSHL): Likewise.
17361 (UNSPEC_FPUCMPSHL): Likewise.
17362 (UNSPEC_FPCMPDESHL): Likewise.
17363 (UNSPEC_FPCMPURSHL): Likewise.
17364 (cpu_feature): New CPU feature `vis4b'.
17365 (dictunpack{8,16,32}): New insns.
17366 (FPCSMODE): New mode iterator.
17367 (fpcscond): New code iterator.
17368 (fpcsucond): Likewise.
17369 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17370 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17371 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17372 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17373 * config/sparc/constraints.md: Define constraints `q' for unsigned
17374 2-bit integer constants and `t' for unsigned 5-bit integer
17376 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17378 (imm5_operand_dictunpack16): Likewise.
17379 (imm5_operand_dictunpack32): Likewise.
17380 (imm2_operand): Likewise.
17381 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17382 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17383 ditunpack* and fpcmp*shl builtins.
17385 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17387 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17388 * config.in: Add HAVE_AS_SPARC6 define.
17389 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17391 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17393 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17394 (CPP_CPU_SPEC): Handle m8.
17395 (ASM_CPU_SPEC): Likewise.
17396 * config/sparc/sparc-opts.h (enum processor_type): Add
17398 * config/sparc/sparc.c (m8_costs): New struct.
17399 (sparc_option_override): Handle TARGET_CPU_m8.
17400 (sparc32_initialize_trampoline): Likewise.
17401 (sparc64_initialize_trampoline): Likewise.
17402 (sparc_issue_rate): Likewise.
17403 (sparc_register_move_cost): Likewise.
17404 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17405 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17406 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17407 (CPP_CPU_SPEC): Handle M8.
17408 (ASM_CPU_SPEC): Likewise.
17409 (AS_M8_FLAG): Define.
17410 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17411 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17412 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17414 * configure: Regenerate.
17415 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17418 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17420 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17422 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17423 ("*movdi_insn_sp32"): Do not set v3pipe.
17424 ("*movsi_insn"): Likewise.
17425 ("*movdi_insn_sp64"): Likewise.
17426 ("*movsf_insn"): Likewise.
17427 ("*movdf_insn_sp32"): Likewise.
17428 ("*movdf_insn_sp64"): Likewise.
17429 ("*zero_extendsidi2_insn_sp64"): Likewise.
17430 ("*sign_extendsidi2_insn"): Likewise.
17431 ("*mov<VM32:mode>_insn"): Likewise.
17432 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17433 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17434 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17435 ("<vlop:code><VL:mode>3"): Likewise.
17436 ("*not_<vlop:code><VL:mode>3"): Likewise.
17437 ("*nand<VL:mode>_vis"): Likewise.
17438 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17439 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17440 ("one_cmpl<VL:mode>2"): Likewise.
17441 ("faligndata<VM64:mode>_vis"): Likewise.
17442 ("alignaddrsi_vis"): Likewise.
17443 ("alignaddrdi_vis"): Likweise.
17444 ("alignaddrlsi_vis"): Likewise.
17445 ("alignaddrldi_vis"): Likewise.
17446 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17447 ("bmaskdi_vis"): Likewise.
17448 ("bmasksi_vis"): Likewise.
17449 ("bshuffle<VM64:mode>_vis"): Likewise.
17450 ("cmask8<P:mode>_vis"): Likewise.
17451 ("cmask16<P:mode>_vis"): Likewise.
17452 ("cmask32<P:mode>_vis"): Likewise.
17453 ("pdistn<P:mode>_vis"): Likewise.
17454 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17456 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17458 * config/sparc/sparc.md ("subtype"): New insn attribute.
17459 ("*wrgsr_sp64"): Set insn subtype.
17460 ("*rdgsr_sp64"): Likewise.
17461 ("alignaddrsi_vis"): Likewise.
17462 ("alignaddrdi_vis"): Likewise.
17463 ("alignaddrlsi_vis"): Likewise.
17464 ("alignaddrldi_vis"): Likewise.
17465 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17466 ("fexpand_vis"): Likewise.
17467 ("fpmerge_vis"): Likewise.
17468 ("faligndata<VM64:mode>_vis"): Likewise.
17469 ("bshuffle<VM64:mode>_vis"): Likewise.
17470 ("cmask8<P:mode>_vis"): Likewise.
17471 ("cmask16<P:mode>_vis"): Likewise.
17472 ("cmask32<P:mode>_vis"): Likewise.
17473 ("fchksm16_vis"): Likewise.
17474 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17475 ("fmean16_vis"): Likewise.
17476 ("fp<plusminus_insn>64_vis"): Likewise.
17477 ("<plusminus_insn>v8qi3"): Likewise.
17478 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17479 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17480 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17481 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17482 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17483 ("*movqi_insn"): Likewise.
17484 ("*movhi_insn"): Likewise.
17485 ("*movsi_insn"): Likewise.
17486 ("movsi_pic_gotdata_op"): Likewise.
17487 ("*movdi_insn_sp32"): Likewise.
17488 ("*movdi_insn_sp64"): Likewise.
17489 ("movdi_pic_gotdata_op"): Likewise.
17490 ("*movsf_insn"): Likewise.
17491 ("*movdf_insn_sp32"): Likewise.
17492 ("*movdf_insn_sp64"): Likewise.
17493 ("*zero_extendhisi2_insn"): Likewise.
17494 ("*zero_extendqihi2_insn"): Likewise.
17495 ("*zero_extendqisi2_insn"): Likewise.
17496 ("*zero_extendqidi2_insn"): Likewise.
17497 ("*zero_extendhidi2_insn"): Likewise.
17498 ("*zero_extendsidi2_insn_sp64"): Likewise.
17499 ("ldfsr"): Likewise.
17500 ("prefetch_64"): Likewise.
17501 ("prefetch_32"): Likewise.
17502 ("tie_ld32"): Likewise.
17503 ("tie_ld64"): Likewise.
17504 ("*tldo_ldub_sp32"): Likewise.
17505 ("*tldo_ldub1_sp32"): Likewise.
17506 ("*tldo_ldub2_sp32"): Likewise.
17507 ("*tldo_ldub_sp64"): Likewise.
17508 ("*tldo_ldub1_sp64"): Likewise.
17509 ("*tldo_ldub2_sp64"): Likewise.
17510 ("*tldo_ldub3_sp64"): Likewise.
17511 ("*tldo_lduh_sp32"): Likewise.
17512 ("*tldo_lduh1_sp32"): Likewise.
17513 ("*tldo_lduh_sp64"): Likewise.
17514 ("*tldo_lduh1_sp64"): Likewise.
17515 ("*tldo_lduh2_sp64"): Likewise.
17516 ("*tldo_lduw_sp32"): Likewise.
17517 ("*tldo_lduw_sp64"): Likewise.
17518 ("*tldo_lduw1_sp64"): Likewise.
17519 ("*tldo_ldx_sp64"): Likewise.
17520 ("*mov<VM32:mode>_insn"): Likewise.
17521 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17522 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17524 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17526 * config/sparc/sparc.md ("type"): New insn type viscmp.
17527 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17529 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17530 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17531 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17532 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17534 ("n7_vis_logical_11cycle"): Likewise.
17535 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17536 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17537 * config/sparc/niagara.md ("niag_vis"): Likewise.
17538 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17539 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17541 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17543 * config/sparc/sparc.md: New instruction type `bmask'.
17544 (bmaskdi_vis): Use the `bmask' type.
17545 (bmasksi_vis): Likewise.
17546 * config/sparc/ultra3.md (us3_array): Likewise.
17547 * config/sparc/niagara7.md (n7_array): Likewise.
17548 * config/sparc/niagara4.md (n4_array): Likewise.
17549 * config/sparc/niagara2.md (niag2_vis): Likewise.
17550 (niag3_vis): Likewise.
17551 * config/sparc/niagara.md (niag_vis): Likewise.
17553 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17555 * ipa-comdats.c: Remove optimize check from gate.
17556 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17557 for functions not optimized.
17558 (ipa_fn_summary_read): Skip optimize check.
17559 (ipa_fn_summary_write): Likewise.
17560 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17562 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17564 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17565 (check_callers): Give up on caller which is not optimized.
17566 (inline_small_functions): Likewise.
17567 (ipa_inline): Do not give up when not optimizing.
17568 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17569 away unoptimizes cdtors.
17570 (whole_program_function_and_variable_visibility): Do
17571 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17572 * ipa.c (process_references): Do not check optimize.
17573 (symbol_table::remove_unreachable_nodes): Update optimize check.
17574 (set_writeonly_bit): Update optimize check.
17575 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17576 (pass_ipa_single_use::gate): Remove.
17578 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17580 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17581 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17582 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17583 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17584 permute_load, permute_store, adjust_extract, adjust_splat,
17585 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17586 replace_swap_with_copy, dump_swap_insn_table,
17587 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17588 recombine_lvx_pattern, recombine_stvx_pattern,
17589 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17590 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17591 to file rs6000-p8swap.c.
17592 * config/rs6000/rs6000-p8swap.c: New file.
17593 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17594 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17595 and rs6000*-*-* targets.
17597 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17599 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17601 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17603 * lto-wrapper.c (merge_and_complain): Do not merge
17604 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17605 fsigned_zeros, ftrapping_math, fwrapv.
17606 (append_compiler_options): Do not track these options.
17607 (append_linker_options): Likewie
17609 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17611 * cgraphunit.c (cgraph_node::finalize_function): When
17612 !flag_toplevel_reorde set no_reorder flag.
17613 (varpool_node::finalize_decl): Likewise.
17614 (symbol_table::compile): Drop no toplevel reorder path.
17616 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17618 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17619 edges; zero probability is not better than uninitialized.
17621 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17623 * asan.h (asan_sanitize_allocas_p): Declare.
17624 * asan.c (asan_sanitize_allocas_p): New function.
17625 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17626 (handle_builtin_alloca): Likewise.
17627 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17628 if !asan_sanitize_allocas_p.
17629 * params.def (asan-instrument-allocas): Add new option.
17630 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17631 * opts.c (common_handle_option): Disable allocas sanitization for
17634 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17636 * asan.c: Include gimple-fold.h.
17637 (get_last_alloca_addr): New function.
17638 (handle_builtin_stackrestore): Likewise.
17639 (handle_builtin_alloca): Likewise.
17640 (asan_emit_allocas_unpoison): Likewise.
17641 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17642 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17643 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17644 (instrument_builtin_call): Pass gimple iterator to
17645 get_mem_refs_of_builtin_call.
17646 (last_alloca_addr): New global.
17647 * asan.h (asan_emit_allocas_unpoison): Declare.
17648 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17649 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17650 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17651 if function calls alloca.
17652 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17653 * gimple-fold.h (replace_call_with_value): Declare.
17654 * internal-fn.c: Include asan.h.
17655 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17656 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17658 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17660 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17661 (C_SELFTEST_FLAGS): New.
17662 (CPP_SELFTEST_FLAGS): New.
17663 (SELFTEST_DEPS): New, from deps of s-selftest.
17664 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17665 (CPP_SELFTEST_DEPS): New.
17666 (selftest): Add dependency on s-selftest-c++.
17667 (s-selftest): Rename to...
17668 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17669 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17670 than SELFTEST_FLAGS.
17671 (selftest-gdb): Rename to...
17672 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17674 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17675 (selftest-valgrind): Rename to...
17676 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17678 (selftest-valgrind): Reintroduce as an alias for
17679 selftest-c-valgrind.
17680 (s-selftest-c++): New.
17681 (selftest-c++-gdb): New.
17682 (selftest-c++-valgrind): New.
17684 2017-07-06 Olivier Hainque <hainque@adacore.com>
17686 * gcc.c (process_command): When deciding if undefined variables
17687 should be ignored when processing specs, accept "gcc -v" as well.
17689 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17691 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17692 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17694 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17696 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17697 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17698 * config/arm/arm-tables.opt: Regenerate.
17699 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17701 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17703 2017-07-06 Carl Love <cel@us.ibm.com>
17705 * ChangeLog: Clean up from mid air collision
17707 2017-07-06 Carl Love <cel@us.ibm.com>
17709 * config/rs6000/rs6000-c.c: Add support for built-in functions
17710 vector signed int vec_subc (vector signed int, vector signed int);
17711 vector signed __int128 vec_subc (vector signed __int128,
17712 vector signed __int128);
17713 vector unsigned __int128 vec_subc (vector unsigned __int128,
17714 vector unsigned __int128);
17715 vector signed int vec_sube (vector signed int, vector signed int,
17716 vector signed int);
17717 vector unsigned int vec_sube (vector unsigned int,
17718 vector unsigned int,
17719 vector unsigned int);
17720 vector signed __int128 vec_sube (vector signed __int128,
17721 vector signed __int128,
17722 vector signed__int128);
17723 vector unsigned __int128 vec_sube (vector unsigned __int128,
17724 vector unsigned __int128,
17725 vector unsigned __int128);
17726 vector signed int vec_subec (vector signed int, vector signed int,
17727 vector signed int);
17728 vector unsigned int vec_subec (vector unsigned int,
17729 vector unsigned int,
17730 vector unsigned int);
17731 vector signed __int128 vec_subec (vector signed __int128,
17732 vector signed __int128,
17733 vector signed__int128);
17734 vector unsigned __int128 vec_subec (vector unsigned __int128,
17735 vector unsigned __int128,
17736 vector unsigned __int128);
17737 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17738 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17739 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17740 BU_ALTIVEC_OVERLOAD_X definitions.
17741 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17742 * doc/extend.texi: Update the built-in documentation file for the new
17743 built-in functions.
17745 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17748 * diagnostic-show-locus.c (layout::layout): Use start and finish
17749 spelling location for the start and finish of each range.
17750 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17751 Add unused aspect param.
17752 * input.c (expand_location_1): Add "aspect" param, and use it
17753 to access the correct part of the location.
17754 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17756 (expand_location_to_spelling_point): Likewise.
17757 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17758 param, and pass it to expand_location_1.
17760 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17762 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17763 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17764 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17765 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17766 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17767 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17768 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17769 _mm_maskz_getmant_ss): New intrinsics.
17770 (__builtin_ia32_getexpss128_mask): Changed to ...
17771 __builtin_ia32_getexpss128_round ... this.
17772 (__builtin_ia32_getexpsd128_mask): Changed to ...
17773 __builtin_ia32_getexpsd128_round ... this.
17774 * config/i386/i386-builtin-types.def
17775 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17776 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17777 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17778 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17779 __builtin_ia32_getmantss_mask_round): New builtins.
17780 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17781 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17782 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17783 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17784 * config/i386/sse.md
17785 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17786 avx512f_sgetexp<mode><mask_scalar_name>
17787 <round_saeonly_scalar_name> ... this.
17788 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17789 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17790 vgetexp<ssescalarmodesuffix>
17791 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17792 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17793 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17794 avx512f_vgetmant<mode><mask_scalar_name>
17795 <round_saeonly_scalar_name> ... this.
17796 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17797 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17798 vgetmant<ssescalarmodesuffix>
17799 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17800 %0<mask_scalar_operand4>, %1, %2
17801 <round_saeonly_scalar_mask_op4>, %3} ... this.
17802 * config/i386/subst.md (mask_scalar_operand4,
17803 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17804 round_saeonly_scalar_nimm_predicate): New subst attributes.
17806 2017-07-06 Julia Koval <julia.koval@intel.com>
17808 * config/i386/i386.c (ix86_erase_embedded_rounding):
17809 Remove code for old rounding pattern.
17811 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17813 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17815 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17817 * doc/sourcebuild.texi (Test Directives, Variants of
17818 dg-require-support): Add documentation for dg-require-stack-check.
17820 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17822 * config/i386/subst.md (mask_scalar, round_scalar,
17823 round_saeonly_scalar): New meta-templates.
17824 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17825 round_scalar_mask_operand3, round_scalar_mask_op3,
17826 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17827 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17828 round_saeonly_scalar_constraint,
17829 round_saeonly_scalar_prefix): New subst attribute.
17830 * config/i386/sse.md
17831 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17832 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17833 <round_scalar_name> ... this.
17834 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17835 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17836 <round_scalar_name> ... this.
17837 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17838 <sse>_vm<code><mode>3<mask_scalar_name>
17839 <round_saeonly_scalar_name> ... this.
17840 (v<plusminus_mnemonic><ssescalarmodesuffix>
17841 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17842 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17843 v<plusminus_mnemonic><ssescalarmodesuffix>
17844 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17845 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17846 (v<multdiv_mnemonic><ssescalarmodesuffix>
17847 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17848 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17849 v<multdiv_mnemonic><ssescalarmodesuffix>
17850 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17851 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17852 (v<maxmin_float><ssescalarmodesuffix>
17853 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17854 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17855 v<maxmin_float><ssescalarmodesuffix>
17856 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17857 %0<mask_scalar_operand3>, %1, %<iptr>2
17858 <round_saeonly_scalar_mask_op3>} ... this.
17860 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17862 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17863 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17865 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17866 Alan Hayward <alan.hayward@arm.com>
17867 David Sherwood <david.sherwood@arm.com>
17869 * combine.c (simplify_if_then_else): Remove "enum" before
17871 * compare-elim.c (can_eliminate_compare): Likewise.
17872 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17874 (aarch64_lookup_simd_builtin_type): Likewise.
17875 (aarch64_simd_builtin_type): Likewise.
17876 (aarch64_init_simd_builtin_types): Likewise.
17877 (aarch64_simd_expand_args): Likewise.
17878 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17880 (aarch64_reverse_mask): Likewise.
17881 (aarch64_simd_emit_reg_reg_move): Likewise.
17882 (aarch64_gen_adjusted_ldpstp): Likewise.
17883 (aarch64_ccmp_mode_to_code): Likewise.
17884 (aarch64_operands_ok_for_ldpstp): Likewise.
17885 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17886 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17888 (aarch64_min_divisions_for_recip_mul): Likewise.
17889 (aarch64_reassociation_width): Likewise.
17890 (aarch64_get_condition_code_1): Likewise.
17891 (aarch64_simd_emit_reg_reg_move): Likewise.
17892 (aarch64_simd_attr_length_rglist): Likewise.
17893 (aarch64_reverse_mask): Likewise.
17894 (aarch64_operands_ok_for_ldpstp): Likewise.
17895 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17896 (aarch64_gen_adjusted_ldpstp): Likewise.
17897 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17899 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17900 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17901 (arm_lookup_simd_builtin_type): Likewise.
17902 (arm_simd_builtin_type): Likewise.
17903 (arm_init_simd_builtin_types): Likewise.
17904 (arm_expand_builtin_args): Likewise.
17905 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17906 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17907 (ft32_setup_incoming_varargs): Likewise.
17908 (ft32_function_arg): Likewise.
17909 (ft32_function_arg_advance): Likewise.
17910 (ft32_pass_by_reference): Likewise.
17911 (ft32_arg_partial_bytes): Likewise.
17912 (ft32_valid_pointer_mode): Likewise.
17913 (ft32_addr_space_pointer_mode): Likewise.
17914 (ft32_addr_space_legitimate_address_p): Likewise.
17915 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17917 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17918 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17919 (iamcu_alignment): Likewise.
17920 (canonicalize_vector_int_perm): Likewise.
17921 (ix86_noce_conversion_profitable_p): Likewise.
17922 (ix86_mpx_bound_mode): Likewise.
17923 (ix86_operands_ok_for_move_multiple): Likewise.
17924 * config/microblaze/microblaze-protos.h
17925 (microblaze_expand_conditional_branch_reg): Likewise.
17926 * config/microblaze/microblaze.c
17927 (microblaze_expand_conditional_branch_reg): Likewise.
17928 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17930 (rs6000_reassociation_width): Likewise.
17931 (rs6000_invalid_binary_op): Likewise.
17932 (fusion_p9_p): Likewise.
17933 (emit_fusion_p9_load): Likewise.
17934 (emit_fusion_p9_store): Likewise.
17935 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17937 (riscv_hard_regno_mode_ok_p): Likewise.
17938 (riscv_address_insns): Likewise.
17939 (riscv_split_symbol): Likewise.
17940 (riscv_legitimize_move): Likewise.
17941 (riscv_function_value): Likewise.
17942 (riscv_hard_regno_nregs): Likewise.
17943 (riscv_expand_builtin): Likewise.
17944 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17945 (riscv_build_integer): Likewise.
17946 (riscv_split_integer): Likewise.
17947 (riscv_legitimate_constant_p): Likewise.
17948 (riscv_cannot_force_const_mem): Likewise.
17949 (riscv_regno_mode_ok_for_base_p): Likewise.
17950 (riscv_valid_base_register_p): Likewise.
17951 (riscv_valid_offset_p): Likewise.
17952 (riscv_valid_lo_sum_p): Likewise.
17953 (riscv_classify_address): Likewise.
17954 (riscv_legitimate_address_p): Likewise.
17955 (riscv_address_insns): Likewise.
17956 (riscv_load_store_insns): Likewise.
17957 (riscv_force_binary): Likewise.
17958 (riscv_split_symbol): Likewise.
17959 (riscv_force_address): Likewise.
17960 (riscv_legitimize_address): Likewise.
17961 (riscv_move_integer): Likewise.
17962 (riscv_legitimize_const_move): Likewise.
17963 (riscv_legitimize_move): Likewise.
17964 (riscv_address_cost): Likewise.
17965 (riscv_subword): Likewise.
17966 (riscv_output_move): Likewise.
17967 (riscv_canonicalize_int_order_test): Likewise.
17968 (riscv_emit_int_order_test): Likewise.
17969 (riscv_function_arg_boundary): Likewise.
17970 (riscv_pass_mode_in_fpr_p): Likewise.
17971 (riscv_pass_fpr_single): Likewise.
17972 (riscv_pass_fpr_pair): Likewise.
17973 (riscv_get_arg_info): Likewise.
17974 (riscv_function_arg): Likewise.
17975 (riscv_function_arg_advance): Likewise.
17976 (riscv_arg_partial_bytes): Likewise.
17977 (riscv_function_value): Likewise.
17978 (riscv_pass_by_reference): Likewise.
17979 (riscv_setup_incoming_varargs): Likewise.
17980 (riscv_print_operand): Likewise.
17981 (riscv_elf_select_rtx_section): Likewise.
17982 (riscv_save_restore_reg): Likewise.
17983 (riscv_for_each_saved_reg): Likewise.
17984 (riscv_register_move_cost): Likewise.
17985 (riscv_hard_regno_mode_ok_p): Likewise.
17986 (riscv_hard_regno_nregs): Likewise.
17987 (riscv_class_max_nregs): Likewise.
17988 (riscv_memory_move_cost): Likewise.
17989 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17990 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17991 (rl78_addr_space_address_mode): Likewise.
17992 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17994 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17995 (rs6000_reassociation_width): Likewise.
17996 (rs6000_invalid_binary_op): Likewise.
17997 (fusion_p9_p): Likewise.
17998 (emit_fusion_p9_load): Likewise.
17999 (emit_fusion_p9_store): Likewise.
18000 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18001 (ok_for_simple_move_operands): Likewise.
18002 (ok_for_simple_move_strict_operands): Likewise.
18003 (ok_for_simple_arith_logic_operands): Likewise.
18004 (visium_legitimize_reload_address): Likewise.
18005 (visium_select_cc_mode): Likewise.
18006 (output_cbranch): Likewise.
18007 (visium_split_double_move): Likewise.
18008 (visium_expand_copysign): Likewise.
18009 (visium_expand_int_cstore): Likewise.
18010 (visium_expand_fp_cstore): Likewise.
18011 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18012 (visium_function_arg): Likewise.
18013 (visium_function_arg_advance): Likewise.
18014 (visium_libcall_value): Likewise.
18015 (visium_setup_incoming_varargs): Likewise.
18016 (visium_legitimate_constant_p): Likewise.
18017 (visium_legitimate_address_p): Likewise.
18018 (visium_legitimize_address): Likewise.
18019 (visium_secondary_reload): Likewise.
18020 (visium_register_move_cost): Likewise.
18021 (visium_memory_move_cost): Likewise.
18022 (prepare_move_operands): Likewise.
18023 (ok_for_simple_move_operands): Likewise.
18024 (ok_for_simple_move_strict_operands): Likewise.
18025 (ok_for_simple_arith_logic_operands): Likewise.
18026 (visium_function_value_1): Likewise.
18027 (rtx_ok_for_offset_p): Likewise.
18028 (visium_legitimize_reload_address): Likewise.
18029 (visium_split_double_move): Likewise.
18030 (visium_expand_copysign): Likewise.
18031 (visium_expand_int_cstore): Likewise.
18032 (visium_expand_fp_cstore): Likewise.
18033 (visium_split_cstore): Likewise.
18034 (visium_select_cc_mode): Likewise.
18035 (visium_split_cbranch): Likewise.
18036 (output_cbranch): Likewise.
18037 (visium_print_operand_address): Likewise.
18038 * expmed.c (flip_storage_order): Likewise.
18039 * expmed.h (emit_cstore): Likewise.
18040 (flip_storage_order): Likewise.
18041 * genrecog.c (validate_pattern): Likewise.
18042 * hsa-gen.c (gen_hsa_addr): Likewise.
18043 * internal-fn.c (expand_arith_overflow): Likewise.
18044 * ira-color.c (allocno_copy_cost_saving): Likewise.
18045 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18046 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18047 (process_invariant_for_inheritance): Likewise.
18048 * lra-eliminations.c (move_plus_up): Likewise.
18049 * omp-low.c (lower_oacc_reductions): Likewise.
18050 * simplify-rtx.c (simplify_subreg): Likewise.
18051 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18052 (TARGET_CHKP_BOUND_MODE): Likewise..
18053 * targhooks.c (default_chkp_bound_mode): Likewise.
18054 (default_setup_incoming_vararg_bounds): Likewise.
18055 * targhooks.h (default_chkp_bound_mode): Likewise.
18056 (default_setup_incoming_vararg_bounds): Likewise.
18057 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18058 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18059 (have_whole_vector_shift): Likewise.
18060 * tree-vect-stmts.c (vectorizable_load): Likewise.
18061 * doc/tm.texi: Regenerate.
18063 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18065 Graceful degrade if Binutils PR21472 is not available.
18068 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18069 .rodata in flash test fails.
18070 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18071 * confgure: Regenerate.
18072 * config.in: Regenerate.
18073 * config/avr/avr.c (avr_asm_named_section)
18074 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18075 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18076 (avr_asm_init_sections): Same.
18078 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18080 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18081 (fma<VH:mode>4_intrinsic): Likewise.
18082 (*fmsub<VCVTF:mode>4): Likewise.
18083 (*fmsub<VH:mode>4_intrinsic): Likewise.
18085 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18088 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18089 Don't depend on "optimize > 0".
18090 (out_movhi_r_mr, out_movqi_mr_r): Same.
18091 (out_movhi_mr_r, out_movqi_r_mr): Same.
18092 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18093 io_address_operand on "optimize > 0".
18095 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18097 * tree-loop-distribution.c: Add general explanantion on the pass.
18098 (generate_loops_for_partition): Mark distributed loop.
18099 (pg_add_dependence_edges): New parameter. Handle alias data
18100 dependence specially and record it in the parameter if asked.
18101 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18102 (init_partition_graph_vertices, add_partition_graph_edge): New.
18103 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18104 (free_partition_graph_vdata, build_partition_graph): New.
18105 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18106 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18107 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18108 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18109 (version_for_distribution_p, finalize_partitions): New.
18110 (distribute_loop): Handle alias data dependence specially. Factor
18111 out loop fusion code as functions and call these functions.
18113 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18115 * tree-loop-distribution.c (classify_partition): New parameter and
18116 better handle reduction statement.
18117 (rdg_build_partitions): Revise comment.
18118 (distribute_loop): Compute statements in all partitions and pass it
18119 to classify_partition.
18121 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18123 * tree-loop-distribution.c (enum partition_type): New.
18124 (struct partition): New field type.
18125 (partition_merge_into): Add parameter. Update partition type.
18126 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18127 (build_rdg_partition_for_vertex): Compute partition type.
18128 (rdg_build_partitions): Dump partition type.
18129 (distribute_loop): Update calls to partition_merge_into.
18131 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18133 * tree-loop-distribution.c (struct ddr_hasher): New.
18134 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18136 (classify_partition): Call get_data_dependence.
18137 (pg_add_dependence_edges): Ditto.
18138 (distribute_loop): Release data dependence hash table.
18140 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18142 * tree-loop-distribution.c (ref_base_address): Delete.
18143 (similar_memory_accesses): Rename ...
18144 (share_memory_accesses): ... to this. Check if partitions access
18145 the same memory reference.
18146 (distribute_loop): Call share_memory_accesses.
18148 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18150 * tree-loop-distribution.c (struct partition): New field recording
18151 its data reference.
18152 (partition_alloc, partition_free): Init and release data refs.
18153 (partition_merge_into): Merge data refs.
18154 (build_rdg_partition_for_vertex): Collect data refs for partition.
18155 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18157 (distribute_loop): Remve data refs from vertice data of partition
18160 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18162 * tree-loop-distribution.c (params.h): Include header file.
18163 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18164 (datarefs_vec): New global var.
18165 (create_rdg_vertices): Use datarefs_vec directly.
18166 (free_rdg): Don't free data references.
18167 (build_rdg): Update use. Don't free data references.
18168 (distribute_loop): Compute global variable for data references.
18169 Bail out if there are too many data references.
18171 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18173 * tree-loop-distribution.c (loop_nest): New global var.
18174 (build_rdg): Use loop directly, rather than loop nest.
18175 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18177 (distribute_loop): Compute global variable loop nest. Update use.
18179 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18181 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18182 (partition_merge_into): New parameter. Dump reason for fusion.
18183 (distribute_loop): Update use of partition_merge_into.
18185 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18187 * tree-loop-distribution.c (bb_top_order_index): New.
18188 (bb_top_order_index_size, bb_top_order_cmp): New.
18189 (stmts_from_loop): Use topological order.
18190 (pass_loop_distribution::execute): Compute and release topological
18191 order for basic blocks.
18193 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18195 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18198 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18200 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18201 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18202 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18203 * internal-fn.def (LOOP_DIST_ALIAS): New.
18204 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18205 (fold_loop_internal_call): ... this.
18206 (vect_loop_dist_alias_call): New function.
18207 (set_uid_loop_bbs): Call fold_loop_internal_call.
18208 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18211 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18214 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18215 Require dead FLAGS_REG at the beginning of a peephole.
18217 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18220 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18221 arguments in the call to __builtin_ia32_sbb_u32.
18222 (_subborrow_u64): Swap _X and _Y arguments in the call to
18223 __builtin_ia32_sbb_u64.
18225 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18228 * tree-vrp.c (compare_assert_loc): Turn into a function template
18229 with stable template parameter. Only test if a->e is NULL,
18230 !a->e == !b->e has been verified already. Use e == NULL or
18231 e != NULL instead of e or ! e tests. If stable is true, don't use
18232 iterative_hash_expr, on the other side allow a or b or both NULL
18233 and sort the NULLs last.
18234 (process_assert_insertions): Sort using compare_assert_loc<false>
18235 instead of compare_assert_loc, later sort using
18236 compare_assert_loc<true> before calling process_assert_insertions_for
18237 in a loop. Use break instead of continue once seen NULL pointer.
18239 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18241 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18242 Cortex-R7 and Cortex-R8 processors.
18244 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18246 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18247 uninitialized while src is not.
18249 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18251 * common/config/arm/arm-common.c: Adjust include path for
18253 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18254 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18255 (arm-cpu-data.h): Likewise.
18256 (arm-cpu-cdata.h): Likewise.
18257 * config/arm/arm-cpu.h: Delete.
18258 * config/arm/arm-cpu-cdata.h: Delete.
18259 * config/arm/arm-cpu-data.h: Delete.
18261 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18263 * config/arm/arm-cpus.in (cortex-a55): New.
18264 (cortex-a75): Likewise.
18265 (cortex-a75.cortex-a55): Likewise.
18266 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18268 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18269 * config/arm/arm-cpu-cdata.h: Regenerate.
18270 * config/arm/arm-cpu-data.h: Regenerate.
18271 * config/arm/arm-cpu.h: Regenerate.
18272 * config/arm/arm-tables.opt: Regenerate.
18273 * config/arm/arm-tune.md: Regenerate.
18275 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18277 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18279 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18281 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18284 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18286 PR tree-optimization/81292
18287 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18288 full_string_p, also call adjust_related_strinfos if the adjustment
18289 is simple, otherwise invalidate related strinfos.
18291 2017-07-04 Martin Liska <mliska@suse.cz>
18294 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18295 newly created variable as DECL_IGNORED_P.
18297 2017-07-04 Martin Liska <mliska@suse.cz>
18300 * ipa-inline.c (inline_small_functions):
18301 Use xstrdup_for_dump.
18303 2017-07-04 Tom de Vries <tom@codesourcery.com>
18305 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18307 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18310 * config/darwin.c (darwin_function_switched_text_sections):
18313 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18315 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18317 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18319 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18321 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18323 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18324 min_profitable_iters, and th as inclusive lower bounds.
18325 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18326 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18327 for min_profitable_iters and min_profitable_estimate.
18328 (vect_transform_loop): Treat th as an inclusive lower bound.
18329 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18331 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18334 * config/darwin.c (darwin_function_switched_text_sections):
18335 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18336 in two pieces, and suppress the use of buf.
18338 2017-07-03 Nathan Sidwell <nathan@acm.org>
18340 * hash-table.h (hash_table_mod1): Fix indentation.
18342 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18344 PR middle-end/81290
18345 * predict.c (force_edge_cold): Be more careful about propagation
18347 * profile-count.h (profile_probability::guessed,
18348 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18350 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18352 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18354 * doc/invoke.texi (rcpc architecture extension): Document it.
18356 2017-07-03 Richard Biener <rguenther@suse.de>
18358 PR tree-optimization/60510
18359 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18360 the scalar reduction PHI and use it.
18361 (vectorizable_reduction): Properly guard the single_defuse_cycle
18362 path for non-SLP reduction chains where we cannot use it.
18363 Rework reduc_def/index and vector type deduction. Rework
18364 vector operand gathering during reduction op code-gen.
18365 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18366 chains dissolve the chain and leave it to non-SLP reduction
18369 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18371 * tree-data-ref.h (dr_alignment): Declare.
18372 * tree-data-ref.c (dr_alignment): New function.
18373 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18374 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18376 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18378 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18380 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18381 and base_misalignment fields.
18382 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18383 * tree-data-ref.c: Include builtins.h.
18384 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18385 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18386 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18387 * tree-vect-data-refs.c: Include tree-cfg.h.
18388 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18389 fields instead of calculating an alignment here.
18390 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18391 innermost_loop_behavior fields.
18393 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18395 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18397 (DR_STEP_ALIGNMENT): New macro.
18398 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18399 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18400 (create_data_ref): Print it.
18401 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18402 to tell whether the step preserves vector (mis)alignment.
18403 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18404 Move the check for an integer step and generalise to all INTEGER_CST.
18405 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18406 Print the outer step alignment.
18408 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18410 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18411 with offset_alignment.
18412 (DR_ALIGNED_TO): Delete.
18413 (DR_OFFSET_ALIGNMENT): New macro.
18414 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18415 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18416 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18417 (create_data_ref): Likewise.
18418 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18419 (vect_analyze_data_refs): Likewise.
18420 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18421 creating dummy innermost behavior.
18423 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18425 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18426 with a "innermost_loop_behavior *" and refeence tree.
18427 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18428 (create_data_ref): Update call accordingly.
18429 * tree-predcom.c (find_looparound_phi): Likewise.
18431 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18433 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18434 fields with dr_wrt_vec_loop.
18435 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18436 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18437 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18438 (vect_dr_behavior): New function.
18439 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18440 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18441 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18442 track whether the step preserves the misalignment.
18443 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18444 Use vect_dr_behavior.
18445 (vect_setup_realignment): Update call accordingly.
18446 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18447 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18448 call to vect_create_addr_base_for_vector_ref.
18449 (vect_create_cond_for_align_checks): Likewise.
18450 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18451 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18452 (vect_recog_mask_conversion_pattern): Likewise.
18453 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18454 (new_stmt_vec_info): Remove redundant zeroing.
18456 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18458 * common/config/arm/arm-common.c (arm_be8_option): New function.
18459 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18460 (ISA_ARMv6): Add isa_bit_be8.
18461 * config/arm/arm.h (arm_be8_option): Add prototype.
18462 (BE8_SPEC_FUNCTION): New define.
18463 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18464 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18465 (mlittle-endian): Similarly.
18466 (mbe8, mbe32): New options.
18467 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18468 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18470 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18472 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18474 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18476 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18477 (cleanup_tree_cfg_bb): Use it.
18478 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18480 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18482 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18485 * loop-doloop.c (add_test): Update profile.
18487 2017-07-03 Martin Liska <mliska@suse.cz>
18490 * sanopt.c (rewrite_usage_of_param): New function.
18491 (sanitize_rewrite_addressable_params): Likewise.
18492 (pass_sanopt::execute): Call rewrite_usage_of_param.
18494 2017-07-03 Richard Biener <rguenther@suse.de>
18496 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18497 back to using VIEW_CONVERT_EXPR.
18499 2017-07-03 Martin Liska <mliska@suse.cz>
18502 * doc/extend.texi: Document when a resolver function is
18503 generated for target_clones.
18505 2017-07-03 Martin Liska <mliska@suse.cz>
18507 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18508 and shadow memory of auto variables which are subject of
18509 use-after-scope sanitization.
18510 (asan_expand_mark_ifn): Add do set only when is_poison.
18512 2016-07-03 Richard Biener <rguenther@suse.de>
18514 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18516 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18517 (vectorizable_reduction): Perform reduction PHI creation when
18518 visiting a reduction PHI and adjust and simplify code generation
18519 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18520 (vect_transform_loop): Visit reduction PHIs.
18521 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18522 defs into the SLP tree.
18523 (vect_build_slp_tree): Reduction defs terminate the recursion.
18524 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18526 (vect_get_vec_defs_for_stmt_copy): Export.
18527 (vect_get_vec_defs): Likewise.
18528 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18530 (vect_get_vec_defs_for_stmt_copy): Declare.
18531 (vect_get_vec_defs): Likewise.
18533 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18535 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18536 parameter with a "loop" parameter and use it instead of the
18537 loop containing DR_STMT. Don't check simple_iv when doing
18538 BB analysis. Describe the two analysis modes in the comment.
18540 2017-07-03 Tom de Vries <tom@codesourcery.com>
18542 PR tree-optimization/69468
18543 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18544 (find_same_succ_bb): Handle ignore_edge_flags.
18546 2017-07-03 Tom de Vries <tom@codesourcery.com>
18548 PR tree-optimization/81192
18549 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18551 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18553 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18555 2017-07-03 Tom de Vries <tom@codesourcery.com>
18557 PR tree-optimization/81192
18558 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18559 BB_SAME_SUCC (bb) == NULL.
18561 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18563 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18566 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18568 * dumpfile.c: Include profile-count.h
18569 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18571 (insert_cond_bb): Update profile.
18572 * tree-cfg.h (insert_cond_bb): Update prototype.
18573 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18574 * tree-dump.c: Do not include tree-cfg.
18576 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18578 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18580 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18582 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18585 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18587 * tree-complex.c (expand_complex_div_wide): update profile.
18589 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18590 Alan Hayward <alan.hayward@arm.com>
18591 David Sherwood <david.sherwood@arm.com>
18593 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18594 (CORETYPES_H): New define.
18595 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18596 (insn-modes-inline.h, s-modes-inline-h): New rules.
18597 (generated_files): Add insn-modes-inline.h.
18598 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18599 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18600 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18601 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18602 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18603 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18604 (build/gencodes.o, build/genconditions.o): Likewise.
18605 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18606 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18607 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18608 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18609 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18610 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18611 * coretypes.h: Include everything up to real.h for generators.
18612 Include insn-modes.h first. Include wide-int-print.h after
18613 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18614 * machmode.h: Don't include insn-modes.h here.
18615 * function-tests.c: Remove includes of signop.h, machmode.h,
18616 double-int.h and wide-int.h.
18618 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18620 * optc-save-gen.awk: Likewise.
18621 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18622 * godump.c: Remove include of wide-int-print.h.
18623 * pretty-print.h: Likewise.
18624 * wide-int-print.cc: Likewise.
18625 * wide-int.cc: Likewise.
18626 * hash-map-tests.c: Remove include of signop.h.
18627 * hash-set-tests.c: Likewise.
18628 * rtl-tests.c: Likewise.
18629 * mkconfig.sh: Remove include of machmode.h.
18630 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18632 (emit_insn_modes_inline_h): ...this new function. Emit the code
18633 into an insn-modes-inline.h header file, adding appropriate
18634 include guards and end comments.
18635 (emit_insn_modes_c_header): Remove include of machmode.h.
18636 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18638 (main): Handle -i flag and call emit_insn_modes_inline_h when
18641 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18643 * tree-ssa-strlen.c (strinfo): Rename the length field to
18644 nonzero_chars. Add a full_string_p field.
18645 (compare_nonzero_chars, zero_length_string_p): New functions.
18646 (get_addr_stridx): Add an offset_out parameter.
18647 Use compare_nonzero_chars.
18648 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18649 (new_strinfo): Update after above changes to strinfo.
18650 (set_endptr_and_length): Set full_string_p.
18651 (get_string_length): Update after above changes to strinfo.
18652 (unshare_strinfo): Update call to new_strinfo.
18653 (maybe_invalidate): Likewise.
18654 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18655 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18656 as a uhwi instead of an shwi. Update after above changes to
18657 strinfo and new_strinfo.
18658 (zero_length_string): Assert that chainsi contains full strings.
18659 Use zero_length_string_p. Update call to new_strinfo.
18660 (adjust_related_strinfos): Update after above changes to strinfo.
18661 Copy full_string_p from origsi.
18662 (adjust_last_stmt): Use zero_length_string_p.
18663 (handle_builtin_strlen): Update after above changes to strinfo and
18664 new_strinfo. Install the lhs as the string length if the previous
18665 entry didn't describe a full string.
18666 (handle_builtin_strchr): Update after above changes to strinfo
18668 (handle_builtin_strcpy): Likewise.
18669 (handle_builtin_strcat): Likewise.
18670 (handle_builtin_malloc): Likewise.
18671 (handle_pointer_plus): Likewise.
18672 (handle_builtin_memcpy): Likewise. Track nonzero characters
18673 that aren't necessarily followed by a nul terminator.
18674 (handle_char_store): Likewise.
18676 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18678 PR tree-optimization/80769
18679 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18680 for malloc and calloc. Document the new invariant that all related
18681 strinfos have delayed lengths or none do.
18682 (verify_related_strinfos): Move earlier in file.
18683 (set_endptr_and_length): New function, split out from...
18684 (get_string_length): ...here. Also set the lengths of related
18686 (zero_length_string): Assert that chainsi has known (rather than
18688 (adjust_related_strinfos): Likewise.
18690 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18692 PR tree-optimization/81136
18693 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18694 assert that two references with the same misalignment have the same
18695 compile-time misalignment if those compile-time misalignments
18698 2017-07-01 Andi Kleen <ak@linux.intel.com>
18700 * print-tree.c (print_node): Print all attributes.
18702 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18704 * cfg.c (scale_bbs_frequencies): New function.
18705 * cfg.h (scale_bbs_frequencies): Declare it.
18706 * cfgloopanal.c (single_likely_exit): Cleanup.
18707 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18709 (scale_loop_profile): Likewise.
18710 (loop_version): Likewise.
18711 (create_empty_loop_on_edge): Update.
18712 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18713 scale_loop_frequencies, scale_loop_profile, loopify,
18714 loop_version): Update prototypes.
18715 * modulo-sched.c (sms_schedule): Update.
18716 * predict.c (unlikely_executed_edge_p): Also check probability.
18717 (probably_never_executed_edge_p): Fix typo.
18718 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18719 * tree-parloops.c (gen_parallel_loop): Update.
18720 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18721 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18722 * tree-ssa-loop-split.c (split_loop): Update.
18723 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18724 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18725 (vect_loop_versioning): Update.
18726 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18728 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18730 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18732 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18734 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18735 to keep profile consistent.
18737 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18739 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18740 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18741 * profile-count.h (max_safe_multiplier): Make unsigned.
18742 (profile_count::guessed_zero): New.
18744 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18746 * bb-reorder.c (fix_up_crossing_landing_pad,
18747 fix_crossing_conditional_branches): Use make_single_succ_edge
18748 to keep profile consistent.
18750 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18752 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18755 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18758 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18759 the right scopes, make sure cond_jump isn't preserved between multiple
18760 iterations. Search for fallthru edge whenever there are 3+ edges and
18761 use find_fallthru_edge for it.
18763 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18765 Patch by Alexander Monakov <amonakov@ispras.ru>
18766 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18767 probabilities consistently.
18769 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18771 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18772 emit_cmp_and_jump_insns.
18774 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18777 * tree-inline.c (expand_call_inline): Combine profile statuses.
18779 2017-06-30 Andrew Pinski <apinski@cavium.com>
18781 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18782 fold_stmt returned true.
18784 2017-06-30 Nathan Sidwell <nathan@acm.org>
18786 * ggc.h (empty_string): Delete.
18787 * cfgexpand.c (expand_asm_stmt): Use plain "".
18788 * optabs.c (expand_asm_memory_barrier): Likewise.
18789 * stringpool.c (empty_string): Delete.
18790 (digit_vector, digit_string): Delete.
18791 (ggc_alloc_string): Use plain "", don't optimize single digit
18792 strings. Use ggc_alloc_atomic.
18794 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18796 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18797 comparison set and one other set, use the cost of the non-comparison
18800 2017-06-30 Nathan Sidwell <nathan@acm.org>
18802 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18805 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18807 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18808 loops. Remove now unneeded calls to gimple_switch_set_label() that
18809 just set removed labels to NULL_TREE.
18811 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18813 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18814 (set_range_info): ...here. Only call set_range_info_raw if domain
18816 (set_nonzero_bits): Call set_range_info_raw.
18817 * tree-ssanames.h (set_range_info_raw): New.
18819 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18822 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18823 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18824 of nonimmediate_operand and <store_mask_constraint> instead of m
18825 for the input operand. For V8FI iterator, always split if input
18826 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18827 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18828 <store_mask_predicate> instead of register_operand and
18829 <store_mask_constraint> instead of v for the input operand. Make
18830 sure both operands aren't MEMs for if not <mask_applied>.
18832 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18834 * lto-wrapper.c (copy_file) Close both file descriptors before
18837 2017-06-30 Martin Liska <mliska@suse.cz>
18840 * multiple_target.c (create_dispatcher_calls): Make ifunc
18841 also for function that don't have calls or are not referenced.
18843 2017-06-30 Richard Biener <rguenther@suse.de>
18845 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18846 analyze the first scalar stmt. Move vector type computation
18847 for the BB case here from ...
18848 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18849 live operation processing in the SLP case properly.
18851 2017-06-30 Richard Biener <rguenther@suse.de>
18853 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18855 2017-06-30 Martin Liska <mliska@suse.cz>
18858 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18859 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18861 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18863 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18866 2017-06-30 Richard Biener <rguenther@suse.de>
18868 PR tree-optimization/81249
18869 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18870 condition reduction result to original scalar type.
18872 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18874 * profile-count.h (enum profile_quality): Fix typos and whitespace
18877 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18879 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18880 type for branch probabilities.
18882 2017-06-29 Julian Brown <julian@codesourcery.com>
18883 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18885 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18886 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18887 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18888 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18890 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18892 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18893 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18894 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18895 CC usage from generic code to here.
18896 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18897 CC usage into the target macros.
18899 2017-06-29 Maya Rashish <coypu@sdf.org>
18901 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18904 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18906 * arm/arm-builtins.c: Include profile-count.h
18907 * except.c (sjlj_emit_function_enter): Use
18908 profile_probability::unlikely.
18910 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18912 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18913 and tocrel_offset be pointer args rather than implicitly using
18915 (legitimate_constant_pool_address_p, rs6000_emit_move,
18916 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18917 tocrel_offset and use in toc_relative_expr_p call.
18918 (print_operand, print_operand_address): Use static tocrel_base_oac
18919 and tocrel_offset_oac.
18920 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18923 2017-06-29 Maya Rashish <coypu@sdf.org>
18925 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18927 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18929 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18930 objects, take into account only the alignment of 'op0' and 'mode1' if
18933 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18935 * ccmp.c (ccmp_tree_comparison_p): New function.
18936 (ccmp_candidate_p): Update to use above function.
18937 (get_compare_parts): New function.
18938 (expand_ccmp_next): Update to use new functions.
18939 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18941 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18942 take mode as argument.
18943 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18944 * expr.c (expand_expr_real_1): Pass mode as argument.
18946 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18948 * combine.c (combine_instructions): Print insns to dump_file, together
18951 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18953 * asan.c (asan_emit_stack_protection): Update.
18954 (create_cond_insert_point): Update.
18955 * auto-profile.c (afdo_propagate_circuit): Update.
18956 * basic-block.h (struct edge_def): Turn probability to
18957 profile_probability.
18958 (EDGE_FREQUENCY): Update.
18959 * bb-reorder.c (find_traces_1_round): Update.
18960 (better_edge_p): Update.
18961 (sanitize_hot_paths): Update.
18962 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18963 (make_single_succ_edge): Update.
18964 (check_bb_profile): Update.
18965 (dump_edge_info): Update.
18966 (update_bb_profile_for_threading): Update.
18967 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18968 probabilitycount to 0.
18969 * cfgbuild.c (compute_outgoing_frequencies): Update.
18970 * cfgcleanup.c (try_forward_edges): Update.
18971 (outgoing_edges_match): Update.
18972 (try_crossjump_to_edge): Update.
18973 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18974 (expand_gimple_tailcall): Update.
18975 (construct_init_block): Use make_single_succ_edge.
18976 (construct_exit_block): Use make_single_succ_edge.
18977 * cfghooks.c (verify_flow_info): Update.
18978 (redirect_edge_succ_nodup): Update.
18979 (split_edge): Update.
18980 (account_profile_record): Update.
18981 * cfgloopanal.c (single_likely_exit): Update.
18982 * cfgloopmanip.c (scale_loop_profile): Update.
18983 (set_zero_probability): Remove.
18984 (duplicate_loop_to_header_edge): Update.
18985 * cfgloopmanip.h (loop_version): Update prototype.
18986 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18987 (force_nonfallthru_and_redirect): Update.
18988 (update_br_prob_note): Update.
18989 (rtl_verify_edges): Update.
18990 (purge_dead_edges): Update.
18991 (rtl_lv_add_condition_to_bb): Update.
18992 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18993 * cgraphunit.c (init_lowered_empty_function): Update.
18994 (cgraph_node::expand_thunk): Update.
18995 * cilk-common.c: Include profile-count.h
18996 * dojump.c (inv): Remove.
18997 (jumpifnot): Update.
18998 (jumpifnot_1): Update.
18999 (do_jump_1): Update.
19001 (do_jump_by_parts_greater_rtx): Update.
19002 (do_compare_rtx_and_jump): Update.
19003 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19004 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19005 * dwarf2cfi.c: Include profile-count.h
19006 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19007 (sjlj_emit_dispatch_table): Likewise.
19008 * explow.c: Include profile-count.h
19009 * expmed.c (emit_store_flag_force): Update.
19010 (do_cmp_and_jump): Update.
19011 * expr.c (compare_by_pieces_d::generate): Update.
19012 (compare_by_pieces_d::finish_mode): Update.
19013 (emit_block_move_via_loop): Update.
19014 (store_expr_with_bounds): Update.
19015 (store_constructor): Update.
19016 (expand_expr_real_2): Update.
19017 (expand_expr_real_1): Update.
19018 * expr.h (try_casesi, try_tablejump): Update prototypes.
19019 * gimple-pretty-print.c (dump_probability): Update.
19020 (dump_profile): New.
19021 (dump_gimple_label): Update.
19022 (dump_gimple_bb_header): Update.
19023 * graph.c (draw_cfg_node_succ_edges): Update.
19024 * hsa-gen.c (convert_switch_statements): Update.
19025 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19026 (find_if_case_1): Update.
19027 (find_if_case_2): Update.
19028 * internal-fn.c (expand_arith_overflow_result_store): Update.
19029 (expand_addsub_overflow): Update.
19030 (expand_neg_overflow): Update.
19031 (expand_mul_overflow): Update.
19032 (expand_vector_ubsan_overflow): Update.
19033 * ipa-cp.c (good_cloning_opportunity_p): Update.
19034 * ipa-split.c (split_function): Use make_single_succ_edge.
19035 * ipa-utils.c (ipa_merge_profiles): Update.
19036 * loop-doloop.c (add_test): Update.
19037 (doloop_modify): Update.
19038 * loop-unroll.c (compare_and_jump_seq): Update.
19039 (unroll_loop_runtime_iterations): Update.
19040 * lra-constraints.c (lra_inheritance): Update.
19041 * lto-streamer-in.c (input_cfg): Update.
19042 * lto-streamer-out.c (output_cfg): Update.
19043 * mcf.c (adjust_cfg_counts): Update.
19044 * modulo-sched.c (sms_schedule): Update.
19045 * omp-expand.c (expand_omp_for_init_counts): Update.
19046 (extract_omp_for_update_vars): Update.
19047 (expand_omp_ordered_sink): Update.
19048 (expand_omp_for_ordered_loops): Update.
19049 (expand_omp_for_generic): Update.
19050 (expand_omp_for_static_nochunk): Update.
19051 (expand_omp_for_static_chunk): Update.
19052 (expand_cilk_for): Update.
19053 (expand_omp_simd): Update.
19054 (expand_omp_taskloop_for_outer): Update.
19055 (expand_omp_taskloop_for_inner): Update.
19056 * omp-simd-clone.c (simd_clone_adjust): Update.
19057 * optabs.c (expand_doubleword_shift): Update.
19058 (expand_abs): Update.
19059 (emit_cmp_and_jump_insn_1): Update.
19060 (expand_compare_and_swap_loop): Update.
19061 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19062 * predict.c (predictable_edge_p): Update.
19063 (edge_probability_reliable_p): Update.
19064 (set_even_probabilities): Update.
19065 (combine_predictions_for_insn): Update.
19066 (combine_predictions_for_bb): Update.
19067 (propagate_freq): Update.
19068 (estimate_bb_frequencies): Update.
19069 (force_edge_cold): Update.
19070 * profile-count.c (profile_count::dump): Add missing space into dump.
19071 (profile_count::debug): Add newline.
19072 (profile_count::differs_from_p): Explicitly convert to unsigned.
19073 (profile_count::stream_in): Update.
19074 (profile_probability::dump): New member function.
19075 (profile_probability::debug): New member function.
19076 (profile_probability::differs_from_p): New member function.
19077 (profile_probability::differs_lot_from_p): New member function.
19078 (profile_probability::stream_in): New member function.
19079 (profile_probability::stream_out): New member function.
19080 * profile-count.h (profile_count_quality): Rename to ...
19081 (profile_quality): ... this one.
19082 (profile_probability): New.
19083 (profile_count): Update.
19084 * profile.c (compute_branch_probabilities): Update.
19085 * recog.c (peep2_attempt): Update.
19086 * sched-ebb.c (schedule_ebbs): Update.
19087 * sched-rgn.c (find_single_block_region): Update.
19088 (compute_dom_prob_ps): Update.
19089 (schedule_region): Update.
19090 * sel-sched-ir.c (compute_succs_info): Update.
19091 * stmt.c (struct case_node): Update.
19092 (do_jump_if_equal): Update.
19093 (get_outgoing_edge_probs): Update.
19094 (conditional_probability): Update.
19095 (emit_case_dispatch_table): Update.
19096 (expand_case): Update.
19097 (expand_sjlj_dispatch_table): Update.
19098 (emit_case_nodes): Update.
19099 * targhooks.c: Update.
19100 * tracer.c (better_p): Update.
19101 (find_best_successor): Update.
19102 * trans-mem.c (expand_transaction): Update.
19103 * tree-call-cdce.c: Update.
19104 * tree-cfg.c (gimple_split_edge): Upate.
19105 (move_sese_region_to_fn): Upate.
19106 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19107 * tree-eh.c (lower_resx): Upate.
19108 (cleanup_empty_eh_move_lp): Upate.
19109 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19110 * tree-inline.c (copy_edges_for_bb): Update.
19111 (copy_cfg_body): Update.
19112 * tree-parloops.c (gen_parallel_loop): Update.
19113 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19114 (gimple_gen_time_profiler): Update.
19115 * tree-ssa-dce.c (remove_dead_stmt): Update.
19116 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19117 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19118 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19119 (unloop_loops): Update.
19120 (try_peel_loop): Update.
19121 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19122 * tree-ssa-loop-split.c (connect_loops): Update.
19123 (split_loop): Update.
19124 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19125 (hoist_guard): Update.
19126 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19127 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19128 (value_replacement): Update.
19129 * tree-ssa-reassoc.c (branch_fixup): Update.
19130 * tree-ssa-tail-merge.c (replace_block_by): Update.
19131 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19132 (create_edge_and_update_destination_phis): Update.
19133 (compute_path_counts): Update.
19134 (recompute_probabilities): Update.
19135 (update_joiner_offpath_counts): Update.
19136 (freqs_to_counts_path): Update.
19137 (duplicate_thread_path): Update.
19138 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19139 (struct switch_conv_info): Update.
19140 (gen_inbound_check): Update.
19141 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19142 (vect_do_peeling): Update.
19143 (vect_loop_versioning): Update.
19144 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19145 (optimize_mask_stores): Update.
19146 * ubsan.c (ubsan_expand_null_ifn): Update.
19147 * value-prof.c (gimple_divmod_fixed_value): Update.
19148 (gimple_divmod_fixed_value_transform): Update.
19149 (gimple_mod_pow2): Update.
19150 (gimple_mod_pow2_value_transform): Update.
19151 (gimple_mod_subtract): Update.
19152 (gimple_mod_subtract_transform): Update.
19153 (gimple_ic): Update.
19154 (gimple_stringop_fixed_value): Update.
19155 (gimple_stringops_transform): Update.
19156 * value-prof.h: Update.
19158 2017-06-29 Carl Love <cel@us.ibm.com>
19160 * config/rs6000/rs6000-c.c: Add support for built-in functions
19161 vector signed int vec_signed (vector float);
19162 vector signed long long vec_signed (vector double);
19163 vector signed int vec_signed2 (vector double, vector double);
19164 vector signed int vec_signede (vector double);
19165 vector signed int vec_signedo (vector double);
19166 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19167 instruction generator.
19168 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19169 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19170 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19172 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19173 vunsignede_v2df): Add define_expands.
19174 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19175 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19176 VEC_UNSIGNEDO): Add definitions.
19177 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19178 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19179 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19180 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19181 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19182 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19183 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19184 vec_unsignede, vec_unsignedo): Add builtin defines.
19185 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19187 * doc/extend.texi: Update the built-in documentation file for the
19188 new built-in functions.
19190 2017-06-29 Richard Biener <rguenther@suse.de>
19192 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19193 reduction chains to LOOP_VINFO_REDUCTIONS.
19194 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19195 SLP reductions after processing reduction chains.
19197 2017-06-29 Nathan Sidwell <nathan@acm.org>
19199 * builtins.c (fold_builtin_FUNCTION): Use
19200 lang_hooks.decl_printable_name.
19202 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19204 PR middle-end/81194
19205 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19206 with only one label.
19207 * stmt.c (expand_case): Assert NCASES is greater than one.
19209 2017-06-29 Richard Biener <rguenther@suse.de>
19211 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19213 (group_case_labels): Likewise.
19214 (find_taken_edge): Push sanity checking on val to workers...
19215 (find_taken_edge_cond_expr): ... here
19216 (find_taken_edge_switch_expr): ... and here, handle cases
19217 with just a default label.
19218 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19219 (group_case_labels): Likewise.
19220 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19221 group_case_labels does anything cleanup the CFG again.
19223 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19225 PR tree-optimization/81196
19226 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19227 exit condition comparing two IVs.
19229 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19231 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19232 profile to the dummy entry at the end of the list of architectures.
19233 * config/arm/arm-cpu-cdata.h: Regenerated.
19235 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19236 Michael Collison <michael.collison@arm.com>
19239 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19241 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19242 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19243 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19244 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19245 of shift when the shift amount is masked with constant equal to
19246 the size of the mode.
19247 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19250 2017-06-29 Martin Liska <mliska@suse.cz>
19252 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19254 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19257 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19260 (edge_growth_cache_entry::edge_growth_cache_entry): New
19262 (reset_edge_growth_cache): Update to use constructor.
19264 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19266 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19267 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19268 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19270 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19272 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19273 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19275 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19277 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19278 (*-linux-uclibc*): Add t-uclibc tmake_file.
19279 * config/t-musl: New.
19280 * config/t-uclibc: New.
19282 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19284 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19286 (gen_comm_data): Emit architectural setting of arch_prof.
19287 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19289 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19290 (armv8-m.base, armv8-m.main): Likewise.
19291 * arm-protos.h (arm_build_target): Add profile field.
19292 (arch_option): Likewise.
19293 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19295 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19296 arm_active_target.profile.
19298 2017-06-28 Richard Biener <rguenther@suse.de>
19300 PR middle-end/81227
19301 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19302 TYPE_OVERFLOW_WRAPS.
19303 * match.pd (negate_expr_p): Likewise.
19304 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19305 fold_build2, not fold_binary.
19307 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19309 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19310 Convert memory address to Pmode.
19311 (aarch64_print_operand): Assert MEM operands are always Pmode.
19313 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19316 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19317 Remove redundant if.
19318 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19319 * config/arm/aarch-common-protos.h
19320 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19321 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19323 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19326 * multiple_target.c (create_dispatcher_calls): Set the default
19327 clone to be static, not public.
19329 2017-06-28 Richard Biener <rguenther@suse.de>
19331 * tree-vect-loop.c (vectorizable_reduction): Move special
19332 cond reduction IV var creation ...
19333 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19334 parameter. Use STMT_VINFO_VECTYPE.
19335 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19338 2017-06-28 Martin Liska <mliska@suse.cz>
19341 * ipa-visibility.c (non_local_p): Handle visibility.
19343 2017-06-28 Martin Liska <mliska@suse.cz>
19346 * common.opt: Add IntegerRange to various options.
19347 * opt-functions.awk (integer_range_info): New function.
19348 * optc-gen.awk: Add integer_range_info to cl_options struct.
19349 * opts-common.c (decode_cmdline_option): Handle
19350 CL_ERR_INT_RANGE_ARG.
19351 (cmdline_handle_error): Likewise.
19352 * opts.c (print_filtered_help): Show valid interval in
19353 when --help is provided.
19354 * opts.h (struct cl_option): Add range_min and range_max fields.
19355 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19357 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19359 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19360 (x * C EQ/NE y * C): New transformation.
19362 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19364 * genmultilib (combination_space): Accept '+' in option names.
19366 2017-06-28 Martin Liska <mliska@suse.cz>
19369 * asan.c (instrument_derefs): Bail out inner references
19370 that are hard register variables.
19372 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19375 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19376 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19378 2017-06-28 Richard Biener <rguenther@suse.de>
19380 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19381 (vect_get_slp_defs): Adjust.
19382 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19384 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19386 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19387 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19388 (vectorizable_reduction): Adjust.
19389 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19391 (vect_get_slp_defs): Likewise.
19392 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19393 (vectorizable_bswap): Adjust.
19394 (vectorizable_call): Likewise.
19395 (vectorizable_conversion): Likewise.
19396 (vectorizable_assignment): Likewise.
19397 (vectorizable_shift): Likewise.
19398 (vectorizable_operation): Likewise.
19399 (vectorizable_store): Likewise.
19400 (vectorizable_condition): Likewise.
19401 (vectorizable_comparison): Likewise.
19403 2017-06-28 Michael Collison <michael.collison@arm.com>
19406 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19408 (arm_gen_movmemqi): Removed unused variable 'i'.
19409 Convert 'for' loop into 'while' loop.
19410 (arm_expand_prologue): Remove last unnecessary set of insn.
19411 (thumb_pop): Remove unused variable 'pushed_words'.
19412 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19414 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19416 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19417 * config/s390/s390-protos.h: Add prototype of
19418 s390_rel_address_ok_p.
19419 * config/s390/s390.c (s390_got_symbol): New function.
19420 (s390_rel_address_ok_p): New function.
19421 (legitimize_pic_address): Use s390_rel_address_ok_p.
19422 (s390_load_got): Use s390_got_symbol.
19423 (s390_option_override): Issue error if
19424 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19425 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19427 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19429 2017-06-27 Andrew Pinski <apinski@cavium.com>
19431 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19432 (X * copysign (1.0, X)): New pattern.
19433 (X * copysign (1.0, -X)): New pattern.
19434 (copysign (-1.0, CST)): New pattern.
19436 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19438 * genmultilib (combination_space): Remove variable.
19439 Validate reuse rules against regular expression for any sequence
19440 of multilib options in any order.
19442 2017-06-27 Michael Collison <michael.collison@arm.com>
19444 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19445 call aarch64_split_simd_combine.
19446 * (aarch64_combine_internal<mode>): Delete pattern.
19447 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19448 Allow register and subreg operands.
19450 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19452 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19453 specific need, just fallback on defaults.
19454 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19456 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19457 Olivier Hainque <hainque@adacore.com>
19459 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19461 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19462 targets. Pick a default if no particular attempt applied.
19463 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19466 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19468 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19469 (x86_64-wrs-vxworks7): Likewise.
19471 2017-06-27 Marek Polacek <polacek@redhat.com>
19474 * ubsan.c (instrument_null): Check get_base_address's result for null.
19476 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19478 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19480 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19482 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19483 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19484 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19485 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19486 New function types.
19487 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19488 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19489 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19490 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19491 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19492 BUILT_IN_FEUPDATEENV): New builtins.
19493 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19494 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19495 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19496 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19498 (builtin_structptr_types): Adjust size.
19499 * tree.c (builtin_structptr_types): Add four entries.
19501 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19502 Olivier Hainque <hainque@adacore.com>
19504 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19505 (TLS_SYM): New local macro, forcing reference to __tls__ on
19506 link command lines for VxWorks 7 RTPs, triggering initialization
19508 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19509 OS features TLS support, true for RTPs on VxWorks 7.
19510 * config/vxworks.c (vxworks_override_options): Setup emutls
19513 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19515 * predict.c (test_prediction_value_range): Use -1U instead of -1
19516 to avoid narrowing conversion warning.
19517 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19518 to avoid narrowing conversion warning.
19519 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19521 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19523 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19525 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19526 64bit configurations.
19527 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19528 (SIZE_TYPE): Likewise.
19529 * config/vxworks.c (vxworks_emutls_var_fields): Use
19530 long_unsigned_type_node instead of unsigned_type_node as the offset
19531 field type, which is "pointer" mode in emutls.c.
19533 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19536 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19538 PR middle-end/81207
19539 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19540 gimple_vuse copying separately from gimple_vdef copying.
19542 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19544 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19547 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19548 Olivier Hainque <hainque@adacore.com>
19550 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19551 for all vxworks7 targets.
19552 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19553 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19554 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19555 variations for VX6/VX7 and 32/64bits later on in ...
19556 (VXWORKS_LIB_SPEC): Leverage new macros.
19557 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19558 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19560 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19562 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19563 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19565 2017-06-26 Carl Love <cel@us.ibm.com>
19567 * config/rs6000/rs6000-c.c: Add support for built-in functions
19568 vector bool char vec_reve (vector bool char);
19569 vector signed char vec_reve (vector signed char);
19570 vector unsigned char vec_reve (vector unsigned char);
19571 vector bool int vec_reve (vector bool int);
19572 vector signed int vec_reve (vector signed int);
19573 vector unsigned int vec_reve (vector unsigned int);
19574 vector bool long long vec_reve (vector bool long long);
19575 vector signed long long vec_reve (vector signed long long);
19576 vector unsigned long long vec_reve (vector unsigned long long);
19577 vector bool short vec_reve (vector bool short);
19578 vector signed short vec_reve (vector signed short);
19579 vector double vec_reve (vector double);
19580 vector float vec_reve (vector float);
19581 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19582 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19583 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19584 (altivec_vreve): New pattern.
19585 * config/rs6000/altivec.h (vec_reve): New define.
19586 * doc/extend.texi (vec_rev): Update the built-in documentation file
19587 for the new built-in functions.
19589 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19591 PR tree-optimization/71815
19592 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19594 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19596 (slsr_process_phi): Likewise.
19597 (replace_uncond_cands_and_profitable_phis): Don't replace a
19598 multiply candidate with a stride of 1 (copy or cast).
19599 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19601 (lowest_cost_path): Likewise.
19602 (total_savings): Likewise.
19604 2017-06-26 Richard Biener <rguenther@suse.de>
19607 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19608 Use def_builtin_pure for all gather builtins.
19610 2017-06-26 Richard Biener <rguenther@suse.de>
19612 PR tree-optimization/81203
19613 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19614 non-dominating BBs.
19616 2017-06-26 Marek Polacek <polacek@redhat.com>
19619 * doc/invoke.texi: Document -Wmultistatement-macros.
19621 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19623 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19624 arm_neon_ok_no_float_abi effective target.
19626 2017-06-26 Richard Biener <rguenther@suse.de>
19628 PR tree-optimization/80928
19629 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19630 (copy_bbs): Set BB_DUPLICATED flag early.
19631 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19633 (execute_on_shrinking_pred): Likewise.
19634 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19635 BB_DUPLICATED blocks.
19636 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19637 iterate over all PHIs considering removal of *gsi.
19639 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19641 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19644 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19646 * config/rs6000/rs6000-string.c: (expand_block_clear,
19647 do_load_for_compare, select_block_compare_mode,
19648 compute_current_alignment, expand_block_compare,
19649 expand_strncmp_align_check, expand_strn_compare,
19650 expand_block_move, rs6000_output_load_multiple)
19651 Move functions related to string/block move/compare
19652 to a separate file.
19653 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19654 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19655 for this function which is now used in two files.
19656 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19657 * config.gcc: Add rs6000-string.o to extra_objs for
19658 targets powerpc*-*-* and rs6000*-*-*.
19660 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19663 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19664 32-bit, since indexed is not valid for DImode.
19665 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19666 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19667 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19668 (define_peephole2 for Altivec d-form store): Likewise.
19671 * multiple_target.c (create_dispatcher_calls): Only create the
19672 dispatcher call if the function is the default clone of a
19673 versioned function.
19675 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19677 PR middle-end/80902
19678 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19679 a call, force the call to not be a tail call.
19681 2017-06-23 Jeff Law <law@redhat.com>
19683 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19686 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19688 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19689 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19690 handling for early expansion of vector shifts (sl,sr,sra,rl).
19691 (builtin_function_type): Add vector shift right instructions
19692 to the unsigned argument list.
19694 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19696 rtl-optimizatoin/79286
19697 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19698 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19699 trap. PIC register plus a const unspec without offset can never trap.
19701 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19703 * tree.h (builtin_structptr_type): New type.
19704 (builtin_structptr_types): Declare new array.
19705 * tree.c (builtin_structptr_types): New array.
19706 (free_lang_data, build_common_tree_nodes): Use it.
19708 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19711 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19714 2017-06-22 Matt Turner <mattst88@gmail.com>
19716 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19717 Lake models to skylake case. Assume skylake for unknown
19718 models with clflushopt.
19720 2017-06-22 Jeff Law <law@redhat.com>
19722 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19723 frame sizes that do not satisfy aarch64_uimm12_shift.
19725 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19727 * profile-count.h (apply_probability,
19728 apply_scale, probability_in): Fix checks for zero.
19730 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19732 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19733 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19735 2016-06-22 Richard Biener <rguenther@suse.de>
19737 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19738 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19739 REDUC_MAX_EXPR support.
19740 (vectorizable_reduction): Likewise.
19741 (vect_create_epilog_for_reduction): Likewise.
19743 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19745 * match.pd (A / (1 << B) -> A >> B): New.
19746 * generic-match-head.c: Include optabs-tree.h.
19747 * gimple-match-head.c: Likewise.
19748 * optabs-tree.h (target_supports_op_p): New.
19749 * optabs-tree.c (target_supports_op_p): New.
19751 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19753 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19754 $gcc_cv_ld --help output.
19755 (gcc_cv_ld_demangle): Likewise.
19756 (gcc_cv_ld_eh_frame_hdr): Likewise.
19757 (gcc_cv_ld_pie): Likewise.
19758 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19759 (gcc_cv_ld_buildid): Likewise.
19760 (gcc_cv_ld_sysroot): Likewise.
19761 (ld_bndplt_support): Likewise.
19762 (ld_pushpopstate_support): Likewise.
19763 * configure: Regenerate.
19764 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19766 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19769 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19770 operands indexes to avoid gap between operands and match_dups.
19772 2017-06-21 Andrew Pinski <apinski@cavium.com>
19774 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19775 Increment Arith_shift and Arith_shift_reg by 1.
19776 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19778 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19779 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19780 (aarch64_strip_extend): Add new argument and test for it.
19781 (aarch64_cheap_mult_shift_p): New function.
19782 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19783 add a cost if it is true.
19784 Update calls to aarch64_strip_extend.
19785 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19787 2017-06-21 Andrew Pinski <apinski@cavium.com>
19789 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19791 (thunderxt88): Likewise.
19792 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19793 (thunderx_prefetch_tune): New variable.
19794 (thunderx2t99_prefetch_tune): Update for the correct values.
19795 (thunderxt88_tunings): New variable.
19796 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19797 generic_prefetch_tune.
19798 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19800 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19802 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19803 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19804 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19805 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19806 (aarch64_atomic_cas<mode>, GPI): Likewise.
19808 2017-06-21 Martin Liska <mliska@suse.cz>
19810 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19811 statements on cold and hot labels.
19812 * predict.c (tree_estimate_probability_bb): Remove the
19813 prediction from this place.
19815 2017-06-21 Martin Liska <mliska@suse.cz>
19817 PR tree-optimization/79489
19818 * gimplify.c (maybe_add_early_return_predict_stmt): New
19820 (gimplify_return_expr): Call the function.
19821 * predict.c (tree_estimate_probability_bb): Remove handling
19823 * predict.def: Update comment about early return predictor.
19824 * gimple-predict.h (is_gimple_predict): New function.
19825 * predict.def: Change default value of early return to 66.
19826 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19828 * passes.def: Put pass_strip_predict_hints to the beginning of
19831 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19833 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19834 FUNCTION_DECL declarations.
19835 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19837 (dwaf2out_decl): Likewise.
19838 * godump.c (go_early_global_decl): Skip call to the real debug hook
19839 for FUNCTION_DECL declarations.
19840 * passes.c (rest_of_decl_compilation): Skip call to the
19841 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19842 -fdump-go-spec is passed.
19844 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19846 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19847 Reorder for compactness.
19848 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19849 (def_builtin_pure, def_builtin_pure2): New functions.
19850 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19852 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19854 * match.pd (nop_convert): New predicate.
19855 ((A +- CST1) +- CST2): Allow some NOP conversions.
19857 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19860 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19861 with ctors/dtors if GOVD_SHARED is set.
19863 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19865 * config/aarch64/aarch64.md (movti_aarch64):
19866 Emit mov rather than orr.
19867 (movtf_aarch64): Likewise.
19868 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19869 Emit mov rather than orr.
19871 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19873 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19874 Swap alternatives, make integer dup more expensive.
19876 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19878 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19879 Return true for non-tls symbols.
19881 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19883 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19884 (cortex-a75): Likewise.
19885 (cortex-a75.cortex-a55): Likewise.
19886 * config/aarch64/aarch64-tune.md: Regenerate.
19887 * doc/invoke.texi (-mtune): Document new values for -mtune.
19889 2017-06-21 Tom de Vries <tom@codesourcery.com>
19891 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19892 stack_size feature.
19893 (Effective-Target Keywords, Other attributes): Suggest using
19894 dg-add-options stack_size feature to get stack limit in stack_size
19895 effective target documentation.
19897 2017-06-21 Julian Brown <julian@codesourcery.com>
19898 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19900 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19901 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19902 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19904 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19905 attribute type list for neon_multiply.
19906 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19907 attribute type list for neon_multiply.
19908 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19909 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19910 attribute type list for neon_multiply.
19911 * config/arm/types.md (crypto_pmull): Add.
19912 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19913 attribute type list.
19915 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19917 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19920 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19922 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19923 to make sure not to dereference a NULL cost_classes_ptr pointer.
19925 2017-06-20 Carl Love <cel@us.ibm.com>
19927 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19928 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19929 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19930 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19931 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19932 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19933 VMULOSW): New enum "unspec" values.
19934 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19935 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19936 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19937 altivec_vmulosw): New patterns.
19938 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19939 VMULOSW): Add definitions.
19941 2017-06-20 Julia Koval <julia.koval@intel.com>
19943 * config/i386/i386.c: Fix rounding expand for new pattern.
19944 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19946 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19948 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19949 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19951 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19953 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19956 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19958 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19959 architecture, then by implementer ID.
19960 * config/aarch64/aarch64-tune.md: Regenerate.
19962 2017-06-20 Richard Biener <rguenther@suse.de>
19964 PR middle-end/81097
19965 * fold-const.c (split_tree): Fold to type before negating.
19967 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19969 * diagnostic-show-locus.c
19970 (selftest::test_fixit_deletion_affecting_newline): New function.
19971 (selftest::diagnostic_show_locus_c_tests): Call it.
19973 2017-06-20 Andreas Schwab <schwab@suse.de>
19976 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19979 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19981 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19982 __ARM_FEATURE_COPROC according to support.
19984 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19986 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19987 Rewritten to avoid overflow for > 32-bit pointers.
19990 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19991 by removing enum keyword.
19992 (ubsan_type_descriptor): Likewise. Formatting fix.
19995 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19996 splitter): Require TARGET_SSE2 in the condition.
19998 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20001 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20002 for doing vector set of SFmode on ISA 3.0.
20003 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20004 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20006 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20007 SFmode value into a V4SF variable that was extracted from another
20008 V4SF variable without converting the element to double precision
20009 and back to single precision vector format.
20010 (vsx_insert_extract_v4sf_p9_2): Likewise.
20012 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20014 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20015 in UWHI to avoid undefined overflow.
20018 * ubsan.h (enum ubsan_encode_value_phase): New.
20019 (ubsan_encode_value): Change second argument to
20020 enum ubsan_encode_value_phase with default value of
20021 UBSAN_ENCODE_VALUE_GENERIC.
20022 * ubsan.c (ubsan_encode_value): Change second argument to
20023 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20024 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20025 create_tmp_var_raw instead of create_tmp_var and use a
20027 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20028 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20029 ubsan_encode_value callers.
20032 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20033 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20034 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20036 2017-06-19 Richard Biener <rguenther@suse.de>
20038 PR middle-end/81118
20039 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20040 estimates if we changed anything.
20042 2017-06-19 Richard Biener <rguenther@suse.de>
20044 PR tree-optimization/80887
20045 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20046 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20047 simplified lookups, then reset mprts_hook.
20048 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20050 (try_to_simplify): Likewise.
20052 2017-06-19 Martin Liska <mliska@suse.cz>
20055 * gimplify.c (gimplify_switch_expr):
20056 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20058 2017-06-19 Martin Liska <mliska@suse.cz>
20060 * doc/install.texi: Document that PGO runs in 4 stages.
20062 2017-06-19 Martin Liska <mliska@suse.cz>
20065 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20066 to dispatcher function name.
20067 * multiple_target.c (replace_function_decl): New function.
20068 (create_dispatcher_calls): Redirect both edges and references.
20070 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20072 * profile-count.c (profile_count::dump): Dump quality.
20073 (profile_count::differs_from_p): Update for unsigned val.
20074 * profile-count.h (profile_count_quality): New enum.
20075 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20077 2017-06-19 Richard Biener <rguenther@suse.de>
20079 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20080 struct function as arg.
20081 (estimate_numbers_of_iterations): Export overload with loop arg.
20082 (free_numbers_of_iterations_estimates_loop): Use an overload of
20083 free_numbers_of_iterations_estimates instead.
20084 * tree-cfg.c (remove_bb): Adjust.
20085 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20086 * tree-parloops.c (gen_parallel_loop): Likewise.
20087 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20089 (tree_unroll_loops_completely): Likewise.
20090 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20091 Use an overload instead and export.
20092 (estimated_loop_iterations): Adjust.
20093 (max_loop_iterations): Likewise.
20094 (likely_max_loop_iterations): Likewise.
20095 (estimate_numbers_of_iterations): Take struct function as arg
20097 (loop_exits_before_overflow): Adjust.
20098 (free_numbers_of_iterations_estimates_loop): Use an overload.
20099 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20100 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20102 2017-06-19 Richard Biener <rguenther@suse.de>
20105 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20106 RANGE_EXPR conservatively.
20108 2017-06-16 Carl Love <cel@us.ibm.com>
20110 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20111 definitions for vec_float, vec_float2, vec_floato,
20112 vec_floate built-ins.
20113 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20114 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20116 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20117 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20118 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20119 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20120 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20121 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20122 vec_floato): Add builtin defines.
20123 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20124 Update the built-in documentation file for the new built-in
20127 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20129 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20130 (mthumb): Mark as the negative of -marm.
20132 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20134 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20136 (ARM Options, -mtune): Document that this accepts the same
20137 extension options as -mcpu.
20138 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20140 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20142 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20143 permitted extensions.
20145 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20147 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20148 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20149 (armv8-m.main): Add option +nodsp.
20150 * config/arm/arm-cpu-cdata.h: Regenerated.
20152 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20154 * config/arm/t-fuchsia: New file.
20155 * config.gcc (arm*-*-fuchsia*): Use it.
20157 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20159 * config/arm/t-symbian: Rewrite for new option infrastructure.
20161 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20163 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20164 (MULTILIB_REQUIRED): Likewise.
20166 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20168 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20169 (MULTILIB_RESUE): Likewise.
20170 (MULTILIB_MATCHES): Likewise.
20171 (MULTLIB_REQUIRED): Likewise.
20173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20175 * config/arm/t-rtems: Rewrite for new option framework.
20177 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20179 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20180 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20181 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20182 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20183 * config/arm/t-multilib: ... here.
20184 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20185 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20186 armv7-a and armv8*-a when A-profile libraries have not been built.
20187 * config/arm/t-rmprofile: Rewrite.
20189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20191 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20192 with a backslash. Remove the backslash after substituting unescaped
20194 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20196 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20198 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20199 the driver to pass through the default setting of -mfloat-abi.
20200 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20202 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20203 (all_feat_combs): New rule.
20204 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20206 * config/arm/t-aprofile: Rewrite.
20208 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20210 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20211 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20212 fpu is not specified by the user/command-line.
20213 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20214 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20215 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20216 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20217 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20218 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20220 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20222 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20223 * config/arm/t-arm-elf: Rewritten.
20225 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20227 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20228 have some floating-point instructions.
20229 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20230 (TARGET_MAYBE_HARD_FLOAT): New macro.
20231 * config/arm/arm-builtins.c (arm_init_builtins): Use
20232 TARGET_MAYBE_HARD_FLOAT.
20233 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20235 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20237 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20238 (configargs.h): Include it.
20239 (arm_print_hint_for_fpu_option): New function.
20240 (arm_parse_fpu_option): New function.
20241 (candidate_extension): New class.
20242 (arm_canon_for_multilib): New function.
20243 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20244 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20245 (ARCH_CANONICAL_SPECS): New macro.
20246 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20248 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20250 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20251 are set after handling multilib fragments. Set target_cpu_default2
20254 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20256 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20258 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20259 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20261 (check_arch): Likewise.
20262 * config/arm/arm.c (arm_configure_build_target): Handle
20263 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20264 options in the default.
20266 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20268 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20269 when an option is an alias of another.
20270 * config/arm/parsecpu.awk (optalias): New parser token.
20271 (gen_comm_data): Mark non-alias options as such. Emit entries
20272 for extension aliases.
20273 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20274 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20275 (armv6kz, armv6zk, armv6t2): Likewise.
20276 (armv7): Make vfpv3-d16 an alias.
20277 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20279 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20280 Sort in canonical order.
20281 (armv8-a): Sort in canonical order.
20282 (armv8.1-a, armv8.2-a): Likewise.
20283 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20285 (cortex-a9): Sort in canonical order.
20286 * config/arm/arm.c (selftests.h): Include it.
20287 (arm_test_cpu_arch_data): New function.
20288 (arm_run_self_tests): New function.
20289 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20290 (targetm): Move declaration to the end of the file.
20291 * arm-cpu-cdata.h: Regenerated.
20293 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20295 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20296 call to target_mode_check describing the type of option passed.
20297 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20298 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20299 arm_parse_cpu_option_name to match parameters against list of
20301 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20302 arm_arch_core_flags data structure.
20303 * config/arm/arm-cpu_cdata.h: Regenerated.
20305 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20307 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20309 (arm_print_hint_for_cpu_option): Likewise.
20310 (arm_print_hint_for_arch_option): Likewise.
20311 (arm_parse_cpu_option_name): Likewise.
20312 (arm_parse_arch_option_name): Likewise.
20313 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20314 of entries in the all_fpus list.
20315 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20316 (arm_parse_cpu_option_name): Declare.
20317 (arm_parse_arch_option_name): Declare.
20318 (arm_parse_option_features): Declare.
20319 (arm_intialize_isa): Declare.
20320 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20322 (gen_comm_data): ... here. Make definitions non-static.
20323 * config/arm/arm-cpu-data.h: Regenerated.
20324 * config/arm/arm-cpu-cdata.h: Regenerated.
20326 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20328 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20329 (cpu_arch_extension): New structure.
20330 (cpu_arch_option, arch_option, cpu_option): New structures.
20331 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20332 architecture types.
20333 (gen_data): Generate new format data tables.
20334 * config/arm/arm.c (cpu_tune): New structure.
20335 (cpu_option, processors): Delete.
20336 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20337 (arm_print_hint_for_cpu_option): ... this and ...
20338 (arm_print_hint_for_arch_option): ... this.
20339 (arm_parse_arch_cpu_name): Delete. Replace with ...
20340 (arm_parse_cpu_option_name): ... this and ...
20341 (arm_parse_arch_option_name): ... this.
20342 (arm_unrecognized_feature): Change type of target parameter to
20344 (arm_parse_arch_cpu_features): Delete. Replace with ...
20345 (arm_parse_option_features): ... this.
20346 (arm_configure_build_target): Rework to use new configuration data
20348 (arm_print_tune_info): Rework for new configuration data tables.
20349 * config/arm/arm-cpu-data.h: Regenerated.
20350 * config/arm/arm-cpu.h: Regenerated.
20352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20354 * Makefile.in (OBJS): Move sbitmap.o from here ...
20355 (OBJS-libcommon): ... to here.
20357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20359 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20360 (ISA_ALL_CRYPTO): New macro.
20361 (ISA_ALL_SIMD): New macro
20362 (ISA_ALL_FP): New macro.
20363 * config/arm/arm.c (fpu_bitlist): Update initializer.
20364 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20366 (arm9e): Add fpu. Add option for nofp
20367 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20368 (arm926ej-s, arm1026ej-s): Likewise.
20369 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20370 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20371 neon-fp16, neon-vfpv4, nofp and nosimd.
20372 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20373 (cortex-a8): Add fpu. Add option for nofp.
20374 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20375 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20376 (cortex-r4f): Add fpu.
20377 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20378 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20380 (cortex-r8): Likewise.
20381 (cortex-m4): Add fpu. Add option for nofp.
20382 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20383 (cortex-a17.cortex-a7): Likewise.
20384 (cortex-a32): Add fpu. Add options for crypto and nofp.
20385 (cortex-a35, cortex-a53): Likewise.
20386 (cortex-a57): Add fpu. Add option for crypto.
20387 (cortex-a72, cortex-a73): Likewise.
20388 (exynos-m1): Likewise.
20389 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20390 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20391 (cortex-m33): Add fpu. Add option for nofp.
20392 * config/arm/arm-cpu-cdata.h: Regenerated
20393 * config/arm/arm-cpu-data.h: Regenerated.
20395 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20397 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20398 (armv5te, armv5tej): Likewise.
20399 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20400 (armv7): Add options fp and vfpv3-d16.
20401 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20402 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20404 (armv7ve): Likewise.
20405 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20406 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20407 (armv8-a): Add nocrypto option.
20408 (armv8.1-a, armv8.2-a): Likewise.
20409 (armv8-m.main): add options fp, fp.dp and nofp.
20411 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20413 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20415 (armv8-a+crc): Delete.
20416 (armv8.1-a): Add options simd, crypto and nofp.
20417 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20418 (armv8.2-a+fp16): Delete.
20419 (armv8-m.main): Add option dsp.
20420 (armv8-m.main+dsp): Delete.
20421 (cortex-a8): Add fpu. Add nofp option.
20422 (cortex-a9): Add fpu. Add nofp and nosimd options.
20423 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20424 link to main cpu and architecture data structures.
20425 (gen_comm_data): Only put isa attributes from the main architecture
20427 (option): New statement for architecture and CPU entries.
20428 * arm.c (struct cpu_option): New structure.
20429 (struct processors): Add entry for options.
20430 (arm_unrecognized_feature): New function.
20431 (arm_parse_arch_cpu_name): Ignore any characters after the first
20433 (arm_parse_arch_cpu_feature): New function.
20434 (arm_configure_build_target): Separate out any CPU and architecture
20435 features and parse separately. Don't error out if -mfpu=auto is
20436 used with only an architecture string.
20437 (arm_print_asm_arch_directives): New function.
20438 (arm_file_start): Call it.
20439 * config/arm/arm-cpu-cdata.h: Regenerated.
20440 * config/arm/arm-cpu-data.h: Likewise.
20441 * config/arm/arm-tables.opt: Likewise.
20443 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20445 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20446 assembler when it is not -mfpu=auto.
20448 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20450 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20451 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20452 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20453 (ASM_CPU_SPEC): Rewrite.
20454 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20455 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20456 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20457 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20458 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20459 copied string is NUL-terminated. Also strip any characters prefixed
20461 (arm_rewrite_selected_arch): New function.
20462 (arm_rewrite_march): New function.
20464 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20466 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20467 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20468 (march, mcpu, mtune): Convert to string-based options.
20469 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20470 (arm_parse_arch_cpu_name): New function.
20471 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20472 identify selected architecture or CPU.
20473 (arm_option_save): New function.
20474 (TARGET_OPTION_SAVE): Redefine.
20475 (arm_option_restore): Restore string options.
20476 (arm_option_print): Print string options.
20478 2017-06-16 Martin Sebor <msebor@redhat.com>
20480 PR tree-optimization/80933
20481 PR tree-optimization/80934
20482 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20483 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20484 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20485 (gimple_fold_builtin): Call them.
20487 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20489 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20490 as unlikely; update profile.
20492 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20494 * predict.c (force_edge_cold): Handle declaring edges impossible
20497 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20499 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20501 (try_unroll_loop_completely): Fix reporting.
20503 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20505 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20507 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20510 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20511 if given a non-constant argument for an intrinsic which requires a
20514 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20516 * profile.c (compare_freqs): New function.
20517 (branch_prob): Sort edge list.
20518 (find_spanning_tree): Assume that the list is priority sorted.
20520 2017-06-16 Richard Biener <rguenther@suse.de>
20522 PR tree-optimization/81090
20523 * passes.def (pass_record_bounds): Remove.
20524 * tree-pass.h (make_pass_record_bounds): Likewise.
20525 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20526 make_pass_record_bounds): Likewise.
20527 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20528 not free niter estimates at the beginning but at the end.
20529 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20531 2017-06-16 Richard Biener <rguenther@suse.de>
20533 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20534 initializer to workaround ICE in host GCC 4.8.
20536 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20538 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20540 (clone_inlined_nodes): Update.
20542 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20544 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20545 prefetch settings, and enable prefetching by default at -O3.
20547 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20549 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20550 Set flag_prefetch_loop_arrays according to tuning data.
20552 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20554 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20555 New tune structure.
20556 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20557 [Unrelated to main purpose of the patch] Place the pointer field last
20558 to enable type checking errors when tune structure are wrongly merged.
20559 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20560 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20561 (thunderx2t99_prefetch_tune): New tune constants.
20562 (tune_params *_tunings): Update all tunings (no functional change).
20563 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20564 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20565 from tunings structures.
20567 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20570 * ubsan.c (instrument_null): Add T argument, use it instead
20571 of computing it based on IS_LHS.
20572 (instrument_object_size): Likewise.
20573 (pass_ubsan::execute): Adjust instrument_null and
20574 instrument_object_size callers to pass gimple_get_lhs or
20575 gimple_assign_rhs1 result to it. Use instrument_null instead of
20576 calling get_base_address and instrument_mem_ref. Handle
20577 aggregate call arguments for object-size sanitization.
20579 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20581 PR tree-optimization/81089
20582 * tree-vrp.c (is_masked_range_test): Validate operands of
20585 2017-06-15 Martin Sebor <msebor@redhat.com>
20588 * dumpfile.c (dump_register): Avoid calling memset to initialize
20589 a class with a default ctor.
20590 * gcc.c (struct compiler): Remove const qualification.
20591 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20592 * hash-table.h: Ditto.
20593 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20595 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20596 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20598 * params.h (struct param_info): Make struct members non-const.
20599 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20600 with default initialization.
20601 * vec.h (vec_copy_construct, vec_default_construct): New helper
20603 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20604 with vec_copy_construct.
20605 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20606 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20607 * doc/invoke.texi (-Wclass-memaccess): Document.
20609 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20611 * emit-rtl.h (is_leaf): Update comment about local
20612 register allocator.
20614 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20617 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20618 for a variable to have a section before checking if the section has a
20620 Set section to.persistent if persistent attribute is set.
20621 Warn if .persistent attribute is used on an automatic variable.
20623 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20625 PR rtl-optimization/80474
20626 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20628 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20630 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20633 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20635 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20636 (casesi): Emit bounds checking as RTL.
20637 (casesi_internal_mips16_<mode>): Remove bounds checking.
20639 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20641 * config/xtensa/xtensa.c (xtensa_option_override): Append
20642 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20643 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20644 xtensa_doloop_hooks): Define unconditionally.
20645 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20647 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20648 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20649 for it in xtensa_option_override.
20650 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20651 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20653 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20655 * doc/cppopts.texi: Document '-' special value to -MF.
20657 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20659 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20660 (cortex_a53_fconst): Likewise.
20661 (cortex_a53_fpmul): Likewise.
20662 (cortex_a53_f_load_64): Likewise.
20663 (cortex_a53_f_load_many): Likewise.
20664 (cortex_a53_advsimd_alu): Likewise.
20665 (cortex_a53_advsimd_alu_q): Likewise.
20666 (cortex_a53_advsimd_mul): Likewise.
20667 (cortex_a53_advsimd_mul_q): Likewise.
20668 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20669 Add missing fmul, r2f_cvt and fconst cases.
20671 2017-06-14 Richard Biener <rguenther@suse.de>
20673 PR middle-end/81088
20674 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20676 (fold_binary_loc): When associating do not treat pre-existing
20677 TREE_OVERFLOW on literal constants as a reason to allow
20678 TREE_OVERFLOW on associated literal constants.
20680 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20682 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20683 (MASK_FEATURES): New macro.
20684 * config/sparc/sparc.c (sparc_option_override): Remove the special
20685 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20687 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20689 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20690 a division of 0 if non-call exceptions are enabled.
20692 2017-06-14 Andrew Pinski <apinski@cavium.com>
20693 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20696 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20697 Improve vector initialization code gen for only variable case.
20699 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20701 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20703 2017-06-14 Richard Biener <rguenther@suse.de>
20705 PR tree-optimization/81083
20706 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20709 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20711 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20712 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20713 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20714 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20715 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20716 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20718 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20720 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20721 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20723 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20725 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20727 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20729 * config/rs6000/t-rtems: Don't handle SPE.
20731 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20733 * config/rs6000/t-linux: Don't handle SPE.
20735 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20737 * config/rs6000/eabispe.h: Delete file.
20739 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20741 * config/rs6000/t-spe: Delete file.
20743 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20745 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20746 (rs6000_legitimate_offset_address_p): Return false for anything in
20747 V2SImode or V2SFmode.
20749 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20751 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20752 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20753 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20755 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20756 (rs6000_legitimate_offset_address_p): Ditto.
20757 (rs6000_emit_move): Ditto.
20758 (rs6000_init_builtins): Remove V4HI_type_node.
20760 2017-06-13 Martin Liska <mliska@suse.cz>
20763 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20764 (gate_asan): Likewise.
20765 * asan.h (asan_no_sanitize_address_p): Remove the function.
20766 (sanitize_flags_p): New function.
20767 * builtins.def: Fix coding style.
20768 * common.opt: Use renamed enum value.
20769 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20770 * doc/extend.texi: Document no_sanitize attribute.
20771 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20772 to SANITIZE_UNDEFINED_NONDEFAULT.
20773 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20774 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20775 Use sanitize_flags_p.
20776 * gimplify.c (gimplify_function_tree): Likewise.
20777 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20778 * opts.c (parse_no_sanitize_attribute): New function.
20779 (common_handle_option): Use renamed enum value.
20780 * opts.h (parse_no_sanitize_attribute): Declare.
20781 * tree.c (sanitize_flags_p): New function.
20782 * tree.h: Declared here.
20783 * tsan.c: Use sanitize_flags_p.
20784 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20785 (instrument_mem_ref): Likewise.
20786 (instrument_bool_enum_load): Likewise.
20787 (do_ubsan_in_current_function): Remove the function.
20788 (pass_ubsan::execute): Use sanitize_flags_p.
20789 * ubsan.h: Remove do_ubsan_in_current_function
20790 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20791 (dump_function_to_file): Use it here.
20793 2017-06-13 Martin Jambor <mjambor@suse.cz>
20795 PR tree-optimization/80803
20796 PR tree-optimization/81063
20797 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20798 (propagate_subaccesses_across_link): Enqueue subtree whenever
20799 necessary instead of relying on the caller.
20801 2017-06-13 Martin Jambor <mjambor@suse.cz>
20803 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20804 that have a first_link.
20805 (sort_and_splice_var_accesses): Do not check first_link before
20807 (subtree_mark_written_and_enqueue): Likewise.
20808 (propagate_all_subaccesses): Likewise and do not stop at first
20809 parent with a first_link.
20811 2017-06-13 Martin Jambor <mjambor@suse.cz>
20813 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20816 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20818 * match.pd: New pattern.
20820 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20822 * tree-vrp.c (is_masked_range_test): New function.
20823 (register_edge_assert_for): Determine ranges for
20826 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20828 PR tree-optimization/67328
20829 * fold-const.c (maskable_range_p): New function.
20830 (build_range_check): Generate bittests if possible.
20832 2017-06-13 Martin Liska <mliska@suse.cz>
20834 * gimple-pretty-print.c (dump_probability): Add new argument.
20835 (dump_edge_probability): Dump both probability and count.
20836 (dump_gimple_label): Likewise.
20837 (dump_gimple_bb_header): Likewise.
20839 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20842 * config/avr/avr-devices.c: Fix indentation.
20843 * config/avr/gen-avr-mmcu-specs.c: Dito.
20845 2017-06-13 Richard Biener <rguenther@suse.de>
20847 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20848 instead get vector type from stmt_info.
20849 (vectorizable_reduction): Adjust. Remove dead code.
20851 2017-06-13 Richard Biener <rguenther@suse.de>
20853 PR middle-end/81065
20854 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20855 case of C * (x * C2 + C3).
20856 (fold_addr_of_array_ref_difference): Properly fold index difference.
20858 2017-06-12 David S. Miller <davem@davemloft.net>
20861 * config/sparc/sparc.md (return expander): Emit frame blockage if
20862 function uses alloca.
20864 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20866 * combine.c (make_field_assignment): Check len rather than the mode
20867 precision when calling force_to_mode.
20869 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20871 Support multilibs and devices that see flash in RAM address range.
20874 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20875 (avr_mcu_t) <flash_pm_offset>: New field.
20876 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20877 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20878 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20879 (AVR_TINY_PM_OFFSET): Remove macro.
20880 * config/avr/avr.opt (-mshort-calls): New option.
20881 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20882 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20883 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20884 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20885 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20886 instead of avr_arch->have_jmp_call.
20887 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20888 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20889 avr_arch->flash_pm_offset to define.
20890 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20891 new field flash_pm_offset. Add entry for avrxmega3.
20892 (avr_texinfo): Add entry for avrxmega3.
20893 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20894 attiny212, attiny214,
20895 attiny412, attiny414, attiny416, attiny417,
20896 attiny814, attiny816, attiny817,
20897 attiny1614, attiny1616, attiny1617,
20898 attiny3214, attiny3216, attiny3217.
20899 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20900 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20901 (avr_print_operand_address) [AVR_TINY]: Same.
20902 (avr_asm_init_sections) <readonly_data_section>: Only patch
20903 callback if avr_arch->flash_pm_offset = 0.
20904 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20905 for rodata if avr_arch->flash_pm_offset != 0.
20906 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20907 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20908 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20909 (m_options): Append opt_rcall.
20910 (m_dirnames): Append dir_rcall.
20911 * config/avr/t-multilib: Regenerate.
20913 * configure.ac [target=avr]: Check whether avrxmega3 default
20914 linker description file works as needed.
20915 * configure: Regenerate.
20916 * doc/avr-mmcu.texi: Regenerate.
20917 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20918 <__AVR_ARCH__>: Document avrxmega3 and 103.
20919 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20920 <__AVR_SHORT_CALLS__>: Document it.
20921 <__AVR_PM_BASE_ADDRESS__>: Document it.
20922 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20923 (AVR Variable Attributes) <progmem>: Document this is
20924 not needed for avrxmega3.
20925 (AVR Named Address Spaces) <__flash>: Dito.
20927 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20929 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20931 2017-06-12 Doug Rupp <rupp@adacore.com>
20933 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20934 Append vxworks-stdint.h to the tm_file list.
20935 * config/vxworks-stdint.h: New file.
20937 2017-06-12 Martin Liska <mliska@suse.cz>
20939 PR tree-optimization/81041
20940 * tree-profile.c (gimple_gen_ic_func_profiler):
20941 Create an extra BB in profile-generate
20942 (gimple_gen_time_profiler): Likewise.
20944 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20946 PR tree-optimization/81003
20947 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20948 (update_range_test): Use it instead of force_gimple_operand_gsi.
20950 2017-06-12 Richard Biener <rguenther@suse.de>
20952 PR tree-optimization/81053
20953 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20954 with backedge value not defined in loop. Simplify def stmt
20957 2017-06-11 Tom de Vries <tom@codesourcery.com>
20960 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20962 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20963 nvptx_cannot_force_const_mem.
20965 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20967 * opts.c (finish_options): Move test for flag_split_stack after
20968 it has been initialized.
20970 2017-06-11 Jason Merrill <jason@redhat.com>
20972 * tree.h (id_equal): New.
20973 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20974 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20975 instead of strcmp of IDENTIFIER_POINTER.
20977 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20979 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20980 (mark_all_inlined_calls_cdtor): Fix formating.
20981 (inline_transform): Rescale profile before inlining.
20983 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20985 * cgraph.h (cgraph_edge::clone): Update prototype.
20986 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20987 (cgraph_node::create_clone): Update.
20988 (cgraph_node::create_version_clone): Update.
20989 * tree-inline.c (copy_bb): Update.
20990 (expand_call_inline): Update.
20992 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20994 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20995 factored out from ...
20996 (rs6000_emit_prologue): ... here.
20998 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21000 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21001 factored out from ...
21002 (rs6000_emit_prologue): ... here.
21004 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21006 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21008 (handle_missing_profiles): Fix computation of tp_first_run.
21009 (counts_to_freqs): Do not touch freqs when count is 0.
21011 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21013 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21016 2017-06-10 Tom de Vries <tom@codesourcery.com>
21018 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21019 attributes): Document signal effective target.
21021 2017-06-10 Tom de Vries <tom@codesourcery.com>
21023 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21024 Document effective target stack_size.
21026 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21028 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21029 to the edit_context if they can be auto-applied.
21031 2017-06-9 Ian Lance Taylor <iant@golang.org>
21033 * opts.c (finish_options): If -fsplit-stack, disable implicit
21034 -forder-blocks-and-partition.
21035 * doc/invoke.texi (Optimize Options): Document that when using
21036 -fsplit-stack -forder-blocks-and-partition is not implicitly
21039 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21041 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21042 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21043 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21044 * builtins.def (abort, trap, unreachable): Declare cold.
21045 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21046 * tree-core.h (ECF_COLD): New.
21047 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21048 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21050 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21052 * predict.c (unlikely_executed_stmt_p): Cleanup.
21054 2017-06-09 Richard Biener <rguenther@suse.de>
21056 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21057 model if the ref is always written to.
21059 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21061 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21063 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21065 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21068 2017-06-09 Tom de Vries <tom@codesourcery.com>
21071 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21072 "target cannot support label values" when encountering LABEL_REF.
21074 2017-06-09 Martin Liska <mliska@suse.cz>
21076 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21077 (gimple_gen_ic_func_profiler): Emit direct comparison
21078 of __gcov_indirect_call_callee with NULL.
21079 (gimple_gen_time_profiler): Change probability from
21080 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21082 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21084 * profile.c (edge_gcov_counts): Turn to pointer.
21085 (compute_branch_probabilities, compute_branch_probabilities): Update.
21086 (branch_prob): Do not clear edge_gcov_count.
21087 * profile.h (edge_gcov_counts): Turn to pointer.
21088 (edge_gcov_count): Update.
21090 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21092 * gimple.h (gimple_check_failed): Mark cold.
21094 2017-06-09 Richard Biener <rguenther@suse.de>
21096 PR tree-optimization/66623
21097 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21098 refactor check_reduction into two parts, properly computing
21099 whether we have to check reduction validity for outer loop
21102 2017-06-09 Richard Biener <rguenther@suse.de>
21104 PR tree-optimization/79483
21105 * graphite-scop-detection.c (order): New global.
21106 (get_order): Compute bb to order mapping that satisfies code
21107 generation constraints.
21108 (cmp_pbbs): New helper.
21109 (build_scops): Start domwalk at entry block, sort generated
21112 2017-06-09 Richard Biener <rguenther@suse.de>
21114 PR middle-end/81007
21115 * ipa-polymorphic-call.c
21116 (ipa_polymorphic_call_context::restrict_to_inner_class):
21117 Skip FIELD_DECLs with error_mark_node type.
21118 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21121 2017-06-09 Martin Liska <mliska@suse.cz>
21123 * predict.c (struct branch_predictor): New struct.
21124 (test_prediction_value_range): New test.
21125 (predict_c_tests): New function.
21126 * selftest-run-tests.c (selftest::run_tests): Run the function.
21127 * selftest.h: Declare new tests.
21129 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21132 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21133 gen_add3_insn did not fail.
21134 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21135 r0, construct that number in a temporary reg and add that reg to r0.
21136 If asked to put the result in r0 as well, fail.
21138 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21140 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21141 for early expansion of vec_eqv.
21143 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21145 PR middle-end/81005
21146 * ubsan.c (instrument_null): Avoid pointless code temporary.
21147 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21149 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21153 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21156 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21159 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21161 * predict.c (unlikely_executed_edge_p): Move ahead.
21162 (probably_never_executed_edge_p): Use it.
21164 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21166 PR middle-end/79988
21167 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21168 gimple_call_builtin_p call.
21170 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21172 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21173 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21174 rtl_check_failed_type2, rtl_check_failed_code1,
21175 rtl_check_failed_code2, rtl_check_failed_code_mode,
21176 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21177 rtvec_check_failed_bounds, rtl_check_failed_flag,
21178 _fatal_insn_not_found, _fatal_insn): Likewise.
21179 * tree.h (tree_contains_struct_check_failed,
21180 tree_check_failed, tree_not_check_failed,
21181 tree_class_check_failed, tree_range_check_failed,
21182 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21183 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21184 tree_operand_check_failed, omp_clause_check_failed,
21185 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21188 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21190 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21191 flag_branch_probabilities.
21192 * ipa-inline.c (edge_badness): Likewise.
21193 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21194 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21195 * predict.c (maybe_hot_frequency_p): Likewise.
21196 (probably_never_executed): Likewise.
21197 * sched-ebb.c (schedule_ebbs): Likewise.
21198 * sched-rgn.c (find_single_block_region): Likewise.
21199 * tracer.c (tail_duplicate): Likewise.
21201 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21203 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21204 longer requires x_flag_profile_use.
21206 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21208 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21209 instead of flag_reorder_blocks_and_partition.
21210 * dbxout.c (dbxout_function_end): Likewise.
21211 * dwarf2out.c (gen_subprogram_die): Likewise.
21212 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21213 * hw-doloop.c (reorg_loops): Likewise.
21214 * varasm.c (assemble_start_function,
21215 assemble_end_function): Likewise.
21216 (decide_function_section): Do not check for
21217 flag_reorder_blocks_and_partition.
21219 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21221 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21223 (chkp_get_hard_register_fake_addr_expr): Ditto.
21224 (chkp_build_addr_expr): Add check for hard reg case.
21225 (chkp_parse_array_and_component_ref): Ditto.
21226 (chkp_find_bounds_1): Ditto.
21227 (chkp_process_stmt): Don't generate bounds store for
21230 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21232 * predict.c (maybe_hot_bb_p): Do not check profile status.
21233 (maybe_hot_edge_p): Likewise.
21234 (probably_never_executed): Check for zero counts even if profile
21236 (unlikely_executed_edge_p): New function.
21237 (unlikely_executed_stmt_p): New function.
21238 (unlikely_executed_bb_p): New function.
21239 (set_even_probabilities): Use unlikely predicates.
21240 (combine_predictions_for_bb): Likewise.
21241 (predict_paths_for_bb): Likewise.
21242 (predict_paths_leading_to_edge): Likewise.
21243 (determine_unlikely_bbs): New function.
21244 (estimate_bb_frequencies): Use it.
21245 (compute_function_frequency): Use zero counts even if profile is
21247 * profile-count.h: Fix typo.
21249 2017-08-08 Julia Koval <julia.koval@intel.com>
21251 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21252 _mm512_mask_cvtsepi16_storeu_epi8,
21253 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21254 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21255 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21256 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21257 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21258 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21259 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21260 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21261 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21262 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21263 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21264 __builtin_ia32_pmovuswb256mem_mask,
21265 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21266 __builtin_ia32_pmovwb512mem_mask): New builtins.
21268 2017-08-08 Julia Koval <julia.koval@intel.com>
21270 PR target/73350,80862
21271 * config/i386/subst.md (round): Fix round pattern.
21272 * config/i386/i386.c (ix86_erase_embedded_rounding):
21273 Fix erasing rounding for the fixed pattern.
21275 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21277 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21279 2017-06-08 Martin Liska <mliska@suse.cz>
21281 PR gcov-profile/80911
21282 * gcov.c (block_info::block_info): New constructor.
21284 2017-06-07 Carl Love <cel@us.ibm.com>
21286 * config/rs6000/rs6000-c: The return type of the following
21287 built-in functions was implemented as int not long long. Fix sign
21288 of return value for the unsigned version of vec_mulo and vec_mule.
21289 vector unsigned long long vec_bperm (vector unsigned long long,
21290 vector unsigned char)
21291 vector signed long long vec_mule (vector signed int,
21293 vector unsigned long long vec_mule (vector unsigned int,
21294 vector unsigned int)
21295 vector signed long long vec_mulo (vector signed int,
21297 vector unsigned long long vec_mulo (vector unsigned int,
21298 vector unsigned int)
21299 * doc/extend.texi: Fix the documentation for the built-in
21302 2017-06-07 Carl Love <cel@us.ibm.com>
21305 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21308 2017-06-07 Carl Love <cel@us.ibm.com>
21310 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21311 support, Generate doublehv for signed int/float for BE case only.
21313 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21315 * doc/invoke.texi (mcx16): Rewrite.
21317 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21319 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21320 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21321 *mov<mode>_softfloat, and an anonymous splitter): Use
21322 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21324 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21326 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21328 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21329 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21330 (rs6000_debug_reg_global): Adjust.
21331 (rs6000_init_hard_regno_mode_ok): Adjust.
21332 (rs6000_dbx_register_number): Adjust.
21333 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21334 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21335 Remove SPE_ACC and SPEFSCR.
21336 (REG_ALLOC_ORDER): Ditto.
21337 (FRAME_POINTER_REGNUM): Change to 111.
21338 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21339 (REG_CLASS_NAMES): Ditto.
21340 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21341 (REGISTER_NAMES): Ditto.
21342 (ADDITIONAL_REG_NAMES): Ditto.
21343 (rs6000_reg_names): Ditto.
21344 * config/rs6000/rs6000.md: Renumber some register number
21347 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21349 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21351 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21352 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21354 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21355 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21356 Delete the SPE high registers.
21357 (REG_ALLOC_ORDER): Ditto.
21358 (enum reg_class): Remove SPE_HIGH_REGS.
21359 (REG_CLASS_NAMES): Ditto.
21360 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21361 (REGISTER_NAMES): Ditto.
21362 (rs6000_reg_names): Ditto.
21363 * doc/tm.texi.in: Remove SPE as example.
21364 * doc/tm.texi: Regenerate.
21366 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21368 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21369 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21370 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21371 * config/rs6000/rs6000.md (type): Remove "brinc".
21373 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21375 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21376 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21377 * config/rs6000/linuxspe.h: Delete file.
21378 * config/rs6000/rs6000.md: Don't include spe.md.
21379 * config/rs6000/spe.h: Delete file.
21380 * config/rs6000/spe.md: Delete file.
21381 * config/rs6000/t-rs6000: Remove spe.md.
21383 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21385 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21386 (reg_or_none500mem_operand): Delete.
21387 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21388 instead of reg_or_none500mem_operand.
21390 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21392 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21393 handling of SPE flags.
21394 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21396 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21398 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21400 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21401 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21402 paired_divv2sf3): Similar.
21403 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21404 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21405 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21407 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21408 Rename the paired_* instruction patterns.
21409 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21411 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21412 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21413 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21414 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21415 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21416 PAIRED_VECTOR_MODE.
21417 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21418 (spe_func_has_64bit_regs_p): Delete.
21419 (spe_expand_predicate_builtin): Delete.
21420 (spe_expand_evsel_builtin): Delete.
21421 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21422 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21423 (invalid_e500_subreg): Delete.
21424 (rs6000_legitimize_address): Always force_reg op2 as well, for
21425 paired single memory accesses.
21426 (rs6000_member_type_forces_blk): Delete.
21427 (rs6000_spe_function_arg): Delete.
21428 (rs6000_expand_unop_builtin): Delete SPE handling.
21429 (rs6000_expand_binop_builtin): Ditto.
21430 (spe_expand_stv_builtin): Delete.
21431 (bdesc_2arg_spe): Delete.
21432 (spe_expand_builtin): Delete.
21433 (spe_expand_predicate_builtin): Delete.
21434 (spe_expand_evsel_builtin): Delete.
21435 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21436 (spe_init_builtins): Delete.
21437 (spe_func_has_64bit_regs_p): Delete.
21438 (savres_routine_name): Delete "info" parameter. Adjust callers.
21439 (rs6000_emit_stack_reset): Ditto.
21440 (rs6000_dwarf_register_span): Delete.
21441 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21442 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21443 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21445 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21447 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21448 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21449 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21450 mulv2sf3, divv2sf3): Delete expanders.
21452 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21454 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21456 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21458 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21459 * config/rs6000/rs6000.c: Ditto.
21461 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21463 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21464 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21465 comparison_operator.
21467 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21469 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21470 * config/rs6000/rs6000.opt: Ditto.
21471 * config/rs6000/t-rtems: Ditto.
21473 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21475 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21476 TARGET_E500_SINGLE by 0, simplify.
21477 * config/rs6000/rs6000.c: Ditto.
21478 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21479 (spe_build_register_parallel): Delete.
21480 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21481 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21482 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21483 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21484 (E500_CONVERT): Delete.
21485 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21487 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21489 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21490 * config/rs6000/dfp.md: Ditto.
21491 (negdd2, *negdd2_fpr): Merge.
21492 (absdd2, *absdd2_fpr): Merge.
21493 (negtd2, *negtd2_fpr): Merge.
21494 (abstd2, *abstd2_fpr): Merge.
21495 * config/rs6000/e500.h: Delete file.
21496 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21497 TARGET_FPRS by 1 and simplify.
21498 * config/rs6000/rs6000-c.c: Ditto.
21499 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21500 TARGET_DF_SPE by 0.
21501 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21503 * config/rs6000/rs6000.md: Ditto.
21504 (floatdidf2, *floatdidf2_fpr): Merge.
21505 (move_from_CR_gt_bit): Delete.
21506 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21507 (E500_CR_IOR_COMPARE): Delete.
21508 (All patterns that require !TARGET_FPRS): Delete.
21509 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21511 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21513 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21515 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21517 * graphds.c (add_edge): Intitialize edge's attached data.
21518 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21519 pointer parameter. Call pointed function on each edge during
21520 graph traversing. Skip traversing the edge when the function
21522 (graphds_dfs, graphds_scc): Ditto.
21523 (for_each_edge): New parameter. Pass the new parameter to callback
21525 * graphds.h (skip_edge_callback): New function pointer type.
21526 (graphds_dfs, graphds_scc): New function pointer parameter.
21527 (graphds_edge_callback, for_each_edge): New parameter.
21529 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21531 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21532 out code checking if runtime alias check is possible to below ...
21533 Call the new function.
21534 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21535 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21537 2017-06-07 Marek Polacek <polacek@redhat.com>
21540 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21541 TYPE_OVERFLOW_WRAPS check.
21543 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21545 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21546 if versioning is required.
21547 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21548 peeling with the check for versioning.
21550 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21552 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21553 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21554 Set true to new parameter if new ssa variable is defined.
21555 (vect_gen_vector_loop_niters): Refactor. Set range information
21556 for the new vector loop bound variable.
21557 (vect_do_peeling): Ditto.
21559 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21561 * tree-affine.c (ssa.h): Include header file.
21562 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21563 has wrapping overflow behavior.
21565 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21567 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21569 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21571 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21572 (tree_to_aff_combination): ... here.
21574 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21576 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21577 reg_pressure model function.
21578 (ivopts_global_cost_for_size): Delete.
21579 (determine_set_costs, iv_ca_recount_cost): Call new model function
21580 ivopts_estimate_reg_pressure.
21582 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21584 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21585 expensive than udiv. Remove floating point cases from mod.
21587 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21589 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21590 Increase idiv cost.
21592 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21594 * config/aarch64/aarch64.md
21595 (copysignsf3): Fix mask generation.
21597 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21599 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21601 (class dump_manager): Add register_dumps method.
21602 * dumpfile.c: Include langhooks.h.
21603 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21604 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21605 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21606 (dump_manager::dump_register): Start with 512 entries instead of 32.
21607 (dump_manager::register_dumps): New method.
21608 * toplev.c (general_init): Instead of invoking register_dumps
21609 langhook, invoke register_dumps method on the dump manager.
21610 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21613 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21615 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21616 Say that # requires an associated define_split to exist, and that
21617 the define_split must be suitable for use after register allocation.
21619 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21621 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21622 (compute_outgoing_frequencies): Also initialize zero counts.
21623 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21624 around loops; preserve more of profile when nothing changes.
21626 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21628 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21630 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21631 * config/arm/arm-cpu-cdata.h: Regenerate.
21632 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21633 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21634 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21635 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21636 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21638 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21640 * config/arm/t-rmprofile: Likewise.
21641 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21643 2017-06-06 David S. Miller <davem@davemloft.net>
21646 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21647 blockage if function uses alloca.
21649 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21651 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21652 New "uid" fields to hold pretty-print IDs of group and ref.
21653 Memory references are now identified as <group_id>:<ref_id>
21654 instead of using [random] addresses.
21655 (dump_mem_details): Simplify, no functional change.
21656 (dump_mem_ref): Simplify and make output more concise.
21657 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21658 (find_or_create_group): Initialize group uid.
21659 (record_ref): Initialize ref uid. Improve debug output.
21660 (prune_group_by_reuse, should_issue_prefetch_p,)
21661 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21662 (mark_nontemporal_store, determine_loop_nest_reuse):
21663 Improve debug output.
21665 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21667 * dbgcnt.def (prefetch): New debug counter.
21668 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21669 (schedule_prefetches): Stop issueing prefetches if debug counter
21672 2017-06-06 Tom de Vries <tom@codesourcery.com>
21674 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21675 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21677 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21679 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21680 Use aarch64_reg_or_zero predicate for operand 4.
21681 (aarch64_compare_and_swap<mode> define_insn_and_split):
21682 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21683 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21685 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21687 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21688 (arm_compute_save_core_reg_mask): This.
21689 (thumb1_compute_save_reg_mask): Rename into ...
21690 (thumb1_compute_save_core_reg_mask): This.
21691 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21692 (arm_compute_frame_layout): Likewise.
21694 2017-06-06 Richard Biener <rguenther@suse.de>
21696 PR tree-optimization/80974
21697 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21698 keep or clear leaders SSA info.
21700 2017-06-06 Tom de Vries <tom@codesourcery.com>
21702 * config/nvptx/nvptx.c (split_mode_p): New function.
21703 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21705 2017-06-06 Tom de Vries <tom@codesourcery.com>
21707 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21709 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21712 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21715 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21717 * shrink-wrap.c (handle_simple_exit): Update profile.
21718 (try_shrink_wrapping): Upate profile.
21720 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21722 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21723 (tree_guess_outgoing_edge_probabilities): New.
21724 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21725 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21727 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21729 * ipa-split.c (split_function): Initialize return bb profile.
21731 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21733 * profile.c (compute_branch_probabilities): Also initialize
21734 EXIT_BLOCK profile.
21736 2017-06-06 Richard Biener <rguenther@suse.de>
21738 PR tree-optimization/80928
21739 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21740 (vect_analyze_loop_operations): Properly guard analysis for
21742 (vect_transform_loop): Likewise.
21743 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21744 (vect_model_induction_cost): Do not cost for pure SLP.
21745 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21746 of induction in inner loop vectorization.
21747 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21748 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21749 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21751 (vect_analyze_slp_cost_1): Cost induction.
21752 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21753 (vect_get_slp_vect_defs): Likewise.
21754 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21755 (vect_transform_stmt): Handle SLP reductions.
21756 * tree-vectorizer.h (vectorizable_induction): Adjust.
21758 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21760 * config/rs6000/rs6000.c (make_resolver_func): Update
21761 init_lowered_empty_function call.
21763 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21765 * doc/invoke.texi: Document the -fprofile-abs-path option.
21766 * common.opt (fprofile-abs-path): New option.
21767 * gcov-io.h (gcov_write_filename): Declare.
21768 * gcov-io.c (gcov_write_filename): New function.
21769 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21770 * profile.c (output_location): Likewise.
21772 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21774 * shring-wrap.c: Revert accidental commit.
21776 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21778 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21780 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21782 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21784 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21785 profile in callgraph edge.
21786 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21787 (apply_scale): Likewise.
21788 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21789 Also scale profile when inlining function with zero profile.
21790 (initialize_cfun): Update exit block profile even when it is zero.
21791 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21792 when profile is read.
21794 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21796 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21797 (CLONE_*): New constants to define the processors we can generate
21798 code for with the target_clone attribute.
21799 (rs6000_clone_map): New array to identify which clone processors
21800 the current program is running on.
21801 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21802 target_clone attribute.
21803 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21804 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21805 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21806 (cpu_expand_builtin): Add support for target_clone attribute.
21807 (rs6000_valid_attribute_p): Allow "default" attribute.
21808 (get_decl_name): New debug function to simplify printing the
21809 current function name in debugging statements.
21810 (rs6000_clone_priority): New functions to support the target_clone
21811 attribute, and be able to generate code to switch between ISA 2.05
21812 through ISA 3.0 (power6 through power9).
21813 (rs6000_compare_version_priority): Likewise.
21814 (rs6000_get_function_versions_dispatcher): Likewise.
21815 (make_resolver_func): Likewise.
21816 (add_condition_to_bb): Likewise.
21817 (dispatch_function_versions): Likewise.
21818 (rs6000_generate_version_dispatcher_body): Likewise.
21819 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21820 (fusion_gpr_load_p): Fix a spacing issue.
21821 * doc/extend.texi (Common Function Attributes): Document that the
21822 PowerPC supports the target_clone attribute.
21824 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21826 * config/arm/arm.h: explain F symbol found in description of ARM
21827 register allocation in its legend.
21829 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21831 * config/mips/frame-header-opt.c: Include profile-count.h.
21832 * config/riscv/riscv.c: Include profile-count.h
21834 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21836 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21838 (sm_set_flag_if_changed): Add bbs field.
21839 (execute_sm_if_changed_flag_set): Pass BBS.
21840 (execute_sm): Update.
21842 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21844 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21847 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21849 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21850 (peephole2): New peephole2 to emit the above.
21851 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21853 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21855 * config/aarch64/aarch64.c (define_peephole2 above
21856 *sub_<shift>_<mode>): New peephole.
21858 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21860 * config/i386/i386.c (make_resolver_func): Update.
21861 * Makefile.in: Add profile-count.h and profile-count.o
21862 * auto-profile.c (afdo_indirect_call): Update to new API.
21863 (afdo_set_bb_count): Update.
21864 (afdo_propagate_edge): Update.
21865 (afdo_propagate_circuit): Update.
21866 (afdo_calculate_branch_prob): Update.
21867 (afdo_annotate_cfg): Update.
21868 * basic-block.h: Include profile-count.h
21869 (struct edge_def): Turn count to profile_count.
21870 (struct basic_block_def): Likewie.
21871 (REG_BR_PROB_BASE): Move to profile-count.h
21872 (RDIV): Move to profile-count.h
21873 * bb-reorder.c (max_entry_count): Turn to profile_count.
21874 (find_traces): Update.
21875 (rotate_loop):Update.
21876 (connect_traces):Update.
21877 (sanitize_hot_paths):Update.
21878 * bt-load.c (migrate_btr_defs): Update.
21879 * cfg.c (RDIV): Remove.
21880 (init_flow): Use alloc_block.
21881 (alloc_block): Uninitialize count.
21882 (unchecked_make_edge): Uninitialize count.
21883 (check_bb_profile): Update.
21884 (dump_edge_info): Update.
21885 (dump_bb_info): Update.
21886 (update_bb_profile_for_threading): Update.
21887 (scale_bbs_frequencies_int): Update.
21888 (scale_bbs_frequencies_gcov_type): Update.
21889 (scale_bbs_frequencies_profile_count): New.
21890 * cfg.h (update_bb_profile_for_threading): Update.
21891 (scale_bbs_frequencies_profile_count): Declare.
21892 * cfgbuild.c (compute_outgoing_frequencies): Update.
21893 (find_many_sub_basic_blocks): Update.
21894 * cfgcleanup.c (try_forward_edges): Update.
21895 (try_crossjump_to_edge): Update.
21896 * cfgexpand.c (expand_gimple_tailcall): Update.
21897 (construct_exit_block): Update.
21898 * cfghooks.c (verify_flow_info): Update.
21899 (dump_bb_for_graph): Update.
21900 (split_edge): Update.
21901 (make_forwarder_block): Update.
21902 (duplicate_block): Update.
21903 (account_profile_record): Update.
21904 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21905 (get_estimated_loop_iterations): Update.
21906 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21907 (single_likely_exit): Update.
21908 * cfgloopmanip.c (scale_loop_profile): Update.
21910 (set_zero_probability): Update.
21911 (lv_adjust_loop_entry_edge): Update.
21912 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21913 (purge_dead_edges): Update.
21914 (rtl_account_profile_record): Update.
21915 * cgraph.c (cgraph_node::create): Uninitialize count.
21916 (symbol_table::create_edge): Uninitialize count.
21917 (cgraph_update_edges_for_call_stmt_node): Update.
21918 (cgraph_edge::dump_edge_flags): Update.
21919 (cgraph_node::dump): Update.
21920 (cgraph_edge::maybe_hot_p): Update.
21921 * cgraph.h: Include profile-count.h
21922 (create_clone), create_edge, create_indirect_edge): Update.
21923 (cgraph_node): Turn count to profile_count.
21924 (cgraph_edge0: Likewise.
21925 (make_speculative, clone): Update.
21926 (create_edge): Update.
21927 (init_lowered_empty_function): Update.
21928 * cgraphclones.c (cgraph_edge::clone): Update.
21929 (duplicate_thunk_for_node): Update.
21930 (cgraph_node::create_clone): Update.
21931 * cgraphunit.c (cgraph_node::analyze): Update.
21932 (cgraph_node::expand_thunk): Update.
21933 * final.c (dump_basic_block_info): Update.
21934 * gimple-streamer-in.c (input_bb): Update.
21935 * gimple-streamer-out.c (output_bb): Update.
21936 * graphite.c (print_global_statistics): Update.
21937 (print_graphite_scop_statistics): Update.
21938 * hsa-brig.c: Include basic-block.h.
21939 * hsa-dump.c: Include basic-block.h.
21940 * hsa-gen.c (T sum_slice): Update.
21941 (convert_switch_statements):Update.
21942 * hsa-regalloc.c: Include basic-block.h.
21943 * ipa-chkp.c (chkp_produce_thunks): Update.
21944 * ipa-cp.c (struct caller_statistics): Update.
21945 (init_caller_stats): Update.
21946 (gather_caller_stats): Update.
21947 (ipcp_cloning_candidate_p): Update.
21948 (good_cloning_opportunity_p): Update.
21949 (get_info_about_necessary_edges): Update.
21950 (dump_profile_updates): Update.
21951 (update_profiling_info): Update.
21952 (update_specialized_profile): Update.
21953 (perhaps_add_new_callers): Update.
21954 (decide_about_value): Update.
21955 (ipa_cp_c_finalize): Update.
21956 * ipa-devirt.c (struct odr_type_warn_count): Update.
21957 (struct decl_warn_count): Update.
21958 (struct final_warning_record): Update.
21959 (possible_polymorphic_call_targets): Update.
21960 (ipa_devirt): Update.
21961 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21962 * ipa-icf.c (sem_function::merge): Update.
21963 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21964 * ipa-inline.c (compute_uninlined_call_time): Update.
21965 (compute_inlined_call_time): Update.
21966 (want_inline_small_function_p): Update.
21967 (want_inline_self_recursive_call_p): Update.
21968 (edge_badness): Update.
21969 (lookup_recursive_calls): Update.
21970 (recursive_inlining): Update.
21971 (inline_small_functions): Update.
21972 (dump_overall_stats): Update.
21973 (dump_inline_stats): Update.
21974 * ipa-profile.c (ipa_profile_generate_summary): Update.
21975 (ipa_propagate_frequency): Update.
21976 (ipa_profile): Update.
21977 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21978 * ipa-utils.c (ipa_merge_profiles): Update.
21979 * loop-doloop.c (doloop_modify): Update.
21980 * loop-unroll.c (report_unroll): Update.
21981 (unroll_loop_runtime_iterations): Update.
21982 * lto-cgraph.c (lto_output_edge): Update.
21983 (lto_output_node): Update.
21984 (input_node): Update.
21985 (input_edge): Update.
21986 (merge_profile_summaries): Update.
21987 * lto-streamer-in.c (input_cfg): Update.
21988 * lto-streamer-out.c (output_cfg): Update.
21989 * mcf.c (create_fixup_graph): Update.
21990 (adjust_cfg_counts): Update.
21991 (sum_edge_counts): Update.
21992 * modulo-sched.c (sms_schedule): Update.
21993 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21994 * predict.c (maybe_hot_count_p): Update.
21995 (probably_never_executed): Update.
21996 (dump_prediction): Update.
21997 (combine_predictions_for_bb): Update.
21998 (propagate_freq): Update.
21999 (handle_missing_profiles): Update.
22000 (counts_to_freqs): Update.
22001 (rebuild_frequencies): Update.
22002 (force_edge_cold): Update.
22003 * predict.h: Include profile-count.h
22004 (maybe_hot_count_p, counts_to_freqs): UPdate.
22005 * print-rtl-function.c: Do not include cfg.h
22006 * print-rtl.c: Include basic-block.h
22007 * profile-count.c: New file.
22008 * profile-count.h: New file.
22009 * profile.c (is_edge_inconsistent): Update.
22010 (correct_negative_edge_counts): Update.
22011 (is_inconsistent): Update.
22012 (set_bb_counts): Update.
22013 (read_profile_edge_counts): Update.
22014 (compute_frequency_overlap): Update.
22015 (compute_branch_probabilities): Update; Initialize and deinitialize
22017 (branch_prob): Update.
22018 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22019 (edge_gcov_count): New.
22020 (bb_gcov_count): New.
22021 * shrink-wrap.c (try_shrink_wrapping): Update.
22022 * tracer.c (better_p): Update.
22023 * trans-mem.c (expand_transaction): Update.
22024 (ipa_tm_insert_irr_call): Update.
22025 (ipa_tm_insert_gettmclone_call): Update.
22026 * tree-call-cdce.c: Update.
22027 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22028 (gimple_duplicate_sese_tail): Update.
22029 (gimple_account_profile_record): Update.
22030 (execute_fixup_cfg): Update.
22031 * tree-inline.c (copy_bb): Update.
22032 (copy_edges_for_bb): Update.
22033 (initialize_cfun): Update.
22034 (freqs_to_counts): Update.
22035 (copy_cfg_body): Update.
22036 (expand_call_inline): Update.
22037 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22038 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22039 (try_unroll_loop_completely): Update.
22040 (try_peel_loop): Update.
22041 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22042 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22043 * tree-ssa-loop-split.c (connect_loops): Update.
22044 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22045 * tree-ssa-reassoc.c (branch_fixup): Update.
22046 * tree-ssa-tail-merge.c (replace_block_by): Update.
22047 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22048 (compute_path_counts): Update.
22049 (update_profile): Update.
22050 (recompute_probabilities): Update.
22051 (update_joiner_offpath_counts): Update.
22052 (estimated_freqs_path): Update.
22053 (freqs_to_counts_path): Update.
22054 (clear_counts_path): Update.
22055 (ssa_fix_duplicate_block_edges): Update.
22056 (duplicate_thread_path): Update.
22057 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22058 (struct switch_conv_info): Update.
22059 * tree-tailcall.c (decrease_profile): Update.
22060 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22061 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22062 * value-prof.c (check_counter): Update.
22063 (gimple_divmod_fixed_value): Update.
22064 (gimple_mod_pow2): Update.
22065 (gimple_mod_subtract): Update.
22066 (gimple_ic_transform): Update.
22067 (gimple_stringop_fixed_value): Update.
22068 * value-prof.h (gimple_ic): Update.
22070 2017-06-02 Carl Love <cel@us.ibm.com>
22072 * config/rs6000/rs6000-c: Add support for built-in functions
22073 vector double vec_doublee (vector signed int);
22074 vector double vec_doublee (vector unsigned int);
22075 vector double vec_doublee (vector float);
22076 vector double vec_doubleh (vector signed int);
22077 vector double vec_doubleh (vector unsigned int);
22078 vector double vec_doubleh (vector float);
22079 vector double vec_doublel (vector signed int);
22080 vector double vec_doublel (vector unsigned int);
22081 vector double vec_doublel (vector float);
22082 vector double vec_doubleo (vector signed int);
22083 vector double vec_doubleo (vector unsigned int);
22084 vector double vec_doubleo (vector float);.
22085 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22086 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22088 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22089 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22090 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22092 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22093 vec_doublel, vec_doubleh.
22094 * doc/extend.texi: Update the built-in documentation file for the
22095 new built-in functions.
22097 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22100 * ipa-inline-analysis.c (free_growth_caches): Set
22101 edge_removal_hook_holder to NULL after removing it.
22103 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22105 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22106 comparision with zero.
22108 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22109 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22110 for early expansion of vec_min and vec_max builtins.
22111 (builtin_function_type): Add min/max unsigned variants to those
22112 identified as having unsigned arguments.
22114 2017-06-02 Olivier Hainque <hainque@adacore.com>
22116 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22118 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22120 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22121 Use VALL_F16 iterator rather than VALL.
22123 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22125 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22126 Emit CBNZ inside loop when doing a strong exchange and comparing
22127 against zero. Generate the CC flags after the loop.
22129 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22131 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22132 (dl_section_ref): New.
22133 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22134 On AIX, append an expression to subtract the size of the
22135 section length to dl_section_ref.
22137 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22139 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22140 for early expansion of vector absolute builtins.
22142 2017-06-02 Richard Biener <rguenther@suse.de>
22144 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22145 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22147 2017-06-02 Richard Biener <rguenther@suse.de>
22149 PR tree-optimization/80948
22150 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22151 stmt order as well.
22153 2017-06-02 Richard Biener <rguenther@suse.de>
22155 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22157 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22158 for inductions as relevant.
22160 2017-06-02 Richard Biener <rguenther@suse.de>
22162 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22163 (vectorizable_induction): ... this. Remove dead code.
22165 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22167 * builtins. (expand_builtin_alloca): Remove second parameter and
22168 infer its value from the first parameter instead.
22169 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22171 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22173 PR rtl-optimization/80903
22174 * loop-doloop.c (add_test): Unshare sequence.
22176 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22178 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22180 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22182 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22184 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22185 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22186 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22187 (xlogue_layout::get_stub_rtx): Make static.
22188 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22189 (xlogue_layout::compute_stub_managed_regs): Rename to...
22190 (xlogue_layout::count_stub_managed_regs): ...this.
22191 (xlogue_layout::is_stub_managed_reg): New function.
22192 (xlogue_layout::m_stub_names): Rename to...
22193 (xlogue_layout::s_stub_names): ...this, make static.
22194 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22195 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22196 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22197 xlogue_layout::s_stub_names): Instantiate statics.
22198 (stub_managed_regs): Remove.
22199 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22200 (disable_call_ms2sysv_xlogues): Rename to...
22201 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22202 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22204 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22205 change after reload_completed.
22206 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22208 (ix86_expand_prologue): Likewise.
22209 (ix86_expand_epilogue): Likewise.
22210 (ix86_expand_split_stack_prologue): Likewise.
22211 (ix86_compute_frame_layout): Remove frame parameter ...
22212 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22213 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22215 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22216 (ix86_frame): Move from here ...
22217 * config/i386/i386.h (ix86_frame): ... to here.
22218 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22219 complete ix86_frame data structure instead. Remove some_ld_name.
22221 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22223 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22224 symbols that hold a DECL_VALUE_EXPR.
22226 2017-06-01 Martin Jambor <mjambor@suse.cz>
22228 PR tree-optimization/80898
22229 * tree-sra.c (process_subtree_disqualification): Removed.
22230 (disqualify_candidate): Do not acll
22231 process_subtree_disqualification.
22232 (subtree_mark_written_and_enqueue): New function.
22233 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22234 RHS has been disqualified and re-queue LHS if necessary. Apart
22235 from that, ignore disqualified RHS.
22237 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22239 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22240 address fetch for z10 or later.
22242 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22244 * config/arc/arc.md (tst_movb): Add guard when splitting.
22246 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22248 * config/arc/arc.c (arc_can_eliminate): Test against
22249 arc_frame_pointer_needed.
22251 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22253 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22254 to prevent store reordering.
22255 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22256 (type): Add block type.
22257 (stack_tie): Define special instruction to be used in
22260 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22262 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22263 constraint. It is not valid for the pattern.
22264 (noncommutative_binary_comparison): Likewise.
22266 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22268 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22271 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22273 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22274 be used by the reg-alloc.
22276 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22278 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22279 reg-alloc when having mul64 or mul32x16 instructions.
22280 (mulsidi3): Likewise.
22281 (umulsidi3): Likewise.
22282 (mulsi32x16): New pattern.
22283 (mulsi64): Likewise.
22284 (mulsidi64): Likewise.
22285 (umulsidi64): Likewise.
22286 (MUL32x16_REG): Define.
22287 (mul64_600): Use MUL32x16_REG.
22288 (mac64_600): Likewise.
22289 (umul64_600): Likewise.
22290 (umac64_600): Likewise.
22292 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22294 * config/arc/arc.md (mulsi3_700): Make it commutative.
22296 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22298 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22300 (*sign_extendsidi2_insn): Likewise for movstosw.
22302 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22304 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22305 the type of the input discriminant value. Convert the
22306 discriminant value of signedness vary.
22308 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22310 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22311 Add to -Wall section.
22313 2017-06-01 Richard Biener <rguenther@suse.de>
22315 PR middle-end/66313
22316 * fold-const.c (fold_plusminus_mult_expr): If the factored
22317 factor may be zero use a wrapping type for the inner operation.
22318 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22319 and handle moved defs.
22320 (process_assignment): Properly guard the unary op case. Return a
22321 tri-state indicating that moving the stmt before the call may allow
22322 to continue. Pass through to_move.
22323 (find_tail_calls): Handle moving unrelated defs before
22326 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22329 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22330 splitter result in the canonical way.
22332 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22334 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22335 also for 32bit target. Update insn attributes.
22336 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22338 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22340 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22341 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22342 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22343 (_mm_maskz_min_ss): New intrinsics.
22345 2017-05-31 Martin Liska <mliska@suse.cz>
22347 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22348 Change comment style to one we normally use.
22349 (vectorizable_reduction): Likewise.
22350 (vectorizable_induction): Likewise.
22351 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22352 (vectorizable_call): Likewise.
22353 (vectorizable_simd_clone_call): Likewise.
22354 (vectorizable_conversion): Likewise.
22355 (vectorizable_assignment): Likewise.
22356 (vectorizable_shift): Likewise.
22357 (vectorizable_operation): Likewise.
22358 (vectorizable_store): Likewise.
22359 (vectorizable_load): Likewise.
22360 * tree-vectorizer.h: Likewise.
22362 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22364 * passes.c (emergency_dump_function): New.
22365 * tree-pass.h (emergency_dump_function): Declare.
22366 * plugin.c (plugins_internal_error_function): Remove.
22367 * plugin.h (plugins_internal_error_function): Remove declaration.
22368 * toplev.c (internal_error_function): New static function. Use it...
22369 (general_init): ...here.
22371 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22373 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22374 (arc_rtx_costs): Add costs for new patterns.
22375 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22376 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22378 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22380 * tree-ssa-strlen.c (get_next_strinfo): New function.
22381 (get_stridx_plus_constant): Use it.
22382 (zero_length_string): Likewise.
22383 (adjust_related_strinfos): Likewise.
22384 (adjust_last_stmt): Likewise.
22386 2017-05-31 Richard Biener <rguenther@suse.de>
22389 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22390 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22392 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22394 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22395 loop_vinfo argument and use of dependence distance vectors.
22396 Check instead whether the two references differ only in their
22397 initial value and assume that they have the same alignment if the
22398 difference is a multiple of the vector alignment.
22399 (vect_analyze_data_refs_alignment): Update call accordingly.
22401 2017-05-31 Martin Liska <mliska@suse.cz>
22404 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22406 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22408 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22409 (create_intersect_range_checks): Move from ...
22410 * tree-data-ref.c (create_intersect_range_checks_index)
22411 (create_intersect_range_checks): ... to here.
22412 (create_runtime_alias_checks): New function factored from ...
22413 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22414 here. Call above function.
22415 * tree-data-ref.h (create_runtime_alias_checks): New function.
22417 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22419 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22420 segment length for dr_b and compute it in wide_int.
22422 2017-05-31 Richard Biener <rguenther@suse.de>
22424 PR tree-optimization/80906
22425 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22426 and pass through iv_map.
22427 (copy_bb_and_scalar_dependences): Adjust.
22428 (translate_pending_phi_nodes): Likewise.
22429 (copy_loop_close_phi_args): Handle code-generating IVs instead
22432 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22434 * diagnostic-color.c (color_dict): Add "type-diff".
22435 (parse_gcc_colors): Update comment.
22436 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22437 -fdiagnostics-show-template-tree and -fno-elide-type.
22438 (GCC_COLORS): Add type-diff to example.
22440 (-fdiagnostics-show-template-tree): New.
22441 (-fno-elide-type): New.
22442 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22443 the pp_format_decoder callback. Call any m_format_postprocessor's
22445 (pretty_printer::pretty_printer): Initialize
22446 m_format_postprocessor.
22447 (pretty_printer::~pretty_printer): Delete any
22448 m_format_postprocessor.
22449 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22450 (class format_postprocessor): New class.
22451 (struct pretty_printer::format_decoder): Document the new parameters.
22452 (struct pretty_printer::m_format_postprocessor): New field.
22453 * tree-diagnostic.c (default_tree_printer): Update for new
22454 bool and const char ** params.
22455 * tree-diagnostic.h (default_tree_printer): Likewise.
22457 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22459 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22460 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22461 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22462 rs6000_gen_cell_microcode code.
22463 (rs6000_final_prescan_insn): Delete.
22464 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22465 "warn-cell-microcode" entries.
22466 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22467 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22468 throughout. Change cc_reg_not_micro_cr0_operand to
22469 cc_reg_not_cr0_operand throughout.
22470 (*extendhi<mode>2_noload): Delete.
22471 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22472 (mwarn-cell-microcode): Delete.
22473 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22474 -mgen-cell-microcode and -mwarn-cell-microcode.
22476 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22479 * config/i386/constraints.md (Yd): New constraint.
22481 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22482 and (?Yd, r) alternatives. Update insn attributes.
22483 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22484 and (?*Yd, r) alternatives. Update insn attributes.
22485 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22487 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22489 * gimplify.c (gimplify_modify_expr): Don't create a
22490 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22493 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22495 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22497 2017-05-30 Richard Biener <rguenther@suse.de>
22499 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22500 and reduc_def fields.
22501 (STMT_VINFO_REDUC_TYPE): New define.
22502 (STMT_VINFO_REDUC_DEF): Likewise.
22503 (vect_force_simple_reduction): Adjust prototype.
22504 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22505 (vect_is_simple_reduction): Remove check_reduction argument.
22506 (vect_force_simple_reduction): Adjust and set
22507 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22508 (vectorizable_reduction): Do not re-do reduction analysis
22509 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22510 * tree-parloops.c (gather_scalar_reductions): Adjust.
22512 2017-05-30 Richard Biener <rguenther@suse.de>
22514 PR middle-end/80901
22515 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22518 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22520 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22521 Introduce unknown_misalignment parameter and remove vf.
22522 (vect_peeling_hash_get_lowest_cost):
22523 Pass unknown_misalignment parameter.
22524 (vect_enhance_data_refs_alignment):
22525 Fix unsupportable data ref treatment.
22527 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22529 * tree-vect-data-refs.c (vect_get_data_access_cost):
22530 Workaround for SLP handling.
22531 (vect_enhance_data_refs_alignment):
22532 Compute costs for doing no peeling at all, compare to the best
22533 peeling costs so far and avoid peeling if cheaper.
22535 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22537 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22538 Return peeling info and set costs to zero for unlimited cost
22540 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22541 with unknown misalignment. Compute and costs for unknown
22542 misalignment, compare them to the costs for known misalignment
22543 and choose the cheapest for peeling.
22545 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22547 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22548 (vect_get_peeling_costs_all_drs): Create function.
22549 (vect_peeling_hash_get_lowest_cost):
22550 Use vect_get_peeling_costs_all_drs.
22551 (vect_peeling_supportable): Create function.
22552 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22554 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22556 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22557 DR_HAS_NEGATIVE_STEP.
22558 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22559 (vect_enhance_data_refs_alignment): Use.
22560 (vect_duplicate_ssa_name_ptr_info): Use.
22561 * tree-vectorizer.h (dr_misalignment): Use.
22562 (known_alignment_for_access_p): Use.
22564 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22567 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22569 (has_section_name): New function.
22571 2017-05-30 Martin Liska <mliska@suse.cz>
22574 * auto-profile.c (get_function_decl_from_block): Fix
22577 2017-05-30 Richard Biener <rguenther@suse.de>
22579 PR middle-end/80876
22580 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22582 2017-05-30 Martin Liska <mliska@suse.cz>
22584 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22585 * dumpfile.h (struct dump_file_info): Remove ctors.
22587 2017-05-30 Martin Liska <mliska@suse.cz>
22589 * predict.def: Fix GNU coding style.
22591 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22593 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22594 Mark 'to' argument with ATTRIBUTE_UNUSED.
22596 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22598 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22599 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22600 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22603 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22605 * doc/install.texi (Options specification): Restore entry of
22606 --enable-sjlj-exceptions.
22608 2017-05-27 Michael Eager <eager@eagercon.com>
22611 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22613 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22615 * config/microblaze/microblaze.h
22616 (FIXED_REGISTERS): Update in macro.
22617 (CALL_USED_REGISTERS): Update in macro.
22619 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22621 * doc/install.texi: Add links to macOS binary distributions.
22623 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22627 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22629 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22631 2017-05-26 Martin Liska <mliska@suse.cz>
22633 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22635 2017-05-26 Martin Liska <mliska@suse.cz>
22637 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22638 always leading ';; '.
22639 (dump_bb_info): Likewise.
22640 (brief_dump_cfg): Likewise.
22641 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22642 * dumpfile.c: Remove usage of TDF_VERBOSE.
22643 * dumpfile.h (enum dump_kind): Likewise.
22644 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22645 * print-tree.c (debug_verbose): Remove.
22646 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22647 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22648 * tree-diagnostic.c (default_tree_printer): Replace
22649 TDF_DIAGNOSTIC with TDF_SLIM.
22651 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22653 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22654 in parameter loop, rather than loop_vinfo.
22655 (create_intersect_range_checks): Ditto.
22656 (vect_create_cond_for_alias_checks): Update call to above functions.
22658 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22660 PR tree-optimization/80815
22661 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22662 for merging runtime alias checks. Handle negative DR_STEPs.
22664 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22666 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22668 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22669 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22670 out code pruning runtime alias checks.
22671 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22672 factored out from above.
22673 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22675 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22677 (prune_runtime_alias_test_list): New decalaration.
22679 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22681 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22682 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22683 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22684 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22685 (operator==, comp_dr_with_seg_len_pair): Ditto.
22686 (vect_prune_runtime_alias_test_list): Ditto.
22688 2017-05-26 Martin Liska <mliska@suse.cz>
22691 * params.def: Bound partial-inlining-entry-probability param.
22693 2017-05-26 Marek Polacek <polacek@redhat.com>
22696 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22699 2017-05-26 Richard Biener <rguenther@suse.de>
22701 PR tree-optimization/80842
22702 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22705 2017-05-26 Richard Biener <rguenther@suse.de>
22707 PR tree-optimization/80844
22708 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22710 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22712 * doc/md.texi (Machine Constraints): Update x86 family
22713 machine constraints section to match 'config/i386/constraints.md'.
22715 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22717 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22719 2017-05-25 Nathan Sidwell <nathan@acm.org>
22721 * doc/invoke.texi (--enable-languages): Update documentation.
22723 2017-05-25 Martin Liska <mliska@suse.cz>
22725 * dumpfile.c: Add TDF_FOLDING.
22726 * dumpfile.h (enum dump_kind): Likewise.
22727 * genmatch.c (dt_simplify::gen_1): Use it.
22729 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22731 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22733 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22735 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22736 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22738 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22740 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22741 * match.pd (X == C): Rewrite it here.
22742 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22743 with_certain_nonzero_bits2): New predicates.
22744 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22746 2017-05-24 Nathan Sidwell <nathan@acm.org>
22748 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22751 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22754 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22756 * config/powerpcspe: New port. Files are copied from the rs6000
22757 port, with "rs6000" in filenames replaced by "powerpcspe".
22759 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22761 PR rtl-optimization/80754
22762 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22764 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22766 * config/sparc/sparc.md (length): Return the correct value for -mflat
22767 sibcalls to match output_sibcall.
22769 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22773 * config/rs6000/rs6000.c (struct machine_function): Add new field
22775 (rs6000_get_separate_components): Init that field, use it.
22776 (rs6000_components_for_bb): Use the field.
22778 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22780 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22782 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22784 PR middle-end/80823
22785 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22787 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22790 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22791 address against address_operand predicate.
22792 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22794 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22796 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22797 parameters passed indirectly.
22799 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22801 * config/i386/i386.md (*movdi_internal): Remove SSE4
22802 alternative 18 (?r, *v). Update insn attributes.
22803 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22804 Update insn attributes.
22805 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22806 Update insn attributes.
22807 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22808 alternative 1 (r, v). Remove isa attribute.
22809 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22810 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22811 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22813 2017-05-23 Tom de Vries <tom@codesourcery.com>
22815 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22818 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22820 * cgraphunit.c (symbol_table::process_new_functions): Update.
22821 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22822 (inline_generate_summary): Rename to ...
22823 (ipa_fn_summary_generate): ... this one.
22824 (inline_read_summary): Rename to ...
22825 (ipa_fn_summary_read): ... this one.
22826 (inline_write_summary): Rename to ...
22827 (ipa_fn_summary_write): ... this one.
22828 (inline_free_summary): Rename to ...
22829 (ipa_free_fn_summary): ... this one.
22830 (pass_data_local_fn_summary, pass_local_fn_summary,
22831 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22832 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22833 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22834 make_pass_ipa_fn_summary): New.
22835 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22836 inline_write_summary, inline_free_summary): Remove.
22837 (ipa_free_fn_summary) : New.
22838 * ipa-inline.c (ipa_inline): Update.
22839 (pass_ipa_inline): Do not generate summaries.
22840 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22842 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22843 and add pass_ipa_fn_summary.
22844 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22846 (make_pass_inline_parameters): Remove.
22848 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22850 * omp-low.c (struct omp_context): Remove "default_kind" member.
22853 * omp-offload.c (execute_oacc_device_lower): Remove the
22854 parallelism dimensions function attributes for unparallelized
22855 OpenACC kernels constructs.
22857 2017-05-23 Martin Liska <mliska@suse.cz>
22859 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22861 (cgraph_edge::make_speculative): Likewise.
22862 (cgraph_edge::resolve_speculation): Likewise.
22863 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22864 (cgraph_node::dump): Likewise.
22865 * cgraph.h: Likewise.
22866 * cgraphunit.c (analyze_functions): Likewise.
22867 (symbol_table::compile): Likewise.
22868 * ipa-cp.c (print_all_lattices): Likewise.
22869 (determine_versionability): Likewise.
22870 (initialize_node_lattices): Likewise.
22871 (ipcp_verify_propagated_values): Likewise.
22872 (estimate_local_effects): Likewise.
22873 (update_profiling_info): Likewise.
22874 (create_specialized_node): Likewise.
22875 (perhaps_add_new_callers): Likewise.
22876 (decide_about_value): Likewise.
22877 (decide_whether_version_node): Likewise.
22878 (identify_dead_nodes): Likewise.
22879 (ipcp_store_bits_results): Likewise.
22880 * ipa-devirt.c (dump_targets): Likewise.
22881 (ipa_devirt): Likewise.
22882 * ipa-icf.c (sem_item::dump): Likewise.
22883 (sem_function::equals): Likewise.
22884 (sem_variable::equals): Likewise.
22885 (sem_item_optimizer::read_section): Likewise.
22886 (sem_item_optimizer::execute): Likewise.
22887 (congruence_class::dump): Likewise.
22888 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22889 (dump_inline_summary): Likewise.
22890 (estimate_node_size_and_time): Likewise.
22891 (inline_analyze_function): Likewise.
22892 * ipa-inline-transform.c (inline_call): Likewise.
22893 * ipa-inline.c (report_inline_failed_reason): Likewise.
22894 (want_early_inline_function_p): Likewise.
22895 (edge_badness): Likewise.
22896 (update_edge_key): Likewise.
22897 (inline_small_functions): Likewise.
22898 * ipa-profile.c (ipa_profile): Likewise.
22899 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22900 (ipa_make_edge_direct_to_target): Likewise.
22901 (remove_described_reference): Likewise.
22902 (ipa_impossible_devirt_target): Likewise.
22903 (propagate_controlled_uses): Likewise.
22904 (ipa_print_node_params): Likewise.
22905 (ipcp_transform_function): Likewise.
22906 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22907 (propagate_pure_const): Likewise.
22908 * ipa-reference.c (generate_summary): Likewise.
22909 (read_write_all_from_decl): Likewise.
22910 (propagate): Likewise.
22911 (ipa_reference_read_optimization_summary): Likewise.
22912 * ipa-utils.c (ipa_merge_profiles): Likewise.
22913 * ipa.c (walk_polymorphic_call_targets): Likewise.
22914 (symbol_table::remove_unreachable_nodes): Likewise.
22915 (ipa_single_use): Likewise.
22916 * passes.c (execute_todo): Likewise.
22917 * predict.c (drop_profile): Likewise.
22918 * symtab.c (symtab_node::get_dump_name): New function.
22919 (symtab_node::dump_name): Likewise.
22920 (symtab_node::dump_asm_name): Likewise.
22921 (symtab_node::dump_references): Likewise.
22922 (symtab_node::dump_referring): Likewise.
22923 (symtab_node::dump_base): Likewise.
22924 (symtab_node::debug_symtab): Likewise.
22925 * tree-sra.c (convert_callers_for_node): Likewise.
22926 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22927 * value-prof.c (init_node_map): Likewise.
22929 2017-05-23 Martin Liska <mliska@suse.cz>
22931 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22932 and symtab_node::debug_symtab to symbol_table::debug.
22933 * cgraphunit.c (analyze_functions): Use the renamed function.
22934 (symbol_table::compile): Likewise.
22935 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22936 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22937 * passes.c (execute_todo): Likewise.
22938 * symtab.c (symbol_table::dump): New function.
22939 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22941 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22943 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22944 that nonconst implies exec.
22946 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22948 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22949 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22950 (inline_edge_summary_vec): Turn into ...
22951 (ipa_call_summaries): ... this one.
22952 (redirect_to_unreachable, edge_set_predicate,
22953 evaluate_properties_for_edge, inline_summary_alloc,
22954 reset_ipa_call_summary, reset_inline_summary,
22955 inline_summary_t::duplicate): Update.
22956 (inline_edge_duplication_hook): Turn to ...
22957 (ipa_call_summary_t::duplicate): ... this one.
22958 (inline_edge_removal_hook): Turn to ...
22959 (ipa_call_summary_t::remove): ... this one.
22960 (dump_inline_edge_summary): Turn to ...
22961 (dump_ipa_call_summary): ... this one.
22962 (estimate_function_body_sizes): Update.
22963 (inline_update_callee_summaries): Update.
22964 (remap_edge_change_prob): Update.
22965 (remap_edge_summaries): Update.
22966 (inline_merge_summary): Update.
22967 (do_estimate_edge_time): Update.
22968 (inline_generate_summary): Update.
22969 (inline_read_section): Update.
22970 (inline_read_summary): Update.
22971 (inline_free_summary): Update.
22972 * ipa-inline.c (can_inline_edge_p): Update.
22973 (compute_inlined_call_time): Update.
22974 (want_inline_small_function_p): Update.
22975 (edge_badness): Update.
22976 (early_inliner): Update.
22977 * ipa-inline.h (inline_edge_summary): Turn to ...
22978 (ipa_call_summary): ... this one.
22979 (ipa_call_summary_t): New class.
22980 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22981 (ipa_call_summaries): New.
22982 (inline_edge_summary): Remove.
22983 (estimate_edge_growth): Update.
22984 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22985 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22986 * ipa-split.c (execute_split_functions): Update.
22987 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22989 2017-05-23 Tom de Vries <tom@codesourcery.com>
22991 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22992 attributes): Document rdrand effective target.
22994 2017-05-23 Tom de Vries <tom@codesourcery.com>
22996 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22997 attributes): Sort alphabetically.
22999 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23001 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23003 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23006 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23007 V2DF/V2DI splat into two separate patterns, one that handles
23008 registers, and the other that only handles memory. Drop support
23009 for splatting from a GPR on ISA 2.07 and then splitting the
23010 splat into direct move and splat.
23011 (vsx_splat_<mode>_reg): Likewise.
23012 (vsx_splat_<mode>_mem): Likewise.
23014 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23016 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23018 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23020 PR middle-end/80809
23021 * omp-low.c (finish_taskreg_remap): New function.
23022 (finish_taskreg_scan): If unit size of ctx->record_type
23023 is non-constant, unshare the size expression and replace
23024 decls in it with possible outer var refs.
23026 PR middle-end/80809
23027 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23028 GOVD_SHARED rather than GOVD_PRIVATE with it.
23029 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23030 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23032 PR middle-end/80853
23033 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23034 as last argument to build_outer_var_ref for pointer bases of array
23035 section reductions.
23037 2017-05-19 Martin Sebor <msebor@redhat.com>
23039 * print-tree.c (print_node): Print DECL_READ_P flag.
23041 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23043 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23044 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23045 * cgraph.c: Likewise.
23046 * cgraphunit.c: Likewise.
23047 * gengtype.c: Likewise.
23048 * ipa-cp.c: Likewise.
23049 * ipa-devirt.c: Likewise.
23050 * ipa-icf.c: Likewise.
23051 * ipa-predicate.c: Likewise.
23052 * ipa-profile.c: Likewise.
23053 * ipa-prop.c: Likewise.
23054 * ipa-split.c: Likewise.
23056 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23057 edge_predicate_pool, dump_inline_hints,
23058 inline_summary::account_size_time, redirect_to_unreachable,
23059 edge_set_predicate, set_hint_predicate,
23060 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23061 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23062 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23063 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23064 ipa_call_summary_t::remove, initialize_growth_caches,
23065 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23066 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23067 mark_modified, unmodified_parm_1, unmodified_parm,
23068 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23069 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23070 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23071 will_be_nonconstant_predicate, record_modified_bb_info,
23072 get_minimal_bb, record_modified, param_change_prob,
23073 phi_result_unknown_predicate, predicate_for_phi_result,
23074 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23075 estimate_function_body_sizes, compute_inline_parameters,
23076 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23077 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23078 inline_update_callee_summaries, remap_edge_change_prob,
23079 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23080 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23081 inline_analyze_function, inline_summary_t::insert,
23082 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23083 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23084 inline_free_summary): Move to ipa-fnsummary.h
23085 (predicate_t): Remove.
23086 * ipa-fnsummary.c: New file.
23087 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23088 (enum inline_hints_vals, inline_hints, agg_position_info,
23089 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23090 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23091 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23092 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23093 inline_read_summary, inline_write_summary, inline_free_summary,
23094 inline_analyze_function, initialize_inline_failed,
23095 inline_merge_summary, inline_update_overall_summary,
23096 compute_inline_parameters): Move to ipa-fnsummary.h
23097 * ipa-fnsummary.h: New file.
23098 * ipa-inline-transform.h: Include ipa-inline.h.
23099 * ipa-inline.c: LIkewise.
23101 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23103 * ipa-inline.c (edge_badness): Use inlined_time instead of
23104 inline_summaries->get.
23106 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23108 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23110 2017-05-22 Nathan Sidwell <nathan@acm.org>
23112 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23113 (fdump-lang): Document 'raw' option.
23114 * dumpfile.h (TDI_tu): Delete.
23115 * dumpfile.c (dump_files): Remove translation-unit.
23116 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23118 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23120 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23121 command option from $(AWK) call.
23122 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23123 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23124 [FORMAT]: Remove handling of variable.
23125 * config/avr/t-multilib: Regenerate.
23127 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23129 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23131 (dump_inline_summary): Do not print self_time.
23132 (estimate_function_body_sizes): Do not set self_time.
23133 (compute_inline_parameters): Likewise.
23134 (inline_read_section, inline_write_summary): Do not stream self_time.
23135 * ipa-inline.h (inline_summary): Drop self_time.
23137 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23139 * ipa-inline-analysis.c (account_size_time): Rename to ...
23140 (inline_summary::account_size_time): ... this one.
23141 (reset_ipa_call_summary): Turn to ...
23142 (ipa_call_summary::reset): ... this one.
23143 (reset_inline_summary): Turn to ...
23144 (inline_summary::reset): ... this one.
23145 (inline_summary_t::remove): Update.
23146 (inline_summary_t::duplicate): Update.
23147 (ipa_call_summary_t::remove): Update.
23148 (dump_inline_summary): Update.
23149 (estimate_function_body_sizes): Update.
23150 (compute_inline_parameters): Update.
23151 (estimate_node_size_and_time): Update.
23152 (inline_merge_summary): Update.
23153 (inline_update_overall_summary): Update.
23154 (inline_read_section): Update.
23155 (inline_write_summary): Update.
23156 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23157 add account_size_time and reset member functions.
23158 (ipa_call_summary): Add reset function.
23159 * ipa-predicate.h (predicate::operator &): Constify.
23161 2017-05-22 Richard Biener <rguenther@suse.de>
23163 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23165 2017-05-19 Jason Merrill <jason@redhat.com>
23167 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23169 2017-05-19 Marek Polacek <polacek@redhat.com>
23172 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23173 TYPE_OVERFLOW_WRAPS checks.
23175 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23177 * tree-core.h (enum omp_clause_default_kind): Add
23178 "OMP_CLAUSE_DEFAULT_PRESENT".
23179 * tree-pretty-print.c (dump_omp_clause): Handle it.
23180 * gimplify.c (enum gimplify_omp_var_data): Add
23181 "GOVD_MAP_FORCE_PRESENT".
23182 (gimplify_adjust_omp_clauses_1): Map it to
23183 "GOMP_MAP_FORCE_PRESENT".
23184 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23186 * gimplify.c (oacc_default_clause): Clarify.
23188 2017-05-19 Nathan Sidwell <nathan@acm.org>
23190 LANG_HOOK_REGISTER_DUMPS
23191 * toplev.c (general_init): Call register dump lang hook.
23192 * doc/invoke.texi: Document -fdump-lang option family.
23193 * dumpfile.c (dump_files): Remove class dump here.
23194 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23195 * dumpfile.h (tree_dump_index): Remove TDI_class.
23196 * langhooks-def.h (lhd_register_dumps): Declare.
23197 (LANG_HOOKS_REGISTER_DUMPS): Define.
23198 (LANG_HOOKS_INITIALIZER): Add it.
23199 * langhooks.c (lhd_register_dumps): Define.
23200 * langhooks.h (struct lang_hooks): Add register_dumps.
23202 2017-05-19 Nathan Sidwell <nathan@acm.org>
23204 * context.h (context::set_passes): New.
23205 * context.c (context::context): Do not create pass manager.
23206 * toplev.c (general_init): Create pass manager here.
23208 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23210 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23211 use this splitter if two add or or instructions would also work for
23212 the constant we want to generate.
23214 2017-05-19 Richard Biener <rguenther@suse.de>
23217 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23218 predicate evaluation.
23220 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23222 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23224 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23227 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23229 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23230 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23231 (inline_edge_summary_vec): Turn into ...
23232 (ipa_call_summaries): ... this one.
23233 (redirect_to_unreachable, edge_set_predicate,
23234 evaluate_properties_for_edge, inline_summary_alloc,
23235 reset_ipa_call_summary, reset_inline_summary,
23236 inline_summary_t::duplicate): Update.
23237 (inline_edge_duplication_hook): Turn to ...
23238 (ipa_call_summary_t::duplicate): ... this one.
23239 (inline_edge_removal_hook): Turn to ...
23240 (ipa_call_summary_t::remove): ... this one.
23241 (dump_inline_edge_summary): Turn to ...
23242 (dump_ipa_call_summary): ... this one.
23243 (estimate_function_body_sizes): Update.
23244 (inline_update_callee_summaries): Update.
23245 (remap_edge_change_prob): Update.
23246 (remap_edge_summaries): Update.
23247 (inline_merge_summary): Update.
23248 (do_estimate_edge_time): Update.
23249 (inline_generate_summary): Update.
23250 (inline_read_section): Update.
23251 (inline_read_summary): Update.
23252 (inline_free_summary): Update.
23253 * ipa-inline.c (can_inline_edge_p): Update.
23254 (compute_inlined_call_time): Update.
23255 (want_inline_small_function_p): Update.
23256 (edge_badness): Update.
23257 (early_inliner): Update.
23258 * ipa-inline.h (inline_edge_summary): Turn to ...
23259 (ipa_call_summary): ... this one.
23260 (ipa_call_summary_t): New class.
23261 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23262 (ipa_call_summaries): New.
23263 (inline_edge_summary): Remove.
23264 (estimate_edge_growth): Update.
23265 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23266 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23267 * ipa-split.c (execute_split_functions): Update.
23268 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23270 2017-05-19 Richard Biener <rguenther@suse.de>
23272 PR middle-end/80764
23273 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23275 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23277 * config/rs6000/rs6000.c (struct machine_function): Add field
23278 fpr_is_wrapped_separately.
23279 (rs6000_get_separate_components): Use 64 components. Handle the
23280 new FPR components.
23281 (rs6000_components_for_bb): Handle the FPR components.
23282 (rs6000_emit_prologue_components): Handle the FPR components.
23283 (rs6000_emit_epilogue_components): Handle the FPR components.
23284 (rs6000_set_handled_components): Handle the FPR components.
23285 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23286 that are already separately shrink-wrapped.
23287 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23288 that are already separately shrink-wrapped.
23290 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23293 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23296 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23297 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23298 cases where the register allocator uses a move and an offsettable
23299 memory operation to/from a FPR register on ISA 2.06/2.07.
23300 (define_peephole2 for Altivec d-form store): Likewise.
23302 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23305 * config/i386/mmx.md (*mov<mode>_internal): Enable
23306 alternatives 11, 12, 13 and 14 also for 32bit targets.
23307 Remove alternatives 15, 16, 17 and 18.
23308 * config/i386/sse.md (vec_concatv2di): Change
23309 alternative (!x, *y) to (x, ?!*Yn).
23311 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23313 * dumpfile.h (enum dump_kind): Remove stray comma.
23315 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23317 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23318 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23319 predicate::num_conditions
23320 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23321 (CHANGED): turn into predicate::changed.
23322 (agg_position_info): Move to ipa-predicate.h
23323 (add_condition, predicate::add_clause, predicate::operator &=,
23324 predicate::or_with, predicate::evaluate, predicate::probability,
23325 dump_condition, dump_clause, predicate::dump,
23326 predicate::remap_after_duplication, predicate::remap_after_inlining,
23327 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23328 (evaluate_conditions_for_known_args): Update.
23329 (set_cond_stmt_execution_predicate): Update.
23330 * ipa-inline.h: Include ipa-predicate.h
23331 (condition, inline_param_summary, conditions, agg_position_info,
23332 predicate): Move to ipa-predicate.h
23333 * ipa-predicate.c: New file.
23334 * ipa-predicate.h: New file.
23336 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23338 * final.c (leaf_function_p): Check we are not in a sequence.
23340 2017-05-18 Martin Liska <mliska@suse.cz>
23342 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23343 * dumpfile.c (dump_register): Use new enum dump_kind.
23344 (get_dump_file_name): Likewise.
23345 (dump_enable_all): Likewise.
23346 (dump_switch_p_1): Likewise.
23347 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23348 * dumpfile.h (enum dump_kind): New enum type.
23349 (struct dump_file_info): Create constructor and
23350 format fields and comments.
23351 * passes.c (pass_manager::register_one_dump_file):
23353 * statistics.c (statistics_early_init): Likewise.
23354 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23355 TDF_TREE with TDF_SLIM.
23356 (gather_memory_references_ref): Likewise.
23358 2017-05-18 Martin Liska <mliska@suse.cz>
23360 * vec.h (struct vnull): Use it.
23362 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23364 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23365 (true_predicate, false_predicate, true_predicate_p,
23366 false_predicate_p): Remove.
23367 (single_cond_predicate, not_inlined_predicate): Turn to member function
23369 (add_condition): Update.
23370 (add_clause): Turn to...
23371 (predicate::add_clause): ... this one; update; allow passing NULL
23373 (and_predicates): Turn to ...
23374 (predicate::operator &=): ... this one.
23375 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23376 (or_predicates): Turn to ...
23377 (predicate::or_with): ... this one.
23378 (evaluate_predicate): Turn to ...
23379 (predicate::evaluate): ... this one.
23380 (predicate_probability): Turn to ...
23381 (predicate::probability): ... this one.
23382 (dump_condition): Update.
23383 (dump_predicate): Turn to ...
23384 (predicate::dump): ... this one.
23385 (account_size_time): Update.
23386 (edge_set_predicate): Update.
23387 (set_hint_predicate): UPdate.
23388 (evaluate_conditions_for_known_args): Update.
23389 (evaluate_properties_for_edge): Update.
23390 (remap_predicate_after_duplication): Turn to...
23391 (predicate::remap_after_duplication): ... this one.
23392 (remap_hint_predicate_after_duplication): Update.
23393 (inline_summary_t::duplicate): UPdate.
23394 (dump_inline_edge_summary): Update.
23395 (dump_inline_summary): Update.
23396 (set_cond_stmt_execution_predicate): Update.
23397 (set_switch_stmt_execution_predicate): Update.
23398 (compute_bb_predicates): Update.
23399 (will_be_nonconstant_expr_predicate): Update.
23400 (will_be_nonconstant_predicate): Update.
23401 (phi_result_unknown_predicate): Update.
23402 (predicate_for_phi_result): Update.
23403 (array_index_predicate): Update.
23404 (estimate_function_body_sizes): Update.
23405 (estimate_node_size_and_time): Update.
23406 (estimate_ipcp_clone_size_and_time): Update.
23407 (remap_predicate): Rename to ...
23408 (predicate::remap_after_inlining): ... this one.
23409 (remap_hint_predicate): Update.
23410 (inline_merge_summary): Update.
23411 (inline_update_overall_summary): Update.
23412 (estimate_size_after_inlining): Update.
23413 (read_predicate): Rename to ...
23414 (predicate::stream_in): ... this one.
23415 (read_inline_edge_summary): Update.
23416 (write_predicate): Rename to ...
23417 (predicate::stream_out): ... this one.
23418 (write_inline_edge_summary): Update.
23419 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23420 (clause_t): Turn to uint32_t
23421 (predicate): Turn to class; implement constructor and operators
23423 (size_time_entry): Update.
23424 (inline_summary): Update.
23425 (inline_edge_summary): Update.
23427 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23429 * fold-const.c (fold_binary_loc): Move transformation...
23430 * match.pd (C - X CMP X): ... here.
23432 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23434 * config/sparc/sparc.c (sparc_option_override): Set function
23435 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23436 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23438 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23440 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23442 2017-05-18 Marek Polacek <polacek@redhat.com>
23445 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23446 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23447 gimple_assign_load_p.
23449 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23451 PR middle-end/80692
23452 * real.c (do_compare): Give decimal_do_compare preference over
23453 comparing just the signs.
23455 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23457 * doc/md.texi (Canonicalization of Instructions): Describe the
23458 canonical form of instructions that inherently set a condition
23461 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23463 PR middle-end/80775
23464 * tree-cfg.c: Move deletion of unreachable case statements to after
23465 the merging of consecutive case labels.
23467 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23469 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23470 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23471 restoring of callee-saved registers.
23473 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23475 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23476 * config/visium/visium.c (single_set_and_flags): Likewise.
23477 * config/visium/visium.md (Substitutions): Likewise.
23479 2017-05-17 Martin Liska <mliska@suse.cz>
23481 * cfg.c: Introduce dump_flags_t type and
23482 use it instead of int type.
23484 * cfghooks.c: Likewise.
23485 * cfghooks.h (struct cfg_hooks): Likewise.
23486 * cfgrtl.c: Likewise.
23487 * cfgrtl.h: Likewise.
23488 * cgraph.c (cgraph_node::get_body): Likewise.
23489 * coretypes.h: Likewise.
23490 * domwalk.c: Likewise.
23491 * domwalk.h: Likewise.
23492 * dumpfile.c (struct dump_option_value_info): Likewise.
23493 (dump_enable_all): Likewise.
23494 (dump_switch_p_1): Likewise.
23495 (opt_info_switch_p): Likewise.
23496 * dumpfile.h (enum tree_dump_index): Likewise.
23497 (struct dump_file_info): Likewise.
23498 * genemit.c: Likewise.
23499 * generic-match-head.c: Likewise.
23500 * gengtype.c (open_base_files): Likewise.
23501 * gimple-pretty-print.c: Likewise.
23502 * gimple-pretty-print.h: Likewise.
23503 * graph.c (print_graph_cfg): Likewise.
23504 * graphite-scop-detection.c (dot_all_sese): Likewise.
23505 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23506 * loop-unroll.c (report_unroll): Likewise.
23507 * passes.c (pass_manager::register_one_dump_file): Likewise.
23508 * print-tree.c: Likewise.
23509 * statistics.c: Likewise.
23510 * tree-cfg.c: Likewise.
23511 * tree-cfg.h: Likewise.
23512 * tree-dfa.c: Likewise.
23513 * tree-dfa.h: Likewise.
23514 * tree-dump.c (dump_function): Likewise.
23515 * tree-dump.h (struct dump_info): Likewise.
23516 * tree-pretty-print.c: Likewise.
23517 * tree-pretty-print.h: Likewise.
23518 * tree-ssa-live.c: Likewise.
23519 * tree-ssa-live.h: Likewise.
23520 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23521 * tree-vect-loop.c: Likewise.
23522 * tree-vect-slp.c: Likewise.
23524 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23525 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23527 PR tree-optimization/80457
23528 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23529 of all arguments to a statement as scalar_to_vec operations.
23530 (vectorizable_call): Adjust call to vect_model_simple_cost for
23532 (vectorizable_conversion): Likewise.
23533 (vectorizable_assignment): Likewise.
23534 (vectorizable_shift): Likewise.
23535 (vectorizable_operation): Likewise.
23536 (vectorizable_comparison): Likewise.
23537 (vect_is_simple_cond): Record the def types for operands.
23538 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23539 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23540 for statement argument count.
23542 2017-05-16 Carl Love <cel@us.ibm.com>
23544 * config/rs6000/rs6000-c: Add support for built-in functions
23545 vector unsigned long long vec_bperm (vector unsigned long long,
23546 vector unsigned char)
23547 vector signed long long vec_mule (vector signed int,
23549 vector unsigned long long vec_mule (vector unsigned int,
23550 vector unsigned int)
23551 vector signed long long vec_mulo (vector signed int,
23553 vector unsigned long long vec_mulo (vector unsigned int,
23554 vector unsigned int)
23555 vector signed char vec_sldw (vector signed char,
23556 vector signed char,
23558 vector unsigned char vec_sldw (vector unsigned char,
23559 vector unsigned char,
23561 vector signed short vec_sldw (vector signed short,
23562 vector signed short,
23564 vector unsigned short vec_sldw (vector unsigned short,
23565 vector unsigned short,
23567 vector signed int vec_sldw (vector signed int,
23570 vector unsigned int vec_sldw (vector unsigned int,
23571 vector unsigned int,
23573 vector signed long long vec_sldw (vector signed long long,
23574 vector signed long long,
23576 vector unsigned long long vec_sldw (vector unsigned long long,
23577 vector unsigned long long,
23579 * config/rs6000/rs6000-c: Add support for built-in functions
23580 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23581 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23582 * doc/extend.texi: Update the built-in documentation for the
23583 new built-in functions.
23585 2017-05-16 Marek Polacek <polacek@redhat.com>
23589 * tree.c (save_expr): Don't fold the expression.
23591 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23593 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23594 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23595 and (?*y,m). Update insn attributes.
23597 2017-05-16 Martin Liska <mliska@suse.cz>
23599 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23600 flags argument of print_gimple_stmt, print_gimple_expr,
23601 print_generic_stmt and print_generic_expr.
23602 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23603 * coretypes.h: Likewise.
23604 * except.c (dump_eh_tree): Likewise.
23605 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23606 * gimple-pretty-print.h: Likewise.
23607 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23608 (backprop::push_to_worklist): Likewise.
23609 (backprop::pop_from_worklist): Likewise.
23610 (backprop::process_use): Likewise.
23611 (backprop::intersect_uses): Likewise.
23612 (note_replacement): Likewise.
23613 * gimple-ssa-store-merging.c
23614 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23615 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23616 (pass_store_merging::execute): Likewise.
23617 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23618 (ssa_base_cand_dump_callback): Likewise.
23619 (dump_incr_vec): Likewise.
23620 (replace_refs): Likewise.
23621 (replace_mult_candidate): Likewise.
23622 (create_add_on_incoming_edge): Likewise.
23623 (create_phi_basis): Likewise.
23624 (insert_initializers): Likewise.
23625 (all_phi_incrs_profitable): Likewise.
23626 (introduce_cast_before_cand): Likewise.
23627 (replace_one_candidate): Likewise.
23628 * gimplify.c (gimplify_expr): Likewise.
23629 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23630 (set_rename): Likewise.
23631 (rename_uses): Likewise.
23632 (copy_loop_phi_nodes): Likewise.
23633 (add_close_phis_to_merge_points): Likewise.
23634 (copy_loop_close_phi_args): Likewise.
23635 (copy_cond_phi_args): Likewise.
23636 (graphite_copy_stmts_from_block): Likewise.
23637 (translate_pending_phi_nodes): Likewise.
23638 * graphite-poly.c (print_pdr): Likewise.
23639 (dump_gbb_cases): Likewise.
23640 (dump_gbb_conditions): Likewise.
23641 (print_scop_params): Likewise.
23642 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23643 (build_cross_bb_scalars_use): Likewise.
23644 (gather_bbs::before_dom_children): Likewise.
23645 * hsa-dump.c (dump_hsa_immed): Likewise.
23646 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23647 (get_replacement_map): Likewise.
23648 * ipa-inline-analysis.c (dump_condition): Likewise.
23649 (estimate_function_body_sizes): Likewise.
23650 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23651 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23652 * ipa-prop.c (ipa_dump_param): Likewise.
23653 (ipa_print_node_jump_functions_for_edge): Likewise.
23654 (ipa_modify_call_arguments): Likewise.
23655 (ipa_modify_expr): Likewise.
23656 (ipa_dump_param_adjustments): Likewise.
23657 (ipa_dump_agg_replacement_values): Likewise.
23658 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23659 * ipa-pure-const.c (check_stmt): Likewise.
23660 (pass_nothrow::execute): Likewise.
23661 * ipa-split.c (execute_split_functions): Likewise.
23662 * omp-offload.c (dump_oacc_loop_part): Likewise.
23663 (dump_oacc_loop): Likewise.
23664 * trans-mem.c (tm_log_emit): Likewise.
23665 (tm_memopt_accumulate_memops): Likewise.
23666 (dump_tm_memopt_set): Likewise.
23667 (dump_tm_memopt_transform): Likewise.
23668 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23669 (print_loop): Likewise.
23670 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23671 (chkp_gather_checks_info): Likewise.
23672 (chkp_get_check_result): Likewise.
23673 (chkp_remove_check_if_pass): Likewise.
23674 (chkp_use_outer_bounds_if_possible): Likewise.
23675 (chkp_reduce_bounds_lifetime): Likewise.
23676 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23677 (chkp_mark_completed_bounds): Likewise.
23678 (chkp_register_incomplete_bounds): Likewise.
23679 (chkp_mark_invalid_bounds): Likewise.
23680 (chkp_maybe_copy_and_register_bounds): Likewise.
23681 (chkp_build_returned_bound): Likewise.
23682 (chkp_get_bound_for_parm): Likewise.
23683 (chkp_build_bndldx): Likewise.
23684 (chkp_get_bounds_by_definition): Likewise.
23685 (chkp_generate_extern_var_bounds): Likewise.
23686 (chkp_get_bounds_for_decl_addr): Likewise.
23687 * tree-chrec.c (chrec_apply): Likewise.
23688 * tree-data-ref.c (dump_data_reference): Likewise.
23689 (dump_subscript): Likewise.
23690 (dump_data_dependence_relation): Likewise.
23691 (analyze_overlapping_iterations): Likewise.
23692 * tree-inline.c (expand_call_inline): Likewise.
23693 (tree_function_versioning): Likewise.
23694 * tree-into-ssa.c (dump_defs_stack): Likewise.
23695 (dump_currdefs): Likewise.
23696 (dump_names_replaced_by): Likewise.
23697 (dump_update_ssa): Likewise.
23698 (update_ssa): Likewise.
23699 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23700 * tree-parloops.c (build_new_reduction): Likewise.
23701 (try_create_reduction_list): Likewise.
23702 (ref_conflicts_with_region): Likewise.
23703 (oacc_entry_exit_ok_1): Likewise.
23704 (oacc_entry_exit_single_gang): Likewise.
23705 * tree-pretty-print.h: Likewise.
23706 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23707 (get_scalar_evolution): Likewise.
23708 (add_to_evolution): Likewise.
23709 (get_loop_exit_condition): Likewise.
23710 (analyze_evolution_in_loop): Likewise.
23711 (analyze_initial_condition): Likewise.
23712 (analyze_scalar_evolution): Likewise.
23713 (instantiate_scev): Likewise.
23714 (number_of_latch_executions): Likewise.
23715 (gather_chrec_stats): Likewise.
23716 (final_value_replacement_loop): Likewise.
23717 (scev_const_prop): Likewise.
23718 * tree-sra.c (dump_access): Likewise.
23719 (disqualify_candidate): Likewise.
23720 (create_access): Likewise.
23721 (reject): Likewise.
23722 (maybe_add_sra_candidate): Likewise.
23723 (create_access_replacement): Likewise.
23724 (analyze_access_subtree): Likewise.
23725 (analyze_all_variable_accesses): Likewise.
23726 (sra_modify_assign): Likewise.
23727 (initialize_constant_pool_replacements): Likewise.
23728 (find_param_candidates): Likewise.
23729 (decide_one_param_reduction): Likewise.
23730 (replace_removed_params_ssa_names): Likewise.
23731 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23732 * tree-ssa-copy.c (dump_copy_of): Likewise.
23733 (copy_prop_visit_cond_stmt): Likewise.
23734 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23735 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23736 (record_equivalences_from_stmt): Likewise.
23737 * tree-ssa-dse.c (compute_trims): Likewise.
23738 (delete_dead_call): Likewise.
23739 (delete_dead_assignment): Likewise.
23740 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23741 (forward_propagate_into_cond): Likewise.
23742 (pass_forwprop::execute): Likewise.
23743 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23744 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23746 (move_computations_worker): Likewise.
23747 (execute_sm): Likewise.
23748 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23749 (remove_exits_and_undefined_stmts): Likewise.
23750 (remove_redundant_iv_tests): Likewise.
23751 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23752 (adjust_iv_update_pos): Likewise.
23753 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23754 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23755 (value_replacement): Likewise.
23756 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23757 * tree-ssa-pre.c (print_pre_expr): Likewise.
23758 (get_representative_for): Likewise.
23759 (create_expression_by_pieces): Likewise.
23760 (insert_into_preds_of_block): Likewise.
23761 (eliminate_insert): Likewise.
23762 (eliminate_dom_walker::before_dom_children): Likewise.
23763 (eliminate): Likewise.
23764 (remove_dead_inserted_code): Likewise.
23765 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23766 * tree-ssa-reassoc.c (get_rank): Likewise.
23767 (eliminate_duplicate_pair): Likewise.
23768 (eliminate_plus_minus_pair): Likewise.
23769 (eliminate_not_pairs): Likewise.
23770 (undistribute_ops_list): Likewise.
23771 (eliminate_redundant_comparison): Likewise.
23772 (update_range_test): Likewise.
23773 (optimize_range_tests_var_bound): Likewise.
23774 (optimize_vec_cond_expr): Likewise.
23775 (rewrite_expr_tree): Likewise.
23776 (rewrite_expr_tree_parallel): Likewise.
23777 (linearize_expr): Likewise.
23778 (break_up_subtract): Likewise.
23779 (linearize_expr_tree): Likewise.
23780 (attempt_builtin_powi): Likewise.
23781 (attempt_builtin_copysign): Likewise.
23782 (transform_stmt_to_copy): Likewise.
23783 (transform_stmt_to_multiply): Likewise.
23784 (dump_ops_vector): Likewise.
23785 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23786 (print_scc): Likewise.
23787 (set_ssa_val_to): Likewise.
23788 (visit_reference_op_store): Likewise.
23789 (visit_use): Likewise.
23790 (sccvn_dom_walker::before_dom_children): Likewise.
23791 (run_scc_vn): Likewise.
23792 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23794 (expr_hash_elt::print): Likewise.
23795 (const_and_copies::pop_to_marker): Likewise.
23796 (const_and_copies::record_const_or_copy_raw): Likewise.
23797 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23798 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23799 (dump_predicates): Likewise.
23800 (find_uninit_use): Likewise.
23801 (warn_uninitialized_phi): Likewise.
23802 (pass_late_warn_uninitialized::execute): Likewise.
23803 * tree-ssa.c (verify_vssa): Likewise.
23804 (verify_ssa): Likewise.
23805 (maybe_optimize_var): Likewise.
23806 * tree-vrp.c (dump_value_range): Likewise.
23807 (dump_all_value_ranges): Likewise.
23808 (dump_asserts_for): Likewise.
23809 (register_edge_assert_for_2): Likewise.
23810 (vrp_visit_cond_stmt): Likewise.
23811 (vrp_visit_switch_stmt): Likewise.
23812 (vrp_visit_stmt): Likewise.
23813 (vrp_visit_phi_node): Likewise.
23814 (simplify_cond_using_ranges_1): Likewise.
23815 (fold_predicate_in): Likewise.
23816 (evrp_dom_walker::before_dom_children): Likewise.
23817 (evrp_dom_walker::push_value_range): Likewise.
23818 (evrp_dom_walker::pop_value_range): Likewise.
23819 (execute_early_vrp): Likewise.
23821 2017-05-16 Richard Biener <rguenther@suse.de>
23823 * dwarf2out.c (loc_list_from_tree_1): Do not create
23824 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23826 2017-05-16 Richard Biener <rguenther@suse.de>
23828 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23830 (note_variable_value_in_expr): If we resolved the decl ref
23831 do not push to the stack.
23833 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23835 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23836 operations in fast-math mode.
23837 (vaddq_f16): Likewise.
23838 (vmul_f16): Likewise.
23839 (vmulq_f16): Likewise.
23840 (vsub_f16): Likewise.
23841 (vsubq_f16): Likewise.
23842 * config/arm/neon.md (add<mode>3): New.
23844 (fma:<VH:mode>3): New. Also remove outdated comment.
23847 2017-05-16 Martin Liska <mliska@suse.cz>
23851 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23852 (odr_types_equivalent_p): Likewise.
23854 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23856 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23858 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23861 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23862 non-interunit SSE move alternatives with '?'.
23863 (zero-extendsidi peephole2): New peephole to skip intermediate
23864 general register in SSE zero-extend sequence.
23866 2017-05-15 Jeff Law <law@redhat.com>
23868 * reorg.c (relax_delay_slots): Create a new variable to hold
23869 the temporary target rather than clobbering TARGET_LABEL.
23871 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23872 missing argument to extract_bit_field call.
23873 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23875 2017-05-15 Martin Liska <mliska@suse.cz>
23878 * gcc.c (process_command): Do not allow empty argument of -o option.
23880 2017-05-15 Renlin Li <renlin.li@arm.com>
23882 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23883 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23884 * config/aarch64/constraints.md (Usf): Add long call check.
23885 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23886 (call_value): Likewise.
23887 (sibcall): Likewise.
23888 (sibcall_value): Likewise.
23890 (call_value_insn): New.
23891 (sibcall_insn): Update rtx pattern.
23892 (sibcall_value_insn): Likewise.
23893 (call_internal): Remove.
23894 (call_value_internal): Likewise.
23895 (sibcall_internal): Likewise.
23896 (sibcall_value_internal): Likewise.
23897 (call_reg): Likewise.
23898 (call_symbol): Likewise.
23899 (call_value_reg): Likewise.
23900 (call_value_symbol): Likewise.
23902 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23905 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23907 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23909 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23910 compatible with CCGOCmode and with CCZmode.
23912 2017-05-14 Martin Sebor <msebor@redhat.com>
23914 PR middle-end/77671
23915 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23916 (gimple_fold_builtin_snprintf): Same.
23917 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23918 (gimple_fold_builtin_snprintf): Same.
23919 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23920 of character types.
23921 (is_call_safe): New function.
23922 (try_substitute_return_value): Call it.
23923 (try_simplify_call): New function.
23924 (pass_sprintf_length::handle_gimple_call): Call it.
23926 2017-05-14 Martin Sebor <msebor@redhat.com>
23928 PR middle-end/80669
23929 * builtins.c (expand_builtin_stpncpy): Simplify.
23931 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23933 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23934 * config/i386/i386.h
23935 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23936 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23937 (struct machine_function): Add new members call_ms2sysv,
23938 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23939 (struct machine_frame_state): New fields sp_realigned and
23940 sp_realigned_offset.
23941 * config/i386/i386.c
23942 (enum xlogue_stub): New enum.
23943 (enum xlogue_stub_sets): New enum.
23944 (class xlogue_layout): New class.
23945 (struct ix86_frame): New fields stack_realign_allocate_offset,
23946 stack_realign_offset and outlined_save_offset. Modify comments to
23947 detail stack layout when using out-of-line stubs.
23948 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23949 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23950 -mcall-ms2sysv-xlogues.
23951 (stub_managed_regs): New static variable.
23952 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23953 registers managed by out-of-line stub.
23954 (disable_call_ms2sysv_xlogues): New function.
23955 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23956 m->call_ms2sysv when appropriate and compute frame layout for
23958 (sp_valid_at, fp_valid_at): New inline functions.
23959 (choose_basereg): New function.
23960 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23962 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23963 Use align parameter of choose_baseaddr to generated aligned SSE movs
23965 (pro_epilogue_adjust_stack): Modify to track
23966 machine_frame_state::sp_realigned.
23967 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23968 (ix86_nsaved_sseregs): Likewise.
23969 (ix86_emit_save_regs): Likewise.
23970 (ix86_emit_save_regs_using_mov): Likewise.
23971 (ix86_emit_save_sse_regs_using_mov): Likewise.
23972 (get_scratch_register_on_entry): Likewise.
23973 (gen_frame_set): New function.
23974 (gen_frame_load): Likewise.
23975 (gen_frame_store): Likewise.
23976 (emit_outlined_ms2sysv_save): Likewise.
23977 (emit_outlined_ms2sysv_restore): Likewise.
23978 (ix86_expand_prologue): Modify stack re-alignment code and call
23979 emit_outlined_ms2sysv_save when appropriate.
23980 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23981 parameter rtx_insn *insn, which allows the function to be used to only
23982 generate the notes.
23983 (ix86_expand_epilogue): Modify validity checks of frame and stack
23984 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23985 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23986 * config/i386/predicates.md
23987 (save_multiple): New predicate.
23988 (restore_multiple): Likewise.
23989 * config/i386/sse.md
23990 (save_multiple<mode>): New pattern.
23991 (save_multiple_realign<mode>): Likewise.
23992 (restore_multiple<mode>): Likewise.
23993 (restore_multiple_and_return<mode>): Likewise.
23994 (restore_multiple_leave_return<mode>): Likewise.
23995 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23997 2017-05-14 Julia Koval <julia.koval@intel.com>
23999 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24000 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24001 (__builtin_ia32_xsetbv): New builtins.
24002 * config/i386/i386.c (ix86_expand_special_args_builtin):
24004 (ix86_expand_builtin): Special expand for new intrinsics.
24005 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24006 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24007 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24009 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24011 * cfganal.c (inverted_post_order_compute): Change argument type
24013 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24014 * df-core.c (rest_of_handle_df_initialize): Adjust.
24015 (rest_of_handle_df_finish): Likewise.
24016 (df_analyze_1): Likewise.
24017 (df_analyze): Likewise.
24018 (loop_inverted_post_order_compute): Change argument to be a vec *.
24019 (df_analyze_loop): Adjust.
24020 (df_get_n_blocks): Likewise.
24021 (df_get_postorder): Likewise.
24022 * df.h (struct df_d): Change field to be a vec.
24023 * lcm.c (compute_laterin): Adjust.
24024 (compute_available): Likewise.
24025 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24026 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24027 * tree-ssa-pre.c (compute_antic): Likewise.
24029 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24031 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24032 (depth_first_search::depth_first_search): Change structure init
24033 function to this constructor.
24034 (depth_first_search::add_bb): Rename function to this member.
24035 (depth_first_search::execute): Likewise.
24036 (flow_dfs_compute_reverse_finish): Adjust.
24038 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24040 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24041 (longest_simple_path): Likewise.
24042 * shrink-wrap.c (spread_components): Likewise.
24043 (disqualify_problematic_components): Likewise.
24044 (emit_common_heads_for_components): Likewise.
24045 (emit_common_tails_for_components): Likewise.
24046 (insert_prologue_epilogue_for_components): Likewise.
24048 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24050 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24053 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24055 * df-core.c (df_set_blocks): Start using auto_bitmap.
24056 (df_compact_blocks): Likewise.
24057 * df-problems.c (df_rd_confluence_n): Likewise.
24058 * df-scan.c (df_insn_rescan_all): Likewise.
24059 (df_process_deferred_rescans): Likewise.
24060 (df_update_entry_block_defs): Likewise.
24061 (df_update_exit_block_uses): Likewise.
24062 (df_entry_block_bitmap_verify): Likewise.
24063 (df_exit_block_bitmap_verify): Likewise.
24064 (df_scan_verify): Likewise.
24065 * lra-constraints.c (lra_constraints): Likewise.
24066 (undo_optional_reloads): Likewise.
24067 (lra_undo_inheritance): Likewise.
24068 * lra-remat.c (calculate_gen_cands): Likewise.
24069 (do_remat): Likewise.
24070 * lra-spills.c (assign_spill_hard_regs): Likewise.
24071 (spill_pseudos): Likewise.
24072 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24073 (bitmap_set_subtract_values): Likewise.
24075 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24077 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24078 management with auto_bitmap.
24079 (fix_inter_tick): Likewise.
24080 (fix_recovery_deps): Likewise.
24081 * ira.c (add_store_equivs): Likewise.
24082 (find_moveable_pseudos): Likewise.
24083 (split_live_ranges_for_shrink_wrap): Likewise.
24084 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24085 (rtx_reuse_manager::seen_def_p): Likewise.
24086 (rtx_reuse_manager::set_seen_def): Likewise.
24087 * print-rtl.h (class rtx_reuse_manager): Likewise.
24089 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24091 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24093 (migrate_btr_def): Likewise.
24094 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24095 * df-core.c (loop_post_order_compute): Likewise.
24096 (loop_inverted_post_order_compute): Likewise.
24097 * hsa-common.h: Likewise.
24098 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24099 * init-regs.c (initialize_uninitialized_regs): Likewise.
24100 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24101 (inline_small_functions): Likewise.
24102 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24103 * ira.c (combine_and_move_insns): Likewise.
24104 (build_insn_chain): Likewise.
24105 * loop-invariant.c (find_invariants): Likewise.
24106 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24107 * predict.c (tree_predict_by_opcode): Likewise.
24108 (predict_paths_leading_to): Likewise.
24109 (predict_paths_leading_to_edge): Likewise.
24110 (estimate_loops_at_level): Likewise.
24111 (estimate_loops): Likewise.
24112 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24113 (spread_components): Likewise.
24114 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24115 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24116 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24117 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24118 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24119 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24120 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24121 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24122 (mark_threaded_blocks): Likewise.
24123 (thread_through_all_blocks): Likewise.
24124 * tree-ssa.c (verify_ssa): Likewise.
24125 (execute_update_addresses_taken): Likewise.
24126 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24128 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24130 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24132 (post_order_compute): Likewise.
24133 (inverted_post_order_compute): Likewise.
24134 (pre_and_rev_post_order_compute_fn): Likewise.
24136 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24138 * genrecog.c (int_set::int_set): Explicitly construct our
24139 auto_vec base class.
24140 * vec.h (auto_vec::auto_vec): New constructor.
24142 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24144 * bitmap.h (class auto_bitmap): New constructor taking
24145 bitmap_obstack * argument.
24147 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24149 * bitmap.h (class auto_bitmap): Change type of m_bits to
24150 bitmap_head, and adjust ctor / dtor and member operators.
24152 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24154 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24155 when returned register mode doesn't match original mode.
24157 2017-05-12 Jeff Law <law@redhat.com>
24158 Jakub Jelinek <jakub@redhat.com>
24160 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24161 we look for cc setter after the compare-elim changes.
24162 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24163 within the vector to match what compare-elim now expects.
24164 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24165 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24167 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24168 after the compare-elim changes.
24169 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24170 the vector to match what compare-elim now expects.
24171 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24172 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24173 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24174 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24175 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24177 * config/visium/visium.c (single_set_and_flags): Fix where
24178 we look for cc setter after the compare-elim changes.
24179 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24180 with the vector to match what compare-elim now expects.
24181 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24182 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24183 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24184 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24185 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24186 (neg<mode>2_insn_set_overflow): Likewise.
24188 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24190 PR middle-end/79794
24191 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24192 maybe_expand_insn call, set ops[0].target. If still set after call,
24193 set alt_rtl. Add extra arg to recursive calls.
24194 (extract_bit_field): Add alt_rtl argument. Pass to
24196 * expmed.h (extract_bit_field): Fix prototype.
24197 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24198 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24199 to extract_bit_field_calls.
24200 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24201 Pass alt_rtl to extract_bit_field calls.
24202 * calls.c (store_unaligned_arguments_into_psuedos)
24203 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24204 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24206 * optabs.h (struct expand_operand): Add target bitfield.
24208 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24210 * compare-elim.c (try_eliminate_compare): Canonicalize
24211 operation with embedded compare to
24212 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24213 (set (reg) (operation)].
24215 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24217 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24220 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24221 cost of adding a carry flag for ADC instruction.
24222 [case MINUS]: Ignore the cost of subtracting a carry flag
24223 for SBB instruction.
24225 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24227 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24229 * config/rs6000/bmiintrin.h: New file.
24230 * config/rs6000/bmi2intrin.h: New file.
24231 * config/rs6000/x86intrin.h: New file.
24233 2017-05-12 Jeff Law <law@redhat.com>
24235 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24238 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24240 PR middle-end/80707
24241 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24243 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24245 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24246 early expansion of vector divide builtins.
24247 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24248 builtins identified as having unsigned arguments.
24250 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24252 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24253 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24254 expansion of vector logical operations (and, andc, or, xor,
24257 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24259 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24260 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24262 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24264 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24265 early GIMPLE expansion of vector multiplies.
24267 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24269 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24270 TARGET_HAVE_MOVT conditional.
24271 (movt splitter): Likewise.
24273 2017-05-12 Richard Biener <rguenther@suse.de>
24275 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24276 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24277 Fold all stmts not inplace.
24279 2017-05-12 Richard Biener <rguenther@suse.de>
24281 PR tree-optimization/80713
24282 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24283 inserted_exprs bit for not removed stmts.
24285 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24287 PR middle-end/69921
24288 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24289 parallelized" attribute for parallelized OpenACC kernels.
24290 * omp-offload.c (execute_oacc_device_lower): Use it.
24292 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24293 Set "oacc kernels" attribute.
24294 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24295 parameter. Adjust all users.
24296 (oacc_fn_attrib_kernels_p): Remove function.
24297 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24298 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24299 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24300 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24301 assert "oacc kernels" attribute is set.
24303 2017-05-11 Carl Love <cel@us.ibm.com>
24305 * config/rs6000/rs6000-c: Add support for built-in functions
24306 vector unsigned char vec_popcnt (vector signed char)
24307 vector unsigned char vec_popcnt (vector unsigned char)
24308 vector unsigned short vec_popcnt (vector signed short)
24309 vector unsigned short vec_popcnt (vector unsigned short)
24310 vector unsigned int vec_popcnt (vector signed int)
24311 vector unsigned int vec_popcnt (vector unsigned int)
24312 vector unsigned long long vec_popcnt (vector signed long long)
24313 vector unsigned long long vec_popcnt (vector unsigned long long)
24314 vector signed long long vec_slo (vector signed long long,
24315 vector signed char)
24316 vector signed long long vec_slo (vector signed long long,
24317 vector unsigned char)
24318 vector unsigned long long vec_slo (vector unsigned long long,
24319 vector signed char)
24320 vector unsigned long long vec_slo (vector unsigned long long,
24321 vector unsigned char)
24322 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24323 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24324 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24325 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24326 * doc/extend.texi: Update the built-in documentation file for the
24327 new built-in functions.
24329 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24331 * attribs.h (sorted_attr_string): Move machine independent
24332 functions for target clone support from the i386 port to common
24333 code. Rename ix86_function_versions to common_function_versions.
24334 Rename make_name to make_unique_name.
24335 (common_function_versions): Likewise.
24336 (make_unique_name): Likewise.
24337 (make_dispatcher_decl): Likewise.
24338 (is_function_default_version): Likewise.
24339 * attribs.c (attr_strcmp): Likewise.
24340 (sorted_attr_string): Likewise.
24341 (common_function_versions): Likewise.
24342 (make_unique_name): Likewise.
24343 (make_dispatcher_decl): Likewise.
24344 (is_function_default_version): Likewise.
24345 * config/i386/i386.c (attr_strcmp): Likewise.
24346 (sorted_attr_string): Likewise.
24347 (ix86_function_versions): Likewise.
24348 (make_name): Likewise.
24349 (make_dispatcher_decl): Likewise.
24350 (is_function_default_version): Likewise.
24351 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24353 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24356 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24357 Account for direct move costs for vec_construct of integer
24360 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24363 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24364 (UNSPEC_STX_ATOMIC): Ditto.
24365 (loaddi_via_sse): New insn.
24366 (storedi_via_sse): Ditto.
24367 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24368 Update corresponding peephole2 patterns.
24369 (atomic_storedi_fpu): Ditto.
24371 2017-05-11 Julia Koval <julia.koval@intel.com>
24373 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24374 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24376 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24377 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24378 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24380 2017-05-11 Nathan Sidwell <nathan@acm.org>
24382 * graphite-poly.c: Include dumpfile.h.
24384 * dumpfle.h (dump_function): Declare here ...
24385 * tree-dump.h (dump_function): ... not here.
24386 * dumpfile.c: #include tree-cfg.h.
24387 (dump_function): Move here from ...
24388 * tree-dump.c (dump_function): ... here.
24389 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24390 * graphite-poly.c: Don't include tree-dump.h.
24391 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24392 * print-tree.c: Likewise.
24393 * stor-layout.c: Likewise.
24394 * tree-nested.c: Likewise.
24396 * dumpfile.c (dump_start): Use TDF_FLAGS.
24397 (dump_enable_all): Fix TDF_KIND check thinko.
24399 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24401 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24402 array entries to represent two legal parameterizations of the
24403 overloaded __builtin_cmpb function, as represented by the
24404 P6_OV_BUILTIN_CMPB constant.
24405 (altivec_resolve_overloaded_builtin): Add special case handling
24406 for the __builtin_cmpb function, as represented by the
24407 P6_OV_BUILTIN_CMPB constant.
24408 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24409 (BU_P6_64BIT_2): New macro.
24410 (BU_P6_OVERLOAD_2): New macro
24411 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24412 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24413 (CMPB): Add overload support to represent both 32-bit and 64-bit
24414 compare-bytes function.
24415 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24416 support for TARGET_CMPB.
24417 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24418 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24419 documentation of the __builtin_cmpb overloaded built-in function.
24421 2017-05-11 Richard Biener <rguenther@suse.de>
24423 PR tree-optimization/80705
24424 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24425 bases are not vectorizable.
24427 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24429 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24430 when counting register pressure.
24432 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24434 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24435 (iv_ca_more_deps): Renamed to ...
24436 (iv_ca_compare_deps): ... this.
24437 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24439 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24441 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24443 (determine_group_iv_costs): ... here.
24444 (find_inv_vars_cb): Record inv var if it's not recorded before.
24446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24448 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24449 (get_shiftadd_cost): Ditto.
24451 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24453 * tree-ssa-address.c: Include header file.
24454 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24456 (add_to_parts): Refactor.
24457 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24458 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24461 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24463 PR tree-optimization/53090
24464 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24466 (extract_cond_operands): Detect condition with IV on both sides
24467 and return COMP_IV_EXPR_2.
24468 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24469 (rewrite_use_compare): Simplify by removing call to function
24470 extract_cond_operands.
24472 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24474 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24475 (extract_cond_operands): Detect condition comparing against non-
24476 invariant bound and return appropriate enum value.
24477 (find_interesting_uses_cond): Update use of extract_cond_operands.
24478 Handle its return value accordingly.
24479 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24481 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24483 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24484 nonlinear iv_use computation in loop invariant sensitive way.
24486 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24488 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24489 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24491 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24493 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24494 (dump_cand): Support iv_cand.inv_exprs.
24495 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24497 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24500 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24502 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24504 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24505 as local function. Include necessary header files.
24506 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24508 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24510 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24512 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24514 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24515 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24516 RSHIFT_EXPR and BIT_NOT_EXPR.
24518 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24520 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24521 (adjust_setup_cost): New parameter supporting round up adjustment.
24522 (struct address_cost_data): Delete.
24523 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24524 (split_address_cost, ptr_difference_cost): Delete.
24525 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24526 (struct ainc_cost_data): New struct.
24527 (get_address_cost_ainc): New function.
24528 (get_address_cost, get_computation_cost): Reimplement.
24529 (determine_group_iv_cost_address): Record inv_expr for all uses of
24531 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24532 (iv_ca_has_deps): Reimplemented to ...
24533 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24535 (iv_ca_extend): Call iv_ca_more_deps.
24537 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24539 * tree-ssa-address.c (struct mem_address): Move to header file.
24540 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24541 * tree-ssa-address.h (struct mem_address): Move from C file.
24542 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24544 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24546 * tree-affine.h (aff_combination_type): New interface.
24547 (aff_combination_zero_p): Remove static.
24548 (aff_combination_const_p): New interface.
24549 (aff_combination_singleton_var_p): New interfaces.
24551 2017-05-11 Richard Biener <rguenther@suse.de>
24553 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24554 Skip unreachable blocks and destinations.
24555 (eliminate): Move stmt removal and fixup ...
24556 (fini_eliminate): ... here. Skip inserted exprs.
24557 (pass_pre::execute): Move fini_pre after fini_eliminate.
24558 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24559 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24560 PRE to get rid of dead code that has invalid SSA form and
24561 split critical edges again.
24563 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24565 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24567 2017-05-11 Richard Biener <rguenther@suse.de>
24569 * passes.c (execute_function_todo): Verify loops if they are
24570 said to be up-to-date.
24571 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24572 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24574 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24577 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24578 handle calling assemble_external ourself.
24581 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24582 modes with zero size. Enhance comment.
24584 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24586 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24587 built-ins for vec_xl and vec_xst with short and char pointer
24590 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24592 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24593 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24594 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24595 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24596 (_mm_maskz_min_round_ss): New intrinsics.
24597 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24598 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24599 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24600 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24601 (__builtin_ia32_minss_mask_round): New builtins.
24602 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24603 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24604 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24606 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24607 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24609 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24612 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24614 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24615 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24616 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24617 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24618 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24619 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24620 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24621 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24622 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24623 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24624 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24625 (__builtin_ia32_mulss_mask_round): New builtins.
24626 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24627 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24628 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24630 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24631 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24633 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24636 2017-05-10 Julia Koval <julia.koval@intel.com>
24638 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24639 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24640 (_mm256_setr_m128i): New intrinsics.
24642 2017-05-10 Julia Koval <julia.koval@intel.com>
24644 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24645 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24646 (_mm_maskz_rcp14_ss): New intrinsics.
24647 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24648 (__builtin_ia32_rcp14ss_mask): New builtins.
24649 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24651 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24653 PR tree-optimization/51513
24654 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24655 (assert_unreachable_fallthru_edge_p): Use it.
24656 (group_case_labels_stmt): Likewise.
24657 * tree-cfg.h: Prototype it.
24658 * stmt.c: Include cfghooks.h and tree-cfg.h.
24659 (emit_case_dispatch_table) <gap_label>: New local variable.
24660 Use it to fill dispatch table gaps.
24661 Test for default_label before updating probabilities.
24662 (expand_case) <default_label>: Remove unneeded initialization.
24663 Test for unreachable default case statement and remove its edge.
24664 Set default_label accordingly.
24665 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24667 2017-05-10 Carl Love <cel@us.ibm.com>
24669 * config/rs6000/rs6000-c: Add support for built-in functions
24670 vector signed char vec_neg (vector signed char)
24671 vector signed short int vec_neg (vector short int)
24672 vector signed int vec_neg (vector signed int)
24673 vector signed long long vec_neg (vector signed long long)
24674 vector float vec_neg (vector float)
24675 vector double vec_neg (vector double)
24676 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24678 * config/rs6000/altivec.h: Add define for vec_neg
24679 * doc/extend.texi: Update the built-in documentation for the
24680 new built-in functions.
24682 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24684 PR tree-optimization/77644
24685 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24687 2017-05-10 Nathan Sidwell <nathan@acm.org>
24689 * dumpfile.h (TDI_lang_all): New.
24690 (TDF_KIND): New. Renumber others
24691 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24693 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24695 (get_dump_file_name): Adjust suffix generation.
24696 (dump_enable_all): Use TDF_KIND.
24697 * doc/invoke.texi (-fdump-lang-all): Document.
24699 * dumpfile.h: Tabify.
24701 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24704 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24705 Move member access before delete.
24707 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24709 * tree-inline.c (expand_call_inline): Split block at stmt
24712 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24715 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24716 are now unused after splitting mov{sf,sd}_hardfloat.
24717 (f32_lr2): Likewise.
24718 (f32_lm): Likewise.
24719 (f32_lm2): Likewise.
24720 (f32_li): Likewise.
24721 (f32_li2): Likewise.
24722 (f32_lv): Likewise.
24723 (f32_sr): Likewise.
24724 (f32_sr2): Likewise.
24725 (f32_sm): Likewise.
24726 (f32_sm2): Likewise.
24727 (f32_si): Likewise.
24728 (f32_si2): Likewise.
24729 (f32_sv): Likewise.
24730 (f32_dm): Likewise.
24731 (f32_vsx): Likewise.
24732 (f32_av): Likewise.
24733 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24734 For movsf, order stores so the VSX stores occur before the GPR
24735 store which encourages the register allocator to use a traditional
24736 FPR instead of a GPR. For movsd, order the stores so that the GPR
24737 store comes before the VSX stores to allow the power6 to work.
24738 This is due to the power6 not having a 32-bit integer store
24739 instruction from a FPR.
24740 (movsf_hardfloat): Likewise.
24741 (movsd_hardfloat): Likewise.
24743 2017-05-09 Martin Sebor <msebor@redhat.com>
24745 PR translation/80280
24746 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24749 PR translation/80280
24750 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24751 data member added in r247778.
24752 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24754 2017-05-09 Nathan Sidwell <nathan@acm.org>
24756 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24758 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24761 2017-05-09 Marek Polacek <polacek@redhat.com>
24763 * doc/invoke.texi: Fix typo.
24765 2017-05-09 Richard Biener <rguenther@suse.de>
24767 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24768 (vrp_val_is_min): Likewise.
24769 (set_value_range_to_value): Likewise.
24770 (set_value_range_to_nonnegative): Likewise.
24771 (gimple_assign_nonzero_p): Likewise.
24772 (gimple_stmt_nonzero_p): Likewise.
24773 (vrp_int_const_binop): Likewise. Remove unreachable case.
24774 (adjust_range_with_scev): Adjust comments.
24775 (compare_range_with_value): Likewise.
24776 (extract_range_from_phi_node): Likewise.
24777 (test_for_singularity): Likewise.
24779 2017-05-09 Richard Biener <rguenther@suse.de>
24781 * tree-vrp.c (get_single_symbol): Add assert that we don't
24782 get overflowed constants as invariant part.
24783 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24784 checks. Use wi::cmp instead of recursing for integer constants.
24785 (compare_values): Just ignore whether we assumed undefined
24786 overflow instead of failing the compare.
24787 (extract_range_for_var_from_comparison_expr): Add comment before the
24788 TREE_NO_WARNING sets.
24789 (test_for_singularity): Likewise.
24790 (extract_range_from_comparison): Do not disable optimization
24791 when we assumed undefined overflow.
24792 (extract_range_basic): Remove init of unused var.
24794 2017-05-09 Richard Biener <rguenther@suse.de>
24796 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24797 (extract_range_from_multiplicative_op_1): Adjust.
24798 (extract_range_from_binary_expr_1): Use int_const_binop.
24800 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24803 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24804 rs6000_store_data_bypass_p in seven define_bypass directives and
24805 in several comments.
24806 * config/rs6000/rs6000-protos.h: Add prototype for
24807 rs6000_store_data_bypass_p function.
24808 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24809 function implements slightly different (rs6000-specific) semantics
24810 than store_data_bypass_p, returning false rather than aborting
24811 with assertion error when arguments do not satisfy the
24812 requirements of store data bypass.
24813 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24814 rs6000_store_data_bypass_p.
24816 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24818 * config/xtensa/xtensa-protos.h
24819 (xtensa_initial_elimination_offset): New declaration.
24820 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24821 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24822 macro definition, add case for FRAME_POINTER_REGNUM when
24823 FRAME_GROWS_DOWNWARD.
24824 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24825 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24826 xtensa_initial_elimination_offset.
24828 2017-05-08 Nathan Sidwell <nathan@acm.org>
24830 * doc/invoke.texi: Alphabetize -fdump options.
24832 2017-05-08 Martin Sebor <msebor@redhat.com>
24834 PR translation/80280
24835 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24837 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24839 * target.def (compute_frame_layout): New optional target hook.
24840 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24841 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24842 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24844 * reload1.c (verify_initial_elim_offsets): Likewise.
24845 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24846 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24847 (arm_get_frame_offsets): Split up into this ...
24848 (arm_compute_frame_layout): ... and this function.
24850 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24852 * config/aarch64/constraints.md (Usa): New constraint.
24853 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24855 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24857 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24858 with_multilib_list after it has been checked.
24860 2017-05-08 Richard Biener <rguenther@suse.de>
24862 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24863 (bitmap_set_subtract_values): Likewise.
24865 2017-05-08 Richard Biener <rguenther@suse.de>
24867 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24868 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24870 (gimple_stmt_nonzero_warnv_p): Rename to ...
24871 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24873 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24874 (extract_range_basic): Adjust, do not disable propagation on
24875 strict overflow sensitive simplification.
24876 (vrp_visit_cond_stmt): Likewise.
24878 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24880 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24881 body size unconditionally.
24883 2017-05-07 Jeff Law <law@redhat.com>
24886 2017-05-06 Jeff Law <law@redhat.com>
24887 PR tree-optimization/78496
24888 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24891 PR tree-optimization/78496
24892 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24893 (simplify_stmt_using_ranges): Call it.
24894 (vrp_dom_walker::before_dom_children): Extract equivalences
24895 from an ASSERT_EXPR with an equality comparison against a
24898 2017-05-06 Jeff Law <law@redhat.com>
24900 PR tree-optimization/78496
24901 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24904 PR tree-optimization/78496
24905 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24906 (simplify_stmt_using_ranges): Call it.
24907 (vrp_dom_walker::before_dom_children): Extract equivalences
24908 from an ASSERT_EXPR with an equality comparison against a
24911 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24913 * lra-constraints.c (lra_copy_reg_equiv): New function.
24914 (split_reg): Use it to copy equivalence information from the
24915 original register to the spill register.
24917 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24919 PR rtl-optimization/75964
24920 * simplify-rtx.c (simplify_const_relational_operation): Remove
24921 invalid handling of comparisons of integer ABS.
24923 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24925 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24926 initialize to zero.
24927 (init_regs): Remove declaration.
24928 (function_arg_advance_32): Initialize error_p as boolean variable.
24930 2017-05-05 Nathan Sidwell <nathan@acm.org>
24932 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24933 lines. Use for (;;).
24935 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24937 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24938 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24939 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24940 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24941 VF=2 that require versioning.
24943 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24945 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24948 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24950 * diagnostic.h (diagnostic_override_option_index): Convert from
24951 macro to inline function.
24953 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24955 * diagnostic.c (last_module_changed_p): New function.
24956 (set_last_module): New function.
24957 (diagnostic_report_current_module): Convert macro usage to
24958 the above functions.
24959 * diagnostic.h (diagnostic_context::last_module): Strengthen
24960 from const line_map * to const line_map_ordinary *.
24961 (diagnostic_last_module_changed): Delete macro.
24962 (diagnostic_set_last_module): Delete macro.
24964 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24966 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24967 with diagnostic_report_diagnostic.
24968 (diagnostic_n_impl_richloc): Likewise.
24969 * diagnostic.h (report_diagnostic): Delete macro.
24970 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24971 with diagnostic_report_diagnostic.
24972 * substring-locations.c (format_warning_va): Likewise.
24974 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24976 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24977 save/restor of format_spec. Move option-printing code to...
24978 (print_option_information): ...this new function, and
24979 reimplement by simply printing to the pretty_printer,
24980 rather than appending to the format string.
24982 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24984 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24985 handling logic into...
24986 (update_effective_level_from_pragmas): ...this new function.
24988 2017-05-04 Andrew Waterman <andrew@sifive.com>
24990 * config/riscv/riscv.opt (mstrict-align): New option.
24991 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24992 (SLOW_UNALIGNED_ACCESS): Define.
24993 (riscv_slow_unaligned_access): Declare.
24994 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24996 (riscv_slow_unaligned_access): New variable.
24997 (rocket_tune_info): Set slow_unaligned_access to true.
24998 (optimize_size_tune_info): Set slow_unaligned_access to false.
24999 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25000 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25001 (riscv_option_override): Set riscv_slow_unaligned_access.
25002 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25004 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25006 * config/riscv/riscv.md: Unify indentation.
25008 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25013 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25015 (extendsi<mode>2): Add support for doing sign extension via
25016 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25017 don't have ISA 3.0 instructions.
25018 (extendsi<mode>2 splitter): Likewise.
25019 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25020 generate the normal insns since SImode can now go in vector
25021 registers. Disallow the special UNSPECs needed for previous
25022 machines to hide SImode being used. Add new insns
25023 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25024 (fix_trunc<mode>si2_stfiwx): Likewise.
25025 (fix_trunc<mode>si2_internal): Likewise.
25026 (fixuns_trunc<mode>si2): Likewise.
25027 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25028 (fctiw<u>z_<mode>_smallint): Likewise.
25029 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25030 of floating point to 32-bit integer from doing a direct move to
25031 the GPR registers to do a store.
25032 (fctiwz_<mode>): Break long line.
25034 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25036 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25037 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25038 (addr_list, addr_offset_valid_p): New.
25039 (split_address_groups): Check offset validity with above function.
25040 (gt-tree-ssa-loop-ivopts.h): Include header file.
25042 2017-05-05 Nathan Sidwell <nathan@acm.org>
25044 * config.gcc (arm*-*-*): Add missing 'fi'.
25046 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25048 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25049 included in -fopt-info does not matter.
25050 * doc/optinfo.texi (-fopt-info): Fix description of default
25051 behavour. Explicitly say order of options included in -fopt-info
25054 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25056 * config.gcc: Allow combinations of aprofile and rmprofile values for
25057 --with-multilib-list.
25058 * config/arm/t-multilib: New file.
25059 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25060 variables. Remove setting of ISA and floating-point ABI in
25061 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25062 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25063 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25064 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25066 * config/arm/t-rmprofile: Likewise except for the matches changes.
25067 * doc/install.texi (--with-multilib-list): Document the combination of
25068 aprofile and rmprofile values and warn about pitfalls in doing that.
25070 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25072 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25073 (movdi_aarch64): Likewise.
25075 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25077 PR tree-optimization/80632
25078 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25080 (build_arrays): Initialize it for virtual phis.
25081 (fix_phi_nodes): Use it for virtual phis.
25083 PR tree-optimization/80558
25084 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25085 [x, y] op z into [x op, y op z] for op & or | if conditions
25088 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25089 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25092 * config/arm/arm.md (use_literal_pool): Remove.
25093 (64-bit immediate split): No longer takes cost into consideration
25094 if arm_disable_literal_pool is enabled.
25095 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25096 used when arm_disable_literal_pool is enabled.
25097 (arm_max_const_double_inline_cost): Remove use of
25098 arm_disable_literal_pool.
25099 (push_minipool_fix): Add assert.
25100 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25101 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25102 (no_literal_pool_sf_immediate): New.
25104 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25106 PR tree-optimization/80613
25107 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25108 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25110 2017-05-05 Richard Biener <rguenther@suse.de>
25112 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25114 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25116 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25117 of this flag from insn conditions due to removal from r247495.
25119 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25121 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25123 (arm_early_store_addr_dep_ptr): Likewise.
25124 * config/arm/aarch-common-protos.h
25125 (arm_early_load_addr_dep_ptr): Add prototype.
25126 (arm_early_store_addr_dep_ptr): Likewise.
25127 * config/arm/cortex-a53.md: Add new bypasses.
25129 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25131 * tree.c (next_type_uid): Change type to unsigned.
25132 (type_hash_canon): Decrement back next_type_uid if
25133 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25134 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25137 2017-05-04 Martin Sebor <msebor@redhat.com>
25139 * builtins.c: Fix a trivial typo in a comment.
25141 PR middle-end/79234
25142 * builtins.c (check_sizes): Adjust to handle reading past the end.
25143 Avoid printing excessive upper bound of ranges. Use %E to print
25144 tree nodes instead of converting them to %wu.
25145 (expand_builtin_memchr): New function.
25146 (compute_dest_size): Rename...
25147 (compute_objsize): ...to this.
25148 (expand_builtin_memcpy): Adjust.
25149 (expand_builtin_mempcpy): Adjust.
25150 (expand_builtin_strcat): Adjust.
25151 (expand_builtin_strcpy): Adjust.
25152 (check_strncat_sizes): Adjust.
25153 (expand_builtin_strncat): Adjust.
25154 (expand_builtin_strncpy): Adjust and simplify.
25155 (expand_builtin_memset): Adjust.
25156 (expand_builtin_bzero): Adjust.
25157 (expand_builtin_memcmp): Adjust.
25158 (expand_builtin): Handle memcmp.
25159 (maybe_emit_chk_warning): Check strncat just once.
25161 2017-05-04 Martin Sebor <msebor@redhat.com>
25163 PR preprocessor/79214
25164 PR middle-end/79222
25165 PR middle-end/79223
25166 * builtins.c (check_sizes): Add inlining context and issue
25167 warnings even when -Wno-system-headers is set.
25168 (check_strncat_sizes): Same.
25169 (expand_builtin_strncat): Same.
25170 (expand_builtin_memmove): New function.
25171 (expand_builtin_stpncpy): Same.
25172 (expand_builtin): Handle memmove and stpncpy.
25174 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25176 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25177 which is not used any more.
25179 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25181 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25183 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25185 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25186 (cortexa53_tunings): Likewise.
25187 (cortexa57_tunings): Likewise.
25188 (cortexa72_tunings): Likewise.
25189 (cortexa73_tunings): Likewise.
25191 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25193 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25194 Set loop alignment to 8.
25196 2017-05-04 Martin Sebor <msebor@redhat.com>
25198 PR translation/80280
25199 * builtins.c (expand_builtin_object_size): Add missing quoting to
25200 %D and like directives.
25201 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25202 (hsa_type_for_tree_type): Same.
25203 (verify_function_arguments): Same.
25204 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25205 * varasm.c (get_section): Same.
25208 2017-05-04 Martin Sebor <msebor@redhat.com>
25210 PR translation/80280
25211 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25213 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25215 * config/aarch64/aarch64.c (generic_addrcost_table):
25216 Change HI/TI mode setting.
25218 2017-05-04 Martin Jambor <mjambor@suse.cz>
25220 PR tree-optimization/80622
25221 * tree-sra.c (comes_initialized_p): New function.
25222 (build_accesses_from_assign): Only set write lazily when
25223 comes_initialized_p is false.
25224 (analyze_access_subtree): Use comes_initialized_p.
25225 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25226 instead of testing for PARM_DECL.
25228 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25230 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25231 constraint on operand 0 to allow more general addressing modes.
25232 Adjust output template.
25233 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25235 * config/aarch64/aarch64-protos.h
25236 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25237 * config/aarch64/constraints.md (Dp): New address constraint.
25238 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25241 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25243 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25244 update use of estimate_ipcp_clone_size_and_time.
25245 (estimate_local_effects): Update use of
25246 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25247 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25248 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25249 Return nonspecialized time.
25251 2017-05-04 Richard Biener <rguenther@suse.de>
25253 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25254 for the last VUSE which def dominates the PHI. Directly call
25256 (get_continuation_for_phi_1): Remove.
25258 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25260 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25261 to explain the use of truncating division. Cap the number of
25262 iterations to the maximum given by nb_iterations_upper_bound,
25265 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25267 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25268 * configure: Regenerate.
25269 * config.in: Regenerate.
25270 * config/i386/driver-mingw32.c: new file.
25271 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25272 * config.host: Link driver-mingw32.o on MinGW host.
25273 * doc/install.texi: Document new --enable-mingw-wildcard configure
25276 2017-05-04 Marek Polacek <polacek@redhat.com>
25278 PR tree-optimization/80612
25279 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25281 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25282 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25284 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25285 (movt splitter): Likewise.
25286 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25287 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25288 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25289 block for Thumb-1 with MOVT.
25290 (thumb2_legitimate_address_p): Move code block ...
25291 (can_avoid_literal_pool_for_label_p): ... into this new function.
25292 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25294 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25295 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25296 "M-profile targets with the MOVT instruction".
25298 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25300 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25301 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25302 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25304 2017-05-04 Martin Liska <mliska@suse.cz>
25306 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25307 variable cond_code.
25309 2017-05-04 Richard Biener <rguenther@suse.de>
25311 * tree.c (array_at_struct_end_p): Handle arrays at struct
25312 end with flexarrays more conservatively. Refactor and treat
25313 arrays of arrays or aggregates more strict. Fix
25314 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25315 * tree.c (array_at_struct_end_p): Adjust prototype.
25316 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25317 * gimple-fold.c (get_range_strlen): Likewise.
25318 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25320 2017-05-04 Richard Biener <rguenther@suse.de>
25322 PR tree-optimization/31130
25323 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25325 (supports_overflow_infinity): Likewise.
25326 (is_negative_overflow_infinity): Likewise.
25327 (is_positive_overflow_infinity): Likewise.
25328 (is_overflow_infinity): Likewise.
25329 (stmt_overflow_infinity): Likewise.
25330 (overflow_infinity_range_p): Likewise.
25331 (usable_range_p): Remove as always returning true.
25332 (make_overflow_infinity): Remove.
25333 (negative_overflow_infinity): Likewise.
25334 (positive_overflow_infinity): Likewise.
25335 (avoid_overflow_infinity): Likewise.
25336 (set_value_range): Adjust accordingly.
25337 (set_value_range_to_nonnegative): Likewise, remove now unused
25338 overflow_infinity arg.
25339 (vrp_operand_equal_p): Adjust.
25340 (update_value_range): Likewise.
25341 (range_int_cst_singleton_p): Likewise.
25342 (operand_less_p): Likewise.
25343 (compare_values_warnv): Likewise.
25344 (extract_range_for_var_from_comparison_expr): Likewise.
25345 (vrp_int_const_binop): Likewise.
25346 (zero_nonzero_bits_from_vr): Likewise.
25347 (extract_range_from_multiplicative_op_1): Likewise.
25348 (extract_range_from_binary_expr_1): Likewise.
25349 (extract_range_from_unary_expr): Likewise.
25350 (extract_range_from_comparison): Likewise.
25351 (extract_range_basic): Likewise.
25352 (adjust_range_with_scev): Likewise.
25353 (compare_ranges): Likewise.
25354 (compare_range_with_value): Likewise.
25355 (dump_value_range): Likewise.
25356 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25358 (simplify_cond_using_ranges): Adjust.
25360 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25362 * brig-builtins.def: Added a builtin for class_f64.
25363 * builtin-types.def: Added a builtin type needed by class_f64.
25365 2017-05-03 Jason Merrill <jason@redhat.com>
25367 * timevar.def: Add TV_CONSTEXPR.
25369 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25371 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25373 2017-05-03 Martin Jambor <mjambor@suse.cz>
25375 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25376 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25377 * ipa-cp.c (ipcp_propagate_stage): Do not call
25378 ipa_update_after_lto_read.
25379 * ipa-inline.c (ipa_inline): Likewise.
25381 2017-05-03 Martin Jambor <mjambor@suse.cz>
25383 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25384 tag. Added a default constructor and a destructor.
25385 (ipa_edge_args_sum_t): New class;
25386 (ipa_edge_args_sum): Declare.
25387 (ipa_edge_args_vector): Remove declaration.
25388 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25389 (ipa_free_edge_args_substructures): Remove declaration.
25390 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25391 (ipa_edge_args_info_available_for_edge_p): Likewise.
25392 * ipa-prop.c (ipa_edge_args_vector): Removed.
25393 (edge_removal_hook_holder): Likewise.
25394 (edge_duplication_hook_holder): Likewise.
25395 (ipa_edge_args_sum): New variable.
25396 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25397 ipa_edge_args_vector.
25398 (ipa_free_edge_args_substructures): Likewise.
25399 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25400 ipa_edge_args_vector.
25401 (ipa_edge_removal_hook): Turned into method
25402 ipa_edge_args_sum_t::remove.
25403 (ipa_edge_duplication_hook): Turned into method
25404 ipa_edge_args_sum_t::duplicate.
25405 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25406 registering edge hooks.
25407 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25408 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25409 ipa_edge_args_sum instead of ipa_edge_args_vector.
25410 * ipa-profile.c (ipa_profile): Likewise.
25412 2017-05-03 Martin Jambor <mjambor@suse.cz>
25414 * symbol-summary.h (function_summary): New method exists.
25415 (function_summary::symtab_removal): Deallocate through release.
25416 (call_summary): New class.
25417 (gt_ggc_mx): New overload.
25418 (gt_pch_nx): Likewise.
25419 (gt_pch_nx): Likewise.
25421 2017-05-03 Jeff Law <law@redhat.com>
25423 PR tree-optimization/78496
25424 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25425 from simplify_cond_using_ranges. Split off code to walk
25426 backwards through casts into ...
25427 (simplify_cond_using_ranges_2): New function.
25428 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25429 (execute_vrp): After identifying jump threads, call
25430 simplify_cond_using_ranges_2.
25432 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25435 * ipa-inline.h (inline_summary): Add ctor.
25436 (create_ggc): Do not use ggc_cleared_alloc.
25438 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25439 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25441 * gcc.c (handle_braces): Support escaping in switch matching
25443 * doc/invoke.texi (Spec Files): Document it.
25444 Remove superfluous @code markup in items.
25446 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25448 * diagnostic-show-locus.c (struct column_range): New struct.
25449 (get_affected_columns): New function.
25450 (get_printed_columns): New function.
25451 (struct correction): New struct.
25452 (correction::ensure_capacity): New function.
25453 (correction::ensure_terminated): New function.
25454 (struct line_corrections): New struct.
25455 (line_corrections::~line_corrections): New dtor.
25456 (line_corrections::add_hint): New function.
25457 (layout::print_trailing_fixits): Reimplement in terms of the new
25459 (selftest::test_overlapped_fixit_printing): New function.
25460 (selftest::diagnostic_show_locus_c_tests): Call it.
25462 2017-05-03 Nathan Sidwell <nathan@acm.org>
25464 Canonicalize canonical type hashing
25465 * tree.h (type_hash_canon_hash): Declare.
25466 * tree.c (type_hash_list, attribute_hash_list): Move into
25467 type_hash_canon_hash.
25468 (build_type_attribute_qual_variant): Break out hash code calc into
25469 type_hash_canon_hash.
25470 (type_hash_canon_hash): New. Generic type hash computation.
25471 (build_range_type_1, build_array_type_1, build_function_type,
25472 build_method_type_directly, build_offset_type, build_complex_type,
25473 make_vector_type): Call it.
25475 2017-05-03 Richard Biener <rguenther@suse.de>
25477 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25478 When all DRs have unknown misaligned do not always peel
25479 when there is a store but apply the same costing model as if
25480 there were only loads.
25482 2017-05-03 Richard Biener <rguenther@suse.de>
25485 PR tree-optimization/80492
25486 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25487 compare_base_decls returning dont-know properly.
25489 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25491 * config/arm/iterators.md (CCSI): New mode iterator.
25492 (arch): New mode attribute.
25493 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25494 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25495 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25496 code iterator for success result mode.
25497 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25498 the corresponding new insn generators.
25500 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25503 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25504 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25506 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25508 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25509 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25510 (DDR_A): Wrap DDR argument in brackets.
25511 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25512 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25513 (DDR_REVERSED_P): Likewise.
25515 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25517 PR tree-optimization/79472
25518 * tree-switch-conversion.c (struct switch_conv_info): Add
25519 contiguous_range and default_case_nonstandard fields.
25520 (collect_switch_conv_info): Compute contiguous_range and
25521 default_case_nonstandard fields, don't clear final_bb if
25522 contiguous_range and only the default case doesn't have the required
25524 (check_all_empty_except_final): Set default_case_nonstandard instead
25525 of failing if contiguous_range and the default case doesn't have empty
25527 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25528 and only the default case doesn't have the required constants. Skip
25530 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25531 if default_case_nonstandard.
25532 (build_constructors): Build constant 1 just once. Assert that default
25533 values aren't inserted in between cases if contiguous_range. Skip
25535 (build_arrays): Skip virtual phis.
25536 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25537 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25538 Handle virtual phis.
25539 (gen_inbound_check): Handle default_case_nonstandard case.
25540 (process_switch): Adjust check_final_bb caller. Call
25541 gather_default_values with the first non-default case instead of
25542 default case if default_case_nonstandard.
25544 2017-05-02 Nathan Sidwell <nathan@acm.org>
25546 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25547 check. Fix formatting.
25549 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25551 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25552 errors when comparing specialized and unspecialized times.
25554 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25556 * diagnostic-show-locus.c
25557 (layout::should_print_annotation_line_p): Make private.
25558 (layout::print_annotation_line): Make private.
25559 (layout::annotation_line_showed_range_p): Make private.
25560 (layout::show_ruler): Make private.
25561 (layout::print_source_line): Make private. Pass in line and
25562 line_width, rather than calling location_get_source_line. Drop
25564 (layout::print_leading_fixits): New method.
25565 (layout::print_any_fixits): Rename to...
25566 (layout::print_trailing_fixits): ...this, and make private.
25567 Don't print newline fixits.
25568 (diagnostic_show_locus): Move logic for printing one row into...
25569 (layout::print_line): ...this new function. Move the
25570 location_get_source_line call and error-handling from
25571 print_source_line to here. Call print_leading_fixits, and rename
25572 print_any_fixits to print_trailing_fixits.
25573 (selftest::test_fixit_insert_containing_newline): Update now that
25574 newlines are partially supported.
25575 (selftest::test_fixit_insert_containing_newline_2): New test.
25576 (selftest::test_fixit_replace_containing_newline): Update comments.
25577 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25578 * edit-context.c (class added_line): New class.
25579 (class edited_line): Describe newline handling in comment.
25580 (edited_line::actually_edited_p): New method.
25581 (edited_line::print_content): Delete redundant decl.
25582 (edited_line::m_predecessors): New field.
25583 (edited_file::print_content): Call edited_line::print_content.
25584 (edited_file::print_diff): Update to support newlines.
25585 (edited_file::print_diff_hunk): Likewise.
25586 (edited_file::print_run_of_changed_lines): New function.
25587 (edited_file::print_diff_line): Convert to...
25588 (print_diff_line): ...this.
25589 (edited_file::get_effective_line_count): New function.
25590 (edited_line::edited_line): Initialize new field m_predecessors.
25591 (edited_line::~edited_line): Clean up m_predecessors.
25592 (edited_line::apply_fixit): Handle newlines.
25593 (edited_line::get_effective_line_count): New function.
25594 (edited_line::print_content): New function.
25595 (edited_line::print_diff_lines): New function.
25596 (selftest::test_applying_fixits_insert_containing_newline): New
25598 (selftest::test_applying_fixits_replace_containing_newline): New
25600 (selftest::insert_line): New function.
25601 (selftest::test_applying_fixits_multiple_lines): Add example of
25603 (selftest::edit_context_c_tests): Call the new tests.
25605 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25607 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25608 parameter cand. Update dump information.
25609 (get_computation_cost): Update uses.
25611 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25613 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25614 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25615 (get_computation_at, rewrite_use_address): Update use of
25616 get_computation_aff.
25618 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25620 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25621 (get_computation): Delete.
25622 (get_computation_cost): Implement like get_computation_cost_at.
25623 Use get_computation_at.
25624 (get_computation_cost_at): Delete.
25625 (rewrite_use_nonlinear_expr): Use get_computation_at.
25626 (rewrite_use_compare, remove_unused_ivs): Ditto.
25628 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25630 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25632 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25634 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25635 (ivopts_global_cost_for_size): Rename parameter and update uses.
25636 (iv_ca_recount_cost): Update uses.
25637 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25638 candidates seperately in n_invs and n_cands.
25639 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25641 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25643 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25644 (find_inv_vars_cb): New.
25645 (find_depends): Renamed to ...
25646 (find_inv_vars): ... this.
25647 (add_candidate_1, force_var_cost): Call find_inv_vars.
25648 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25650 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25652 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25653 inv_vars. Add inv_exprs.
25654 (struct iv_cand): Rename depends_on to inv_vars.
25655 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25656 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25657 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25658 (dump_cand): Dump inv_vars.
25659 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25660 (record_invariant, find_depends, add_candidate_1): Ditto.
25661 (set_group_iv_cost, force_var_cost): Ditto.
25662 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25663 (get_computation_cost_at, get_computation_cost): Ditto.
25664 (determine_group_iv_cost_generic): Ditto.
25665 (determine_group_iv_cost_address): Ditto.
25666 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25667 (determine_group_iv_costs): Ditto.
25668 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25669 (iv_ca_set_remove_invariants): Renamed to ...
25670 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25671 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25672 (iv_ca_set_add_invariants): Renamed to ...
25673 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25674 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25675 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25676 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25677 (create_new_ivs): Remove useless dump.
25679 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25681 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25683 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25684 (iv_ca_set_no_cp, create_new_iv): Ditto.
25686 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25688 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25690 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25692 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25693 function tree_check2.
25695 2017-05-02 Martin Liska <mliska@suse.cz>
25697 * doc/gcov.texi: Add missing preposition.
25698 * gcov.c (function_info::function_info): Properly fill up
25699 all member variables.
25701 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25703 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25705 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25707 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25709 2017-05-02 Martin Liska <mliska@suse.cz>
25712 * lto-streamer-in.c (lto_read_tree_1): Remove
25713 LTO_STREAMER_DEBUG.
25714 * lto-streamer.c (struct tree_hash_entry): Likewise.
25715 (struct tree_entry_hasher): Likewise.
25716 (tree_entry_hasher::hash): Likewise.
25717 (tree_entry_hasher::equal): Likewise.
25718 (lto_streamer_init): Likewise.
25719 (lto_orig_address_map): Likewise.
25720 (lto_orig_address_get): Likewise.
25721 (lto_orig_address_remove): Likewise.
25722 * lto-streamer.h: Likewise.
25723 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25724 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25726 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25728 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25729 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25730 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25731 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25732 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25733 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25734 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25735 (mm_maskz_sub_ss): New intrinsics.
25736 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25737 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25738 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25739 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25740 (__builtin_ia32_subss_mask_round): New builtins.
25741 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25742 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25743 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25745 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25746 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25748 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25751 2017-05-02 Martin Jambor <mjambor@suse.cz>
25753 PR tree-optimization/78687
25754 * tree-sra.c (access): New field parent.
25755 (process_subtree_disqualification): New function.
25756 (disqualify_candidate): Call it.
25757 (build_accesses_from_assign): Reset write flag if creating an
25759 (build_access_subtree): Fill in parent field and also prpagate
25760 down grp_write flag.
25761 (create_artificial_child_access): New parameter set_grp_write, set
25762 grp_write to its value.
25763 (propagate_subaccesses_across_link): Also propagate grp_write flag
25765 (propagate_all_subaccesses): Push the closest parent back to work
25766 queue if add_access_to_work_queue returned true.
25768 2017-05-02 Richard Biener <rguenther@suse.de>
25770 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25771 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25772 -fstrict-overflow documentation.
25773 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25774 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25775 flag_strict_overflow.
25776 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25777 * lto-opts.c (lto_write_options): Do not stream it.
25778 * lto-wrapper.c (merge_and_complain): Do not handle it.
25779 * opts.c (default_options_table): Do not set -fstrict-overflow.
25780 (finish_options): Likewise do not clear it when sanitizing.
25781 * simplify-rtx.c (simplify_const_relational_operation): Do not
25782 test flag_strict_overflow.
25784 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25786 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25787 using enabled attribute.
25788 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25789 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25790 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25791 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25792 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25793 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25794 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25795 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25796 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25797 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25799 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25801 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25803 2017-05-02 Richard Biener <rguenther@suse.de>
25805 PR tree-optimization/80591
25807 2017-04-10 Richard Biener <rguenther@suse.de>
25809 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25812 2017-05-02 Richard Biener <rguenther@suse.de>
25814 PR tree-optimization/80549
25815 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25816 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25817 headers if they do not have a preheader.
25819 2017-05-02 Martin Liska <mliska@suse.cz>
25822 * common.opt: Fix typo.
25823 * doc/invoke.texi: Likewise.
25825 2017-05-01 Jan Beulich <jbeulich@suse.com>
25827 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25828 swapping, add (x,x,m,x,n) alternative.
25830 2017-05-01 Nathan Sidwell <nathan@acm.org>
25832 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25833 unnecessary unadjusted_alignment check.
25835 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25838 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25840 * gimplify.c (gimplify_cilk_detach): New function.
25841 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25842 * tree-core.h: Document EXPR_CILK_SPAWN.
25843 * tree.h (EXPR_CILK_SPAWN): Define.
25845 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25847 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25848 to use new fixit_hint representation, using the "replace" logic.
25849 (get_line_span_for_fixit_hint): Likewise.
25850 (layout::print_any_fixits): Likewise.
25851 (selftest::test_one_liner_many_fixits): Rename to...
25852 (selftest::test_one_liner_many_fixits_1): ...this, and update
25853 comment and expected output to reflect that the multiple fix-it
25854 hints are now consolidated into one insertion.
25855 (selftest::test_one_liner_many_fixits_2): New test.
25856 (selftest::test_diagnostic_show_locus_one_liner): Update for
25858 (selftest::test_fixit_consolidation): Update for fix-it API
25860 * diagnostic.c (print_parseable_fixits): Likewise.
25861 * edit-context.c (edited_line::m_line_events): Convert from
25862 auto_vec <line_event *> to auto_vec <line_event>.
25863 (class line_event): Convert from abstract base class to a concrete
25864 class, taking over the role of replace_event.
25865 (class insert_event): Delete.
25866 (class replace_event): Rename to class line_event. Convert to
25868 (edit_context::add_fixits): Reimplement.
25869 (edit_context::apply_insert): Delete.
25870 (edit_context::apply_replace): Rename to...
25871 (edit_context::apply_fixit): ...this. Convert to half-open range.
25872 (edited_file::apply_insert): Delete.
25873 (edited_file::apply_replace): Rename to...
25874 (edited_file::apply_fixit): ...this.
25875 (edited_line::~edited_line): Drop deletion of events.
25876 (edited_line::apply_insert): Delete.
25877 (edited_line::apply_replace): Rename to...
25878 (edited_line::apply_fixit): ...this. Convert to half-open range.
25879 Update for change to type of m_line_events.
25880 * edit-context.h (edit_context::apply_insert): Delete.
25881 (edit_context::apply_replace): Rename to...
25882 (edit_context::apply_fixit): ...this.
25884 2017-05-01 Martin Sebor <msebor@redhat.com>
25886 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25889 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25892 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25893 __get_cpuid_max returns 0.
25894 (__get_cpuid_count): Ditto.
25896 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25898 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25899 replacement expression is another instance of one of its arguments.
25901 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25904 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25905 check for stack push/pop autoinc.
25906 * config/i386/i386.c (ix86_agi_dependent): Return false
25907 if the only reason why modified_in_p returned true is that
25908 addr is SP based and set_insn is a push or pop.
25910 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25912 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25915 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25918 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25919 (account_size_time): Use two predicates - exec_pred and
25921 (evaluate_conditions_for_known_args): Compute both clause and
25923 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25924 (inline_summary_t::duplicate): Update.
25925 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25927 (compute_inline_parameters): Likewise.
25928 (estimate_edge_size_and_time): Update caluclation of time.
25929 (estimate_node_size_and_time): Compute both time and nonspecialized
25931 (estimate_ipcp_clone_size_and_time): Update.
25932 (inline_merge_summary): Update.
25933 (do_estimate_edge_time): Update.
25934 (do_estimate_edge_size): Update.
25935 (do_estimate_edge_hints): Update.
25936 (inline_read_section, inline_write_summary): Stream both new predicates.
25937 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25939 (compute_inlined_call_time): Cleanup.
25940 (big_speedup_p): Update.
25941 (edge_badness): Update.
25942 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25943 (size_time_entry): Replace predicate by exec_predicate and
25944 nonconst_predicate.
25945 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25946 (estimate_edge_time): Return also nonspec_time.
25947 (reset_edge_growth_cache): Update.
25949 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25951 PR rtl-optimization/80491
25952 * ifcvt.c (noce_process_if_block): When looking for x setter
25953 with missing else_bb, don't check only the insn right before
25954 cond_earliest, but look for the last insn that x is modified in
25955 within the same bb.
25957 PR rtl-optimization/80491
25958 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25960 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25962 PR tree-optimization/80487
25963 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25965 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25967 PR tree-optimization/79697
25968 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25969 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25970 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25972 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25973 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25975 2017-04-28 Martin Sebor <msebor@redhat.com>
25977 PR tree-optimization/80523
25978 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25979 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25981 (maybe_warn, format_directive, parse_directive): Use new functions.
25982 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25984 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25986 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25988 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25990 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25991 target_header_dir): Set correctly.
25992 * configure: Regenerated.
25993 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25994 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25995 instead of SYSTEM_HEADER_DIR.
25997 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25999 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26000 (estimate_local_effects): Likewise.
26001 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26002 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26003 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26004 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26005 do_estimate_edge_time, estimate_edge_time): Likewise.
26006 * ipa-inline-analysis.c (estimate_node_size_and_time,
26007 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26008 (estimate_time_after_inlining): Remove.
26010 2017-04-28 Martin Liska <mliska@suse.cz>
26012 * doc/gcov.texi: Enhance documentation of gcov.
26014 2017-04-28 Martin Liska <mliska@suse.cz>
26016 * doc/gcov.texi: Sort options in alphabetic order.
26017 * doc/gcov-dump.texi: Likewise.
26018 * doc/gcov-tool.texi: Likewise.
26019 * gcov.c (print_usage): Likewise.
26020 * gcov-dump.c (print_usage): Likewise.
26021 * gcov-tool.c (print_merge_usage_message): Likewise.
26022 (print_rewrite_usage_message): Likewise.
26023 (print_overlap_usage_message): Likewise.
26025 2017-04-28 Martin Liska <mliska@suse.cz>
26027 PR gcov-profile/53915
26028 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26030 2017-04-28 Martin Liska <mliska@suse.cz>
26032 PR gcov-profile/79891
26033 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26034 is marked by compiler as living on a line.
26035 (get_cycles_count): Remove usage of the union.
26036 (output_intermediate_file): Likewise.
26037 (find_source): Fix GNU coding style.
26038 (accumulate_line_counts): Remove old non-all block mode.
26039 (output_lines): Remove usage of the union.
26040 * profile.c (output_location): Include all BBs, even if
26041 belonging to a same line (and file) as a previous BB.
26043 2017-04-28 Martin Liska <mliska@suse.cz>
26045 * gcov.c (process_args): Handle new argument 'w'.
26046 (read_graph_file): Assign ID to BBs.
26047 (output_branch_count): Display BB # if verbose flag is set.
26048 (output_lines): Likewise for arcs.
26049 (print_usage): Add '--verbose' option help.
26050 * doc/gcov.texi: Document --verbose (-w) option.
26052 2017-04-28 Martin Liska <mliska@suse.cz>
26054 * gcov.c (struct block_location_info): New struct.
26055 (process_file): Fill up the new structure.
26056 (read_graph_file): Replace usage of encoding by the newly added
26058 (add_line_counts): Likewise.
26059 (accumulate_line_counts): Remove usage of the union.
26060 (function_info::function_info): New function.
26061 (function_info::~function_info): Likewise.
26062 (process_file): Call delete instead of release_function.
26063 (release_function): Release the function.
26064 (release_structures): Call delete instead of release_function.
26065 (solve_flow_graph): Replace usage of num_blocks.
26066 (find_exception_blocks): Likewise.
26067 (output_lines): Fix GNU coding style.
26069 2017-04-28 Martin Liska <mliska@suse.cz>
26072 * coverage.c (coverage_remove_note_file): New function.
26073 * coverage.h: Declare the function.
26074 * toplev.c (finalize): Clean if an error has been seen.
26076 2017-04-28 Martin Liska <mliska@suse.cz>
26078 PR gcov-profile/80031
26079 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26080 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26081 * gcov.c (read_graph_file): Read just number of blocks.
26082 * profile.c (branch_prob): Do not stream 0 flags per a basic
26085 2017-04-28 Martin Liska <mliska@suse.cz>
26087 * gcov-dump.c (tag_*): Add new argument to declarations.
26088 (dump_gcov_file): Likewise.
26089 (tag_blocks): Add and use new argument depth.
26090 (tag_arcs): Likewise.
26091 (tag_lines): Likewise.
26092 (tag_counters): Likewise.
26093 (tag_summary): Likewise.
26094 (dump_working_sets): Use depth to do a proper indentation.
26096 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26099 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26100 * symtab.c (symtab_node::debug_symtab): Move definition here.
26102 2017-04-28 Richard Biener <rguenther@suse.de>
26104 * lto-streamer.h (LTO_major_version): Bump to 7.
26106 2017-04-28 Richard Biener <rguenther@suse.de>
26108 * tree-vrp.c (assert_info): New struct.
26109 (add_assert_info): New helper.
26110 (register_edge_assert_for_2): Refactor to add asserts to a vector
26112 (register_edge_assert_for_1): Likewise.
26113 (register_edge_assert_for): Likewise.
26114 (finish_register_edge_assert_for): New helper actually registering
26115 asserts where live on edge.
26116 (find_conditional_asserts): Adjust.
26117 (find_switch_asserts): Likewise.
26118 (evrp_dom_walker::try_find_new_range): Generalize.
26119 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26121 2017-04-27 Marek Polacek <polacek@redhat.com>
26124 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26125 arg10 and arg11 to itype.
26127 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26129 * doc/extend.texi (Object Size Checking): Improve grammar.
26131 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26134 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26135 that the logic for permitting reciprocal estimates matches that
26138 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26141 * tree.c (type_cache_hasher::equal): Only compare
26142 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26143 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26144 non-aggregate element types.
26145 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26146 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26148 2017-04-27 Richard Biener <rguenther@suse.de>
26150 PR middle-end/80533
26151 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26152 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26153 keeping a reference to a trailing array.
26155 2017-04-27 Richard Biener <rguenther@suse.de>
26157 PR middle-end/80539
26158 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26159 being in loop-closed SSA form conservatively.
26160 (chrec_fold_multiply_poly_poly): Likewise.
26162 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26164 PR middle-end/79665
26165 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26166 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26168 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26171 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26172 (aarch64_function_arg_alignment): Return unsigned int again, but still
26173 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26174 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26175 Don't emit -Wpsabi note.
26176 (aarch64_function_arg_boundary): Likewise.
26177 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26180 2017-04-26 Nathan Sidwell <nathan@acm.org>
26182 * tree.h (crc32_unsigned_n): Declare.
26183 (crc32_unsigned, crc32_unsigned): Make inline.
26184 * tree.c (crc32_unsigned_bits): Replace with ...
26185 (crc32_unsigned_n): ... this.
26186 (crc32_unsigned, crc32_byte): Remove.
26187 (crc32_string): Remove unnecessary braces.
26189 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26191 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26192 * ipa-inline-analysis.c (MAX_TIME): Remove.
26193 (account_size_time): Use sreal for time.
26194 (dump_inline_summary): Update.
26195 (estimate_function_body_sizes): Update.
26196 (estimate_edge_size_and_time): Update.
26197 (estimate_calls_size_and_time): Update.
26198 (estimate_node_size_and_time): Update.
26199 (inline_merge_summary): Update.
26200 (inline_update_overall_summary): Update.
26201 (estimate_time_after_inlining): Update.
26202 (inline_read_section): Update.
26203 (inline_write_summary): Update.
26204 * ipa-inline.c (compute_uninlined_call_time): Update.
26205 (compute_inlined_call_time): Update.
26206 (recursive_inlining): Update.
26207 (inline_small_functions): Update.
26208 (dump_overall_stats): Update.
26209 * ipa-inline.h: Include sreal.h.
26210 (size_time_entry): Turn time to sreal.
26211 (inline_summary): Turn self_time nad time to sreal.
26213 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26215 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26217 (sreal::stream_out, sreal::stream_in): New.
26218 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26220 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26222 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26225 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26228 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26230 (dimode_scalar_chain::compute_convert_gain): Ditto.
26231 (dimode_scalar_chain::make_vector_copies): Ditto.
26232 (dimode_scalar_chain::convert_reg): Ditto.
26233 (dimode_scalar_chain::convert_insn): Ditto.
26234 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26235 (VI248_AVX512BW_1): New mode iterator.
26236 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26237 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26240 2017-04-25 Martin Sebor <msebor@redhat.com>
26242 PR tree-optimization/80497
26243 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26244 constants are representable in HOST_WIDE_INT.
26245 (parse_directive): Ditto.
26247 2017-04-25 Martin Sebor <msebor@redhat.com>
26250 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26251 (new_zero_array): Adjust signature.
26252 (dom_info::dom_init): Used unsigned rather that size_t.
26253 (dom_info::dom_info): Same.
26255 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26256 Jakub Jelinek <jakub@redhat.com>
26259 * config/arm/arm.c: Include gimple.h.
26260 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26261 returns negative, increment ncrn only if it returned positive.
26262 (arm_needs_doubleword_align): Return int instead of bool,
26263 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26264 members, but if there is any such non-FIELD_DECL
26265 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26266 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26267 returns negative, increment nregs only if it returned positive.
26268 (arm_setup_incoming_varargs): Likewise.
26269 (arm_function_arg_boundary): Emit -Wpsabi note if
26270 arm_needs_doubleword_align returns negative, return
26271 DOUBLEWORD_ALIGNMENT only if it returned positive.
26273 2017-04-25 Marek Polacek <polacek@redhat.com>
26276 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26277 first argument to type.
26279 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26282 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26283 type checks to test for compatibility instead of equality.
26285 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26286 Jakub Jelinek <jakub@redhat.com>
26289 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26291 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26292 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26293 the alignment computation, but return their maximum in warn_alignment.
26294 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26295 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26297 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26298 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26301 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26303 * config/arc/simdext.md (dmpyh): Fix typo.
26305 2017-04-25 Richard Biener <rguenther@suse.de>
26307 PR tree-optimization/80492
26308 * alias.c (compare_base_decls): Handle registers with asm
26309 specification conservatively.
26310 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26311 compare_base_decls returning dont-know properly.
26313 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26315 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26316 (legitimate_offset_address_p): New function.
26317 (arc_legitimate_address_p): Use above function.
26319 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26321 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26323 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26325 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26326 ACCH registers whenever they are available.
26328 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26330 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26331 double regs fix when not used.
26333 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26335 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26337 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26338 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26340 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26342 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26343 when emitting short ADD instructions.
26345 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26347 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26349 (cmpsi_cc_c_insn): Likewise.
26350 (cbranchsi4_scratch): Compute proper instruction length using
26351 compact_hreg_operand.
26352 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26354 2017-04-25 Richard Biener <rguenther@suse.de>
26356 PR middle-end/80509
26357 * passes.c (pass_manager::pass_manager): Initialize
26358 m_name_to_pass_map.
26360 2017-04-25 Richard Biener <rguenther@suse.de>
26362 PR tree-optimization/79201
26363 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26365 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26368 * config/s390/vector.md: Split MEM->GPR vector moves for
26369 non-s_operand addresses.
26371 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26374 * config/s390/predicates.md (reload_const_wide_int_operand): New
26376 * config/s390/s390.md ("movti"): Remove d/P alternative.
26377 ("movti_bigconst"): New pattern definition.
26379 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26382 * s390-protos.h (s390_expand_cs_hqi): Removed.
26383 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26384 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26385 modes as well as CCZ1mode and CCZmode.
26386 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26387 signature of s390_emit_compare_and_swap.
26388 (s390_expand_cs_hqi): Likewise, make static.
26389 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26390 compare-and-swap, in some cases.
26391 (s390_expand_cs): Wrapper function.
26392 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26394 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26395 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26396 patterns for small and large integers. Forbid symref memory operands.
26397 Move expander to s390.c. Require cc register.
26398 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26399 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26400 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26401 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26402 symref memory operands. Remove CC mode and call s390_match_ccmode
26404 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26406 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26408 * config/s390/s390.md (define_peephole2): New peephole to help
26409 combining the load-and-test pattern with volatile memory.
26411 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26413 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26414 with CCZmode for TARGET_Z196.
26416 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26418 PR rtl-optimization/80501
26419 * combine.c (make_compound_operation_int): Set subreg_code to SET
26420 even for AND with mask of the sign bit of mode.
26422 PR rtl-optimization/80500
26423 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26424 sum's initial value.
26426 2017-04-25 Julian Brown <julian@codesourcery.com>
26427 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26429 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26431 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26433 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26435 2017-04-25 Julian Brown <julian@codesourcery.com>
26436 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26438 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26439 (thunderx2t99_sha): New Reservation.
26441 2017-04-25 Julian Brown <julian@codesourcery.com>
26442 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26444 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26445 type for 1-element load.
26447 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26449 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26451 2017-04-24 Martin Jambor <mjambor@suse.cz>
26453 PR tree-optimization/80293
26454 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26455 char arrays not totally scalarizable if it is false.
26456 (analyze_all_variable_accesses): Pass correct value in the new
26457 parameter. Add a statistics counter.
26459 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26461 PR middle-end/79931
26462 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26464 2017-04-24 Richard Biener <rguenther@suse.de>
26466 PR tree-optimization/80494
26467 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26468 out for complex types.
26470 2017-04-24 Richard Biener <rguenther@suse.de>
26472 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26473 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26474 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26475 (DFS): Adjust and never fail.
26476 (sccvn_dom_walker::fail): Remove.
26477 (sccvn_dom_walker::before_dom_children): Adjust.
26478 (run_scc_vn): Likewise and never fail.
26479 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26480 (pass_fre::execute): Likewise.
26482 2017-04-24 Richard Biener <rguenther@suse.de>
26484 PR tree-optimization/79725
26485 * tree-ssa-sink.c (statement_sink_location): Return whether
26486 failure reason was zero uses. Move that check later.
26487 (sink_code_in_bb): Deal with zero uses by removing the stmt
26490 2017-04-24 Richard Biener <rguenther@suse.de>
26493 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26494 pointer-based references.
26496 2017-04-24 Richard Biener <rguenther@suse.de>
26499 * pass_manager.h (pass_manager::operator new): Remove.
26500 (pass_manager::operator delete): Likewise.
26501 * passes.c (pass_manager::operator new): Remove.
26502 (pass_manager::operator delete): Likewise.
26503 (pass_manager::pass_manager): Zero individual pass members.
26505 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26508 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26509 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26510 Check "XEXP (src, 1)" operand here.
26511 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26512 Check "XEXP (src, 1)" operand here.
26513 (dimode_scalar_chain::make_vector_copies): Detect count register
26514 of a shift instruction. Zero extend count register from QImode
26515 to DImode to satisfy vector shift pattern count operand predicate.
26516 Substitute vector shift count operand with a DImode copy.
26517 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26520 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26522 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26523 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26524 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26525 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26526 (UNSPEC_NOREX_MEM): Remove definition.
26528 2017-04-21 Richard Biener <rguenther@suse.de>
26530 PR tree-optimization/79547
26531 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26532 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26533 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26534 without any constraints.
26536 2017-04-21 Richard Biener <rguenther@suse.de>
26538 PR tree-optimization/78847
26539 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26541 2017-04-21 Richard Biener <rguenther@suse.de>
26543 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26544 (build_distinct_type_copy): Likewise.
26545 (build_variant_type_copy): Likewise.
26546 * tree.c (build_qualified_type): Pass down mem-stat info.
26547 (build_distinct_type_copy): Likewise.
26548 (build_variant_type_copy): Likewise.
26550 2017-04-21 Richard Biener <rguenther@suse.de>
26552 PR tree-optimization/80237
26553 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26555 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26556 for a simplified result.
26558 2016-04-21 Richard Biener <rguenther@suse.de>
26560 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26561 sth as strict as a simple_iv but a chrec without symbols and an
26562 operand defined in the loop we are peeling (and not some subloop).
26563 (propagate_constants_for_unrolling): Propagate all constants.
26565 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26568 * config/i386/i386.c (print_reg): Remove assert for disalowed
26569 regno values, call output_operand_lossage instead.
26571 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26574 * config/i386/constraints.md (Yc): New register constraint.
26575 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26576 Use Yc constraint for alternative 2 of operand 0. Remove
26577 preferred_for_speed attribute.
26579 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26581 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26582 lastprivate clauses in SIMT case.
26584 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26586 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26588 2017-04-20 Richard Biener <rguenther@suse.de>
26590 PR tree-optimization/57796
26591 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26592 as N scalar stores.
26593 (vect_model_load_cost): Cost gathers as N scalar loads.
26595 2017-04-20 Richard Biener <rguenther@suse.de>
26597 * ggc-page.c (ggc_allocated_p): Rename to ...
26598 (safe_lookup_page_table_entry): ... this and return the lookup
26600 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26602 2017-04-20 Richard Biener <rguenther@suse.de>
26604 PR tree-optimization/80453
26605 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26606 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26607 from the conditions.
26608 (vn_phi_eq): Pass them down.
26609 (vn_phi_lookup): Record them.
26610 (vn_phi_insert): Likewise.
26612 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26614 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26615 uninitialized variable warning to avoid buffer overrun.
26617 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26620 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26621 is suppressed for '{ 0 }' in C.
26623 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26625 * BASE-VER: Set to 8.0.0.
26627 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26629 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26630 priority .init_array and .fini_array section with SECTION_NOTYPE
26633 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26635 PR middle-end/80423
26636 * tree.h (build_array_type): Add typeless_storage default argument.
26637 * tree.c (type_cache_hasher::equal): Also compare
26638 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26639 (build_array_type): Add typeless_storage argument, set
26640 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26642 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26643 (build_array_type): Likewise. Add typeless_storage argument.
26645 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26646 Jakub Jelinek <jakub@redhat.com>
26648 PR tree-optimization/80426
26649 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26650 operation on symbolic operands, also compute the overflow for the
26651 invariant part when the operation degenerates into a negation.
26653 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26656 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26657 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26660 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26662 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26665 * config/avr/avr.c (tree.h): Include it.
26666 (cgraph.h): Include it.
26667 (avr_encode_section_info): Don't warn for uninitialized progmem
26668 variable if it's just an alias.
26670 2017-04-19 Richard Biener <rguenther@suse.de>
26673 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26674 when needed by AutoPGO.
26676 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26679 * doc/lto.texi: Remove an extra 'that'.
26681 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26683 PR rtl-optimization/80429
26684 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26685 are only used in debug insns.
26687 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26688 Vladimir Makarov <vmakarov@redhat.com>
26690 * config/sparc/predicates.md (input_operand): Add comment. Return
26691 true for any memory operand when LRA is in progress.
26692 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26694 2017-04-18 Jeff Law <law@redhat.com>
26697 * mips.md ({return,simple_return}_internal): Do not overwrite
26700 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26702 PR tree-optimization/80443
26703 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26704 instead of adding 1, subtract -1 and similarly instead of subtracting
26707 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26709 PR rtl-optimization/80357
26710 * haifa-sched.c (tmp_bitmap): New variable.
26711 (model_recompute): Handle duplicate use records.
26712 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26713 (free_global_sched_pressure_data): Free it.
26715 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26718 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26719 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26720 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26721 instead of SYSTEM_HEADER_DIR.
26723 2017-04-18 Jeff Law <law@redhat.com>
26725 PR middle-end/80422
26726 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26727 predecessors after walking up the insn chain.
26729 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26732 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26733 sizetype type into debug info.
26735 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26738 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26739 unneeded test for TARGET_UPPER_REGS_SF.
26740 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26742 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26745 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26746 instead of gsi_after_labels.
26748 2017-04-18 Jeff Law <law@redhat.com>
26750 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26754 2017-04-13 Jeff Law <law@redhat.com>
26755 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26756 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26758 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26761 * config/avr/avr.c (intl.h): Include it.
26762 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26764 2017-04-18 Martin Liska <mliska@suse.cz>
26766 PR gcov-profile/78783
26767 * gcov-tool.c (gcov_output_files): Validate that destination
26768 file is either removed by the tool or by a user.
26770 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26771 Guy Benyei <guybe@mellanox.com>
26773 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26774 block, and do not negate it, the stored id is already negative.
26776 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26778 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26780 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26783 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26784 masks of options that should be turned off if the VSX vector
26785 options are turned off.
26786 (OTHER_P8_VECTOR_MASKS): Likewise.
26787 (OTHER_VSX_VECTOR_MASKS): Likewise.
26788 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26789 rs6000_disable_incompatible_switches to validate no type switches
26791 (rs6000_incompatible_switch): New function to disallow turning on
26792 other vector options if -mno-vsx, -mno-power8-vector, or
26793 -mno-power9-vector are specified.
26795 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26797 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26799 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26801 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26802 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26803 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26804 (ARG_POINTER_CFA_OFFSET): Likewise.
26806 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26808 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26809 conditions to take advantage of various optimizations.
26811 2017-04-13 Jeff Law <law@redhat.com>
26813 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26814 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26815 (zero_extendsidi2_dext): Likewise.
26817 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26820 * fold-const.c (fold_ternary_loc): Revert
26821 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26824 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26826 PR rtl-optimization/80343
26827 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26828 new scratch pseudo.
26830 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26833 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26834 to ubsan_encode_value.
26836 2017-04-13 Jeff Law <law@redhat.com>
26838 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26839 appearing in DEBUG_INSNs.
26841 2017-04-13 Martin Liska <mliska@suse.cz>
26843 PR gcov-profile/80413
26844 * gcov-io.c (gcov_write_string): Copy to buffer just when
26845 allocated size is greater than zero.
26847 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26850 * dwarf2out.c (decls_for_scope): Ignore declarations of
26851 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26853 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26856 * ipa-visibility.c (non_local_p): Fix typos.
26857 (localize_node): When localizing symbol in same comdat group,
26858 dissolve the group only when we know external symbols are going
26860 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26862 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26864 PR tree-optimization/79390
26865 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26866 order does not result in usable sequence, retry with reversed operand
26872 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26873 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26874 op0 instead of fold_convert_loc (loc, type, arg0).
26876 2017-04-12 Jeff Law <law@redhat.com>
26878 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26879 has a delay slot in the generated code.
26881 * config/cris/cris.md (cris_preferred_reload_class): Return
26882 GENNONACR_REGS rather than GENERAL_REGS.
26884 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26887 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26888 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26889 signedness of the result type.
26891 2017-04-12 Richard Biener <rguenther@suse.de>
26892 Jeff Law <law@redhat.com>
26894 PR tree-optimization/80359
26895 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26896 trim stores to TARGET_MEM_REFs.
26898 2017-04-12 Richard Biener <rguenther@suse.de>
26900 PR tree-optimization/79390
26901 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26902 threading case even more.
26904 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26907 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26908 for quad_address_p for TImode, instead of just not indexed_address.
26910 2017-04-12 Richard Biener <rguenther@suse.de>
26911 Bernd Edlinger <bernd.edlinger@hotmail.de>
26913 PR middle-end/79671
26914 * alias.c (component_uses_parent_alias_set_from): Handle
26915 TYPE_TYPELESS_STORAGE.
26916 (get_alias_set): Likewise.
26917 * tree-core.h (tree_type_common): Add typeless_storage flag.
26918 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26919 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26920 for types containing members with TYPE_TYPELESS_STORAGE.
26921 (place_field): Likewise.
26922 (layout_type): Likewise for ARRAY_TYPE.
26923 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26924 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26925 TYPE_TYPELESS_STORAGE.
26926 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26928 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26931 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26932 first argument to type.
26934 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26938 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26939 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26940 (rs6000_expand_binop_builtin): Likewise.
26941 (rs6000_expand_ternop_builtin): Likewise; also add missing
26942 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26943 vshasigma built-ins.
26944 * doc/extend.texi: Document that vec_xxpermdi's third argument
26945 must be a constant.
26947 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26949 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26950 Use shift_const cost parameter when calculating gain of STV shifts.
26952 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26954 PR rtl-optimization/70478
26955 * lra-constraints.c (process_alt_operands): Check memory for
26956 disfavoring memory insn operand.
26958 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26960 PR middle-end/80100
26961 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26962 left shift in unsigned HOST_WIDE_INT type.
26964 PR rtl-optimization/80385
26965 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26966 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26969 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26970 if they have any depend clauses.
26972 2017-04-11 Martin Liska <mliska@suse.cz>
26975 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26976 * ipa-split.c (split_function): Create a local comdat symbol
26977 if caller is in a comdat group.
26979 2017-04-11 Martin Liska <mliska@suse.cz>
26982 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26985 2017-04-11 Martin Sebor <msebor@redhat.com>
26987 PR middle-end/80364
26988 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26989 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26991 (directive::set_width, directive::set_precision, format_character):
26993 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26996 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26999 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27000 conflict, set target->arch_name instead of target->cpu_name.
27002 2017-04-11 Richard Biener <rguenther@suse.de>
27004 PR tree-optimization/80374
27005 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27006 build_zero_cst, remove fold_convertible_p check again.
27008 2017-04-11 Martin Liska <mliska@suse.cz>
27011 * ubsan.c (instrument_object_size): Do not instrument register
27014 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27017 * config/i386/i386-builtin-types.def
27018 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27019 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27020 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27021 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27022 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27023 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27024 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27025 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27026 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27027 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27028 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27029 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27030 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27031 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27032 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27033 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27034 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27035 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27036 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27037 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27038 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27039 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27040 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27041 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27042 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27043 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27044 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27045 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27046 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27047 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27048 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27049 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27050 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27051 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27052 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27053 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27054 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27055 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27056 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27057 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27058 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27059 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27060 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27061 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27062 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27063 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27065 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27066 flag to second_arg_count, handle 4 argument function type _COUNT
27067 aliases, handle second_arg_count on second argument rather than last.
27069 2017-04-10 Jeff Law <law@redhat.com>
27071 PR tree-optimization/80374
27072 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27073 record anything if we can not convert integer_zero_node to the
27076 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27079 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27080 Enhance special handling given to the TARGET_P9_MINMAX option in
27081 relation to certain other options.
27083 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27085 PR tree-optimization/80153
27086 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27087 remove POINTER_PLUS_EXPR's base part directly, rather than through
27090 2017-04-10 Richard Biener <rguenther@suse.de>
27091 Bin Cheng <bin.cheng@arm.com>
27093 PR tree-optimization/80153
27094 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27095 the first element of pointer type aff_tree. Build result expr in
27097 (add_elt_to_tree): Convert to type unconditionally. Remove other
27098 fold_convert calls.
27099 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27100 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27102 2017-04-10 Richard Biener <rguenther@suse.de>
27104 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27107 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27109 PR rtl-optimization/70478
27110 * lra-constraints.c (curr_small_class_check): New.
27111 (update_and_check_small_class_inputs): New.
27112 (process_alt_operands): Update curr_small_class_check. Disfavor
27113 alternative insn memory operands. Check available regs for small
27116 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27119 * config/mips/mips.opt (-mvirt): Update description.
27120 * doc/invoke.texi (-mvirt): Likewise.
27122 2017-04-10 Richard Biener <rguenther@suse.de>
27124 PR middle-end/80362
27125 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27126 looking for NEGATE_EXPR in -A / -B to A / B folding.
27128 2017-04-10 Martin Liska <mliska@suse.cz>
27130 PR gcov-profile/80224
27131 * gcov.c (print_usage): Fix usage string.
27132 (get_gcov_intermediate_filename): Remove.
27133 (output_gcov_file): Use both for normal and intermediate format.
27134 (generate_results): Do not initialize special file for
27135 intermediate format.
27137 2017-04-10 Richard Biener <rguenther@suse.de>
27139 PR tree-optimization/80304
27140 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27143 2017-04-10 Nathan Sidwell <nathan@acm.org>
27146 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27147 (rs6000_init_builtins): Use it.
27149 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27151 * config/arm/arm.md (<mrc>): Add mode to SET source.
27152 (<mrrc>): Likewise.
27154 2017-04-10 Richard Biener <rguenther@suse.de>
27156 PR middle-end/80344
27157 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27159 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27162 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27163 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27164 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27165 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27166 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27167 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27168 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27169 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27170 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27171 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27172 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27173 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27174 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27175 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27176 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27177 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27178 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27179 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27180 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27181 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27182 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27183 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27184 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27186 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27188 PR rtl-optimization/70478
27189 * lra-constraints.c: Reverse the last patch.
27191 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27193 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27194 Add comment for WCHAR_T.
27196 2017-04-08 Martin Liska <mliska@suse.cz>
27199 2017-04-07 Martin Liska <mliska@suse.cz>
27202 * ipa-split.c (split_function): Add function part to a same comdat
27205 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27208 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27210 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27212 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27213 (vec_pairing): ...to this.
27214 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27215 (rs6000_sched_init): Adjust for name change.
27216 (struct rs6000_sched_context): Likewise.
27217 (rs6000_init_sched_context): Likewise.
27218 (rs6000_set_sched_context): Likewise.
27220 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27226 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27228 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27229 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27230 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27232 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27234 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27236 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27238 PR rtl-optimization/70703
27239 * ira-color.c (update_conflict_hard_regno_costs): Use
27240 int64_t instead of HOST_WIDE_INT.
27242 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27244 PR rtl-optimization/70478
27245 * lra-constraints.c (process_alt_operands): Disfavor alternative
27246 insn memory operands.
27248 2017-04-07 Jeff Law <law@redhat.com>
27250 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27251 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27253 2017-04-07 Martin Liska <mliska@suse.cz>
27256 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27257 Show error message instead of an ICE.
27259 2017-04-07 Martin Liska <mliska@suse.cz>
27262 * ipa-split.c (split_function): Add function part to a same comdat
27265 2017-04-07 Richard Biener <rguenther@suse.de>
27267 PR middle-end/80341
27268 * tree.c (get_unwidened): Also handle ! for_type case for
27270 * convert.c (do_narrow): Split out from ...
27271 (convert_to_integer_1): ... here. Do not pass final truncation
27272 type to get_unwidened for TRUNC_DIV_EXPR.
27274 2017-04-07 Richard Biener <rguenther@suse.de>
27276 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27278 (aff_combination_const): Adjust.
27279 (aff_combination_scale): Likewise.
27280 (aff_combination_add_elt): Likewise.
27281 (aff_combination_add_cst): Likewise.
27282 (aff_combination_convert): Likewise.
27283 (add_elt_to_tree): Likewise. Remove unused argument.
27284 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27286 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27288 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27290 * config/arm/arm.c (arm_default_short_enums): Use
27291 ARM_DEFAULT_SHORT_ENUMS.
27292 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27294 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27297 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27298 members with redundant out-of-class redeclaration.
27300 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27303 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27304 * config/i386/i386.md (*zero_extendsidi2):
27305 Add (?*x,*x) and (?*v,*v) alternatives.
27307 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27310 * config/i386/i386.c (ix86_expand_builtin)
27311 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27312 mode from insn data. Convert operands to insn operand mode.
27313 Copy operands that don't satisfy insn predicate to a register.
27315 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27317 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27320 2017-04-06 Richard Biener <rguenther@suse.de>
27322 PR tree-optimization/80334
27323 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27324 preserve alignment of accesses.
27326 2017-04-06 Richard Biener <rguenther@suse.de>
27328 PR tree-optimization/80262
27329 * tree-sra.c (build_ref_for_offset): Preserve address-space
27331 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27332 Drop useless address-space information on MEM_REF offsets.
27334 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27336 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27338 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27340 PR rtl-optimization/70703
27341 * ira-color.c (update_conflict_hard_regno_costs): Use
27342 HOST_WIDE_INT instead of long.
27344 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27347 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27348 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27350 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27351 for x86_64 target. Handle -m3dnowa option.
27353 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27355 PR rtl-optimization/70703
27356 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27357 (update_conflict_hard_regno_costs): Use long instead of unsigned
27358 arithmetic for cost calculation.
27360 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27361 Bernd Edlinger <bernd.edlinger@hotmail.de>
27364 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27367 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27370 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27371 ptr_mode with Pmode throughout.
27372 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27373 into probe_stack_range and use DImode.
27375 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27378 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27379 call_eh_return is true.
27381 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27383 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27384 Initialize last_match_fntype_index.
27386 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27389 * tree-nvr.c: Include internal-fn.h.
27390 (pass_return_slot::execute): Ignore internal calls without
27393 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27394 Richard Biener <rguenther@suse.de>
27397 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27398 captures used multiple times, except for the last use.
27399 * generic-match-head.c: Include gimplify.h.
27401 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27403 PR tree-optimization/79390
27404 * target.h (struct noce_if_info): Declare.
27405 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27406 * target.def (noce_conversion_profitable_p): New target hook.
27407 * ifcvt.h (struct noce_if_info): New type, moved from ...
27408 * ifcvt.c (struct noce_if_info): ... here.
27409 (noce_conversion_profitable_p): Renamed to ...
27410 (default_noce_conversion_profitable_p): ... this. No longer
27412 (noce_try_store_flag_constants, noce_try_addcc,
27413 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27414 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27415 instead of noce_conversion_profitable_p.
27416 * config/i386/i386.c: Include ifcvt.h.
27417 (ix86_option_override_internal): Don't override
27418 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27419 (ix86_noce_conversion_profitable_p): New function.
27420 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27421 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27422 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27423 * doc/tm.texi: Regenerated.
27425 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27427 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27430 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27433 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27434 instructions for small multiply cores.
27436 2017-04-04 Jeff Law <law@redhat.com>
27438 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27440 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27441 that are not set by the loop over the elements.
27443 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27446 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27447 int mode, convert_modes it to mode as unsigned, otherwise use
27448 lowpart_subreg to mode rather than SImode.
27449 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27450 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27451 Use DImode instead of SImode for the shift count operand.
27452 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27455 2017-04-04 Richard Biener <rguenther@suse.de>
27457 PR middle-end/80281
27458 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27459 arithmetic done for the negate or the plus. Simplify.
27460 (A - (-B) -> A + B): Likewise.
27461 * fold-const.c (split_tree): Make sure to not negate pointers.
27463 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27465 PR rtl-optimization/60818
27466 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27467 a compare of comparisons with the thing compared if this results
27468 in a different machine mode.
27470 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27472 * alias.c (base_alias_check): Fix typo in comment.
27473 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27474 * cgraphunit.c (symbol_table::compile): Likewise.
27475 * collect2.c (maybe_run_lto_and_relink): Likewise.
27476 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27477 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27478 * config/avr/avr.c (avr_map_op_t): Likewise.
27479 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27480 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27481 * config/epiphany/epiphany.md (movcc): Likewise.
27482 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27483 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27485 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27486 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27487 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27488 * config/sh/sh.c (sh_rtx_costs): Likewise.
27489 * fold-const.c (fold_truth_andor): Likewise.
27490 * genautomata.c (collapse_flag): Likewise.
27491 * gengtype.h (struct type::u::s): Likewise.
27492 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27493 * input.c (FORMAT_AMOUNT): Likewise.
27494 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27495 (known_aggs_to_agg_replacement_list): Likewise.
27496 * ipa-inline-analysis.c: Likewise.
27497 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27498 * ipa-polymorphic-call.c
27499 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27500 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27501 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27503 * modulo-sched.c (apply_reg_moves): Likewise.
27504 * omp-expand.c (build_omp_regions_1): Likewise.
27505 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27506 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27507 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27508 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27509 * value-prof.c: Likewise.
27510 * var-tracking.c (val_reset): Likewise.
27512 2017-04-03 Richard Biener <rguenther@suse.de>
27514 PR tree-optimization/80275
27515 * fold-const.c (split_address_to_core_and_offset): Handle
27518 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27520 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27521 descriptors is at least equal to that of functions.
27523 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27525 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27527 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27530 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27531 (mov<IMOD4:mode>): New expander.
27532 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27534 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27536 PR rtl-optimization/79405
27537 * fwprop.c (propagations_left): New variable.
27538 (forward_propagate_into): Decrement it.
27539 (fwprop_init): Initialize it.
27540 (fw_prop): If the variable has reached zero, stop propagating.
27541 (fwprop_addr): Ditto.
27543 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27546 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27547 a FUNCTION_DECL, pass it as decl instead of origin to
27550 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27552 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27555 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27558 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27559 TARGET_VSX_SMALL_INTEGER.
27561 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27563 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27564 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27566 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27568 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27569 extraction from odd-numbered MSA register.
27571 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27573 PR middle-end/80173
27574 * expmed.c (store_bit_field_1): Don't attempt to create
27575 a word subreg out of hard registers wider than word if they
27576 have HARD_REGNO_NREGS of 1 for their mode.
27578 PR middle-end/80163
27579 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27580 conversions to integer types wider than word and pointer.
27583 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27584 (rtx_equal_for_cselib_p): Pass 0 to it.
27585 * cselib.c (cselib_hasher::equal): Likewise.
27586 (rtx_equal_for_cselib_1): Add depth argument. If depth
27587 is 128, don't look up VALUE locs and punt. Increment
27588 depth in recursive calls when walking VALUE locs.
27590 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27592 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27593 (make_gcov_file_name): Use the canonical path name for generating
27595 (read_line): Fix handling of files with ascii null bytes.
27597 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27599 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27600 to initialise a vector register instead
27601 of using a const_int.
27603 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27605 PR translation/80189
27606 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27607 diagnostic messages.
27609 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27612 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27613 (dfp_diex_<mode>): Update mode of operand 1.
27614 * doc/extend.texi (dxex, dxexq): Document change to return type.
27615 (diex, diexq): Document change to argument type.
27617 2017-03-30 Martin Jambor <mjambor@suse.cz>
27620 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27621 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27622 it reflects the signature changes performed at the callee side.
27623 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27624 to cgraph_build_function_type_skip_args.
27625 (build_function_decl_skip_args): Adjust call to the above function.
27627 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27630 * config/i386/sse.md
27631 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27632 register as dest whenever it is a MEM not rtx_equal_p to the
27633 corresponding dup operand, and when forcing into reg move the
27634 reg into the memory afterwards.
27635 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27636 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27637 for the force_reg mode.
27638 (avx512vl_vextractf128<mode>): Use register as dest either
27639 always when a MEM, or when it is a MEM not rtx_equal_p to the
27640 corresponding dup operand, or even not when it is a CONST_VECTOR
27641 depending on the mode and lo vs. hi.
27642 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27644 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27645 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27646 Likewise. Require that operands[2] is even.
27647 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27648 Remove extraneous parens. Require that operands[2] is a multiple
27650 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27651 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27652 disallow memory then.
27654 2017-03-30 Richard Biener <rguenther@suse.de>
27656 PR tree-optimization/77498
27657 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27658 to non-constants over backedges.
27660 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27662 PR rtl-optimization/80233
27663 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27664 as last_combined_insn. Do not test for BARRIER_P separately.
27666 2017-03-29 Andreas Schwab <schwab@suse.de>
27669 * calls.c (prepare_call_address): Convert funexp to Pmode before
27670 copying to temp reg.
27672 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27674 PR tree-optimization/80158
27675 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27676 Handle possible future case of more than one alternate
27678 (replace_rhs_if_not_dup): Likewise.
27679 (replace_one_candidate): Likewise.
27681 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27683 PR rtl-optimization/80193
27684 * ira.c (ira): Do not check allocation for LRA.
27686 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27688 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27689 (nvptx_output_simt_exit): Declare.
27690 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27691 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27692 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27693 (nvptx_declare_function_name): ...here. Emit declaration of local
27694 memory space buffer for omp_simt_enter insn.
27695 (nvptx_output_unisimt_switch): New.
27696 (nvptx_output_softstack_switch): New.
27697 (nvptx_output_simt_enter): New.
27698 (nvptx_output_simt_exit): New.
27699 * config/nvptx/nvptx.h (struct machine_function): New fields
27700 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27701 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27702 (UNSPECV_SIMT_EXIT): Ditto.
27703 (omp_simt_enter_insn): New insn.
27704 (omp_simt_enter): New expansion.
27705 (omp_simt_exit): New insn.
27706 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27708 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27709 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27710 (expand_GOMP_SIMT_EXIT): New.
27711 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27712 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27713 (GOMP_SIMT_EXIT): Ditto.
27714 * target-insns.def (omp_simt_enter): New insn.
27715 (omp_simt_exit): Ditto.
27716 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27718 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27719 (lower_rec_input_clauses): Likewise.
27720 (lower_lastprivate_clauses): Handle SIMT privatization.
27722 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27723 (ompdevlow_adjust_simt_enter): New.
27724 (find_simtpriv_var_op): New.
27725 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27726 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27728 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27729 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27730 (copy_decl_for_dup_finish): Ditto.
27732 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27734 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27737 * config/i386/i386.c (ix86_option_override_internal): Always
27738 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27740 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27742 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27744 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27746 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27747 mark new edge's irreducible flag accordign to it.
27748 (vect_do_peeling): Check loop preheader edge's irreducible flag
27749 and pass it to function slpeel_add_loop_guard.
27751 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27753 PR tree-optimization/80218
27754 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27755 Update block frequencies and counts.
27757 2017-03-28 Richard Biener <rguenther@suse.de>
27759 PR tree-optimization/78644
27760 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27761 of a simplification result we may not use it at all.
27763 2017-03-28 Richard Biener <rguenther@suse.de>
27766 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27767 without arguments, generate default definition of a SSA name.
27769 2017-03-28 Richard Biener <rguenther@suse.de>
27771 PR middle-end/80222
27772 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27773 TYPE_REF_CAN_ALIAS_ALL references.
27774 * fold-const.c (fold_indirect_ref_1): Likewise.
27776 2017-03-28 Martin Liska <mliska@suse.cz>
27779 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27780 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27782 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27783 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27785 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27786 (EXTRA_SPECS): Define.
27787 (SUBTARGET_EXTRA_SPECS): Likewise.
27788 (SUBTARGET_CPP_SPEC): Likewise.
27789 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27790 SUBTARGET_EXTRA_SPECS.
27791 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27793 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27795 * config/arc/simdext.md (vst64_insn): Update pattern.
27796 (vld32wh_insn): Likewise.
27797 (vld32wl_insn): Likewise.
27798 (vld64_insn): Likewise.
27799 (vld32_insn): Likewise.
27801 2017-03-28 Marek Polacek <polacek@redhat.com>
27804 * fold-const.c (fold_comparison): Use protected_set_expr_location
27805 instead of SET_EXPR_LOCATION.
27807 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27809 * tree.c (add_expr): Avoid name lookup warning.
27811 2017-03-27 Jeff Law <law@redhat.com>
27813 PR tree-optimization/80216
27814 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27815 function name. Limit recursion depth.
27816 (record_temporary_equivalences): Corresponding changes.
27818 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27820 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27823 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27826 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27828 * cfgcleanup.c (reg_note_cfa_p): New array.
27829 (insns_have_identical_cfa_notes): New function.
27830 (old_insns_match_p): Don't cross-jump in between /f
27831 and non-/f instructions. If both i1 and i2 are frame related,
27832 verify all CFA notes, their order and content.
27834 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27837 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27838 HImode and SImode with zero extend to DImode to one insn.
27839 (bswap<mode>2_extenddi): Likewise.
27840 (bswapsi2_extenddi): Likewise.
27841 (bswaphi2_extendsi): Likewise.
27842 (bswaphi2): Combine bswap HImode and SImode into one insn.
27843 Separate memory insns from swapping register.
27844 (bswapsi2): Likewise.
27845 (bswap<mode>2): Likewise.
27846 (bswaphi2_internal): Delete, no longer used.
27847 (bswapsi2_internal): Likewise.
27848 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27849 store, and gpr<-gpr swap insns.
27850 (bswap<mode>2_store): Likewise.
27851 (bswaphi2_reg): Register only splitter, combine with the splitter.
27852 (bswaphi2 splitter): Likewise.
27853 (bswapsi2_reg): Likewise.
27854 (bswapsi2 splitter): Likewise.
27855 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27856 the insns into load, store, and register/register insns.
27857 (bswapdi2_ldbrx): Likewise.
27858 (bswapdi2_load): Likewise.
27859 (bswapdi2_store): Likewise.
27860 (bswapdi2_reg): Likewise.
27862 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27864 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27865 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27867 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27870 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27872 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27873 special handling for target option conflicts between dform
27874 options (-mpower9-dform, -mpower9-dform-vector,
27875 -mpower9-dform-scalar) and -mno-direct-move.
27877 2017-03-27 Richard Biener <rguenther@suse.de>
27879 PR tree-optimization/80181
27880 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27882 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27884 * config/arc/predicates.md (move_double_src_operand): Replace the
27885 call to move_double_src_operand with a call to address_operand.
27887 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27889 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27890 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27891 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27893 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27895 * config/arc/predicates.md (long_immediate_loadstore_operand):
27896 Consider scaled addresses cases.
27898 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27900 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27901 restored when in interrupt.
27902 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27903 doesn't have delay slot.
27905 2017-03-27 Richard Biener <rguenther@suse.de>
27908 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27909 inlined thunk clones.
27911 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27914 * asan.c (instrument_derefs): Copy over last operand from
27915 original COMPONENT_REF to the new COMPONENT_REF with
27916 DECL_BIT_FIELD_REPRESENTATIVE.
27917 * ubsan.c (instrument_object_size): Likewise.
27919 2017-03-27 Richard Biener <rguenther@suse.de>
27921 PR tree-optimization/80170
27922 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27923 sure DR/SCEV didnt fold in constants we do not see when looking
27924 at the reference base alignment.
27926 2017-03-27 Richard Biener <rguenther@suse.de>
27928 PR middle-end/80171
27929 * gimple-fold.c (fold_ctor_reference): Properly guard against
27930 NULL return value from canonicalize_constructor_val.
27932 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27935 * config/i386/i386.c (ix86_expand_builtin)
27936 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27937 flags reg setting and flags reg using instructions.
27938 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27939 clobbering instructions to zero extend op2.
27941 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27943 * doc/install.texi (Configuration) <--with-aix-soname>:
27944 Update link to AIX ld.
27946 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27948 PR rtl-optimization/80160
27949 PR rtl-optimization/80159
27950 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27951 reg_alternate_class into account.
27953 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27956 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27957 to consider in curr_insn_transform.
27959 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27961 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27962 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27963 and emit_mode_inner.
27965 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27967 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27968 argument to the overloaded builtin variants. Use the new flag to
27969 deprecate certain builtin variants.
27970 * config/s390/s390-builtin-types.def: Add new builtin types.
27971 * config/s390/s390-builtins.h: Support new flags field for
27972 overloaded builtins.
27973 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27974 (s390_macro_to_expand): Enable vector float data type.
27975 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27976 builtins by incrementing the __VEC__ version number.
27977 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27979 (s390_resolve_overloaded_builtin): Emit error messages depending
27980 on the builtin flags.
27981 * config/s390/s390.c (s390_expand_builtin): Support additional
27982 flags argument. Change error message to match the messages
27983 emitted in s390-c.c.
27984 * config/s390/s390.md: New UNSPEC_* constants.
27985 (op_type): Add new instruction types.
27986 * config/s390/vecintrin.h: Add new builtins and test data class
27988 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27989 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27990 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27991 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27992 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27993 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27995 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27996 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27997 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27998 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28000 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28001 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28002 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28003 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28005 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28006 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28007 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28008 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28009 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28010 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28011 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28013 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28014 ("vec_scatter_element<V_HW_4:mode>_DI")
28015 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28016 ("vec_fpint<mode>", "vflls")
28017 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28018 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28019 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28020 ("*vec_cmphe<mode>_cc"): ... these.
28022 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28023 mode constant instead of magic value.
28025 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28027 * config/s390/s390.c (s390_expand_vec_compare): Support other
28028 vector floating point modes than just V2DF.
28029 (s390_expand_vcond): Likewise.
28030 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28031 (s390_cannot_change_mode_class): Prevent mode changes between TF
28032 and V1TF in vector registers.
28033 * config/s390/s390.md (DF, SF): New mode attributes.
28034 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28035 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28036 SFmode support for VRs.
28037 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28039 (VFT, VF_HW): New mode iterators.
28040 (vw, sdx): New mode attributes.
28041 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28042 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28043 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28044 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28045 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28046 also the new vector floating point modes. Renaming to ...
28048 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28049 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28050 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28051 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28052 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28053 ("vec_unordered<mode>"): ... these.
28055 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28056 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28057 ("*vec_extendv2df"): New insn definitions.
28059 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28061 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28062 ("mulditi3_2", "*muldi3_sign"): New patterns.
28063 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28064 rename the pattern definition.
28066 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28068 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28070 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28072 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28074 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28075 instruction if possible.
28076 * config/s390/vector.md (vec_halfnumelts): New mode
28078 ("*vec_vllezlf<mode>"): New pattern.
28080 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28082 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28083 ("popcountv4si2", "popcountv2di2"): Rename to ...
28084 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28085 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28087 ("popcount<mode>2_vxe"): New pattern.
28089 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28091 * common/config/s390/s390-common.c (processor_flags_table): Add
28093 * config.gcc: Add arch12.
28094 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28095 Default to arch12 for unknown CPU model numbers.
28096 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28097 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28098 PROCESSOR_max sanity check.
28099 * config/s390/s390-opts.h (enum processor_type): Add
28101 * config/s390/s390.c (processor_table): Add arch12.
28102 (s390_expand_builtin): Add check for B_VXE flag.
28103 (s390_issue_rate): Add PROCESSOR_ARCH12.
28104 (s390_get_sched_attrmask): Likewise.
28105 (s390_get_unit_mask): Likewise.
28106 (s390_sched_score): Enable z13 scheduling for arch12.
28107 (s390_sched_reorder): Likewise.
28108 (s390_sched_variable_issue): Likewise.
28109 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28111 (s390_tune_attr): Use z13 scheduling also for arch12.
28112 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28113 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28114 (TARGET_VXE_P): New macros.
28115 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28116 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28117 * config/s390/s390.opt: Add arch12 as processor_type.
28119 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28121 * config/s390/s390.md
28122 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28123 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28124 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28126 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28127 Rename expanders to ...
28129 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28130 ("fixuns_truncdddi2_emu"): ... these.
28132 ("fixuns_trunc<mode>si2_emu"): New expander.
28134 ("*fixuns_truncdfdi2_z13"): Rename to ...
28135 ("*fixuns_truncdfdi2_vx"): ... this.
28137 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28139 * config/s390/2964.md: Remove the single element vector compare
28140 instructions which are no longer used.
28141 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28143 (s390_canonicalize_comparison): Remove handling of DFmode
28145 (s390_expand_vec_compare_scalar): Remove function.
28146 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28147 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28149 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28151 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28153 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28155 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28156 will anyway by matched by mov<mode>_64dfp.
28158 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28160 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28161 vlef/vstef. Add missing operand to vleif.
28163 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28165 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28166 pair for all vector types with 64 bit elements.
28167 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28168 * config/s390/vector.md (V_HW_64): ... here.
28169 (V_128_NOSINGLE): New mode iterator.
28170 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28171 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28172 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28173 ("*vec_load_pairv2di"): Change to ...
28174 ("*vec_load_pair<mode>"): ... this one.
28176 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28178 * config/s390/constraints.md: Add comments.
28179 (jKK): Reject element sizes > 8 bytes.
28180 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28182 * config/s390/s390.md: Add the s_operand checks formerly in
28183 s390_split_ok_p to various splitters where they are still
28185 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28186 for 128 bit vectors. Plus two splitters.
28188 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28190 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28193 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28196 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28197 error if the boundary argument is not constant.
28199 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28201 PR rtl-optimization/80112
28202 * loop-doloop.c (doloop_condition_get): Don't check condition
28203 if cmp isn't SET with IF_THEN_ELSE src.
28205 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28207 PR tree-optimization/80158
28208 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28209 replacing a candidate statement, also replace it for the
28210 candidate's alternate interpretation.
28211 (replace_rhs_if_not_dup): Likewise.
28212 (replace_one_candidate): Likewise.
28214 2017-03-24 Richard Biener <rguenther@suse.de>
28216 PR tree-optimization/80167
28217 * graphite-isl-ast-to-gimple.c
28218 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28220 (translate_isl_ast_to_gimple::get_rename): Likewise.
28222 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28224 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28225 handling of certain combinations of target options, including the
28226 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28227 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28229 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28232 * config/arm/arm.md (*load_multiple): Add reload_completed to
28233 matching condition.
28235 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28236 Richard Biener <rguenth@suse.de>
28238 PR tree-optimization/79908
28239 PR tree-optimization/80136
28240 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28241 been cast away, gimplify_and_add suffices.
28243 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28245 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28247 2017-03-23 Richard Biener <rguenther@suse.de>
28249 PR tree-optimization/80032
28250 * gimplify.c (gimple_push_cleanup): Forced unconditional
28251 cleanups still have to go to the conditional_cleanups
28254 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28256 PR tree-optimization/80072
28257 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28259 (next_operand_entry_id): Change type to unsigned int.
28260 (sort_by_operand_rank): Make sure to return the right return value
28261 even if unsigned fields are bigger than INT_MAX.
28262 (struct oecount): Change cnt and id type to unsigned int.
28263 (oecount_hasher::equal): Formatting fix.
28264 (oecount_cmp): Make sure to return the right return value
28265 even if unsigned fields are bigger than INT_MAX.
28266 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28269 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28270 TREE_READONLY on result if writing it more than once.
28273 * doc/invoke.texi (-fsanitize=thread): Document that with
28274 -fnon-call-exceptions atomics are not able to throw
28278 * tsan.c: Include tree-eh.h.
28279 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28280 maybe_clean_or_replace_eh_stmt where needed.
28281 (instrument_memory_accesses): Add cfg_changed argument.
28282 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28283 if it returned true.
28284 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28286 PR rtl-optimization/63191
28287 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28288 wrapper function, moved the whole old content into ...
28289 (ix86_delegitimize_address_1): ... this. New inline function.
28290 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28291 true as last argument instead of ix86_delegitimize_address.
28293 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28295 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28296 cortexa57_branch_cost.
28298 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28300 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28302 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28305 * doc/md.texi (Constraints): Document wA constraint.
28306 * config/rs6000/constraints.md (wA): New.
28307 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28308 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28309 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28310 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28312 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28315 * gimplify.c (is_oacc_declared): New function.
28316 (oacc_default_clause): Use it to set default flags for acc declared
28317 variables inside parallel regions.
28318 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28319 declared variables.
28320 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28321 declare attribute to any decl as necessary.
28323 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28326 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28327 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28328 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28329 (arm_arch_lpae): This.
28330 * config/arm/arm.c (arm_arch7ve): Rename into ...
28331 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28332 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28335 2017-03-22 Martin Liska <mliska@suse.cz>
28338 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28339 error message instead of an ICE.
28341 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28343 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28345 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28347 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28349 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28352 2017-03-21 Martin Sebor <msebor@redhat.com>
28354 * doc/extend.texi: Use "cannot" instead of "can't."
28355 * doc/hostconfig.texi: Same.
28356 * doc/install.texi: Same.
28357 * doc/invoke.texi: Same.
28358 * doc/loop.texi: Same.
28359 * doc/md.texi: Same.
28360 * doc/objc.texi: Same.
28361 * doc/rtl.texi: Same.
28362 * doc/tm.texi: Same.
28363 * doc/tm.texi.in: Same.
28364 * doc/trouble.texi: Same.
28366 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28369 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28370 (collect_checksum_attributes): Set it.
28371 (die_checksum_ordered): Use it.
28373 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28375 PR tree-optimization/79908
28376 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28377 change: For a VA_ARG whose LHS has been cast away, use
28378 force_gimple_operand to construct the side effects.
28380 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28382 PR translation/80001
28383 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28384 more amenable to translation.
28385 (oacc_loop_auto_partitions): Likewise.
28387 2017-03-21 Marek Polacek <polacek@redhat.com>
28388 Martin Sebor <msebor@redhat.com>
28390 PR tree-optimization/80109
28391 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28392 on INTEGRAL_TYPE_P.
28394 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28395 Segher Boessenkool <segher@kernel.crashing.org>
28398 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28399 check reg_used_between_p between insn and one of succ or succ2
28400 depending on if succ is artificial insn not inserted into insn
28403 2017-03-21 Martin Liska <mliska@suse.cz>
28405 PR gcov-profile/80081
28406 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28407 * doc/gcc.texi: Include gcov-dump stuff.
28408 * doc/gcov-dump.texi: New file.
28410 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28412 PR rtl-optimization/79150
28413 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28414 conditional jump, if the jump is the last insn of the loop.
28416 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28417 Richard Biener <rguenth@suse.de>
28419 PR tree-optimization/79908
28420 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28421 been cast away, use force_gimple_operand to construct the side
28424 2017-03-21 Martin Liska <mliska@suse.cz>
28426 PR libfortran/79956
28427 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28430 2017-03-21 Brad Spengler <spender@grsecurity.net>
28433 * plugin.c (htab_hash_plugin): New function.
28434 (add_new_plugin): Use it and adjust.
28435 (parse_plugin_arg_opt): Adjust.
28436 (init_one_plugin): Likewise.
28438 2017-03-21 Richard Biener <rguenther@suse.de>
28440 PR tree-optimization/80032
28441 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28442 if set force the cleanup to happen unconditionally.
28443 (gimplify_target_expr): Push inserted clobbers with force_uncond
28444 to avoid them being removed by control-dependent DCE.
28446 2017-03-21 Richard Biener <rguenther@suse.de>
28448 PR tree-optimization/80122
28449 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28450 va_arg_pack_len when the inlined call stmt requires pack
28452 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28454 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28457 * tsan.c (instrument_builtin_call): If the memory model argument
28458 is not a constant, assume it is valid.
28461 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28464 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28466 PR rtl-optimization/79910
28467 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28468 if its dest is used by an insn before I2 (other than the combined
28469 insns themselves, which are properly handled already).
28471 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28474 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28476 * combine.c (record_used_regs): New static function.
28477 (try_combine): Handle situations where there is an additional
28478 instruction between I2 and I3 which needs to have a LOG_LINK
28482 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28484 * combine.c (try_combine): Delete redundant i1 test. Call
28485 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28487 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28490 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28491 alternatives 13/14.
28493 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28495 PR tree-optimization/80054
28496 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28497 the optimization if a PHI or any of its arguments is not dominated
28498 by the candidate's basis. Use gphi* rather than gimple* as
28500 (replace_profitable_candidates): Clean up a gimple* variable that
28501 should be a gphi* variable.
28503 2017-03-20 Martin Sebor <msebor@redhat.com>
28506 * doc/extend.texi (attribute constructor): Document present limitation.
28508 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28511 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28512 __POWER9_VECTOR__ #ifdef control, change template definition to
28513 use Power9-specific built-in function.
28514 (vec_any_eq): Likewise.
28515 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28516 to control outcomes from this test.
28517 (vector_ae_<mode>p): For VEC_F modes, likewise.
28519 2017-03-20 Ian Lance Taylor <iant@google.com>
28521 * config/i386/i386.c (ix86_function_regparm): Save an extra
28522 register for -fsplit-stack with DECL_STATIC_CHAIN.
28524 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28527 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28528 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28530 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28532 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28534 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28537 2017-03-20 Marek Polacek <polacek@redhat.com>
28540 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28542 2017-03-20 Richard Biener <rguenther@suse.de>
28544 PR tree-optimization/80113
28545 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28546 allocate extra SSA name for PHI def.
28547 (add_close_phis_to_outer_loops): Likewise.
28548 (add_close_phis_to_merge_points): Likewise.
28549 (copy_loop_close_phi_args): Likewise.
28550 (copy_cond_phi_nodes): Likewise.
28552 2017-03-20 Martin Liska <mliska@suse.cz>
28554 PR middle-end/79753
28555 * tree-chkp.c (chkp_build_returned_bound): Do not build
28556 returned bounds for a LHS that's not a BOUNDED_P type.
28558 2017-03-20 Martin Liska <mliska@suse.cz>
28562 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28563 COMPLEX_CST and VECTOR_CST.
28565 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28568 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28569 target operand. A new splitter adds the clobber statement in case
28570 the target operand is dead anyway.
28572 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28574 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28575 to age-old versions of binutils and glibc.
28577 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28579 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28581 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28583 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28585 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28587 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28588 requirement for binutils 2.13.
28590 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28592 * combine.c (try_combine): Delete redundant i1 test. Call
28593 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28595 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28597 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28598 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28600 <riscv64-*-elf>: Re-arrange section
28601 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28602 <riscv32-*-linux>: Likewise.
28603 <riscv64-*-elf>: Likewise
28604 <riscv64-*-linux>: Likewise.
28606 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28609 * aarch64.opt(verbose-cost-dump): Fix typo.
28611 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28614 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28615 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28617 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28619 * reload.c (find_reloads): When reloading a nonoffsettable address,
28620 use RELOAD_OTHER for it and its address reloads.
28622 PR rtl-optimization/79910
28623 * combine.c (record_used_regs): New static function.
28624 (try_combine): Handle situations where there is an additional
28625 instruction between I2 and I3 which needs to have a LOG_LINK
28628 2017-03-17 Jeff Law <law@redhat.com>
28630 PR tree-optimization/71437
28631 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28632 conditional in the hash table first.
28633 (vrp_dom_walker::before_dom_children): Extract condition from
28634 ASSERT_EXPR. Record condition, its inverion and any implied
28635 conditions as well.
28637 2017-03-17 Marek Polacek <polacek@redhat.com>
28638 Markus Trippelsdorf <markus@trippelsdorf.de>
28640 PR tree-optimization/80079
28641 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28644 2017-03-17 Richard Biener <rguenther@suse.de>
28646 PR middle-end/80075
28647 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28648 Properly verify the LHS before the RHS possibly claims to be
28650 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28653 2017-03-17 Martin Jambor <mjambor@suse.cz>
28655 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28656 (List of -O2 options): Likewise.
28657 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28660 2017-03-17 Tom de Vries <tom@codesourcery.com>
28662 * gcov-dump.c (print_usage): Print bug_report_url.
28664 2017-03-17 Richard Biener <rguenther@suse.de>
28666 PR middle-end/80050
28667 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28668 (parser::peek): Likewise.
28670 2017-03-17 Richard Biener <rguenther@suse.de>
28672 PR tree-optimization/80048
28673 * sese.c (free_sese_info): Properly release rename_map and
28674 copied_bb_map elements.
28676 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28678 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28679 Add linked-list forward and backlinks. Insert on
28680 construction, remove on destruction.
28681 (class pass_store_merging): Add m_stores_head field.
28682 (pass_store_merging::terminate_and_process_all_chains):
28683 Iterate over m_stores_head list.
28684 (pass_store_merging::terminate_all_aliasing_chains):
28686 (pass_store_merging::execute): Check for debug stmts first.
28687 Push new chains onto the m_stores_head stack.
28689 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28692 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28693 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28694 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28696 2017-03-16 Jeff Law <law@redhat.com>
28698 PR tree-optimization/71437
28699 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28700 member function. Implementation moved into after_dom_children
28701 member function and into the threader's thread_outgoing_edges
28703 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28704 some code into new thread_outgoing_edges.
28705 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28706 definition. Simplify marker handling (do it here). Assume we always
28707 have the available expression and the const/copies tables.
28708 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28710 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28711 * tree-vrp.c (equiv_stack): No longer file scoped.
28712 (vrp_dom_walker): New class.
28713 (vrp_dom_walker::before_dom_children): New member function.
28714 (vrp_dom_walker::after_dom_children): Likewise.
28715 (identify_jump_threads): Setup domwalker. Use it rather than
28716 walking edges in a random order by hand. Simplify setup/finalization.
28717 (finalize_jump_threads): Remove.
28718 (vrp_finalize): Do not call identify_jump_threads here.
28719 (execute_vrp): Do it here instead and call thread_through_all_blocks
28722 PR tree-optimization/71437
28723 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28725 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28727 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28728 (dom_opt_dom_walker::thread_across_edge): Remove
28729 handle_dominating_asserts argument. All callers changed.
28730 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28731 changes. Remove calls to lhs_of_dominating_assert. Other
28732 uses of handle_dominating_asserts turn into unconditional code
28733 (simplify_control_stmt_condition_1): Likewise.
28734 (simplify_control_stmt_condition): Likewise.
28735 (thread_through_normal_block, thread_across_edge): Likewise.
28736 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28737 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28738 object if it is not an SSA_NAME.
28739 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28740 before calling into the VRP specific simplifiers.
28741 (identify_jump_threads): Remove handle_dominating_asserts
28744 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28747 * tree-diagnostic.c (default_tree_printer): No longer static.
28748 * tree-diagnostic.h (default_tree_printer): New prototype.
28750 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28752 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28753 Change ins into fmov.
28755 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28757 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28758 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28759 Use h_con constraint for operand 1.
28760 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28761 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28763 2017-03-15 Jeff Law <law@redhat.com>
28765 PR tree-optimization/71437
28766 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28767 (record_temporary_equivalences): Use it.
28769 PR tree-optimization/71437
28770 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28771 tree-ssa-scopedtables.
28772 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28773 (record_conditions, record_cond, vuse_eq): Likewise.
28774 (record_edge_info): Adjust to API tweak of record_conditions.
28775 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28776 (record_temporary_equivalences, optimize_stmt): Likewise.
28777 (eliminate_redundant_computations): Likewise.
28778 (record_equivalences_from_stmt): Likewise.
28779 * tree-ssa-scopedtables.c: Include options.h and params.h.
28780 (vuse_eq): New function, moved from tree-ssa-dom.c
28781 (build_and_record_new_cond): Likewise.
28782 (record_conditions): Likewise. Accept vector of conditions rather
28783 than edge_equivalence structure for first argument.
28784 for the first argument.
28785 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28786 from tree-ssa-dom.c.
28787 (avail_exprs_stack::record_cond): Likewise.
28788 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28789 from tree-ssa-dom.c.
28790 (avail_exprs_stack): Add new member functions lookup_avail_expr
28792 (record_conditions): Declare.
28794 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28797 * lra-constraints.c (process_alt_operands): Increase reject for
28798 reloading an input/output operand.
28800 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28803 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28804 insns to convert from signed/unsigned char/short to IEEE 128-bit
28806 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28808 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28811 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28812 subreg of inner mode for values already in registers.
28814 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28816 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28817 iteration reg is used after the loop.
28819 2017-03-14 Martin Sebor <msebor@redhat.com>
28821 PR tree-optimization/79800
28822 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28823 precision in negative-positive range.
28824 (format_floating): Call non-const overload with adjusted precision.
28826 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28829 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28832 2017-03-14 Martin Sebor <msebor@redhat.com>
28834 PR middle-end/80020
28835 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28836 * builtins.def (aligned_alloc): Use it.
28839 * Makefile.in (GTFILES): Add calls.c.
28840 * calls.c: Include "gt-calls.h".
28842 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28844 PR rtl-optimization/79728
28845 * regs.h (struct target_regs): New field
28846 x_contains_allocatable_regs_of_mode.
28847 (contains_allocatable_regs_of_mode): New macro.
28848 * reginfo.c (init_reg_sets_1): Initialize it, and change
28849 contains_reg_of_mode so it includes global regs as well.
28850 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28851 rather than contains_regs_of_mode.
28853 2017-03-14 Martin Liska <mliska@suse.cz>
28855 * doc/invoke.texi: Document options that can't be combined with
28856 -fcheck-pointer-bounds.
28858 2017-03-14 Martin Liska <mliska@suse.cz>
28860 PR middle-end/79831
28861 * doc/invoke.texi (-Wchkp): Document the option.
28863 2017-03-14 Martin Liska <mliska@suse.cz>
28865 * Makefile.in: Install gcov-dump.
28867 2017-03-14 Martin Liska <mliska@suse.cz>
28869 * multiple_target.c (expand_target_clones): Bail out for
28870 an invalid attribute.
28872 2017-03-14 Richard Biener <rguenther@suse.de>
28874 * alias.c (struct alias_set_entry): Pack properly.
28875 * cfgloop.h (struct loop): Likewise.
28876 * cse.c (struct set): Likewise.
28877 * ipa-utils.c (struct searchc_env): Likewise.
28878 * loop-invariant.c (struct invariant): Likewise.
28879 * lra-remat.c (struct cand): Likewise.
28880 * recog.c (struct change_t): Likewise.
28881 * rtl.h (struct address_info): Likewise.
28882 * symbol-summary.h (function_summary): Likewise.
28883 * tree-loop-distribution.c (struct partition): Likewise.
28884 * tree-object-size.c (struct object_size_info): Likewise.
28885 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28886 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28887 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28888 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28889 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28890 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28891 (struct _stmt_vec_info): Likewise.
28893 2017-03-14 Martin Liska <mliska@suse.cz>
28896 * multiple_target.c (create_dispatcher_calls): Check that
28897 a target can create a function dispatcher.
28899 2017-03-14 Martin Liska <mliska@suse.cz>
28902 * multiple_target.c (expand_target_clones): Drop local.local
28903 flag for default implementation.
28905 2017-03-14 Richard Biener <rguenther@suse.de>
28907 PR tree-optimization/80030
28908 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28910 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28912 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28913 gcc_fallthrough() instead of __attribute__((fallthrough));
28915 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28917 * doc/gcc.texi: Remove "up" link to (DIR).
28918 * doc/gccint.texi: Ditto.
28920 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28922 * doc/install.texi (Specific) <avr>: Remove reference to
28925 2017-03-13 Jeff Law <law@redhat.com>
28927 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28928 attribute rather than comments.
28930 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28931 match_scratch operand is highest.
28933 2017-03-13 Martin Liska <mliska@suse.cz>
28935 PR middle-end/78339
28936 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28937 is a CHKP clone, use original declaration.
28939 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28941 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28942 (arc_conditional_register_usage): Use a different allocation order
28943 when optimizing for size.
28944 * common/config/arc/arc-common.c (arc_option_optimization_table):
28945 Section anchors default on when optimizing for size.
28947 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28949 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28951 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28953 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28954 * config/arc/arc.md (cpu_facility): Add cd variant.
28955 (*movqi_insn): Add code density variant.
28956 (*movhi_insn): Likewise.
28957 (*movqi_insn): Likewise.
28958 (*addsi3_mixed): Likewise.
28959 (subsi3_insn): Likewise.
28961 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28963 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28965 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28967 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28968 expressions with MINUS and UNARY ops.
28970 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28973 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28975 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28976 between vec_select and vector argument.
28977 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28978 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28979 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28980 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28981 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28982 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28984 2017-03-13 Richard Biener <rguenther@suse.de>
28987 * params.def (vect-max-peeling-for-alignment): Fix typo.
28989 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28991 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28992 issue that only occurred with binutils below 2.18.
28994 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28996 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28997 refer to binutils 2.11/2.12 minimum.
28999 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29001 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29002 ftp.kernel.org and simplify binutils requirement.
29004 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29006 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29008 (Optimize Options): Ditto. Also remove redundancy.
29010 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29012 PR translation/79848
29013 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29015 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29016 to G_ to avoid double translation.
29018 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29020 PR translation/79923
29021 * auto-profile.c (get_combined_location): Convert leading
29022 character of diagnostics to lower case and remove trailing period.
29023 (read_profile): Likewise for various diagnostics.
29024 * config/arm/arm.c (arm_option_override): Remove trailing period
29025 from various diagnostics.
29026 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29027 (msp430_expand_delay_cycles): Likewise.
29029 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29032 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29033 full command-line argument, rather than just "str".
29034 (aarch64_validate_march): Likewise.
29035 (aarch64_validate_mtune): Likewise.
29037 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29039 PR rtl-optimization/78911
29040 * lra-assigns.c (must_not_spill_p): New function.
29041 (spill_for): Use it.
29043 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29045 PR tree-optimization/79981
29046 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29047 ATOMIC_COMPARE_EXCHANGE ifn result.
29048 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29049 IFN_ATOMIC_COMPARE_EXCHANGE.
29051 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29054 * opts.c (parse_sanitizer_options): Add missing question mark to
29055 "did you mean" message.
29057 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29059 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29061 (VMULEUH_UNS): Likewise.
29062 (VMULOUB_UNS): Likewise.
29063 (VMULOUH_UNS): Likewise.
29064 * config/rs6000/rs6000.c (builtin_function_type): Remove
29065 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29067 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29070 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29071 x with result of extra_parsing_for_operand_code_0.
29072 (function_reader::extra_parsing_for_operand_code_0): Convert
29073 return type from void to rtx, returning x. When reading
29074 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29075 larger size containing struct block_symbol.
29077 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29079 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29080 -mfloat128-hardware without -m64.
29082 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29085 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29086 entries to the case statement that marks unsigned arguments to
29087 overloaded functions.
29089 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29091 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29092 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29094 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29097 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29098 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29100 2017-03-10 Martin Liska <mliska@suse.cz>
29104 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29105 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29108 2017-03-10 Olivier Hainque <hainque@adacore.com>
29110 * tree-switch-conversion (array_value_type): Start by resetting
29111 candidate type to it's main variant.
29113 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29115 PR rtl-optimization/79909
29116 * combine.c (try_combine): Use simplify_replace_rtx on individual
29117 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29118 of the whole CALL_INSN_FUNCTION_USAGE.
29120 PR tree-optimization/79972
29121 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29122 get_range_info on SSA_NAMEs. Formatting fixes.
29124 2017-03-10 Richard Biener <rguenther@suse.de>
29125 Jakub Jelinek <jakub@redhat.com>
29127 PR tree-optimization/77975
29128 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29129 edge to be constant.
29130 (get_val_for): For constant x return it. Formatting fix.
29131 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29132 would use the same bases as the current one.
29134 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29136 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29137 instead of vec_select for V1TImode.
29138 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29140 (VSX_LE_128): Add V1TI to this mode iterator.
29141 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29142 (*vsx_le_perm_store_<mode>): Likewise.
29143 (pre-reload splitter for VSX stores): Likewise.
29144 (post-reload splitter for VSX stores): Likewise.
29145 (*vsx_xxpermdi2_le_<mode>): Likewise.
29146 (*vsx_lxvd2x2_le_<mode>): Likewise.
29147 (*vsx_stxvd2x2_le_<mode>): Likewise.
29149 2017-03-09 Michael Eager <eager@eagercon.com>
29151 Correct failures with --enable-checking=yes,rtl.
29153 * config/microblaze/microblaze.c (microblaze_expand_shift):
29154 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29155 test for const0_rtx.
29156 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29157 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29159 2017-03-09 Richard Biener <rguenther@suse.de>
29161 PR tree-optimization/79977
29162 * graphite-scop-detection.c (scop_detection::merge_sese):
29163 Handle the case of extra exits to blocks dominating the entry.
29165 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29167 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29170 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29172 PR rtl-optimization/79949
29173 * lra-constraints.c (process_alt_operands): Check memory when
29174 trying to predict a cycle. Print about the overall increase.
29176 2017-03-09 Richard Biener <rguenther@suse.de>
29178 PR middle-end/79971
29179 * gimple-expr.c (useless_type_conversion_p): Preserve
29180 TYPE_SATURATING for fixed-point types.
29182 2017-03-09 Richard Biener <rguenther@suse.de>
29185 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29186 alignment of BLKmode params.
29188 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29191 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29192 (VALL_NO_V2Q): Likewise.
29194 * config/aarch64/aarch64-simd.md
29195 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29197 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29198 VALL_NO_V2Q mode iterator.
29199 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29201 2017-03-09 Martin Liska <mliska@suse.cz>
29203 PR tree-optimization/79631
29204 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29205 tree_int_cst_sign_bit just for INTEGER constants.
29207 2017-03-09 Martin Liska <mliska@suse.cz>
29211 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29214 2017-03-09 Marek Polacek <polacek@redhat.com>
29217 * tree.c (inchash::add_expr): Handle TREE_VEC.
29219 2017-03-09 Martin Liska <mliska@suse.cz>
29222 (chkp_narrow_size_and_offset): New function.
29223 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29224 (void chkp_parse_bit_field_ref): New function.
29225 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29226 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29228 2017-03-09 Martin Liska <mliska@suse.cz>
29231 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29232 (chkp_find_bounds_1): Remove gcc_unreachable.
29234 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29237 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29238 BUILT_IN_SYNC*, determine the access type from the size suffix and
29239 always build a MEM_REF with that type. Handle forgotten
29240 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29243 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29244 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29245 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29246 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29247 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29248 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29249 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29250 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29251 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29252 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29253 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29254 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29255 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29256 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29257 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29258 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29259 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29260 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29261 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29262 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29263 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29264 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29265 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29266 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29267 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29268 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29269 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29270 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29271 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29272 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29273 definitions outside of __OPTIMIZE__ guarded section.
29276 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29277 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29278 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29279 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29282 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29284 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29285 ("vfenez<mode>"): Add missing constraints.
29287 2017-03-08 Martin Sebor <msebor@redhat.com>
29290 * config/nds32/nds32.c (nds32_option_override):
29291 Fix misspelled diagnostic.
29293 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29296 * gimplify.c (gimplify_omp_for): Replace index var in outer
29297 taskloop statement with an artificial variable and add
29298 OMP_CLAUSE_PRIVATE clause for it.
29300 2017-03-08 Richard Biener <rguenther@suse.de>
29302 PR tree-optimization/79955
29303 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29304 for accesses that are completely outside of the variable.
29306 2017-03-08 Andrew Haley <aph@redhat.com>
29308 PR tree-optimization/79943
29309 * tree-ssa-loop-split.c (compute_new_first_bound): When
29310 calculating the new upper bound, (END-BEG) should be added, not
29313 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29315 * config/avr/avr.md (setmemhi): Make sure match_dup
29316 operand number comes before match_scratch.
29318 2017-03-08 Richard Biener <rguenther@suse.de>
29320 PR tree-optimization/79920
29321 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29322 with ncopies == 1 to ...
29323 (vect_transform_slp_perm_load): ... here. Properly compute
29324 all element loads by iterating VF times over the group. Do
29325 not handle ncopies (computed in a broken way) in
29326 vect_create_mask_and_perm.
29328 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29331 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29332 is a uniform vector, use uniform_vector_p return value instead of
29333 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29335 2017-03-07 Marek Polacek <polacek@redhat.com>
29337 PR middle-end/79809
29338 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29339 (alloca_call_type): Likewise.
29341 2017-03-07 Martin Liska <mliska@suse.cz>
29343 * gcov.c (process_args): Put comment to correct location.
29345 2017-03-07 Martin Liska <mliska@suse.cz>
29347 PR middle-end/68270
29348 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29349 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29350 (chkp_narrow_bounds_for_field): Likewise.
29351 (chkp_parse_array_and_component_ref): Pass one more argument to
29354 2017-03-07 Richard Biener <rguenther@suse.de>
29356 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29359 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29361 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29362 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29364 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29367 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29368 to end of description.
29369 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29371 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29373 PR rtl-optimization/79901
29374 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29376 (*avx512f_<code><mode>3<mask_name>): ... this.
29377 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29378 iterator instead of VI8_AVX2_AVX512BW.
29380 PR rtl-optimization/79901
29381 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29382 min/max expander, expand it using expand_vec_cond_expr.
29385 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29388 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29391 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29392 to void * for PCH reasons.
29393 * dwarf2out.c (output_loc_operands, output_die): Cast
29394 v.val_vec.array to unsigned char *.
29396 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29399 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29402 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29404 PR rtl-optimization/79571
29405 * lra-constraints.c (process_alt_operands): Calculate static
29406 reject and subtract it from overall when only addresses will be
29409 2017-03-06 Julia Koval <julia.koval@intel.com>
29412 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29413 incoming stack boundary to 128 for 64-bit targets.
29415 2017-03-06 Richard Biener <rguenther@suse.de>
29417 PR tree-optimization/79894
29418 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29419 to NULL after folding it.
29421 2017-03-06 Richard Biener <rguenther@suse.de>
29423 PR tree-optimization/79824
29424 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29425 check disabling peeling for gaps.
29427 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29429 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29430 attributes): Document gettimeofday.
29432 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29434 * config/s390/s390.c (s390_option_override_internal): Set
29435 PARAM_MIN_VECT_LOOP_BOUND
29437 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29439 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29440 * config/s390/s390.md: Likewise.
29442 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29445 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29446 (<avx2_avx512>_perm<mode>): Rename to ...
29447 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29449 (<avx512>_perm<mode>_mask): Rename to ...
29450 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29452 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29453 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29454 instead of VI8F_256_512.
29455 (avx512f_perm<mode>): New define_expand.
29456 (avx512f_perm<mode>_mask): Likewise.
29457 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29458 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29460 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29462 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29463 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29465 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29467 2017-03-06 Martin Liska <mliska@suse.cz>
29470 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29471 when having a SSA NAME w/o VAR_DECL assigned to it.
29473 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29475 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29476 msa_dpsub_<su>_d): Fix MODE for vec_select.
29478 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29480 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29482 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29484 2017-03-06 Richard Biener <rguenther@suse.de>
29486 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29487 * plugin.c (register_plugin_info): Likewise.
29488 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29490 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29492 * config/i386/sse.md (sse_storehps, sse_storelps,
29493 avx_<castmode><avxsizesuffix>_<castmode>,
29494 avx512f_<castmode><avxsizesuffix>_<castmode>,
29495 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29496 in condition that at least one operand is not a MEM.
29498 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29500 PR middle-end/79805
29501 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29502 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29504 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29505 gimple_call_nothrow_p flag based on whether original builtin can throw.
29506 If it can, emit following stmts on the fallthrough edge.
29507 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29508 don't create new bb if inserting just debug stmts on the edge, try to
29509 insert them on the fallthru bb or just reset debug stmts.
29511 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29514 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29515 restore recog_data (including the operand rtxes inside it) around
29516 the call to get_insn_template.
29518 2017-03-03 Martin Sebor <msebor@redhat.com>
29520 PR tree-optimization/79699
29521 * context.c (context::~context): Free MPFR caches to avoid
29522 a memory leak on program exit.
29524 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29526 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29527 Use wide_int::ulow () instead of .elt (0).
29529 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29531 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29532 (*pushxf): Limit oF constraint to 32bit targets and add oC
29533 constraint for 64bit targets.
29534 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29535 (*pushdf): Change rmF constraint to rmC.
29537 2017-03-03 Martin Liska <mliska@suse.cz>
29539 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29540 Remove unused variable.
29542 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29545 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29546 is a memory operand, increase num_memory.
29547 (ix86_expand_args_builtin): Likewise.
29549 2017-03-03 Jan Hubicka <jh@suse.cz>
29552 * ipa-devirt.c (maybe_record_node): Properly handle
29553 __cxa_pure_virtual visibility.
29555 2017-03-03 Martin Liska <mliska@suse.cz>
29557 PR tree-optimization/79803
29558 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29560 (pass_loop_prefetch::execute): Disabled optimization if an
29561 assumption about L1 cache size is not met.
29563 2017-03-03 Martin Liska <mliska@suse.cz>
29565 PR rtl-optimization/79574
29566 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29567 (hash_scan_set): Likewise.
29568 (dump_hash_table): Likewise.
29569 (hoist_code): Likewise.
29571 2017-03-03 Richard Biener <rguenther@suse.de>
29573 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29574 in place of uhigh ().
29575 (fixed_convert_from_real): Likewise.
29577 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29580 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29582 2017-03-03 Richard Biener <rguenther@suse.de>
29584 PR middle-end/79818
29585 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29586 TYPE_OVERFLOW_UNDEFINED check.
29588 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29590 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29592 (vector_ae_<mode>_p): Likewise.
29593 (vector_nez_<mode>_p): Likewise.
29594 (vector_ne_v2di_p): Likewise.
29595 (vector_ae_v2di_p): Likewise.
29596 (vector_ne_<mode>_p): Likewise.
29597 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29599 (vsx_tsqrt<mode>2_fe): Likewise.
29601 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29604 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29606 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29608 PR rtl-optimization/79780
29609 * cprop.c (one_cprop_pass): When second and further conditional trap
29610 in a single basic block is turned into an unconditional trap, turn it
29611 into a deleted note to avoid RTL verification failures.
29613 2017-03-02 Richard Biener <rguenther@suse.de>
29615 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29617 2017-03-02 Richard Biener <rguenther@suse.de>
29619 PR tree-optimization/79345
29621 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29622 param and abort the walk, returning -1 if it is hit.
29623 (walk_aliased_vdefs): Take a limit param and pass it on.
29624 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29625 defaulting to 0 and return a signed int.
29626 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29627 (check_defs): New helper.
29628 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29629 about uninitialized memory.
29630 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29631 bogus uninitialized warning.
29632 (fixed_convert_from_real): Likewise.
29634 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29636 PR tree-optimization/66768
29637 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29638 iv_use if base object can't be determined.
29640 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29642 PR tree-optimization/79345
29643 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29644 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29645 (get_pattern_stats): Initialize it.
29646 * genemit.c (gen_expand): Verify match_scratch numbers come after
29647 match_operand/match_dup numbers.
29648 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29649 match_scratch numbers.
29650 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29652 * config/s390/s390.md (trunctdsd2): Likewise.
29654 2017-03-02 Richard Biener <rguenther@suse.de>
29656 * wide-int.h (wide_int_storage::operator=): Implement in terms
29659 2017-03-02 Richard Biener <rguenther@suse.de>
29661 PR tree-optimization/79777
29662 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29663 the to insert expression to sth existing.
29665 2017-03-01 Martin Sebor <msebor@redhat.com>
29667 PR middle-end/79692
29668 * gimple-ssa-sprintf.c
29669 (directive::known_width_and_precision): New function.
29670 (format_integer): Use it.
29671 (get_mpfr_format_length): Consider the full range of precision
29672 when computing %g output with the # flag. Set the likely byte
29673 count to 3 rather than 1 when precision is indeterminate.
29674 (format_floating): Correct the lower bound of precision.
29676 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29678 * doc/invoke.texi: Document default code model for 64-bit Linux.
29680 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29683 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29684 udiv rather than div since input pattern is unsigned.
29686 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29688 * config/i386/i386.c (print_reg): Warn for values of
29689 unsupported size in integer register.
29691 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29694 * config/rs6000/predicates.md (current_file_function_operand): Do
29695 not allow self calls to be local if the function is replaceable.
29697 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29700 * config/rs6000/altivec.h (vec_ctz and others): Change the
29701 preprocessor macro that controls conditional compilation from
29702 _ARCH_PWR9 to __POWER9_VECTOR__.
29703 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29704 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29705 control (instead of _ARCH_PWR9 control) so that template
29706 definition uses power9-specific function.
29707 (vec_any_eq): Likewise.
29708 (vec_all_ne): Change macro definition to use a power9-specific
29709 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29710 _ARCH_PWR9 control).
29711 (vec_any_eq) Likewise.
29712 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29713 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29714 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29715 support for xvcmpnedp instruction.
29716 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29717 macro expansion so that Power9 implementation of vec_all_ne does
29718 not use the AltiVec predicate framework.
29719 (VCMPNEH_P): Likewise.
29720 (VCMPNEW_P): Likewise.
29721 (VCMPNED_P): Likewise.
29722 (VCMPNEFP_P): Likewise.
29723 (VCMPNEDP_P): Likewise.
29724 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29725 implementation of vec_any_eq to not use AltiVec predicate
29727 (VCMPAEH_P): Likewise.
29728 (VCMPAEW_P): Likewise.
29729 (VCMPAED_P): Likewise.
29730 (VCMPAEFP_P): Likewise.
29731 (VCMPAEDP_P): Likewise.
29732 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29733 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29734 not use the AltiVec predicate framework.
29735 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29736 of vec_any_eq to not use AltiVec predicate framework.
29737 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29738 support for predefined __POWER9_VECTOR__ macro to indicate that
29739 Power9 instruction selection is enabled.
29740 (altivec_overloaded_builtins): Remove extraneous
29741 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29742 function argument types RS6000_BTI_bool_V16QI and
29743 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29744 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29745 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29746 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29747 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29748 Power9 for implementations of vec_cmpne. Change the signature for
29749 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29750 (representing vec_all_ne) to remove the previously described first
29751 argument of type RS6000_BTI_INTSI, as this was an artifact of
29752 reliance on the AltiVec predicate framework, which is no longer
29753 used in the implementation of these functions. Add
29754 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29755 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29756 since, unlike the AltiVec predicate framework implementation, we
29757 do not share function descriptors between vec_alle and vec_anyeq.
29758 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29759 set of modes that receive special treatment even when
29760 TARGET_P9_VECTOR is true. The special treatment emits code that
29761 does not depend on Power9 instructions.
29762 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29763 define_expand to not rely on AltiVec predicate framework.
29764 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29766 (vector_ne_v2di_p): Change this define_expand to not rely on
29767 AltiVec predicate framework.
29768 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29770 (vector_ne_<mode>_p): Change this define_expand to not rely on
29771 AltiVec predicate framework.
29772 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29774 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29775 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29776 define_insn pattern.
29777 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29778 define_insn pattern because the xvcmpne<VSs>. instruction is not
29780 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29781 instruction is not supported.
29783 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29785 * config/nvptx/nvptx.c: Include intl.h.
29787 2017-03-01 Martin Jambor <mjambor@suse.cz>
29790 * ipa-prop.h (ipa_bits): Removed field known.
29791 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29792 to pointers. Adjusted their comments to warn about their sharing.
29793 (ipcp_transformation_summary): Change bits to a vector of pointers.
29794 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29795 (ipa_get_ipa_bits_for_value): Declare.
29796 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29797 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29798 (ipa_bits_hash_table): Likewise.
29799 (ipa_vr_ggc_hash_traits): Likewise.
29800 (ipa_vr_hash_table): Likewise.
29801 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29802 being pointers and vr_known being removed.
29803 (ipa_set_jf_unknown): Likewise.
29804 (ipa_get_ipa_bits_for_value): New function.
29805 (ipa_set_jfunc_bits): Likewise.
29806 (ipa_get_value_range): New overloaded functions.
29807 (ipa_set_jfunc_vr): Likewise.
29808 (ipa_compute_jump_functions_for_edge): Use the above functions to
29809 construct bits and vr parts of jump functions.
29810 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29811 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29813 (ipcp_grow_transformations_if_necessary): Also allocate
29814 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29816 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29817 them. Fix too long lines.
29818 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29819 vr_known being removed.
29820 (ipa_read_jump_function): Use new setter functions to construct bits
29821 and vr parts of jump functions or set them to NULL.
29822 (write_ipcp_transformation_info): Adjust for bits being pointers.
29823 (read_ipcp_transformation_info): Likewise.
29824 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29826 Include gt-ipa-prop.h.
29827 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29829 (ipcp_store_bits_results): Likewise.
29830 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29831 Do not write to existing jump functions but use a temporary instead.
29833 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29836 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29837 attempt to use its first operand as BIT_FIELD_REF base.
29839 2017-03-01 Richard Biener <rguenther@suse.de>
29841 PR middle-end/79721
29842 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29843 interpolating formula in wrapping arithmetic.
29844 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29846 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29848 PR tree-optimization/79734
29849 * tree-vect-generic.c (expand_vector_condition): Optimize
29850 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29851 Handle VEC_COND_EXPR where comparison has different inner width from
29852 type's inner width.
29854 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29856 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29857 markup, and similar issues. Remove @opindex entries for things
29858 that aren't options. Add missing -mmpy-option entries.
29860 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29862 PR tree-optimization/79737
29863 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29864 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29865 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29866 instead of byte_size. Formatting fix.
29867 (shift_bytes_in_array_right): Formatting fix.
29869 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29872 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29873 condition on optimize for the leaf function test.
29875 2017-02-28 Martin Liska <mliska@suse.cz>
29878 * read-rtl-function.c (function_reader::handle_unknown_directive):
29879 Bail out when one uses -flto.
29881 2017-02-28 Martin Liska <mliska@suse.cz>
29883 * common.opt: Replace space with tabular for options of <number>
29885 * config/i386/i386.opt: Show <number> value for
29886 -mlarge-data-threshold.
29887 * opts.c (print_filtered_help): Do not display number in hexadecimal
29890 2017-02-28 Martin Liska <mliska@suse.cz>
29892 * common.opt: Fix --help=option -Q for options which are of
29895 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29897 * config/i386/i386.c (print_reg): Error out for values
29898 of 8-bit size in invalid integer register.
29900 2017-02-28 Martin Sebor <msebor@redhat.com>
29902 PR tree-optimization/79691
29903 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29905 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29908 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29909 gcc_unreachable with output_operand_lossage.
29911 2017-02-28 Richard Biener <rguenther@suse.de>
29913 PR tree-optimization/79740
29914 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29916 (visit_nary_op): Insert the nary into the hashtable if we
29917 pattern-matched sth.
29918 * tree-ssa-pre.c (eliminate_insert): Robustify.
29920 2017-02-28 Richard Biener <rguenther@suse.de>
29922 PR middle-end/79731
29923 * fold-const.c (decode_field_reference): Reject out-of-bound
29926 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29928 * config/i386/i386.c: Include intl.h.
29929 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29930 instead of just cond ? "..." : "...".
29931 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29932 * coverage.c (read_counts_file): Likewise.
29933 * omp-offload.c: Include intl.h.
29934 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29935 of just cond ? "..." : "...".
29936 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29937 of just cond ? "..." : "...".
29939 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29942 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29944 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29945 'tune for' CPU name.
29946 * config/arm/arm-cpu-data.h: Regenerated.
29948 2017-02-28 Richard Biener <rguenther@suse.de>
29950 PR tree-optimization/79732
29951 * tree-inline.c (expand_call_inline): Do not shadow var.
29953 2017-02-28 Richard Biener <rguenther@suse.de>
29955 PR tree-optimization/79723
29956 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29957 address-space properly.
29959 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29961 * doc/optinfo.texi (Optimization groups): Fix option used for
29963 * doc/invoke.texi (-fopt-info): Document "omp".
29964 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29965 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29966 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29967 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29968 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29970 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29972 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29973 OPTGROUP_OMP with "omp".
29975 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29978 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29979 for arithmetic shift of unsigned V2DI.
29981 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29983 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29984 arc/linux.h headers.
29985 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29986 (LINK_SPEC): Likewise.
29987 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29988 (EXTRA_SPECS): Likewise.
29989 (STARTFILE_SPEC): Likewise.
29990 (ENDFILE_SPEC): Likewise.
29991 (LIB_SPEC): Likewise.
29992 (TARGET_SDATA_DEFAULT): Likewise.
29993 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29994 (MULTILIB_DEFAULTS): Likewise.
29995 (DWARF2_UNWIND_INFO): Likewise.
29996 * config/arc/big.h: New file.
29997 * config/arc/elf.h: Likewise.
29998 * config/arc/linux.h: Likewise.
29999 * config/arc/t-uClibc: Remove.
30001 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30003 PR tree-optimization/77536
30004 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30005 (tree_transform_and_unroll_loop): Use above function to compute the
30006 estimated niter of unrolled loop and use it when scaling profile.
30007 Also use count info rather than frequency if it's non-zero.
30008 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30009 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30010 (vect_transform_loop): Call above function.
30012 2017-02-27 Richard Biener <rguenther@suse.de>
30014 PR tree-optimization/45397
30015 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30016 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30017 (visit_nary_op): Add pattern matching for CSEing sign-changed
30018 or truncated operations with wider ones.
30020 2017-02-27 Richard Biener <rguenther@suse.de>
30022 PR tree-optimization/79690
30023 * tree-vect-stmts.c (vectorizable_store): Use vector type
30024 built from the DR with address-space.
30026 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30028 * doc/invoke.texi (Optimize Options): Refine the description
30029 of asan-use-after-return.
30031 2017-02-25 Alan Modra <amodra@gmail.com>
30033 PR rtl-optimization/79584
30034 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30035 base, not ad->base_term, the reg within base. Remove assertion
30036 that ad->base == ad->base_term. Replace gen_int_mode using
30037 bogus mode with const0_rtx.
30039 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30041 PR middle-end/79396
30042 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30043 FMA_EXPR like tcc_binary or tcc_unary.
30045 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30048 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30050 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30051 (output_loc_operands): Handle DW_OP_call_ref and
30052 DW_OP_GNU_variable_value.
30053 (struct variable_value_struct): New type.
30054 (struct variable_value_hasher): Likewise.
30055 (variable_value_hash): New variable.
30056 (string_types): Remove.
30057 (copy_loc_descr): New function.
30058 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30059 (prepend_loc_descr_to_each): New function.
30060 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30061 instead of add_loc_descr_to_each if the first argument is single
30062 location list and the second has multiple.
30063 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30064 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30065 when looking for variable value which doesn't have other location info.
30066 (loc_list_from_tree): Formatting fix.
30067 (gen_array_type_die): Simplify DW_AT_string_length handling.
30068 (adjust_string_types): Remove.
30069 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30070 string_types. Call resolve_variable_values.
30071 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30072 (resolve_addr_in_expr): Likewise. Add A argument.
30073 (copy_deref_exprloc): Remove deref argument. Adjust for the
30074 original expression being DW_OP_GNU_variable_value with optionally
30075 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30076 optionally after it.
30077 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30078 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30079 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30080 (variable_value_hasher::hash, variable_value_hasher::equal): New
30082 (resolve_variable_value_in_expr, resolve_variable_value,
30083 resolve_variable_values, note_variable_value_in_expr,
30084 note_variable_value): New functions.
30085 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30088 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30091 * opts.h (handle_generated_option): Add GENERATED_P argument.
30092 * opts-common.c (handle_option): Adjust function comment.
30093 (handle_generated_option): Add GENERATED_P argument, pass it to
30095 (control_warning_option): Pass false to handle_generated_option
30097 * opts.c (maybe_default_option): Pass true to handle_generated_option
30099 * optc-gen.awk: Likewise.
30101 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30103 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30104 a REG, look at the REG it is a SUBREG of.
30105 (splitter for cmpeqsi_t): Ditto.
30107 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30109 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30110 the special USEs with the pattern of the insn, not the insn itself.
30112 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30115 * doc/invoke.texi: Document -mload-store-pairs.
30117 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30118 Sandra Loosemore <sandra@codesourcery.com>
30120 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30121 argument isn't a CONST_INT.
30122 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30123 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30124 (nios2_validate_compare): Bypass alternate compare logic if *op2
30125 is not a CONST_INT.
30126 (ldstwm_operation_p): Return false if first_base is not a REG or
30127 if first_offset is not a CONST_INT.
30129 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30131 * config/cris/cris.md: Use correct operand in a define_peephole2.
30133 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30135 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30137 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30139 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30140 this_insn if it is an INSN or JUMP_INSN.
30141 (force_offsettable): Look at base, not at addr.
30142 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30143 on things that aren't necessarily CONST_INTs.
30145 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30147 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30148 -mfpmath=sse is the default also for x86-32 targets with SSE2
30149 instruction set when @option{-ffast-math} is enabled
30151 2017-02-24 Jeff Law <law@redhat.com>
30153 PR rtl-optimizatoin/79286
30154 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30157 2017-02-24 Richard Biener <rguenther@suse.de>
30159 PR tree-optimization/79389
30160 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30163 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30165 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30166 function comment to reflect reality.
30167 (loop_exits_before_overflow): Fix typo in function description.
30169 2017-02-24 Richard Biener <rguenther@suse.de>
30171 PR tree-optimization/79389
30172 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30173 properly that a threading opportunity exists. Detect conditional
30174 copy/constant propagation opportunities.
30176 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30178 * config/visium/visium.md (type): Add trap.
30179 (b): New mode attribute.
30180 (*btst): Rename into...
30181 (*btst<mode>): ...this and adjust.
30182 (*cbranchsi4_btst_insn): Rename into...
30183 (*cbranch<mode>4_btst_insn): ...this and adjust.
30184 (trap): New define_insn.
30186 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30188 PR tree-optimization/79389
30189 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30190 (noce_reversed_cond_code): New function.
30191 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30192 reversed_comparison_code. Formatting fix.
30193 (noce_try_store_flag): Test rev_cond != NULL in addition to
30194 reversed_comparison_code.
30195 (noce_try_store_flag_constants): Likewise.
30196 (noce_try_store_flag_mask): Likewise.
30197 (noce_try_addcc): Use rev_cond if non-NULL instead of
30198 reversed_comparison_code.
30199 (noce_try_cmove_arith): Likewise. Formatting fixes.
30200 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30201 (noce_find_if_block): Initialize rev_cond.
30202 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30203 instead of false as last argument never attempt to reverse it
30206 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30208 PR tree-optimization/79663
30209 * tree-predcom.c (combine_chains): Process refs in reverse order
30210 only for ZERO length chains, and add explaining comment.
30212 2017-02-23 Jeff Law <law@redhat.com>
30214 PR tree-optimization/79578
30215 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30216 in call to operand_equal_p.
30218 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30221 * config/i386/cpuid.h: Fix another undefined behavior.
30223 2017-02-23 Richard Biener <rguenther@suse.de>
30225 PR tree-optimization/79683
30226 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30227 vector types for data-refs.
30229 2017-02-23 Martin Liska <mliska@suse.cz>
30231 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30233 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30235 PR middle-end/79665
30236 * internal-fn.c (get_range_pos_neg): Moved to ...
30237 * tree.c (get_range_pos_neg): ... here. No longer static.
30238 * tree.h (get_range_pos_neg): New prototype.
30239 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30240 are known to be in between 0 and signed maximum inclusive, try to
30241 expand both unsigned and signed divmod and use the cheaper one from
30244 2017-02-22 Jeff Law <law@redhat.com>
30246 PR tree-optimization/79578
30247 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30248 to compare base operands.
30250 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30253 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30254 gpc_reg_operand instead of fpr_reg_operand.
30256 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30258 * config/mips/mips.c (mips_return_in_memory): Force FP
30259 vector types to be returned in memory for o32 ABI.
30261 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30263 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30264 instead of DW_TAG_member for static data member declarations and don't
30265 set no_linkage_name for static inline data members.
30266 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30269 2017-02-22 Martin Liska <mliska@suse.cz>
30271 * doc/invoke.texi: Replace inequality signs with square brackets
30274 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30277 * lra-constraints.c (simplify_operand_subreg): Handle
30278 WORD_REGISTER_OPERATIONS targets.
30280 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30283 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30284 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30285 elimination by swapping fld*.
30287 2017-02-22 Richard Biener <rguenther@suse.de>
30289 PR tree-optimization/79673
30290 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30291 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30292 irrelevant address-space qualifiers and avoiding a
30293 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30295 2017-02-22 Richard Biener <rguenther@suse.de>
30297 PR tree-optimization/79666
30298 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30299 to not symbolically negate if that may introduce undefined
30302 2017-02-22 Martin Liska <mliska@suse.cz>
30305 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30306 * data-streamer-out.c (streamer_write_gcov_count_stream):
30308 * value-prof.c (stream_out_histogram_value): Make assert more
30309 precise based on type of counter.
30311 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30314 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30315 Use nonimmediate_operand instead of memory_operand for operand 1.
30316 (float-extend standard_x87sse_constant_load splitter): Ditto.
30318 2017-02-21 Jeff Law <law@redhat.com>
30320 PR tree-optimization/79621
30321 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30322 blocks with edges to themselves.
30324 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30327 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30328 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30329 Use gimple_call_builtin_p.
30332 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30333 on temporarily removed DEBUG_INSNs.
30335 PR tree-optimization/79649
30336 * tree-loop-distribution.c (classify_partition): Give up on
30337 non-generic address space loads/stores.
30339 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30341 * doc/loop.texi (Loop manipulation): Remove nonexistent
30342 tree_ssa_loop_version from the documentation.
30343 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30345 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30348 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30349 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30350 * config/rs6000/rs6000.c: Include except.h.
30351 (rs6000_expand_split_stack_prologue): Call
30352 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30354 2017-02-21 Martin Jambor <mjambor@suse.cz>
30357 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30358 have been analyzed.
30360 2017-02-21 Martin Jambor <mjambor@suse.cz>
30362 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30363 for backward compatibility only.
30364 * doc/invoke.texi (Option Summary): Remove all references to
30365 -fipa-cp-alignment.
30367 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30371 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30373 * lra-constraints.c (curr_insn_transform): Handle
30374 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30376 2017-02-21 Martin Liska <mliska@suse.cz>
30378 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30380 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30382 PR translation/79638
30383 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30385 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30388 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30389 (arm_function_ok_for_sibcall): Return false for an indirect call by
30390 descriptor if all the argument registers are used.
30391 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30392 alignment of the function.
30394 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30396 PR tree-optimization/61441
30397 * simplify-rtx.c (simplify_const_unary_operation): For
30398 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30399 the sNaN unmodified.
30401 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30403 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30404 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30405 instead of SYSTEM_HEADER_DIR.
30407 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30408 Martin Liška <mliska@suse.cz>
30410 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30411 Fix typos and grammar, use active voice, and clarify.
30413 2017-02-20 Marek Polacek <polacek@redhat.com>
30415 PR middle-end/79537
30416 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30419 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30421 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30424 * config/i386/i386.c (ix86_expand_builtin): Handle
30425 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30426 ix86_builtins_isa[fcode].isa as a requirement of those
30427 flags and any other flag in the bitmask.
30428 (ix86_init_mmx_sse_builtins): Use 0 instead of
30429 ~OPTION_MASK_ISA_64BIT as mask.
30430 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30431 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30432 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30433 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30435 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30438 * lra-constraints.c (split_reg): Check requested split mode
30439 is supported by the register.
30441 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30443 * lra-constraints.c (simplify_operand_subreg): Remove early
30446 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30449 * lra-constraints.c (curr_insn_transform): Tighten condition
30450 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30452 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30455 * lra-constraints.c (curr_insn_transform): Handle
30456 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30458 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30461 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30463 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30465 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30468 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30471 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30473 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30476 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30478 * final.c (last_columnnum, override_columnnum): New variables.
30479 (final_start_function): Set last_columnnum, pass it to begin_prologue
30480 hook and pass 0 to dwarf2out_begin_prologue.
30481 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30482 to source_line debug hook.
30483 (notice_source_line): Compute last_columnnum and for debug_column_info
30484 return true on column changes.
30485 * debug.h (struct gcc_debug_hooks): Add column argument to
30486 source_line and begin_prologue hooks.
30487 (debug_nothing_int_charstar_int_bool): Remove prototype.
30488 (debug_nothing_int_int_charstar,
30489 debug_nothing_int_int_charstar_int_bool): New prototypes.
30490 (dwarf2out_begin_prologue): Add column argument.
30491 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30492 begin_prologue hooks.
30493 (debug_nothing_int_charstar_int_bool): Remove.
30494 (debug_nothing_int_int_charstar,
30495 debug_nothing_int_int_charstar_int_bool): New functions.
30496 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30497 through to dwarf2out_source_line.
30498 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30499 (dwarf2out_source_line): Add column argument, emit it if requested.
30500 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30502 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30503 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30504 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30505 through to dwarf2out_begin_prologue.
30506 (vmsdbgout_source_line): Add column argument, pass it through to
30507 dwarf2out_source_line.
30508 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30509 dbxout_source_line caller.
30510 (dbxout_source_line): Add column argument.
30512 * common.opt (gno-column-info, gcolumn-info): New options.
30513 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30514 (check_die): Also test for multiple DW_AT_decl_column attributes.
30515 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30516 DW_AT_decl_column if requested.
30517 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30519 (gen_variable_die): Likewise.
30520 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30521 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30524 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30525 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30526 (ix86_handle_option): Handle OPT_m3dnowa.
30527 * doc/invoke.texi (-m3dnowa): Document.
30528 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30529 -m3dnowa instead of -m3dnow -march=athlon.
30532 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30533 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30535 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30538 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30539 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30540 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30541 generator for vsx_xxpermdi_<mode>_be.
30542 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30543 force big-endian semantics.
30544 (vsx_xxpermdi_<mode>_be): New define_expand with same
30545 implementation as previous version of vsx_xxpermdi_<mode>.
30547 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30549 PR tree-optimization/79327
30550 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30551 variable, its initialization and use.
30553 2017-02-17 Julia Koval <julia.koval@intel.com>
30555 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30556 (OPTION_MASK_ISA_PKU_UNSET): New.
30557 (ix86_handle_option): Handle -mrdpid.
30558 * config/i386/cpuid.h (bit_RDPID): New.
30559 * config/i386/driver-i386.c (host_detect_local_cpu):
30560 Detect RDPID feature.
30561 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30562 * config/i386/i386-c.c (ix86_target_macros_internal):
30564 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30565 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30566 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30567 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30568 * config/i386/i386.md (define_insn "rdpid"): New.
30569 * config/i386/i386.opt Add -mrdpid.
30570 * config/i386/immintrin.h (_rdpid_u32): New.
30572 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30574 PR rtl-optimization/79541
30575 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30576 instead of transforming it into USE.
30578 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30580 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30581 If HONOR_SNANS (SFmode) force the input to a register.
30582 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30583 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30584 an frsp or similar insn.
30586 2017-02-17 Martin Liska <mliska@suse.cz>
30588 PR rtl-optimization/79577
30589 * params.def (selsched-max-sched-times): Increase minimum to 1.
30591 2017-02-17 Martin Liska <mliska@suse.cz>
30593 PR rtl-optimization/79574
30594 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30596 2017-02-17 Martin Liska <mliska@suse.cz>
30598 PR tree-optimization/79529
30599 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30600 ssa_defined_default_def_p to handle cases which are implicitly
30602 * tree-ssa.c (ssa_defined_default_def_p): New function.
30603 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30604 which are implicitly defined.
30605 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30607 2017-02-17 Richard Biener <rguenther@suse.de>
30609 PR middle-end/79576
30610 * params.def (max-ssa-name-query-depth): Limit to 10.
30612 2017-02-17 Richard Biener <rguenther@suse.de>
30614 PR tree-optimization/79552
30615 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30618 2017-02-17 Richard Biener <rguenther@suse.de>
30621 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30623 2017-02-17 Marek Polacek <polacek@redhat.com>
30625 PR middle-end/79536
30626 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30627 (fold_negate_expr): New wrapper.
30629 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30631 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30632 Correct terminology and de-emphasize pre-standard behavior.
30634 2017-02-16 Alan Modra <amodra@gmail.com>
30636 PR rtl-optimization/79286
30637 * ira.c (def_dominates_uses): New function.
30638 (update_equiv_regs): Don't create an equivalence for insns that
30639 may trap where the register def does not dominate the use.
30641 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30643 PR rtl-optimization/78127
30644 * lra.c (lra): Call lra_eliminate before finish the loop after
30647 2017-02-16 Richard Biener <rguenther@suse.de>
30649 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30651 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30652 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30653 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30654 (isl_val_int_from_wi): New function.
30655 (extract_affine_gmp): Rename to ...
30656 (extract_affine_wi): ... this, take a widest_int.
30657 (extract_affine_int): Just wrap extract_affine_wi.
30658 (add_param_constraints): Use isl_val_int_from_wi.
30659 (add_loop_constraints): Likewise, and extract_affine_wi.
30661 2017-02-15 Jeff Law <law@redhat.com>
30663 PR middle-end/79521
30664 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30665 ira_init_register_move_cost_if_necessary.
30667 2017-02-15 Martin Sebor <msebor@redhat.com>
30669 PR middle-end/32003
30670 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30671 removed in a prior commit.
30673 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30675 PR tree-optimization/79347
30676 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30677 counters during peeling.
30679 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30681 * Makefile.in (site.exp): Remove "set ISLVER".
30683 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30686 * real.c (real_from_integer): Call real_convert even for decimal.
30688 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30691 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30693 2017-02-14 Andrew Pinski <apinski@cavium.com>
30695 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30696 cores and change the partno/implementer to be correct.
30697 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30698 the 'B" as the implementer.
30699 * config/aarch64/aarch64-tune.md: Regenerate.
30701 2017-02-14 Carl Love <cel@us.ibm.com>
30703 * config/rs6000/rs6000.c: Add case statement entry to make the
30704 xvcvuxdsp built-in argument unsigned.
30705 * config/rs6000/vsx.md: Fix the source and return operand types so they
30706 match the instruction definitions from the ISA document. Fix typo
30707 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30710 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30713 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30714 member early_clobber_alts.
30715 * lra-lives.c (reg_early_clobber_p): New.
30716 (process_bb_lives): Use it.
30717 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30718 (debug_operand_data): Initialize early_clobber_alts.
30719 (setup_operand_alternative): Set up early_clobber_alts.
30720 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30721 alternatives to new_insn_reg.
30722 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30724 (lra_update_insn_regno_info): Pass the new arg.
30726 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30728 PR middle-end/79505
30729 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30730 (new_oacc_loop_raw): Don't clear already cleared fields.
30733 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30734 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30735 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30737 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30740 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30742 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30745 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30746 the extra instruction to the right place to store 128-bit constant
30749 2017-02-14 Martin Sebor <msebor@redhat.com>
30751 PR middle-end/79448
30752 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30753 warning for strings of unknown length.
30755 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30757 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30759 2017-02-14 Jeff Law <law@redhat.com>
30762 * ira-costs.c (scan_one_insn): Initialize register move costs
30763 for pseudos seen in USE/CLOBBER insns.
30765 PR tree-optimization/79095
30766 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30767 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30768 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30769 if the operands are known to be not equal, then the resulting range
30771 (intersect_ranges): If the new range is ~[0,0] and the old range is
30772 wide, then prefer ~[0,0].
30773 * tree-vrp.c (overflow_comparison_p_1): New function.
30774 (overflow_comparison_p): New function.
30775 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30776 if NAME is used in an overflow test.
30777 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30778 overflow check that can be expressed as an equality test, then adjust
30779 ops to be that equality test.
30781 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30783 * config/s390/s390-builtin-types.def: Remove flags argument.
30784 * config/s390/s390.c (s390_init_builtins): Likewise.
30786 2017-02-14 Martin Liska <mliska@suse.cz>
30788 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30789 vector. Fix trailing white spaces.
30791 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30793 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30796 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30798 PR rtl-optimization/68664
30799 * config/arm/arm.c (arm_sched_can_speculate_insn):
30800 New function. Declare prototype.
30801 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30803 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30805 PR rtl-optimization/68664
30806 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30808 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30810 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30812 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30813 max skip bytes for function, loop and jump.
30815 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30817 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30818 ABS_EXPR for gimple dump.
30820 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30823 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30825 PR tree-optimization/79408
30826 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30827 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30828 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30829 also if rhs1 is INTEGER_CST.
30831 2017-02-14 Richard Biener <rguenther@suse.de>
30833 PR middle-end/79432
30834 * tree-into-ssa.c (insert_phi_nodes): When the function can
30835 have abnormal edges rewrite SSA names with broken use-def
30836 dominance out of SSA and register them for PHI insertion.
30838 2017-02-13 Martin Sebor <msebor@redhat.com>
30840 PR middle-end/79496
30841 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30842 clearing info.nowrite flag when snprintf size argument is a range.
30844 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30846 * cprop.c (cprop_jump): Add missing space in string literal.
30847 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30848 (get_constraint_for_component_ref): Likewise.
30849 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30850 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30851 * lra-constraints.c (process_alt_operands): Likewise.
30852 * ipa-inline.c (inline_small_functions): Likewise.
30853 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30854 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30855 * trans-mem.c (diagnose_tm_1_op): Likewise.
30856 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30857 (grid_parallel_clauses_gridifiable): Likewise.
30859 * config/nvptx/mkoffload.c (process): Add space in between
30862 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30863 "MOD4_SSE_REGS" and "ALL_REGS".
30865 * spellcheck.c (test_data): Add , in between "foo" and "food".
30867 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30870 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30871 boundary crossing check and subsequent code generation agree.
30873 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30875 * config/aarch64/aarch64.c (has_memory_op): Delete.
30876 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30879 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30881 PR rtl-optimization/79388
30882 PR rtl-optimization/79450
30883 * combine.c (distribute_notes): When removing TEM_INSN for which
30884 corresponding dest has last value recorded, invalidate that last
30887 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30889 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30890 of explicit '@'. Add missing assembly comment marker on branch costs
30893 2017-02-13 Nathan Sidwell <nathan@acm.org>
30895 * gengtype-lex.l (<in_struct>): Add '/'.
30897 2017-02-13 Martin Liska <mliska@suse.cz>
30900 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30902 2017-02-13 Richard Biener <rguenther@suse.de>
30904 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30906 * configure: Re-generate.
30907 * config.in: Likewise.
30908 * graphite-dependences.c: Simplify as if
30909 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30910 * graphite-isl-ast-to-gimple.c: Likewise.
30911 * graphite-optimize-isl.c: Likewise.
30912 * graphite-poly.c: Likewise.
30913 * graphite-sese-to-poly.c: Likewise.
30914 * graphite.h: Likewise.
30915 * toplev.c: Include isl/version.h and use isl_version () for
30916 printing the ISL version.
30917 * doc/install.texi: Update ISL requirement.
30919 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30921 * doc/standards.texi (Standards): Update reference to
30924 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30926 * doc/extend.texi (Named Address Spaces): sourceware.org now
30928 * doc/install.texi (Binaries): Ditto.
30931 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30933 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30934 standard terminology "stringize"/"stringizing" throughout.
30935 * doc/cppinternals.texi: Likewise.
30937 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30939 * doc/extend.texi: Fix some spelling mistakes and typos.
30940 * doc/invoke.texi: Likewise.
30942 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30945 * params.def (inline-min-speedup) Change from 10 to 8.
30947 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30949 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30952 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30955 * ipa-inline-analysis.c (get_minimal_bb): New function.
30956 (record_modified): Use it.
30957 (remap_edge_change_prob): Handle also ancestor functions.
30959 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30961 * doc/contrib.texi (Contributors): Remove broken link into
30962 the Mauve CVS repository.
30964 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30966 PR middle-end/79454
30967 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30968 result computation whenever lhs doesn't have vector mode, not
30969 just when it has BLKmode.
30971 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30973 * doc/makefile.texi (profiledbootstrap): Refer to the
30974 installation instructions only in textual form.
30976 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30979 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30981 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30983 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30984 (Specific): Update mingw-w64 reference.
30986 (Specific): Remove broken link to Renesas RX processor.
30988 2017-02-10 Richard Biener <rguenther@suse.de>
30990 * toplev.c (process_options): Do not mention obsolete graphite
30991 options when printing sorry message about missing graphite support.
30992 Mention -floop-nest-optimize.
30994 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30996 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30997 (vtst_p16): Likewise.
30998 (vtstq_p8): Likewise.
30999 (vtstq_p16): Likewise.
31001 (vtstq_p64): Likewise.
31002 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31003 (vset_lane_p64): New.
31004 (vsetq_lane_p64): New.
31006 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31008 PR tree-optimization/79411
31009 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31010 stmt operands are SSA_NAMEs used in abnormal phis.
31011 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31014 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31017 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31020 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31022 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31024 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31026 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31030 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31031 not arbitrary TREE_CONSTANT.
31034 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31035 "omp declare target link" attribute unless is_global_var.
31036 * omp-offload.c (find_link_var_op): Likewise.
31038 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31039 Chung-Lin Tang <cltang@codesourcery.com>
31041 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31043 (gimplify_adjust_omp_clauses): Don't delete TILE.
31044 (gimplify_omp_for): Deal with TILE.
31045 * internal-fn.c (expand_GOACC_TILE): New function.
31046 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31048 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31049 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31051 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31052 avoid DIV for outermost collapse var.
31053 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31054 Remove out of date comments, fix whitespace.
31055 * omp-general.c (omp_extract_for_data): Deal with tiling.
31056 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31057 adjust OLF_DIM_BASE value.
31058 (struct omp_for_data): Add tiling field.
31059 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31060 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31061 for auto loops. Remove default auto determining, moved to
31062 oacc_loop_fixed_partitions.
31063 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31064 stmts, add e_mask field.
31065 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31066 (oacc_thread_numbers): Use oacc_dim_call.
31067 (oacc_xform_tile): New.
31068 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31069 (finish_oacc_loop): Adjust for ifns vector.
31070 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31071 add case for GOACC_TILE fns.
31072 (oacc_loop_xform_loop): Delete.
31073 (oacc_loop_process): Iterate over call list directly, and add
31074 handling for GOACC_TILE fns.
31075 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31077 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31078 vector partitioning to outer loops. Assign 2 partitions to loops
31079 when available. Add TILE handling.
31080 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31081 (execite_oacc_device_lower): Process GOACC_TILE fns,
31082 ignore unknown specs.
31083 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31084 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31085 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31087 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31089 * configure.ac (ACX_BUGURL): Update.
31090 * configure: Regenerate.
31092 2017-02-09 Richard Biener <rguenther@suse.de>
31094 PR tree-optimization/69823
31095 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31096 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31098 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31100 * config/arc/arc-c.def: Add __NPS400__ definition.
31101 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31102 (TARGET_NPS400): Define.
31104 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31106 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31108 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31109 pointer, arch_info.
31110 (arc_cpu_types): Fill the arch_info field with a pointer into the
31111 arc_arch_types table.
31112 (arc_selected_cpu): Declare.
31113 * config/arc/arc.c (arc_selected_cpu): Make global.
31114 (arc_selected_arch): Delete.
31115 (arc_base_cpu): Delete.
31116 (arc_override_options): Remove references to deleted variables,
31117 update access to arch information.
31118 (ARC_OPT): Update access to arch information.
31119 (ARC_OPTX): Likewise.
31120 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31121 (TARGET_ARC600): Update access to arch information.
31122 (TARGET_ARC601): Likewise.
31123 (TARGET_ARC700): Likewise.
31124 (TARGET_EM): Likewise.
31125 (TARGET_HS): Likewise.
31126 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31129 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31132 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31133 condition/operands for integer GE/LE/GEU/LEU operations.
31135 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31137 PR translation/79397
31138 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31141 2017-02-08 Martin Jambor <mjambor@suse.cz>
31144 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31145 whether allocation happened.
31146 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31147 nothing was allocated.
31149 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31151 PR tree-optimization/79408
31152 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31153 constant, but SSA_NAME with a known integer range, use the minimum
31154 of that range instead of op1 to determine if modulo can be replaced
31155 with its first operand.
31157 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31159 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31161 2017-02-08 Richard Biener <rguenther@suse.de>
31163 PR tree-optimization/71824
31164 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31165 Check all loops contained in the merged region.
31167 2017-02-07 Andrew Pinski <apinski@cavium.com>
31169 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31171 2017-02-07 Andrew Pinski <apinski@cavium.com>
31173 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31174 (thunderxt88): Likewise.
31175 (thunderxt81): Disable LSE and change v8.1 to v8.
31176 (thunderxt83): Likewise.
31178 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31179 Richard Biener <rguenther@suse.de>
31181 PR middle-end/79399
31182 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31183 type from int to size_t.
31184 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31186 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31188 PR rtl-optimization/79386
31189 * cprop.c (bypass_conditional_jumps): Initialize
31190 bypass_last_basic_block already before splitting bbs after
31191 unconditional traps...
31192 (bypass_conditional_jumps): ... rather than here.
31195 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31196 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31197 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31198 fix -masm=intel patterns.
31200 2017-02-07 Richard Biener <rguenther@suse.de>
31202 PR tree-optimization/79256
31203 PR middle-end/79278
31204 * builtins.c (get_object_alignment_2): Use min_align_of_type
31205 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31206 and ADJUST_FIELD_ALIGN.
31208 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31210 * doc/tm.texi: Regenerate.
31211 * stor-layout.c (layout_decl): Adjust.
31212 (update_alignment_for_field): Likewise.
31213 (place_field): Likewise.
31214 (min_align_of_type): Likewise.
31215 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31216 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31217 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31218 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31219 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31220 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31221 * config/i386/i386.c (x86_field_alignment): Likewise.
31222 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31223 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31224 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31225 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31226 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31227 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31231 2017-01-30 Richard Biener <rguenther@suse.de>
31233 PR tree-optimization/79256
31234 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31235 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31238 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31240 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31241 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31242 builtins to SImode and emit a zero-extend, if necessary.
31244 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31246 * docs/invoke.texi (RISC-V Options): Alphabetize.
31248 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31250 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31253 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31255 * config/riscv/riscv.c: New file.
31256 * common/config/riscv/riscv-common.c: Likewise.
31257 * config.gcc: Likewise.
31258 * config/riscv/constraints.md: Likewise.
31259 * config/riscv/elf.h: Likewise.
31260 * config/riscv/generic.md: Likewise.
31261 * config/riscv/linux.h: Likewise.
31262 * config/riscv/multilib-generator: Likewise.
31263 * config/riscv/peephole.md: Likewise.
31264 * config/riscv/pic.md: Likewise.
31265 * config/riscv/predicates.md: Likewise.
31266 * config/riscv/riscv-builtins.c: Likewise.
31267 * config/riscv/riscv-c.c: Likewise.
31268 * config/riscv/riscv-ftypes.def: Likewise.
31269 * config/riscv/riscv-modes.def: Likewise.
31270 * config/riscv/riscv-opts.h: Likewise.
31271 * config/riscv/riscv-protos.h: Likewise.
31272 * config/riscv/riscv.h: Likewise.
31273 * config/riscv/riscv.md: Likewise.
31274 * config/riscv/riscv.opt: Likewise.
31275 * config/riscv/sync.md: Likewise.
31276 * config/riscv/t-elf-multilib: Likewise.
31277 * config/riscv/t-linux: Likewise.
31278 * config/riscv/t-linux-multilib: Likewise.
31279 * config/riscv/t-riscv: Likewise.
31280 * configure.ac: Likewise.
31281 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31282 Waterman as RISC-V maintainers.
31283 * doc/install.texi: Add RISC-V entries.
31284 * doc/invoke.texi: Add RISC-V options section.
31285 * doc/md.texi: Add RISC-V constraints section.
31286 * configure: Regenerated.
31288 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31291 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31292 false values to be constant vectors with all 0 or all 1 bits set.
31293 (vcondu<mode><mode>): Likewise.
31294 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31296 (fpmask_comparison_operator): Update comment.
31297 (vecint_comparison_operator): New predicate.
31298 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31299 vector conditionals when the true and false values are constant
31300 vectors with all 0 bits or all 1 bits set.
31302 2017-02-06 Martin Sebor <msebor@redhat.com>
31304 PR tree-optimization/79376
31305 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31307 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31309 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31310 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31311 to simplify split condition.
31313 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31315 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31316 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31319 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31321 PR rtl-optimization/68664
31322 * target.def (can_speculate_insn): New hook.
31323 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31324 * doc/tm.texi: Regenerate.
31325 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31326 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31327 (rs6000_sched_can_speculate_insn): New function.
31329 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31331 PR tree-optimization/79284
31332 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31333 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31334 vectorizable_mask_load_store, vectorizable_operation,
31335 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31336 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31337 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31338 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31339 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31340 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31341 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31342 is_gimple_assign (stmt). Replace another such test with
31343 is_gimple_assign (stmt).
31345 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31348 * config/avr/avr.c (rtl-iter.h): Include it.
31349 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31350 (avr_legitimate_combined_insn): ...and implementation.
31352 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31354 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31355 * config/s390/s390.c (s390_const_operand_ok)
31356 (s390_canonicalize_comparison, s390_extract_part)
31357 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31358 (s390_contiguous_bitmask_p, s390_rtx_costs)
31359 (legitimize_pic_address): Likewise.
31360 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31361 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31362 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31363 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31364 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31366 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31368 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31369 REGNO($0) == REGNO($1).
31371 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31373 * config/s390/linux.h(SIZE_TYPE): Add comment.
31375 2017-02-06 Julian Brown <julian@codesourcery.com>
31376 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31377 Virendra Pathak <virendra.pathak@broadcom.com>
31379 * config/aarch64/aarch64-cores.def: Change the scheduler
31381 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31382 * config/aarch64/thunderx2t99.md: New file.
31384 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31386 * doc/standards.texi (Go Language): Update link to language
31389 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31391 * tree-eh.c (lower_resx): Sanitize profile.
31392 (cleanup_empty_eh_move_lp): Likewise.
31394 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31397 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31399 * cfgloopmanip.h (loop_version): Update prototype.
31400 * modulo-sched.c (sms_schedule): Update call of loop_version.
31401 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31402 * tree-parloops.c (gen_parallel_loop): Likewise.
31403 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31404 * tree-ssa-loop-split.c (split_loop): Likewise.
31405 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31406 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31408 2017-02-05 Martin Liska <mliska@suse.cz>
31411 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31413 (print_operand_address): Initialize a struct to zero.
31415 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31417 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31418 garbage collector only in textual form.
31420 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31422 * doc/extend.texi (x86 specific memory model extensions for
31423 transactional memory): Simplify a phrase.
31425 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31428 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31429 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31430 (atomic_storedi_1): Likewise.
31432 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31434 PR tree-optimization/79338
31435 * tree-parloops.c (gather_scalar_reductions): Don't call
31436 vect_analyze_loop_form for loop->inner before destroying loop's
31439 2017-02-03 Martin Sebor <msebor@redhat.com>
31441 PR tree-optimization/79327
31442 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31443 when precision has resulted in leading zeros.
31444 (format_integer): Adjust the likely counter to assume an unknown
31445 argument that may be zero is non-zero.
31447 2017-02-03 Jason Merrill <jason@redhat.com>
31450 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31451 avoid copying non-taken branch.
31453 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31455 PR tree-optimization/79340
31456 * tree-vect-loop.c (vectorizable_reduction): Release
31457 vec_defs elements after safe_splicing them into other vectors.
31460 PR tree-optimization/79327
31461 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31462 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31464 (format_integer): Use wide_int_to_tree instead of build_int_cst
31465 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31466 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31467 of shortest and longest sequence.
31469 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31471 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31472 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31474 2017-02-03 Walter Lee <walt@tilera.com>
31477 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31478 after initial stackframe link reg save.
31479 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31481 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31484 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31485 wu for stxssp alternative.
31487 2017-02-03 Martin Sebor <msebor@redhat.com>
31489 PR tree-optimization/79352
31490 * gimple-fold.c (get_range_strlen): Add argument.
31491 (get_range_strlen): Change return type to bool.
31492 (get_maxval_strlen): Pass in a dummy argument.
31493 * gimple-fold.h (get_range_strlen): Change return type to bool.
31494 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31495 * tree.h (array_at_struct_end_p): Add argument.
31496 * tree.c (array_at_struct_end_p): Handle it.
31498 2017-02-03 Martin Liska <mliska@suse.cz>
31501 * multiple_target.c (create_dispatcher_calls): Redirect edge
31502 from a caller of a dispatcher.
31503 (expand_target_clones): Make the clones local.
31504 (ipa_target_clone): Do both target clones and resolvers.
31505 (ipa_dispatcher_calls): Remove the pass.
31506 (pass_dispatcher_calls::gate): Likewise.
31507 (make_pass_dispatcher_calls): Likewise.
31508 * passes.def (pass_target_clone): Put as very first IPA early
31511 2017-02-03 Martin Liska <mliska@suse.cz>
31513 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31514 in case of a function with ifunc attribute.
31516 2017-02-03 Martin Liska <mliska@suse.cz>
31518 * cgraph.c (cgraph_node::dump): Dump function version info.
31519 * symtab.c (symtab_node::dump_base): Add missing new line.
31521 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31523 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31524 (ifcombine_ifandif): Use it.
31526 2017-02-03 Martin Liska <mliska@suse.cz>
31528 * doc/invoke.texi: Document default value for
31529 use-after-scope-direct-emission-threshold.
31531 2017-02-03 Martin Liska <mliska@suse.cz>
31533 PR tree-optimization/79339
31534 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31535 (format_floating): Likewise.
31537 2017-02-03 Martin Liska <mliska@suse.cz>
31540 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31542 (ipa_node_params_t::remove): Likewise.
31543 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31544 initialization from removed ipa_node_params_t::insert.
31545 (ipa_node_params::~ipa_node_params): Move from removed
31546 ipa_node_params_t::release.
31547 * symbol-summary.h (symbol_summary::m_released): New member.
31548 Do not release a summary twice. Do not allow to call finalizer
31549 for types of a summary that live in GGC memory.
31551 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31553 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31556 2017-02-02 Martin Sebor <msebor@redhat.com>
31558 PR middle-end/79275
31559 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31560 (format_string): Tighten up the range of output for non-constant
31561 strings and correct the expected range for wide non-constant strings.
31563 2017-02-02 Martin Sebor <msebor@redhat.com>
31565 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31567 PR middle-end/32003
31568 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31570 (-fdump-tree-@var): Add to index and document how to come up
31571 with pass-specific option and dump file names.
31572 (-fdump-passes): Clarify where to look for output.
31574 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31576 PR middle-end/77445
31577 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31578 statistics of the analyzed path; allow threading for speed when
31579 any of BBs along the path are optimized for speed.
31581 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31583 PR middle-end/78468
31584 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31585 settings of the virtual registers.
31588 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31590 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31591 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31594 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31596 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31597 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31599 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31601 * config/s390/s390.md: Add missing comments with the expanded
31603 * config/s390/vector.md: Likewise.
31604 * config/s390/vx-builtins.md: Likewise.
31606 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31609 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31610 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31611 conditions on a single line.
31613 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31615 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31616 __S390_VX__ to __VX__.
31618 2017-02-01 Andrew Pinski <apinski@cavium.com>
31620 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31621 stmt_info to record_stmt_cost.
31622 (vect_get_known_peeling_cost): Pass stmt_info if known to
31624 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31625 cpu_vector_cost field into
31626 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31627 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31628 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31629 splitting of scalar_stmt_cost and vec_stmt_cost.
31630 (thunderx_vector_cost): Likewise.
31631 (cortexa57_vector_cost): LIkewise.
31632 (exynosm1_vector_cost): Likewise.
31633 (xgene1_vector_cost): Likewise.
31634 (thunderx2t99_vector_cost): Improve after the splitting of the two
31636 (aarch64_builtin_vectorization_cost): Update for the splitting of
31637 scalar_stmt_cost and vec_stmt_cost.
31639 2017-02-01 Torvald Riegel <triegel@redhat.com>
31640 Richard Henderson <rth@redhat.com>
31642 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31643 conditional on existance of a fast atomic load.
31644 * optabs-query.c (can_atomic_load_p): New function.
31645 * optabs-query.h (can_atomic_load_p): Declare it.
31646 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31647 no fast atomic load is available for the particular size of access.
31648 (expand_atomic_compare_and_swap): Likewise.
31649 (expand_atomic_load): Likewise.
31650 (expand_atomic_store): Likewise.
31651 (expand_atomic_fetch_op): Likewise.
31652 * testsuite/lib/target-supports.exp
31653 (check_effective_target_sync_int_128): Remove x86 because it provides
31654 no fast atomic load.
31655 (check_effective_target_sync_int_128_runtime): Likewise.
31657 2017-02-01 Richard Biener <rguenther@suse.de>
31659 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31660 (graphite_transform_loops): Provide opt-info for optimized nests.
31661 * tree-parloop.c (parallelize_loops): Provide opt-info for
31662 parallelized loops.
31664 2017-02-01 Richard Biener <rguenther@suse.de>
31666 PR middle-end/79315
31667 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31668 was not set before.
31670 2017-02-01 Richard Biener <rguenther@suse.de>
31672 PR tree-optimization/71824
31673 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31674 Verify the loops are valid in the merged SESE region.
31675 (scop_detection::can_represent_loop_1): Check analyzing the
31676 evolution of the number of iterations in the region succeeds.
31678 2017-01-31 Ian Lance Taylor <iant@golang.org>
31680 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31681 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31683 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31685 PR preprocessor/79210
31686 * input.c (get_substring_ranges_for_loc): Replace line_width
31687 assertion with error-handling.
31689 2017-01-31 Richard Biener <rguenther@suse.de>
31691 PR tree-optimization/77318
31692 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31693 (create_pw_aff_from_tree): Take loop parameter.
31694 (add_condition_to_pbb): Pass loop of the condition to
31695 create_pw_aff_from_tree.
31697 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31699 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31700 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31702 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31706 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31708 (convert_int_to_float128): Likewise.
31709 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31710 (convert_int_to_float128): Likewise.
31711 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31712 (UNSPEC_IEEE128_CONVERT): Likewise.
31713 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31714 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31715 Use local variables for IBM extended format.
31716 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31717 (fix_trunc<mode>si2_fprs): Likewise.
31718 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31719 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31720 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31721 to know that we can now have integers of all sizes in vector
31723 (fix<uns>_<mode>di2_hw): Likewise.
31724 (float<uns>_<mode>si2_hw): Likewise.
31725 (fix_<mode>si2_hw): Likewise.
31726 (fixuns_<mode>si2_hw): Likewise.
31727 (float<uns>_<mode>di2_hw): Likewise.
31728 (float_<mode>di2_hw): Likewise.
31729 (float_<mode>si2_hw): Likewise.
31730 (floatuns_<mode>di2_hw): Likewise.
31731 (floatuns_<mode>si2_hw): Likewise.
31732 (xscvqp<su>wz_<mode>): Delete, no longer used.
31733 (xscvqp<su>dz_<mode>): Likewise.
31734 (xscv<su>dqp_<mode>): Likewise.
31735 (ieee128_mfvsrd_64bit): Likewise.
31736 (ieee128_mfvsrd_32bit): Likewise.
31737 (ieee128_mfvsrwz): Likewise.
31738 (ieee128_mtvsrw): Likewise.
31739 (ieee128_mtvsrd_64bit): Likewise.
31740 (ieee128_mtvsrd_32bit): Likewise.
31742 2017-01-31 Martin Liska <mliska@suse.cz>
31745 * ipa-prop.c (ipa_free_all_node_params): Call release method
31746 instead of ~sumbol_summary to not to trigger double times
31749 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31751 PR tree-optimization/71691
31752 * bitmap.h (class auto_bitmap): New.
31753 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31754 is_maybe_undefined instead of ssa_undefined_value_p.
31756 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31758 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31759 __S390_ARCH_LEVEL__ to __ARCH__.
31761 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31763 PR tree-optimization/79267
31764 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31765 if should_remove_lhs_p is true.
31767 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31770 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31771 (add_alignment_attribute): New.
31772 (base_type_die): Add alignment attribute.
31773 (subrange_type_die): Likewise.
31774 (modified_type_die): Likewise.
31775 (gen_array_type_die): Likewise.
31776 (gen_descr_array_type_die: Likewise.
31777 (gen_enumeration_type_die): Likewise.
31778 (gen_subprogram_die): Likewise.
31779 (gen_variable_die): Likewise.
31780 (gen_field_die): Likewise.
31781 (gen_ptr_to_mbr_type_die): Likewise.
31782 (gen_struct_or_union_type_die): Likewise.
31783 (gen_subroutine_type_die): Likewise.
31784 (gen_typedef_die): Likewise.
31785 (base_type_cmp): Compare alignment attribute.
31787 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31790 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31791 (setb_unsigned) New pattern for setb with CCUNS.
31792 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31793 subfc./subfe sequence to avoid overflow problems. Generate a
31794 shorter sequence with cmpld/setb for power9.
31795 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31796 for generating subfc. instruction.
31797 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31798 now uses this instruction.
31800 2017-01-30 Ian Lance Taylor <iant@google.com>
31803 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31804 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31806 2017-01-30 Martin Sebor <msebor@redhat.com>
31808 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31809 Move constant to the right of a relational operator.
31810 (get_mpfr_format_length, format_character, format_string): Ditto.
31811 (should_warn_p, maybe_warn): Same.
31813 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31815 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31818 * asan.c (get_translation_unit_decl): Remove function.
31819 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31821 2017-01-30 Martin Liska <mliska@suse.cz>
31823 PR gcov-profile/79259
31824 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31825 -fprofile-generate.
31827 2017-01-30 Martin Liska <mliska@suse.cz>
31830 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31831 Initialize variables with NULL value.
31833 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31836 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31838 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31840 2017-01-30 Richard Biener <rguenther@suse.de>
31842 PR tree-optimization/79276
31843 * tree-vrp.c (process_assert_insertions): Properly adjust common
31844 when removing a duplicate.
31846 2017-01-30 Richard Biener <rguenther@suse.de>
31848 PR tree-optimization/79256
31849 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31850 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31852 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31854 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31857 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31858 ("*r<noxa>sbg_<mode>_sll_bitmask")
31859 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31860 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31861 Use contiguous_bitmask_nowrap_operand.
31863 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31866 * config/rs6000/altivec.h (vec_xl): Revise #define.
31867 (vec_xst): Likewise.
31869 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31871 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31873 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31875 PR rtl-optimization/79194
31876 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31877 traps before call to bypass_conditional_jumps.
31879 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31881 PR tree-optimization/71374
31882 * lra-constraints.c (check_conflict_input_operands): New.
31883 (match_reload): Use it.
31885 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31888 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31889 account to calculate conflict_set.
31891 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31893 PR rtl-optimization/78559
31894 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31895 other_insn in combine.
31897 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31899 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31900 uint16_type_node for BT_UINT16.
31902 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31904 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31905 "RTL Tests" to menu.
31906 (GIMPLE Tests): New node.
31907 (RTL Tests): New node.
31909 2017-01-27 Richard Biener <rguenther@suse.de>
31911 PR tree-optimization/79245
31912 * tree-loop-distribution.c (distribute_loop): Apply cost
31913 modeling also to detected patterns.
31915 2017-01-27 Richard Biener <rguenther@suse.de>
31917 PR tree-optimization/71433
31918 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31919 (compare_assert_loc): New function.
31920 (process_assert_insertions): Sort and optimize assert locations
31921 to remove duplicates and push down identical assertions on
31922 edges to their destination block.
31924 2017-01-27 Richard Biener <rguenther@suse.de>
31926 PR tree-optimization/79244
31927 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31928 out SSA names even if abnormal.
31930 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31932 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31933 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31934 instead of MPFR_RNDN.
31936 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31939 * arm.c (arm_option_override): Don't call build_target_option_node
31940 until after doing all option overrides.
31941 (arm_valid_target_attribute_tree): Likewise.
31943 2017-01-27 Martin Liska <mliska@suse.cz>
31945 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31946 for {cd}tors and C++ {cd}tors.
31948 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31950 * config/s390/s390.md ("*setmem_long_and")
31951 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31953 2017-01-26 Martin Sebor <msebor@redhat.com>
31955 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31958 2017-01-26 Martin Sebor <msebor@redhat.com>
31960 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31961 HAVE_DFmode before using XFmode or DFmode.
31962 (parse_directive): Avoid using the z length modifier to avoid
31963 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31965 PR middle-end/78703
31966 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31967 to accept adjustment as an array.
31968 (get_int_range): New function.
31969 (struct directive): Make width and prec arrays.
31970 (directive::set_width, directive::set_precision): Call get_int_range.
31971 (format_integer, format_floating): Handle width and precision ranges.
31972 (format_string, parse_directive): Same.
31974 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31977 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31978 comdat_type_p set, just clone them, but keep the children in the
31982 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31983 which have direct callers with -fvar-tracking-assignments enabled
31985 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31986 inside of type units.
31988 2017-01-26 Martin Sebor <msebor@redhat.com>
31990 PR middle-end/78703
31991 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31993 (struct format_result): Replace number_chars, number_chars_min,
31994 and number_chars_max with a single member of struct result_range.
31996 (format_result::operator+=): Adjust.
31997 (struct fmtresult): Remove bounded. Handle likely and unlikely
31999 (fmtresult::adjust_for_width_or_precision): New function.
32000 (fmtresult:type_max_digits): New function.
32001 (bytes_remaining): Handle likely and unlikely counters.
32002 (min_bytes_remaining): Remove.
32003 (format_percent): Simplify.
32004 (format_integer, format_floating): Set likely and unlikely counters.
32005 (get_string_length, format_character, format_string): Same.
32006 (format_plain, should_warn_p): New function.
32007 (maybe_warn): Call should_warn_p. Update diagnostic messages
32008 and handle those for all directives, including plain strings.
32009 (format_directive): Handle likely and unlikely counters.
32010 Remove unnecessary quoting from diagnostics. Add an informational
32012 (add_bytes): Remove.
32013 (pass_sprintf_length::compute_format_length): Simplify.
32014 (try_substitute_return_value): Handle likely and unlikely counters.
32016 2017-01-26 Carl Love <cel@us.ibm.com>
32018 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32019 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32021 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32024 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32025 endianess for subregs into account.
32026 * lra-constraints.c (lra_constraints): Do risky transformations
32027 always on the first iteration.
32028 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32029 last_call_used_reg_set.
32030 (process_bb_lives): Define and use last_call_used_reg_set.
32031 * lra.c (lra): Always continue after lra_constraints on the first
32034 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32036 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32038 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32040 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32042 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32043 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32044 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32045 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32046 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32047 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32048 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32049 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32050 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32052 2017-01-26 Marek Polacek <polacek@redhat.com>
32055 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32056 for the third operand.
32058 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32060 PR middle-end/79236
32061 * omp-low.c (struct omp_context): Add simt_stmt field.
32062 (scan_omp_for): Return omp_context *.
32063 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32064 context to the _simt_ SIMD stmt.
32065 (lower_omp_for): For combined SIMD with sibling _simt_
32066 SIMD, make sure to use the same decls in _looptemp_
32067 clauses as in the sibling.
32069 2017-01-26 David Sherwood <david.sherwood@arm.com>
32071 PR middle-end/79212
32072 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32075 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32078 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32079 emit fld b; fld a; if possible.
32081 * brig-builtins.def: Update copyright years.
32082 * config/arm/arm_acle_builtins.def: Update copyright years.
32084 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32087 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32088 constraint instead of o for the stxsd instruction.
32090 2017-01-25 Carl Love <cel@us.ibm.com>
32092 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32093 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32095 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32097 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32099 2017-01-25 Richard Biener <rguenther@suse.de>
32101 PR tree-optimization/69264
32102 * target.def (vector_alignment_reachable): Improve documentation.
32103 * doc/tm.texi: Regenerate.
32104 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32106 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32107 earlier changes with respect to TYPE_USER_ALIGN.
32108 (vector_alignment_reachable_p): Likewise. Improve dumping.
32110 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32113 * config/arm/arm.md (xordi3): Force constant operand into a register
32116 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32118 * doc/invoke.texi (-fstore-merging): Correct default optimization
32119 levels at which it is enabled.
32120 (-O): Move -fstore-merging from list to...
32123 2017-01-25 Richard Biener <rguenther@suse.de>
32126 * omp-expand.c: Include debug.h.
32127 (expand_omp_taskreg): Make sure to generate early debug before
32128 outlining anything from a function.
32129 (expand_omp_target): Likewise.
32130 (grid_expand_target_grid_body): Likewise.
32132 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32135 * asan.c (get_translation_unit_decl): New function.
32136 (asan_add_global): Extract modules file name from globals
32137 TRANSLATION_UNIT_DECL name.
32139 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32142 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32143 for long calls with APCS frame and VFP.
32145 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32147 * cfg.c (original_copy_tables_initialized_p): New function.
32148 * cfg.h (original_copy_tables_initialized_p): New decl.
32149 * cfgrtl.c (relink_block_chain): Guard the call to
32150 free_original_copy_tables with a call to
32151 original_copy_tables_initialized_p.
32152 * cgraph.h (symtab_node::native_rtl_p): New decl.
32153 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32154 (symtab_node::needed_p): Don't assert for early assembly output
32155 for __RTL functions.
32156 (cgraph_node::finalize_function): Set "force_output" for __RTL
32158 (cgraph_node::analyze): Bail out early for __RTL functions.
32159 (analyze_functions): Update assertion to support __RTL functions.
32160 (cgraph_node::expand): Bail out early for __RTL functions.
32161 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32163 * function.h (struct function): Update comment for field
32165 * gimple-expr.c: Include "tree-pass.h".
32166 (gimple_has_body_p): Return false for __RTL functions.
32167 * Makefile.in (OBJS): Add run-rtl-passes.o.
32168 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32170 (gcc::pass_manager::get_clean_slate): New accessor.
32171 * passes.c: Include "insn-addr.h".
32172 (should_skip_pass_p): Add logging. Update logic for running
32173 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32174 property-provider override so it is only done for gimple passes.
32176 (skip_pass): New function.
32177 (execute_one_pass): Call skip_pass when skipping passes.
32178 * read-md.c (md_reader::read_char): Support filtering
32179 the input to a subset of line numbers.
32180 (md_reader::md_reader): Initialize fields
32181 m_first_line and m_last_line.
32182 (md_reader::read_file_fragment): New function.
32183 * read-md.h (md_reader::read_file_fragment): New decl.
32184 (md_reader::m_first_line): New field.
32185 (md_reader::m_last_line): New field.
32186 * read-rtl-function.c (function_reader::create_function): Only
32187 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32188 curr_properties. Set DECL_INITIAL to a dummy block.
32189 (read_rtl_function_body_from_file_range): New function.
32190 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32192 * run-rtl-passes.c: New file.
32193 * run-rtl-passes.h: New file.
32195 2017-01-24 Jeff Law <law@redhat.com>
32197 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32200 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32202 PR tree-optimization/79159
32203 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32204 (record_nonwrapping_iv): Improve boundary using above function if no
32205 value range information.
32207 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32208 Martin Jambor <mjambor@suse.cz>
32210 * brig-builtins.def: New file.
32211 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32212 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32213 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32214 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32215 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32216 * builtin-types.def (BT_INT8): New.
32217 (BT_INT16): Likewise.
32218 (BT_UINT8): Likewise.
32219 (BT_UINT16): Likewise.
32220 (BT_FN_ULONG): Likewise.
32221 (BT_FN_UINT_INT): Likewise.
32222 (BT_FN_UINT_ULONG): Likewise.
32223 (BT_FN_UINT_LONG): Likewise.
32224 (BT_FN_UINT_PTR): Likewise.
32225 (BT_FN_ULONG_PTR): Likewise.
32226 (BT_FN_INT8_FLOAT): Likewise.
32227 (BT_FN_INT16_FLOAT): Likewise.
32228 (BT_FN_UINT32_FLOAT): Likewise.
32229 (BT_FN_UINT16_FLOAT): Likewise.
32230 (BT_FN_UINT8_FLOAT): Likewise.
32231 (BT_FN_UINT64_FLOAT): Likewise.
32232 (BT_FN_UINT16_UINT32): Likewise.
32233 (BT_FN_UINT32_UINT16): Likewise.
32234 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32235 (BT_FN_INT_PTR_INT): Likewise.
32236 (BT_FN_UINT_PTR_UINT): Likewise.
32237 (BT_FN_LONG_PTR_LONG): Likewise.
32238 (BT_FN_ULONG_PTR_ULONG): Likewise.
32239 (BT_FN_VOID_UINT64_UINT64): Likewise.
32240 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32241 (BT_FN_INT8_INT8_INT8): Likewise.
32242 (BT_FN_INT16_INT16_INT16): Likewise.
32243 (BT_FN_INT_INT_INT): Likewise.
32244 (BT_FN_UINT_FLOAT_UINT): Likewise.
32245 (BT_FN_FLOAT_UINT_UINT): Likewise.
32246 (BT_FN_ULONG_UINT_UINT): Likewise.
32247 (BT_FN_ULONG_UINT_PTR): Likewise.
32248 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32249 (BT_FN_UINT_UINT_UINT): Likewise.
32250 (BT_FN_VOID_UINT_PTR): Likewise.
32251 (BT_FN_UINT_UINT_PTR: Likewise.
32252 (BT_FN_UINT32_UINT64_PTR): Likewise.
32253 (BT_FN_INT_INT_UINT_UINT): Likewise.
32254 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32255 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32256 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32257 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32258 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32259 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32260 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32261 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32262 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32263 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32264 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32265 * doc/frontends.texi: List BRIG FE.
32266 * doc/install.texi (Testing): Add BRIG tesring requirements.
32267 * doc/invoke.texi (Overall Options): Mention BRIG.
32268 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32270 2017-01-24 Richard Biener <rguenther@suse.de>
32272 PR translation/79208
32273 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32275 2017-01-24 Martin Jambor <mjambor@suse.cz>
32278 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32279 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32280 and known_contexts.
32282 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32284 PR middle-end/79123
32285 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32286 casts from signed to unsigned really don't have a range.
32288 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32290 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32291 GMP_RNDx for compatiblity.
32293 2017-01-24 Martin Liska <mliska@suse.cz>
32296 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32297 that would prevent us to call alloca with -1 as argument.
32299 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32301 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32302 Avoid -Wformat-security warning.
32304 2017-01-23 Andrew Pinski <apinski@cavium.com>
32306 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32309 2017-01-23 Martin Sebor <msebor@redhat.com>
32311 PR middle-end/78703
32312 * gimple-ssa-sprintf.c (warn_level): New global.
32313 (format_integer): Use it here and throughout the rest of the file.
32314 Use the same switch to compute sign as base.
32315 (maybe_warn): New function.
32316 (format_directive): Factor out warnings into maybe_warn.
32317 Add debugging output. Use warn_level.
32318 (add_bytes): Use warn_level.
32319 (pass_sprintf_length::compute_format_length): Add debugging output.
32320 (try_substitute_return_value): Same.
32321 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32323 PR middle-end/78703
32324 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32325 (struct fmtresult, format_integer, format_floating): Adjust.
32326 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32327 (get_string_length, format_string,format_directive): Same.
32328 (pass_sprintf_length::compute_format_length): Same.
32329 (try_substitute_return_value): Simplify slightly.
32331 PR middle-end/78703
32332 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32333 (fmtresult::operator+=): Outlined.
32334 (struct fmtresult): Add ctors.
32335 (struct conversion_spec): Rename...
32336 (struct directive): ...to this. Add and remove data members.
32337 (directive::set_width, directive::set_precision): New functions.
32338 (format_percent): Use fmtresult ctor.
32339 (get_width_and_precision): Remove.
32340 (format_integer): Make naming changes. Avoid computing width and
32342 (format_floating): Same. Adjust indentation.
32343 (format_character, format_none): New functions.
32344 (format_string): Moved character handling to format_character.
32345 (format_directive): Remove arguments, change return type.
32346 (parse_directive): New function.
32347 (pass_sprintf_length::compute_format_length): Move directive
32348 parsing to parse_directive.
32350 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32352 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32353 (assign_assembler_name_if_needed): ... this.
32354 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32355 (assign_assembler_name_if_needed): ... this.
32356 (free_lang_data_in_cgraph): Adjust callers.
32357 * cgraphunit.c (cgraph_node::analyze): Likewise.
32358 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32361 2017-01-23 Richard Biener <rguenther@suse.de>
32363 PR tree-optimization/79088
32364 PR tree-optimization/79188
32365 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32366 resetting loop bounds after last path deletion. Reset loop
32367 bounds of the target loop, make code match the comments.
32368 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32369 Make sure loops need no fixups.
32371 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32373 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32374 exponent support with double type for first argument.
32375 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32376 type returned by __builtin_vec_extract_sig,
32377 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32378 functions from "vector int" to "vector unsigned int" or from
32379 "vector long long int" to "vector unsigned long long int".
32380 Changed type returned by __builtin_vec_extract_exp,
32381 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32382 functions from "vector int" to "vector unsigned int" or from
32383 "vector long long int" to "vector unsigned long long int".
32384 Changed return type of __builtin_vec_test_data_class,
32385 __builtin_vec_test_data_class_sp, and
32386 __builtin_vec_test_data_class_dp from "vector int" to
32387 "vector bool int" or from "vector long long int" to "vector bool
32388 long long int" and changed second argument type from "unsigned
32389 int" to "int". Added new overloaded function forms "vector float
32390 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32391 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32392 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32393 double, vector unsigned long long int)" and "vector double
32394 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32395 long int)". Changed return type of
32396 __builtin_scalar_test_data_class and
32397 __builtin_scalar_test_data_class_sp and
32398 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32399 int" and changed second argument from "unsigned int" to "int".
32400 Changed type returned by __builtin_scalar_test_neg,
32401 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32402 from "int" to "bool int". Added new overloaded function form
32403 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32404 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32405 exponent double-precision with floating point first argument.
32406 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32407 documentation of scalar_test_data_class, scalar_test_neg,
32408 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32409 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32410 vec_test_data_class built-in functions to reflect refinements in
32411 their type signatures.
32413 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32415 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32417 (aarch64_elf_asm_destructor): Likewise.
32419 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32421 PR rtl-optimization/78634
32422 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32423 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32424 * ifcvt.c (noce_try_cmove): Add missing cost check.
32426 PR rtl-optimization/71724
32427 * combine.c (if_then_else_cond): Look for situations where it is
32428 beneficial to undo the work of one of the recursive calls.
32430 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32432 PR tree-optimization/70754
32433 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32434 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32435 combined stmt before it if not NULL.
32436 (combine_chains): Process refs reversely and compute dominance point
32439 2017-01-23 Martin Liska <mliska@suse.cz>
32441 PR tree-optimization/79196
32442 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32443 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32445 (strlen_optimize_stmt): Call the renamed function.
32447 2017-01-23 Michael Matz <matz@suse.de>
32449 PR tree-optimization/78384
32450 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32452 2017-01-23 Richard Biener <rguenther@suse.de>
32454 PR tree-optimization/79186
32455 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32456 both incoming edges before moving an assert.
32458 2017-01-23 Martin Jambor <mjambor@suse.cz>
32460 * ipa-prop.c (load_from_param_1): Removed.
32461 (load_from_unmodified_param): Bits from load_from_param_1 put back
32463 (load_from_param): Removed.
32464 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32465 with stmt. Reverted back to use of load_from_unmodified_param.
32467 2017-01-23 Martin Jambor <mjambor@suse.cz>
32470 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32471 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32472 field a pointer to garbage collected vector, mark lattices and
32473 ipcp_orig_node with GTY((skip)).
32474 (ipa_get_param_count): Adjust to descriptors being a pointer.
32475 (ipa_get_param): Likewise.
32476 (ipa_get_type): Likewise.
32477 (ipa_get_param_move_cost): Likewise.
32478 (ipa_set_param_used): Likewise.
32479 (ipa_get_controlled_uses): Likewise.
32480 (ipa_set_controlled_uses): Likewise.
32481 (ipa_is_param_used): Likewise.
32482 (ipa_node_params_t): Move into garbage collector. New methods insert
32484 (ipa_node_params_sum): Annotate wth GTY(()).
32485 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32487 (ipa_load_from_parm_agg): Adjust declaration.
32488 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32489 * ipa-profile.c (ipa_profile): Likewise.
32490 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32491 (ipa_populate_param_decls): Make descriptors parameter garbage
32493 (ipa_dump_param): Adjust to descriptors being a pointer.
32494 (ipa_alloc_node_params): Likewise.
32495 (ipa_initialize_node_params): Likewise.
32496 (load_from_param_1): Make descriptors parameter garbage collected.
32497 (load_from_unmodified_param): Likewise.
32498 (load_from_param): Likewise.
32499 (ipa_load_from_parm_agg): Likewise.
32500 (ipa_node_params::~ipa_node_params): Removed.
32501 (ipa_free_all_node_params): Remove call to delete operator.
32502 (ipa_node_params_t::insert): New.
32503 (ipa_node_params_t::remove): Likewise.
32504 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32505 copy known_csts and known_contexts vectors.
32506 (ipa_read_node_info): Adjust to descriptors being a pointer.
32507 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32509 (ipcp_transform_function): Make descriptors variable garbage
32512 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32514 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32515 * config/i386/avx512dqintrin.h: Ditto.
32516 * config/i386/avx512fintrin.h: Ditto.
32517 * config/i386/i386.c: Handle new builtins.
32518 * config/i386/i386-builtin.def: Add new builtins.
32519 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32520 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32522 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32523 Martin Liska <mliska@suse.cz>
32525 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32526 * asan.c (asan_expand_poison_ifn): Support stores and use
32527 appropriate ASAN report function.
32528 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32529 * internal-fn.def (ASAN_POISON_USE): Declare.
32530 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32531 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32532 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32533 ASAN_POISON calls w/o LHS.
32534 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32535 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32536 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32537 * gimplify.c (asan_poison_variables): Add attribute
32538 use_after_scope_memory to variables that really needs to live
32540 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32541 having the attribute.
32543 2017-01-23 Martin Liska <mliska@suse.cz>
32545 * asan.c (create_asan_shadow_var): New function.
32546 (asan_expand_poison_ifn): Likewise.
32547 * asan.h (asan_expand_poison_ifn): New declaration.
32548 * internal-fn.c (expand_ASAN_POISON): Likewise.
32549 * internal-fn.def (ASAN_POISON): New builtin.
32550 * sanopt.c (pass_sanopt::execute): Expand
32551 asan_expand_poison_ifn.
32552 * tree-inline.c (copy_decl_for_dup_finish): Make function
32554 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32555 * tree-ssa.c (is_asan_mark_p): New function.
32556 (execute_update_addresses_taken): Rewrite local variables
32557 (identified just by use-after-scope as addressable) into SSA.
32559 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32561 * doc/install.texi (Specific): opensource.apple.com uses https
32562 now. Remove trailing slash.
32564 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32566 * README.Portability: Remove note on an Irix compatibility issue.
32568 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32570 * gcov.c (INCLUDE_ALGORITHM): Define.
32571 (INCLUDE_VECTOR): Define.
32572 No longer include <vector> and <algorithm> directly.
32574 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32576 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32578 * doc/invoke.texi (Code Gen Options): Ditto.
32580 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32583 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32585 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32587 rtl-optimization/79125
32588 * cprop.c (local_cprop_pass): Handle cases where we make an
32589 unconditional trap.
32591 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32595 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32596 read from, for big endian.
32598 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32600 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32601 register pauth builtins for LP64 only.
32603 2017-01-20 Marek Polacek <polacek@redhat.com>
32606 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32609 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32611 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32613 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32614 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32615 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32617 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32620 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32621 in big-endian mode when they are not a single duplicated value.
32623 2017-01-20 Richard Biener <rguenther@suse.de>
32625 * BASE-VER: Bump to 7.0.1.
32627 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32629 * omp-low.c (omplow_simd_context): New struct. Use it...
32630 (lower_rec_simd_input_clauses): ...here and...
32631 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32632 references to idx, lane, max_vf, is_simt.
32634 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32636 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32639 2017-01-20 Martin Jambor <mjambor@suse.cz>
32641 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32642 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32644 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32645 (GTFILES): Rename hsa.c to hsa-common.c.
32646 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32647 * hsa-dump.c: Likewise.
32648 * hsa-gen.c: Likewise.
32649 * hsa-regalloc.c: Likewise.
32650 * ipa-hsa.c: Likewise.
32651 * omp-expand.c: Likewise.
32652 * omp-low.c: Likewise.
32653 * toplev.c: Likewise.
32655 2017-01-20 Marek Polacek <polacek@redhat.com>
32658 * doc/invoke.texi: Document -Wduplicated-branches.
32659 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32660 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32661 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32662 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32663 return 0 only when not OEP_LEXICOGRAPHIC.
32664 (fold_build_cleanup_point_expr): Use the expression
32665 location when building CLEANUP_POINT_EXPR.
32666 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32667 * tree.c (add_expr): Handle error_mark_node.
32669 2017-01-20 Martin Liska <mliska@suse.cz>
32672 * tree-profile.c (init_ic_make_global_vars): Do not call
32674 (gimple_init_gcov_profiler): Likewise.
32676 2017-01-20 Martin Liska <mliska@suse.cz>
32679 * cgraph.h (maybe_create_reference): Remove argument and
32681 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32683 * ipa-cp.c (create_specialized_node): Likewise.
32684 * symtab.c (symtab_node::maybe_create_reference): Handle
32685 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32687 2017-01-20 Martin Liska <mliska@suse.cz>
32689 * read-rtl-function.c (function_reader::create_function): Use
32690 build_decl instread of build_decl_stat.
32692 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32694 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32695 * config/i386/avx512dqintrin.h: Ditto.
32696 * config/i386/avx512fintrin.h: Ditto.
32697 * config/i386/i386-builtin-types.def: Add new types.
32698 * config/i386/i386.c: Handle new types.
32699 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32700 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32701 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32702 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32703 (__builtin_ia32_kshiftridi): New.
32704 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32706 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32710 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32711 define to rs6000_init_stack_protect_guard.
32712 (rs6000_init_stack_protect_guard): New function.
32714 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32715 Yunqiang Su <yunqiang.su@imgtec.com>
32717 * config.gcc (supported_defaults): Add madd4.
32718 (with_madd4): Add validation.
32719 (all_defaults): Add madd4.
32720 * config/mips/mips.opt (mmadd4): New option.
32721 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32723 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32725 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32726 (ISA_HAS_FUSED_MADD4): Likewise.
32727 * doc/invoke.texi (-mmadd4): Document the new option.
32728 * doc/install.texi (--with-madd4): Document the new option.
32730 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32732 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32733 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32734 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32735 (aarch64_init_pauth_hint_builtins): New.
32736 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32737 (aarch64_expand_builtin): Expand new builtins.
32739 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32741 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32742 * combine-stack-adj.c (no_unhandled_cfa): Handle
32743 REG_CFA_TOGGLE_RA_MANGLE.
32744 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32745 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32746 info for return address signing.
32747 (aarch64_expand_epilogue): Likewise.
32749 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32751 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32752 * config/aarch64/aarch64-protos.h
32753 (aarch64_return_address_signing_enabled): New declaration.
32754 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32756 (aarch64_expand_prologue): Sign return address before it's pushed onto
32758 (aarch64_expand_epilogue): Authenticate return address fetched from
32760 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32761 (aarch64_attributes): New function attributes for "sign-return-address".
32762 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32763 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32764 ("*do_return"): Generate combined instructions according to key index.
32765 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32766 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32768 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32769 * config/aarch64/aarch64.opt (msign-return-address=): New.
32770 * doc/extend.texi (AArch64 Function Attributes): Documents
32771 "sign-return-address=".
32772 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32774 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32776 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32777 overall option summary.
32779 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32781 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32782 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32783 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32784 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32786 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32788 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32789 -mpower9-minmax by default for -mcpu=power9.
32790 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32791 128-bit floating point.
32793 2017-01-20 Alan Modra <amodra@gmail.com>
32795 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32796 optimizing for size.
32798 2017-01-20 Alan Modra <amodra@gmail.com>
32801 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32802 for strcmp and strncmp from corresponding builtin decl.
32804 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32806 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32807 instead of i386/rtems-64.h.
32808 * config/i386/rtems-64.h: Remove.
32810 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32814 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32816 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32818 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32820 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32821 Change int to HOST_WIDE_INT.
32822 * config/aarch64/aarch64-protos.h
32823 (aarch64_simd_gen_const_vector_dup): Likewise.
32824 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32826 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32828 * langhooks-def.h (lhd_type_for_size): New decl.
32829 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32830 * langhooks.c (lhd_type_for_size): New function, taken from
32833 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32835 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32836 define_bypass for CR latency.
32837 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32838 (power9-alu2): Add define_bypass for CR latency.
32840 (power9-mul): Update insn latency.
32841 (power9-mul-compare): Update insn latency, bypass latency and remove
32844 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32846 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32848 * config/aarch64/aarch64.md
32849 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32850 aarch64_nopcrelative_literal_loads.
32851 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32853 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32855 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32856 TARGET_LOONGSON_3A.
32857 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32859 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32862 * config.gcc (supported_defaults): Add lxc1-sxc1.
32863 (with_lxc1_sxc1): Add validation.
32864 (all_defaults): Add lxc1-sxc1.
32865 * config/mips/mips.opt (mlxc1-sxc1): New option.
32866 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32868 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32869 __mips_no_lxc1_sxc1.
32870 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32871 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32872 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32874 2017-01-19 Richard Biener <rguenther@suse.de>
32876 PR tree-optimization/72488
32877 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32878 sure to restore SSA info.
32879 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32881 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32883 PR rtl-optimization/79121
32884 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32885 of the inner type when shifting an extended value.
32887 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32890 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32891 interposable aliases.
32893 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32896 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32897 Use the evmergelohi instruction.
32898 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32899 (mov_sitf_e500_subreg8_2_be): Likewise.
32900 (mov_sitf_e500_subreg12_2_le): Likewise.
32901 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32902 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32903 (mov_sitf_e500_subreg8_2_le): Likewise.
32904 (mov_sitf_e500_subreg12_2_be): Likewise.
32906 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32908 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32909 attribute from vecsimple to vecperm.
32910 (altivec_vbpermq2): Likewise.
32912 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32915 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32917 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32918 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32919 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32920 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32921 case where N arg is SIZE_MAX.
32922 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32923 (cmpstrsi): Add pattern.
32925 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32927 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32928 __builtin_vec_revb builtins.
32929 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32930 built-in functions to support generation of the ISA 3.0 XXBR<x>
32931 vector byte reverse instructions.
32932 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32933 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32934 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32935 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32936 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32937 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32938 (P9V_BUILTIN_VEC_REVB): Likewise.
32939 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32940 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32941 (p9_xxbrq_v16qi): Likewise.
32942 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32943 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32944 (p9_xxbrh_v8hi): Likewise.
32945 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32946 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32947 vec_revb built-in functions.
32949 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32951 PR rtl-optimization/78952
32952 * config/i386/i386.md (any_extract): New code iterator.
32953 (*insvqi_2): Use any_extract for source operand.
32954 (*insvqi_3): Use any_shiftrt for source operand.
32956 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32958 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32960 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32962 2017-01-18 Matthias Klose <doko@ubuntu.com>
32964 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32966 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32968 * config/rs6000/altivec.h (vec_bperm): Change #define.
32969 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32970 (altivec_vbpermq2): New define_insn.
32971 (altivec_vbpermd): Likewise.
32972 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32973 function interface.
32974 (VBPERMD): Likewise.
32975 (VBPERM): New polymorphic function interface.
32976 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32977 Add entries for P9V_BUILTIN_VEC_VBPERM.
32978 * doc/extend.texi: Add interfaces for vec_bperm.
32980 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32982 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32983 first letter of error messages.
32984 (s390_resolve_overloaded_builtin): Likewise.
32985 * config/s390/s390.c (s390_expand_builtin): Likewise.
32986 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32987 (s390_valid_target_attribute_inner_p): Likewise.
32988 * config/s390/s390.md ("tabort"): Likewise.
32990 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32992 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32993 (ISA_AVOID_DIV_HILO): New macro.
32994 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32995 (ISA_HAS_DDIV): Likewise.
32997 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32999 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33001 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33003 * doc/invoke.texi (fabi-version): Spelling fix.
33005 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33008 * doc/invoke.texi (fabi-version): Mention mangling fix for
33011 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33014 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33016 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33019 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33020 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33022 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33024 (stack_protect_test): Ditto.
33025 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33026 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33028 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33029 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33030 -mstack-protector-guard-offset=.
33031 (RS/6000 and PowerPC Options): Ditto.
33033 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33035 * config/i386/i386.h (MASK_CLASS_P): New define.
33036 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33037 there are no registers from different register sets also when
33038 mask registers are used. Update function comment.
33039 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33040 to (*k/*r) and (*k/*km) alternatives.
33042 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33044 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33045 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33046 (EH_RETURN_HANDLER_RTX): New define.
33047 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33048 Force frame pointer in EH return functions.
33049 (aarch64_expand_epilogue): Add barrier for eh_return.
33050 (aarch64_final_eh_return_addr): Remove.
33051 (aarch64_eh_return_handler_rtx): New function.
33052 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33054 (aarch64_eh_return_handler_rtx): New prototype.
33056 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33058 * config/rs6000/altivec.h (vec_rlmi): New #define.
33059 (vec_vrlnm): Likewise.
33060 (vec_rlnm): Likewise.
33061 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33062 (UNSPEC_VRLNM): Likewise.
33063 (VIlong): New mode iterator.
33064 (altivec_vrl<VI_char>mi): New define_insn.
33065 (altivec_vrl<VI_char>nm): Likewise.
33066 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33068 (VRLDNM): Likewise.
33069 (RLNM): New polymorphic function entry.
33070 (VRLWMI): New monomorphic function entry.
33071 (VRLDMI): Likewise.
33072 (RLMI): New polymorphic function entry.
33073 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33074 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33075 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33078 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33081 * dwarf2out.c (field_byte_offset): Restore the
33082 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33083 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33084 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33087 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33090 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33092 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33095 * dwarf2out.c (add_data_member_location_attribute): For constant
33096 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33097 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33098 DW_AT_byte_size attributes.
33100 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33102 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33103 after forcing to constant memory when the code model is medium.
33105 2017-01-17 Julia Koval <julia.koval@intel.com>
33108 * config/i386/avx512fintrin.h
33109 (_mm512_i32gather_ps): Change __addr type to void const*.
33110 (_mm512_mask_i32gather_ps): Ditto.
33111 (_mm512_i32gather_pd): Ditto.
33112 (_mm512_mask_i32gather_pd): Ditto.
33113 (_mm512_i64gather_ps): Ditto.
33114 (_mm512_mask_i64gather_ps): Ditto.
33115 (_mm512_i64gather_pd): Ditto.
33116 (_mm512_mask_i64gather_pd): Ditto.
33117 (_mm512_i32gather_epi32): Ditto.
33118 (_mm512_mask_i32gather_epi32): Ditto.
33119 (_mm512_i32gather_epi64): Ditto.
33120 (_mm512_mask_i32gather_epi64): Ditto.
33121 (_mm512_i64gather_epi32): Ditto.
33122 (_mm512_mask_i64gather_epi32): Ditto.
33123 (_mm512_i64gather_epi64): Ditto.
33124 (_mm512_mask_i64gather_epi64): Ditto.
33125 (_mm512_i32scatter_ps): Change __addr type to void*.
33126 (_mm512_mask_i32scatter_ps): Ditto.
33127 (_mm512_i32scatter_pd): Ditto.
33128 (_mm512_mask_i32scatter_pd): Ditto.
33129 (_mm512_i64scatter_ps): Ditto.
33130 (_mm512_mask_i64scatter_ps): Ditto.
33131 (_mm512_i64scatter_pd): Ditto.
33132 (_mm512_mask_i64scatter_pd): Ditto.
33133 (_mm512_i32scatter_epi32): Ditto.
33134 (_mm512_mask_i32scatter_epi32): Ditto.
33135 (_mm512_i32scatter_epi64): Ditto.
33136 (_mm512_mask_i32scatter_epi64): Ditto.
33137 (_mm512_i64scatter_epi32): Ditto.
33138 (_mm512_mask_i64scatter_epi32): Ditto.
33139 (_mm512_i64scatter_epi64): Ditto.
33140 (_mm512_mask_i64scatter_epi64): Ditto.
33141 * config/i386/avx512pfintrin.h
33142 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33143 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33144 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33145 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33146 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33147 (_mm512_prefetch_i32scatter_ps): Ditto.
33148 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33149 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33150 (_mm512_prefetch_i64scatter_pd): Ditto.
33151 (_mm512_prefetch_i64scatter_ps): Ditto.
33152 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33153 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33154 * config/i386/avx512vlintrin.h
33155 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33156 (_mm_mmask_i32gather_ps): Ditto.
33157 (_mm256_mmask_i32gather_pd): Ditto.
33158 (_mm_mmask_i32gather_pd): Ditto.
33159 (_mm256_mmask_i64gather_ps): Ditto.
33160 (_mm_mmask_i64gather_ps): Ditto.
33161 (_mm256_mmask_i64gather_pd): Ditto.
33162 (_mm_mmask_i64gather_pd): Ditto.
33163 (_mm256_mmask_i32gather_epi32): Ditto.
33164 (_mm_mmask_i32gather_epi32): Ditto.
33165 (_mm256_mmask_i32gather_epi64): Ditto.
33166 (_mm_mmask_i32gather_epi64): Ditto.
33167 (_mm256_mmask_i64gather_epi32): Ditto.
33168 (_mm_mmask_i64gather_epi32): Ditto.
33169 (_mm256_mmask_i64gather_epi64): Ditto.
33170 (_mm_mmask_i64gather_epi64): Ditto.
33171 (_mm256_i32scatter_ps): Change __addr type to void*.
33172 (_mm256_mask_i32scatter_ps): Ditto.
33173 (_mm_i32scatter_ps): Ditto.
33174 (_mm_mask_i32scatter_ps): Ditto.
33175 (_mm256_i32scatter_pd): Ditto.
33176 (_mm256_mask_i32scatter_pd): Ditto.
33177 (_mm_i32scatter_pd): Ditto.
33178 (_mm_mask_i32scatter_pd): Ditto.
33179 (_mm256_i64scatter_ps): Ditto.
33180 (_mm256_mask_i64scatter_ps): Ditto.
33181 (_mm_i64scatter_ps): Ditto.
33182 (_mm_mask_i64scatter_ps): Ditto.
33183 (_mm256_i64scatter_pd): Ditto.
33184 (_mm256_mask_i64scatter_pd): Ditto.
33185 (_mm_i64scatter_pd): Ditto.
33186 (_mm_mask_i64scatter_pd): Ditto.
33187 (_mm256_i32scatter_epi32): Ditto.
33188 (_mm256_mask_i32scatter_epi32): Ditto.
33189 (_mm_i32scatter_epi32): Ditto.
33190 (_mm_mask_i32scatter_epi32): Ditto.
33191 (_mm256_i32scatter_epi64): Ditto.
33192 (_mm256_mask_i32scatter_epi64): Ditto.
33193 (_mm_i32scatter_epi64): Ditto.
33194 (_mm_mask_i32scatter_epi64): Ditto.
33195 (_mm256_i64scatter_epi32): Ditto.
33196 (_mm256_mask_i64scatter_epi32): Ditto.
33197 (_mm_i64scatter_epi32): Ditto.
33198 (_mm_mask_i64scatter_epi32): Ditto.
33199 (_mm256_i64scatter_epi64): Ditto.
33200 (_mm256_mask_i64scatter_epi64): Ditto.
33201 (_mm_i64scatter_epi64): Ditto.
33202 (_mm_mask_i64scatter_epi64): Ditto.
33203 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33204 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33205 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33206 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33207 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33208 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33209 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33210 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33211 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33212 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33213 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33214 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33215 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33216 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33217 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33218 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33219 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33220 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33221 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33222 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33223 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33224 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33225 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33226 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33227 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33228 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33229 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33230 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33231 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33232 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33233 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33234 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33235 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33236 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33237 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33238 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33239 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33240 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33241 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33242 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33243 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33244 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33245 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33246 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33247 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33248 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33249 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33250 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33251 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33252 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33253 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33254 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33255 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33256 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33257 definitions accordingly.
33259 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33260 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33263 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33266 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33269 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33270 hard regs for inner regno.
33272 2017-01-17 Martin Liska <mliska@suse.cz>
33275 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33276 assumption and add comment.
33278 2017-01-17 Nathan Sidwell <nathan@acm.org>
33280 * ipa-visibility.c (localize_node): New function, broken out of ...
33281 (function_and_variable_visibility): ... here. Call it.
33283 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33285 PR middle-end/77445
33286 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33287 correctly set frequency of oudgoing edge.
33288 (duplicate_thread_path): Fix profile updating.
33290 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33293 * configure.ac: Add GCC_BASE_VER.
33294 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33295 version from BASE-VER file.
33296 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33297 (gcc.o): Depend on $(BASEVER).
33298 * common.opt (dumpfullversion): New option.
33299 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33300 * doc/invoke.texi: Document -dumpfullversion.
33301 * doc/install.texi: Document --with-gcc-major-version-only.
33302 * configure: Regenerated.
33304 2017-01-17 Richard Biener <rguenther@suse.de>
33306 PR tree-optimization/71433
33307 * tree-vrp.c (register_new_assert_for): Merge same asserts
33308 on all incoming edges.
33309 (process_assert_insertions_for): Handle insertions at the
33312 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33314 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33315 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33317 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33320 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33323 2017-01-17 Alan Modra <amodra@gmail.com>
33326 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33327 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33328 symbolic stack limit when pic.
33330 2017-01-16 Martin Sebor <msebor@redhat.com>
33332 PR tree-optimization/78608
33333 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33335 2017-01-16 Jeff Law <law@redhat.com>
33338 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33339 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33340 for several include directories that may be relative to sysroot.
33341 * config/i386/x-mingw32 (gplus_includedir): Define.
33342 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33343 (native_system_includedir): Likewise.
33344 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33345 override if TARGET_SYSTEM_ROOT is defined.
33346 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33348 PR tree-optimization/79090
33349 PR tree-optimization/33562
33350 PR tree-optimization/61912
33351 PR tree-optimization/77485
33352 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33353 and computed trims into the dump file.
33355 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33357 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33359 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33362 * gimplify.c (gimplify_init_constructor): If want_value and
33363 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33367 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33368 sequence. Formatting fixes.
33369 (doloop_optimize): Formatting fixes.
33372 * gcc.c (debug_level_greater_than_spec_func): New function.
33373 (static_spec_functions): Add debug-level-gt spec function.
33374 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33376 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33377 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33378 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33379 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33380 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33381 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33383 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33385 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33386 QImode fixups to general and mask registers only.
33388 2017-01-16 Carl Love <cel@us.ibm.com>
33390 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33391 for built-in functions
33392 vector signed char vec_nabs (vector signed char)
33393 vector signed short vec_nabs (vector signed short)
33394 vector signed int vec_nabs (vector signed int)
33395 vector signed long long vec_nabs (vector signed long long)
33396 vector float vec_nabs (vector float)
33397 vector double vec_nabs (vector double)
33398 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33400 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33401 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33402 * doc/extend.texi: Update the documentation file for the new built-in
33405 2017-01-16 Martin Sebor <msebor@redhat.com>
33407 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33410 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33412 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33413 UNSPEC_VSX__XXSPLTD to require special splat handling.
33415 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33418 * system.h: Poison strndup.
33420 2017-01-16 Alan Modra <amodra@gmail.com>
33423 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33426 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33428 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33430 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33432 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33433 call recog here. Assert that INSN_CODE (insn) is non-negative.
33435 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33438 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33440 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33441 in the currently scheduled RTL fragment.
33443 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33445 PR rtl-optimization/78751
33446 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33449 2017-01-14 Jeff Law <law@redhat.com>
33451 PR tree-optimization/79090
33452 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33453 variable length stores.
33454 (compute_trims): Delete dead assignment to *trim_tail.
33455 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33458 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33460 PR rtl-optimization/78626
33461 PR rtl-optimization/78727
33462 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33463 of a block, and split such blocks after everything else is finished.
33465 2017-01-14 Alan Modra <amodra@gmail.com>
33468 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33469 target legitimate_combined_insn.
33470 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33471 (rs6000_legitimate_combined_insn): New function.
33472 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33474 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33475 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33476 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33478 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33480 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33482 2017-01-13 Jeff Law <law@redhat.com>
33484 PR tree-optimization/33562
33485 PR tree-optimization/61912
33486 PR tree-optimization/77485
33487 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33489 (delete_dead_assignment): Likewise.
33490 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33491 statement to delete_dead_call and delete_dead_assignment.
33493 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33496 * substring-locations.c (format_warning_va): Strengthen case 1 so
33497 that both endpoints of the substring must be within the format
33498 range for just the substring to be printed.
33500 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33502 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33503 * config/i386/i386.c (ix86_target_string): Add missing options
33504 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33505 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33506 flags_other and ix86_target_other to flags2_other. Display unknown
33508 (ix86_valid_target_attribute_inner_p): Add missing options and
33509 reorder options by implied ISAs, as in ix86_target_string.
33511 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33513 * hash-table.h (hash_table::too_empty_p): New function.
33514 (hash_table::expand): Use it.
33515 (hash_table::traverse): Likewise.
33516 (hash_table::empty_slot): Use sizeof (value_type) instead of
33517 sizeof (PTR) to convert bytes to elements. Shrink the table
33518 if the current size is excessive for the current number of
33521 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33523 * ira-costs.c (record_reg_classes): Break from the inner loop
33524 early once alt_fail is known to be true. Update outer loop
33525 handling accordingly.
33527 2017-01-13 Jeff Law <law@redhat.com>
33529 * tree-ssa-dse.c (decrement_count): New function.
33530 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33531 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33532 when we know the partially dead statement is a mem* function.
33534 PR tree-optimization/61912
33535 PR tree-optimization/77485
33536 * tree-ssa-dse.c: Include expr.h.
33537 (maybe_trim_constructor_store): New function.
33538 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33540 PR tree-optimization/33562
33541 PR tree-optimization/61912
33542 PR tree-optimization/77485
33543 * doc/invoke.texi: Document new dse-max-object-size param.
33544 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33545 * tree-ssa-dse.c: Include params.h.
33546 (dse_store_status): New enum.
33547 (initialize_ao_ref_for_dse): New, partially extracted from
33549 (valid_ao_ref_for_dse, normalize_ref): New.
33550 (setup_live_bytes_from_ref, compute_trims): Likewise.
33551 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33552 (maybe_trim_partially_dead_store): Likewise.
33553 (maybe_trim_complex_store): Likewise.
33554 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33555 Track what bytes live from the original store. Return tri-state
33556 for dead, partially dead or live.
33557 (dse_dom_walker): Add constructor, destructor and new private members.
33558 (delete_dead_call, delete_dead_assignment): New extracted from
33560 (dse_optimize_stmt): Make a member of dse_dom_walker.
33561 Use initialize_ao_ref_for_dse.
33563 PR tree-optimization/33562
33564 PR tree-optimization/61912
33565 PR tree-optimization/77485
33566 * sbitmap.h (bitmap_count_bits): Prototype.
33567 (bitmap_clear_range, bitmap_set_range): Likewise.
33568 * sbitmap.c (bitmap_clear_range): New function.
33569 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33571 2017-01-13 Martin Liska <mliska@suse.cz>
33574 * function.c (set_cfun): Add new argument force.
33575 * function.h (set_cfun): Likewise.
33576 * ipa-inline-transform.c (inline_call): Use the function when
33577 strict alising from is dropped for function we inline to.
33579 2017-01-13 Richard Biener <rguenther@suse.de>
33581 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33582 for dumping GIMPLE INTEGER_CSTs.
33584 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33586 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33587 to 201112L since C++17.
33589 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33592 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33593 if -fsanitize=kernel-address is present.
33595 2017-01-13 Richard Biener <rguenther@suse.de>
33597 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33598 as _Literal ( type ) number in case usual suffixes do not
33599 preserve all information.
33601 2017-01-13 Richard Biener <rguenther@suse.de>
33603 PR tree-optimization/77283
33604 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33605 and ssa-iterators.h.
33606 (is_feasible_trace): Implement a cost model based on joiner
33609 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33612 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33613 char or short to __float128/_Float128 directly.
33615 2017-01-12 Martin Sebor <msebor@redhat.com>
33617 to -Wformat-overflow.
33618 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33619 (min_bytes_remaining): Same.
33620 (get_string_length): Same.
33621 (format_string): Same.
33622 (format_directive): Same.
33624 (pass_sprintf_length::handle_gimple_call): Same.
33626 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33628 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33629 info.nowrite calls with no lhs that can't throw. Return bool
33630 whether gsi_remove has been called or not.
33631 (pass_sprintf_length::handle_gimple_call): Return bool whether
33632 try_substitute_return_value called gsi_remove. Formatting fix.
33633 (pass_sprintf_length::execute): Don't use gsi_remove if
33634 handle_gimple_call returned true.
33637 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33638 be removed due to side-effects, don't remove following barrier nor
33639 turn the successor edge into fallthru edge.
33641 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33644 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33645 element-reversing loads and stores as not swappable.
33647 2017-01-12 Nathan Sidwell <nathan@acm.org>
33648 Nicolai Stange <nicstange@gmail.com>
33650 * combine.c (try_combine): Don't ignore result of overlap checking
33651 loop. Combine overlap & asm check into single loop.
33653 2017-01-12 Richard Biener <rguenther@suse.de>
33655 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33656 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33658 2017-01-12 Richard Biener <rguenther@suse.de>
33660 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33661 and TS_TARGET_OPTION directly derive from TS_BASE.
33662 * tree-core.h (tree_optimization_option): Derive from tree_base.
33663 (tree_target_option): Likewise.
33665 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33667 * config/i386/i386.c (memory_address_length): Increase len
33668 only when rip_relative_addr_p returns false.
33670 2017-01-11 Julia Koval <julia.koval@intel.com>
33672 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33673 (OPTION_MASK_ISA_SGX_SET): New.
33674 (ix86_handle_option): Handle OPT_msgx.
33675 * config.gcc: Added sgxintrin.h.
33676 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33677 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33678 * config/i386/i386.c (ix86_target_string): Add -msgx.
33680 (ix86_option_override_internal): Handle new options.
33681 (ix86_valid_target_attribute_inner_p): Add sgx.
33682 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33683 * config/i386/i386.opt: Add msgx.
33684 * config/i386/sgxintrin.h: New file.
33685 * config/i386/x86intrin.h: Add sgxintrin.h.
33687 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33690 * fold-const.c (maybe_nonzero_address): Return 1 for function
33692 (tree_single_nonzero_warnv_p): Don't handle function local objects
33696 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33699 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33702 * opts.c: Include "spellcheck.h"
33703 (struct string_fragment): New struct.
33704 (struct edit_distance_traits<const string_fragment &>): New
33706 (get_closest_sanitizer_option): New function.
33707 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33709 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33711 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33713 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33714 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33715 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33716 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33717 for initial die_offset if dwarf_split_debug_info.
33718 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33719 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33721 (output_skeleton_debug_sections): Formatting fix. Use
33722 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33723 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33725 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33727 * config/arm/cortex-a53.md: Add bypasses for
33728 cortex_a53_r2f_cvt.
33729 (cortex_a53_r2f): Only use for transfers.
33730 (cortex_a53_f2r): Likewise.
33731 (cortex_a53_r2f_cvt): Add reservation for conversions.
33732 (cortex_a53_f2r_cvt): Likewise.
33734 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33736 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33737 to all inlined functions, change static to extern.
33739 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33742 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33744 (arm_assemble_integer): Likewise.
33746 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33748 * config.gcc: Use new awk script to check CPU, FPU and architecture
33749 parameters for --with-... options.
33750 * config/arm/parsecpu.awk: New file
33751 * config/arm/arm-cpus.in: New file.
33752 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33754 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33756 * config/arm/t-arm: Update dependency rules.
33757 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33758 of processing .def files.
33759 * config/arm/genopt.sh: Deleted.
33760 * config/arm/gentune.sh: Deleted.
33761 * config/arm/arm-cores.def: Deleted.
33762 * config/arm/arm-arches.def: Deleted.
33763 * config/arm/arm-fpus.def: Deleted.
33764 * config/arm/arm-tune.md: Regenerated.
33765 * config/arm/arm-tables.opt: Regenerated.
33766 * config/arm/arm-cpu.h: New generated file.
33767 * config/arm/arm-cpu-data.h: New generated file.
33768 * config/arm/arm-cpu-cdata.h: New generated file.
33770 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33773 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33775 (input_varpool_node): Unpack dynamically_initialized bit.
33777 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33779 PR rtl-optimization/79032
33780 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33781 the alignment of the adjusted memory reference against that of MODE,
33782 instead of the alignment of the original memory reference.
33784 2017-01-11 Martin Jambor <mjambor@suse.cz>
33786 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33788 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33789 decorated functions.
33791 2017-01-11 Richard Biener <rguenther@suse.de>
33793 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33794 set range/nonnull info for PHI results. Do not set it on
33795 stmts marked for removal.
33797 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33799 * expr.c (store_field): In the bitfield case, fetch the return value
33800 from the registers before applying a single big-endian adjustment.
33801 Always do a final load for a BLKmode value not larger than a word.
33803 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33806 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33807 that we correctly handle column numbers greater than
33808 LINE_MAP_MAX_COLUMN_NUMBER.
33810 2017-01-10 Martin Sebor <msebor@redhat.com>
33812 PR middle-end/78245
33813 * gimple-ssa-sprintf.c (get_destination_size): Call
33814 {init,fini}object_sizes.
33815 * tree-object-size.c (addr_object_size): Adjust.
33816 (pass_through_call): Adjust.
33817 (pass_object_sizes::execute): Adjust.
33818 * tree-object-size.h (fini_object_sizes): Declare.
33820 2017-01-10 Martin Sebor <msebor@redhat.com>
33822 PR tree-optimization/78775
33823 * builtins.c (get_size_range): Move...
33824 * calls.c: ...to here.
33825 (alloc_max_size): Accept zero argument.
33826 (operand_signed_p): Remove.
33827 (maybe_warn_alloc_args_overflow): Call get_size_range.
33828 * calls.h (get_size_range): Declare.
33830 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33832 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33833 from TI's devices.csv file as of September 2016.
33834 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33836 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33838 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33839 * doc/invoke.texi: Likewise.
33840 * doc/md.texi: Likewise.
33841 * doc/objc.texi: Likewise.
33843 2017-01-10 Joshua Conner <joshconner@google.com>
33845 * config/arm/fuchsia-elf.h: New file.
33846 * config/fuchsia.h: New file.
33847 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33848 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33850 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33852 2016-01-10 Richard Biener <rguenther@suse.de>
33854 PR tree-optimization/79034
33855 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33856 Propagate out degenerate PHIs in the joiner.
33858 2017-01-10 Martin Liska <mliska@suse.cz>
33860 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33861 (sort_congruence_classes_by_decl_uid): Likewise.
33862 (sort_congruence_class_groups_by_decl_uid): Likewise.
33863 (sem_item_optimizer::merge_classes): Sort class, groups in these
33864 classes and members in the groups by DECL_UID of declarations.
33865 This would make merge operations stable.
33867 2017-01-10 Martin Liska <mliska@suse.cz>
33869 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33870 usage of m_classes_vec.
33871 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33872 (sem_item_optimizer::get_group_by_hash): Likewise.
33873 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33874 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33875 (sem_item_optimizer::verify_classes): Likewise.
33876 (sem_item_optimizer::process_cong_reduction): Likewise.
33877 (sem_item_optimizer::dump_cong_classes): Likewise.
33878 (sem_item_optimizer::merge_classes): Likewise.
33879 * ipa-icf.h (congruence_class_hash): Rename from
33880 congruence_class_group_hash. Remove declaration of m_classes_vec.
33882 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33884 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33885 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33886 * config.gcc: Add avx512vpopcntdqintrin.h.
33887 * config/i386/avx512vpopcntdqintrin.h: New.
33888 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33889 * config/i386/i386-builtin-types.def: Add new types.
33890 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33891 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33892 __builtin_ia32_vpopcountq_v8di_mask): New.
33893 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33894 __AVX512VPOPCNTDQ__.
33895 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33896 (PTA_AVX512VPOPCNTDQ): Define.
33897 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33898 TARGET_AVX512VPOPCNTDQ_P): Define.
33899 * config/i386/i386.opt: Add mavx512vpopcntdq.
33900 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33901 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33903 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33905 PR middle-end/77484
33906 * predict.def (PRED_CALL): Set to 67.
33908 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33910 * expr.c (store_field): In the bitfield case, if the value comes from
33911 a function call and is of an aggregate type returned in registers, do
33912 not modify the field mode; extract the value in all cases if the mode
33913 is BLKmode and the size is not larger than a word.
33915 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33918 * config/i386/cpuid.h: Fix undefined behavior.
33920 2017-01-04 Jeff Law <law@redhat.com>
33922 PR tree-optimization/79007
33923 PR tree-optimization/67955
33924 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33925 conservative for pt.null when flag_non_call_exceptions is on.
33927 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33929 PR translation/79019
33930 PR translation/79020
33931 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33932 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33933 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33935 * config/avr/avr.opt (maccumulate-args): Likewise.
33936 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33937 * common.opt (freport-bug): Likewise.
33938 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33939 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33940 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33941 translatable string.
33942 * config/i386/i386.c (function_value_32): Likewise.
33943 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33944 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33946 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33947 * common/config/msp430/msp430-common.c (msp430_handle_option):
33949 * symtab.c (symtab_node::verify_base): Likewise.
33950 * opts.c (set_debug_level): Likewise.
33951 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33952 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33953 missing whitespace to translatable strings.
33954 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33955 * config/sh/superh.h: Likewise.
33956 * config/i386/xopintrin.h: Likewise.
33957 * config/i386/znver1.md: Likewise.
33958 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33959 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33960 * double-int.h (struct double_int): Likewise.
33961 * double-int.c (div_and_round_double): Likewise.
33962 * wide-int.cc: Likewise.
33963 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33964 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33965 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33966 (crossjumps_occurred): ... this.
33967 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33970 PR tree-optimization/78899
33971 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33972 returning bool return struct loop *, NULL for failure and the new
33974 (versionable_outer_loop_p): Don't version outer loop if it has
33975 dont_vectorized bit set.
33976 (tree_if_conversion): When versioning outer loop, ensure
33977 tree_if_conversion is performed also on the inner loop of the
33978 non-vectorizable outer loop copy.
33979 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33980 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33981 prevent vectorization of it.
33982 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33983 the outer loop vectorization of the non-scalar version is attempted
33984 before vectorization of the inner loop in scalar version. If
33985 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33986 vectorization of its inner loop.
33987 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33988 has 2 inner loops, rename also on edges from bb whose single pred
33989 is outer_loop->header. Fix typo in function comment.
33991 2017-01-09 Martin Sebor <msebor@redhat.com>
33994 * asan.c (asan_emit_stack_protection): Increase local buffer size
33995 to avoid snprintf truncation warning.
33997 2017-01-09 Andrew Pinski <apinski@cavium.com>
33999 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34000 to reference thunderx2t99 for the tuning structure
34001 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34003 (thunderx2t99_extra_costs): This.
34004 * config/aarch64/aarch64-tune.md: Regenerate.
34005 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34006 (vulcan_addrcost_table): This.
34007 (vulcan_regmove_cost): Rename to ...
34008 (thunderx2t99_regmove_cost): This.
34009 (vulcan_vector_cost): Rename to ...
34010 (thunderx2t99_vector_cost): this.
34011 (vulcan_branch_cost): Rename to ...
34012 (thunderx2t99_branch_cost): This.
34013 (vulcan_tunings): Rename to ...
34014 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34015 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34017 2017-01-09 Martin Jambor <mjambor@suse.cz>
34021 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34022 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34023 (propagate_vr_accross_jump_function): Use the above function for all
34024 value range computations for pass-through jump functions and type
34025 converasion from explicit value range values.
34026 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34027 parameters from TYPE_ARG_TYPES.
34028 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34029 (ipa_write_node_info): Stream type of the actual argument.
34030 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34032 2017-01-09 Martin Liska <mliska@suse.cz>
34035 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34036 (lookup_compiler): Do not show error message with have_E.
34038 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34040 PR tree-optimization/78938
34041 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34042 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34043 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34044 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34047 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34049 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34052 2017-01-09 Richard Biener <rguenther@suse.de>
34054 PR tree-optimization/78997
34055 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34056 name condition properly.
34058 2017-01-09 Richard Biener <rguenther@suse.de>
34061 * dwarf2out.c (is_cxx): New overload with context.
34062 (is_naming_typedef_decl): Use it.
34064 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34066 * invoke.texi (Option Summary): Correct spacing in option lists
34067 and add line breaks to fix over-long lines.
34069 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34071 PR middle-end/17660
34073 * extend.texi (Common Variable Attributes): Add xref to GCC
34074 Internals manual to explain mode attribute keywords.
34076 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34079 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34080 and Preprocessor Options.
34081 (Options for Linking): Document -pthread here....
34082 (RS/6000 and PowerPC Options): ...not here.
34083 (Solaris 2 Options): ...or here.
34084 * doc/cppopts.texi: Document -pthread.
34086 2017-01-08 Martin Sebor <msebor@redhat.com>
34088 PR middle-end/77708
34089 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34090 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34091 New member functions.
34092 (format_directive): Used them.
34094 (pass_sprintf_length::handle_gimple_call): Same.
34095 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34096 to avoid truncation for any argument.
34097 (extract_affine_mul): Same.
34098 * tree.c (get_file_function_name): Same.
34100 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34102 PR middle-end/77484
34103 * predict.def (PRED_INDIR_CALL): Set to 86.
34105 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34107 PR preprocessor/54124
34108 * doc/cppopts.texi: Reformat -d subtable to list the full name
34109 of the options. Add cross-reference to the docs for the general
34110 compiler -d options.
34111 * doc/invoke.texi (Developer Options): Add cross-reference to the
34112 preprocessor-specific -d option documentation.
34114 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34116 PR preprocessor/13498
34117 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34118 redudant material, and reflect new command-line options.
34119 (System Headers): Likewise.
34121 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34123 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34124 -isystem, and -idirafter. Copy-edit.
34125 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34126 default for -ftrack-macro-expansion. Delete obsolete and
34127 badly-formatted implementation details about -fdebug-cpp output.
34128 * doc/cppwarnopts.texi: Copy-edit.
34130 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34133 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34134 that the transition from a max line width >= 1<<10 to narrower
34135 lines works correctly.
34137 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34139 * doc/options.texi (PerFunction): New.
34140 * opt-functions.awk (switch_flags): Map both Optimization and
34141 PerFunction to CL_OPTIMIZATION.
34142 * opth-gen.awk: Test for PerFunction flag along with
34144 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34145 it only when the latter is present. Skip those that don't in
34146 the hash function generator.
34147 * common.opt (fvar-tracking): Mark as PerFunction instead of
34149 (fvar-tracking-assignments): Likewise.
34150 (fvar-tracking-assignments-toggle): Likewise.
34151 (fvar-tracking-uninit): Likewise.
34153 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34155 PR translation/79018
34156 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34159 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34162 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34163 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34164 TARGET_LONG_JUMP_TABLE_OFFSETS.
34165 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34166 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34167 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34168 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34169 * config/m68k/m68k.md (tablejump expander): Likewise.
34170 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34171 TARGET_LONG_JUMP_TABLE_OFFSETS.
34172 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34173 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34175 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34176 David Holsgrove <david.holsgrove@xilinx.com>
34178 * common/config/microblaze/microblaze-common.c
34179 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34180 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34182 * config/microblaze/microblaze.c (microblaze_must_save_register)
34183 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34185 (microblaze_eh_return): New function.
34186 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34187 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34188 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34189 * config/microblaze/microblaze.md (eh_return): New pattern.
34191 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34193 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34194 GCC_DIAGNOSTIC_STRINGIFY): Define.
34196 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34198 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34200 * config/arm/arm.md (<mcrr>): New.
34202 * config/arm/arm.c (arm_arch5te): New.
34203 (arm_option_override): Set arm_arch5te.
34204 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34206 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34207 (arm_mcrr_qualifiers): ... this. New.
34208 (MRRC_QUALIFIERS): Define to...
34209 (arm_mrrc_qualifiers): ... this. New.
34210 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34212 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34213 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34214 (MRRCI, mrrc, MRRC): New.
34215 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34216 VUNSPEC_MRRC2): New.
34218 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34220 * config/arm/arm.md (<mcr>): New.
34222 * config/arm/arm.c (arm_coproc_builtin_available): Add
34223 support for mcr, mrc, mcr2 and mrc2.
34224 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34225 (arm_mcr_qualifiers): ... this. New.
34226 (MRC_QUALIFIERS): Define to ...
34227 (arm_mrc_qualifiers): ... this. New.
34228 (MCR_QUALIFIERS): Define to ...
34229 (arm_mcr_qualifiers): ... this. New.
34230 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34232 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34233 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34234 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34235 VUNSPEC_MRC2): New.
34237 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34239 * config/arm/arm.md (*ldc): New.
34243 * config/arm/arm.c (arm_coproc_builtin_available): Add
34244 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34245 (arm_coproc_ldc_stc_legitimate_address): New.
34246 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34247 'qualifier_const_pointer'.
34248 (LDC_QUALIFIERS): Define to...
34249 (arm_ldc_qualifiers): ... this. New.
34250 (STC_QUALIFIERS): Define to...
34251 (arm_stc_qualifiers): ... this. New.
34252 * config/arm/arm-protos.h
34253 (arm_coproc_ldc_stc_legitimate_address): New.
34254 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34255 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34256 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34257 stc2, stcl, stc2l): New.
34258 * config/arm/constraints.md (Uz): New.
34259 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34260 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34261 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34262 VUNSPEC_STC2L): New.
34264 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34266 * config/arm/arm.md (<cdp>): New.
34267 * config/arm/arm.c (neon_const_bounds): Rename this ...
34268 (arm_const_bounds): ... this.
34269 (arm_coproc_builtin_available): New.
34270 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34271 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34272 (CDP_QUALIFIERS): Define to...
34273 (arm_cdp_qualifiers): ... this. New.
34275 (arm_expand_builtin_args): Add case for 6 arguments.
34276 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34277 (arm_const_bounds): ... this.
34278 (arm_coproc_builtin_available): New.
34279 * config/arm/arm_acle.h (__arm_cdp): New.
34281 * config/arm/arm_acle_builtins.def (cdp): New.
34283 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34284 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34285 'arm_const_bounds'.
34286 * config/arm/types.md (coproc): New.
34287 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34288 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34289 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34290 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34292 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34294 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34295 (UBINOP_QUALIFIERS): New.
34297 (acle_builtin_data): New. Change comment.
34298 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34299 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34300 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34301 arm_acle_builtins.def.
34302 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34303 (arm_init_acle_builtins): New.
34304 (CRC32_BUILTIN): Remove.
34305 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34306 crc32cb, crc32ch and crc32cw.
34307 (arm_init_crc32_builtins): Remove.
34308 (arm_init_builtins): Use arm_init_acle_builtins rather
34309 than arm_init_crc32_builtins.
34310 (arm_expand_acle_builtin): New.
34311 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34312 * config/arm/arm_acle_builtins.def: New.
34314 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34316 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34317 (arm_builtin_datum): ... this.
34318 (arm_init_neon_builtin): Rename to ...
34319 (arm_init_builtin): ... this. Add a new parameters PREFIX
34320 and USE_SIG_IN_NAME.
34321 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34322 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34323 'arm_builtin_datum'.
34324 (arm_init_vfp_builtins): Likewise.
34325 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34326 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34327 (arm_expand_neon_args): Rename to ...
34328 (arm_expand_builtin_args): ... this. Rename builtin_arg
34329 enum values and differentiate between ARG_BUILTIN_MEMORY
34330 and ARG_BUILTIN_NEON_MEMORY.
34331 (arm_expand_neon_builtin_1): Rename to ...
34332 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34333 values, arm_expand_builtin_args and add bool parameter NEON.
34334 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34335 (arm_expand_vfp_builtin): Likewise.
34336 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34338 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34340 PR middle-end/77484
34341 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34342 * predict.c (tree_estimate_probability_bb): Reverse direction of
34343 polymorphic call predictor.
34345 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34347 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34348 (determine_pass_name_match): ...this new function and...
34349 (should_skip_pass_p): ...this new function.
34351 2017-01-06 Nathan Sidwell <nathan@acm.org>
34353 * ipa-visibility.c (function_and_variable_visibility): Reformat
34354 comments and long lines. Remove extrneous if.
34355 * symtab.c (symtab_node::make_decl_local): Fix code format.
34356 (symtab_node::set_section_for_node): Fix comment typo.
34358 2017-01-06 Martin Liska <mliska@suse.cz>
34361 * lra-constraints.c: Rename invariant to lra_invariant.
34362 * predict.c (set_even_probabilities): Initialize e to NULL.
34364 2017-01-05 Martin Sebor <msebor@redhat.com>
34366 PR tree-optimization/78910
34367 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34368 (format_integer): Correct off-by-one error in the handling
34369 of precision with negative numbers in signed conversions..
34371 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34373 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34375 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34377 PR tree-optimization/71016
34378 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34379 factor_out_conditional_conversion. Formatting fix.
34380 (factor_out_conditional_conversion): Add cond_stmt argument.
34381 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34382 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34385 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34387 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34388 read-rtl-function.o, and selftest-rtl.o.
34389 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34390 (selftest::aarch64_test_loading_full_dump): New function.
34391 (selftest::aarch64_run_selftests): New function.
34392 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34393 selftest::aarch64_run_selftests.
34394 * config/i386/i386.c
34395 (selftest::ix86_test_loading_dump_fragment_1): New function.
34396 (selftest::ix86_test_loading_call_insn): New function.
34397 (selftest::ix86_test_loading_full_dump): New function.
34398 (selftest::ix86_test_loading_unspec): New function.
34399 (selftest::ix86_run_selftests): Call the new functions.
34400 * emit-rtl.c (maybe_set_max_label_num): New function.
34401 * emit-rtl.h (maybe_set_max_label_num): New decl.
34402 * function.c (instantiate_decls): Guard call to
34403 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34404 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34406 * gensupport.c (gen_reader::gen_reader): Pass "false"
34407 for new "compact" param of rtx_reader.
34408 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34409 rather than an empty string for NULL strings.
34410 * read-md.c: Potentially include config.h rather than bconfig.h.
34411 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34412 (have_error): New global, copied from errors.c.
34413 (md_reader::read_name): Rename to...
34414 (md_reader::read_name_1): ...this, adding "out_loc" param,
34415 and converting "missing name or number" to returning false, rather
34417 (md_reader::read_name): Reimplement in terms of read_name_1.
34418 (md_reader::read_name_or_nil): New function.
34419 (md_reader::read_string): Handle "(nil)" by returning NULL.
34420 (md_reader::md_reader): Add new param "compact".
34421 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34422 (md_reader::read_file): New method.
34423 * read-md.h (md_reader::md_reader): Add new param "compact".
34424 (md_reader::read_file): New method.
34425 (md_reader::is_compact): New accessor.
34426 (md_reader::read_name): Convert return type from void to file_location.
34427 (md_reader::read_name_or_nil): New decl.
34428 (md_reader::read_name_1): New decl.
34429 (md_reader::m_compact): New field.
34430 (noop_reader::noop_reader): Pass "false" for new "compact" param
34432 (rtx_reader::rtx_reader): Add new "compact" param.
34433 (rtx_reader::read_rtx_operand): Make virtual and convert return
34434 type from void to rtx.
34435 (rtx_reader::read_until): New decl.
34436 (rtx_reader::handle_any_trailing_information): New virtual function.
34437 (rtx_reader::postprocess): New virtual function.
34438 (rtx_reader::finalize_string): New virtual function.
34439 (rtx_reader::m_in_call_function_usage): New field.
34440 (rtx_reader::m_reuse_rtx_by_id): New field.
34441 * read-rtl-function.c: New file.
34442 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34443 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34444 (selftest::verify_three_block_rtl_cfg): New decl.
34445 * read-rtl-function.h: New file.
34446 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34447 For host, include function.h, memmodel.h, and emit-rtl.h.
34448 (one_time_initialization): New function.
34449 (struct compact_insn_name): New struct.
34450 (compact_insn_names): New array.
34451 (find_code): Handle insn codes in compact dumps.
34452 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34453 (bind_subst_iter_and_attr): Likewise.
34454 (add_condition_to_string): Likewise.
34455 (add_condition_to_rtx): Likewise.
34456 (apply_attribute_uses): Likewise.
34457 (add_current_iterators): Likewise.
34458 (apply_iterators): Likewise.
34459 (initialize_iterators): Guard usage of apply_subst_iterator with
34460 #ifdef GENERATOR_FILE.
34461 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34462 (md_reader::read_mapping): Likewise.
34463 (add_define_attr_for_define_subst): Likewise.
34464 (add_define_subst_attr): Likewise.
34465 (read_subst_mapping): Likewise.
34466 (check_code_iterator): Likewise.
34467 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34469 (one_time_initialization): New function.
34470 (rtx_reader::read_until): New method.
34471 (read_flags): New function.
34472 (parse_reg_note_name): New function.
34473 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34474 Handle reuse_rtx ids.
34475 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34476 Add parsing support for RTL dumps, mirroring the special-cases in
34477 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34478 values, and calling handle_any_trailing_information.
34479 (rtx_reader::read_rtx_operand): Convert return type from void
34480 to rtx, returning return_rtx. Handle case 'e'. Call
34481 finalize_string on XSTR and XTMPL fields.
34482 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34483 "(nil)" values were omitted. Call the postprocess vfunc on the
34485 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34486 class ctor. Initialize m_in_call_function_usage. Call
34487 one_time_initialization.
34488 * rtl-tests.c (selftest::test_uncond_jump): Call
34489 set_new_first_and_last_insn.
34490 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34491 * selftest-rtl.c: New file.
34492 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34493 (selftest::get_insn_by_uid): New decl.
34494 * selftest-run-tests.c (selftest::run_tests): Call
34495 read_rtl_function_c_tests.
34496 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34497 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34500 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34502 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34503 operands in a special way. Assert that pos+len <= mode precision.
34505 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34507 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34508 3 argument Alias with unlimited for the negative form.
34509 (fno-vect-cost-model): Removed.
34511 2017-01-05 Martin Liska <mliska@suse.cz>
34513 * hsa-gen.c (gen_hsa_divmod): New function.
34514 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34516 2017-01-05 Martin Liska <mliska@suse.cz>
34519 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34522 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34524 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34525 small constant length operands.
34527 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34529 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34530 between loop iterations.
34532 2017-01-05 Martin Liska <mliska@suse.cz>
34535 * gimplify.c (gimplify_decl_expr): Compare to
34536 asan_poisoned_variables instread of checking flags.
34537 (gimplify_target_expr): Likewise.
34538 (gimplify_expr): Likewise.
34539 (gimplify_function_tree): Conditionally initialize
34540 asan_poisoned_variables.
34542 2017-01-04 Jeff Law <law@redhat.com>
34544 PR tree-optimizatin/78812
34545 * rtl.h (contains_mem_rtx_p): Prototype.
34546 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34547 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34548 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34549 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34550 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34552 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34554 * input.c (assert_char_at_range): Default-initialize actual_range.
34556 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34558 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34559 to match the caller.
34561 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34563 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34564 insns after final jump in test to emit dummy move.
34566 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34568 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34569 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34571 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34573 * multiple_target.c (create_dispatcher_calls): Init e_next.
34574 * tree-ssa-loop-split.c (split_loop): Init border.
34575 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34578 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34583 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34584 (altivec_register_operand): Do not return true if the operand
34585 contains a SUBREG mixing SImode and SFmode.
34586 (vsx_register_operand): Likewise.
34587 (vsx_reg_sfsubreg_ok): New predicate.
34588 (vfloat_operand): Do not return true if the operand contains a
34589 SUBREG mixing SImode and SFmode.
34590 (vint_operand): Likewise.
34591 (vlogical_operand): Likewise.
34592 (gpc_reg_operand): Likewise.
34593 (int_reg_operand): Likewise.
34594 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34595 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34596 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34598 (rs6000_emit_move_si_sf_subreg): New helper function.
34599 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34600 fixup SUBREGs involving SImode and SFmode.
34601 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34602 numbers for the new peephole2 optimization.
34603 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34604 the GLIBC math library that do AND/IOR/XOR operations on single
34605 precision floating point.
34606 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34607 target macros to say whether we need to avoid SUBREGs mixing
34609 (TARGET_ALLOW_SF_SUBREG): Likewise.
34610 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34611 (UNSPEC_SI_FROM_SF): Likewise.
34612 (iorxor): Change spacing.
34613 (and_ior_xor): New iterator for AND, IOR, and XOR.
34614 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34615 (movdi_from_sf_zero_ext): Likewise.
34616 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34617 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34618 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34619 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34620 (fms<mode>4): Likewise.
34621 (fnma<mode>4): Likewise.
34622 (fnms<mode>4): Likewise.
34623 (nfma<mode>4): Likewise.
34624 (nfms<mode>4): Likewise.
34626 2017-01-04 Marek Polacek <polacek@redhat.com>
34629 * doc/invoke.texi: Document -Wpointer-compare.
34631 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34633 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34636 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34637 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34640 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34643 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34644 documentation of the powerpc_popcntb_ok attribute.
34645 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34646 code to issue warning messages if a requested CPU configuration is
34647 not supported by the binary (assembler and loader) toolchain.
34648 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34649 made to define a built-in function that has been disabled.
34650 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34651 made to define a built-in function that has been disabled.
34652 (altivec_init_builtins): Add comment explaining why definition
34653 of the DST built-in functions is not preceded by an assertion
34654 check. Add assertions to prevent ICE if attempts are made to
34655 define an altivec predicate or an abs* built-in function that has
34657 (htm_init_builtins): Add comment explaining why definition of the
34658 htm built-in functions is not preceded by an assertion check.
34660 2017-01-04 Jeff Law <law@redhat.com>
34662 PR tree-optimizatin/67955
34663 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34664 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34665 the points-to solution does not include pt_null. Use DECL_PT_UID
34668 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34670 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34671 Use gen_int_mode instead of gen_lopwart for const_int operands.
34673 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34675 PR tree-optimization/71563
34676 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34677 out of range value - has low bits known to be zero.
34679 2017-01-04 Alan Modra <amodra@gmail.com>
34681 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34682 * configure: Regenerate.
34683 * config.in: Regenerate.
34685 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34688 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34689 a substring of the message, but strcmp with the whole message. Ifdef
34690 ENABLE_NLS, translate the message first using dgettext.
34692 2017-01-03 Jeff Law <law@redhat.com>
34694 PR tree-optimizatin/78856
34695 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34696 (mark_threaded_blocks): Remove code to truncate thread paths that
34697 cross multiple loop headers. Instead invalidate the cached loop
34698 iteration information and handle case of a thread path walking
34699 into an irreducible region.
34701 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34704 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34705 assertions. Add support for doing the signbit if the IEEE 128-bit
34706 floating point value is in a GPR.
34707 * config/rs6000/rs6000.md (Fsignbit): Delete.
34708 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34709 Update the length attribute if the value is in a GPR.
34710 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34711 the sign or zero extension instruction, since the value is always 0/1.
34712 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34715 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34716 extracting SImode to a GPR register so that we can generate a
34717 store, limit the vector to be in a traditional Altivec register
34718 for the vextuwrx instruction.
34720 2017-01-03 Ian Lance Taylor <iant@google.com>
34722 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34724 2017-01-03 Martin Sebor <msebor@redhat.com>
34726 PR tree-optimization/78696
34727 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34728 precision. Use MPFR for %f for greater fidelity. Correct handling
34730 (pass_sprintf_length::compute_format_length): Set width and precision
34731 specified by asrerisk to void_node for vararg functions.
34732 (try_substitute_return_value): Adjust dump output.
34734 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34736 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34738 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34740 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34741 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34742 -mlra/-mno-lra was passed to the compiler.
34744 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34746 PR rtl-optimization/65618
34747 * emit-rtl.c (try_split): Move initialization of "before" and
34748 "after" to just before the call to emit_insn_after_setloc.
34750 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34752 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34754 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34756 * dwarf2out.c (gen_enumeration_type_die): When
34757 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34759 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34761 PR tree-optimization/78965
34762 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34763 Change first argument from const call_info & to call_info &. For %n
34764 set info.nowrite to false.
34766 PR middle-end/78901
34767 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34768 possibly throwing calls.
34770 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34771 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34772 and fns handling, rather than in a separate case SSA_NAME.
34774 2017-01-02 Jeff Law <law@redhat.com>
34776 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34777 fixes for first_period and second_period variables.
34779 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34782 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34783 (*insvqi_1): New insn pattern.
34784 (*insvqi_1_mem_rex64): Ditto.
34785 (*insvqi_2): Ditto.
34786 (*insvqi_3): Rename from *insvqi.
34788 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34790 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34792 * doc/cfg.texi (Edges): Remove reference to Java.
34793 (Maintaining the CFG): Ditto.
34795 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34797 PR middle-end/77674
34798 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34799 transparent aliases.
34801 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34803 PR middle-end/77484
34804 * predict.def (PRED_CALL): Update hitrate.
34805 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34806 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34807 into direct/indirect/polymorphic variants.
34809 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34811 Update copyright years.
34813 * gcc.c (process_command): Update copyright notice dates.
34814 * gcov-dump.c (print_version): Ditto.
34815 * gcov.c (print_version): Ditto.
34816 * gcov-tool.c (print_version): Ditto.
34817 * gengtype.c (create_file): Ditto.
34818 * doc/cpp.texi: Bump @copying's copyright year.
34819 * doc/cppinternals.texi: Ditto.
34820 * doc/gcc.texi: Ditto.
34821 * doc/gccint.texi: Ditto.
34822 * doc/gcov.texi: Ditto.
34823 * doc/install.texi: Ditto.
34824 * doc/invoke.texi: Ditto.
34826 Copyright (C) 2017 Free Software Foundation, Inc.
34828 Copying and distribution of this file, with or without modification,
34829 are permitted in any medium without royalty provided the copyright
34830 notice and this notice are preserved.