1 2018-09-21 David Malcolm <dmalcolm@redhat.com>
3 PR tree-optimization/87309
4 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
5 calls with pflags and alt_flags.
6 (selftest::test_capture_of_dump_calls): Add test of interaction of
7 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
9 2018-09-21 Olivier Hainque <hainque@adacore.com>
11 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
13 2018-09-21 Olivier Hainque <hainque@adacore.com>
15 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
17 2018-09-21 Olivier Hainque <hainque@adacore.com>
19 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
20 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
21 (VXWORKS_LIBS_RTP): Minor reordering.
23 2018-09-21 Olivier Hainque <hainque@adacore.com>
25 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
26 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
28 2018-09-21 Olivier Hainque <hainque@adacore.com>
30 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
31 (PTRDIFF_TYPE): Likewise.
33 2018-09-21 Olivier Hainque <hainque@adacore.com>
35 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
36 triplet, similar to support for VxWorks7.
37 * config/vxworks-dummy.h: Provide a default definition
38 of TARGET_VXWORKS64 to 0.
40 2018-09-21 Olivier Hainque <hainque@adacore.com>
42 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
43 * config/vxworks-dummy.h: here.
45 2018-09-21 Olivier Hainque <hainque@adacore.com>
47 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
49 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
50 Bo Zhou <zbo.zhou@hisilicon.com>
52 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
53 * config/aarch64/aarch64-tune.md: Regenerated.
54 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
55 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
56 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
58 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
59 Julian Brown <julian@codesourcery.com>
61 * builtins.c (get_builtin_sync_mem): Handle address spaces.
63 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
65 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
66 if the call takes a static chain.
68 2018-09-21 Martin Liska <mliska@suse.cz>
70 * auto-profile.c (autofdo_source_profile::read): Do not
72 (read_profile): Do not add working sets.
73 (read_autofdo_file): Remove sum_all.
74 (afdo_callsite_hot_enough_for_early_inline): Remove const
76 * coverage.c (struct counts_entry): Remove gcov_summary.
77 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
78 do not support GCOV_TAG_PROGRAM_SUMMARY.
79 (get_coverage_counts): Remove summary and expected
81 * coverage.h (get_coverage_counts): Likewise.
82 * doc/gcov-dump.texi: Remove -w option.
83 * gcov-dump.c (dump_working_sets): Remove.
84 (main): Do not support '-w' option.
85 (print_usage): Likewise.
86 (tag_summary): Likewise.
87 * gcov-io.c (gcov_write_summary): Do not dump
89 (gcov_read_summary): Likewise.
90 (gcov_histo_index): Remove.
91 (gcov_histogram_merge): Likewise.
92 (compute_working_sets): Likewise.
93 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
95 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
96 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
97 (GCOV_HISTOGRAM_SIZE): Remove.
98 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
99 (struct gcov_summary): Simplify rapidly just
100 to runs and sum_max fields.
101 (gcov_histo_index): Remove.
102 (NUM_GCOV_WORKING_SETS): Likewise.
103 (compute_working_sets): Likewise.
104 * gcov-tool.c (print_overlap_usage_message): Remove
106 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
107 (output_lines): Remove program related line.
108 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
109 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
111 (input_profile_summary): Do not read it.
112 (merge_profile_summaries): And do not merge it.
113 (input_symtab): Do not call removed function.
114 * modulo-sched.c (sms_schedule): Do not print sum_max.
115 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
116 removed when histogram method was invented.
117 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
119 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
121 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
122 and dump selected value.
123 * profile.c (add_working_set): Remove.
124 (get_working_sets): Likewise.
125 (find_working_set): Likewise.
126 (get_exec_counts): Do not work with working sets.
127 (read_profile_edge_counts): Do not inform as sum_max is removed.
128 (compute_branch_probabilities): Likewise.
129 (compute_value_histograms): Remove argument for call of
131 * profile.h: Do not make gcov_summary const.
133 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
135 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
137 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
139 PR tree-optimization/86990
140 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
141 Check that the entire merged store group is made of constants only for
144 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
146 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
147 (VTABLE_VERIFICATION_SPEC): Likewise.
148 (SANITIZER_EARLY_SPEC): Likewise.
149 (SANITIZER_SPEC): Likewise.
150 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
151 * doc/invoke.texi (Link Options): Document -r.
153 2018-09-20 Richard Biener <rguenther@suse.de>
156 * gimplify.c (gimplify_expr): Retain alignment of
157 addressable lvalue in dereference.
159 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
162 * configure.ac: Check for .loc is_stmt support.
163 * configure, config.in: Rebuilt.
164 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
167 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
169 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
170 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
173 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
175 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
177 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
180 2018-09-20 Marek Polacek <polacek@redhat.com>
182 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
184 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
186 PR tree-optimization/87288
187 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
188 into account when determining PEELING_FOR_NITERS.
190 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
192 PR tree-optimization/86877
193 * tree-vect-loop.c (vect_analyze_loop_2): Call
194 vect_verify_datarefs_alignment.
196 2018-09-19 Marek Polacek <polacek@redhat.com>
198 * doc/invoke.texi: Document -Wclass-conversion.
200 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
202 * config/pa/pa.c (pa_adjust_priority): Delete.
203 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
205 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
206 (atomic_storehi): Likewise.
207 (atomic_storesi): Likewise.
208 (atomic_loaddi): Restore compare and swap exchange loop code.
210 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
212 PR rtl-optimization/86902
213 * combine.c (try_combine): When changing the CC mode used, don't change
214 an unrelated mode in other_insn to that new CC mode.
216 2018-09-19 David Malcolm <dmalcolm@redhat.com>
218 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
219 with %T in place of calls to dump_generic_expr.
220 (prune_runtime_alias_test_list): Likewise.
221 (create_runtime_alias_checks): Likewise.
222 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
223 (vect_analyze_data_ref_dependence): Likewise.
224 (vect_slp_analyze_data_ref_dependence): Likewise.
225 (vect_record_base_alignment): Likewise. Use %G in place of call
227 (vect_compute_data_ref_alignment): Likewise.
228 (verify_data_ref_alignment): Likewise.
229 (vect_find_same_alignment_drs): Likewise.
230 (vect_analyze_group_access_1): Likewise.
231 (vect_analyze_data_ref_accesses): Likewise.
232 (dependence_distance_ge_vf): Likewise.
233 (dump_lower_bound): Likewise.
234 (vect_prune_runtime_alias_test_list): Likewise.
235 (vect_find_stmt_data_reference): Likewise.
236 (vect_analyze_data_refs): Likewise.
237 (vect_create_addr_base_for_vector_ref): Likewise.
238 (vect_create_data_ref_ptr): Likewise.
239 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
240 (vect_can_advance_ivs_p): Likewise.
241 (vect_update_ivs_after_vectorizer): Likewise.
242 (vect_gen_prolog_loop_niters): Likewise.
243 (vect_prepare_for_masked_peels): Likewise.
244 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
245 (vect_determine_vectorization_factor): Likewise.
246 (vect_is_simple_iv_evolution): Likewise.
247 (vect_analyze_scalar_cycles_1): Likewise.
248 (vect_analyze_loop_operations): Likewise.
249 (report_vect_op): Likewise.
250 (vect_is_slp_reduction): Likewise.
251 (check_reduction_path): Likewise.
252 (vect_is_simple_reduction): Likewise.
253 (vect_create_epilog_for_reduction): Likewise.
254 (vect_finalize_reduction:): Likewise.
255 (vectorizable_induction): Likewise.
256 (vect_transform_loop_stmt): Likewise.
257 (vect_transform_loop): Likewise.
258 (optimize_mask_stores): Likewise.
259 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
260 (vect_split_statement): Likewise.
261 (vect_recog_over_widening_pattern): Likewise.
262 (vect_recog_average_pattern): Likewise.
263 (vect_determine_min_output_precision_1): Likewise.
264 (vect_determine_precisions_from_range): Likewise.
265 (vect_determine_precisions_from_users): Likewise.
266 (vect_mark_pattern_stmts): Likewise.
267 (vect_pattern_recog_1): Likewise.
268 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
269 (vect_record_max_nunits): Likewise.
270 (vect_build_slp_tree_1): Likewise.
271 (vect_build_slp_tree_2): Likewise.
272 (vect_print_slp_tree): Likewise.
273 (vect_analyze_slp_instance): Likewise.
274 (vect_detect_hybrid_slp_stmts): Likewise.
275 (vect_detect_hybrid_slp_1): Likewise.
276 (vect_slp_analyze_operations): Likewise.
277 (vect_slp_analyze_bb_1): Likewise.
278 (vect_transform_slp_perm_load): Likewise.
279 (vect_schedule_slp_instance): Likewise.
280 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
281 (vect_mark_stmts_to_be_vectorized): Likewise.
282 (vect_init_vector_1): Likewise.
283 (vect_get_vec_def_for_operand): Likewise.
284 (vect_finish_stmt_generation_1): Likewise.
285 (vect_check_load_store_mask): Likewise.
286 (vectorizable_call): Likewise.
287 (vectorizable_conversion): Likewise.
288 (vectorizable_operation): Likewise.
289 (vectorizable_load): Likewise.
290 (vect_analyze_stmt): Likewise.
291 (vect_is_simple_use): Likewise.
292 (vect_get_vector_types_for_stmt): Likewise.
293 (vect_get_mask_type_for_stmt): Likewise.
294 * tree-vectorizer.c (increase_alignment): Likewise.
296 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
298 * doc/rtl.texi: Adjust vec_select description.
299 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
300 non-constant selectors.
302 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
304 * config/aarch64/aarch64-protos.h
305 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
306 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
307 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
308 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
309 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
310 (AARCH64_FL_PROFILE): Move index so flags are ordered.
311 (AARCH64_ISA_RCPC8_4): New flag.
312 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
313 to aarch64_offset_9bit_signed_unscaled_p.
314 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
316 * config/aarch64/constraints.md (Ust): New constraint.
317 * config/aarch64/predicates.md.
318 (aarch64_9bit_offset_memory_operand): New predicate.
319 (aarch64_rcpc_memory_operand): New predicate.
321 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
323 PR rtl-optimization/87361
324 * rtlanal.c (nonzero_bits1): Revert accidental change.
326 2018-09-19 Richard Biener <rguenther@suse.de>
328 PR tree-optimization/87349
329 PR tree-optimization/87342
330 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
332 2018-09-18 Marek Polacek <polacek@redhat.com>
334 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
335 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
337 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
339 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
341 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
343 PR rtl-optimization/86882
344 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
346 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
348 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
349 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
351 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
354 * doc/invoke.texi (Link Options): Fix formatting and grammar.
356 2018-09-18 Richard Biener <rguenther@suse.de>
359 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
360 (compute_samebase_partition_bases): Likewise.
361 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
362 (gimple_can_coalesce_p): Simplify.
364 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
366 Handle a library implementation of ffs calling __builtin_ffs.
367 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
368 (mmix_init_libfuncs): New function: make __builtin_ffs expand
371 2018-09-17 David Malcolm <dmalcolm@redhat.com>
373 * diagnostic-show-locus.c (class layout_range): Add field
375 (layout_range::layout_range): Add "original_idx" param and use it
376 to initialize new field.
377 (make_range): Use 0 for original_idx.
378 (layout::layout): Pass in index to calls to
379 maybe_add_location_range.
380 (layout::maybe_add_location_range): Add param "original_idx" and
381 pass it on to layout_range.
382 (layout::print_any_labels): Pass on range->m_original_idx to
384 (gcc_rich_location::add_location_if_nearby): Use 0 for
386 * gcc-rich-location.h (text_range_label::get_text): Update for new
388 (range_label_for_type_mismatch::get_text): Likewise.
390 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
392 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
394 2018-09-17 David Malcolm <dmalcolm@redhat.com>
396 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
397 format_string_diagnostic_t.
398 (fmtwarn_n): Likewise.
399 * substring-locations.c
400 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
401 (format_warning_n_va): Convert to...
402 (format_string_diagnostic_t::emit_warning_n_va): ...this.
403 (format_warning_va): Convert to...
404 (format_string_diagnostic_t::emit_warning_va): ...this.
405 (format_warning_at_substring): Convert to...
406 (format_string_diagnostic_t::emit_warning): ...this.
407 (format_warning_at_substring_n): Convert to...
408 (format_string_diagnostic_t::emit_warning_n): ...this.
409 * substring-locations.h (class format_string_diagnostic_t): New
411 (format_warning_va): Convert to
412 format_string_diagnostic_t::emit_warning_va.
413 (format_warning_n_va): Convert to
414 format_string_diagnostic_t::emit_warning_n_va.
415 (format_warning_at_substring): Convert to
416 format_string_diagnostic_t::emit_warning.
417 (format_warning_at_substring_n): Convert to
418 format_string_diagnostic_t::emit_warning_n.
420 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
421 Bernd Schmidt <bernds_cb1@t-online.de>
423 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
426 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
428 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
429 operand 0 predicate to nonimmediate operand.
430 (rint<mode>2_frndint): Remove insn pattern.
431 (rint<mode>2): Change operand 1 predicate to general_operand.
432 Extend operand 1 to XFmode and generate rintxf2 insn.
433 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
434 Do not use X87MODEF mode macro.
435 (frndintxf2_<rounding>_i387): Rename from
436 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
437 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
438 to XFmode and generate significandxf3 insn.
440 2018-09-17 Richard Biener <rguenther@suse.de>
442 PR tree-optimization/87328
443 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
444 visiting unexecutable backedges when not iterating.
445 (do_rpo_vn): Mark all edges not executable even when not
448 2018-09-17 Martin Jambor <mjambor@suse.cz>
451 * doc/invoke.texi (Warning Options): Likewise.
453 2018-09-17 Richard Biener <rguenther@suse.de>
455 PR tree-optimization/87301
456 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
457 clean EH info from leftover copy assignments.
459 2018-09-17 Martin Liska <mliska@suse.cz>
461 PR gcov-profile/85871
462 * gcov.c (output_intermediate_file): Fix out of bounds
465 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
467 * config/arc/arc.c: Object attributes for core4 not reflected
469 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
472 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
474 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
476 2018-09-17 Martin Liska <mliska@suse.cz>
478 * doc/gcov.texi: Document new option --use-hotness-colors.
479 * gcov.c (struct source_info): Declare new field.
480 (source_info::source_info): Set default for maximum_count.
481 (print_usage): Add new -q option.
482 (process_args): Process it.
483 (accumulate_line_info): Save src->maximum_count.
484 (output_line_beginning): Make color line number if
485 flag_use_hotness_colors is set.
486 (output_line_details): Pass default argument value.
487 (output_lines): Pass src->maximum_count.
489 2018-09-17 Martin Liska <mliska@suse.cz>
491 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
492 Use processor_names table.
493 * config/i386/i386.c (ix86_default_align): Use
494 processor_cost_table for alignment values.
495 (ix86_option_override_internal): Use processor_names.
496 (ix86_function_specific_print): Likewise.
497 * config/i386/i386.h (struct processor_costs):
498 Add alignment values.
499 (struct ptt): Remove and replace with const char *.
500 * config/i386/x86-tune-costs.h (struct processor_costs):
501 Declare default alignments for all costs.
503 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
505 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
506 symbolics or VR_VARYING ranges for ABS_EXPR.
507 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
508 when range will wrap.
510 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
513 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
514 before and after a JUMP_TABLE_DATA.
516 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
519 * dojump.c (do_compare_and_jump): Canonicalize function pointers
520 when one operand is a function pointer. Use POINTER_TYPE_P and
521 FUNC_OR_METHOD_TYPE_P.
522 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
523 * fold-const.c (build_range_check): Likewise.
524 * match.pd (simple_comparison): Likewise.
526 2018-09-14 David Malcolm <dmalcolm@redhat.com>
529 * spellcheck.c (get_edit_distance_cutoff): New function.
530 (selftest::test_edit_distance_unit_test_oneway): Rename to...
531 (selftest::test_get_edit_distance_one_way): ...this.
532 (selftest::test_get_edit_distance_unit): Rename to...
533 (selftest::test_get_edit_distance_both_ways): ...this.
534 (selftest::test_edit_distances): Move tests to this new function,
535 and test some more pairs of strings. Update for above renaming.
536 (selftest::get_old_cutoff): New function.
537 (selftest::test_get_edit_distance_cutoff): New function.
538 (selftest::assert_suggested_for): New function.
539 (ASSERT_SUGGESTED_FOR): New macro.
540 (selftest::assert_not_suggested_for): New function.
541 (ASSERT_NOT_SUGGESTED_FOR): New macro.
542 (selftest::test_suggestions): New function.
543 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
544 tests to selftest::test_edit_distances and call it. Add calls to
545 selftest::test_get_edit_distance_cutoff and
546 selftest::test_suggestions.
547 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
548 (best_match::consider): Replace hard-coded cutoff calculation with
550 (best_match::get_cutoff): New declaration.
551 (best_match::get_best_meaningful_candidate): Likewise.
553 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
555 * builtins.c (fold_builtin_strlen): Remove TODO comment.
557 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
560 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
562 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
563 terminated string literal.
565 2018-09-14 Martin Sebor <msebor@redhat.com>
567 * builtins.c (unterminated_array): Handle ARRAY_REF.
568 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
569 * builtins.h (unterminated_array): Declare extern.
570 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
572 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
575 2018-09-14 Martin Sebor <msebor@redhat.com>
576 Jeff Law <law@redhat.com>
578 * builtins.c (unterminated_array): New.
579 (expand_builtin_strcpy): Adjust.
580 (expand_builtin_strcpy_args): Detect unterminated arrays.
581 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
583 * gimple-fold.h (get_maxval_strlen): Add argument.
584 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
586 * gimple-fold.c (get_range_strlen): Add argument.
587 (get_maxval_strlen): Adjust.
588 * gimple-fold.h (get_range_strlen): Add argument.
590 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
592 * config/i386/movdirintrin.h: Fix copyright year.
594 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
596 * reg-stack.c: Include regs.h.
597 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
598 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
599 FIRST_STACK_REG, not DFmode.
600 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
601 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
602 (change stack): Default register mode to the reg_raw_mode of
603 FIRST_STACK_REG, not DFmode.
604 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
605 (*swapxf): Rename from swapxf.
607 2018-09-14 Carl Love <cel@us.ibm.com>
609 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
610 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
612 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
615 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
618 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
621 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
623 2018-09-14 Jason Merrill <jason@redhat.com>
625 Fix --enable-gather-detailed-mem-stats.
626 * hash-table.c (hash_table_usage): Change from variable to function.
627 * hash-table.h: Adjust.
628 * Makefile.in: Add missing dependencies on hash-table.h.
630 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
632 PR tree-optimization/87259
634 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
635 execute_cse_reciprocals_1 has tried transforming.
637 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
639 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
640 VR_VARYING for PLUS/MINUS_EXPR.
642 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
644 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
647 2018-09-14 Richard Biener <rguenther@suse.de>
650 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
651 bits for the merged partition.
653 2018-09-13 Martin Sebor <msebor@redhat.com>
654 Bernd Edlinger <bernd.edlinger@hotmail.de>
656 * builtins.h (c_srlen): Add argument.
657 * builtins.c (warn_string_no_nul): New function.
658 (c_strlen): Add argument and use it. Update recursive calls.
659 Pass DECL argument to string_constant to get info on non
660 terminated strings. Update *NONSTR as needed.
661 (fold_builtin_strlen): Add argument to calls to c_strlen.
662 Warn for unterminated arrays.
663 (warn_string_no_null): Add prototype.
664 * expr.c (string_constant): Update arguments. Update recursive
665 calls appropriately. Detect missing NUL terminator and outermost
666 declaration its missing in.
667 Improve checks for arrays with nonzero lower bound or elements
668 that are not a single byte. Simplify offset computation.
669 Simplify checks for non-NUL terminated strings.
670 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
671 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
673 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
675 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
677 * fold-const.c (c_getstr): Fix function comment. Remove unused third
678 argument. Fix range checks.
679 * fold-const.h (c_getstr): Adjust protoype.
680 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
681 string is constant but contains no NUL byte.
683 * expr.c (string_constant): Adjust function comment.
684 Remove bogus check for zero termination.
686 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
688 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
689 (get_constant_size): Don't make STRING_CSTs larger than they are.
690 (check_string_literal): New check function for STRING_CSTs.
691 (output_constant): Use it.
693 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
696 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
698 2018-09-13 Richard Biener <rguenther@suse.de>
700 PR tree-optimization/87263
701 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
702 (struct unwind_state): Add max_rpo field.
703 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
704 Compute max_rpo, the max RPO number a block can be backwards reached
705 from. Re-write non-iterating mode to a RPO ordered worklist approach,
706 separating it from the iterating mode.
708 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
710 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
711 (rfs_decision): New scheduling decision.
713 2018-09-13 Richard Biener <rguenther@suse.de>
716 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
717 (vn_nary_op_insert_pieces_predicated): Do not write useless
718 valid_dominated_by_p entry outside of the allocated storage.
720 2018-09-13 Omar Sandoval <osandov@osandov.com>
721 Tom de Vries <tdevries@suse.de>
724 * dwarf2out.c (is_c): New function.
725 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
727 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
730 * config/aarch64/aarch64.md (*aarch64_bfxil):
732 * config/aarch64/constraints.md (Ulc): Define.
733 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
735 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
738 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
740 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
741 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
742 aarch64_layout_frame call.
743 (aarch64_expand_epilogue): Likewise.
744 (aarch64_initial_elimination_offset): Likewise.
745 (aarch64_get_separate_components): Likewise.
746 (aarch64_use_return_insn_p): Likewise.
747 (aarch64_layout_frame): Remove unneeded check.
749 2018-09-13 Jakub Jelinek <jakub@redhat.com>
751 * configure.ac: Only append
752 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
753 gcc_config_arguments if it was never reconfigured or last reconfigure
754 was with different arguments.
755 * configure: Regenerated.
757 2018-09-13 Jakub Jelinek <jakub@redhat.com>
758 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
761 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
762 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
764 2018-09-13 Jakub Jelinek <jakub@redhat.com>
766 PR tree-optimization/87287
767 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
768 X % C == 0 to X % (unsigned) C == 0 optimization to ...
769 * match.pd (X % C == 0): ... here. New optimization.
771 2018-09-12 Jakub Jelinek <jakub@redhat.com>
774 * expr.h (maybe_optimize_mod_cmp): Declare.
775 * expr.c (mod_inv): New function.
776 (maybe_optimize_mod_cmp): New function.
777 (do_store_flag): Use it.
778 * cfgexpand.c (expand_gimple_cond): Likewise.
780 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
781 Julian Brown <julian@codesourcery.com>
784 * gimplify.c (gimplify_scan_omp_clauses): Set
785 target_firstprivatize_array_bases in OpenACC parallel and kernels
786 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
787 OpenACC data regions.
789 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
791 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
792 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
794 2018-09-12 Richard Biener <rguenther@suse.de>
796 PR tree-optimization/87280
797 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
798 edge but unreachable target.
799 (do_rpo_vn): For conservatively handling a PHI only mark
800 the backedge executable but not the block reachable.
802 2018-09-12 Richard Biener <rguenther@suse.de>
804 PR tree-optimization/87266
805 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
808 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
810 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
812 ("trunc<BFP:mode><DFP_ALL:mode>2")
813 ("trunc<DFP_ALL:mode><BFP:mode>2")
814 ("extend<BFP:mode><DFP_ALL:mode>2")
815 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
816 according to the target operand type.
818 2018-09-12 Jakub Jelinek <jakub@redhat.com>
819 Andreas Krebbel <krebbel@linux.ibm.com>
821 PR tree-optimization/86844
822 * gimple-ssa-store-merging.c
823 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
824 there are any overlapping stores in between them, make sure they are
825 also coalesced or we give up completely.
827 2018-09-12 Jakub Jelinek <jakub@redhat.com>
830 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
831 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
833 2018-09-12 Tom de Vries <tdevries@suse.de>
835 * common.opt (gdescribe-dies): Add option.
836 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
837 attribute for artifical and nameless decls.
838 (dwarf2out_register_external_die): Add description attribute to
839 external reference die.
840 (add_desc_attribute): New functions.
841 (gen_subprogram_die): Add description attribute to
842 DW_TAG_call_site_parameter.
843 * tree-pretty-print.c (print_generic_expr_to_str): New function.
844 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
845 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
847 (@item -gdescribe-dies): Add.
849 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
851 * tree-vrp.c (vrp_shift_undefined_p): Remove.
852 (extract_range_from_binary_expr_1: Call
853 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
854 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
857 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
859 * gimple-ssa-warn-alloca.c
860 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
861 field for ALLOCA_BOUND_*_LARGE.
863 2018-09-11 Nathan Sidwell <nathan@acm.org>
865 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
867 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
869 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
870 for clobbers. Remove obsolete comment.
872 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
874 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
875 mpxchk, mpxld and mpxst types.
876 (define_attr length_immediate): Remove all processing of mpx types.
877 (define_attr prefix_0f): Ditto.
878 (define_attr memory): Ditto.
880 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
882 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
883 (log<mode>2): Change operand 1 predicate to general_operand.
884 Extend operand 1 to XFmode and generate logxf3 insn.
885 (log10<mode>2): Change operand 1 predicate to general_operand.
886 Extend operand 1 to XFmode and generate log10xf3 insn.
887 (log2<mode>2): Change operand 1 predicate to general_operand.
888 Extend operand 1 to XFmode and generate log2xf3 insn.
889 (fyl2xp1_extend<mode>xf3_i387): Remove.
890 (log1p<mode>2): Change operand 1 predicate to general_operand.
891 Extend operand 1 to XFmode and generate log1pxf3 insn.
892 (fxtract_extend<mode>xf3_i387): Remove.
893 (logb<mode>2): Change operand 1 predicate to general_operand.
894 Extend operand 1 to XFmode and generate logbxf3 insn.
895 (ilogb<mode>2): Change operand 1 predicate to general_operand.
896 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
897 (significand<mode>2): Change operand 1 predicate to general_operand.
898 Extend operand 1 to XFmode and generate significandxf3 insn.
900 2018-09-11 Nathan Sidwell <nathan@acm.org>
902 * gcc.c (perror_with_name, pfatal_with_name): Delete.
903 (load_specs): Use fatal_error.
904 (DELETE_IF_ORDINARY, process_command): Use error.
905 (execute, run_attempt): Use fatal_error.
907 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
909 * diagnostic-core.h (sorry_at): New prototype.
910 * diagnostic.c (sorry_at): New function.
912 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
914 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
915 by zero as VR_UNDEFINED.
917 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
919 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
920 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
921 (<sincos>mode2): New expander.
922 (sincos_extend<mode>xf3_i387): Remove insn pattern.
923 (sincos -> sin, cos splitters): Remove splitter patterns.
924 (sincos<mode>3): Change operand 2 predicate to general_operand.
925 Extend operand 2 to XFmode and generate sincosxf3 insn.
926 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
927 Change operand 3 predicate to const1_operand.
928 (fptan_extend<mode>xf4_i387): Remove insn pattern.
929 (tanxf2): Update operands in the call to fptanxf4_i387.
930 (tan<mode>2): Change operand 1 predicate to general_operand.
931 Extend operand 1 to XFmode and generate tanxf3 insn.
932 (atan2xf3): Rename from *fpatanxf3_i387.
933 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
934 (atan2xf3): Remove expander.
935 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
936 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
937 (atan<mode>2): Change operand 1 predicate to general_operand.
938 Extend operand 1 to XFmode and generate atanxf3 insn.
940 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
942 * config/i386/i386.md (x87/SSE constant load splitter): Use
943 memory_operand instead of nonimmediate_operand for input operand
946 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
948 * config/i386/i386.md (float partial SSE register stall splitter): Move
949 splitter near its instruction pattern.
950 (float_extend partial SSE register stall splitter): Ditto.
951 (float_truncate partial SSE register stall splitter): Ditto.
953 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
956 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
957 to speculation_safe_value_not_needed.
960 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
961 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
962 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
963 leaf_function_p, instead use has_hard_reg_initial_val.
965 2018-09-09 Nathan Sidwell <nathan@acm.org>
967 * gcc.h (pfatal_with_name): Don't declare here.
968 * gcc.c (pfatal_with_name): Make static.
970 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
972 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
975 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
977 PR rtl-optimization/85458
978 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
979 priority hook to reduce the priority of EXPR.
981 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
983 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
984 DImode for x87 on 32bit targets. Conditionally disable x87 modes
985 with X87_ENABLE_FLOAT. Remove preparation code.
986 (*float<SWI48:mode><MODEF:mode>2): Rename from
987 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
988 math using "enabled" attribute.
989 (*floatdi<MODEF:mode>2_i387): Rename from
990 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
991 enable for 32bit targets only.
992 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
994 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
995 as operand 1 predicate. Rewrite as define_insn_and_split.
996 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
998 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
1000 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
1001 to fallthru to FLOAT case.
1003 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
1006 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
1007 around folding of vec_sl to handle out of range shift values.
1009 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
1011 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
1012 Update callers to gen_fix_trunc<mode>_i387_fisttp
1013 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
1014 nonimmediate_operand.
1015 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
1016 and corresponding splitters.
1017 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
1018 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
1019 (fix_truncdi_i387_with_temp): Remove insn pattern
1020 and corresponding splitters.
1021 (fix_trunc<mode>_i387): Change operand 0 predicate to
1022 nonimmediate_operand.
1023 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
1024 and corresponding splitters.
1025 (*fistdi2_1): Remove.
1027 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
1028 (lrintxfdi2): Remove expander. Reimplement as define_insn.
1029 (*fist<mode>2_1): Remove.
1030 (fist<mode>2): Ditto.
1031 (fist<mode>2_with_temp): Remove insn pattern and corresponding
1033 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
1034 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
1035 (fistdi2_<rounding>): Change operand 0 predicate to
1036 nonimmediate_operand.
1037 (fistdi2_<rounding>_with_temp): Remove insn pattern
1038 and corresponding splitters.
1039 (fist<mode>2_<rounding>): Change operand 0 predicate to
1040 nonimmediate_operand.
1041 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
1042 and corresponding splitters.
1044 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
1046 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1048 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
1051 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
1053 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1054 early gimple folding of vec_splat().
1055 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
1056 * gimple-fold.h: Add an extern define for tree_vec_extract().
1058 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
1060 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
1061 wrappers around TREE_TYPE comparisons.
1063 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
1066 * config/s390/predicates.md: Add nonsym_memory_operand.
1067 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
1068 contains a SYMBOL_REF, load it into an intermediate pseudo.
1069 (s390_emit_compare_and_swap): Legitimize operand.
1070 * config/s390/s390.md: Use the new nonsym_memory_operand
1071 with UNSPECV_CAS patterns.
1073 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
1076 * config/s390/s390-passes.def: New file.
1077 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
1079 (make_pass_s390_early_mach): Add declaration.
1080 * config/s390/s390.c (make_pass_s390_early_mach):
1081 (s390_option_override): Remove dynamic registration.
1082 * config/s390/t-s390: Add s390-passes.def.
1084 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
1086 * config/s390/s390.c (s390_decompose_constant_pool_ref):
1087 Remove UNSPEC_LTREL_BASE check.
1088 (annotate_constant_pool_refs): Likewise.
1089 (find_constant_pool_ref): Likewise.
1090 (find_ltrel_base): Removed.
1091 (replace_ltrel_base): Removed.
1092 (s390_mainpool_finish): Remove replace_ltrel_base call.
1093 (s390_chunkify_start): Remove pending LTREL_BASE logic.
1094 (s390_chunkify_finish): Remove replace_ltrel_base call.
1095 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
1097 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
1100 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
1101 to speculation_safe_value_not_needed.
1103 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
1104 Bernd Schmidt <bernds_cb1@t-online.de>
1106 * config/nvptx/nvptx-opts.h: New file.
1107 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
1108 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
1110 (TARGET_SM35): New macro.
1111 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
1113 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
1115 (misa=): New option.
1116 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
1118 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
1120 * config/i386/i386.md (truncdfsf2): Remove expander.
1121 (truncdfsf2_with_temp): Ditto.
1122 (truncxf<mode>2): Ditto.
1123 (*truncdfsf_fast_mixed): Remove insn pattern.
1124 (*truncdfsf_fast_i387): Ditto.
1125 (*truncdfsf_mixed): Ditto.
1126 (*truncdfsf_i387): Ditto.
1127 (*truncdfsf2_i387_1): Ditto.
1128 (*truncxfsf2_mixed): Ditto.
1129 (*truncxfdf2_mixed): Ditto.
1130 (*truncxf<mode>2_i387_noop): Ditto. Update callers
1131 to call gen_truncxf<mode>2 instead.
1132 (*truncxf<mode>2_i387): Remove.
1133 (reg->reg splitters): Remove splitter pattern.
1134 (reg->mem splitters): Ditto.
1136 (truncdfsf2): New insn pattern.
1137 (truncxf<mode>2): Ditto.
1139 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1141 * tree-ssa-math-opts.c (is_mult_by): New function.
1142 (is_square_of): Use the above.
1143 (optimize_recip_sqrt): New function.
1144 (pass_cse_reciprocals::execute): Use the above.
1146 2018-09-05 Richard Biener <rguenther@suse.de>
1149 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
1150 to zero-init the emplaced vec.
1152 2018-09-05 Martin Liska <mliska@suse.cz>
1154 PR tree-optimization/87205
1155 * tree-switch-conversion.c (pass_lower_switch::execute):
1156 Group cases for switch statements.
1158 2018-09-05 Richard Biener <rguenther@suse.de>
1160 PR tree-optimization/87217
1161 * tree-ssa-sccvn.c (vuse_valueize): New.
1162 (vn_reference_lookup_pieces): Use it.
1163 (vn_reference_lookup): Likewise.
1165 2018-09-05 Nathan Sidwell <nathan@acm.org>
1168 * stor-layout.c (place_field): Scan forwards to check last
1169 bitfield when ms_bitfield_placement is in effect.
1171 2018-09-05 Richard Biener <rguenther@suse.de>
1174 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
1177 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
1178 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
1180 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
1181 * config.gcc (extra_objs): Build it.
1182 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
1184 * config/aarch64/aarch64-passes.def
1185 (pass_tag_collision_avoidance): New pass.
1186 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
1187 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
1188 (aarch64_classify_address): Remove static qualifier.
1189 (aarch64_address_info, aarch64_address_type): Move to...
1190 * config/aarch64/aarch64-protos.h: ... here.
1191 (make_pass_tag_collision_avoidance): New function.
1192 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
1195 2018-09-05 Martin Liska <mliska@suse.cz>
1197 * doc/gcov.texi: Update documentation of humar
1199 * gcov.c (format_count): Print one decimal place, it provides
1200 more fine number of situations like '1G' vs. '1.4G'.
1202 2018-09-05 Martin Liska <mliska@suse.cz>
1205 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
1206 * optc-gen.awk: Allow 'Var' for Deprecated options in order
1207 to generate a MASK value.
1209 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
1212 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
1213 if frame pointer isn't used.
1214 (compute_frame_pointer_to_fb_displacement): Likewise.
1216 2018-09-04 Jakub Jelinek <jakub@redhat.com>
1219 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
1220 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
1221 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
1222 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
1223 and OPTION_MASK_ISA_XSAVEC_UNSET.
1225 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
1227 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
1228 XOR operations in NAND case.
1230 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
1232 * wide-int-range.cc (wide_int_range_convert): New.
1233 * wide-int-range.h (wide_int_range_convert): New.
1234 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
1235 code into wide_int_range_convert.
1236 (extract_range_into_wide_ints): Do not munge anti range constants
1237 into the entire domain. Just return the range back.
1239 2018-09-04 Martin Liska <mliska@suse.cz>
1241 * genmatch.c (output_line_directive): Add new argument
1243 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
1245 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
1247 * doc/invoke.texi (Option Summary): Add whitespace.
1249 * doc/invoke.texi (Option Summary): Add -Waligned-new.
1251 2018-09-04 Richard Biener <rguenther@suse.de>
1253 PR tree-optimization/87211
1254 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
1255 backedge value we're supposed to treat as VARYING also number
1256 the PHI to VARYING in case it got a different value-number already.
1258 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
1260 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
1261 (extract_range_from_binary_expr_1): Do not call
1262 vrp_can_optimize_bit_op.
1263 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
1265 (wide_int_range_get_mask_and_bounds): New.
1266 (wide_int_range_optimize_bit_op): New.
1267 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
1268 (wide_int_range_bit_and): Same.
1269 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
1270 (wide_int_range_optimize_bit_op): New.
1271 (wide_int_range_get_mask_and_bounds): New.
1273 2018-09-04 Richard Biener <rguenther@suse.de>
1275 PR tree-optimization/87176
1276 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
1277 variable. When value-numbering a virtual PHI node make sure
1278 to not value-number to the backedge value.
1280 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
1282 * doc/extend.texi (Long Long, Hex Floats): Document support for
1283 long long and hex floats in more recent versions of ISO C++.
1285 2018-09-03 Richard Biener <rguenther@suse.de>
1287 PR tree-optimization/87177
1288 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
1291 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1293 * bb-reorder.c (edge_order): Convert to C-qsort-style
1294 tri-state comparator.
1295 (reorder_basic_blocks_simple): Change std::stable_sort to gcc_stablesort.
1297 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1299 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
1300 tri-state comparator.
1301 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
1303 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1305 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
1306 (mergesort): ... here as maximum count for using netsort.
1307 (gcc_qsort): Set nlim to 3 if stable sort is requested.
1308 (gcc_stablesort): New.
1309 * system.h (gcc_stablesort): Declare.
1311 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1313 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
1314 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
1315 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
1317 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
1319 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
1320 lxsdx and stxsdx alternatives.
1321 (*mov<mode>_hardfloat64): Ditto.
1322 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
1324 2018-09-03 Richard Biener <rguenther@suse.de>
1326 PR tree-optimization/87200
1327 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
1330 2018-09-03 Martin Liska <mliska@suse.cz>
1332 PR tree-optimization/87201
1333 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
1334 Fix parenthesis in an expression.
1336 2018-09-03 Richard Biener <rguenther@suse.de>
1338 PR tree-optimization/87197
1339 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
1340 visited. CSE the VN_INFO hashtable lookup.
1342 PR tree-optimization/87169
1343 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
1344 iterating make sure there's no extra backedges from irreducible
1345 regions feeding the header. Mark the destination block
1348 2018-09-03 Martin Liska <mliska@suse.cz>
1351 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
1352 * common/common-targhooks.c (default_get_valid_option_values):
1354 * common/common-targhooks.h (default_get_valid_option_values):
1356 * common/config/i386/i386-common.c: Move processor_target_table
1358 (ix86_get_valid_option_values): New function.
1359 (TARGET_GET_VALID_OPTION_VALUES): New macro.
1360 * config/i386/i386.c (struct ptt): Move to i386-common.c.
1361 (PTA_*): Move all defined masks into i386-common.c.
1362 (ix86_function_specific_restore): Use new processor_cost_table.
1363 * config/i386/i386.h (struct ptt): Moved from i386.c.
1364 (struct pta): Likewise.
1365 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
1366 * doc/tm.texi.in: Likewise.
1367 * opt-suggestions.c (option_proposer::suggest_option):
1368 Pass prefix to build_option_suggestions.
1369 (option_proposer::get_completions): Likewise.
1370 (option_proposer::build_option_suggestions): Use the new target
1372 * opts.c (struct option_help_tuple): New struct.
1373 (print_filtered_help): Use the new target hook.
1375 2018-09-03 Martin Liska <mliska@suse.cz>
1378 * predict.c (set_even_probabilities): Add likely_edges
1379 argument and handle cases where we have precisely one
1381 (combine_predictions_for_bb): Catch also likely_edges.
1382 (tree_predict_by_opcode): Handle gswitch statements.
1383 * tree-cfg.h (find_case_label_for_value): New declaration.
1384 (find_taken_edge_switch_expr): Likewise.
1385 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
1386 Find pivot in decision tree based on probabily, not by number of
1389 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
1391 * doc/standards.texi (Standards): Update Objective-C reference.
1393 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1395 * doc/install.texi (Prerequisites): Update link for MPC.
1397 2018-09-01 Michael Matz <matz@suse.de>
1399 PR tree-optimization/87074
1400 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
1401 PHIs for outer-loop uses.
1403 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1405 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
1406 * doc/invoke.texi (C Dialect Options): Ditto.
1408 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1410 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
1412 2018-08-31 Richard Biener <rguenther@suse.de>
1414 PR tree-optimization/87168
1415 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
1416 (rpo_elim::eliminate_avail): When OP was not visited it must
1419 2018-08-31 David Malcolm <dmalcolm@redhat.com>
1421 * tree-vrp.c (copy_value_range): Convert param "from" from
1422 "value_range *" to "const value_range *".
1423 (range_is_null): Likewise for param "vr".
1424 (range_int_cst_p): Likewise.
1425 (range_int_cst_singleton_p): Likewise.
1426 (symbolic_range_p): Likewise.
1427 (value_ranges_intersect_p): Likewise for both params.
1428 (value_range_nonnegative_p): Likewise for param "vr".
1429 (value_range_constant_singleton): Likewise.
1430 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
1431 (extract_range_into_wide_ints): Likewise for param "vr".
1432 (extract_range_from_multiplicative_op): Likewise for params "vr0"
1434 (vrp_can_optimize_bit_op): Likewise.
1435 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
1437 (extract_range_from_unary_expr): Likewise.
1438 (debug_value_range): Likewise for param "vr".
1439 (value_range::dump): Add "const" qualifier.
1440 (vrp_prop::check_array_ref): Convert local "vr" from
1441 "value_range *" to "const value_range *".
1442 (vrp_prop::check_mem_ref): Likewise.
1443 (vrp_prop::visit_stmt): Likewise for local "old_vr".
1444 (vrp_intersect_ranges_1): Likewise for param "vr_1".
1445 (vrp_intersect_ranges): Likewise.
1446 (simplify_stmt_for_jump_threading): Likewise for local "vr".
1447 (vrp_prop::vrp_finalize): Likewise.
1448 * tree-vrp.h (value_range::dump): Add "const" qualifier.
1449 (vrp_intersect_ranges): Add "const" qualifier to params as above.
1450 (extract_range_from_unary_expr): Likewise.
1451 (value_range_constant_singleton): Likewise.
1452 (symbolic_range_p): Likewise.
1453 (copy_value_range): Likewise.
1454 (extract_range_from_binary_expr_1): Likewise.
1455 (range_int_cst_p): Likewise.
1456 (vrp_set_zero_nonzero_bits): Likewise.
1457 (range_int_cst_singleton_p): Likewise.
1459 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
1461 * config/aarch64/arm_neon.h (vabsd_s64): New.
1462 (vnegd_s64): Likewise.
1464 2018-08-31 Martin Jambor <mjambor@suse.cz>
1466 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
1468 2018-08-31 Martin Liska <mliska@suse.cz>
1470 * ipa-icf.c (sem_item::add_type): Use
1471 sem_item::m_type_hash_cache.
1472 * ipa-icf.h: Move the cache from sem_item_optimizer
1475 2018-08-31 Nathan Sidwell <nathan@acm.org>
1477 * doc/extend.texi (Backwards Compatibility): Remove implicit
1478 extern C leeway of () being (...).
1480 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1482 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
1484 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
1488 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
1490 2018-08-31 Jakub Jelinek <jakub@redhat.com>
1493 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
1494 gen_int_mode. Formatting fixes.
1496 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
1498 * target.def (custom_function_descriptors): Improve documentation.
1499 * doc/tm.texi.in (Trampolines): Expand discussion of function
1500 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
1501 beginning of the section.
1502 * doc/tm.texi: Regenerated.
1504 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
1506 * cfg.h (class auto_edge_flag): Spell out the template-id of the
1507 base class in the initializer list. This is a workaround for
1508 building with older GCC.
1509 (class auto_bb_flag): Likewise.
1511 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
1513 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
1514 (altivec_vcmpequ<VI_char>_p): Remove star.
1515 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
1517 (expand_strncmp_vec_sequence): New function.
1518 (emit_final_str_compare_vec): New function.
1519 (expand_strn_compare): Add support for vector strncmp.
1520 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
1521 length specification to bytes.
1522 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
1523 (vcmpnezb_p): New pattern.
1524 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
1525 for option -mstring-compare-inline-limit.
1527 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
1529 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
1530 (PTA_SKYLAKE): Add PTA_AES.
1531 (PTA_GOLDMONT): Likewise.
1533 2018-08-29 Jan Hubicka <jh@suse.cz>
1536 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
1537 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
1539 2018-08-29 Jan Hubicka <jh@suse.cz>
1541 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
1543 (hash_tree): Do not visit TYPE_STUB_DECL.
1544 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
1545 stream TYPE_STUB_DECL.
1546 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
1547 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
1548 after free_lang_data.
1549 (type_in_anonymous_namespace_p): Likewise.
1551 2018-08-29 Jan Hubicka <jh@suse.cz>
1553 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
1554 comment that it has to be even number.
1555 (class sreal): Change m_sig type to int32_t.
1556 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
1557 int64_t for temporary calculations.
1558 (sreal_verify_basics): Drop one bit from minimum and maximum.
1560 2018-08-30 Richard Biener <rguenther@suse.de>
1562 PR tree-optimization/87147
1563 * tree-ssa-sccvn.c (SSA_VISITED): New function.
1564 (visit_phi): When the degenerate result is from the backedge and
1565 we didn't visit its definition yet drop to VARYING.
1566 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
1568 2018-08-29 Jan Hubicka <jh@suse.cz>
1570 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
1572 (hash_tree): Likewise.
1574 2018-08-29 Jan Hubicka <jh@suse.cz>
1576 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
1577 and TYPE_NEXT_REF_TO.
1579 2018-08-29 Jan Hubicka <jh@suse.cz>
1581 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
1582 comment that it has to be even number.
1583 (class sreal): Change m_sig type to int32_t.
1584 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
1585 int64_t for temporary calculations.
1586 (sreal_verify_basics): Drop one bit from minimum and maximum.
1588 2018-08-30 Tamar Christina <tamar.christina@arm.com>
1590 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
1592 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
1595 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
1596 if to_add is negative.
1598 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
1601 * builtins.c (c_strlen): Improve range checks.
1603 2018-08-29 Martin Sebor <msebor@redhat.com>
1604 Jeff Law <law@redhat.com>
1606 PR tree-optimization/86714
1607 PR tree-optimization/86711
1608 * builtins.c (c_strlen): Add arguments to call to string_constant.
1609 * expr.c (string_constant): Add argument. Detect missing nul
1610 terminator and outermost declaration it's missing in.
1611 * expr.h (string_constant): Add argument.
1612 * fold-const.c (read_from_constant_string): Add arguments to call to
1614 (c_getstr): Likewise.
1615 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
1617 * tree-ssa-strlen.c (get_stridx): Likewise.
1619 2018-08-29 Jan Hubicka <jh@suse.cz>
1621 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
1622 Do not stream DECL_VINDEX.
1623 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
1624 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
1625 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
1627 2018-08-29 Richard Biener <rguenther@suse.de>
1629 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
1630 virtual operands that are not default defs to honor region
1632 (rpo_vn_valueize): Remove ineffective code here.
1634 2018-08-29 Richard Biener <rguenther@suse.de>
1636 PR tree-optimization/87132
1637 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
1638 when skipping defs reachable over backedges.
1640 2018-08-29 Richard Biener <rguenther@suse.de>
1642 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
1643 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
1644 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
1645 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
1646 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
1647 (vn_reference_lookup_3): Remove use of const_parms.
1648 (free_rpo_vn): Do not free const_parms.
1649 (do_rpo_vn): Do not call init_const_parms.
1650 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
1651 SSA_NAME_POINTS_TO_READONLY_MEMORY.
1652 (call_may_clobber_ref_p_1): Likewise.
1654 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
1657 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
1659 (-ftree-scev-cprop): Document.
1661 2018-08-29 Jan Hubicka <jh@suse.cz>
1663 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
1665 (sreal constructor): Update.
1666 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
1667 sreal:operator/): Update.
1669 2018-08-29 Martin Liska <mliska@suse.cz>
1671 * tree-switch-conversion.c (switch_conversion::expand):
1672 Strenghten assumption about gswitch statements.
1674 2018-08-29 Richard Biener <rguenther@suse.de>
1676 PR tree-optimization/87117
1677 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
1678 re-value-number released SSA VDEFs.
1680 2018-08-29 Richard Biener <rguenther@suse.de>
1682 PR tree-optimization/87126
1683 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
1685 2018-08-28 Jim Wilson <jimw@sifive.com>
1687 * config/riscv/pic.md: Rewrite.
1688 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
1690 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
1691 (SOFTF, default_load, softload, softstore): New.
1693 2018-08-28 Jeff Law <law@redhat.com>
1695 * fold-const.c (fold_binary_loc): Remove recently added assert.
1697 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
1699 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
1700 to OP parmeter of generated function.
1702 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
1704 PR tree-optimization/87009
1705 * match.pd: Add boolean optimizations.
1707 2018-08-28 Martin Sebor <msebor@redhat.com>
1710 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
1711 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
1712 (pass_walloca::gate): Use it.
1713 (alloca_call_type): Same.
1714 (pass_walloca::execute): Same.
1715 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
1717 2018-08-28 David Malcolm <dmalcolm@redhat.com>
1719 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
1720 GCC_VERSION for usage of "__gcc_dump_printf__" format from
1723 2018-08-28 Richard Biener <rguenther@suse.de>
1725 PR tree-optimization/87124
1726 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
1727 constants before looking up avail.
1729 2018-08-28 Jakub Jelinek <jakub@redhat.com>
1732 * calls.c (maybe_warn_nonstring_arg): Punt early if
1733 warn_stringop_overflow is zero. Don't call get_range_strlen
1734 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
1735 Swap comparison operands to have constants on rhs. Only use
1736 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
1737 increment lenrng[0].
1739 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
1741 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
1742 use of tree_to_shwi. Remove duplicated test for the size being
1743 a whole number of bytes.
1745 2018-08-28 Richard Biener <rguenther@suse.de>
1747 PR tree-optimization/87117
1748 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
1749 Handle removed stmt without LHS (GIMPLE_NOP).
1751 2018-08-28 Richard Biener <rguenther@suse.de>
1753 PR tree-optimization/87117
1754 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
1755 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
1757 2018-08-28 Richard Biener <rguenther@suse.de>
1759 PR tree-optimization/87117
1760 * tree-ssa-pre.c (compute_avail): Do not make expressions
1761 with predicated values available.
1762 (get_expr_value_id): Assert we do not run into predicated value
1765 2018-08-28 Richard Biener <rguenther@suse.de>
1767 PR tree-optimization/87117
1768 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
1769 get virtual operands.
1770 (get_expr_operands): Handle STRING_CST like other decls.
1772 2018-08-28 Martin Liska <mliska@suse.cz>
1774 * tree.h: Update documentation of fndecl_built_in_p
1778 2018-08-27 Jeff Law <law@redhat.com>
1779 PR tree-optimization/87110
1780 * tree-ssa-dse.c (compute_trims): Handle non-constant
1783 2018-08-27 Martin Sebor <msebor@redhat.com>
1785 PR tree-optimization/86914
1786 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
1788 2018-08-27 Martin Sebor <msebor@redhat.com>
1790 PR tree-optimization/87112
1791 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
1792 the type of the bound argument.
1794 2018-08-27 Jeff Law <law@redhat.com>
1796 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
1797 type does not have a TYPE_SIZE_UNIT.
1799 2018-08-27 Steve Ellcey <sellcey@cavium.com>
1801 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
1802 with include of backend.h.
1804 2018-08-27 Richard Biener <rguenther@suse.de>
1806 PR tree-optimization/86927
1807 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
1808 use const cond reduction code.
1810 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
1812 PR tree-optimization/85758
1813 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
1815 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1818 * diagnostic-show-locus.c (class layout_range): Update for
1819 conversion of show_caret_p to a tri-state.
1820 (layout_range::layout_range): Likewise.
1821 (make_range): Likewise.
1822 (layout::maybe_add_location_range): Likewise.
1823 (layout::should_print_annotation_line_p): Don't show annotation
1824 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
1825 (layout::get_state_at_point): Update for conversion of
1826 show_caret_p to a tri-state. Bail out early for
1827 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
1828 underlining or source colorization.
1829 (gcc_rich_location::add_location_if_nearby): Update for conversion
1830 of show_caret_p to a tri-state.
1831 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
1832 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
1834 (selftest::test_one_liner_labels): Likewise.
1835 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
1836 conversion of show_caret_p to a tri-state.
1837 * pretty-print.c (text_info::set_location): Likewise.
1838 * pretty-print.h (text_info::set_location): Likewise.
1839 * substring-locations.c (format_warning_n_va): Likewise.
1840 * tree-diagnostic.c (default_tree_printer): Likewise.
1841 * tree-pretty-print.c (newline_and_indent): Likewise.
1843 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1846 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
1847 line above for line-insertion fix-it hints.
1848 (selftest::test_fixit_insert_containing_newline): Update the
1849 expected results, and add a test with line-numbering enabled.
1851 2018-08-27 Martin Liska <mliska@suse.cz>
1854 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
1855 params with DECL_HAS_VALUE_EXPR_P.
1857 2018-08-27 Martin Liska <mliska@suse.cz>
1859 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
1860 selected expansion strategy.
1862 2018-08-27 Martin Liska <mliska@suse.cz>
1864 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
1865 * builtins.c (is_builtin_fn): Likewise.
1866 * attribs.c (diag_attr_exclusions): Use new function
1867 fndecl_built_in_p and remove check for FUNCTION_DECL if
1869 (builtin_mathfn_code): Likewise.
1870 (fold_builtin_expect): Likewise.
1871 (fold_call_expr): Likewise.
1872 (fold_builtin_call_array): Likewise.
1873 (fold_call_stmt): Likewise.
1874 (set_builtin_user_assembler_name): Likewise.
1875 (is_simple_builtin): Likewise.
1876 * calls.c (gimple_alloca_call_p): Likewise.
1877 (maybe_warn_nonstring_arg): Likewise.
1878 * cfgexpand.c (expand_call_stmt): Likewise.
1879 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
1880 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
1881 (cgraph_node::verify_node): Likewise.
1882 * cgraphclones.c (build_function_decl_skip_args): Likewise.
1883 (cgraph_node::create_clone): Likewise.
1884 * config/arm/arm.c (arm_insert_attributes): Likewise.
1885 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
1886 * dse.c (scan_insn): Likewise.
1887 * expr.c (expand_expr_real_1): Likewise.
1888 * fold-const.c (operand_equal_p): Likewise.
1889 (fold_binary_loc): Likewise.
1890 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1891 * gimple-low.c (lower_stmt): Likewise.
1892 * gimple-pretty-print.c (dump_gimple_call): Likewise.
1893 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Likewise.
1894 * gimple.c (gimple_build_call_from_tree): Likewise.
1895 (gimple_call_builtin_p): Likewise.
1896 (gimple_call_combined_fn): Likewise.
1897 * gimplify.c (gimplify_call_expr): Likewise.
1898 (gimple_boolify): Likewise.
1899 (gimplify_modify_expr): Likewise.
1900 (gimplify_addr_expr): Likewise.
1901 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
1902 * ipa-cp.c (determine_versionability): Likewise.
1903 * ipa-fnsummary.c (compute_fn_summary): Likewise.
1904 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
1905 * ipa-split.c (visit_bb): Likewise.
1906 (split_function): Likewise.
1907 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
1908 * lto-cgraph.c (input_node): Likewise.
1909 * lto-streamer-out.c (write_symbol): Likewise.
1910 * omp-low.c (setjmp_or_longjmp_p): Likewise.
1911 (lower_omp_1): Likewise.
1912 * predict.c (strip_predict_hints): Likewise.
1913 * print-tree.c (print_node): Likewise.
1914 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
1915 * trans-mem.c (is_tm_irrevocable): Likewise.
1916 (is_tm_load): Likewise.
1917 (is_tm_simple_load): Likewise.
1918 (is_tm_store): Likewise.
1919 (is_tm_simple_store): Likewise.
1920 (is_tm_abort): Likewise.
1921 (tm_region_init_1): Likewise.
1922 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
1923 * tree-cfg.c (verify_gimple_call): Likewise.
1924 (move_stmt_r): Likewise.
1925 (stmt_can_terminate_bb_p): Likewise.
1926 * tree-eh.c (lower_eh_constructs_2): Likewise.
1927 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
1928 * tree-inline.c (remap_gimple_stmt): Likewise.
1929 (copy_bb): Likewise.
1930 (estimate_num_insns): Likewise.
1931 (fold_marked_statements): Likewise.
1932 * tree-sra.c (scan_function): Likewise.
1933 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
1934 (optimize_stack_restore): Likewise.
1935 (pass_fold_builtins::execute): Likewise.
1936 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1937 (mark_all_reaching_defs_necessary_1): Likewise.
1938 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
1939 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
1940 (pass_forwprop::execute): Likewise.
1941 * tree-ssa-loop-im.c (stmt_cost): Likewise.
1942 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1943 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
1944 * tree-ssa-strlen.c (get_string_length): Likewise.
1945 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
1946 (find_func_aliases_for_call): Likewise.
1947 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
1948 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
1949 * tree-tailcall.c (find_tail_calls): Likewise.
1950 * tree.c (need_assembler_name_p): Likewise.
1951 (free_lang_data_in_decl): Likewise.
1952 (get_call_combined_fn): Likewise.
1953 * ubsan.c (is_ubsan_builtin_p): Likewise.
1954 * varasm.c (incorporeal_function_p): Likewise.
1955 * tree.h (DECL_BUILT_IN): Remove and replace with
1957 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
1958 (fndecl_built_in_p): New.
1960 2018-08-27 Martin Liska <mliska@suse.cz>
1962 PR tree-optimization/86847
1963 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
1964 Dump also subtree probability.
1965 (switch_decision_tree::do_jump_if_equal): New function.
1966 (switch_decision_tree::emit_case_nodes): Handle special
1967 situations in balanced tree that can be emitted much simpler.
1968 Fix calculation of probabilities that happen in tree expansion.
1969 * tree-switch-conversion.h (struct cluster): Add
1971 (struct simple_cluster): Likewise.
1972 (struct case_tree_node): Add new function has_child.
1973 (do_jump_if_equal): New.
1975 2018-08-27 Martin Liska <mliska@suse.cz>
1977 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
1978 Add new argument to bit_test_cluster constructor.
1979 (bit_test_cluster::emit): Set bits really number of values
1981 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
1982 probability argument.
1983 * tree-switch-conversion.h (struct bit_test_cluster):
1984 Add m_handles_entire_switch.
1986 2018-08-27 Martin Liska <mliska@suse.cz>
1988 PR tree-optimization/86702
1989 * tree-switch-conversion.c (jump_table_cluster::emit):
1990 Make probabilities even for values in jump table
1991 according to number of cases handled.
1992 (switch_decision_tree::compute_cases_per_edge): Pass
1993 argument to reset_out_edges_aux function.
1994 (switch_decision_tree::analyze_switch_statement): Likewise.
1995 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
1998 2018-08-27 Martin Liska <mliska@suse.cz>
2000 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
2001 cfun argument explicitly.
2002 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
2003 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
2004 function gimple_switch_default_bb.
2005 (convert_switch_statements):
2007 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
2008 * stmt.c (label_to_block_fn): Use label_to_block and pass
2009 cfun argument explicitly and use gimple_switch_label_bb.
2010 (expand_case): Likewise.
2011 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
2012 cfun argument explicitly. Likewise.
2013 (make_edges_bb): Likewise.
2014 (make_cond_expr_edges): Likewise.
2015 (get_cases_for_edge): Likewise.
2016 (make_gimple_switch_edges): Likewise.
2017 (label_to_block_fn): Likewise.
2018 (label_to_block): Likewise.
2019 (make_goto_expr_edges): Likewise.
2020 (make_gimple_asm_edges): Likewise.
2021 (main_block_label): Likewise.
2022 (group_case_labels_stmt): Likewise.
2023 (find_taken_edge_computed_goto): Likewise.
2024 (find_taken_edge_switch_expr): Likewise.
2025 (gimple_verify_flow_info): Likewise.
2026 (gimple_redirect_edge_and_branch): Likewise.
2027 (gimple_switch_label_bb): New function.
2028 (gimple_switch_default_bb): Likewise.
2029 (gimple_switch_edge): Likewise.
2030 (gimple_switch_default_edge): Likewise.
2031 * tree-cfg.h (label_to_block_fn): Remove and replace ...
2032 (label_to_block): ... with this.
2033 (gimple_switch_label_bb): New.
2034 (gimple_switch_default_bb): Likewise.
2035 (gimple_switch_edge): Likewise.
2036 (gimple_switch_default_edge): Likewise.
2037 * tree-cfgcleanup.c (convert_single_case_switch): Use
2038 new gimple functions and pass new argument to label_to_block.
2039 (cleanup_control_flow_bb):
2040 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
2041 cfun argument explicitly.
2042 (make_eh_edges): Likewise.
2043 (redirect_eh_dispatch_edge): Likewise.
2044 (lower_resx): Likewise.
2045 (lower_eh_dispatch): Likewise.
2046 (maybe_remove_unreachable_handlers): Likewise.
2047 (unsplit_eh): Likewise.
2048 (cleanup_empty_eh): Likewise.
2049 (verify_eh_edges): Likewise.
2050 (verify_eh_dispatch_edge): Likewise.
2051 * tree-ssa-dom.c (record_edge_info): Likewise.
2052 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
2053 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
2054 (thread_through_normal_block): Likewise.
2055 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
2056 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
2057 * tree-switch-conversion.c (switch_conversion::collect): Use new
2059 (switch_conversion::check_final_bb): Likewise.
2060 (switch_conversion::gather_default_values): Pass new argument
2062 (switch_conversion::build_constructors): Likewise.
2063 (switch_decision_tree::compute_cases_per_edge): Use new
2064 gimple_switch_edge function.
2065 (switch_decision_tree::analyze_switch_statement): Pass new argument
2067 (switch_decision_tree::try_switch_expansion): Use
2068 gimple_switch_default_edge.
2069 * tree-vrp.c (find_switch_asserts): Pass new argument
2071 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
2072 (vr_values::simplify_switch_using_ranges): Likewise.
2074 2018-08-27 Richard Biener <rguenther@suse.de>
2076 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
2077 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
2079 * tree-ssa-sccvn.h (struct vn_pval): New structure.
2080 (struct vn_nary_op_s): Add unwind_to member. Add
2081 predicated_values flag and put result into a union together
2082 with a linked list of vn_pval.
2083 (struct vn_ssa_aux): Add name member to make maintaining
2084 a map of SSA name to vn_ssa_aux possible. Remove no longer
2085 needed info, dfsnum, low, visited, on_sccstack, use_processed
2086 and range_info_anti_range_p members.
2087 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
2088 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
2090 (vn_valueize): New global.
2091 (vn_context_bb): Likewise.
2092 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
2093 VN_INFO_PTR_INFO): Remove.
2094 * tree-ssa-sccvn.c: ... (rewrite)
2095 (pass_fre::execute): For -O2+ initialize loops and run
2096 RPO VN in optimistic mode (iterating). For -O1 and -Og
2097 run RPO VN in non-optimistic mode.
2098 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
2099 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
2100 * doc/invoke.texi (sccvn-max-scc-size): Remove.
2101 (rpo-vn-max-loop-depth): Document.
2102 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
2103 when valuezing the VUSE signals we walked out of the region.
2104 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
2105 (phi_translate): Set VN context block to use for availability
2107 (compute_avail): Likewise.
2108 (pre_valueize): New function.
2109 (pass_pre::execute): Adjust to the RPO VN API.
2111 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
2112 (propagate_constants_for_unrolling): Remove.
2113 (tree_unroll_loops_completely): Perform value-numbering
2114 on the unrolled bodies loop parent.
2116 2018-08-27 Richard Biener <rguenther@suse.de>
2118 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
2119 for partial antic compute.
2121 2018-08-27 Jakub Jelinek <jakub@redhat.com>
2123 PR rtl-optimization/87065
2124 * combine.c (simplify_if_then_else): Formatting fix.
2125 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
2127 (known_cond): Don't return const_true_rtx for vector modes. Use
2128 CONST0_RTX instead of const0_rtx. Formatting fixes.
2130 2018-08-27 Martin Liska <mliska@suse.cz>
2132 PR gcov-profile/87069
2133 * gcov.c (process_file): Record files already processed
2134 and warn about a file being processed multiple times.
2136 2018-08-27 Martin Liska <mliska@suse.cz>
2139 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2140 Set default values for x_aarch64_*_string strings.
2141 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
2142 prefix. For -mabi do not print '=ABI' in help and use
2143 <option_value> format for -msve-vector-bits and -moverride
2146 2018-08-26 Jeff Law <law@redhat.com>
2148 * config/mips/frame-header-opt.c: Include "backend.h" rather than
2151 2018-08-26 Marek Polacek <polacek@redhat.com>
2153 PR c++/87029, Implement -Wredundant-move.
2154 * doc/invoke.texi: Document -Wredundant-move.
2156 2018-08-25 Martin Sebor <msebor@redhat.com>
2158 PR tree-optimization/87059
2159 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
2160 to the same type as the other.
2161 * fold-const.c (fold_binary_loc): Assert expectation.
2163 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
2165 * config/darwin.c (machopic_legitimize_pic_address): Clean up
2166 extraneous parentheses, dead code section and formatting.
2168 2018-08-24 David Malcolm <dmalcolm@redhat.com>
2171 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
2172 wide enough for jumps in the line-numbering to be visible.
2173 (layout::print_gap_in_line_numbering): New member function.
2174 (layout::calculate_line_spans): When using line numbering, merge
2175 line spans that are only 1 line apart.
2176 (diagnostic_show_locus): When printing line numbers, show gaps in
2177 line numbering directly, rather than printing headers.
2178 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
2179 line-numbering with multiple line spans.
2180 (selftest::test_fixit_insert_containing_newline_2): Add test of
2181 line-numbering, in which the spans are close enough to be merged.
2183 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
2185 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
2186 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
2187 * tree-vrp.c (range_is_nonnull): Remove.
2188 (range_includes_zero_p): Accept value_range instead of min/max.
2189 (extract_range_from_binary_expr_1): Do not early bail on
2191 Use range_includes_zero_p instead of range_is_nonnull.
2192 (extract_range_from_unary_expr): Use range_includes_zero_p instead
2193 of range_is_nonnull.
2194 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
2195 special case VR_ANTI_RANGE.
2196 (vrp_finalize): Same.
2197 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
2199 (range_is_nonnull): Remove.
2200 * vr-values.c (vrp_stmt_computes_nonzero): Use
2201 range_includes_zero_p instead of range_is_nonnull.
2202 (extract_range_basic): Pass value_range to range_includes_zero_p
2203 instead of range_is_nonnull.
2205 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
2207 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
2208 * emit-rtl.h (rtl_data): Remove return_bnd.
2209 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
2210 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
2211 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
2212 (POINTER_BOUNDS_MODE): Remove definition.
2213 (make_pointer_bounds_mode): Remove.
2214 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
2215 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
2216 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
2217 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
2218 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
2219 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
2220 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
2222 * config/i386/i386-modes.def (BND32, BND64): Remove.
2223 * config/i386/i386.c (dbx_register_map): Remove bound registers.
2224 (dbx64_register_map): Ditto.
2225 (svr4_dbx_register_map): Ditto.
2226 (indirect_thunk_bnd_needed): Remove.
2227 (indirect_thunks_bnd_used): Ditto.
2228 (indirect_return_bnd_needed): Ditto.
2229 (indirect_return_via_cx_bnd): Ditto.
2230 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
2231 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
2232 (output_indirect_thunk): Ditto. Remove need_prefix argument.
2233 (output_indirect_thunk_function): Remove handling of
2234 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
2235 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
2236 (ix86_save_reg): Remove handling of crtl->return_bnd.
2237 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
2238 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
2239 and UNSPEC_BNDLX_ADDR.
2240 (ix86_output_indirect_branch_via_reg): Remove handling of
2241 indirect_thunk_prefix_bnd.
2242 (ix86_output_indirect_branch_via_push): Ditto.
2243 (ix86_output_function_return): Ditto.
2244 (ix86_output_indirect_function_return): Ditto.
2245 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
2246 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
2247 (CALL_USED_REGISTERS): Ditto.
2248 (REG_ALLOC_ORDER): Update for removal of bound registers.
2249 (HI_REGISTER_NAMES): Ditto.
2250 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
2251 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
2252 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
2253 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
2254 (FIRST_PSEUDO_REG): Update.
2255 (BND): Remove mode iterator.
2256 * config/i386/predicates.md (bnd_mem_operator): Remove.
2258 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
2260 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
2263 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
2265 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
2266 the case in which the permute needs only a single element and
2267 repeats for every vector of the result. Extend that case to
2268 handle variable-length vectors.
2269 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
2271 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
2274 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
2275 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
2277 2018-08-24 Richard Biener <rguenther@suse.de>
2279 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
2280 bb_flags_allocated members.
2281 (auto_flag): New RAII class for allocating flags.
2282 (auto_edge_flag): New RAII class for allocating edge flags.
2283 (auto_bb_flag): New RAII class for allocating bb flags.
2284 * cfgloop.c (verify_loop_structure): Allocate temporary edge
2286 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
2287 in favor of temporarily allocated BB flag.
2288 * hsa-brig.c: Re-order includes.
2289 * hsa-dump.c: Likewise.
2290 * hsa-regalloc.c: Likewise.
2291 * print-rtl.c: Likewise.
2292 * profile-count.c: Likewise.
2294 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
2297 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
2300 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
2303 * wide-int-range.cc (wide_int_range_div): Do not ignore result
2304 from wide_int_range_multiplicative_op.
2306 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2308 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
2311 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
2313 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
2316 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
2318 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
2319 full GENERIC expression used for replacement.
2321 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
2323 * tree-vrp.c (abs_extent_range): Remove.
2324 (extract_range_into_wide_ints): Pass wide ints by reference.
2325 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
2326 Pass wide ints by reference in all calls to
2327 extract_range_into_wide_ints.
2328 * wide-int-range.cc (wide_int_range_div): New.
2329 * wide-int-range.h (wide_int_range_div): New.
2330 (wide_int_range_includes_zero_p): New.
2331 (wide_int_range_zero_p): New.
2333 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
2335 * config/aarch64/aarch64.md (arches): New enum.
2336 (arch): New enum attr.
2337 (arch_enabled): New attr.
2338 (enabled): Now uses arch_enabled only.
2339 (simd, sve, fp16): Removed attribute.
2340 (fp): Attr now defined in terms of 'arch'.
2341 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
2342 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
2343 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
2344 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
2345 attributes into 'arch'.
2346 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
2347 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
2348 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
2349 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
2350 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
2351 'simd' attribute into 'arch'.
2352 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
2353 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
2354 Convert use of 'fp' attribute to 'arch'.
2355 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
2356 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
2358 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
2359 (different modes) Merge 'fp' and 'simd' into 'arch'.
2360 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
2363 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
2365 PR rtl-optimization/87026
2366 * expmed.c (canonicalize_comparison): If we can no longer create
2367 pseudoregisters, don't.
2369 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
2372 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
2374 * config/arm/arm.c (speculation_barrier_libfunc): New static
2376 (arm_init_libfuncs): Initialize it.
2377 (arm_emit_speculation_barrier): New function.
2378 * config/arm/arm.md (speculation_barrier): Call
2379 arm_emit_speculation_barrier for architectures that do not have
2381 (speculation_barrier_insn): Only match on Armv7 or later.
2383 2018-08-23 Richard Biener <rguenther@suse.de>
2386 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
2389 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
2391 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
2392 of single-vector TBLs.
2393 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
2396 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
2399 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
2400 aarch64_evpc_tbl guard.
2402 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
2404 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
2407 2018-08-22 Martin Sebor <msebor@redhat.com>
2410 * tree-pretty-print.c (pretty_print_string): Add argument.
2411 (dump_generic_node): Call to pretty_print_string with string size.
2413 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
2415 PR rtl-optimization/86771
2416 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
2417 of two SETs into those two SETs, one to be placed at i2, if that SETs
2418 destination is modified between i2 and i3.
2420 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2422 PR tree-optimization/86725
2423 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
2425 (vect_analyze_scalar_cycles_1): Check it.
2427 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2429 PR tree-optimization/86725
2430 * tree-vect-loop.c (vect_is_simple_reduction): When treating
2431 an outer loop phi as a double reduction, make sure that the
2432 single user of the phi result is an inner loop phi.
2434 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2436 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
2437 grouped stores with gaps to a strided group.
2439 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2441 * tree-vect-stmts.c (get_group_load_store_type)
2442 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
2443 first statement in a group.
2445 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2447 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
2448 the sequence used in gcc/gcc.c.
2450 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2453 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
2456 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2458 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
2459 Darwin10-specific unwinder-shim.
2460 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
2461 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
2462 New to cater for Darwin10 Rosetta.
2464 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2466 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
2467 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
2469 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2474 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
2475 (dwarf2out_switch_text_section): Generate a local label for the
2476 second function sub-section and apply it as the second FDE start
2478 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
2479 second sub-section start.
2481 2018-08-22 Richard Biener <rguenther@suse.de>
2483 PR tree-optimization/86988
2484 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
2486 2018-08-22 Richard Biener <rguenther@suse.de>
2488 PR tree-optimization/86945
2489 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
2491 2018-08-22 Alexandre Oliva <oliva@adacore.com>
2493 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
2494 a comment about how uses of r2 for .sdata2 come about.
2496 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
2498 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
2500 2018-08-21 Marek Polacek <polacek@redhat.com>
2502 PR c++/86981, Implement -Wpessimizing-move.
2503 * doc/invoke.texi: Document -Wpessimizing-move.
2505 2018-08-21 Jan Hubicka <jh@suse.cz>
2507 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
2508 * tree.h (is_redundant_typedef): Remove.
2509 * dwarf2out.c (is_redundant_typedef): Turn into static function.
2511 2018-08-21 Jan Hubicka <jh@suse.cz>
2513 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
2516 2018-08-21 Tamar Christina <tamar.christina@arm.com>
2518 * expmed.c (extract_low_bits): Reject invalid subregs early.
2520 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2523 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
2526 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2528 * config/vxworks.h: Guard vxworks_asm_out_constructor and
2529 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
2530 * config/vxworks.c: Likewise.
2532 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2534 * config/vxworks.c: Set targetm.have_ctors_dtors
2535 if HAVE_INITFINI_ARRAY_SUPPORT.
2536 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
2537 if HAVE_INITFINI_ARRAY_SUPPORT.
2539 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2541 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
2542 default search path for VxWorks < 7.
2544 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2546 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
2547 (get_format_string): Refer to c_getstr.
2549 2018-08-21 Tom de Vries <tdevries@suse.de>
2551 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
2552 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
2553 (debuginfo_early_stop): Declare.
2554 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
2555 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
2556 (debuginfo_early_stop): New function.
2557 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
2558 and debuginfo_early_stop.
2559 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
2560 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
2561 (general_init): Call debuginfo_early_init.
2562 (finalize): Call debuginfo_fini.
2563 (do_compile): Call debuginfo_init.
2564 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
2566 (@item -fdump-debug, @item -fdump-earlydebug): Add.
2568 2018-08-21 Tom de Vries <tdevries@suse.de>
2570 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
2571 flag_dump_noaddr and flag_dump_unnumbered.
2573 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
2575 * wide-int-range.cc (wide_int_range_abs): New.
2576 (wide_int_range_order_set): Rename from wide_int_range_min_max.
2577 * wide-int-range.h (wide_int_range_abs): New.
2578 (wide_int_range_min_max): New.
2579 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
2580 case to call wide_int_range_abs.
2581 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
2582 (extract_range_from_abs_expr): Delete.
2584 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
2587 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
2588 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
2591 2018-08-20 Nathan Sidwell <nathan@acm.org>
2592 Jeff Law <law@redhat.com>
2594 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
2595 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
2597 2018-08-20 David Malcolm <dmalcolm@redhat.com>
2600 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
2601 (decl_attributes): Likewise.
2602 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
2604 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
2605 * diagnostic-core.h (class auto_diagnostic_group): New class.
2606 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
2607 (diagnostic_report_diagnostic): Handle the first diagnostics within
2609 (emit_diagnostic): Add auto_diagnostic_group instance.
2611 (inform_n): Likewise.
2612 (warning): Likewise.
2613 (warning_at): Likewise.
2614 (warning_n): Likewise.
2615 (pedwarn): Likewise.
2616 (permerror): Likewise.
2618 (error_n): Likewise.
2619 (error_at): Likewise.
2621 (fatal_error): Likewise.
2622 (internal_error): Likewise.
2623 (internal_error_no_backtrace): Likewise.
2624 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
2625 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
2626 * diagnostic.h (struct diagnostic_context): Add fields
2627 "diagnostic_group_nesting_depth",
2628 "diagnostic_group_emission_count", "begin_group_cb",
2630 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
2631 Add auto_diagnostic_group instance(s).
2632 (find_explicit_erroneous_behavior): Likewise.
2633 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
2634 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
2635 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
2636 (gimplify_va_arg_expr): Likewise.
2637 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
2638 (HSA_SORRY_AT): Likewise.
2639 * ipa-devirt.c (compare_virtual_tables): Likewise.
2640 (warn_odr): Likewise.
2641 * multiple_target.c (expand_target_clones): Likewise.
2642 * opts-common.c (cmdline_handle_error): Likewise.
2643 * reginfo.c (globalize_reg): Likewise.
2644 * substring-locations.c (format_warning_n_va): Likewise.
2645 * tree-inline.c (expand_call_inline): Likewise.
2646 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
2647 * tree-ssa-loop-niter.c
2648 (do_warn_aggressive_loop_optimizations): Likewise.
2649 * tree-ssa-uninit.c (warn_uninit): Likewise.
2650 * tree.c (warn_deprecated_use): Likewise.
2652 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
2655 * config/i386/i386.md (eh_return): Always update EH return
2656 address in word_mode.
2658 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
2660 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
2661 TARGET_SPLIT_COMPLEX_ARG is defined.
2663 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2665 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
2667 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2670 * expr.c (expand_assignment): Assert that bitpos is positive.
2671 (store_field): Likewise
2672 (expand_expr_real_1): Make sure that bitpos is positive.
2673 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
2676 2018-08-20 Nathan Sidwell <nathan@acm.org>
2678 * Makefile.in (CPP_ID_DATA_H): Delete.
2679 (CPP_INTERNAL_H): Don't add it.
2680 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
2681 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
2683 2018-08-20 Richard Biener <rguenther@suse.de>
2685 PR tree-optimization/78655
2686 * tree-vrp.c (extract_range_from_binary_expr_1): Make
2687 pointer + offset nonnull if either operand is nonnull work.
2689 2018-08-20 Tom de Vries <tdevries@suse.de>
2691 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
2692 unless the referenced die describes the added property using
2693 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
2694 Otherwise, add a DW_AT_location to the referenced die.
2696 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
2699 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
2700 register_operand when calling ix86_set_reg_reg_cost.
2701 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
2702 Set *total to 0 for operands that satisfy x86_64_immediate_operand
2703 predicate and to 1 otherwise.
2705 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
2707 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
2708 emit a diagnostic that it is not supported and reset the option.
2709 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
2710 supported and consume the option. (ASM_FINAL_SPEC): New.
2712 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2714 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
2717 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
2719 C-SKY port: Documentation
2721 * doc/extend.texi (C-SKY Function Attributes): New section.
2722 * doc/invoke.texi (Option Summary): Add C-SKY options.
2723 (C-SKY Options): New section.
2724 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
2726 2018-08-17 Jojo <jijie_rong@c-sky.com>
2727 Huibin Wang <huibin_wang@c-sky.com>
2728 Sandra Loosemore <sandra@codesourcery.com>
2729 Chung-Lin Tang <cltang@codesourcery.com>
2731 C-SKY port: Backend implementation
2733 * config/csky/*: New.
2734 * common/config/csky/*: New.
2736 2018-08-17 Jojo <jijie_rong@c-sky.com>
2737 Huibin Wang <huibin_wang@c-sky.com>
2738 Sandra Loosemore <sandra@codesourcery.com>
2739 Chung-Lin Tang <cltang@codesourcery.com>
2740 Andrew Jenner <andrew@codesourcery.com>
2742 C-SKY port: Configury
2744 * config.gcc (csky-*-*): New.
2745 * configure.ac: Add csky to targets for dwarf2 debug_line support.
2746 * configure: Regenerated.
2748 2018-08-17 David Malcolm <dmalcolm@redhat.com>
2750 * dump-context.h: Include "dumpfile.h".
2751 (dump_context::dump_printf_va): Convert final param from va_list
2752 to va_list *. Convert from ATTRIBUTE_PRINTF to
2753 ATTRIBUTE_GCC_DUMP_PRINTF.
2754 (dump_context::dump_printf_loc_va): Likewise.
2755 * dumpfile.c: Include "stringpool.h".
2756 (make_item_for_dump_printf_va): Delete.
2757 (make_item_for_dump_printf): Delete.
2758 (class dump_pretty_printer): New class.
2759 (dump_pretty_printer::dump_pretty_printer): New ctor.
2760 (dump_pretty_printer::emit_items): New member function.
2761 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
2763 (dump_pretty_printer::emit_item): New member function.
2764 (dump_pretty_printer::stash_item): New member function.
2765 (dump_pretty_printer::format_decoder_cb): New member function.
2766 (dump_pretty_printer::decode_format): New member function.
2767 (dump_context::dump_printf_va): Reimplement in terms of
2768 dump_pretty_printer.
2769 (dump_context::dump_printf_loc_va): Convert final param from va_list
2771 (dump_context::begin_scope): Reimplement call to
2772 make_item_for_dump_printf.
2773 (dump_printf): Update for change to dump_printf_va.
2774 (dump_printf_loc): Likewise.
2775 (selftest::test_capture_of_dump_calls): Convert "stmt" from
2776 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
2777 with %T, %E, and %G.
2778 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
2779 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
2780 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
2781 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
2782 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
2783 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
2784 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
2785 within a dump_printf_loc call to "%wu".
2786 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
2787 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
2788 missing space after "=".
2789 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
2790 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
2791 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
2792 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
2793 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
2794 duplicate "vectorized" from message.
2796 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
2798 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
2799 polyNxK_t element's TYPE_STRING_FLAG.
2801 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2803 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
2804 (they were unnamed before). Fix comments.
2806 2018-08-17 Nathan Sidwell <nathan@acm.org>
2808 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
2810 2018-08-17 Richard Biener <rguenther@suse.de>
2812 PR tree-optimization/86841
2813 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
2815 2018-08-17 Martin Liska <mliska@suse.cz>
2817 * common.opt: Remove Warn, Init and Report for options with
2818 Ignore/Deprecated flag. Warning is done automatically for
2820 * config/i386/i386.opt: Likewise.
2821 * config/ia64/ia64.opt: Likewise.
2822 * config/rs6000/rs6000.opt: Likewise.
2823 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
2824 Remove usage of flag_check_pointer_bounds.
2825 * lto-wrapper.c (merge_and_complain): Do not handle
2826 OPT_fcheck_pointer_bounds.
2827 (append_compiler_options): Likewise.
2828 * opt-functions.awk: Do not handle Deprecated.
2829 * optc-gen.awk: Check that Var, Report and Init are not
2830 used for an option with Ignore/Deprecated flag.
2831 * opts-common.c (decode_cmdline_option): Do not report
2833 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
2835 * opts.h (struct cl_option): Remove cl_deprecated flag.
2836 (CL_ERR_DEPRECATED): Remove error enum value.
2838 2018-08-17 Richard Biener <rguenther@suse.de>
2841 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
2842 across a va-arg-pack using call adjust its return value accordingly.
2844 2018-08-16 Martin Sebor <msebor@redhat.com>
2846 PR tree-optimization/86853
2847 * gimple-ssa-sprintf.c (struct format_result): Rename member.
2848 (struct fmtresult): Add member and initialize it in ctors.
2849 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
2850 (format_string): Handle %S the same as %ls. Set MAYFAIL.
2851 (format_directive): Set POSUNDER4K when MAYFAIL is set.
2852 (parse_directive): Handle %C same as %c.
2853 (sprintf_dom_walker::compute_format_length): Adjust.
2854 (is_call_safe): Adjust.
2856 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
2858 * builtins.c (c_strlen): Add new parameter eltsize. Use it
2859 for determining how to count the elements.
2860 * builtins.h (c_strlen): Adjust prototype.
2861 * expr.c (string_constant): Add new parameter mem_size.
2862 Set *mem_size appropriately.
2863 * expr.h (string_constant): Adjust protoype.
2864 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
2865 * gimple-fold.h (get_range_strlen): Adjust prototype.
2866 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
2867 (format_string): Call get_string_length with eltsize.
2869 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2871 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
2872 to emit the span, rather than setting it as the prefix.
2874 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2876 * diagnostic-show-locus.c (layout::start_annotation_line): Add
2877 "margin_char" parameter, defaulting to space. Use it in place
2878 of pp_space for the initial part of the margin.
2879 (layout::print_leading_fixits): Use '+' when filling the margin
2880 of line-insertion fix-it hints.
2882 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2884 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
2887 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2889 * config/rs6000/altivec.md: Don't set length attribute to the default
2891 * config/rs6000/darwin.md: Ditto.
2892 * config/rs6000/dfp.md: Ditto.
2893 * config/rs6000/htm.md: Ditto.
2894 * config/rs6000/rs6000.md: Ditto.
2895 * config/rs6000/sync.md: Ditto.
2896 * config/rs6000/vsx.md: Ditto.
2898 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2900 * config/rs6000/altivec.md: Don't set length attribute to the default
2901 value, for branch instructions.
2902 * config/rs6000/darwin.md: Ditto.
2903 * config/rs6000/rs6000.md: Ditto.
2905 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2907 * config/rs6000/rs6000.md (length): Always define as const_int 4.
2908 (unnamed conditional branch define_insn): Set length to 4 or 8
2909 depending on offset.
2910 (<bd>_<mode>): Similar, for alternative 0.
2911 (<bd>tf_<mode>): Ditto.
2913 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2915 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
2917 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
2919 * doc/rtl.texi: Replace old RTX class names with new names.
2922 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
2924 * expmed.h (canonicalize_comparison): New declaration.
2925 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
2926 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
2927 * optabs.c (prepare_cmp_insn): Likewise.
2928 * rtl.h (unsigned_condition_p): New function which checks if a
2929 comparison operator is unsigned.
2931 2018-08-16 Nathan Sidwell <nathan@acm.org>
2933 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
2934 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
2936 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2939 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
2940 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
2941 (mov<mov>): ..this and enable unconditionally.
2943 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2945 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
2947 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
2949 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
2950 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
2951 "Common" with "Target".
2953 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2955 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
2956 * doc/invoke.texi (mmitigate-rop): Remove.
2957 * config/i386/i386.c: Do not include "regrename.h".
2958 (ix86_rop_should_change_byte_p, reg_encoded_number)
2959 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
2961 (ix86_reorg): Remove call to ix86_mitigate_rop.
2962 * config/i386/i386.md (attr "modrm_class"): Remove.
2963 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
2964 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
2965 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
2967 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2969 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
2970 allow folding of mergeh() and mergel() for the float and double types.
2971 (fold_mergehl_helper): Rework to handle building a permute tree
2974 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2976 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
2979 2018-08-15 David Malcolm <dmalcolm@redhat.com>
2981 * common.opt (fdiagnostics-show-labels): New option.
2982 * diagnostic-show-locus.c (class layout_range): Add field
2984 (class layout): Add field "m_show_labels_p".
2985 (layout_range::layout_range): Add param "label" and use it to
2987 (make_range): Pass in NULL for new "label" param of layout_range's
2989 (layout::layout): Initialize m_show_labels_p.
2990 (layout::maybe_add_location_range): Pass in loc_range->m_label
2991 when constructing layout_range instances.
2992 (struct line_label): New struct.
2993 (layout::print_any_labels): New member function.
2994 (layout::print_line): Call it if label-printing is enabled.
2995 (selftest::test_one_liner_labels): New test.
2996 (selftest::test_diagnostic_show_locus_one_liner): Call it.
2997 * diagnostic.c (diagnostic_initialize): Initialize
2998 context->show_labels_p.
2999 * diagnostic.h (struct diagnostic_context): Add field
3001 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
3002 -fno-diagnostics-show-labels.
3003 * dwarf2out.c (gen_producer_string): Add
3004 OPT_fdiagnostics_show_labels to the ignored options.
3005 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
3007 (gcc_rich_location::maybe_add_expr): Likewise.
3008 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
3009 label" param, defaulting to NULL.
3010 (gcc_rich_location::add_expr): Add "label" param.
3011 (gcc_rich_location::maybe_add_expr): Likewise.
3012 (class text_range_label): New class.
3013 (class range_label_for_type_mismatch): New class.
3014 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
3015 of format_warning_va.
3016 (fmtwarn_n): Likewise for new params of format_warning_n_va.
3017 * lto-wrapper.c (merge_and_complain): Add
3018 OPT_fdiagnostics_show_labels to the "pick one setting" options.
3019 (append_compiler_options): Likewise to the dropped options.
3020 (append_diag_options): Likewise to the passed-on options.
3021 * opts.c (common_handle_option): Handle the new option.
3022 * selftest-diagnostic.c
3023 (test_diagnostic_context::test_diagnostic_context): Enable
3025 * substring-locations.c: Include "gcc-rich-location.h".
3026 (format_warning_n_va): Add "fmt_label" and "param_label" params
3027 and use them as appropriate.
3028 (format_warning_va): Add "fmt_label" and "param_label" params,
3029 passing them on to format_warning_n_va.
3030 (format_warning_at_substring): Likewise.
3031 (format_warning_at_substring_n): Likewise.
3032 * substring-locations.h (format_warning_va): Add "fmt_label" and
3033 "param_label" params.
3034 (format_warning_n_va): Likewise.
3035 (format_warning_at_substring): Likewise.
3036 (format_warning_at_substring_n): Likewise.
3037 * toplev.c (general_init): Initialize global_dc->show_labels_p.
3039 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
3042 * builtins.c (expand_builtin_memcmp): Do not expand the call
3043 when overflow is detected.
3045 2018-08-15 Martin Sebor <msebor@redhat.com>
3047 PR tree-optimization/71625
3048 * config/aarch64/aarch64-builtins.c
3049 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
3051 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
3053 * config/s390/s390.c (s390_reorg): Remove loop.
3055 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
3058 (darwin_function_switched_text_sections): Delete.
3059 * gcc/config/darwin.h
3060 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
3062 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
3065 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
3066 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
3067 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
3069 2018-08-15 Martin Liska <mliska@suse.cz>
3071 PR tree-optimization/86925
3072 * predict.c (expr_expected_value_1): When taking
3073 later predictor, assign also probability.
3074 Use fold_build2_initializer_loc in order to fold
3075 the expression in -frounding-math.
3077 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
3079 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
3081 (expand_vec_perm_1): Try the new method.
3083 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
3086 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
3087 Check whether lra_live_max_point is 0 before dividing.
3089 2018-08-14 Martin Sebor <msebor@redhat.com>
3091 PR tree-optimization/86650
3092 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
3093 (vrp_prop::check_mem_ref): Same.
3095 2018-08-13 Liu Hao <lh_mouse@126.com>
3097 * pretty-print.c (eat_esc_sequence): Swap the foreground and
3098 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
3099 and clear it thereafter, as it only works for DBCS.
3101 2018-08-13 Liu Hao <lh_mouse@126.com>
3103 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
3104 handle returned by _get_osf_handle().
3106 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
3108 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
3109 for folding vec_perm.
3111 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
3113 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
3114 Add support for gimple-folding of vec_pack() and vec_unpack()
3117 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
3119 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
3120 vec_xst variants to the list.
3121 (rs6000_gimple_fold_builtin): Add support for folding unaligned
3122 vector loads and stores.
3124 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
3126 * config.gcc (rs6000-ibm-aix4.x): Delete.
3127 (rs6000-ibm-aix5.1): Delete.
3128 (rs6000-ibm-aix5.2): Delete.
3129 (rs6000-ibm-aix5.3): Delete.
3130 * config/rs6000/aix43.h: Delete.
3131 * config/rs6000/aix51.h: Delete.
3132 * config/rs6000/aix52.h: Delete.
3133 * config/rs6000/t-aix43: Delete.
3135 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
3137 * config/s390/s390.c (s390_decompose_constant_pool_ref):
3139 (s390_decompose_address): Factor out constant pool ref
3142 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3144 * config/nds32/nds32-predicates.c
3145 (nds32_can_use_bclr_p): Change return type as bool.
3146 (nds32_can_use_bset_p): Ditto.
3147 (nds32_can_use_btgl_p): Ditto.
3148 (nds32_can_use_bitci_p): Ditto.
3149 * config/nds32/nds32-protos.h
3150 (nds32_can_use_bclr_p): Change declaration.
3151 (nds32_can_use_bset_p): Ditto.
3152 (nds32_can_use_btgl_p): Ditto.
3153 (nds32_can_use_bitci_p): Ditto.
3155 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3157 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
3158 Support -msched-prolog-epilog option.
3159 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
3161 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3163 * common/config/nds32/nds32-common.c
3164 (nds32_option_optimization_table): Enalbe -malways-align.
3166 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3168 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
3170 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
3171 OPT_misr_secure_ case.
3172 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
3173 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
3174 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
3176 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
3177 (nds32_isr_info): New field security_level.
3178 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
3179 * config/nds32/nds32.md (return_internal): Consider critical attribute.
3180 * config/nds32/nds32.opt (misr-secure): New option.
3181 * config/nds32/nds32_init.inc: New file.
3182 * config/nds32/nds32_isr.h: New file.
3184 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
3186 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
3187 Update comment for atomic instructions.
3188 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
3189 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
3191 (atomic_loaddi): Revise fence expansion to only emit fence prior to
3192 load for __ATOMIC_SEQ_CST model.
3193 (atomic_loaddi_1): Remove float register target.
3194 (atomic_storedi): Handle CONST_INT values.
3195 (atomic_storedi_1): Remove float register source. Add special case
3197 (memory_barrier): New expander and insn.
3199 2018-08-11 Jakub Jelinek <jakub@redhat.com>
3201 PR tree-optimization/86835
3202 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
3203 new_stmt after def_gsi, make sure to insert new_square_stmt after
3204 that stmt, not 2 stmts before it.
3206 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
3209 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
3212 2018-08-10 Martin Liska <mliska@suse.cz>
3215 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
3217 * builtins.c (expand_builtin_expect_with_probability):
3219 (expand_builtin_expect_with_probability): New function.
3220 (build_builtin_expect_predicate): Add new argumnet probability
3221 for BUILT_IN_EXPECT_WITH_PROBABILITY.
3222 (fold_builtin_expect):
3225 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
3226 * builtins.h (fold_builtin_expect): Set new argument.
3227 * doc/extend.texi: Document __builtin_expect_with_probability.
3228 * doc/invoke.texi: Likewise.
3229 * gimple-fold.c (gimple_fold_call): Pass new argument.
3230 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
3231 also BUILT_IN_EXPECT_WITH_PROBABILITY.
3232 * predict.c (get_predictor_value): New function.
3233 (expr_expected_value): Add new argument probability. Assume
3234 that predictor and probability are always non-null.
3235 (expr_expected_value_1): Likewise. For __builtin_expect and
3236 __builtin_expect_with_probability set probability. Handle
3237 combination in binary expressions.
3238 (tree_predict_by_opcode): Simplify code by simply calling
3239 get_predictor_value.
3240 (pass_strip_predict_hints::execute): Add handling of
3241 BUILT_IN_EXPECT_WITH_PROBABILITY.
3242 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
3244 * tree.h (DECL_BUILT_IN_P): New function.
3246 2018-08-10 Martin Liska <mliska@suse.cz>
3248 PR tree-optimization/85799
3249 * passes.def: Add argument for pass_strip_predict_hints.
3250 * predict.c (class pass_strip_predict_hints): Add new argument
3252 (strip_predictor_early): New function.
3253 (pass_strip_predict_hints::execute): Call the function to
3255 (strip_predict_hints): New function.
3256 * predict.def: Fix comment.
3258 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
3260 * Makefile.in: Clarify which tm.texi to copy over to assert the
3261 right to grant a GFDL license for all.
3263 2018-08-09 Jeff Law <law@redhat.com>
3265 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
3268 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
3270 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
3273 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3275 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
3276 reductions for variable-length vectors.
3278 2018-08-09 David Malcolm <dmalcolm@redhat.com>
3281 * common.opt (fdiagnostics-show-line-numbers): New option.
3282 * diagnostic-show-locus.c (class layout): Add fields
3283 "m_show_line_numbers_p" and "m_linenum_width";
3284 (num_digits): New function.
3285 (test_num_digits): New function.
3286 (layout::layout): Initialize new fields. Update m_x_offset
3287 logic to handle any left margin.
3288 (layout::print_source_line): Print line number when requested.
3289 (layout::start_annotation_line): New member function.
3290 (layout::print_annotation_line): Call it.
3291 (layout::print_leading_fixits): Likewise.
3292 (layout::print_trailing_fixits): Likewise. Update calls to
3293 move_to_column for new parameter.
3294 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
3295 it to potentially call start_annotation_line.
3296 (layout::show_ruler): Call start_annotation_line.
3297 (selftest::test_line_numbers_multiline_range): New selftest.
3298 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
3299 and selftest::test_line_numbers_multiline_range.
3300 * diagnostic.c (diagnostic_initialize): Initialize
3301 show_line_numbers_p.
3302 * diagnostic.h (struct diagnostic_context): Add field
3303 "show_line_numbers_p".
3304 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
3305 -fno-diagnostics-show-line-numbers.
3306 * dwarf2out.c (gen_producer_string): Add
3307 OPT_fdiagnostics_show_line_numbers to the ignored options.
3308 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
3309 one setting" options.
3310 (append_compiler_options): Likewise to the dropped options.
3311 (append_diag_options): Likewise to the passed-on options.
3312 * opts.c (common_handle_option): Handle the new option.
3313 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
3315 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
3317 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
3318 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
3319 third argument of type "const signed char" to descriptions of
3320 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
3321 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
3322 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
3323 __builtin_bcdsub_ov functions.
3325 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3327 PR tree-optimization/86858
3328 * tree-vect-loop.c (vect_is_simple_reduction): Restore
3329 flow_bb_inside_loop_p calls.
3331 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3333 PR tree-optimization/86871
3334 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
3335 instead of gimple_assign_lhs.
3337 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
3340 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
3341 register constraint to operand 0.
3342 (add<mode>3_carryinC): Likewise.
3343 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
3345 2018-08-09 Martin Liska <mliska@suse.cz>
3348 * common.opt: Remove extra line.
3350 2018-08-09 Martin Liska <mliska@suse.cz>
3352 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
3353 at the end of a line, make first letter capital and end up
3354 a sentence with a dot.
3355 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
3356 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
3357 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
3358 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
3359 (PARAM_MAX_ISL_OPERATIONS): Likewise.
3360 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
3361 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
3362 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
3363 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
3364 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
3365 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
3366 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
3367 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
3368 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
3369 (PARAM_TREE_REASSOC_WIDTH): Likewise.
3370 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
3371 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
3372 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
3374 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
3377 * config/s390/s390.c (s390_option_override_internal): Reduce the
3378 stack-clash-protection-probe-interval param if it would be too big
3381 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
3384 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
3385 * config/m68k/m68k.c (handle_move_double): Don't call
3386 m68k_final_prescan_insn.
3387 (m68k_adjust_decorated_operand): Renamed from
3388 m68k_final_prescan_insn, remove first and third operand and
3390 (print_operand): Call it.
3391 (print_operand_address): Call it.
3393 2018-08-08 Nathan Sidwell <nathan@acm.org>
3395 * diagnostic.c (diagnostic_report_current_module): Use
3396 linemap_included_from & linemap_included_from_linemap.
3398 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
3400 * config/aarch64/aarch64-cores.def: Add phecda core.
3401 * config/aarch64/aarch64-tune.md: Regenerate.
3402 * doc/invoke.texi: Add phecda core.
3404 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
3407 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
3409 * config/s390/s390.md ("movti"): Add more alternatives for
3410 constant to GPR copies.
3412 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
3414 * config/s390/s390.c: Fix whitespace damage throughout the file.
3415 * config/s390/s390.h: Likewise.
3416 * config/s390/tpf.h: Likewise.
3418 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
3420 * config/s390/s390.c (s390_loadrelative_operand_p):
3421 Remove TARGET_CPU_ZARCH usages.
3422 (s390_rtx_costs): Likewise.
3423 (s390_legitimate_constant_p): Likewise.
3424 (s390_cannot_force_const_mem): Likewise.
3425 (legitimate_reload_constant_p): Likewise.
3426 (s390_preferred_reload_class): Likewise.
3427 (legitimize_pic_address): Likewise.
3428 (legitimize_tls_address): Likewise.
3429 (s390_split_branches): Removed.
3430 (s390_add_execute): Removed.
3431 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
3432 (s390_mainpool_start): Likewise.
3433 (s390_mainpool_finish): Likewise.
3434 (s390_mainpool_cancel): Removed.
3435 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
3436 (s390_chunkify_cancel): Likewise.
3437 (s390_return_addr_rtx): Likewise.
3438 (s390_register_info): Remove split_branches_pending_p uages.
3439 (s390_optimize_register_info): Likewise.
3440 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
3441 split_branches_pending_p usages.
3442 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
3443 (s390_load_got): Likewise.
3444 (s390_expand_split_stack_prologue): Likewise.
3445 (output_asm_nops): Likewise.
3446 (s390_function_profiler): Likewise.
3447 (s390_emit_call): Likewise.
3448 (s390_conditional_register_usage): Likewise.
3449 (s390_optimize_prologue): Likewise.
3450 (s390_reorg): Remove TARGET_CPU_ZARCH and
3451 split_branches_pending_p usages.
3452 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
3454 (s390_output_indirect_thunk_function): Likewise.
3455 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
3456 (TARGET_CPU_ZARCH_P): Removed.
3457 (struct machine_function): Remove split_branches_pending_p.
3458 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
3460 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
3462 * common/config/s390/s390-common.c (processor_flags_table):
3464 * config.gcc: Remove with_arch/with_tune support.
3465 * config/s390/2064.md: Remove cpu attribute comparisons.
3466 * config/s390/driver-native.c (s390_host_detect_local_cpu):
3468 * config/s390/linux.h (ASM_SPEC):
3469 Remove -march support.
3470 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
3471 Use a table to get an arch level.
3472 * config/s390/s390-opts.h (enum processor_type):
3474 * config/s390/s390.c
3475 (processor_table): Remove entries, add arch_level values.
3476 (s390_issue_rate): Remove cases.
3477 (s390_option_override): Adjust
3478 s390_option_override_internal() call.
3479 (s390_option_override_internal): Remove deprecation warning.
3480 (s390_valid_target_attribute_tree): Adjust
3481 s390_option_override_internal() call.
3482 * config/s390/s390.h (struct s390_processor):
3483 Share with s390-c.c, add arch_level field.
3484 * config/s390/s390.md:
3485 Remove occurrences in cpu attribute.
3486 * config/s390/s390.opt: Remove -march/-mtune support.
3487 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
3488 * doc/invoke.texi: Remove deprecation warning.
3490 2018-08-08 Luis Machado <luis.machado@linaro.org>
3492 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
3494 (qdf24xx_tunings): Set vector cost structure to
3495 qdf24xx_vector_cost.
3497 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
3498 <register_sextend>: Set to 3.
3500 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
3503 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
3504 * config/aarch64/aarch64-simd.md
3505 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
3506 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
3507 (aarch64_frecpx<mode>): ...this new pattern.
3508 * config/aarch64/aarch64-simd-builtins.def: Remove comment
3509 about aarch64_frecp<FRECP:frecp_suffix><mode>.
3511 2018-08-07 Martin Liska <mliska@suse.cz>
3514 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
3515 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
3516 * predict.def (PRED_MALLOC_NONNULL): New predictor.
3517 * doc/extend.texi: Document that malloc attribute adds
3520 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
3523 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3524 Define to speculation_safe_value_not_needed.
3526 2018-08-06 Jeff Law <law@redhat.com>
3528 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
3529 the vr_values instance to cprop_into_stmt.
3530 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
3531 (cprop_operand): Also query EVRP to determine if OP is a constant.
3533 2018-08-06 Nathan Sidwell <nathan@acm.org>
3535 * diagnostic.c (diagnostic_report_current_module): Reroll
3536 included-at loop. Translate text.
3538 2018-08-06 David Malcolm <dmalcolm@redhat.com>
3540 * function-tests.c (selftest::test_expansion_to_rtl): Call
3541 free_after_compilation.
3543 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3545 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
3547 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
3549 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
3550 loops with memory block operations from getting unrolled.
3552 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
3555 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3556 Define to speculation_safe_value_not_needed.
3558 2018-08-06 Jeff Law <law@redhat.com>
3560 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
3563 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3566 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
3567 with all enabled __intN types.
3569 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
3571 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3573 * alias.c (record_set): Check for clobber high.
3574 * cfgexpand.c (expand_gimple_stmt): Likewise.
3575 * combine-stack-adj.c (single_set_for_csa): Likewise.
3576 * combine.c (find_single_use_1): Likewise.
3577 (set_nonzero_bits_and_sign_copies): Likewise.
3578 (get_combine_src_dest): Likewise.
3579 (is_parallel_of_n_reg_sets): Likewise.
3580 (try_combine): Likewise.
3581 (record_dead_and_set_regs_1): Likewise.
3582 (reg_dead_at_p_1): Likewise.
3583 (reg_dead_at_p): Likewise.
3584 * dce.c (deletable_insn_p): Likewise.
3585 (mark_nonreg_stores_1): Likewise.
3586 (mark_nonreg_stores_2): Likewise.
3587 * df-scan.c (df_find_hard_reg_defs): Likewise.
3588 (df_uses_record): Likewise.
3589 (df_get_call_refs): Likewise.
3590 * dwarf2out.c (mem_loc_descriptor): Likewise.
3591 * haifa-sched.c (haifa_classify_rtx): Likewise.
3592 * ira-build.c (create_insn_allocnos): Likewise.
3593 * ira-costs.c (scan_one_insn): Likewise.
3594 * ira.c (equiv_init_movable_p): Likewise.
3595 (rtx_moveable_p): Likewise.
3596 (interesting_dest_for_shprep): Likewise.
3597 * jump.c (mark_jump_label_1): Likewise.
3598 * postreload-gcse.c (record_opr_changes): Likewise.
3599 * postreload.c (reload_cse_simplify): Likewise.
3600 (struct reg_use): Add source expr.
3601 (reload_combine): Check for clobber high.
3602 (reload_combine_note_use): Likewise.
3603 (reload_cse_move2add): Likewise.
3604 (move2add_note_store): Likewise.
3605 * print-rtl.c (print_pattern): Likewise.
3606 * recog.c (decode_asm_operands): Likewise.
3607 (store_data_bypass_p): Likewise.
3608 (if_test_bypass_p): Likewise.
3609 * regcprop.c (kill_clobbered_value): Likewise.
3610 (kill_set_value): Likewise.
3611 * reginfo.c (reg_scan_mark_refs): Likewise.
3612 * reload1.c (maybe_fix_stack_asms): Likewise.
3613 (eliminate_regs_1): Likewise.
3614 (elimination_effects): Likewise.
3615 (mark_not_eliminable): Likewise.
3616 (scan_paradoxical_subregs): Likewise.
3617 (forget_old_reloads_1): Likewise.
3618 * reorg.c (find_end_label): Likewise.
3619 (try_merge_delay_insns): Likewise.
3620 (redundant_insn): Likewise.
3621 (own_thread_p): Likewise.
3622 (fill_simple_delay_slots): Likewise.
3623 (fill_slots_from_thread): Likewise.
3624 (dbr_schedule): Likewise.
3625 * resource.c (update_live_status): Likewise.
3626 (mark_referenced_resources): Likewise.
3627 (mark_set_resources): Likewise.
3628 * rtl.c (copy_rtx): Likewise.
3629 * rtlanal.c (reg_referenced_p): Likewise.
3630 (single_set_2): Likewise.
3631 (noop_move_p): Likewise.
3632 (note_stores): Likewise.
3633 * sched-deps.c (sched_analyze_reg): Likewise.
3634 (sched_analyze_insn): Likewise.
3636 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3638 * cse.c (invalidate_reg): New function extracted from...
3639 (invalidate): ...here.
3640 (canonicalize_insn): Check for clobber high.
3641 (invalidate_from_clobbers): invalidate clobber highs.
3642 (invalidate_from_sets_and_clobbers): Likewise.
3643 (count_reg_usage): Check for clobber high.
3644 (insn_live_p): Likewise.
3645 * cselib.c (cselib_expand_value_rtx_1):Likewise.
3646 (cselib_invalidate_regno): Check for clobber in setter.
3647 (cselib_invalidate_rtx): Pass through setter.
3648 (cselib_invalidate_rtx_note_stores):
3649 (cselib_process_insn): Check for clobber high.
3650 * cselib.h (cselib_invalidate_rtx): Add operand.
3652 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3654 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
3655 (mark_not_eliminable): Likewise.
3656 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
3657 * lra-lives.c (process_bb_lives): Check for clobber high.
3658 * lra.c (new_insn_reg): Remember clobber highs.
3659 (collect_non_operand_hard_regs): Check for clobber high.
3660 (lra_set_insn_recog_data): Likewise.
3661 (add_regs_to_insn_regno_info): Likewise.
3662 (lra_update_insn_regno_info): Likewise.
3664 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3666 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
3667 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
3669 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3671 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
3672 (copy_insn_1): Likewise.
3673 (gen_hard_reg_clobber_high): New gen function.
3674 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
3675 * genemit.c (gen_exp): Likewise.
3676 (gen_emit_seq): Pass through info.
3677 (gen_insn): Check for CLOBBER_HIGH.
3678 (gen_expand): Pass through info.
3679 (gen_split): Likewise.
3680 (output_add_clobbers): Likewise.
3681 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
3682 (remove_clobbers): Likewise.
3683 * rtl.h (gen_hard_reg_clobber_high): New declaration.
3685 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3687 * doc/rtl.texi (clobber_high): Add.
3688 (parallel): Add in clobber high
3689 * rtl.c (rtl_check_failed_code3): Add function.
3690 * rtl.def (CLOBBER_HIGH): Add expression.
3691 * rtl.h (RTL_CHECKC3): Add macro.
3692 (rtl_check_failed_code3): Add declaration.
3693 (XC3EXP): Add macro.
3695 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
3698 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
3699 cfun->machine->max_used_stack_alignment if needed.
3701 2018-08-04 Martin Sebor <msebor@redhat.com>
3703 PR tree-optimization/86571
3704 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
3707 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
3710 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3713 2018-08-03 Jeff Law <law@redhat.com>
3716 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3717 Define to speculation_safe_value_not_needed.
3719 2018-08-03 David Malcolm <dmalcolm@redhat.com>
3721 * doc/gcov.texi (-x): Remove duplicate "to".
3722 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
3723 (-Wif-not-aligned): Remove duplicate "is".
3724 (-flto): Remove duplicate "the".
3725 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
3726 duplicate "v5.00.b".
3727 (MSP430 Options): Remove duplicate "and" from the description
3728 of "-mgprel-sec=regexp".
3729 (x86 Options): Remove duplicate copies of "vmldLog102" and
3730 vmlsLog104 from description of "-mveclibabi=type".
3732 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
3734 * internal-fn.h (first_commutative_argument): Declare.
3735 * internal-fn.c (first_commutative_argument): New function.
3736 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
3737 restrictions for pattern statements. Use first_commutative_argument
3738 to look for commutative operands in calls to internal functions.
3740 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
3742 * Makefile.in (wide-int-range.o): New.
3743 * tree-vrp.c: Move all the wide_int_* functions to...
3744 * wide-int-range.cc: ...here.
3745 * tree-vrp.h: Move all the wide_int_* prototypes to...
3746 * wide-int-range.h: ...here.
3748 2018-08-03 Tom de Vries <tdevries@suse.de>
3750 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
3752 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
3753 * except.c (output_function_exception_table): Do early exit if
3754 targetm_common.except_unwind_info (&global_options) == UI_NONE.
3756 2018-08-03 Martin Liska <mliska@suse.cz>
3758 * predict.c (dump_prediction): Change to 2 digits
3761 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
3763 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
3765 (falkor_am_1_gtov_gtov): ... a new insn reservation.
3767 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
3769 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
3770 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
3771 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
3772 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
3773 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
3775 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3777 * diagnostic-show-locus.c (diagnostic_show_locus): Use
3778 pp_take_prefix when saving the existing prefix.
3779 * diagnostic.c (diagnostic_append_note): Likewise.
3780 * langhooks.c (lhd_print_error_function): Likewise.
3781 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
3782 param's type. Free the existing prefix.
3783 (pp_take_prefix): New function.
3784 (pretty_printer::pretty_printer): Drop the prefix parameter.
3785 Rename the length parameter to match the comment.
3786 (pretty_printer::~pretty_printer): Free the prefix.
3787 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
3789 (struct pretty_printer): Drop the "const" from "prefix" field's
3790 type and clarify memory management.
3791 (pp_set_prefix): Drop the "const" from the 2nd param.
3792 (pp_take_prefix): New decl.
3794 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3796 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
3797 for word_mode_ok here instead of passing as argument.
3798 (expand_block_compare): Change select_block_compare_mode() call.
3799 (expand_strncmp_gpr_sequence): New function.
3800 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
3802 2018-08-02 Jeff Law <law@redhat.com>
3805 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3806 Define to speculation_safe_value_not_needed.
3809 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3810 Define to speculation_safe_value_not_needed.
3812 2018-08-02 Tom de Vries <tdevries@suse.de>
3815 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
3816 function. Return UI_TARGET unconditionally.
3817 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
3818 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
3820 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3822 * genemit.c (print_overload_test): Fix typo.
3824 2018-08-02 Richard Biener <rguenther@suse.de>
3826 PR tree-optimization/86816
3827 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
3828 which checks for value availability before querying it.
3829 (gvn_uses_equal): Use it.
3830 (same_succ_hash): Likewise.
3831 (gimple_equal_p): Likewise.
3833 2018-08-02 Nick Clifton <nickc@redhat.com>
3836 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3837 Define to speculation_safe_value_not_needed.
3840 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3841 Define to speculation_safe_value_not_needed.
3844 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3845 Define to speculation_safe_value_not_needed.
3848 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3849 Define to speculation_safe_value_not_needed.
3852 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3853 Define to speculation_safe_value_not_needed.
3856 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3857 Define to speculation_safe_value_not_needed.
3860 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3861 Define to speculation_safe_value_not_needed.
3864 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3865 Define to speculation_safe_value_not_needed.
3868 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3869 speculation_safe_value_not_needed.
3872 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
3873 to speculation_safe_value_not_needed.
3875 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3877 * doc/md.texi: Expand the documentation of instruction names
3878 to mention port-local uses. Document '@' in pattern names.
3879 * read-md.h (overloaded_instance, overloaded_name): New structs.
3881 (md_reader::handle_overloaded_name): New member function.
3882 (md_reader::get_overloads): Likewise.
3883 (md_reader::m_first_overload): New member variable.
3884 (md_reader::m_next_overload_ptr): Likewise.
3885 (md_reader::m_overloads_htab): Likewise.
3886 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
3887 m_next_overload_ptr and m_overloads_htab.
3888 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
3889 (get_mode_token, get_code_token, get_int_token): New functions.
3890 (map_attr_string): Add an optional argument that passes back
3891 the associated iterator.
3892 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
3893 (md_reader::handle_overloaded_name, add_overload_instance): New
3895 (apply_iterators): Handle '@' names. Report an error if '@'
3896 is used without iterators.
3897 (initialize_iterators): Initialize the new iterator_group fields.
3898 * genopinit.c (handle_overloaded_code_for)
3899 (handle_overloaded_gen): New functions.
3900 (main): Use them to print declarations of maybe_code_for_* and
3901 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
3902 * genemit.c (print_overload_arguments, print_overload_test)
3903 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
3904 (main): Use it to print definitions of maybe_code_for_* and
3905 maybe_gen_* functions.
3906 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
3907 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
3908 instead of explicit mode checks.
3909 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
3910 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
3911 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
3912 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
3913 (aarch64_expand_compare_and_swap): Likewise
3914 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
3915 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
3916 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
3917 (aarch64_constant_pool_reload_icode): Delete.
3918 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
3919 instead of aarch64_constant_pool_reload_icode. Use
3920 code_for_aarch64_reload_mov instead of explicit mode checks.
3921 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
3922 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
3923 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
3924 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
3925 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
3926 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
3927 (aarch64_atomic_load_op_code): Delete.
3928 (aarch64_emit_atomic_load_op): Likewise.
3929 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
3930 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
3931 instead of aarch64_emit_atomic_load_op.
3932 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
3933 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
3934 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
3935 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
3936 character before the pattern name.
3937 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3938 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
3939 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
3940 (aarch64_frecps<mode>): Likewise.
3941 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
3942 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
3943 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
3944 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
3945 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
3947 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3949 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3950 Allow HFmode constants if TARGET_FP_F16INST.
3952 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
3955 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
3956 No longer check last store for clobber of address register.
3958 2018-08-02 Martin Liska <mliska@suse.cz>
3960 PR gcov-profile/86817
3961 * gcov.c (process_all_functions): New function.
3963 (process_file): Move functions processing to
3964 process_all_functions.
3966 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3968 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
3969 "const" to the "gimple *" and "rtx_insn *" parameters.
3970 * dumpfile.h (dump_user_location_t::dump_user_location_t):
3972 (dump_location_t::dump_location_t): Likewise.
3974 2018-08-01 Martin Sebor <msebor@redhat.com>
3976 PR tree-optimization/86650
3977 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
3978 rather than a "gcall *". Directly pass the data of interest
3979 to percent_K_format, rather than building a temporary CALL_EXPR
3981 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
3982 (gimple_fold_builtin_strncat): Adjust.
3983 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
3984 gcall* argument with gimple*.
3985 * gimple-ssa-warn-restrict.c (check_call): Same.
3986 (wrestrict_dom_walker::before_dom_children): Same.
3987 (builtin_access::builtin_access): Same.
3988 (check_bounds_or_overlap): Same
3989 (maybe_diag_overlap): Same.
3990 (maybe_diag_offset_bounds): Same.
3991 * tree-diagnostic.c (default_tree_printer): Move usage of
3992 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
3994 * tree-pretty-print.c (percent_K_format): Add argument.
3995 * tree-pretty-print.h: Add argument.
3996 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
3997 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
3998 (maybe_diag_stxncpy_trunc): Same.
3999 (handle_builtin_stxncpy): Same.
4000 (handle_builtin_strcat): Same.
4002 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4004 * match.pd: Optimise pointer range checks.
4006 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4008 PR tree-optimization/86758
4009 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
4010 to remove pattern statements.
4012 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4014 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
4015 result of dfs_enumerate_from when constructing stmt_vec_infos,
4016 instead of additionally calling get_loop_body.
4018 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4020 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
4022 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
4023 When creating an iv, assert that the step is not known to be zero.
4024 (vect_setup_realignment): Update call accordingly.
4025 * tree-vect-stmts.c (vectorizable_store): Likewise.
4026 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
4028 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4030 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
4031 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
4032 (vectorizable_reduction): Likewise.
4033 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4034 (vect_detect_hybrid_slp_stmts): Likewise.
4035 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4037 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
4039 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
4040 (wide_int_set_zero_nonzero_bits): ...this.
4041 (zero_nonzero_bits_from_vr): Rename to...
4042 (vrp_set_zero_nonzero_bits): ...this.
4043 (extract_range_from_multiplicative_op_1): Abstract wide int
4045 (wide_int_range_multiplicative_op): ...here.
4046 (extract_range_from_binary_expr_1): Extract wide int binary
4047 operations into their own functions.
4048 (wide_int_range_lshift): New.
4049 (wide_int_range_can_optimize_bit_op): New.
4050 (wide_int_range_shift_undefined_p): New.
4051 (wide_int_range_bit_xor): New.
4052 (wide_int_range_bit_ior): New.
4053 (wide_int_range_bit_and): New.
4054 (wide_int_range_trunc_mod): New.
4055 (extract_range_into_wide_ints): New.
4056 (vrp_shift_undefined_p): New.
4057 (extract_range_from_multiplicative_op): New.
4058 (vrp_can_optimize_bit_op): New.
4059 * tree-vrp.h (value_range::dump): New.
4060 (wide_int_range_multiplicative_op): New.
4061 (wide_int_range_lshift):New.
4062 (wide_int_range_shift_undefined_p): New.
4063 (wide_int_range_bit_xor): New.
4064 (wide_int_range_bit_ior): New.
4065 (wide_int_range_bit_and): New.
4066 (wide_int_range_trunc_mod): New.
4067 (zero_nonzero_bits_from_bounds): Rename to...
4068 (wide_int_set_zero_nonzero_bits): ...this.
4069 (zero_nonzero_bits_from_vr): Rename to...
4070 (vrp_set_zero_nonzero_bits): ...this.
4071 (range_easy_mask_min_max): Rename to...
4072 (wide_int_range_can_optimize_bit_op): this.
4073 * vr-values.c (simplify_bit_ops_using_ranges): Rename
4074 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
4076 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4078 * tree-vectorizer.h (vect_orig_stmt): New function.
4079 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
4080 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
4081 (vect_create_epilog_for_reduction): Likewise.
4082 (vectorizable_live_operation): Likewise.
4083 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
4084 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
4085 * tree-vect-stmts.c (vectorizable_call): Likewise.
4086 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
4088 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4090 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
4092 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
4093 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
4094 (vect_transform_loop): Likewise.
4095 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4097 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4099 * tree-vectorizer.h (vect_schedule_slp): Return void.
4100 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4101 (vect_schedule_slp): Likewise.
4103 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4105 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
4107 (vect_transform_loop): Update calls accordingly. Schedule SLP
4108 instances before the main loop, if any exist.
4110 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4112 PR tree-optimization/86749
4113 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
4114 If the lhs is used in a COND_EXPR, check that it is being used
4115 as the "then" or "else" value.
4117 2018-08-01 Tom de Vries <tdevries@suse.de>
4120 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
4121 speculation_safe_value_not_needed.
4123 2018-08-01 Richard Biener <rguenther@suse.de>
4125 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
4128 2018-08-01 Martin Liska <mliska@suse.cz>
4130 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
4131 format how successful transformation is dumped.
4132 (gimple_mod_pow2_value_transform): Likewise.
4133 (gimple_mod_subtract_transform): Likewise.
4134 (gimple_stringops_transform): Likewise.
4136 2018-08-01 Martin Liska <mliska@suse.cz>
4139 * value-prof.c (interesting_stringop_to_profile_p):
4140 Simplify the code and add BUILT_IN_MEMMOVE.
4141 (gimple_stringops_transform): Likewise.
4143 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
4145 * config/aarch64/aarch64-simd.md
4146 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
4147 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
4148 use GPI iterator instead of SI mode.
4150 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
4152 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
4153 rs6000_speculation_barrier.
4154 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
4155 new barrier pattern name.
4157 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
4159 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
4160 (speculation_barrier): New insn.
4162 2018-08-01 Richard Biener <rguenther@suse.de>
4165 * graphite.h: Include isl/id.h and isl/space.h to allow build
4168 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
4171 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
4172 mode for COFF targets.
4173 * defaults.h (TARGET_COFF): Define.
4174 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
4175 TARGET_COFF): Define.
4176 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
4177 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
4178 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
4180 2018-07-31 Alexandre Oliva <oliva@adacore.com>
4182 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
4183 * gimple-streamer-out.c (output_bb): Save it.
4184 * lto-streamer-in.c (input_struct_function_base): Restore
4185 instance discriminator if available. Create map on demand.
4186 * lto-streamer-out.c (output_struct_function_base): Save it if
4188 * final.c (decl_to_instance_map): Document LTO strategy.
4190 2018-07-31 Alexandre Oliva <oliva@adacore.com>
4191 Olivier Hainque <hainque@adacore.com>
4193 * debug.h (decl_to_instance_map_t): New type.
4194 (decl_to_instance_map): Declare.
4195 (maybe_create_decl_to_instance_map): New inline function.
4196 * final.c (bb_discriminator, last_bb_discriminator): New statics,
4197 to track basic block discriminators.
4198 (final_start_function_1): Initialize them.
4199 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
4201 (decl_to_instance_map): New variable.
4202 (map_decl_to_instance, maybe_set_discriminator): New functions.
4203 (notice_source_line): Set discriminator.
4205 2018-07-31 Ian Lance Taylor <iant@golang.org>
4207 * targhooks.c (default_have_speculation_safe_value): Add
4210 2018-07-31 David Malcolm <dmalcolm@redhat.com>
4212 * dump-context.h: Include "pretty-print.h".
4213 (dump_context::refresh_dumps_are_enabled): New decl.
4214 (dump_context::emit_item): New decl.
4215 (class dump_context): Add fields "m_test_pp" and
4217 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
4218 (temp_dump_context::get_dumped_text): New decl.
4219 (class temp_dump_context): Add field "m_pp".
4220 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
4221 (dump_context::refresh_dumps_are_enabled): ...and add a test for
4223 (set_dump_file): Update for above change.
4224 (set_alt_dump_file): Likewise.
4225 (dump_loc): New overload, taking a pretty_printer *.
4226 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
4227 to any test pretty-printer.
4228 (make_item_for_dump_gimple_stmt): New function, adapted from
4229 optinfo::add_gimple_stmt.
4230 (dump_context::dump_gimple_stmt): Call it, and use the result,
4231 eliminating the direct usage of dump_file and alt_dump_file in
4232 favor of indirectly using them via emit_item.
4233 (make_item_for_dump_gimple_expr): New function, adapted from
4234 optinfo::add_gimple_expr.
4235 (dump_context::dump_gimple_expr): Call it, and use the result,
4236 eliminating the direct usage of dump_file and alt_dump_file in
4237 favor of indirectly using them via emit_item.
4238 (make_item_for_dump_generic_expr): New function, adapted from
4240 (dump_context::dump_generic_expr): Call it, and use the result,
4241 eliminating the direct usage of dump_file and alt_dump_file in
4242 favor of indirectly using them via emit_item.
4243 (make_item_for_dump_printf_va): New function, adapted from
4244 optinfo::add_printf_va.
4245 (make_item_for_dump_printf): New function.
4246 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
4247 and use the result, eliminating the direct usage of dump_file and
4248 alt_dump_file in favor of indirectly using them via emit_item.
4249 (make_item_for_dump_dec): New function.
4250 (dump_context::dump_dec): Call it, and use the result,
4251 eliminating the direct usage of dump_file and alt_dump_file in
4252 favor of indirectly using them via emit_item.
4253 (make_item_for_dump_symtab_node): New function, adapted from
4254 optinfo::add_symtab_node.
4255 (dump_context::dump_symtab_node): Call it, and use the result,
4256 eliminating the direct usage of dump_file and alt_dump_file in
4257 favor of indirectly using them via emit_item.
4258 (dump_context::begin_scope): Reimplement, avoiding direct usage
4259 of dump_file and alt_dump_file in favor of indirectly using them
4261 (dump_context::emit_item): New member function.
4262 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
4263 Set up test pretty-printer on the underlying context. Call
4264 refresh_dumps_are_enabled.
4265 (temp_dump_context::~temp_dump_context): Call
4266 refresh_dumps_are_enabled.
4267 (temp_dump_context::get_dumped_text): New member function.
4268 (selftest::verify_dumped_text): New function.
4269 (ASSERT_DUMPED_TEXT_EQ): New macro.
4270 (selftest::test_capture_of_dump_calls): Run all tests twice, with
4271 and then without optinfo enabled. Add uses of
4272 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
4273 * dumpfile.h: Update comment for the dump_* API.
4274 * optinfo-emit-json.cc
4275 (selftest::test_building_json_from_dump_calls): Update for new
4276 param for temp_dump_context ctor.
4277 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
4278 and "m_owned" field.
4279 (optinfo_item::~optinfo_item): Likewise.
4280 (optinfo::add_item): New member function.
4281 (optinfo::emit): Update comment.
4282 (optinfo::add_string): Delete.
4283 (optinfo::add_printf): Delete.
4284 (optinfo::add_printf_va): Delete.
4285 (optinfo::add_gimple_stmt): Delete.
4286 (optinfo::add_gimple_expr): Delete.
4287 (optinfo::add_tree): Delete.
4288 (optinfo::add_symtab_node): Delete.
4289 (optinfo::add_dec): Delete.
4290 * optinfo.h (class dump_context): New forward decl.
4291 (optinfo::add_item): New decl.
4292 (optinfo::add_string): Delete.
4293 (optinfo::add_printf): Delete.
4294 (optinfo::add_printf_va): Delete.
4295 (optinfo::add_gimple_stmt): Delete.
4296 (optinfo::add_gimple_expr): Delete.
4297 (optinfo::add_tree): Delete.
4298 (optinfo::add_symtab_node): Delete.
4299 (optinfo::add_dec): Delete.
4300 (optinfo::add_poly_int): Delete.
4301 (optinfo_item::optinfo_item): Remove "owned" param.
4302 (class optinfo_item): Remove field "m_owned".
4304 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4307 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
4308 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
4309 requested variable alignment.
4310 (expand_one_ssa_partition): Likewise.
4311 (expand_one_var): Likewise.
4313 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4315 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4316 to speculation_safe_value_not_needed.
4318 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4320 * targhooks.h (speculation_safe_value_not_needed): New prototype.
4321 * targhooks.c (speculation_safe_value_not_needed): New function.
4322 * target.def (have_speculation_safe_value): Update documentation.
4323 * doc/tm.texi: Regenerated.
4325 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4327 * config/aarch64/iterators.md (ALLI_TI): New iterator.
4328 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
4330 (despeculate_copy<ALLI:mode>_insn): New insn.
4331 (despeculate_copyti_insn): New insn.
4332 (despeculate_simple<ALLI:mode>): New insn
4333 (despeculate_simpleti): New insn.
4334 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
4336 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
4337 aarch64_speculation_safe_value.
4338 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
4340 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4342 * config/aarch64/aarch64-speculation.cc: New file.
4343 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
4344 before pass_reorder_blocks.
4345 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
4347 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
4348 X14 and X15 when tracking speculation.
4349 * config/aarch64/aarch64.md (register name constants): Add
4350 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
4351 (unspec): Add UNSPEC_SPECULATION_TRACKER.
4352 (speculation_barrier): New insn attribute.
4353 (cmp<mode>): Allow SP in comparisons.
4354 (speculation_tracker): New insn.
4355 (speculation_barrier): Add speculation_barrier attribute.
4356 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
4357 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
4358 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
4360 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4362 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
4363 aarch64_track_speculation is true.
4364 (tb<optab><mode>1): Likewise.
4365 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
4366 generate CB[N]Z when tracking speculation.
4367 (aarch64_split_compare_and_swap): Likewise.
4368 (aarch64_split_atomic_op): Likewise.
4370 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4372 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
4374 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4376 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
4377 (speculation_barrier): New insn.
4379 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4381 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
4382 * config/arm/arm.md (speculation_barrier): New expand.
4383 (speculation_barrier_insn): New pattern.
4385 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4387 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
4388 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
4389 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
4390 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
4392 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
4393 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
4394 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
4395 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
4396 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
4397 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
4398 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
4399 * builtins.c (expand_speculation_safe_value): New function.
4400 (expand_builtin): Call it.
4401 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
4402 * doc/extend.texi: Document __builtin_speculation_safe_value.
4403 * doc/md.texi: Document "speculation_barrier" pattern.
4404 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
4405 TARGET_HAVE_SPECULATION_SAFE_VALUE.
4406 * doc/tm.texi: Regenerated.
4407 * target.def (have_speculation_safe_value, speculation_safe_value): New
4409 * targhooks.c (default_have_speculation_safe_value): New function.
4410 (default_speculation_safe_value): New function.
4411 * targhooks.h (default_have_speculation_safe_value): Add prototype.
4412 (default_speculation_safe_value): Add prototype.
4414 2018-07-31 David Malcolm <dmalcolm@redhat.com>
4416 * dump-context.h (dump_context::dump_loc): New decl.
4417 * dumpfile.c (dump_context::dump_loc): New member function.
4418 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
4419 and dump_gimple_stmt.
4420 (dump_context::dump_gimple_expr_loc): Likewise, using
4422 (dump_context::dump_generic_expr_loc): Likewise, using
4424 (dump_context::dump_printf_loc_va): Likewise, using
4426 (dump_context::begin_scope): Explicitly using the global function
4427 "dump_loc", rather than the member function.
4429 2018-07-31 Martin Sebor <msebor@redhat.com>
4431 PR tree-optimization/86741
4432 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
4434 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
4436 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
4437 depend on whether prefetch instructions will be emitted or not.
4438 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
4439 will be emitted or not.
4440 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
4441 (TARGET_SETMEM_PFD): New macros.
4443 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4445 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
4446 (NULL_STMT_VEC_INFO): Delete.
4447 (stmt_vec_info::operator*): Likewise.
4448 (stmt_vec_info::operator gimple *): Likewise.
4449 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
4450 of NULL_STMT_VEC_INFO.
4451 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4452 (vect_reassociating_reduction_p): Likewise.
4453 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
4454 (vectorizable_store): Likewise.
4455 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
4456 (vec_info::free_stmt_vec_infos): Likewise.
4458 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4460 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
4461 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
4463 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4465 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
4466 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
4467 (vec_info::free_stmt_vec_info): New private member functions.
4468 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
4469 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
4470 * tree-parloops.c (gather_scalar_reductions): Remove calls to
4471 set_stmt_vec_info_vec and free_stmt_vec_infos.
4472 * tree-vect-loop.c (_loop_vec_info): Remove call to
4473 set_stmt_vec_info_vec.
4474 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
4475 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
4476 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
4477 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
4478 (vec_info::free_stmt_vec_info): ...these new functions. Remove
4479 assignments in {vec_info::,}new_stmt_vec_info that are redundant
4480 with the clearing in the xcalloc.
4481 (stmt_vec_info_vec): Delete.
4482 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
4483 (vectorize_loops): Likewise.
4484 (vec_info::~vec_info): Remove argument from call to
4485 free_stmt_vec_infos.
4486 (vec_info::add_stmt): Remove vinfo argument from call to
4489 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4491 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
4492 rather than a gimple stmt.
4493 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
4494 information for pattern statements when passed the original
4495 statement; instead wait to be passed the pattern statement itself.
4496 Don't call set_vinfo_for_stmt here.
4497 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
4498 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
4499 stmt_vec_infos here.
4500 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
4501 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
4502 stmt_vec_infos entry.
4504 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4506 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
4507 * tree-vectorizer.c (vec_info::replace_stmt): New function.
4508 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
4509 * tree-vect-stmts.c (vectorizable_call): Likewise.
4510 (vectorizable_simd_clone_call): Likewise.
4512 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4514 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
4515 * tree-vectorizer.c (vec_info::remove_stmt): New function.
4516 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
4517 * tree-vect-loop.c (vect_transform_loop): Likewise.
4518 * tree-vect-slp.c (vect_schedule_slp): Likewise.
4519 * tree-vect-stmts.c (vect_remove_stores): Likewise.
4521 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4523 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
4524 (vect_dr_stmt): Delete.
4525 * tree-vectorizer.c (vec_info::lookup_dr): New function.
4526 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
4528 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
4529 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
4530 (vect_verify_datarefs_alignment, vect_peeling_supportable)
4531 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
4532 (vect_analyze_data_refs): Likewise.
4533 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
4535 (vect_find_same_alignment_drs): Likewise.
4536 (vect_slp_analyze_node_dependences): Update calls accordingly.
4537 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
4538 instead of DR_VECT_AUX.
4539 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
4540 of a vector data references. Use vec_info::lookup_dr instead of
4542 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
4543 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
4544 instead of DR_VECT_AUX.
4546 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4548 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
4550 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4552 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4553 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4554 (vect_gen_prolog_loop_niters): Likewise.
4556 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4558 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
4559 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
4560 (vect_known_alignment_in_bytes, vect_dr_behavior)
4561 (vect_get_scalar_dr_size): Take references as dr_vec_infos
4562 instead of data_references. Update calls to other routines for
4563 which the same change has been made.
4564 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
4565 dr_vec_infos instead of stmt_vec_infos.
4566 (vect_analyze_data_ref_dependence): Update call accordingly.
4567 (vect_slp_analyze_data_ref_dependence)
4568 (vect_record_base_alignments): Use DR_VECT_AUX.
4569 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
4570 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
4571 (vector_alignment_reachable_p, vect_get_data_access_cost)
4572 (vect_peeling_supportable, vect_analyze_group_access_1)
4573 (vect_analyze_group_access, vect_analyze_data_ref_access)
4574 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
4575 (vect_compile_time_alias, vect_small_gap_p)
4576 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
4577 (vect_supportable_dr_alignment): Take references as dr_vec_infos
4578 instead of data_references. Update calls to other routines for
4579 which the same change has been made.
4580 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
4581 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4582 (vect_slp_analyze_and_verify_node_alignment)
4583 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
4584 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4585 (vect_setup_realignment): Use dr_vec_infos. Update calls after
4587 (_vect_peel_info::dr): Replace with...
4588 (_vect_peel_info::dr_info): ...this new field.
4589 (vect_peeling_hash_get_most_frequent)
4590 (vect_peeling_hash_choose_best_peeling): Update accordingly.
4591 (vect_peeling_hash_get_lowest_cost):
4592 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
4593 routines for which the same change has been made.
4594 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
4596 * tree-vect-loop-manip.c (get_misalign_in_elems)
4597 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
4599 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4600 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
4601 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
4602 (get_group_load_store_type, get_negative_load_store_type)
4603 (vect_get_data_ptr_increment, vectorizable_store)
4604 (vectorizable_load): Likewise.
4605 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
4606 Update calls to other routines for which the same change has been made.
4608 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4610 * tree-vectorizer.h (vec_info::move_dr): New member function.
4611 (dataref_aux): Rename to...
4612 (dr_vec_info): ...this and add "dr" and "stmt" fields.
4613 (_stmt_vec_info::dr_aux): Update accordingly.
4614 (_stmt_vec_info::data_ref_info): Delete.
4615 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
4616 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
4617 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
4618 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
4620 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
4621 (STMT_VINFO_DR_INFO): New macro.
4622 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
4623 (set_dr_misalignment): Update after rename of dataref_aux.
4624 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
4625 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
4626 initialization of STMT_VINFO_DATA_REF.
4627 * tree-vectorizer.c (vec_info::move_dr): New function.
4628 * tree-vect-patterns.c (vect_recog_bool_pattern)
4629 (vect_recog_mask_conversion_pattern)
4630 (vect_recog_gather_scatter_pattern): Use it.
4631 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
4632 the "dr" and "stmt" fields of dr_vec_info instead of
4633 STMT_VINFO_DATA_REF.
4635 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4637 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
4638 (is_pattern_stmt_p): Use it.
4639 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
4640 on pattern statements.
4642 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4644 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
4645 original stmt as a stmt_vec_info rather than a gimple stmt.
4646 (vect_pattern_recog_1): Take the statement directly as a
4647 stmt_vec_info, rather than via a gimple_stmt_iterator.
4648 Update call to vect_mark_pattern_stmts.
4649 (vect_pattern_recog): Update calls accordingly.
4651 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4653 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
4654 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
4655 a vect_def_type for the first argument.
4656 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
4657 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
4658 operand if it isn't defined by a vectorized statement.
4659 (vect_build_gather_load_calls): Remove the mask_dt argument and
4660 update calls to vect_get_vec_def_for_stmt_copy.
4661 (vectorizable_bswap): Likewise the dt argument.
4662 (vectorizable_call): Update calls to vectorizable_bswap and
4663 vect_get_vec_def_for_stmt_copy.
4664 (vectorizable_simd_clone_call, vectorizable_assignment)
4665 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
4666 (vectorizable_comparison): Update calls to
4667 vect_get_vec_def_for_stmt_copy.
4668 (vectorizable_store): Likewise. Remove now-unnecessary calls to
4670 (vect_get_loop_based_defs): Remove dt argument and update call
4671 to vect_get_vec_def_for_stmt_copy.
4672 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
4673 and vect_get_vec_def_for_stmt_copy.
4674 (vectorizable_load): Update calls to vect_build_gather_load_calls
4675 and vect_get_vec_def_for_stmt_copy.
4676 * tree-vect-loop.c (vect_create_epilog_for_reduction)
4677 (vectorizable_reduction, vectorizable_live_operation): Update calls
4678 to vect_get_vec_def_for_stmt_copy.
4680 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4682 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
4683 and gimple stmt arguments with a stmt_vec_info.
4684 (vect_record_base_alignments): Update calls accordingly.
4685 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
4686 and gimple stmt arguments with a stmt_vec_info.
4687 (vect_build_slp_tree_1): Remove vinfo argument and update call
4688 to vect_record_max_nunits.
4689 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
4690 and vect_record_max_nunits.
4692 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4694 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
4695 file and take a stmt_vec_info instead of a gimple stmt.
4696 (supportable_widening_operation, vect_finish_replace_stmt)
4697 (vect_finish_stmt_generation, vect_get_store_rhs)
4698 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
4699 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
4700 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
4701 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
4702 (vect_create_data_ref_ptr, bump_vector_ptr)
4703 (vect_permute_store_chain, vect_setup_realignment)
4704 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
4705 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
4706 (vectorizable_reduction, vectorizable_induction)
4707 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
4708 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
4709 than gimple stmts as arguments.
4710 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
4711 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
4712 (can_group_stmts_p, vect_check_gather_scatter)
4713 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4714 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
4715 (vect_permute_load_chain, vect_shift_permute_load_chain)
4716 (vect_transform_grouped_load)
4717 (vect_record_grouped_load_vectors): Likewise.
4718 * tree-vect-loop.c (vect_fixup_reduc_chain)
4719 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
4720 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
4721 (vectorizable_reduction, vectorizable_induction)
4722 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
4723 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
4724 (vect_get_load_store_mask): Likewise.
4725 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
4726 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
4727 * tree-vect-stmts.c (vect_mark_relevant)
4728 (is_simple_and_all_uses_invariant)
4729 (exist_non_indexing_operands_for_use_p, process_use)
4730 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
4731 (vect_get_vec_def_for_operand, vect_get_vec_defs)
4732 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
4733 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
4734 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
4735 (get_negative_load_store_type, get_load_store_type)
4736 (vect_check_load_store_mask, vect_check_store_rhs)
4737 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
4738 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4739 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4740 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4741 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
4742 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
4743 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
4744 (supportable_widening_operation): Likewise.
4746 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4748 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
4749 a stmt_vec_info instead of a gcall.
4750 (vect_check_gather_scatter): Update call accordingly.
4751 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
4753 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
4754 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
4755 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
4756 instead of a gimple stmt.
4757 (vect_transform_loop): Update calls accordingly.
4758 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
4759 stmt_vec_infos instead of gimple stmts.
4760 (vect_analyze_slp_instance): Update use accordingly.
4761 * tree-vect-stmts.c (read_vector_array, write_vector_array)
4762 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
4763 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
4764 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
4765 (vect_gen_widened_results_half, vect_get_loop_based_defs)
4766 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
4767 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
4768 down to subroutines.
4770 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4772 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
4773 of the worklist from a vector of gimple stmts to a vector of
4775 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4776 (vect_mark_stmts_to_be_vectorized): Likewise
4778 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4780 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
4781 statement before passing it to vect_analyze_stmt.
4782 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
4783 the chain of phi vector definitions. Track the exit phi via its
4785 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
4786 STMT_VINFO_REDUC_DEF.
4787 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
4788 stmt_vec_infos to handle the statement chains.
4789 (vect_get_slp_defs): Record the first statement in the node
4790 using a stmt_vec_info.
4791 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
4792 statements here and pass their stmt_vec_info down to subroutines.
4793 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
4794 down to vect_finish_stmt_generation.
4795 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
4796 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
4797 stmt_vec_infos to subroutines.
4798 (vect_remove_stores): Use stmt_vec_infos to handle the statement
4801 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4803 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
4804 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
4805 (vect_permute_store_chain, vect_setup_realignment)
4806 (vect_permute_load_chain, vect_shift_permute_load_chain)
4807 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
4808 stmts internally, and when passing values to other vectorizer routines.
4809 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4810 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
4811 (vect_analyze_loop_operations, get_initial_def_for_reduction)
4812 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
4813 (vectorizable_reduction, vectorizable_induction)
4814 (vectorizable_live_operation, vect_transform_loop_stmt)
4815 (vect_transform_loop): Likewise.
4816 * tree-vect-patterns.c (vect_reassociating_reduction_p)
4817 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
4818 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
4819 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4820 (vect_slp_analyze_node_operations_1): Likewise.
4821 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4822 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
4823 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
4824 (vect_finish_stmt_generation_1, get_group_load_store_type)
4825 (get_load_store_type, vect_build_gather_load_calls)
4826 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4827 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4828 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4829 (vectorizable_store, vectorizable_load, vectorizable_condition)
4830 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
4831 (supportable_widening_operation): Likewise.
4832 (vect_get_vector_types_for_stmt): Likewise.
4833 * tree-vectorizer.h (vect_dr_behavior): Likewise.
4835 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4837 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4838 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
4839 (vect_permute_store_chain, vect_permute_load_chain)
4840 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
4841 repeated stmt_vec_info lookups.
4842 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4843 (vect_update_ivs_after_vectorizer): Likewise.
4844 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
4845 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4846 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
4847 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4848 (vect_bb_slp_scalar_cost): Likewise.
4849 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
4851 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4853 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
4854 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
4855 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
4856 to get gassigns and gcalls, rather than operating on generc gimple
4858 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
4859 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
4860 (vectorizable_load, vect_analyze_stmt): Likewise.
4861 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
4863 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4865 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
4866 return stmt_vec_infos rather than gimple stmts. Do not accept
4868 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
4870 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
4871 Update use of get_later_stmt.
4872 (vect_get_constant_vectors): Update call accordingly.
4873 (vect_schedule_slp_instance): Likewise
4874 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
4875 (vect_slp_analyze_instance_dependence): Likewise.
4876 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
4878 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4880 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
4881 (stmt_info_for_cost::stmt_info): ...this new field.
4882 (add_stmt_costs): Update accordingly.
4883 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4884 (vect_get_known_peeling_cost): Likewise.
4885 (vect_estimate_min_profitable_iters): Likewise.
4886 * tree-vect-stmts.c (record_stmt_cost): Likewise.
4888 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4890 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
4891 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4892 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4894 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
4896 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4898 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
4899 a gimple stmt to a stmt_vec_info.
4900 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
4902 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4904 * tree-vectorizer.h (vec_info::grouped_stores): Change from
4905 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4906 (_loop_vec_info::reduction_chains): Likewise.
4907 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
4909 * tree-vect-slp.c (vect_analyze_slp): Likewise.
4911 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4913 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
4914 a gimple stmt to a stmt_vec_info.
4915 (_stmt_vec_info::next_element): Likewise.
4916 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
4917 (vect_slp_analyze_and_verify_node_alignment)
4918 (vect_analyze_group_access_1, vect_analyze_group_access)
4919 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
4920 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
4921 (vect_supportable_dr_alignment): Update accordingly.
4922 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
4923 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
4924 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
4925 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
4926 * tree-vect-slp.c (vect_build_slp_tree_1)
4927 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
4928 (vect_split_slp_store_group, vect_analyze_slp_instance)
4929 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
4930 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
4931 (get_group_load_store_type, get_load_store_type)
4932 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
4933 (vect_transform_stmt, vect_remove_stores): Likewise.
4935 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4937 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
4939 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4940 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
4941 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
4942 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
4943 (vector_alignment_reachable_p, vect_get_data_access_cost)
4944 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
4945 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
4946 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4947 (vect_analyze_group_access_1, vect_analyze_group_access)
4948 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
4949 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
4950 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
4951 result of vect_dr_stmt and use the stmt_vec_info instead of
4952 the associated gimple stmt.
4953 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4954 (vect_gen_prolog_loop_niters): Likewise.
4955 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4957 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4959 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
4960 to a vec<stmt_vec_info>.
4961 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
4962 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
4964 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
4965 to a vec<stmt_vec_info>.
4966 (bst_traits::value_type, bst_traits::value_type): Likewise.
4967 (bst_traits::hash): Update accordingly.
4968 (vect_get_and_check_slp_defs): Change the stmts parameter from
4969 a vec<gimple *> to a vec<stmt_vec_info>.
4970 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
4971 (vect_build_slp_tree): Likewise.
4972 (vect_build_slp_tree_2): Likewise. Update uses of
4973 SLP_TREE_SCALAR_STMTS.
4974 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
4975 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
4976 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
4977 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
4978 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
4979 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
4980 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
4981 (vect_get_constant_vectors, vect_get_slp_defs)
4982 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
4983 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
4984 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
4985 instead of gimple stmts.
4986 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
4987 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
4988 (vect_slp_analyze_instance_dependence): Update uses of
4989 SLP_TREE_SCALAR_STMTS.
4990 (vect_slp_analyze_and_verify_node_alignment): Likewise.
4991 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
4992 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
4993 (get_initial_defs_for_reduction): Likewise.
4994 (vect_create_epilog_for_reduction): Likewise.
4995 (vectorize_fold_left_reduction): Likewise.
4996 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
4997 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
4998 (can_vectorize_live_stmts): Likewise.
5000 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5002 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
5003 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
5004 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
5006 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
5008 (gather_scalar_reductions): Update after above interface changes.
5009 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
5010 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
5012 (vect_force_simple_reduction): Likewise.
5013 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
5014 LOOP_VINFO_REDUCTIONS.
5015 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
5017 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5019 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
5020 a gimple stmt to a stmt_vec_info.
5021 * tree-vect-loop.c (vect_active_double_reduction_p)
5022 (vect_force_simple_reduction, vectorizable_reduction): Update
5025 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5027 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
5028 vec<gimple *> to a vec<stmt_vec_info>.
5029 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
5030 the reduction_phis argument from a vec<gimple *> to a
5032 (vectorizable_reduction): Likewise the phis local variable that
5033 is passed to vect_create_epilog_for_reduction. Update for new type
5034 of SLP_TREE_VEC_STMTS.
5035 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
5036 (vectorizable_live_operation): Likewise.
5037 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
5038 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
5040 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5042 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
5043 a gimple stmt to a stmt_vec_info.
5044 (vectorizable_condition, vectorizable_live_operation)
5045 (vectorizable_reduction, vectorizable_induction): Pass back the
5046 vectorized statement as a stmt_vec_info.
5047 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
5048 use of STMT_VINFO_VEC_STMT.
5049 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
5050 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
5051 as stmt_vec_infos rather than gimple stmts.
5052 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
5054 (vectorizable_live_operation): Likewise.
5055 (vectorizable_reduction, vectorizable_induction): Likewise,
5056 updating use of STMT_VINFO_VEC_STMT.
5057 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
5058 of STMT_VINFO_VEC_STMT.
5059 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
5060 (vectorizable_simd_clone_call, vectorizable_conversion)
5061 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
5062 (vectorizable_store, vectorizable_load, vectorizable_condition)
5063 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
5064 from a gimple stmt to a stmt_vec_info.
5065 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
5066 pointer to a stmt_vec_info to the vectorizable_* routines.
5068 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5070 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
5071 a gimple stmt to a stmt_vec_info.
5072 (is_pattern_stmt_p): Update accordingly.
5073 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
5074 (vect_record_grouped_load_vectors): Likewise.
5075 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
5076 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
5077 (vect_model_reduction_cost): Likewise.
5078 (vect_create_epilog_for_reduction): Likewise.
5079 (vectorizable_reduction, vectorizable_induction): Likewise.
5080 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
5081 Return the stmt_vec_info for the pattern statement.
5082 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
5083 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
5084 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
5085 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
5086 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
5087 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
5088 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
5089 (free_stmt_vec_info, vect_is_simple_use): Likewise.
5091 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5093 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
5094 (vect_finish_stmt_generation): Likewise.
5095 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
5096 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
5097 (vect_build_gather_load_calls): Use the return value of the above
5098 functions instead of a separate call to vinfo_for_stmt. Use narrow
5099 scopes for the input gimple stmt and wider scopes for the associated
5100 stmt_vec_info. Use vec_info::lookup_def when setting these
5101 stmt_vec_infos from an SSA_NAME definition.
5102 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
5103 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
5104 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
5105 (vectorizable_store, vectorizable_load, vectorizable_condition)
5106 (vectorizable_comparison): Likewise.
5107 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
5108 (vectorizable_reduction): Likewise.
5110 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5112 * tree-vectorizer.h (vect_is_simple_use): Add an optional
5113 stmt_vec_info * parameter before the optional gimple **.
5114 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
5115 (process_use, vect_get_vec_def_for_operand_1): Update callers.
5116 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
5117 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5118 (vectorizable_live_operation): Likewise.
5119 * tree-vect-patterns.c (type_conversion_p): Likewise.
5120 (vect_look_through_possible_promotion): Likewise.
5121 (vect_recog_rotate_pattern): Likewise.
5122 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
5124 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5126 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
5127 a typedef to a wrapper class.
5128 (NULL_STMT_VEC_INFO): New macro.
5129 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
5130 (stmt_vec_info::operator*): New function.
5131 (stmt_vec_info::operator gimple *): Likewise.
5132 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
5133 (add_stmt_costs): Likewise.
5134 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
5135 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
5136 (vect_get_known_peeling_cost): Likewise.
5137 (vect_estimate_min_profitable_iters): Likewise.
5138 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
5139 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
5140 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
5141 (vectorizable_store, free_stmt_vec_infos): Likewise.
5142 (new_stmt_vec_info): Change return type of xcalloc to
5145 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5147 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
5148 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
5149 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
5150 a single_imm_use-based sequence.
5151 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
5153 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5155 * tree-vectorizer.h (vec_info::lookup_def): Declare.
5156 * tree-vectorizer.c (vec_info::lookup_def): New function.
5157 * tree-vect-patterns.c (vect_get_internal_def): Use it.
5158 (vect_widened_op_tree): Likewise.
5159 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
5160 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
5161 (vectorizable_reduction): Likewise.
5162 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
5164 (vect_is_slp_reduction): Update calls accordingly. Use
5165 vec_info::lookup_def.
5166 (vect_is_simple_reduction): Likewise
5167 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
5169 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5171 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
5172 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
5173 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
5175 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
5176 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
5177 (vect_update_vf_for_slp, vect_analyze_loop_operations)
5178 (vect_is_slp_reduction, vectorizable_induction)
5179 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
5180 * tree-vect-patterns.c (vect_init_pattern_stmt):
5181 (vect_determine_min_output_precision_1, vect_determine_precisions)
5182 (vect_pattern_recog): Likewise.
5183 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
5184 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
5185 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
5186 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
5187 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
5188 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
5189 info field from a loop to a loop_vec_info.
5191 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5193 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
5194 (vec_info::add_stmt): Declare.
5195 * tree-vectorizer.c (vec_info::add_stmt): New function.
5196 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
5197 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
5198 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
5199 (vectorizable_induction): Likewise.
5200 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
5201 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
5202 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
5203 (vectorizable_load): Likewise.
5204 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
5205 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
5206 (vect_recog_gather_scatter_pattern): Likewise.
5207 (append_pattern_def_seq): Likewise. Remove a check that is
5208 performed by add_stmt itself.
5210 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5212 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
5213 which make_ssa_name was called with new_stmt before new_stmt
5216 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5218 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
5220 (vect_is_slp_reduction): ...here...
5221 (vect_is_simple_reduction): ...and here. Remove repetition of tests
5222 that are already known to be false.
5224 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5226 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
5227 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
5228 STMT_VINFO_NUM_SLP_USES when it's true.
5229 (vect_free_slp_instance): Add a final_p parameter and pass it to
5231 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
5232 (vect_analyze_slp_instance): Likewise.
5233 (vect_slp_analyze_operations): Likewise.
5234 (vect_slp_analyze_bb_1): Likewise.
5235 * tree-vectorizer.c (vec_info): Likewise.
5236 * tree-vect-loop.c (vect_transform_loop): Likewise.
5238 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5240 * tree-vect-loop.c (vectorizable_reduction): Assert that the
5241 function is not called for second and subsequent members of
5244 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5246 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
5247 cases for nested loops from here to ...
5248 (vect_create_epilog_for_reduction): ...here. Only call
5249 vect_is_simple_use for inner-loop reductions.
5251 2018-07-31 Martin Liska <mliska@suse.cz>
5253 PR gcov-profile/85338
5254 PR gcov-profile/85350
5255 PR gcov-profile/85372
5256 * profile.c (struct location_triplet): New.
5257 (struct location_triplet_hash): Likewise.
5258 (output_location): Do not output a BB that
5259 is already recorded for a line.
5260 (branch_prob): Use streamed_locations.
5262 2018-07-31 Martin Liska <mliska@suse.cz>
5264 PR gcov-profile/85370
5265 * coverage.c (coverage_begin_function): Do not mark target
5266 clones as artificial functions.
5268 2018-07-31 Martin Liska <mliska@suse.cz>
5270 PR gcov-profile/83813
5271 PR gcov-profile/84758
5272 PR gcov-profile/85217
5273 PR gcov-profile/85332
5274 * profile.c (branch_prob): Do not record GOTO expressions
5275 for GIMPLE statements which locations are already streamed.
5277 2018-07-31 Olivier Hainque <hainque@adacore.com>
5279 * gcc.c (handle_spec_function): Accept a soft_matched_part
5280 argument, as do_spec_1. Pass it down to ...
5281 (eval_spec_function): Accept a soft_matched_part argument,
5282 and pass it down to ...
5283 (do_spec_2): Accept a soft_matched_part argument, and pass
5284 it down to do_spec_1.
5285 (do_spec_1): Pass soft_matched_part to handle_spec_function.
5286 (handle_braces): Update call to handle_spec_function.
5287 (driver::set_up_specs): Update calls to do_spec_2.
5288 (compare_debug_dump_opt_spec_function): Likewise.
5289 (compare_debug_self_opt_spec_function): Likewise.
5291 2018-07-31 Olivier Hainque <hainque@adacore.com>
5293 * common.opt (nolibc): New option.
5294 * doc/invoke.texi (Link Options): Document it.
5295 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
5296 * config/alpha/linux.h: Likewise.
5297 * config/arc/elf.h: Likewise.
5298 * config/arm/uclinux-elf.h: Likewise.
5299 * config/arm/unknown-elf.h: Likewise.
5300 * config/avr/avrlibc.h: Likewise.
5301 * config/bfin/bfin.h: Likewise.
5302 * config/bfin/linux.h: Likewise.
5303 * config/bfin/uclinux.h: Likewise.
5304 * config/darwin.h: Likewise.
5305 * config/darwin10.h: Likewise.
5306 * config/darwin12.h: Likewise.
5307 * config/gnu-user.h: Likewise.
5308 * config/lm32/uclinux-elf.h: Likewise.
5309 * config/pa/pa-hpux11.h: Likewise.
5310 * config/pa/pa64-hpux.h: Likewise.
5311 * config/sparc/sparc.h: Likewise.
5313 2018-07-31 Olivier Hainque <hainque@adacore.com>
5315 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
5316 undefined variables.
5318 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
5321 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
5324 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
5326 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
5327 terminated string literal.
5329 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
5331 PR rtl-optimization/85160
5332 * combine.c (is_just_move): New function.
5333 (try_combine): Allow combining two instructions into two if neither of
5334 the original instructions was a move.
5336 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
5339 * doc/extend.texi (Global Register Variables): Discourage use of type
5341 (Local Register Variables): Likewise.
5343 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
5345 PR tree-optimization/86506
5346 * hwint.h (ceil_log2): Resync with hwint.c implementation.
5348 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
5351 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
5352 hard_regno, make sure no insn between `from` and `to` clobbers it.
5354 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
5355 Tom de Vries <tdevries@suse.de>
5357 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
5358 (PTX_DEFAULT_RUNTIME_DIM): ... this.
5359 (nvptx_goacc_validate_dims): Set default worker and gang dims to
5360 PTX_DEFAULT_RUNTIME_DIM.
5361 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
5363 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
5365 * config/pa/pa.c (pa_output_addr_vec): Align address table.
5366 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
5367 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
5369 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
5371 * config/rs6000/constraints.md (wG constraint): Delete, no longer
5373 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
5374 predicate to reflect toc fusion has been deleted.
5375 (toc_fusion_mem_raw): Delete, no longer used.
5376 (toc_fusion_mem_wrapped): Likewise.
5377 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
5379 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
5380 Delete, no longer used.
5381 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
5382 meant to be used for toc fusion.
5383 (rs6000_debug_print_mode): Delete toc fusion debugging.
5384 (rs6000_debug_reg_global): Likewise.
5385 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
5386 fusion and secondary reload support that were never used.
5387 (rs6000_option_override_internal): Delete TOC fusion, that was only
5388 partially defined, and it did not work unless you also used the
5390 (rs6000_legitimate_address_p): Delete TOC fusion support.
5391 (rs6000_opt_masks): Likewise.
5392 (fusion_wrap_memory_address): Delete function, no longer used.
5393 (fusion_split_address); Delete TOC fusion support.
5394 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
5395 longer used with toc fusion being deleted.
5396 (TARGET_TOC_FUSION_FP): Likewise.
5397 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
5399 (toc fusion spliter): Delete TOC fusion support.
5400 (toc_fusionload_<mode>): Likewise.
5401 (toc_fusionload_di): Likewise.
5402 (fusion_gpr_load_<mode>): Delete generator function, this insn no
5403 longer needs to be named. Rename predicate to delete TOC fusion.
5404 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
5405 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
5406 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
5407 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
5408 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
5410 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5412 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
5413 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
5414 __int128 in built-in function prototypes.
5415 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
5416 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
5418 2018-07-27 Martin Sebor <msebor@redhat.com>
5420 PR tree-optimization/86696
5421 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
5422 types, including enums.
5423 (handle_char_store): Be prepared for the above function to fail.
5425 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
5427 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
5428 when optimization level is lower than 2 or optimize for size.
5430 2018-07-26 Martin Sebor <msebor@redhat.com>
5432 PR tree-optimization/86043
5433 PR tree-optimization/86042
5434 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
5435 (get_string_cst_length): Rename...
5436 (get_min_string_length): ...to this. Add argument.
5437 (handle_char_store): Extend to handle multi-character stores by
5439 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
5440 * tree.h (initializer_zerop): Add argument.
5442 2018-07-26 Jakub Jelinek <jakub@redhat.com>
5445 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
5446 declare target to variables if they have always,{to,from,tofrom} map
5449 2018-07-26 Martin Liska <mliska@suse.cz>
5452 * lto-wrapper.c: Add linker_output as prefix
5453 for ltrans_output_file.
5455 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
5457 PR rtl-optimization/85805
5458 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
5459 value for hard registers if that was written in the same mode.
5461 2018-07-26 Martin Liska <mliska@suse.cz>
5463 PR gcov-profile/86536
5464 * gcov.c (format_gcov): Use printf format %.*f directly
5465 and do not handle special values.
5467 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5469 * common/config/arc/arc-common.c (arc_option_optimization_table):
5470 Update default optimizations for size.
5472 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5474 * config/arc/arc.md (movsf_insn): Add short instruction selection.
5475 * config/arc/constraints.md (CfZ): New constraint.
5476 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
5477 (subsf3_fpu): Likewise.
5478 (cmpsf_fpu): Likewise.
5479 (cmpsf_fpu_uneq): Likewise.
5481 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5483 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
5485 (arc_is_uncached_mem_p): uncached applies to both the variable and
5488 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5490 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
5493 2018-07-25 David Malcolm <dmalcolm@redhat.com>
5495 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
5496 field "m_scopes" from vec to auto_vec.
5498 2018-07-25 Martin Liska <mliska@suse.cz>
5500 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
5503 2018-07-25 Richard Biener <rguenther@suse.de>
5506 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
5507 special wrt context_die late.
5508 (gen_subprogram_die): Re-use DIEs in local scope.
5510 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
5512 PR tree-optimization/86644
5513 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
5515 2018-07-25 Martin Liska <mliska@suse.cz>
5518 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
5519 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
5521 2018-07-25 Martin Liska <mliska@suse.cz>
5524 * params.def: Explain ASan abbreviation and provide
5525 a documentation link.
5527 2018-07-24 Martin Sebor <msebor@redhat.com>
5529 PR tree-optimization/86622
5530 PR tree-optimization/86532
5531 * builtins.h (string_length): Declare.
5532 * builtins.c (c_strlen): Correct handling of non-constant offsets.
5533 (check_access): Be prepared for non-constant length ranges.
5534 (string_length): Make extern.
5535 * expr.c (string_constant): Only handle the minor non-constant
5536 array index. Use string_constant to compute the length of
5537 a generic string constant.
5539 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
5541 PR tree-optimization/86618
5542 * tree-vect-stmts.c (vectorizable_call): Don't take the address
5543 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
5545 2018-07-24 David Malcolm <dmalcolm@redhat.com>
5547 PR tree-optimization/86636
5548 * json.cc (json::object::set): Fix comment. Add assertions.
5549 (json::array::append): Move here from json.h. Add comment and an
5551 (json::string::string): Likewise.
5552 * json.h (json::array::append): Move to json.cc.
5553 (json::string::string): Likewise.
5554 * optinfo-emit-json.cc
5555 (optrecord_json_writer::impl_location_to_json): Assert that we
5556 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
5557 wrapper around it. Expand the location once, rather than three
5559 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
5560 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
5562 (optrecord_json_writer::optinfo_to_json): Likewise, in four
5563 places. Fix some overlong lines.
5565 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
5567 * config/aarch64/aarch64-simd.md
5568 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
5569 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
5570 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
5571 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
5572 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
5573 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
5574 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
5575 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
5576 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
5578 2018-07-24 Jakub Jelinek <jakub@redhat.com>
5581 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
5582 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
5583 and abs_d == d, do the power of two handling if profitable.
5585 2018-07-24 Richard Biener <rguenther@suse.de>
5587 * match.pd: Add BIT_FIELD_REF canonicalizations.
5589 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5592 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
5594 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5596 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
5597 terminated STRING_CST object.
5599 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5601 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
5603 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
5605 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
5606 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
5607 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
5608 the elements into a register.
5609 (rs6000_split_v4si_init_di_reg): Delete.
5610 (rs6000_split_v4si_init): Delete.
5611 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
5612 (vsx_init_v4si): Rewrite as a define_expand.
5614 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
5616 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
5617 zero_extend argument from memory): New.
5619 2018-07-22 Martin Sebor <msebor@redhat.com>
5622 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
5623 diagnosing calls with unknown arguments unless -Walloca-larger-than
5624 is restricted to less than PTRDIFF_MAX bytes.
5626 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
5628 * doc/gcov.texi (Invoking Gcov): Editorial changes.
5630 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5632 * pretty-print.c (text_info::set_location): Remove redundant
5633 "line_table" parameter from call to rich_location::set_range.
5635 2018-07-20 Martin Sebor <msebor@redhat.com>
5638 * builtins.c (expand_builtin_alloca): Adjust.
5639 * calls.c (alloc_max_size): Simplify.
5640 * cgraphunit.c (cgraph_node::expand): Adjust.
5641 * common.opt (larger_than_size, warn_frame_larger_than): Remove
5643 (frame_larger_than_size): Same.
5644 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
5645 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
5647 * doc/invoke.texi (GCC Command Options): Document option arguments.
5648 Explain byte-size arguments and suffixes.
5649 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
5650 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
5651 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
5652 * doc/options.texi (UInteger): Expand.
5653 (Host_Wide_Int, ByteSize): Document new properties.
5654 * final.c (final_start_function_1): Include sizes in an error message.
5655 * function.c (frame_offset_overflow): Same.
5656 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
5657 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
5658 Diagnose unbounded alloca calls only for limits of less than
5660 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
5661 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
5663 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
5664 only for limits of less than PTRDIFF_MAX.
5665 * langhooks-def.h (lhd_handle_option): Change function argument
5667 * langhooks.c (lhd_handle_option): Same.
5668 * langhooks.h (handle_option): Same.
5669 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
5671 (var_type, var_type_struct): Same.
5672 (var_set): Handle ByteSize flag.
5673 * optc-gen.awk: Add comments to output to ease debugging. Make
5674 use of HOST_WIDE_INT where appropriate.
5675 * opts-gen-save.awk: Use %lx to format unsigned long.
5676 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
5677 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
5678 arguments. Parse bytes-size suffixes.
5679 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
5680 (enum_value_to_arg): Same.
5681 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
5682 (handle_option): Adjust.
5683 (generate_option): Change function argument to HOST_WIDE_INT.
5684 (cmdline_handle_error): Adjust.
5685 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
5686 (set_option): Change function argument to HOST_WIDE_INT.
5687 (option_enabled): Handle cl_host_wide_int.
5688 (get_option_state): Handle CLVC_SIZE.
5689 (control_warning_option): Same.
5690 * opts.c (common_handle_option): Change function argument to
5691 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
5692 OPT_Wvla_larger_than_.
5693 * opts.h (enum cl_var_type): Add an enumerator.
5694 * stor-layout.c (layout_decl): Print a more meaningful warning.
5695 * toplev.c (output_stack_usage): Adjust.
5697 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
5699 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
5700 call to inline_expand_builtin_string_cmp.
5701 (expand_builtin_strcmp): Likewise.
5702 (expand_builtin_strncmp): Likewise.
5703 (inline_string_cmp): Delete the last parameter, change char_type_node
5704 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
5706 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
5707 the inlining expansion on target where the type of the call has same or
5708 narrower precision than unsigned char.
5710 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5712 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
5713 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
5714 * common.opt (fsave-optimization-record): New option.
5715 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
5716 * doc/invoke.texi (-fsave-optimization-record): New option.
5717 * dumpfile.c: Include "optinfo-emit-json.h".
5718 (struct kv_pair): Move to coretypes.h.
5719 (optgroup_options): Make non-static.
5720 (dump_context::end_scope): Call
5721 optimization_records_maybe_pop_dump_scope.
5722 * dumpfile.h (optgroup_options): New decl.
5723 * json.cc: New file.
5725 * optinfo-emit-json.cc: New file.
5726 * optinfo-emit-json.h: New file.
5727 * optinfo.cc: Include "optinfo-emit-json.h".
5728 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
5729 (optinfo_enabled_p): Check optimization_records_enabled_p.
5730 (optinfo_wants_inlining_info_p): Likewise.
5731 * optinfo.h: Update comment.
5732 * profile-count.c (profile_quality_as_string): New function.
5733 * profile-count.h (profile_quality_as_string): New decl.
5734 (profile_count::quality): New accessor.
5735 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
5736 and optinfo_emit_json_cc_tests.
5737 * selftest.h (selftest::json_cc_tests): New decl.
5738 (selftest::optinfo_emit_json_cc_tests): New decl.
5739 * toplev.c: Include "optinfo-emit-json.h".
5740 (compile_file): Call optimization_records_finish.
5741 (do_compile): Call optimization_records_start.
5742 * tree-ssa-live.c: Include optinfo.h.
5743 (remove_unused_scope_block_p): Retain inlining information if
5744 optinfo_wants_inlining_info_p returns true.
5746 2018-07-20 Richard Biener <rguenther@suse.de>
5749 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
5750 to cover -flto-partition=none.
5752 2018-07-20 Martin Liska <mliska@suse.cz>
5754 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
5755 (get_decl_source_range): Remove unused function.
5757 2018-07-20 Richard Biener <rguenther@suse.de>
5759 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
5760 (struct vn_phi_s): Likewise.
5761 (struct vn_reference_s): Likewise.
5762 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
5763 for searching the slot of an entry known to be in the hash itself.
5764 (vn_phi_hasher::equal): Likewise.
5765 (vn_reference_hasher::equal): Likewise.
5766 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
5768 (optimistic_info, current_info): Remove, keeping only valid_info.
5769 (vn_reference_lookup_1): Remove fallback lookup.
5770 (vn_reference_lookup_2): Likewise.
5771 (vn_nary_op_lookup_1): Likewise.
5772 (vn_phi_lookup): Likewise.
5773 (vn_nary_build_or_lookup_1): Make sure to not chain the built
5775 (vn_reference_insert): Adjust, chain the inserted hash element
5776 at last_inserted_ref.
5777 (vn_reference_insert_pieces): Likewise.
5778 (visit_reference_op_call): Likewise.
5779 (vn_nary_op_insert_into): Chain the inserted hash element at
5781 (vn_nary_op_insert_pieces): Adjust.
5782 (vn_nary_op_insert): Likewise.
5783 (vn_nary_op_insert_stmt): Likewise.
5784 (vn_phi_insert): Adjust, chain the inserted hash element at
5786 (process_scc): Remove clearing and copying the optimistic
5787 table. Instead remove elements inserted during an optimistic
5788 iteration from the single table we maintain.
5789 (init_scc_vn): Adjust.
5790 (free_scc_vn): Likewise.
5791 (sccvn_dom_walker::record_cond): Likewise.
5792 (sccvn_dom_walker::after_dom_children): Likewise.
5794 2018-07-19 Martin Sebor <msebor@redhat.com>
5796 PR tree-optimization/84047
5797 PR tree-optimization/83776
5798 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
5799 (check_array_bounds): Call it.
5801 2018-07-19 Martin Sebor <msebor@redhat.com>
5803 * align.h (align_flags): Use member initialization.
5805 2018-07-19 David Malcolm <dmalcolm@redhat.com>
5807 * Makefile.in (OBJS): Add optinfo.o.
5808 * coretypes.h (class symtab_node): New forward decl.
5809 (struct cgraph_node): New forward decl.
5810 (class varpool_node): New forward decl.
5811 * dump-context.h: New file.
5812 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
5814 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
5815 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
5816 (set_alt_dump_file): Likewise.
5817 (dump_context::~dump_context): New dtor.
5818 (dump_gimple_stmt): Move implementation to...
5819 (dump_context::dump_gimple_stmt): ...this new member function.
5820 Add the stmt to any pending optinfo, creating one if need be.
5821 (dump_gimple_stmt_loc): Move implementation to...
5822 (dump_context::dump_gimple_stmt_loc): ...this new member function.
5823 Start a new optinfo and add the stmt to it.
5824 (dump_gimple_expr): Move implementation to...
5825 (dump_context::dump_gimple_expr): ...this new member function.
5826 Add the stmt to any pending optinfo, creating one if need be.
5827 (dump_gimple_expr_loc): Move implementation to...
5828 (dump_context::dump_gimple_expr_loc): ...this new member function.
5829 Start a new optinfo and add the stmt to it.
5830 (dump_generic_expr): Move implementation to...
5831 (dump_context::dump_generic_expr): ...this new member function.
5832 Add the tree to any pending optinfo, creating one if need be.
5833 (dump_generic_expr_loc): Move implementation to...
5834 (dump_context::dump_generic_expr_loc): ...this new member
5835 function. Add the tree to any pending optinfo, creating one if
5837 (dump_printf): Move implementation to...
5838 (dump_context::dump_printf_va): ...this new member function. Add
5839 the text to any pending optinfo, creating one if need be.
5840 (dump_printf_loc): Move implementation to...
5841 (dump_context::dump_printf_loc_va): ...this new member function.
5842 Start a new optinfo and add the stmt to it.
5843 (dump_dec): Move implementation to...
5844 (dump_context::dump_dec): ...this new member function. Add the
5845 value to any pending optinfo, creating one if need be.
5846 (dump_context::dump_symtab_node): New member function.
5847 (dump_context::get_scope_depth): New member function.
5848 (dump_context::begin_scope): New member function.
5849 (dump_context::end_scope): New member function.
5850 (dump_context::ensure_pending_optinfo): New member function.
5851 (dump_context::begin_next_optinfo): New member function.
5852 (dump_context::end_any_optinfo): New member function.
5853 (dump_context::s_current): New global.
5854 (dump_context::s_default): New global.
5855 (dump_scope_depth): Delete global.
5856 (dumpfile_ensure_any_optinfo_are_flushed): New function.
5857 (dump_symtab_node): New function.
5858 (get_dump_scope_depth): Reimplement in terms of dump_context.
5859 (dump_begin_scope): Likewise.
5860 (dump_end_scope): Likewise.
5861 (selftest::temp_dump_context::temp_dump_context): New ctor.
5862 (selftest::temp_dump_context::~temp_dump_context): New dtor.
5863 (selftest::verify_item): New function.
5864 (ASSERT_IS_TEXT): New macro.
5865 (ASSERT_IS_TREE): New macro.
5866 (ASSERT_IS_GIMPLE): New macro.
5867 (selftest::test_capture_of_dump_calls): New test.
5868 (selftest::dumpfile_c_tests): Call it.
5869 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
5870 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
5871 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
5872 descriptive comment.
5873 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
5874 (dump_node, dump_bb): Move these unrelated decls.
5875 (class dump_manager): Add leading comment.
5876 * optinfo.cc: New file.
5877 * optinfo.h: New file.
5879 2018-07-19 Michael Collison <michael.collison@arm.com>
5880 Richard Henderson <rth@redhat.com>
5882 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
5883 (subti): Handle op1 zero.
5884 (subvti4, usub4ti4): New.
5885 (*sub<GPI>3_compare1_imm): New.
5886 (sub<GPI>3_carryinCV): New.
5887 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
5888 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
5890 2018-07-19 Michael Collison <michael.collison@arm.com>
5891 Richard Henderson <rth@redhat.com>
5893 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
5894 (addti3): Create simpler code if low part is already known to be 0.
5895 (addvti4, uaddvti4): New.
5896 (*add<GPI>3_compareC_cconly_imm): New.
5897 (*add<GPI>3_compareC_cconly): New.
5898 (*add<GPI>3_compareC_imm): New.
5899 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
5900 handle constants within this pattern..
5901 (*add<GPI>3_compareV_cconly_imm): New.
5902 (*add<GPI>3_compareV_cconly): New.
5903 (*add<GPI>3_compareV_imm): New.
5904 (add<GPI>3_compareV): New.
5905 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
5906 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
5907 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
5908 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
5909 with 'comparison' operator.
5910 (*add<GPI>3_compareV_cconly_imm): Ditto.
5911 (*add<GPI>3_compareV_cconly): Ditto.
5912 (*add<GPI>3_compareV_imm): Ditto.
5913 (add<GPI>3_compareV): Ditto.
5914 (add<mode>3_carryinC): Ditto.
5915 (*add<mode>3_carryinC_zero): Ditto.
5916 (*add<mode>3_carryinC): Ditto.
5917 (add<mode>3_carryinV): Ditto.
5918 (*add<mode>3_carryinV_zero): Ditto.
5919 (*add<mode>3_carryinV): Ditto.
5921 2018-07-19 Michael Collison <michael.collison@arm.com>
5922 Richard Henderson <rth@redhat.com>
5924 * config/aarch64/aarch64-modes.def (CC_V): New.
5925 * config/aarch64/aarch64-protos.h
5926 (aarch64_addti_scratch_regs): Declare
5927 (aarch64_subvti_scratch_regs): Declare.
5928 (aarch64_expand_subvti): Declare.
5929 (aarch64_gen_unlikely_cbranch): Declare
5930 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
5931 for signed overflow using CC_Vmode.
5932 (aarch64_get_condition_code_1): Handle CC_Vmode.
5933 (aarch64_gen_unlikely_cbranch): New function.
5934 (aarch64_addti_scratch_regs): New function.
5935 (aarch64_subvti_scratch_regs): New function.
5936 (aarch64_expand_subvti): New function.
5938 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5940 * config/aarch64/aarch64-option-extensions.def: New entry for profile
5942 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
5943 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
5946 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5949 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
5950 address check not strict.
5952 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5954 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
5956 (store_pair_lanes<mode>): Likewise.
5957 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
5958 enum value 'ADDR_QUERY_LDP_STP_N'.
5959 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
5960 (aarch64_print_address_internal): Add declaration.
5961 (aarch64_print_ldpstp_address): Remove.
5962 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
5963 (aarch64_print_operand): Change printing of 'y'.
5964 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
5965 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
5966 'true' rather than '1'.
5967 * config/aarch64/constraints.md (Uml): Likewise.
5968 (Uml): Rename to Umn.
5971 2018-07-19 Richard Biener <rguenther@suse.de>
5973 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
5975 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
5976 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
5977 (vn_reference_hasher): Likewise.
5978 (struct vn_tables_s): Remove obstack and alloc-pool members.
5979 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
5980 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
5981 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
5982 (vn_reference_insert_pieces): Likewise.
5983 (alloc_vn_nary_op_noinit): Adjust.
5984 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
5985 (vn_phi_eq): Adjust.
5986 (shared_lookup_phiargs): Remove.
5987 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
5988 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
5989 (visit_reference_op_call): Likewise.
5990 (copy_nary, copy_phi, copy_reference): Remove.
5991 (process_scc): Rewind the obstack when iterating. Do not
5992 copy the elements to valid_info but just move them from one
5993 hashtable to the other.
5994 (allocate_vn_table): Adjust.
5995 (free_vn_table): Likewise.
5996 (init_scc_vn): Likewise.
5997 (free_scc_vn): Likewise.
5999 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
6002 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
6003 indirect_return as function type attribute.
6004 (ix86_attribute_table): Change indirect_return to function
6006 * doc/extend.texi: Update indirect_return attribute.
6008 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
6010 * wide-int.h (widest2_int): New.
6011 * gimple-fold.c (arith_overflowed_p): Use it.
6012 * tree.h (widest2_int_cst): New.
6013 * tree-vrp.c (wide_int_binop_overflow): Rename from
6014 vrp_int_const_binop.
6015 Rewrite to work on trees.
6016 (extract_range_from_multiplicative_op_1): Abstract code to...
6017 (wide_int_range_min_max): ...here.
6018 (wide_int_range_cross_product): ...and here.
6019 (extract_range_from_binary_expr_1): Abstract overflow code to...
6020 (wide_int_range_mult_wrapping): ...here.
6021 * tree-vrp.h (wide_int_range_cross_product): New.
6022 (wide_int_range_mult_wrapping): New.
6024 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
6025 Julia Koval <julia.koval@intel.com>
6027 * config/i386/x86-tune-costs.h (skylake_memcpy,
6028 skylake_memset): Replace rep_prefix with unrolling for size 512.
6030 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
6033 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
6034 comparision with EQ_EXPR in last stmt.
6036 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
6038 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
6039 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
6040 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
6041 previously known as "PowerPC AltiVec Built-in Functions". Move
6042 some material to new subsubsections "PowerPC AltiVec Built-in
6043 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
6045 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
6046 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
6047 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
6048 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
6050 2018-07-18 Richard Biener <rguenther@suse.de>
6052 PR tree-optimization/86557
6053 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
6056 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
6058 * config/s390/s390.c (s390_function_profiler): Generate CFI.
6060 2018-07-17 Jeff Law <law@redhat.com>
6062 * config/arm/arm.c (get_label_padding): Update for recent
6063 changes to label_to_alignment.
6065 PR tree-optimization/86010
6066 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
6068 * config/mips/mips.c (vr4130_align_insns): Update for recent
6069 changes to label_to_alignment.
6071 * config/frv/frv.c (frv_label_align): Update for recent changes
6072 to label_to_alignment.
6074 * config/nios2/nios2.c (nios2_label_align): Update for recent
6075 changes which dropped ALIGN_LABELS_LOG.
6077 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
6079 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
6080 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
6082 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6084 * config/arc/arc.c (arc_label_align): Use align_labels instead of
6085 deprecated align_labels_log.
6087 2018-07-17 Richard Biener <rguenther@suse.de>
6090 * dwarf2out.c (init_sections_and_labels): Always generate
6091 a debug_line_str_section for early LTO debug.
6092 (dwarf2out_finish): Reset debug_line_str_hash output early.
6093 Bump counter for extra dwarf5 .debug_loc labels to not conflict
6095 (dwarf2out_early_finish): Output debug_line_str.
6097 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
6099 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
6100 index register on z196 or later.
6102 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
6104 * config/s390/s390.c (s390_default_align): Set default function
6106 (s390_override_options_after_change): Call s390_default align.
6107 (s390_option_override_internal): Call s390_default_align.
6108 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
6110 2018-07-17 Jakub Jelinek <jakub@redhat.com>
6113 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
6114 to _looptemp_ clauses, other than the first two.
6116 2018-07-17 Martin Liska <mliska@suse.cz>
6118 * opts.c: Do not enable OPT_falign_* for -Os.
6120 2018-07-17 Martin Liska <mliska@suse.cz>
6122 * align.h (MAX_CODE_ALIGN): New.
6123 (MAX_CODE_ALIGN_VALUE): New.
6124 * common/config/i386/i386-common.c (ix86_handle_option):
6125 (MAX_CODE_ALIGN): Moved to align.h.
6126 * final.c (MAX_CODE_ALIGN): Likewise.
6127 * opts.c (parse_and_check_align_values):
6128 (MAX_CODE_ALIGN): Likewise.
6129 (MAX_CODE_ALIGN_VALUE): Likewise.
6131 2018-07-17 Martin Liska <mliska@suse.cz>
6133 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
6134 in order to fulfil coding style.
6135 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
6136 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6137 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6138 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
6139 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
6140 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
6141 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
6142 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
6143 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6145 2018-07-17 Martin Liska <mliska@suse.cz>
6147 * align.h: New file.
6148 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
6150 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
6151 align_flags of label_to_alignment.
6152 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
6154 * config/m68k/m68k.c: Do not use removed align_labels_value and
6156 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
6157 (LOOP_ALIGN): Likewise.
6158 (LABEL_ALIGN): Likewise.
6159 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
6160 Remove not used macro.
6161 (rs6000_loop_align): Change return type to align_flags.
6162 (rs6000_loop_align_max_skip): Remove.
6163 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
6164 Change return type to align_flags.
6165 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
6166 Remove not used macro.
6167 (rs6000_loop_align): Change return type to align_flags.
6168 (rs6000_loop_align_max_skip): Remove.
6169 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
6170 * config/rx/rx-protos.h (rx_align_for_label): Make it
6172 * config/rx/rx.c (rx_align_for_label): Change return type
6174 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
6176 into align_flags class.
6177 (LABEL_ALIGN): Likewise.
6178 (LOOP_ALIGN): Likewise.
6179 * config/s390/s390.c (s390_label_align): Use align_flags
6181 (s390_asm_output_function_label): Likewise.
6182 * config/sh/sh.c (sh_override_options_after_change):
6183 Use align_flags class directly without macros.
6184 (find_barrier): Likewise.
6185 (barrier_align): Likewise.
6186 (sh_loop_align): Likewise.
6187 * config/spu/spu.c (spu_option_override):
6188 Use align_flags_tuple::get_value instead of removed macros.
6189 (spu_sched_init): Likewise.
6190 * config/spu/spu.h (GTY): Likewise.
6191 * config/visium/visium.c (visium_option_override):
6192 Set "8" as default secondary alignment.
6193 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
6194 in order to guarantee secondary alignment of 8.
6195 * coretypes.h: Include align.h header file.
6196 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
6197 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
6198 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
6199 * doc/tm.texi.in: Likewise.
6200 * final.c (struct label_alignment): Remove not used structure.
6201 (LABEL_ALIGN): Change type to align_flags.
6202 (LOOP_ALIGN): Likewise.
6203 (JUMP_ALIGN): Likewise.
6204 (default_loop_align_max_skip): Remove.
6205 (default_label_align_max_skip): Likewise.
6206 (default_jump_align_max_skip): Likewise.
6207 (default_label_align_after_barrier_max_skip):
6208 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
6209 (LABEL_TO_MAX_SKIP): Remove.
6210 (label_to_alignment): Return align_flags type instead of integer.
6211 (label_to_max_skip): Remove.
6212 (align_fuzz): Use align_flags type.
6213 (compute_alignments): Use align_flags type and use align_flags::max
6214 to combine multiple alignments.
6215 (grow_label_align): Grow vec instead of C array.
6216 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
6217 (shorten_branches): Use align_flags type and use align_flags::max
6218 to combine multiple alignments.
6219 (final_scan_insn_1): Remove usage of secondary alignment that comes
6220 from label alignment, but instead use proper secondary alignment
6221 which is computed in grow_label_align.
6222 * flags.h (struct align_flags_tuple): Move to align.h.
6223 (struct align_flags): Likewise.
6224 (state_align_loops): Rename to align_loops.
6225 (state_align_jumps): Rename to align_jumps.
6226 (state_align_labels): Rename to align_labels.
6227 (state_align_functions): Rename to align_functions.
6228 (align_loops_log): Remove.
6229 (align_jumps_log): Remove.
6230 (align_labels_log): Remove.
6231 (align_functions_log): Remove.
6232 (align_loops_max_skip): Remove.
6233 (align_jumps_max_skip): Remove.
6234 (align_labels_max_skip): Remove.
6235 (align_functions_max_skip): Remove.
6236 (align_loops_value): Remove.
6237 (align_jumps_value): Remove.
6238 (align_labels_value): Remove.
6239 (align_functions_value): Remove.
6240 * output.h (label_to_alignment): Change return type to align_flags.
6241 (label_to_max_skip): Remove.
6242 * target.def: Remove loop_align_max_skip, label_align_max_skip,
6243 jump_align_max_skip macros.
6244 * targhooks.h (default_loop_align_max_skip): Remove.
6245 (default_label_align_max_skip): Likewise.
6246 (default_jump_align_max_skip): Likewise.
6247 (default_label_align_after_barrier_max_skip): Remove.
6248 * toplev.c (read_log_maxskip): Use ::normalize function.
6249 (parse_N_M): Remove not used argument and also call ::normalize.
6250 (parse_alignment_opts): Do not pass unused arguments.
6251 * varasm.c (assemble_start_function): Use directly align_functions
6252 instead of removed macros.
6253 * system.h: Do not poison removed macros.
6255 2018-07-17 Jakub Jelinek <jakub@redhat.com>
6258 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
6259 and cond temporaries don't have reference type if iterator has
6260 pointer type. For init use &for_pre_body instead of pre_p if
6261 for_pre_body is non-empty.
6263 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6265 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
6266 double-double modes to SFmode directly directly.
6267 (trunc<mode>sf2_fprs): Delete.
6269 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6271 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
6272 for conversions between IFmode and the decimal floating point modes.
6273 (init_float128_ieee): Use the correct names for conversions between
6274 KFmode and the decimal floating point modes.
6276 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6278 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
6279 for the conversions between TDmode and IFmode.
6280 (init_float128_ieee): Use more correct names for the conversions
6281 between TDmode and KFmode.
6283 2018-07-16 Jakub Jelinek <jakub@redhat.com>
6285 PR tree-optimization/86526
6286 * builtins.c (expand_builtin_memcmp): Formatting fixes.
6287 (inline_expand_builtin_string_cmp): Likewise.
6288 (inline_string_cmp): Likewise. Use c_readstr instead of
6289 builtin_memcpy_read_str. Add unit_mode temporary.
6291 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6294 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
6295 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
6297 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
6299 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
6300 Alphabetize prototypes of built-in functions, separating out
6301 built-in functions that are listed in this section but should be
6302 described elsewhere.
6304 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
6307 * expmed.c (emit_store_flag): Do not emit setcc followed by a
6308 conditional move when trapping comparison was split to a
6309 non-trapping one (and vice versa).
6311 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6313 * config/s390/s390.c (s390_function_profiler): Generate nops
6314 instead of profiler call sequences.
6315 * config/s390/s390.opt: Add the new option.
6317 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6319 * config/s390/s390.c (s390_function_profiler): Generate
6320 __mcount_loc section.
6321 * config/s390/s390.opt: Add the new option.
6323 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6325 * common.opt: Add the new warning.
6326 * config/s390/s390.c (s390_function_profiler): Emit "brasl
6327 %r0,__fentry__" when -mfentry is specified.
6328 (s390_option_override_internal): Disallow -mfentry for 31-bit
6330 * config/s390/s390.opt: Add the new option.
6332 2018-07-16 Richard Biener <rguenther@suse.de>
6335 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
6336 for function-local FUNCTION_DECL and RESULT_DECL immediately.
6338 2018-07-16 Martin Liska <mliska@suse.cz>
6341 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
6344 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
6346 * config/arc/arcHS.md: Update ARCHS scheduling rules.
6348 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
6350 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
6352 * config/arc/arc-cpus.def (hs4x): New cpu.
6354 * config/arc/arc-tables.opt: Regenerate.
6355 * config/arc/arc.c (arc_sched_issue_rate): New function.
6356 (TARGET_SCHED_ISSUE_RATE): Define.
6357 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
6358 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
6360 (attr tune): Add ARCHS4x tune values.
6361 (attr tune_dspmpy): Define.
6362 (*tst): Correct instruction type.
6363 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
6364 * config/arc/arcHS4x.md: New file.
6365 * config/arc/fpu.md: Update instruction type attributes.
6366 * config/arc/t-multilib: Regenerate.
6368 2018-07-16 Tom de Vries <tdevries@suse.de>
6371 * var-tracking.c (vt_initialize): Fix pre_dec handling.
6373 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6375 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
6378 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
6380 PR tree-optimization/86514
6381 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
6382 conversion to a boolean type from a type with greater precision.
6384 2018-07-16 Tom de Vries <tdevries@suse.de>
6386 * var-tracking.c (vt_initialize): Print adjusted insn slim if
6387 dump_flags request TDF_SLIM.
6389 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
6391 * fold-const.c (int_const_binop_1): Abstract...
6392 (wide_int_binop): ...wide int code here.
6393 (poly_int_binop): ...poly int code here.
6394 Abstract the rest of int_const_binop_1 into int_const_binop.
6395 * fold-const.h (wide_int_binop): New.
6396 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
6397 Remove useless PLUS/MINUS_EXPR case.
6398 (zero_nonzero_bits_from_vr): Move wide int code...
6399 (zero_nonzero_bits_from_bounds): ...here.
6400 (extract_range_from_binary_expr_1): Move mask optimization code...
6401 (range_easy_mask_min_max): ...here.
6402 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
6403 (range_easy_mask_min_max): New.
6405 2018-07-15 Jeff Law <law@redhat.com>
6408 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
6411 2018-07-14 Jim Wilson <jimw@sifive.com>
6413 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
6415 2018-07-14 Paul Koning <ni1d@arrl.net>
6417 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
6419 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
6421 * lto-streamer-out.c (copy_function_or_variable): Dump info about
6424 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
6425 Steve Munroe <munroesj52@gmail.com>
6427 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
6428 (_mm_andnot_si128): Likewise.
6429 (_mm_or_si128): Likewise.
6430 (_mm_xor_si128): Likewise.
6432 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
6435 * builtins.c (expand_builtin_memcmp): Inline the calls first
6436 when result_eq is false.
6437 (expand_builtin_strcmp): Inline the calls first.
6438 (expand_builtin_strncmp): Likewise.
6439 (inline_string_cmp): New routine. Expand a string compare
6440 call by using a sequence of char comparison.
6441 (inline_expand_builtin_string_cmp): New routine. Inline expansion
6442 a call to str(n)cmp/memcmp.
6443 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
6445 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
6447 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
6449 * config/arm/driver-arm.c: Include arm-native.h.
6450 (host_detect_local_cpu): Use auto-generated data tables.
6451 (vendors, arm_cpu_table): Delete. Move part information to ...
6452 * config/arm/arm-cpus.in: ... here.
6453 * config/arm/parsecpu.awk (gen_native): New function.
6454 (vendor, part): New CPU fields.
6455 (END): Add support for building the native CPU detection tables.
6456 * config/arm/t-arm (arm-native.h): Add build rule.
6457 (driver-arm.o): Add dependency on arm-native.h.
6459 2018-07-13 Richard Biener <rguenther@suse.de>
6462 * match.pd (addr1 - addr2): Allow either of the operand to
6465 2018-07-13 Tom de Vries <tdevries@suse.de>
6467 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
6470 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
6472 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
6473 arrays instead of numbered variables.
6475 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
6477 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
6478 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
6480 2018-07-13 Richard Biener <rguenther@suse.de>
6483 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
6484 instead of get_context_die.
6486 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
6487 Richard Biener <rguenther@suse.de>
6490 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
6491 that the loop latch destination where phi is defined.
6493 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
6495 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
6496 (riscv_expand_epilogue): Add assertion to check interrupt mode.
6497 (riscv_set_current_function): Extract getting interrupt type to new
6499 (riscv_get_interrupt_type): New function.
6500 (riscv_merge_decl_attributes): New function, checking interrupt type is
6502 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
6504 2018-07-12 Paul Koning <ni1d@arrl.net>
6506 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
6509 2018-07-12 Paul Koning <ni1d@arrl.net>
6511 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
6512 zero reference, add doloop_end instead.
6513 * doc/md.texi (decrement_and_branch_until_zero): Remove.
6514 (Looping patterns): Remove decrement_and_branch_until_zero. Add
6515 detail for doloop_end.
6517 2018-07-12 Martin Sebor <msebor@redhat.com>
6520 * attribs.c (decl_attributes): Reject conflicting attributes before
6521 calling attribute handlers.
6523 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
6525 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
6527 (gcc::dump_manager::get_dump_file_name): likewise.
6528 (dump_begin): Likewise.
6529 * dumpfile.h (dump_begin): Update prototype.
6530 (gcc::dump_manager::get_dump_file_name,
6531 gcc::dump_manager::get_dump_file_name): Update prototype.
6533 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6535 * internal-fn.h (vectorizable_internal_fn_p): New function.
6536 * tree-vect-slp.c (compatible_calls_p): Likewise.
6537 (vect_build_slp_tree_1): Remove nops argument. Handle calls
6538 to internal functions.
6539 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
6541 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6543 * fold-const.h (inverse_conditions_p): Declare.
6544 * fold-const.c (inverse_conditions_p): New function.
6545 * match.pd: Use inverse_conditions_p. Add folds of view_converts
6546 that test the inverse condition of a conditional internal function.
6547 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
6548 * internal-fn.c (internal_fn_mask_index): Handle conditional
6550 (vectorized_internal_fn_supported_p): New function.
6551 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
6552 (any_pred_load_store): Replace with...
6553 (need_to_predicate): ...this new variable.
6554 (redundant_ssa_names): New variable.
6555 (ifcvt_can_use_mask_load_store): Move initial checks to...
6556 (ifcvt_can_predicate): ...this new function. Handle tree codes
6557 for which a conditional internal function exists.
6558 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
6559 instead of ifcvt_can_use_mask_load_store. Update after variable
6561 (predicate_load_or_store): New function, split out from
6562 predicate_mem_writes.
6563 (check_redundant_cond_expr): New function.
6564 (value_available_p): Likewise.
6565 (predicate_rhs_code): Likewise.
6566 (predicate_mem_writes): Rename to...
6567 (predicate_statements): ...this. Use predicate_load_or_store
6568 and predicate_rhs_code.
6569 (combine_blocks, tree_if_conversion): Update after above name changes.
6570 (ifcvt_local_dce): Handle redundant_ssa_names.
6571 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
6572 general conditional functions.
6573 * tree-vect-stmts.c (vectorizable_call): Likewise.
6575 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6576 Alan Hayward <alan.hayward@arm.com>
6577 David Sherwood <david.sherwood@arm.com>
6579 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
6580 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
6581 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
6582 plus and minus and convert them into IFN_COND_FMA-based sequences.
6583 (convert_mult_to_fma): Handle conditional plus and minus.
6585 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6587 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
6588 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
6589 (cond_fnms_optab): New optabs.
6590 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
6592 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
6593 * internal-fn.h (get_conditional_internal_fn): Declare.
6594 (get_unconditional_internal_fn): Likewise.
6595 * internal-fn.c (cond_ternary_direct): New macro.
6596 (expand_cond_ternary_optab_fn): Likewise.
6597 (direct_cond_ternary_optab_supported_p): Likewise.
6598 (FOR_EACH_COND_FN_PAIR): Likewise.
6599 (get_conditional_internal_fn): New function.
6600 (get_unconditional_internal_fn): Likewise.
6601 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
6602 (gimple_match_op::gimple_match_op): Add a new overload for 5
6604 (gimple_match_op::set_op): Likewise.
6605 (gimple_resimplify5): Declare.
6606 * genmatch.c (decision_tree::gen): Generate simplifications for
6608 * gimple-match-head.c (gimple_simplify): Define an overload for
6609 5 operands. Handle calls with 5 arguments in the top-level overload.
6610 (convert_conditional_op): Handle conversions from unconditional
6611 internal functions to conditional ones.
6612 (gimple_resimplify5): New function.
6613 (build_call_internal): Pass a fifth operand.
6614 (maybe_push_res_to_seq): Likewise.
6615 (try_conditional_simplification): Try converting conditional
6616 internal functions to unconditional internal functions.
6617 Handle 3-operand unconditional forms.
6618 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
6619 Define ternary equivalents of the current rules for binary conditional
6621 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
6623 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
6624 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
6625 (optab): Handle them.
6626 (SVE_COND_FP_TERNARY): New int iterator.
6627 (sve_fmla_op, sve_fmad_op): New int attributes.
6628 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
6629 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
6630 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
6632 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6634 * target.def (preferred_else_value): New target hook.
6635 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
6636 * doc/tm.texi: Regenerate.
6637 * targhooks.h (default_preferred_else_value): Declare.
6638 * targhooks.c (default_preferred_else_value): New function.
6639 * internal-fn.h (conditional_internal_fn_code): Declare.
6640 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
6641 (get_conditional_internal_fn): Use it.
6642 (conditional_internal_fn_code): New function.
6643 * gimple-match.h (gimple_match_cond): New struct.
6644 (gimple_match_op): Add a cond member function.
6645 (gimple_match_op::gimple_match_op): Update all forms to take a
6647 * genmatch.c (expr::gen_transform): Use the same condition as res_op
6648 for the suboperation, but don't specify a particular else_value.
6649 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
6650 (visit_nary_op, visit_reference_op_load): Pass
6651 gimple_match_cond::UNCOND to the gimple_match_op constructor.
6652 * gimple-match-head.c: Include tree-eh.h
6653 (convert_conditional_op): New function.
6654 (maybe_resimplify_conditional_op): Likewise.
6655 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
6656 (gimple_resimplify2): Likewise.
6657 (gimple_resimplify3): Likewise.
6658 (gimple_resimplify4): Likewise.
6659 (maybe_push_res_to_seq): Return null for conditional operations.
6660 (try_conditional_simplification): New function.
6661 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
6663 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
6665 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
6667 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
6669 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
6671 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
6673 (hash_tree): Do not hash DECL_FCONTEXT
6674 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
6675 Do not stream DECL_FCONTEXT.
6676 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
6677 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
6679 2018-07-12 Richard Biener <rguenther@suse.de>
6682 * dwarf2out.c (gen_block_die): Only output blocks when they have
6683 at least one !DECL_IGNORED_P variable.
6685 2018-07-12 Richard Biener <rguenther@suse.de>
6688 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
6689 Remove -mieee-fp handling.
6691 2018-07-12 Richard Biener <rguenther@suse.de>
6693 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
6694 left-over from last patch.
6696 2018-07-12 Jakub Jelinek <jakub@redhat.com>
6698 PR tree-optimization/86492
6699 * gimple-ssa-store-merging.c
6700 (imm_store_chain_info::coalesce_immediate_stores): Call
6701 check_no_overlap even for the merge_overlapping case. Formatting fix.
6703 2018-07-12 Richard Biener <rguenther@suse.de>
6706 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
6707 move possibly trapping operations into the conditional.
6709 2018-07-12 Richard Biener <rguenther@suse.de>
6711 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
6712 (vn_lookup_simplify_result): Remove recursion limit applied
6714 (vn_nary_build_or_lookup_1): Adjust.
6715 (try_to_simplify): Likewise.
6716 * gimple-match-head.c (gimple_resimplify1): Instead apply one
6718 (gimple_resimplify2): Likewise.
6719 (gimple_resimplify3): Likewise.
6720 (gimple_resimplify4): Likewise.
6722 2018-07-11 Jakub Jelinek <jakub@redhat.com>
6724 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
6725 Use __mmask64 type instead of __mmask8 for __M argument.
6726 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
6727 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
6729 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
6730 __mmask16 for __M argument.
6731 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
6732 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
6733 to __mmask16 instead of __mmask8.
6734 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
6735 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
6736 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
6737 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
6738 instead of __mmask16 for __U argument.
6739 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
6740 __mmask16 instead of __mmask8 for __U argument.
6741 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
6743 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
6745 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
6747 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
6749 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
6751 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
6753 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
6755 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
6757 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
6759 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
6761 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
6762 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
6763 return type as well as __M argument type and all casts from __mmask8
6765 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
6766 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
6767 return type as well as __M argument type and all casts from __mmask8
6769 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
6770 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
6771 return type as well as __M argument type and all casts from __mmask8
6773 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
6774 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
6775 return type as well as __M argument type and all casts from __mmask8
6777 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
6778 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
6781 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
6783 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
6784 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
6787 2018-07-11 Paul Koning <ni1d@arrl.net>
6789 * doc/md.texi (define_subst): Document how multiple occurrences of
6790 the same argument in the replacement pattern are handled.
6792 2018-07-11 Paul Koning <ni1d@arrl.net>
6794 * doc/extend.texi (Common Variable Attributes): Move "mode" into
6796 (Common Type Attributes): Add "mode" attribute.
6798 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
6800 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
6801 stream DECL_ORIGINAL_TYPE.
6802 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
6803 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
6804 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6805 Do not walk original type.
6806 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
6808 (write_ts_decl_non_common_tree_pointers): Do not stream
6810 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
6811 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
6813 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6815 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
6816 thread twice from the same starting edge.
6818 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6820 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
6821 * gimple.c (gimple_call_nonnull_result_p): ...here...
6822 (gimple_call_nonnull_arg): ...and here.
6823 * gimple.h (gimple_call_nonnull_result_p): New.
6824 (gimple_call_nonnull_arg): New.
6826 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
6828 * config/arm/arm-cpus.in: Move information from fpu field of each
6829 cpu definition to the isa field.
6830 * config/arm/parsecpu.awk (fpu): Delete match rule.
6831 (gen_comm_data): Don't add bits from the CPU's FPU entry.
6833 2018-07-11 Richard Biener <rguenther@suse.de>
6836 * dwarf2out.c (init_sections_and_labels): Use
6837 output_asm_line_debug_info consistently.
6838 (dwarf2out_early_finish): Likewise.
6839 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
6842 2018-07-11 Richard Biener <rguenther@suse.de>
6844 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
6845 Rework father_bb setting in a way to avoid propagating constants
6846 multiple times on a loop body.
6848 2018-07-10 Mark Wielaard <mark@klomp.org>
6851 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
6853 2018-07-10 Richard Biener <rguenther@suse.de>
6855 * hash-map.h (hash_map::iterator::operator*): Return
6856 references to key and value.
6858 2018-07-10 Jakub Jelinek <jakub@redhat.com>
6861 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
6862 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
6863 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
6864 (gimplify_omp_for): For composite loops, move outer
6865 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
6866 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
6867 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
6868 TREE_LIST for both the original class iterator and the "last" helper
6869 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
6870 loop, remember has_decl_expr from outer composite loops for the
6871 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
6873 2018-07-09 Martin Sebor <msebor@redhat.com>
6877 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
6878 accessing implicitly initialized array elements.
6879 * expr.c (string_constant): Handle string initializers of
6880 character arrays within aggregates.
6881 * gimple-fold.c (fold_array_ctor_reference): Add argument.
6882 Store element offset. As a special case, handle zero size.
6883 (fold_nonarray_ctor_reference): Same.
6884 (fold_ctor_reference): Add argument. Store subobject offset.
6885 * gimple-fold.h (fold_ctor_reference): Add argument.
6887 2018-07-09 Paul Koning <ni1d@arrl.net>
6889 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
6890 (pdp11_insn_cost): New function.
6891 (pdp11_md_asm_adjust): New function.
6892 (TARGET_INVALID_WITHIN_DOLOOP): Define.
6893 (pdp11_rtx_costs): Update to match machine better.
6894 (output_addr_const_pdp11): Correct format mismatch warnings.
6895 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
6896 * config/pdp11/pdp11.md: General change to add base_cost and/or
6897 length attributes for use by new pdp11_insn_cost function.
6898 (MIN_BRANCH): Correct definition.
6900 (doloop_end): Use standard pattern name for looping pattern.
6901 (doloop_end_nocc): New.
6902 (movsf): Add another constraint alternative.
6903 (zero_extendqihi2): Add constraint alternatives for not in place
6905 (zero_extendhisi2): Remove.
6906 (shift patterns): Add CC handling variants.
6910 (define_peephole2): New peephole to recognize mov that sets CC for
6913 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6915 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
6916 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
6919 2018-07-09 Jakub Jelinek <jakub@redhat.com>
6922 * real.c (real_nextafter): Return true if result is denormal.
6924 2018-07-09 Martin Liska <mliska@suse.cz>
6926 * common.opt: Add back wrongly removed attribute.
6928 2018-07-09 Richard Biener <rguenther@suse.de>
6931 * dwarf2out.c (gen_block_die): For an early generated DIE
6932 always output high/low PC attributes.
6934 2018-07-09 Tom de Vries <tdevries@suse.de>
6936 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
6937 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
6938 onto VAR_DECL with abstract origin.
6940 2018-07-07 Jim Wilson <jimw@sifive.com>
6942 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
6944 2018-07-07 Tom de Vries <tdevries@suse.de>
6946 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
6948 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
6950 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
6952 (combine_bound): Use wide-int overflow calculation instead of
6954 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
6956 * fold-const.c (int_const_binop_2): Same.
6957 (extract_muldiv_1): Same.
6958 (fold_div_compare): Same.
6959 (fold_abs_const): Same.
6961 * poly-int.h (add): Same.
6965 * predict.c (predict_iv_comparison): Same.
6966 * profile-count.c (slow_safe_scale_64bit): Same.
6967 * simplify-rtx.c (simplify_const_binary_operation): Same.
6968 * tree-chrec.c (tree_fold_binomial): Same.
6969 * tree-data-ref.c (split_constant_offset_1): Same.
6970 * tree-if-conv.c (idx_within_array_bound): Same.
6971 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
6972 * tree-ssa-phiopt.c (minmax_replacement): Same.
6973 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
6974 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
6975 * vr-values.c (vr_values::adjust_range_with_scev): Same.
6976 * wide-int.cc (wi::add_large): Same.
6977 (wi::mul_internal): Same.
6978 (wi::sub_large): Same.
6979 (wi::divmod_internal): Same.
6980 * wide-int.h: Change overflow type to overflow_type for neg, add,
6981 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
6982 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
6983 mul_internal, divmod_internal.
6984 (overflow_type): New enum.
6985 (accumulate_overflow): New.
6987 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
6989 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
6990 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
6992 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6994 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
6995 argument is checked for zero before entering loop, avoid checking again.
6997 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6999 * gimplify.h (generic_expr_could_trap_p): Set as global function.
7000 * gimplify.h (generic_expr_could_trap_p): Likwise.
7001 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
7003 2018-07-06 Jakub Jelinek <jakub@redhat.com>
7005 PR tree-optimization/86401
7006 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
7007 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
7008 (fold_bit_and_mask): ... here. New helper function for match.pd.
7009 * fold-const.h (fold_bit_and_mask): Declare.
7010 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
7012 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
7015 * target.def (translate_mode_attribute): New hook.
7016 * targhooks.h (default_translate_mode_attribute): Declare.
7017 * targhooks.c (default_translate_mode_attribute): New function.
7018 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
7019 * doc/tm.texi: Regenerate.
7020 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
7021 (rs6000_translate_mode_attribute): New function.
7023 2018-07-06 Paul Koning <ni1d@arrl.net>
7025 * doc/md.texi (define_split): Document DONE and FAIL.
7026 (define_peephole2): Ditto.
7028 2018-07-05 Jeff Law <law@redhat.com>
7030 PR tree-optimization/86010
7031 * tree-ssa-dse.c (compute_trims): More aggressively trim at
7032 both the head and tail of mem* and str* calls.
7034 2018-07-05 Jim Wilson <jimw@sifive.com>
7036 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
7038 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
7040 * config/aarch64/aarch64-simd.md: correct flags text for
7041 MIN_EXPR replacement.
7043 2018-07-05 James Clarke <jrtc27@jrtc27.com>
7045 * configure: Regenerated.
7047 2018-07-05 Carl Love <cel@us.ibm.com>
7049 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
7050 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
7051 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
7052 VSX_BUILTIN_DOUBLEL_V4SF.
7054 2018-07-05 Martin Sebor <msebor@redhat.com>
7057 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
7058 than its domain to compute its the upper bound of a char array.
7060 2018-07-05 Nathan Sidwell <nathan@acm.org>
7062 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
7063 * doc/cpp.texi: Update comment.
7064 * doc/tm.texi: Rebuilt.
7065 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
7066 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
7067 * doc/extend.texi (Backwards Compatibility): Clarify it is system
7068 headers affected by extern "C".
7069 * system.h: Poison NO_IMPLICIT_EXTERN_C.
7070 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
7071 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
7072 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
7073 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
7074 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
7075 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
7076 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
7077 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
7078 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
7079 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
7080 config/sparc/sp64-elf.h, config/spu/spu.h,
7081 config/stormy16/stormy16.h, config/v850/v850.h,
7082 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
7083 define NO_IMPLICIT_EXTERN_C.
7084 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
7086 2018-07-05 Tamar Christina <tamar.christina@arm.com>
7089 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
7090 instead of GET_MODE_SIZE when comparing Units.
7092 2018-07-05 Tamar Christina <tamar.christina@arm.com>
7095 * rtlanal.c (set_noop_p): Constrain on mode change,
7096 include hard-reg-set.h
7098 2018-07-05 Tamar Christina <tamar.christina@arm.com>
7100 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
7102 2018-07-05 Jakub Jelinek <jakub@redhat.com>
7105 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
7108 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
7110 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
7112 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
7115 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
7117 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
7119 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
7122 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
7124 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
7126 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
7127 neg_*_op* variables.
7129 2018-07-04 Martin Liska <mliska@suse.cz>
7131 * tree-switch-conversion.c: Define
7132 max_ratio_for_speed and max_ratio_for_size constants.
7134 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
7135 Martin Liska <mliska@suse.cz>
7140 * common.opt: Rename align options with 'str_' prefix.
7141 * common/config/i386/i386-common.c (set_malign_value): New
7143 (ix86_handle_option): Use it to set -falign-* options/
7144 * config/aarch64/aarch64-protos.h (struct tune_params): Change
7145 type from int to string.
7146 * config/aarch64/aarch64.c: Update default values from int
7148 * config/alpha/alpha.c (alpha_override_options_after_change):
7150 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
7151 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7152 max skip conditionally.
7153 * config/i386/freebsd.h (SUBALIGN_LOG): New.
7154 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7155 max skip conditionally.
7156 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7157 max skip conditionally.
7158 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
7159 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7160 max skip conditionally.
7161 * config/i386/i386.c (struct ptt): Change type from int to
7163 (ix86_default_align): Set default values.
7164 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
7165 max skip conditionally.
7166 * config/i386/iamcu.h (SUBALIGN_LOG): New.
7167 (ASM_OUTPUT_MAX_SKIP_ALIGN):
7168 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
7169 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7170 max skip conditionally.
7171 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
7172 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
7173 * config/i386/x86-64.h (SUBALIGN_LOG): New.
7174 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7175 max skip conditionally.
7176 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
7177 * config/ia64/ia64.c (ia64_option_override): Set default values
7178 for alignment options.
7179 * config/m68k/m68k.c: Handle new str_align_* options.
7180 * config/mips/mips.c (mips_set_compression_mode): Change
7182 (mips_option_override): Set default values for options.
7183 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
7185 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7187 * config/rx/rx.c (rx_option_override): Likewise.
7188 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
7189 (LABEL_ALIGN): Use align_labels_log.
7190 (LOOP_ALIGN): Use align_loops_align.
7191 * config/s390/s390.c (s390_asm_output_function_label): Use new
7193 * config/sh/sh.c (sh_override_options_after_change):
7194 Change type of constants.
7195 * config/spu/spu.c (spu_sched_init): Likewise.
7196 * config/sparc/sparc.c (sparc_option_override): Set default
7198 * config/visium/visium.c (visium_option_override): Likewise.
7199 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
7200 emit p2align format with last argument if it's not needed.
7201 * doc/invoke.texi: Document extended format of -falign-*.
7202 * final.c: Use align_labels alignment.
7203 * flags.h (struct target_flag_state): Change type to use
7205 (struct align_flags_tuple): New.
7206 (struct align_flags): Likewise.
7207 (align_loops_log): Redefine macro to use new types.
7208 (align_loops_max_skip): Redefine macro to use new types.
7209 (align_jumps_log): Redefine macro to use new types.
7210 (align_jumps_max_skip): Redefine macro to use new types.
7211 (align_labels_log): Redefine macro to use new types.
7212 (align_labels_max_skip): Redefine macro to use new types.
7213 (align_functions_log): Redefine macro to use new types.
7214 (align_loops): Redefine macro to use new types.
7215 (align_jumps): Redefine macro to use new types.
7216 (align_labels): Redefine macro to use new types.
7217 (align_functions): Redefine macro to use new types.
7218 (align_functions_max_skip): Redefine macro to use new types.
7219 (align_loops_value): New macro.
7220 (align_jumps_value): New macro.
7221 (align_labels_value): New macro.
7222 (align_functions_value): New macro.
7223 * function.c (invoke_set_current_function_hook): Propagate
7224 alignment values from flags to global variables default in
7226 * ipa-icf.c (sem_function::equals_wpa): Use
7227 cl_optimization_option_eq instead of memcmp.
7228 * lto-streamer.h (cl_optimization_stream_out): Support streaming
7230 (cl_optimization_stream_in): Likewise.
7231 * optc-save-gen.awk: Support strings in cl_optimization.
7232 * opth-gen.awk: Likewise.
7233 * opts.c (finish_options): Remove error checking of invalid
7235 (MAX_CODE_ALIGN): Remove.
7236 (MAX_CODE_ALIGN_VALUE): Likewise.
7237 (parse_and_check_align_values): New function.
7238 (check_alignment_argument): Likewise.
7239 (common_handle_option): Use check_alignment_argument.
7240 * opts.h (parse_and_check_align_values): Declare.
7241 * toplev.c (init_alignments): Remove.
7242 (read_log_maxskip): New.
7243 (parse_N_M): Likewise.
7244 (parse_alignment_opts): Likewise.
7245 (backend_init_target): Remove usage of init_alignments.
7246 * toplev.h (parse_alignment_opts): Declare.
7247 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
7249 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
7250 * tree.c (cl_option_hasher::equal): New.
7251 * varasm.c: Use new global macros.
7253 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
7255 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
7256 Use a simpler align directive also if MAXSKIP = ALIGN-1.
7257 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7258 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7259 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7260 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
7261 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
7262 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
7263 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
7264 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7265 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7266 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7267 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7269 2018-07-04 Martin Liska <mliska@suse.cz>
7270 Jonathan Wakely <jwakely@redhat.com>
7272 * coverage.c: Use correct type.
7273 * doc/invoke.texi: Language correction.
7275 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
7278 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
7279 ENDBRANCH for non-tail call which may return via indirect branch.
7280 * doc/extend.texi: Document indirect_return attribute.
7282 2018-07-03 Martin Sebor <msebor@redhat.com>
7284 PR tree-optimization/86274
7285 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
7287 (format_floating): Correct handling of infinities and NaNs.
7289 2018-07-03 Martin Sebor <msebor@redhat.com>
7291 * print-tree.c (print_real_cst): New function.
7292 (print_node_brief): Call it.
7293 (print_node): Ditto.
7295 2018-07-03 Jeff Law <law@redhat.com>
7297 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
7298 into a single pattern.
7300 * config/h8300/h8300.md (ors code_iterator): New.
7301 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
7302 a single pattern and single splitter.
7303 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
7304 (iorqi3_1, xorqi3_1): Likewise.
7305 (iorqi3, xorqi3 expanders): Similarly.
7307 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
7308 (movmd_internal) into a single pattern using the P mode iterator.
7309 (movmd splitters): Similarly.
7310 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
7311 (movsd splitters): Similarly.
7313 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
7314 ADDB, ADDW and ADDL into a single ADD attribute which selects the
7315 right table based on the size of the operand.
7316 * config/h8300/h8300.md (length_table): Corresponding changes. All
7317 references to "addb", "addw" and "addl" changed to "add".
7318 (btst patterns): Merge two variants into a single pattern.
7319 (tstqi, tsthi): Likewise.
7320 (addhi3_incdec, addsi3_incdec): Likewise.
7321 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
7322 (mulhi3, mulsi3): Likewise.
7323 (udivhi3, udivsi3): Likewise.
7324 (divhi3, divsi3): Likewise.
7325 (andorqi3, andorhi3, andorsi3): Likewise.
7327 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
7330 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
7331 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
7333 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7335 PR tree-optimization/85694
7336 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
7337 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
7339 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
7340 (<u>avg<mode>3_ceil): New patterns.
7342 2018-07-03 David Malcolm <dmalcolm@redhat.com>
7344 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
7345 scan-tree-dump directive.
7346 * gcc.dg/vect/slp-perm-2.c: Likewise.
7347 * gcc.dg/vect/slp-perm-3.c: Likewise.
7348 * gcc.dg/vect/slp-perm-5.c: Likewise.
7349 * gcc.dg/vect/slp-perm-6.c: Likewise.
7350 * gcc.dg/vect/slp-perm-7.c: Likewise.
7351 * gcc.dg/vect/slp-perm-8.c: Likewise.
7353 2018-07-03 Marek Polacek <polacek@redhat.com>
7356 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
7358 2018-07-03 Richard Biener <rguenther@suse.de>
7361 * tree-ssa-structalias.c (find_func_clobbers): Properly
7362 handle indirect calls.
7364 2018-07-03 Jeff Law <law@redhat.com>
7366 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
7367 (shifts): New code iterator.
7368 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
7369 expander. Fix HImode handling on H8/SX.
7370 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
7371 (subqi3, subhi3, subsi3 expanders): Likewise.
7372 (andqi3, andhi3, andsi3 expanders): Likewise.
7373 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
7374 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
7375 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
7376 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
7377 (zero_extendqihi2, zero_extendqisi2): Likewise.
7378 (extendqihi2, extendqisi2): Likewise.
7379 (rotlqi3, rotlhi3, rotlsi3): Likewise.
7380 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
7381 (rotlqi3_1, rotlhi3_1): Likewise.
7382 (logicalhi3_sn, logicalsi3_sn): Likewise.
7383 (logicalhi3, logicalsi3): Likewise.
7385 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7387 * tree-vect-patterns.c (vect_recog_rotate_pattern)
7388 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7389 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
7390 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
7391 type to append_pattern_def_seq instead of creating a stmt_vec_info
7393 (build_mask_conversion): Likewise. Remove vinfo argument.
7394 (vect_add_conversion_to_patterm): Likewise, renaming to...
7395 (vect_add_conversion_to_pattern): ...this.
7396 (vect_recog_mask_conversion_pattern): Update call to
7397 build_mask_conversion. Pass the vector type to
7398 append_pattern_def_seq here too.
7399 (vect_recog_gather_scatter_pattern): Update call to
7400 vect_add_conversion_to_pattern.
7402 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7404 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
7405 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7406 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
7407 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
7408 STMT_VINFO_PATTERN_DEF_SEQ to null here.
7409 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
7410 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
7411 append_pattern_def_seq instead of new_pattern_def_seq.
7412 (vect_recog_divmod_pattern): Do both of the above.
7413 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
7416 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7418 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
7419 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
7420 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
7421 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
7422 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
7423 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7424 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
7425 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
7426 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
7427 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
7428 parameter with a single stmt_vec_info.
7429 (vect_recog_func_ptr): Likewise.
7430 (vect_recog_gather_scatter_pattern): Likewise, folding in...
7431 (vect_try_gather_scatter_pattern): ...this.
7432 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
7433 the stmt_vec_info of the statement to be matched. Don't clear
7434 STMT_VINFO_RELATED_STMT.
7435 (vect_pattern_recog): Update call accordingly.
7437 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7439 PR tree-optimization/85694
7440 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
7441 (uavgM3_ceil): Document new optabs.
7442 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
7443 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
7445 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
7446 (savg_ceil_optab): New optabs.
7447 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
7448 (vect_vect_recog_func_ptrs): Add it.
7449 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
7450 constant directly from the associated lhs.
7452 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7454 * tree-vect-patterns.c (vect_split_statement): New function.
7455 (vect_convert_input): Use it to try to split an existing cast.
7457 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7459 * poly-int.h (print_hex): New function.
7460 * dumpfile.h (dump_dec, dump_hex): Declare.
7461 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
7462 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
7463 min_input_precision, operation_precision and operation_sign.
7464 * tree-vect-patterns.c (vect_get_range_info): New function.
7465 (vect_same_loop_or_bb_p, vect_single_imm_use)
7466 (vect_operation_fits_smaller_type): Delete.
7467 (vect_look_through_possible_promotion): Add an optional
7468 single_use_p parameter.
7469 (vect_recog_over_widening_pattern): Rewrite to use new
7470 stmt_vec_info infomration. Handle one operation at a time.
7471 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
7472 (vect_truncatable_operation_p, vect_set_operation_type)
7473 (vect_set_min_input_precision): New functions.
7474 (vect_determine_min_output_precision_1): Likewise.
7475 (vect_determine_min_output_precision): Likewise.
7476 (vect_determine_precisions_from_range): Likewise.
7477 (vect_determine_precisions_from_users): Likewise.
7478 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
7479 (vect_vect_recog_func_ptrs): Put over_widening first.
7481 (vect_pattern_recog): Call vect_determine_precisions.
7483 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7485 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
7486 statements that have been replaced by further pattern statements.
7487 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
7489 2018-07-03 Richard Biener <rguenther@suse.de>
7491 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
7492 always set *dt. Dump vectype in vectype overload.
7493 * dumpfile.h (dump_gimple_expr): New function.
7494 (dump_gimple_expr_loc): Likewise.
7495 * dumpfile.c (dump_gimple_expr): New function.
7496 (dump_gimple_expr_loc): Likewise.
7498 2018-07-02 Jeff Law <law@redhat.com>
7500 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
7501 the H8/300, H8/300H and H8/S variants into a single pattern.
7502 (movhi_h8300, movqi_h8300hs): Similarly.
7503 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
7504 (QHI mode iterator): New.
7506 * config/h8300/h8300.md: Remove trailing whitespace.
7508 2018-07-02 Jim Wilson <jimw@sifive.com>
7510 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
7511 instead of emit_insn for interrupt returns.
7512 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
7513 (riscv_sret, riscv_uret): Likewise.
7515 2018-07-02 David Malcolm <dmalcolm@redhat.com>
7517 * pretty-print.c (selftest::test_pp_format): Move save and restore
7518 of quotes to class auto_fix_quotes, and add an instance.
7519 * selftest.c: Include "intl.h".
7520 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
7521 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
7522 * selftest.h (selftest::auto_fix_quotes): New class.
7524 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
7526 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
7527 (aarch64_sve_prepare_conditional_op): Remove.
7528 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
7529 Allow aarch64_simd_reg_or_zero as select operand; remove
7530 the aarch64_sve_prepare_conditional_op call.
7531 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
7532 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
7533 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
7534 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
7535 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
7536 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
7537 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
7538 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
7539 and a splitters to match all of the *_any patterns.
7540 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
7542 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
7543 (SVE_COND_FP_BINARY_REV): Remove.
7544 (sve_int_op_rev, sve_fp_op_rev): New.
7545 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
7546 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
7547 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
7548 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
7549 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
7550 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
7551 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
7552 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
7553 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
7555 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
7556 Remove match_dup 1 from the inner unspec.
7557 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
7559 * config/aarch64/aarch64.md (movprfx): New attr.
7560 (length): Default movprfx to 8.
7561 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
7562 (*madd<SVE_I>, *msub<SVE_I): Likewise.
7563 (*<su>mul<SVE_I>3_highpart): Likewise.
7564 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
7565 (*v<ASHIFT><SVE_I>3): Likewise.
7566 (*<su><MAXMIN><SVE_I>3): Likewise.
7567 (*<su><MAXMIN><SVE_F>3): Likewise.
7568 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
7569 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
7570 (*div<SVE_F>4): Likewise.
7572 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
7574 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
7577 2018-07-02 Richard Biener <rguenther@suse.de>
7579 PR tree-optimization/86363
7580 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
7581 memset argument refers to a non-variable address.
7583 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
7585 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
7586 {PLUS,MINUS}_EXPR code to...
7587 (adjust_symbolic_bound): ...here,
7588 (combine_bound): ...here,
7589 (set_value_range_with_overflow): ...and here.
7591 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
7593 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
7595 (extract_range_from_abs_expr): ...here.
7597 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
7599 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
7600 -fno-omit-frame-pointer when not optimizing.
7602 2018-07-02 Martin Liska <mliska@suse.cz>
7605 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
7606 (propagate_nothrow): Likewise.
7608 2018-07-02 Martin Liska <mliska@suse.cz>
7611 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
7613 2018-07-02 David Malcolm <dmalcolm@redhat.com>
7615 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
7616 function in r262149, changing "loc" param from source_location to
7617 const dump_location_t &.
7618 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
7619 declaration, as above.
7621 2018-07-01 Paul Koning <ni1d@arrl.net>
7623 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
7624 -munit-asm, -mgnu-asm, -mdec-asm.
7625 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
7626 (pdp11_output_labelref): New.
7627 (pdp11_output_def): New.
7628 (pdp11_output_addr_vec_elt): New.
7629 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
7630 %# and %@ format codes.
7631 (pdp11_option_override): New.
7632 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
7633 (pdp11_output_ident): New.
7634 (pdp11_asm_named_section): New.
7635 (pdp11_asm_init_sections): New.
7636 (pdp11_file_start): New.
7637 (pdp11_file_end): New.
7638 (output_ascii): Use .ascii/.asciz for -mdec-asm.
7639 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
7640 %o, like %c but octal.
7641 (pdp11_option_override): New.
7642 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
7644 (DATA_SECTION_ASM_OP): Ditto.
7645 (READONLY_DATA_SECTION_ASM_OP): New.
7646 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
7647 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
7648 (ASM_OUTPUT_LABELREF): Ditto.
7649 (ASM_OUTPUT_DEF): Ditto.
7650 (ASM_OUTPUT_EXTERNAL): New.
7651 (ASM_OUTPUT_SOURCE_FILENAME): New.
7652 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
7653 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
7654 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
7655 %# and %@ format codes.
7656 * config/pdp11/pdp11.opt (mgnu-asm): New.
7657 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
7658 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
7659 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
7661 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
7663 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
7664 dereferencing path[] beyond its length.
7666 (debug_all_paths): New.
7667 (rewire_first_differing_edge): New.
7668 (adjust_paths_after_duplication): New.
7669 (duplicate_thread_path): Call adjust_paths_after_duplication.
7671 (thread_through_all_blocks): Add new argument to
7672 duplicate_thread_path.
7674 2018-06-30 Jim Wilson <jimw@sifive.com>
7676 * config/riscv/predicates.md (p2m1_shift_operand): New.
7677 (high_mask_shift_operand): New.
7678 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
7679 pattern using p2m1_shift_operand.
7680 (lshsi3_zero_extend_3+2): New combiner pattern using
7681 high_mask_shift_operand.
7683 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7685 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
7687 (vect_recog_rotate_pattern): ...here.
7688 (vect_convert_input): Try to insert casts of invariants in the
7690 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
7691 preheader to be empty.
7693 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7695 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
7696 vector type. If given, install it in the new statement's
7698 (vect_element_precision): New function.
7699 (vect_unpromoted_value): New struct.
7700 (vect_unpromoted_value::vect_unpromoted_value): New function.
7701 (vect_unpromoted_value::set_op): Likewise.
7702 (vect_look_through_possible_promotion): Likewise.
7703 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
7704 (vect_widened_op_tree, vect_convert_input): Likewise.
7705 (vect_convert_inputs, vect_convert_output): Likewise.
7706 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
7707 to handle the optional cast of the multiplication result and
7708 vect_widened_op_tree to detect the widened multiplication itself.
7709 Do not require the input and output of promotion casts to have
7710 the same sign, but base the signedness of the operation on the
7711 input rather than the result. If the pattern includes two
7712 promotions, check that those promotions have the same sign.
7713 Do not restrict the MULT_EXPR handling to a double-width result;
7714 handle quadruple-width results and wider. Use vect_convert_inputs
7715 to convert the inputs to the common type.
7716 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
7717 to handle the optional cast of the ABS result. Also allow a sign
7718 change or a sign extension between the ABS and MINUS.
7719 Use vect_widened_op_tree to detect the widened subtraction and use
7720 vect_convert_inputs to convert the inputs to the common type.
7721 (vect_handle_widen_op_by_const): Delete.
7722 (vect_recog_widen_op_pattern): New function.
7723 (vect_recog_widen_mult_pattern): Use it.
7724 (vect_recog_widen_shift_pattern): Likewise.
7725 (vect_recog_widen_sum_pattern): Use
7726 vect_look_through_possible_promotion to handle the promoted
7729 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7731 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
7732 the containing gimple_seq *.
7733 * gimple-iterator.h (gsi_for_stmt): Declare it.
7734 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
7735 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
7736 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7737 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7738 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
7740 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
7742 (vect_mark_pattern_stmts): ...here. Handle cases in which the
7743 statement being replaced is part of an existing pattern
7744 definition sequence, inserting the new pattern statements before
7746 (vect_pattern_recog_1): Don't return a bool. If the statement
7747 is already part of a pattern, instead apply pattern matching
7748 to the pattern definition statements. Don't clear the
7749 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
7750 (vect_pattern_recog): Don't break after the first match;
7751 continue processing the pattern definition statements instead.
7752 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
7754 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7756 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
7757 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7758 (vect_recog_widen_sum_pattern): Use it.
7760 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7762 * tree-vect-loop.c (vectorizable_reduction): Assert that the
7763 phi is not a pattern statement and has not been replaced by
7764 a pattern statement.
7765 * tree-vect-patterns.c (type_conversion_p): Don't check
7766 STMT_VINFO_IN_PATTERN_P.
7767 (vect_recog_vector_vector_shift_pattern): Likewise.
7768 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
7769 the pattern statement rather than the original statement; check
7770 directly for a WIDEN_MULT_EXPR here.
7771 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
7772 vect_is_simple_use to return the pattern statement rather
7773 than the original statement; use is_pattern_stmt_p to check
7774 for such a pattern statement.
7775 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
7776 to return the pattern statement rather than the original statement;
7777 don't do the same transformation here.
7778 (vect_is_simple_use): If the defining statement has been replaced
7779 by a pattern statement, return the pattern statement instead.
7780 Remove the corresponding (local) transformation from the vectype
7783 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7785 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
7786 end and default to null.
7787 * tree-vect-loop.c (vect_create_epilog_for_reduction)
7788 (vectorizable_reduction): Update calls accordingly, dropping the
7789 gimple ** argument if the passed-back statement isn't needed.
7790 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
7791 (vect_recog_rotate_pattern): Likewise.
7792 (vect_recog_mask_conversion_pattern): Likewise.
7793 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
7794 (vect_mask_constant_operand_p): Likewise.
7795 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
7796 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
7797 (get_group_load_store_type, get_load_store_type): Likewise.
7798 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
7799 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
7800 (vectorizable_conversion, vectorizable_assignment): Likewise.
7801 (vectorizable_shift, vectorizable_operation): Likewise.
7802 (vectorizable_store, vect_is_simple_cond): Likewise.
7803 (vectorizable_condition, vectorizable_comparison): Likewise.
7804 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
7805 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
7806 and move it to the end. Cope with null def_stmt_outs.
7808 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7810 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
7812 2018-06-29 Jeff Law <law@redhat.com>
7814 * config/v850/v850.c (v850_legitimate_address_p): Handle large
7815 displacements for TARGET_V850E2V3 and newer.
7816 (TARGET_LRA_P): Remove. Defaults to LRA now.
7817 * config/v850/v850.md (sign23byte_load): Remove.
7818 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
7819 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
7821 2018-06-29 Martin Liska <mliska@suse.cz>
7824 * coverage.c (coverage_init): Mangle full path name.
7825 * doc/invoke.texi: Document the change.
7826 * gcov-io.c (mangle_path): New.
7827 * gcov-io.h (mangle_path): Likewise.
7828 * gcov.c (mangle_name): Use mangle_path for path mangling.
7830 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7832 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
7833 if starting source register is not even.
7835 2018-06-29 Martin Liska <mliska@suse.cz>
7837 PR tree-optimization/86263
7838 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
7839 Make edge redirection.
7841 2018-06-29 David Malcolm <dmalcolm@redhat.com>
7843 * dumpfile.c (dump_loc): Add indentation based on scope depth.
7844 (dump_scope_depth): New variable.
7845 (get_dump_scope_depth): New function.
7846 (dump_begin_scope): New function.
7847 (dump_end_scope): New function.
7848 * dumpfile.h (get_dump_scope_depth): New declaration.
7849 (dump_begin_scope): New declaration.
7850 (dump_end_scope): New declaration.
7851 (class auto_dump_scope): New class.
7852 (AUTO_DUMP_SCOPE): New macro.
7853 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
7856 2018-06-29 Richard Biener <rguenther@suse.de>
7858 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
7859 compute_all_dependences succeeds.
7860 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
7861 exceed --param loop-max-datarefs-for-datadeps.
7863 2018-06-29 Jakub Jelinek <jakub@redhat.com>
7865 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
7867 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
7870 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
7871 alternative 0 in preferred_for_speed attribute.
7873 2018-06-28 Paul Koning <ni1d@arrl.net>
7875 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
7876 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
7877 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
7878 * config/pdp11/pdp11.md: Correct "length" attribute calculation
7879 for shift insn patterns.
7881 2018-06-28 David Malcolm <dmalcolm@redhat.com>
7883 * cgraph.c (cgraph_node::get_body): Replace assignments to
7884 "dump_file" with calls to set_dump_file.
7885 * dumpfile.c (alt_dump_file): Make static, and group with...
7886 (alt_flags): ...this definition.
7887 (dumps_are_enabled): New variable.
7888 (refresh_dumps_are_enabled): New function.
7889 (set_dump_file): New function.
7890 (set_alt_dump_file): New function.
7891 (gcc::dump_manager::dump_start): Replace assignments to
7892 "dump_file" and "alt_dump_file" with calls to set_dump_file and
7894 (gcc::dump_manager::dump_finish): Likewise.
7895 * dumpfile.h (alt_dump_file): Delete decl.
7896 (dumps_are_enabled): New variable decl.
7897 (set_dump_file): New function decl.
7898 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
7900 * tree-nested.c (lower_nested_functions): Replace assignments to
7901 "dump_file" with calls to set_dump_file.
7903 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7905 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
7906 goto_locus of each outgoing edge of each basic block.
7908 2018-06-28 Richard Biener <rguenther@suse.de>
7910 * dwarf2out.c (decl_scope_table): Remove.
7911 (push_decl_scope): Likewise.
7912 (pop_decl_scope): Likewise.
7913 (gen_type_die_for_member): Do not call push/pop_decl_scope.
7914 (gen_struct_or_union_type_die): Likewise.
7915 (gen_tagged_type_die): Likewise.
7916 (dwarf2out_init): Do not initialize decl_scope_table.
7917 (dwarf2out_c_finalize): Do not free it.
7919 2018-06-28 Richard Biener <rguenther@suse.de>
7921 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
7922 deciding whether to not re-use a DIE.
7924 2018-06-28 Richard Biener <rguenther@suse.de>
7926 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
7927 DW_AT_abstract_origin attribute.
7929 2018-06-28 Martin Liska <mliska@suse.cz>
7931 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
7932 Use newly introduced constants.
7933 * tree-switch-conversion.h (struct jump_table_cluster):
7934 Define max_ratio_for_size and max_ratio_for_speed.
7936 2018-06-28 Martin Liska <mliska@suse.cz>
7938 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7939 Add new checking assert to catch invalid state.
7940 (jump_table_cluster::can_be_handled): Handle single case
7942 (jump_table_cluster::is_beneficial): Bail out for such case.
7943 (bit_test_cluster::find_bit_tests):
7944 Add new checking assert to catch invalid state.
7945 (bit_test_cluster::can_be_handled): Handle single case
7947 (bit_test_cluster::is_beneficial): Bail out for such case.
7948 (switch_decision_tree::analyze_switch_statement):
7951 2018-06-28 Martin Liska <mliska@suse.cz>
7953 * common.opt: Introduce -completion option.
7954 * gcc.c (driver_handle_option): Handle it.
7955 (driver::main): Print completions if completion
7957 * opt-suggestions.c (option_proposer::get_completions):
7959 (option_proposer::suggest_completion): Likewise.
7960 (option_proposer::find_param_completions): Likewise.
7961 (verify_autocompletions): Likewise.
7962 (test_completion_valid_options): Likewise.
7963 (test_completion_valid_params): Likewise.
7964 (in_completion_p): Likewise.
7965 (empty_completion_p): Likewise.
7966 (test_completion_partial_match): Likewise.
7967 (test_completion_garbage): Likewise.
7968 (opt_proposer_c_tests): Likewise.
7969 * opt-suggestions.h: Declare new functions.
7970 * opts.c (common_handle_option): Handle OPT__completion_.
7971 * selftest-run-tests.c (selftest::run_tests): Add
7972 opt_proposer_c_tests.
7973 * selftest.c (assert_str_startswith): New.
7974 * selftest.h (assert_str_startswith): Likewise.
7975 (opt_proposer_c_tests): New.
7976 (ASSERT_STR_STARTSWITH): Likewise.
7978 2018-06-28 Martin Liska <mliska@suse.cz>
7980 * Makefile.in: Add opt-suggestions.o.
7981 * gcc-main.c: Include opt-suggestions.h.
7982 * gcc.c (driver::driver): Likewise.
7983 (driver::~driver): Remove m_option_suggestions.
7984 (driver::build_option_suggestions): Moved to option_proposer.
7985 (driver::suggest_option): Likewise.
7986 (driver::handle_unrecognized_options): Use option_proposer.
7987 * gcc.h (class driver): Add new memver m_option_proposer.
7988 * opt-suggestions.c: New file.
7989 * opt-suggestions.h: New file.
7991 2018-06-28 Martin Liska <mliska@suse.cz>
7993 * vec.h (class auto_string_vec): New (moved from auto_argvec).
7994 (auto_string_vec::~auto_string_vec): Likewise.
7996 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7998 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
7999 prevent_decl_creation_for_types fields up and add reset_location field.
8000 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
8001 statement if id->reset_location is true.
8002 (copy_edges_for_bb): Do not set goto_locus on the new edges if
8003 id->reset_location is true.
8004 (copy_phis_for_bb): Force input_location on the arguments if
8005 id->reset_location is true.
8006 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
8007 is set on the function to be inlined.
8009 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
8011 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
8013 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
8015 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
8016 registers for Pmode.
8017 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
8018 hard registers for the clobbered pseudo.
8020 2018-06-27 Paul Koning <ni1d@arrl.net>
8022 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
8023 mutually exclusive options.
8024 * config/pdp11/constraints.md (h): New constraint.
8025 (O): Update definition to match shift code generation.
8026 (D): New constraint.
8027 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
8029 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
8031 (output_jump): Change arguments.
8032 (pdp11_fixed_cc_regs): New function.
8033 (pdp11_cc_mode): Ditto.
8034 (pdp11_expand_shift): Ditto.
8035 (pdp11_assemble_shift): Ditto.
8036 (pdp11_small_shift): Ditto.
8037 (pdp11_branch_cost): Remove.
8038 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
8040 (pdp11_register_move_cost): Update for CC registers.
8041 (pdp11_rtx_costs): Add case for LSHIFTRT.
8042 (pdp11_output_jump): Add CCNZ mode conditional branches.
8043 (notice_update_cc_on_set): Remove.
8044 (pdp11_cc_mode): New function.
8045 (simple_memory_operand): Correct pre/post decrement case.
8046 (no_side_effect_operand): New function.
8047 (pdp11_regno_reg_class): Add CC_REGS class.
8048 (pdp11_fixed_cc_regs): New function.
8049 (pdp11_small_shift): New function.
8050 (pdp11_expand_shift): New function to expand shift insns.
8051 (pdp11_assemble_shift): New function to output shifts.
8052 (pdp11_branch_cost): Remove.
8053 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
8054 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
8055 (WCHAR_TYPE): Ditto.
8056 (PTRDIFF_TYPE): Ditto.
8057 (ADJUST_INSN_LENGTH): New macro.
8058 (FIXED_REGISTERS): Add CC registers.
8059 (CALL_USED_REGISTERS): Ditto.
8061 (REG_CLASS_NAMES): Ditto.
8062 (REG_CLASS_CONTENTS): Ditto.
8063 (SELECT_CC_MODE): Use new function.
8064 (TARGET_FLAGS_REGNUM): New macro.
8065 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
8066 (cc0_reg_rtx): Remove.
8067 (CC_STATUS_MDEP): Remove.
8068 (CC_STATUS_MDEFP_INIT): Remove.
8069 (CC_IN_FPU): Remove.
8070 (NOTICE_UPDATE_CC): Remove.
8071 (REGISTER_NAMES): Add CC registers.
8072 (BRANCH_COST): Change to constant 1.
8073 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
8075 * config/pdp11/pdp11.opt (mbcopy): Remove.
8076 (mbcopy-builtin): Remove.
8077 (mbranch-cheap): Remove.
8078 (mbranch-expensive): Remove.
8079 * config/pdp11/predicates.md (expand_shift_operand): Update to
8080 match shift code generation.
8081 (ccnz_operator): New predicate.
8082 * doc/invoke.texi (PDP-11 Options): Remove deleted options
8083 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
8084 Remove non-existent option -mabshi, -mno-abshi. Document mutually
8086 * doc/md.texi (PDP-11): Document new D and h constraints. Update
8087 description of O constraint.
8089 2018-06-27 Jeff Law <law@redhat.com>
8090 Austin Law <austinklaw@gmail.com>
8092 * config/v850/v850.md (addsi3_set_flags): New pattern.
8093 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
8094 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
8095 (zero_extendhisi2_v850_set_flags): Likewise.
8096 (zero_extendqisi2_v850_set_flags): Likewise.
8097 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
8098 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
8099 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
8101 * config/v850/v850-protos.h (notice_update_cc): Remove.
8102 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
8103 (v850_print_operand): Handle 'D' and "d".
8104 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
8105 Add handling of arithmetic/logical operations compared against zero.
8106 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
8107 Do not look at v850_compare_op, instead get mode from last argument.
8108 (v850_gen_compare): Remove
8109 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
8110 after reload for prologue insns.
8111 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
8113 (construct_save_jarl): Likewise.
8114 (TARGET_FLAGS_REGNUM): Define.
8115 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
8116 (NOTICE_UPDATE_CC): Remove.
8117 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
8118 than cc0. Conditionalize on reload_completed.
8119 (cmpsi_insn, setfcc_insn): Likewise.
8120 (tst1 splitter): Turn into define_and_split which sets the flags
8122 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
8123 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
8124 (cstoresf4, cstoredf4): Clobber the flags.
8125 (cmpsi, cmpsf, cmpdf): Remove expanders.
8126 (setf_insn): Remove pattern.
8127 (addsi3): Turn into define_and_split which clobbers the flags after
8128 reload and a suitable pattern (addsi3_clobber_flags) for use after
8130 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
8131 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
8132 (ashrsi3, ashrsi3_v850e2): Likewise.
8133 (bins): Clobber the flags.
8134 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
8135 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
8136 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
8137 (call_value_internal_short, call_value_internal_long): Likewise.
8138 (callt_save_interrupt, callt_return_interrupt): Likewise.
8139 (save_interrupt, return_interrupt): Likewise.
8140 (callt_save_all_interrupt, save_all_interrupt): Likewise.
8141 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
8142 (restore_all_interrupt, _restore_all_interrupt): Likewise.
8143 (All FP comparisons): Only allow after reload has completed.
8145 (divh, divhu): Tweak output template.
8146 (branch_z_normal, branch_z_invert): Remove
8147 (branch_nz_normal, branch_nz_invert): Likewise.
8148 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
8150 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
8151 * config/v850/v850.c (notice_update_cc): Remove.
8152 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
8153 (CC_NO_CARRY): Likewise.
8154 (NOTICE_UPDATE_CC): Define to nothing.
8155 * config/v850/v850.md: Remove block comment on cc0 handling
8156 Remove "cc" attribute from all patterns. Remove cc_status handling
8157 from all patterns. Minor formatting fixes.
8159 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8161 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
8162 (cortex-a76.cortex-a55): Likewise.
8163 * config/aarch64/aarch64-tune.md: Regenerate.
8164 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
8165 cortex-a76.cortex-a55.
8167 2018-06-27 Jeff Law <law@redhat.com>
8169 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
8170 (MULTILIB_DIRNAMES): Similarly.
8172 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
8174 * gimple.h (gimple_return_retbnd): Delete.
8175 (gimple_return_set_retbnd): Likewise.
8176 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
8177 gimple_return_set_retbnd.
8178 * gimple-pretty-print.c (dump_gimple_return): Remove call to
8179 gimple_return_retbnd and adjust.
8180 * tree-inline.h (struct copy_body_data): Remove retbnd field.
8181 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
8182 Explicitly return NULL in a couple more cases. Move assertion
8183 on debug statements and remove unreachable code.
8184 (reset_debug_binding): Do not test id->retbnd.
8185 (expand_call_inline): Do not set it.
8187 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8189 * configure.ac: Add --disable-gcov option.
8190 * configure: Regenerate.
8191 * Makefile.in: Honour @enable_gcov@.
8192 * doc/install.texi: Document --disable-gcov.
8194 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8196 * config/arm/arm-cpus.in (cortex-a76): New entry.
8197 (cortex-a76.cortex-a55): Likewise.
8198 * config/arm/arm-tables.opt: Regenerate.
8199 * config/arm/arm-tune.md: Likewise.
8200 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
8201 * doc/invoke.texi (ARM Options): Document cortex-a76 and
8202 cortex-a76.cortex-a55.
8204 2018-06-27 Tamar Christina <tamar.christina@arm.com>
8207 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
8209 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
8211 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
8213 (EPILOGUE_USES): Likewise.
8215 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
8217 * tree-inline.c (remap_location): New function extracted from...
8218 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
8219 (copy_phis_for_bb): ...here. Call remap_location.
8220 (copy_cfg_body): Adjust call to copy_edges_for_bb.
8222 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
8224 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
8225 unaligned vsx for 16B memset.
8227 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
8230 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
8231 ieee128_float_type_node to long_double_type_node unless
8232 TARGET_LONG_DOUBLE_128 is set.
8234 2018-06-26 David Malcolm <dmalcolm@redhat.com>
8236 * cfgloop.c (get_loop_location): Convert return type from
8237 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
8238 by implicit construction from rtx_insn *, and using
8239 dump_user_location_t::from_function_decl for the fallback case.
8240 * cfgloop.h (get_loop_location): Convert return type from
8241 location_t to dump_user_location_t.
8242 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
8243 dump_printf_loc to pass in a dump_location_t rather than a
8244 location_t, via the gimple stmt.
8245 * coverage.c (get_coverage_counts): Update calls to
8246 dump_printf_loc to pass in dump_location_t rather than a
8248 * doc/optinfo.texi (Dump types): Convert example of
8249 dump_printf_loc from taking "locus" to taking "insn". Update
8250 description of the "_loc" calls to cover dump_location_t.
8251 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
8253 (dump_user_location_t::dump_user_location_t): New constructors,
8254 from gimple *stmt and rtx_insn *.
8255 (dump_user_location_t::from_function_decl): New function.
8256 (dump_loc): Make static.
8257 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
8258 const dump_location_t &.
8259 (dump_generic_expr_loc): Delete.
8260 (dump_printf_loc): Convert param "loc" from location_t to
8261 const dump_location_t &.
8262 (selftest::test_impl_location): New function.
8263 (selftest::dumpfile_c_tests): New function.
8264 * dumpfile.h: Include "profile-count.h".
8265 (class dump_user_location_t): New class.
8266 (struct dump_impl_location_t): New struct.
8267 (class dump_location_t): New class.
8268 (dump_printf_loc): Convert 2nd param from source_location to
8269 const dump_location_t &.
8270 (dump_generic_expr_loc): Delete.
8271 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
8272 const dump_location_t &.
8273 * gimple-fold.c (fold_gimple_assign): Update call to
8274 dump_printf_loc to pass in a dump_location_t rather than a
8275 location_t, via the gimple stmt.
8276 (gimple_fold_call): Likewise.
8277 * gimple-loop-interchange.cc
8278 (loop_cand::analyze_iloop_reduction_var): Update for change to
8279 check_reduction_path.
8280 (tree_loop_interchange::interchange): Update for change to
8282 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
8283 change in return-type of find_loop_location.
8284 (graphite_regenerate_ast_isl): Likewise.
8285 * graphite-optimize-isl.c (optimize_isl): Likewise.
8286 * graphite.c (graphite_transform_loops): Likewise.
8287 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
8288 pass in a dump_location_t rather than a location_t, via the
8290 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8291 * ipa.c (walk_polymorphic_call_targets): Likewise.
8292 * loop-unroll.c (report_unroll): Convert "locus" param from
8293 location_t to dump_location_t.
8294 (decide_unrolling): Update for change to get_loop_location's
8296 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
8297 location_t to dump_user_location_t.
8298 (grid_find_single_omp_among_assignments_1): Updates calls to
8299 dump_printf_loc to pass in a dump_location_t rather than a
8300 location_t, via the gimple stmt.
8301 (grid_parallel_clauses_gridifiable): Convert "tloc" from
8302 location_t to dump_location_t. Updates calls to dump_printf_loc
8303 to pass in a dump_location_t rather than a location_t, via the
8305 (grid_inner_loop_gridifiable_p): Likewise.
8306 (grid_dist_follows_simple_pattern): Likewise.
8307 (grid_gfor_follows_tiling_pattern): Likewise.
8308 (grid_target_follows_gridifiable_pattern): Likewise.
8309 (grid_attempt_target_gridification): Convert initialization
8310 of local "grid" from memset to zero-initialization; FIXME: does
8311 this require C++11? Update call to dump_printf_loc to pass in a
8312 optinfo_location rather than a location_t, via the gimple stmt.
8313 * profile.c (read_profile_edge_counts): Updates call to
8314 dump_printf_loc to pass in a dump_location_t rather than a
8316 (compute_branch_probabilities): Likewise.
8317 * selftest-run-tests.c (selftest::run_tests): Call
8319 * selftest.h (dumpfile_c_tests): New decl.
8320 * tree-loop-distribution.c (pass_loop_distribution::execute):
8321 Update for change in return type of find_loop_location.
8322 * tree-parloops.c (parallelize_loops): Likewise.
8323 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
8324 "locus" from location_t to dump_user_location_t.
8325 (canonicalize_loop_induction_variables): Likewise.
8326 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
8327 for change in return type of find_loop_location.
8328 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
8329 to dump_printf_loc to pass in a dump_location_t rather than a
8330 location_t, via the stmt.
8331 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
8333 * tree-vect-loop-manip.c (find_loop_location): Convert return
8334 type from source_location to dump_user_location_t.
8335 (vect_do_peeling): Update for above change.
8336 (vect_loop_versioning): Update for change in type of
8338 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
8339 from location_t to dump_user_location_t.
8340 (vect_estimate_min_profitable_iters): Update for change in type
8342 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
8343 location_t to dump_location_t.
8344 (vect_slp_bb): Update for change in type of vect_location.
8345 * tree-vectorizer.c (vect_location): Convert from source_location
8346 to dump_user_location_t.
8347 (try_vectorize_loop_1): Update for change in vect_location's type.
8348 (vectorize_loops): Likewise.
8349 (increase_alignment): Likewise.
8350 * tree-vectorizer.h (vect_location): Convert from source_location
8351 to dump_user_location_t.
8352 (find_loop_location): Convert return type from source_location to
8353 dump_user_location_t.
8354 (check_reduction_path): Convert 1st param from location_t to
8355 dump_user_location_t.
8356 * value-prof.c (check_counter): Update call to dump_printf_loc to
8357 pass in a dump_user_location_t rather than a location_t; update
8358 call to error_at for change in type of "locus".
8359 (check_ic_target): Update call to dump_printf_loc to
8360 pass in a dump_user_location_t rather than a location_t, via the
8363 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
8365 * config/s390/s390.h (enum processor_flags): Do not use
8366 default tune parameter when -march was specified.
8368 2018-06-26 Jakub Jelinek <jakub@redhat.com>
8371 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
8372 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
8375 2018-06-26 Richard Biener <rguenther@suse.de>
8377 PR tree-optimization/86287
8379 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
8380 (vect_analyze_loop): Initialize n_stmts.
8382 2018-06-26 Richard Biener <rguenther@suse.de>
8385 * fold-const.c (fold_convertible_p): Pointer extension
8388 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
8391 * dwarf2out.c (loc_list_has_views): Adjust comments.
8392 (dw_loc_list): Split single cross-partition range with
8395 2018-06-25 Jeff Law <law@redhat.com>
8397 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
8398 on -mbig-switch by default.
8400 * config/v850/predicates.md (const_float_1_operand): Fix match_code
8402 (const_float_0_operand): Remove unused predicate.
8403 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
8404 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
8405 (recipsf2): New expander. Original pattern now called
8407 (recipdf2, recipdf2_insn): Similarly.
8408 (rsqrtsf2, rsqrtsf2_insn): Similarly
8409 (rsqrtdf2, rsqrtdf2_insn): Similarly
8411 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
8413 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
8414 Simplify logic for FreeBSD (twice).
8416 2018-06-25 Martin Sebor <msebor@redhat.com>
8418 PR tree-optimization/86204
8419 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
8420 a strnlen result if it's less than the length of the string.
8422 2018-06-25 Martin Sebor <msebor@redhat.com>
8424 PR tree-optimization/85700
8425 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
8426 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
8427 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
8429 2018-06-25 Martin Sebor <msebor@redhat.com>
8431 * doc/extend.texi (Zero-length arrays): Update and clarify.
8433 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
8435 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
8436 added IEEE/IBM long double multilib support on PowerPC little
8437 endian Linux systems.
8438 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
8439 (MULTILIB_DEFAULTS): Likewise.
8440 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8442 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
8443 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
8444 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
8446 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
8449 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
8450 (REORDER_45): Likewise.
8452 2018-06-25 Jeff Law <law@redhat.com>
8454 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
8455 dividend to 32 bits. Adjust length.
8456 (udivmodhi4): Cleanup output template. Fix length.
8458 2018-06-25 Carl Love <cel@us.ibm.com>
8460 * config/rs6000/vsx.md: Change word selector to prefered location.
8462 2018-06-25 Richard Biener <rguenther@suse.de>
8464 PR tree-optimization/86304
8465 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
8466 epilogue-if-converted loops as well.
8468 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
8470 * lto-section-out.c (lto_begin_section): Do not print section
8471 name for noaddr and unnumbered dumps.
8473 2018-06-25 Richard Biener <rguenther@suse.de>
8475 * tree-vectorizer.h (struct vec_info_shared): New structure
8476 with parts split out from struct vec_info and loop_nest from
8477 struct _loop_vec_info.
8478 (struct vec_info): Adjust accordingly.
8479 (struct _loop_vec_info): Likewise.
8480 (LOOP_VINFO_LOOP_NEST): Adjust.
8481 (LOOP_VINFO_DATAREFS): Likewise.
8482 (LOOP_VINFO_DDRS): Likewise.
8483 (struct _bb_vec_info): Likewise.
8484 (BB_VINFO_DATAREFS): Likewise.
8485 (BB_VINFO_DDRS): Likewise.
8486 (struct _stmt_vec_info): Add dr_aux member.
8487 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
8488 (DR_MISALIGNMENT_UNINITIALIZED): New.
8489 (set_dr_misalignment): Adjust.
8490 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
8491 (vect_analyze_loop): Adjust prototype.
8492 (vect_analyze_loop_form): Likewise.
8493 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
8494 Compute dependences lazily.
8495 (vect_record_base_alignments): Use shared datarefs/ddrs.
8496 (vect_verify_datarefs_alignment): Likewise.
8497 (vect_analyze_data_refs_alignment): Likewise.
8498 (vect_analyze_data_ref_accesses): Likewise.
8499 (vect_analyze_data_refs): Likewise.
8500 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
8501 constructor parameter for shared part.
8502 (vect_analyze_loop_form): Pass in shared part and adjust.
8503 (vect_analyze_loop_2): Pass in storage for the number of
8504 stmts. Move loop nest finding to the caller. Compute
8506 (vect_analyze_loop): Pass in shared part.
8507 (vect_transform_loop): Verify shared datarefs are unchanged.
8508 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
8509 constructor parameter for shared part.
8510 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
8511 (vect_slp_bb): Verify shared datarefs are unchanged before
8513 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
8515 (new_stmt_vec_info): Initialize DR_AUX misalignment to
8516 DR_MISALIGNMENT_UNINITIALIZED.
8517 * tree-vectorizer.c (vec_info::vec_info): Add constructor
8518 parameter for shared part.
8519 (vec_info::~vec_info): Adjust.
8520 (vec_info_shared::vec_info_shared): New.
8521 (vec_info_shared::~vec_info_shared): Likewise.
8522 (vec_info_shared::save_datarefs): Likewise.
8523 (vec_info_shared::check_datarefs): Likewise.
8524 (try_vectorize_loop_1): Construct shared part live for analyses
8525 of a single loop for multiple vector sizes.
8526 * tree-parloops.c (gather_scalar_reductions): Adjust.
8528 2018-06-25 Richard Biener <rguenther@suse.de>
8530 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
8531 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
8532 (vect_analyze_data_refs): Remove similar code from here and
8533 simplify accordingly.
8535 2018-06-25 Richard Biener <rguenther@suse.de>
8537 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
8538 for reverse storage order accesses rather than asserting
8539 they cannot happen here.
8541 2018-06-25 Tom de Vries <tdevries@suse.de>
8544 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
8545 Use data16 instead of .byte for insn prefix.
8547 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
8550 * parser.c (make_char_string_pack): Pass this literal chars
8551 through cpp_interpret_string.
8552 (cp_parser_userdef_numeric_literal): Check the result of
8553 make_char_string_pack.
8555 2018-06-24 Maya Rashish <coypu@sdf.org>
8557 * ginclude/stddef.h: Simplify conditions around avoiding
8558 re-definition of __size_t.
8560 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8562 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
8563 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
8565 2018-06-22 Maya Rashish <coypu@sdf.org>
8567 * doc/invoke.texi (mno-fancy-math-387): Update for changes
8568 made to OpenBSD and NetBSD through the years.
8570 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
8572 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8573 behavior of vec_pack (vector double, vector double) to match
8574 behavior of vec_float2 (vector double, vector double).
8576 2018-06-22 Olivier Hainque <hainque@adacore.com>
8578 * gimplify.c (gimplify_function_tree): Prevent creation
8579 of a trampoline for the address of the current function
8580 passed to entry/exit instrumentation hooks.
8582 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
8585 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
8588 2018-06-22 Martin Liska <mliska@suse.cz>
8590 PR tree-optimization/86263
8591 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8592 Bail out if is_enabled is false.
8593 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
8595 (jump_table_cluster::is_enabled): New function.
8597 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8599 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
8600 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
8601 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
8602 (lto_input_ts_binfo_tree_pointers): Likewise.
8603 * tree-streamer-out.c (streamer_write_tree_bitfields,
8604 write_ts_binfo_tree_pointers): Likewise.
8605 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
8607 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8609 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
8611 2018-06-22 Martin Liska <mliska@suse.cz>
8613 * symbol-summary.h (get): Make it pure and inline move
8614 functionality from ::get function.
8615 (get): Remove and inline into ::get and ::get_create.
8616 (get_create): Move code from ::get function.
8618 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8621 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
8622 -x assembler-with-cpp.
8624 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8626 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
8627 _FILE_OFFSET_BITS=64 for C++.
8629 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
8631 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
8632 conversion insn that shows up when pr85657-3.c is compiled using
8633 IEEE 128-bit long double.
8634 (neg<mode>2_internal): Use the correct mode to check whether the
8635 mode is IBM extended.
8636 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
8637 multiply and divide external functions from being created more
8640 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8642 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
8644 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
8645 the edge can be forwarded.
8646 (cfg_layout_merge_blocks): Likewise.
8648 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8650 * except.c (finish_eh_generation): Commit edge insertions only after
8651 the EH edges have been redirected from post-landing to landing pads.
8653 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8655 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
8656 create_tmp_var_for to create the FRAME decl.
8657 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
8659 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8661 * tree-inline.c (copy_edges_for_bb): Minor tweak.
8662 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
8663 debug statement when resetting its value.
8664 (expand_call_inline): Copy the locus of the call onto the assignment
8665 of the return value, if any. Use local variable in more cases.
8667 2018-06-21 Martin Liska <mliska@suse.cz>
8669 * ipa-pure-const.c (propagate_nothrow): Use
8670 funct_state_summaries->get.
8671 (dump_malloc_lattice): Likewise.
8672 (propagate_malloc): Likewise.
8674 2018-06-21 Richard Biener <rguenther@suse.de>
8676 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
8677 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
8678 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
8680 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
8681 BLOCK_ABSTRACT_ORIGIN unconditionally.
8683 2018-06-21 David Malcolm <dmalcolm@redhat.com>
8685 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
8687 * ipa-reference.c (ipa_reference_c_finalize): Delete
8688 ipa_ref_opt_sum_summaries and set it to NULL.
8690 2018-06-21 Tom de Vries <tdevries@suse.de>
8692 PR tree-optimization/85859
8693 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
8694 test with comment from bb_no_side_effects_p.
8696 2018-06-21 Richard Biener <rguenther@suse.de>
8698 PR tree-optimization/86232
8699 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
8700 max for constant niter.
8702 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8704 * config/aarch64/aarch64-simd.md
8705 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
8707 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8709 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
8710 Make opernads of the unspec commutative.
8712 2018-06-21 Richard Biener <rguenther@suse.de>
8714 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
8715 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8716 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
8717 (vect_analyze_data_ref_dependence): Re-order checks to deal with
8719 (vect_record_base_alignments): Do not record base alignment
8720 for gathers or scatters.
8721 (vect_compute_data_ref_alignment): Drop return value that is always
8722 true. Bail out early for gathers or scatters.
8723 (vect_enhance_data_refs_alignment): Bail out early for gathers
8725 (vect_find_same_alignment_drs): Likewise.
8726 (vect_analyze_data_refs_alignment): Remove dead code.
8727 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8728 (vect_analyze_data_refs): For possible gathers or scatters do
8729 not create an alternate DR, just check their possible validity
8730 and mark them. Adjust DECL_NONALIASED handling to not rely
8732 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
8733 update inits of gathers or scatters.
8734 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
8735 Also copy gather/scatter flag to pattern vinfo.
8737 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
8739 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8740 behavior of vec_packsu (vector unsigned long long, vector unsigned
8741 long long) to match behavior of vec_packs with same signature.
8743 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
8744 Thomas Schwinge <thomas@codesourcery.com>
8745 Cesar Philippidis <cesar@codesourcery.com>
8747 * gimplify.c (gimplify_scan_omp_clauses): Add support for
8748 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8749 (gimplify_adjust_omp_clauses): Likewise.
8750 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
8751 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
8752 (gimplify_omp_target_update): Update handling of acc update and
8754 * omp-low.c (install_var_field): Remove unused parameter
8755 base_pointers_restrict.
8756 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
8757 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
8759 (omp_target_base_pointers_restrict_p): Delete.
8760 (scan_omp_target): Update call to scan_sharing_clauses.
8761 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
8763 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
8764 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8765 (convert_local_omp_clauses): Likewise.
8766 * tree-pretty-print.c (dump_omp_clause): Likewise.
8767 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
8769 (omp_clause_code_name): Likewise.
8771 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8774 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
8777 PR tree-optimization/86231
8778 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
8779 anti-range don't overwrite *vr0min before using it to compute *vr0max.
8781 2018-06-20 Tom de Vries <tdevries@suse.de>
8783 PR tree-optimization/86097
8784 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
8785 iv type if signedness of iv type is not the same as that of *nit.
8787 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8789 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
8790 EDGE_EH edges, verify they are all EDGE_EH.
8792 2018-06-20 Maya Rashish <coypu@sdf.org>
8794 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
8796 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8798 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
8799 * config/aarch64/aarch64.c (xgene1_tunings): Add
8800 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
8801 (aarch64_mode_valid_for_sched_fusion_p):
8802 Allow 16-byte modes.
8803 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
8804 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
8806 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
8808 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
8809 * config/aarch64/iterators.md (VQ2): New mode iterator.
8811 2018-06-20 Martin Liska <mliska@suse.cz>
8813 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
8814 Change default ratio from 10 to 8.
8816 2018-06-20 Martin Liska <mliska@suse.cz>
8818 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8820 (bit_test_cluster::find_bit_tests): Likewise.
8821 (switch_decision_tree::analyze_switch_statement): Find clusters.
8822 * tree-switch-conversion.h (struct jump_table_cluster): Document
8825 2018-06-20 Martin Liska <mliska@suse.cz>
8827 * tree-switch-conversion.c (switch_conversion::collect):
8828 Record m_uniq property.
8829 (switch_conversion::expand): Bail out for special conditions.
8830 (group_cluster::~group_cluster): New.
8831 (group_cluster::group_cluster): Likewise.
8832 (group_cluster::dump): Likewise.
8833 (jump_table_cluster::emit): New.
8834 (switch_decision_tree::fix_phi_operands_for_edges): New.
8835 (struct case_node): Remove struct.
8836 (jump_table_cluster::can_be_handled): New.
8837 (case_values_threshold): Moved to header.
8838 (reset_out_edges_aux): Likewise.
8839 (jump_table_cluster::is_beneficial): New.
8840 (bit_test_cluster::can_be_handled): Likewise.
8841 (add_case_node): Remove.
8842 (bit_test_cluster::is_beneficial): New.
8843 (case_bit_test::cmp): New.
8844 (bit_test_cluster::emit): New.
8845 (expand_switch_as_decision_tree_p): Remove.
8846 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
8847 (fix_phi_operands_for_edge): Likewise.
8848 (switch_decision_tree::analyze_switch_statement): New.
8849 (compute_cases_per_edge): Move ...
8850 (switch_decision_tree::compute_cases_per_edge): ... here.
8851 (try_switch_expansion): Likewise.
8852 (switch_decision_tree::try_switch_expansion): Likewise.
8853 (record_phi_operand_mapping): Likewise.
8854 (switch_decision_tree::record_phi_operand_mapping): Likewise.
8855 (emit_case_decision_tree): Likewise.
8856 (switch_decision_tree::emit): Likewise.
8857 (balance_case_nodes): Likewise.
8858 (switch_decision_tree::balance_case_nodes): Likewise.
8859 (dump_case_nodes): Likewise.
8860 (switch_decision_tree::dump_case_nodes): Likewise.
8861 (emit_jump): Likewise.
8862 (switch_decision_tree::emit_jump): Likewise.
8863 (emit_cmp_and_jump_insns): Likewise.
8864 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
8865 (emit_case_nodes): Likewise.
8866 (switch_decision_tree::emit_case_nodes): Likewise.
8867 (conditional_probability): Remove.
8868 * tree-switch-conversion.h (enum cluster_type): New.
8870 (struct cluster): Likewise.
8871 (cluster::cluster): Likewise.
8872 (struct simple_cluster): Likewise.
8873 (simple_cluster::simple_cluster): Likewise.
8874 (struct group_cluster): Likewise.
8875 (struct jump_table_cluster): Likewise.
8876 (struct bit_test_cluster): Likewise.
8877 (struct min_cluster_item): Likewise.
8878 (struct case_tree_node): Likewise.
8879 (case_tree_node::case_tree_node): Likewise.
8880 (jump_table_cluster::case_values_threshold): Likewise.
8881 (struct case_bit_test): Likewise.
8882 (struct switch_decision_tree): Likewise.
8883 (struct switch_conversion): Likewise.
8884 (switch_decision_tree::reset_out_edges_aux): Likewise.
8886 2018-06-20 Martin Liska <mliska@suse.cz>
8888 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
8889 (hoist_edge_and_branch_if_true): Likewise.
8890 (expand_switch_using_bit_tests_p): Likewise.
8891 (struct case_bit_test): Likewise.
8892 (case_bit_test_cmp): Likewise.
8893 (emit_case_bit_tests): Likewise.
8894 (switch_conversion::switch_conversion): New class.
8895 (struct switch_conv_info): Remove old struct.
8896 (collect_switch_conv_info): More to ...
8897 (switch_conversion::collect): ... this.
8898 (check_range): Likewise.
8899 (switch_conversion::check_range): Likewise.
8900 (check_all_empty_except_final): Likewise.
8901 (switch_conversion::check_all_empty_except_final): Likewise.
8902 (check_final_bb): Likewise.
8903 (switch_conversion::check_final_bb): Likewise.
8904 (create_temp_arrays): Likewise.
8905 (switch_conversion::create_temp_arrays): Likewise.
8906 (free_temp_arrays): Likewise.
8907 (gather_default_values): Likewise.
8908 (switch_conversion::gather_default_values): Likewise.
8909 (build_constructors): Likewise.
8910 (switch_conversion::build_constructors): Likewise.
8911 (constructor_contains_same_values_p): Likewise.
8912 (switch_conversion::contains_same_values_p): Likewise.
8913 (array_value_type): Likewise.
8914 (switch_conversion::array_value_type): Likewise.
8915 (build_one_array): Likewise.
8916 (switch_conversion::build_one_array): Likewise.
8917 (build_arrays): Likewise.
8918 (switch_conversion::build_arrays): Likewise.
8919 (gen_def_assigns): Likewise.
8920 (switch_conversion::gen_def_assigns): Likewise.
8921 (prune_bbs): Likewise.
8922 (switch_conversion::prune_bbs): Likewise.
8923 (fix_phi_nodes): Likewise.
8924 (switch_conversion::fix_phi_nodes): Likewise.
8925 (gen_inbound_check): Likewise.
8926 (switch_conversion::gen_inbound_check): Likewise.
8927 (process_switch): Use the newly created class.
8928 (switch_conversion::expand): New.
8929 (switch_conversion::~switch_conversion): New.
8930 * tree-switch-conversion.h: New file.
8932 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8934 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
8935 tree-vect-patterns.c.
8936 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
8937 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
8938 (vect_recog_sad_pattern): Likewise.
8939 (vect_recog_widen_sum_pattern): Likewise.
8940 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
8941 (vect_recog_widen_shift_pattern): Remove the type_in argument.
8942 (vect_recog_rotate_pattern): Likewise.
8943 (vect_recog_mult_pattern): Likewise.
8944 (vect_recog_vector_vector_shift_pattern): Likewise.
8945 (vect_recog_divmod_pattern): Likewise.
8946 (vect_recog_mixed_size_cond_pattern): Likewise.
8947 (vect_recog_bool_pattern): Likewise.
8948 (vect_recog_mask_conversion_pattern): Likewise.
8949 (vect_try_gather_scatter_pattern): Likewise.
8950 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
8951 (vect_recog_over_widening_pattern): Likewise.
8952 (vect_recog_gather_scatter_pattern): Likewise.
8953 (vect_recog_func_ptr): Move from tree-vectorizer.h
8954 (vect_vect_recog_func_ptrs): Move further down the file.
8955 (vect_recog_func): Likewise. Remove the third argument.
8956 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
8957 (vect_pattern_recog_1): Expect the pattern function to do any
8958 necessary target tests. Also expect it to provide a vector type.
8959 Remove the type_in handling.
8961 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8963 * tree-vect-patterns.c (vect_pattern_detected): New function.
8964 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
8965 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
8966 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
8967 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
8968 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
8969 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
8970 (vect_recog_mask_conversion_pattern)
8971 (vect_try_gather_scatter_pattern): Likewise.
8973 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8975 * tree-vect-patterns.c (vect_get_internal_def): New function.
8976 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
8977 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
8978 (search_type_for_mask_1): Use it.
8980 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8982 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8983 redundant WIDEN_SUM_EXPR handling.
8984 (vect_recog_sad_pattern): Likewise.
8986 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8988 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8989 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
8990 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
8991 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
8992 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
8994 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8996 * tree-vect-stmts.c (vectorizable_call): Make sure that we
8997 use the stmt_vec_info of the original bb statement for the
8998 new zero assignment, even if the call is part of a pattern.
9000 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9002 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
9003 that the sequence is attached to the original statement rather
9004 than the pattern statement.
9005 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
9006 PATTERN_DEF_SEQ from the original statement rather than
9007 the main pattern statement.
9008 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
9009 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9010 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
9012 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9014 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
9015 definition statements before the early exit for statements that aren't
9017 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
9019 (vect_transform_loop): ...here. Process pattern definition
9020 statements without first checking whether the main pattern
9021 statement is live or relevant.
9023 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
9025 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
9026 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
9028 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
9030 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
9031 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
9032 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
9033 (expand_block_compare): Change select_block_compare_mode call.
9034 (expand_strncmp_align_check): Use new functions, fix comment.
9035 (emit_final_str_compare_gpr): New function.
9036 (expand_strn_compare): Refactor and clean up code.
9037 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
9039 2018-06-19 Tony Reix <tony.reix@atos.com>
9040 Damien Bergamini <damien.bergamini@atos.com>
9041 David Edelsohn <dje.gcc@gmail.com>
9043 * collect2.c (static_obj): New variable.
9044 (static_libs): New variable.
9045 (is_in_list): Uncomment declaration.
9046 (main): Track AIX libraries linked statically.
9047 (is_in_list): Uncomment definition.
9048 (scan_prog_file): Don't add AIX shared libraries initializer
9049 to constructor list if linking statically.
9051 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
9053 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
9055 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
9057 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
9059 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
9062 2018-06-19 Martin Liska <mliska@suse.cz>
9064 * config/i386/i386.c (ix86_can_inline_p): Do not use
9065 ipa_fn_summaries::get_create.
9066 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
9068 (devirtualization_time_bonus): Likewise.
9069 (ipcp_propagate_stage): Likewise.
9070 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
9071 (edge_set_predicate): Likewise.
9072 (evaluate_conditions_for_known_args): Likewise.
9073 (evaluate_properties_for_edge): Likewise.
9074 (ipa_call_summary::reset): Tranform to ...
9075 (ipa_call_summary::~ipa_call_summary): ... this.
9076 (ipa_fn_summary::reset): Transform to ...
9077 (ipa_fn_summary::~ipa_fn_summary): ... this.
9078 (ipa_fn_summary_t::remove): Rename to ...
9079 (ipa_fn_summary_t::remove_callees): ... this.
9080 (ipa_fn_summary_t::duplicate): Use placement new
9081 instead of memory copy.
9082 (ipa_call_summary_t::duplicate): Likewise.
9083 (ipa_call_summary_t::remove): Remove.
9084 (dump_ipa_call_summary): Change get_create to get.
9085 (ipa_dump_fn_summary): Dump only when summary exists.
9086 (analyze_function_body): Use symbol_summary::get instead
9088 (compute_fn_summary): Likewise.
9089 (estimate_edge_devirt_benefit): Likewise.
9090 (estimate_edge_size_and_time): Likewise.
9091 (inline_update_callee_summaries): Likewise.
9092 (remap_edge_change_prob): Likewise.
9093 (remap_edge_summaries): Likewise.
9094 (ipa_merge_fn_summary_after_inlining): Likewise.
9095 (write_ipa_call_summary): Likewise.
9096 (ipa_fn_summary_write): Likewise.
9097 (ipa_free_fn_summary): Likewise.
9098 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
9099 (struct ipa_call_summary): Likewise.
9100 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
9102 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
9103 (estimate_size_after_inlining): Likewise.
9104 (estimate_growth): Likewise.
9105 (growth_likely_positive): Likewise.
9106 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
9107 (inline_call): Likewise.
9108 * ipa-inline.c (caller_growth_limits): Likewise.
9109 (can_inline_edge_p): Likewise.
9110 (can_inline_edge_by_limits_p): Likewise.
9111 (compute_uninlined_call_time): Likewise.
9112 (compute_inlined_call_time): Likewise.
9113 (want_inline_small_function_p): Likewise.
9114 (edge_badness): Likewise.
9115 (update_caller_keys): Likewise.
9116 (update_callee_keys): Likewise.
9117 (inline_small_functions): Likewise.
9118 (inline_to_all_callers_1): Likewise.
9119 (dump_overall_stats): Likewise.
9120 (early_inline_small_functions): Likewise.
9121 (early_inliner): Likewise.
9122 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9123 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
9124 * ipa-pure-const.c (malloc_candidate_p): Likewise.
9125 * ipa-split.c (execute_split_functions): Likewise.
9126 * symbol-summary.h: Likewise.
9127 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
9129 2018-06-19 Richard Biener <rguenther@suse.de>
9131 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
9132 (vectorize_loops): ... here. Fix dbgcnt handling.
9133 (try_vectorize_loop): Wrap try_vectorize_loop_1.
9135 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
9138 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
9139 ieee128 argument takes up only one (vector) register, not two (floating
9142 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
9144 * gimplify.c (gimplify_init_constructor): Really never clear for an
9145 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
9147 2018-06-19 Richard Biener <rguenther@suse.de>
9149 PR tree-optimization/86179
9150 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
9151 after failed recognition.
9153 2018-06-18 Martin Sebor <msebor@redhat.com>
9156 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
9157 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
9158 Handle integer subtraction.
9159 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
9160 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
9162 2018-06-18 David Malcolm <dmalcolm@redhat.com>
9164 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
9165 param from rtx to rtx_insn *.
9166 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
9168 (frv_ifcvt_modify_insn): Likwise.
9169 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
9170 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
9171 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
9172 as_a <rtx_insn *> cast to local "unprotected_region" once
9173 it's been established that it's not NULL or pc_rtx.
9174 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
9175 param "sethi" from rtx to rtx_insn *.
9176 (nds32_group_float_insns): Likewise for param "insn".
9177 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
9179 (vax_output_int_subtract): Likewise.
9180 * config/vax/vax.c (vax_output_int_add): Likewise for param
9182 (vax_output_int_subtract): Likewise.
9183 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
9184 (emit_pattern_after): Likewise for param "after".
9185 (emit_insn_after): Likewise.
9186 (emit_jump_insn_after): Likewise.
9187 (emit_call_insn_after): Likewise.
9188 (emit_debug_insn_after): Likewise.
9189 (emit_pattern_before): Likewise for param "before".
9190 (emit_insn_before): Likewise.
9191 (emit_jump_insn_before): Likewise.
9192 * final.c (get_insn_template): Likewise for param "insn", removing
9194 * output.h (get_insn_template): Likewise for 2nd param.
9195 * rtl.h (emit_insn_before): Likewise.
9196 (emit_jump_insn_before): Likewise.
9197 (emit_debug_insn_before_noloc): Likewise.
9198 (emit_insn_after): Likewise.
9199 (emit_jump_insn_after): Likewise.
9200 (emit_call_insn_after): Likewise.
9201 (emit_debug_insn_after): Likewise.
9202 (set_insn_deleted): Likewise for param.
9204 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
9207 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
9208 floating point modes, so that IFmode is numerically greater than
9209 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
9210 to declare the ordering. This prevents IFmode from being
9211 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
9212 machine. Include rs6000-modes.h to share the fractional values
9213 between genmodes* and the rest of the compiler.
9217 * config/rs6000/rs6000-modes.h: New file.
9218 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
9219 meaning of rs6000_long_double_size so that 126..128 selects an
9220 appropriate 128-bit floating point type.
9221 (rs6000_option_override_internal): Likewise.
9222 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
9223 (TARGET_LONG_DOUBLE_128): Change the meaning of
9224 rs6000_long_double_size so that 126..128 selects an appropriate
9225 128-bit floating point type.
9226 (LONG_DOUBLE_TYPE_SIZE): Update comment.
9227 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
9228 source and destination to match the standard usage.
9229 (truncifkf2): Likewise.
9230 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
9231 ISA 2.07 to use an explicit clobber, instead of passing in a
9233 (copysign<mode>3_soft): Likewise.
9235 2018-06-18 David Malcolm <dmalcolm@redhat.com>
9237 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
9238 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
9239 (vect_slp_analyze_instance_dependence): Likewise.
9240 (vect_enhance_data_refs_alignment): Likewise.
9241 (vect_analyze_data_refs_alignment): Likewise.
9242 (vect_slp_analyze_and_verify_instance_alignment
9243 (vect_analyze_data_ref_accesses): Likewise.
9244 (vect_prune_runtime_alias_test_list): Likewise.
9245 (vect_analyze_data_refs): Likewise.
9246 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
9247 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
9248 (vect_analyze_scalar_cycles_1): Likewise.
9249 (vect_get_loop_niters): Likewise.
9250 (vect_analyze_loop_form_1): Likewise.
9251 (vect_update_vf_for_slp): Likewise.
9252 (vect_analyze_loop_operations): Likewise.
9253 (vect_analyze_loop): Likewise.
9254 (vectorizable_induction): Likewise.
9255 (vect_transform_loop): Likewise.
9256 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
9257 * tree-vect-slp.c (vect_analyze_slp): Likewise.
9258 (vect_make_slp_decision): Likewise.
9259 (vect_detect_hybrid_slp): Likewise.
9260 (vect_slp_analyze_operations): Likewise.
9261 (vect_slp_bb): Likewise.
9262 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
9263 (vectorizable_bswap): Likewise.
9264 (vectorizable_call): Likewise.
9265 (vectorizable_simd_clone_call): Likewise.
9266 (vectorizable_conversion): Likewise.
9267 (vectorizable_assignment): Likewise.
9268 (vectorizable_shift): Likewise.
9269 (vectorizable_operation): Likewise.
9270 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
9272 2018-06-18 Martin Sebor <msebor@redhat.com>
9274 PR tree-optimization/81384
9275 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
9276 * builtins.c (expand_builtin_strnlen): New function.
9277 (expand_builtin): Call it.
9278 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
9279 * builtins.def (BUILT_IN_STRNLEN): New.
9280 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
9281 Warn for bounds in excess of maximum object size.
9282 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
9283 single-value ranges. Handle strnlen.
9284 (handle_builtin_strlen): Handle strnlen.
9285 (strlen_check_and_optimize_stmt): Same.
9286 * doc/extend.texi (Other Builtins): Document strnlen.
9288 2018-06-18 Maya Rashish <coypu@sdf.org>
9290 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
9291 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
9292 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
9294 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
9296 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
9298 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9300 * tree.c (escaped_string::escape): Replace cast to char * by
9301 const_cast<char *> (unescaped).
9303 2018-06-18 Nick Clifton <nickc@redhat.com>
9306 * tree.c (escaped_string): New class. Converts an unescaped
9307 string into its escaped equivalent.
9308 (warn_deprecated_use): Use the new class to convert the
9309 deprecation message, if present.
9310 (test_escaped_strings): New self test.
9311 (test_c_tests): Add test_escaped_strings.
9312 * doc/extend.texi (deprecated): Add a note that the
9313 deprecation message is affected by the -fmessage-length
9314 option, and that control characters will be escaped.
9315 (#pragma GCC error): Document this pragma.
9316 (#pragma GCC warning): Likewise.
9317 * doc/invoke.texi (-fmessage-length): Document this option's
9318 effect on the #warning and #error preprocessor directives and
9319 the deprecated attribute.
9321 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
9323 * tree.c (decl_value_expr_lookup): Revert latest change.
9324 (decl_value_expr_insert): Likewise.
9326 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
9328 * gimplify.c (nonlocal_vlas): Delete.
9329 (nonlocal_vla_vars): Likewise.
9330 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
9332 (gimplify_body): Do not create and destroy nonlocal_vlas.
9333 * tree-nested.c: Include diagnostic.h.
9334 (use_pointer_in_frame): Tweak.
9335 (lookup_field_for_decl): Add assertion and declare the transformation.
9336 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
9337 internal error when the reference is in a wrong context. Do not
9338 create a debug decl by default.
9339 (note_nonlocal_block_vlas): Delete.
9340 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
9341 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
9342 create a debug decl by default.
9343 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
9344 call is in a wrong context.
9345 (fixup_vla_decls): New function.
9346 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
9347 debug variables were created.
9348 * tree.c (decl_value_expr_lookup): Add checking assertion.
9349 (decl_value_expr_insert): Likewise.
9351 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
9354 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
9355 * tree-scalar-evolution.c (interpret_expr): Likewise.
9356 (expression_expensive_p): Likewise.
9357 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
9358 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
9359 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
9360 (ssa_defined_by_minus_one_stmt_p): New.
9362 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
9365 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
9366 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
9367 * dojump.c (do_jump): Likewise.
9368 * expr.c (expand_expr_real_2): Check operand type's sign.
9369 * fold-const.c (const_unop): Handle ABSU_EXPR.
9370 (fold_abs_const): Likewise.
9371 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
9372 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
9373 (strip_sign_op_1): Likesise.
9374 * match.pd: Add new pattern to generate ABSU_EXPR.
9375 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
9376 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
9377 * tree-eh.c (operation_could_trap_helper_p): Likewise.
9378 * tree-inline.c (estimate_operator_cost): Likewise.
9379 * tree-pretty-print.c (dump_generic_node): Likewise.
9380 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
9381 * tree.def (ABSU_EXPR): New.
9383 2018-06-16 Jakub Jelinek <jakub@redhat.com>
9386 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
9387 documented as preserved for backward compatibility only.
9388 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
9390 PR rtl-optimization/86108
9391 * bb-reorder.c (create_forwarder_block): Renamed to ...
9392 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
9393 jump from new landing pad to the second part.
9394 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
9397 2018-06-15 Jakub Jelinek <jakub@redhat.com>
9400 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
9401 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
9402 Only call store_expr for halves if the mode is the same.
9405 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
9406 Fix up comment formatting.
9408 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9410 * typed-splay-tree.h (typed_splay_tree::remove): New function.
9411 (typed_splay_tree::closure,
9412 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
9413 (typed_splay_tree::typed_splay_tree,
9414 typed_splay_tree::operator =): Declared private.
9415 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
9416 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
9417 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
9418 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
9419 typed_splay_tree::splay_tree_splay,
9420 typed_splay_tree::splay_tree_foreach_helper,
9421 typed_splay_tree::splay_tree_insert,
9422 typed_splay_tree::splay_tree_remove,
9423 typed_splay_tree::splay_tree_lookup,
9424 typed_splay_tree::splay_tree_predecessor,
9425 typed_splay_tree::splay_tree_successor,
9426 typed_splay_tree::splay_tree_min,
9427 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
9428 (typed_splay_tree::root, typed_splay_tree::comp,
9429 typed_splay_tree::delete_key,
9430 typed_splay_tree::delete_value): New data members.
9431 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
9432 typed_splay_tree::remove.
9434 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
9436 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
9437 -mginv and -mno-ginv to the assembler.
9438 * config/mips/mips.opt (-mcrc): New option.
9440 * doc/invoke.text (-mcrc): Document.
9443 2018-06-15 Nick Clifton <nickc@redhat.com>
9446 * tree.c (escaped_string): New class. Converts an unescaped
9447 string into its escaped equivalent.
9448 (warn_deprecated_use): Use the new class to convert the
9449 deprecation message, if present.
9450 (test_escaped_strings): New self test.
9451 (test_c_tests): Add test_escaped_strings.
9452 * doc/extend.texi (deprecated): Add a note that the
9453 deprecation message is affected by the -fmessage-length
9454 option, and that control characters will be escaped.
9455 (#pragma GCC error): Document this pragma.
9456 (#pragma GCC warning): Likewise.
9457 * doc/invoke.texi (-fmessage-length): Document this option's
9458 effect on the #warning and #error preprocessor directives and
9459 the deprecated attribute.
9461 2018-06-15 Richard Biener <rguenther@suse.de>
9463 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
9464 here, also noting vector size used.
9465 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
9466 size used in MSG_OPTIMIZED_LOCATIONS dump.
9467 (pass_slp_vectorize::execute): Adjust.
9469 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
9472 * config/arc/arc.c (arc_return_address_register): Fix
9475 2018-06-15 Richard Biener <rguenther@suse.de>
9478 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
9479 leave useless conversion stripping to force_gimple_operand_gsi.
9480 (gimplify_build2): Likewise.
9481 (gimplify_build1): Likewise.
9483 2018-06-15 Richard Biener <rguenther@suse.de>
9486 * tree-cfg.c (move_stmt_op): unshare invariant addresses
9487 before adjusting their block.
9489 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
9491 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
9492 multilibs for *-*-rtems*.
9493 * config/riscv/t-rtems: New file.
9495 2018-06-14 Jakub Jelinek <jakub@redhat.com>
9498 * match.pd ((A +- CST1) +- CST2): Punt if last resort
9499 unsigned_type_for returns NULL.
9502 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
9503 subregs of multi-word pseudos unless the float mode has word size.
9505 2018-06-14 Richard Biener <rguenther@suse.de>
9508 * tree-vect-generic.c (build_word_mode_vector_type): Remove
9509 duplicate and harmful type_hash_canon.
9510 * tree.c (type_hash_canon): Assert we didn't find ourselves.
9512 2018-06-14 Richard Biener <rguenther@suse.de>
9515 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
9518 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
9520 * config/rtems.h (STDINT_LONG32): Define.
9522 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
9523 Prachi Godbole <prachi.godbole@imgtec.com>
9525 * config/mips/mips-cpus.def: Define P6600.
9526 * config/mips/mips-tables.opt: Regenerate.
9527 * config/mips/mips.c (mips_ucbranch_type): New enum.
9528 (mips_rtx_cost_data): Add support for P6600.
9529 (mips_issue_rate): Likewise.
9530 (mips_multipass_dfa_lookahead): Likewise.
9531 (mips_avoid_hazard): Likewise.
9532 (mips_reorg_process_insns): Likewise.
9533 (mips_classify_branch_p6600): New function.
9534 * config/mips/mips.h (TUNE_P6600): New define.
9535 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
9536 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
9537 * config/mips/mips.md: Include p6600.md.
9538 (processor): Add p6600.
9539 * config/mips/p6600.md: New file.
9540 * doc/invoke.texi: Add p6600 to supported architectures.
9542 2018-06-13 Martin Sebor <msebor@redhat.com>
9544 PR tree-optimization/86114
9545 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
9547 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
9549 2018-06-13 Richard Biener <rguenther@suse.de>
9551 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
9552 Properly set vector type of the intermediate stmt.
9553 * tree-vect-stmts.c (vectorizable_operation): The destination
9554 var always has vectype_out type.
9556 2018-06-13 Jeff Law <law@redhat.com>
9558 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
9559 integer 0 for argument to print_rtl_with_bb.
9560 (rl78_reorg): Likewise.
9562 2018-06-13 David Malcolm <dmalcolm@redhat.com>
9564 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
9565 from rtx to rtx_insn *.
9566 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
9568 (add_sched_insns_for_speculation): Likewise for local "target",
9569 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
9570 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
9571 from rtx_insn ** to rtx_code_label **.
9572 (reorg_emit_nops): Likewise.
9573 (c6x_reorg): Likewise for local "call_labels".
9574 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
9576 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
9577 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
9578 the loops over LABEL_REFS.
9579 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
9581 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
9582 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
9583 (split_branches): Strengthen local "olabel" from rtx to
9584 rtx_insn *, adding a safe_as_a cast.
9585 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
9587 (add_insn_after): Likewise for first two params.
9588 (add_insn_before): Likewise.
9589 (remove_insn): Likewise for param.
9590 (emit_pattern_before_noloc): Likewise for second and third params.
9591 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
9592 (emit_call_insn_before_noloc): Likewise.
9593 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
9595 (emit_barrier_before): Likewise.
9596 (emit_label_before): Strengthen "label" param from "rtx" to
9597 "rtx_code_label *". Strengthen "before" param from "rtx" to
9599 (emit_insn_after_1): Strengthen "after" param from "rtx" to
9601 (emit_pattern_after_noloc): Likewise.
9602 (emit_insn_after_noloc): Likewise.
9603 (emit_jump_insn_after_noloc): Likewise.
9604 (emit_call_insn_after_noloc): Likewise.
9605 (emit_debug_insn_after_noloc): Likewise.
9606 (emit_barrier_after): Likewise.
9607 (emit_label_after): Likewise for both params.
9608 (emit_pattern_after_setloc): Likewise for "after" param. Convert
9609 "loc" param from "int" to "location_t".
9610 (emit_insn_after_setloc): Likewise.
9611 (emit_jump_insn_after_setloc): Likewise.
9612 (emit_call_insn_after_setloc): Likewise.
9613 (emit_debug_insn_after_setloc): Likewise.
9614 (emit_pattern_before_setloc): Likewise for "before" param. Convert
9615 "loc" param from "int" to "location_t".
9616 (emit_pattern_before): Convert NULL_RTX to NULL.
9617 (emit_insn_before_setloc): Convert "loc" param from "int" to
9619 (emit_jump_insn_before_setloc): Likewise.
9620 (emit_call_insn_before_setloc): Likewise.
9621 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
9622 rtx_insn *. Convert "loc" param from "int" to "location_t".
9623 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
9624 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
9625 Convert 3rd param from "int" to "location_t".
9626 (emit_barrier_before, emit_barrier_after, next_real_insn):
9627 Strengthen param from rtx to rtx_insn *.
9628 (emit_label_before): Strengthen 1st param from "rtx" to
9629 "rtx_code_label *". Strengthen 2nd param from "rtx" to
9631 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
9632 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
9633 Strengthen 2nd param from "rtx" to "rtx_insn *".
9634 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
9635 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
9636 Likewise. Convert 3rd param from "int" to "location_t".
9637 (emit_label_after): Strengthen 1st param from "rtx" to
9639 (next_real_insn, remove_insn): Strengthen param from "rtx" to
9641 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
9642 from "rtx" to "rtx_insn *".
9644 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
9646 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
9647 bodies streamed in with -Q.
9648 * dumpfile.c (dump_files): Add lto-stream-out dump file.
9649 * dumpfile.h (tree_dump_index): Add lto_stream_out.
9650 * gimple-streamer-out.c: Include gimple-pretty-print.h
9651 (output_bb): Dump stmts streamed.
9652 * lto-section-out.c: Include print-tree.h
9653 (lto_begin_section): Dump sections created.
9654 (lto_output_decl_index): Dump decl encoded.
9655 * lto-streamer-out.c: Include print-tree.h
9656 (create_output_block): Dump output block created.
9657 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
9658 (output_function): Dump function output.
9659 (output_constructor): Dump constructor streamed.
9660 (write_global_stream): Output indexes encoded.
9661 (produce_asm_for_decls): Dump streams encoded.
9662 * lto-streamer.c (streamer_dump_file): New global var.
9663 * lto-streamer.h (streamer_dump_file): Declare.
9664 * passes.c (ipa_write_summaries): Initialize streamer dump.
9665 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
9668 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
9671 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
9672 offsets for register save directives. Emit a second batch of save
9673 directives, if need be, when the function accesses prior frames.
9675 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9677 * config/arc/fpu.md (fmasf4): Force operand to register.
9678 (fnmasf4): Likewise.
9680 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9682 * config/arc/arc-protos.h (arc_pad_return): Remove.
9683 * config/arc/arc.c (machine_function): Remove force_short_suffix
9685 (arc_print_operand): Adjust printing of '&'.
9686 (arc_verify_short): Remove conditional printing of short suffix.
9687 (arc_final_prescan_insn): Remove reference to size_reason.
9688 (pad_return): New function.
9689 (arc_reorg): Call pad_return.
9690 (arc_pad_return): Remove.
9691 (arc_init_machine_status): Remove reference to force_short_suffix.
9692 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
9693 (attr length): When attribute iscompact is true force to 2
9694 regardless; in the case of maybe check if we want to force the
9695 instruction to have 4 bytes length.
9696 (nopv): Change it to generate 4 byte long nop as well.
9697 (blockage): New pattern.
9698 (simple_return): Remove call to arc_pad_return.
9699 (p_return_i): Likewise.
9701 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9703 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
9705 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9707 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
9710 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9712 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
9713 for ARC700 and ARCv2.
9715 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
9718 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
9719 operands[2] instead of operands[1].
9722 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9724 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
9725 case, check whether the outer register overlaps an unallocatable
9726 register, not just whether it fits the required class.
9728 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9730 * poly-int.h (can_div_trunc_p): Add new overload in which all values
9732 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
9733 (memrefs_conflict_p): Likewise.
9734 (init_alias_analysis): Likewise.
9735 * cfgexpand.c (expand_debug_expr): Likewise.
9736 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
9737 * cse.c (fold_rtx): Likewise.
9738 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
9739 * expr.c (emit_block_move_hints): Likewise.
9740 (clear_storage_hints, push_block, emit_push_insn): Likewise.
9741 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
9742 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
9743 (emit_group_store): Likewise.
9744 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
9745 to read the PRE/POST_MODIFY increment.
9746 * calls.c (store_one_arg): Use strip_offset.
9747 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
9749 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
9751 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
9752 (simplify_binary_operation_1): Extend CONST_INT handling to
9754 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
9755 than a HOST_WIDE_INT.
9756 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
9758 (adjust_mems, add_stores): Update accodingly.
9759 (vt_canonicalize_addr): Track polynomial offsets.
9760 (emit_note_insn_var_location): Likewise.
9761 (vt_add_function_parameter): Likewise.
9762 (vt_initialize): Likewise.
9764 2018-06-12 Jeff Law <law@redhat.com>
9766 * config.gcc (alpha*-*-freebsd*): Remove.
9767 * config/alpha/freebsd.h: Remove.
9769 2018-06-12 David Malcolm <dmalcolm@redhat.com>
9772 * spellcheck-tree.c (levenshtein_distance): Rename to...
9773 (get_edit_distance): ...this, and update for underlying renaming.
9774 * spellcheck-tree.h (levenshtein_distance): Rename to...
9775 (get_edit_distance): ...this.
9776 * spellcheck.c (levenshtein_distance): Rename to...
9777 (get_edit_distance): ...this. Convert from Levenshtein distance
9778 to Damerau-Levenshtein distance by supporting transpositions of
9779 adjacent characters. Rename "v1" to "v_next" and "v0" to
9781 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
9782 (selftest::test_edit_distance_unit_test_oneway): ...this, and
9783 update for underlying renaming.
9784 (selftest::levenshtein_distance_unit_test): Rename to...
9785 (selftest::test_get_edit_distance_unit): ...this, and update for
9786 underlying renaming.
9787 (selftest::test_find_closest_string): Add example from PR 69968
9788 where transposition helps
9789 (selftest::test_metric_conditions): Update for renaming.
9790 (selftest::test_metric_conditions): Likewise.
9791 (selftest::spellcheck_c_tests): Likewise.
9792 * spellcheck.h (levenshtein_distance): Rename both overloads to...
9793 (get_edit_distance): ...this.
9794 (best_match::consider): Update for renaming.
9796 2018-06-12 Martin Sebor <msebor@redhat.com>
9798 PR tree-optimization/85259
9799 * builtins.c (compute_objsize): Handle constant offsets.
9800 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
9801 true iff a warning has been issued.
9802 * gimple.h (gimple_nonartificial_location): New function.
9803 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
9804 gimple_nonartificial_location and handle -Wno-system-headers.
9805 (handle_builtin_stxncpy): Same.
9807 2018-06-12 Martin Sebor <msebor@redhat.com>
9810 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
9812 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9814 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9815 BUILTIN_VEC_XST entries for pointer to double and long long.
9817 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
9820 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9822 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9825 2018-06-12 Martin Liska <mliska@suse.cz>
9827 * doc/options.texi: Document IntegerRange.
9829 2018-06-12 Martin Liska <mliska@suse.cz>
9831 * config/i386/i386.opt: Make MPX-related options as Deprecated.
9832 * opt-functions.awk: Handle Deprecated flag.
9833 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
9835 (read_cmdline_option): Report warning for a deprecated option.
9836 * opts.h (struct cl_option): Add new field cl_deprecated.
9837 (CL_ERR_DEPRECATED): New.
9839 2018-06-12 Martin Liska <mliska@suse.cz>
9841 * doc/options.texi: Document Deprecated option flag.
9843 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9845 * config/arc/arc-arch.h (arc_extras): New enum.
9846 (arc_cpu_t):Add field extra.
9847 (arc_cpu_types): Consider the extras.
9848 * config/arc/arc-cpus.def: Add extras info.
9849 * config/arc/arc-opts.h (processor_type): Consider extra field.
9850 * config/arc/arc.c (arc_override_options): Handle extra field.
9852 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9854 * config/arc/arc-arch.h: Update ARC_OPTX macro.
9855 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
9857 * config/arc/arc.c (arc_init): Update pic warning.
9858 (irq_range): Update irq range parsing warnings.
9859 (arc_override_options): Update various warning messages.
9860 (arc_handle_aux_attribute): Likewise.
9862 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
9864 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
9866 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9868 * doc/sourcebuild.texi: Document usage of line number 0 in verify
9869 compiler messages directives.
9871 2018-06-12 Matthew Fortune <mfortune@gmail.com>
9873 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
9874 * config/mips/mips-tables.opt: Regenerate.
9875 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
9877 * doc/invoke.texi: Document -march=i6500.
9879 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
9881 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
9882 (i6400_gpmul): Add cpu_unit.
9883 (i6400_gpdiv): Likewise.
9884 (i6400_msa_add_d): Update reservations.
9885 (i6400_msa_int_add) Likewise.
9886 (i6400_msa_short_logic3) Likewise.
9887 (i6400_msa_short_logic2) Likewise.
9888 (i6400_msa_short_logic) Likewise.
9889 (i6400_msa_move) Likewise.
9890 (i6400_msa_cmp) Likewise.
9891 (i6400_msa_short_float2) Likewise.
9892 (i6400_msa_div_d) Likewise.
9893 (i6400_msa_long_logic1) Likewise.
9894 (i6400_msa_long_logic2) Likewise.
9895 (i6400_msa_mult) Likewise.
9896 (i6400_msa_long_float2) Likewise.
9897 (i6400_msa_long_float4) Likewise.
9898 (i6400_msa_long_float5) Likewise.
9899 (i6400_msa_long_float8) Likewise.
9900 (i6400_fpu_fadd): Include frint type.
9901 (i6400_fpu_store): New define_insn_reservation.
9902 (i6400_fpu_load): Likewise.
9903 (i6400_fpu_move): Likewise.
9904 (i6400_fpu_fcmp): Likewise.
9905 (i6400_fpu_fmadd): Likewise.
9906 (i6400_int_mult): Include imul3nc type and update reservation.
9907 (i6400_int_div): Include idiv3 type and update reservation.
9908 (i6400_int_load): Update to check type not move_type.
9909 (i6400_int_store): Likewise.
9910 (i6400_int_prefetch): Set zero latency.
9912 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
9914 * gcc.c: Document new %@{...} sequence.
9915 (LINK_COMMAND_SPEC): Use it for the -L switches.
9916 (cpp_unique_options): Use it for the -I switches.
9917 (at_file_argbuf): New global variable.
9918 (in_at_file): Likewise.
9919 (alloc_args): Create at_file_argbuf.
9920 (clear_args): Truncate at_file_argbuf.
9921 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
9922 (open_at_file): New function.
9923 (close_at_file): Likewise.
9924 (create_at_file): Delete.
9925 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
9928 (validate_switches_from_spec): Deal with %@{...} sequence.
9929 (validate_switches): Likewise.
9930 (driver::finalize): Call clear_args.
9932 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
9934 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
9936 2018-06-11 Martin Sebor <msebor@redhat.com>
9938 * doc/invoke.texi (-Wall): List -Wc++17-compat.
9939 (Wno-class-memaccess): Add @opindex.
9940 (Wno-templates, Wno-multiple-inheritance): Same.
9941 (Wno-virtual-inheritance, Wno-namespaces): Same.
9942 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
9943 (Wno-format-overflow, Wno-format-truncation): Same.
9944 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
9945 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
9946 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
9947 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
9948 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
9949 (Wno-misspelled-isr): Same.
9951 2018-06-11 Martin Sebor <msebor@redhat.com>
9953 * PR tree-optimization/86083
9954 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
9956 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
9958 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
9960 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
9963 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
9964 on the correct operand.
9965 (*movdi_internal64): Ditto.
9967 2018-06-11 Martin Liska <mliska@suse.cz>
9969 PR tree-optimization/86089
9970 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
9972 2018-06-11 Julia Koval <julia.koval@intel.com>
9974 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
9975 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
9976 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
9978 2018-06-11 Olivier Hainque <hainque@adacore.com>
9980 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
9981 for Ada with strict dwarf2.
9983 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
9986 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
9989 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
9991 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
9993 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
9995 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
9998 2018-06-08 Martin Liska <mliska@suse.cz>
10000 * tree-cfg.h (debug_function): Fix argument type to match
10003 2018-06-08 Martin Liska <mliska@suse.cz>
10005 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
10006 Remove usage of MPX-related (and removed) fields.
10007 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
10009 2018-06-08 David Malcolm <dmalcolm@redhat.com>
10011 * cfg.c (debug): Use TDF_NONE rather than 0.
10012 * cfghooks.c (debug): Likewise.
10013 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
10014 (struct dump_option_value_info): Convert to...
10015 (struct kv_pair): ...this template type.
10016 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
10018 (optinfo_verbosity_options): Likewise.
10019 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
10021 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
10022 than int for "optgroup_flags" param.
10023 (dump_generic_expr_loc): Use dump_flags_t rather than int for
10025 (dump_dec): Likewise.
10026 (dump_finish): Use TDF_NONE rather than 0.
10027 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
10028 rather than int for "optgroup_flags" param. Use TDF_NONE rather
10029 than 0. Update for change to option_ptr.
10030 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
10031 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
10032 0. Update for changes to optinfo_verbosity_options and
10034 (opt_info_switch_p): Convert optgroup_flags from int to
10036 (dump_basic_block): Use dump_flags_t rather than int
10037 for "dump_kind" param.
10038 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
10039 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
10040 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
10041 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
10042 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
10043 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
10044 TDF_NONE): Convert from macros to...
10045 (enum dump_flag): ...this new enum.
10046 (dump_flags_t): Update to use enum.
10047 (operator|, operator&, operator~, operator|=, operator&=):
10048 Implement for dump_flags_t.
10049 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
10050 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
10051 Convert from macros to...
10052 (enum optgroup_flag): ...this new enum.
10053 (optgroup_flags_t): New typedef.
10054 (operator|, operator|=): Implement for optgroup_flags_t.
10055 (struct dump_file_info): Convert field "alt_flags" to
10056 dump_flags_t. Convert field "optgroup_flags" to
10058 (dump_basic_block): Use dump_flags_t rather than int for param.
10059 (dump_generic_expr_loc): Likewise.
10060 (dump_dec): Likewise.
10061 (dump_register): Convert param "optgroup_flags" to
10063 (opt_info_enable_passes): Likewise.
10064 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
10066 * gimple-pretty-print.c (debug): Likewise.
10067 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
10068 (merged_store_group::apply_stores): Likewise.
10069 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
10070 * gimple.c (verify_gimple_pp): Likewise.
10071 * graphite-poly.c (print_pbb_body): Likewise.
10072 * passes.c (pass_manager::register_one_dump_file): Convert
10073 local "optgroup_flags" to optgroup_flags_t.
10074 * print-tree.c (print_node): Use TDF_NONE rather than 0.
10076 (debug_body): Likewise.
10077 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
10078 to optgroup_flags_t.
10079 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
10081 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
10082 (convert_mult_to_fma): Likewise.
10083 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
10084 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
10085 * tree-vect-data-refs.c (dump_lower_bound): Convert param
10086 "dump_kind" to dump_flags_t.
10088 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
10090 * config/rs6000/rs6000.c (min, max): Delete.
10092 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
10094 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
10095 -mabi=spe and -mabi=no-spe.
10097 2018-06-08 Martin Liska <mliska@suse.cz>
10099 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
10100 where we expect an existing summary.
10102 2018-06-08 Martin Liska <mliska@suse.cz>
10104 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
10105 * ipa-inline.h (estimate_edge_growth): Likewise.
10107 2018-06-08 Martin Liska <mliska@suse.cz>
10109 * cgraph.c (function_version_hasher::hash): Use
10110 cgraph_node::get_uid ().
10111 (function_version_hasher::equal):
10112 * cgraph.h (cgraph_node::get_uid): New method.
10113 * ipa-inline.c (update_caller_keys): Use
10114 cgraph_node::get_uid ().
10115 (update_callee_keys): Likewise.
10116 * ipa-utils.c (searchc): Likewise.
10117 (ipa_reduced_postorder): Likewise.
10118 * lto-cgraph.c (input_node): Likewise.
10119 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
10120 * symbol-summary.h (symtab_insertion): Likewise.
10121 (symtab_removal): Likewise.
10122 (symtab_duplication): Likewise.
10123 * tree-pretty-print.c (dump_function_header): Likewise.
10124 * tree-sra.c (convert_callers_for_node): Likewise.
10126 2018-06-08 Martin Liska <mliska@suse.cz>
10128 * cgraph.c (symbol_table::create_edge): Always assign a new
10130 (symbol_table::free_edge): Do not recycle numbers.
10131 * cgraph.h (cgraph_edge::get): New method.
10132 * symbol-summary.h (symtab_removal): Use it.
10133 (symtab_duplication): Likewise.
10134 (call_summary::hashable_uid): Remove.
10136 2018-06-08 Martin Liska <mliska@suse.cz>
10138 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
10139 (initialize_growth_caches): Remove.
10140 (free_growth_caches): Likewise.
10141 (do_estimate_edge_time): Use edge_growth_cache.
10142 (do_estimate_edge_size): Likewise.
10143 (do_estimate_edge_hints): Likewise.
10144 * ipa-inline.c (reset_edge_caches): Likewise.
10145 (recursive_inlining): Likewise.
10146 (inline_small_functions): Likewise.
10147 * ipa-inline.h (initialize_growth_caches): Remove.
10148 (estimate_edge_size): Likewise.
10149 (estimate_edge_time): Likewise.
10150 (estimate_edge_hints): Likewise.
10151 (reset_edge_growth_cache): Likewise.
10152 * symbol-summary.h (call_summary::remove): New method.
10154 2018-06-08 Martin Liska <mliska@suse.cz>
10156 * ipa-cp.c (class edge_clone_summary): New summary.
10157 (grow_edge_clone_vectors): Remove.
10158 (ipcp_edge_duplication_hook): Remove.
10159 (class edge_clone_summary_t): New call_summary class.
10160 (ipcp_edge_removal_hook): Remove.
10161 (edge_clone_summary_t::duplicate): New function.
10162 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
10163 (create_specialized_node): Likewise.
10164 (ipcp_driver): Initialize edge_clone_summaries and do not
10167 2018-06-08 Martin Liska <mliska@suse.cz>
10169 * symbol-summary.h (get): New function.
10170 (call_summary::m_initialize_when_cloning): New class member.
10172 2018-06-08 Martin Liska <mliska@suse.cz>
10174 * cgraph.c (cgraph_node::remove): Do not recycle uid.
10175 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
10176 (symbol_table::allocate_cgraph_symbol): Do not set uid.
10177 * passes.c (uid_hash_t): Record removed_nodes by their uids.
10178 (remove_cgraph_node_from_order): Use the removed_nodes set.
10179 (do_per_function_toporder): Likwise.
10180 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
10181 instead of summary_uid.
10182 (symtab_removal): Likewise.
10183 (symtab_duplication): Likewise.
10185 2018-06-08 Martin Liska <mliska@suse.cz>
10187 * ipa-cp.c (ipcp_store_bits_results): Use
10188 ipcp_transformation_sum.
10189 (ipcp_store_vr_results): Likewise.
10190 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
10192 (ipcp_transformation_initialize): ... this.
10193 (ipa_set_node_agg_value_chain):
10194 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
10195 (write_ipcp_transformation_info): Likewise.
10196 (read_ipcp_transformation_info): Likewise.
10197 (ipcp_update_bits): Likewise.
10198 (ipcp_update_vr): Likewise.
10199 (ipcp_transform_function): Likewise.
10200 * ipa-prop.h: Rename ipcp_transformation_summary to
10201 ipcp_transformation.
10202 (class ipcp_transformation_t): New function summary.
10203 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
10204 (ipa_get_agg_replacements_for_node): Likewise.
10206 2018-06-08 Martin Liska <mliska@suse.cz>
10208 * ipa-pure-const.c (struct funct_state_d): Do it class instead
10210 (class funct_state_summary_t): New function_summary class.
10211 (has_function_state): Remove.
10212 (get_function_state): Likewise.
10213 (set_function_state): Likewise.
10214 (add_new_function): Likewise.
10215 (funct_state_summary_t::insert): New function.
10216 (duplicate_node_data): Remove.
10217 (remove_node_data): Remove.
10218 (funct_state_summary_t::duplicate): New function.
10219 (register_hooks): Create new funct_state_summaries.
10220 (pure_const_generate_summary): Use it.
10221 (pure_const_write_summary): Likewise.
10222 (pure_const_read_summary): Likewise.
10223 (propagate_pure_const): Likewise.
10224 (propagate_nothrow): Likewise.
10225 (dump_malloc_lattice): Likewise.
10226 (propagate_malloc): Likewise.
10227 (execute): Do not register hooks, just remove summary
10229 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
10232 2018-06-08 Martin Liska <mliska@suse.cz>
10234 * ipa-reference.c (remove_node_data): Remove.
10235 (duplicate_node_data): Likewise.
10236 (class ipa_ref_var_info_summary_t): New class.
10237 (class ipa_ref_opt_summary_t): Likewise.
10238 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
10239 (get_reference_optimization_summary): Use
10240 ipa_ref_opt_sum_summaries.
10241 (set_reference_vars_info): Remove.
10242 (set_reference_optimization_summary): Likewise.
10243 (ipa_init): Create summaries.
10244 (init_function_info): Use function summary.
10245 (ipa_ref_opt_summary_t::duplicate): New function.
10246 (ipa_ref_opt_summary_t::remove): New function.
10247 (get_read_write_all_from_node): Fix GNU coding style.
10248 (propagate): Use function summary.
10249 (write_node_summary_p): Fix GNU coding style.
10250 (stream_out_bitmap): Likewise.
10251 (ipa_reference_read_optimization_summary): Use function summary.
10252 (ipa_reference_c_finalize): Do not release hooks.
10254 2018-06-08 Martin Liska <mliska@suse.cz>
10256 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
10257 (analyze_function_body): Extract multiple calls of get_create.
10258 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
10259 * ipa-inline.c (recursive_inlining): Use ::get method.
10260 * ipa-inline.h (estimate_edge_growth): Likewise.
10262 2018-06-08 Martin Liska <mliska@suse.cz>
10264 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
10266 (hsa_function_summary::hsa_function_summary): Use the new enum
10268 (hsa_gpu_implementation_p): Use hsa_summaries::get.
10269 * hsa-gen.c (hsa_get_host_function): Likewise.
10270 (get_brig_function_name): Likewise.
10271 * ipa-hsa.c (process_hsa_functions): Likewise.
10272 (ipa_hsa_write_summary): Likewise.
10273 * symbol-summary.h (symtab_duplication): Use ::get function/
10274 (get): New function.
10276 2018-06-08 Martin Liska <mliska@suse.cz>
10278 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
10280 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
10281 (hsa_register_kernel): Likewise.
10282 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
10283 * hsa-gen.c (hsa_get_host_function): Likewise.
10284 (get_brig_function_name): Likewise.
10285 (generate_hsa): Likewise.
10286 (pass_gen_hsail::execute): Likewise.
10287 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
10288 (devirtualization_time_bonus): Likewise.
10289 (ipcp_propagate_stage): Likewise.
10290 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
10291 (edge_set_predicate): Likewise.
10292 (evaluate_conditions_for_known_args): Likewise.
10293 (evaluate_properties_for_edge): Likewise.
10294 (ipa_fn_summary::reset): Likewise.
10295 (ipa_fn_summary_t::duplicate): Likewise.
10296 (dump_ipa_call_summary): Likewise.
10297 (ipa_dump_fn_summary): Likewise.
10298 (analyze_function_body): Likewise.
10299 (compute_fn_summary): Likewise.
10300 (estimate_edge_devirt_benefit): Likewise.
10301 (estimate_edge_size_and_time): Likewise.
10302 (estimate_calls_size_and_time): Likewise.
10303 (estimate_node_size_and_time): Likewise.
10304 (inline_update_callee_summaries): Likewise.
10305 (remap_edge_change_prob): Likewise.
10306 (remap_edge_summaries): Likewise.
10307 (ipa_merge_fn_summary_after_inlining): Likewise.
10308 (ipa_update_overall_fn_summary): Likewise.
10309 (read_ipa_call_summary): Likewise.
10310 (inline_read_section): Likewise.
10311 (write_ipa_call_summary): Likewise.
10312 (ipa_fn_summary_write): Likewise.
10313 (ipa_free_fn_summary): Likewise.
10314 * ipa-hsa.c (process_hsa_functions): Likewise.
10315 (ipa_hsa_write_summary): Likewise.
10316 (ipa_hsa_read_section): Likewise.
10317 * ipa-icf.c (sem_function::merge): Likewise.
10318 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
10319 (do_estimate_edge_time): Likewise.
10320 (estimate_size_after_inlining): Likewise.
10321 (estimate_growth): Likewise.
10322 (growth_likely_positive): Likewise.
10323 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
10324 (inline_call): Likewise.
10325 * ipa-inline.c (caller_growth_limits): Likewise.
10326 (can_inline_edge_p): Likewise.
10327 (can_inline_edge_by_limits_p): Likewise.
10328 (compute_uninlined_call_time): Likewise.
10329 (compute_inlined_call_time): Likewise.
10330 (want_inline_small_function_p): Likewise.
10331 (edge_badness): Likewise.
10332 (update_caller_keys): Likewise.
10333 (update_callee_keys): Likewise.
10334 (recursive_inlining): Likewise.
10335 (inline_small_functions): Likewise.
10336 (inline_to_all_callers_1): Likewise.
10337 (dump_overall_stats): Likewise.
10338 (early_inline_small_functions): Likewise.
10339 (early_inliner): Likewise.
10340 * ipa-inline.h (estimate_edge_growth): Likewise.
10341 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10342 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
10343 * ipa-prop.h (IPA_NODE_REF): Likewise.
10344 (IPA_EDGE_REF): Likewise.
10345 * ipa-pure-const.c (malloc_candidate_p): Likewise.
10346 (propagate_malloc): Likewise.
10347 * ipa-split.c (execute_split_functions): Likewise.
10348 * symbol-summary.h: Rename get to get_create.
10350 (get_create): Likewise.
10351 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
10353 2018-06-08 Martin Liska <mliska@suse.cz>
10355 * symbol-summary.h (release): Move definition out of class
10357 (symtab_removal): Likewise.
10358 (symtab_duplication): Likewise.
10360 2018-06-08 Martin Liska <mliska@suse.cz>
10362 * symbol-summary.h (function_summary): Move constructor
10363 implementation out of class declaration.
10364 (release): Likewise.
10365 (symtab_insertion): Likewise.
10366 (symtab_removal): Likewise.
10367 (symtab_duplication): Likewise.
10370 2018-06-08 Martin Liska <mliska@suse.cz>
10372 * Makefile.in: Remove support for MPX (macros, related functions,
10373 fields in cgraph_node, ...).
10374 * builtin-types.def (BT_BND): Likewise.
10375 (BT_FN_BND_CONST_PTR): Likewise.
10376 (BT_FN_CONST_PTR_BND): Likewise.
10377 (BT_FN_VOID_PTR_BND): Likewise.
10378 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
10379 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
10380 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
10381 (expand_builtin_mempcpy_with_bounds): Likewise.
10382 (expand_builtin_memset_with_bounds): Likewise.
10383 (expand_builtin_memset_args): Likewise.
10384 (std_expand_builtin_va_start): Likewise.
10385 (expand_builtin): Likewise.
10386 (expand_builtin_with_bounds): Likewise.
10387 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
10388 (DEF_LIB_BUILTIN_CHKP): Likewise.
10389 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
10390 (DEF_CHKP_BUILTIN): Likewise.
10391 (BUILT_IN_MEMCPY): Likewise.
10392 (BUILT_IN_MEMMOVE): Likewise.
10393 (BUILT_IN_MEMPCPY): Likewise.
10394 (BUILT_IN_MEMSET): Likewise.
10395 (BUILT_IN_STPCPY): Likewise.
10396 (BUILT_IN_STRCAT): Likewise.
10397 (BUILT_IN_STRCHR): Likewise.
10398 (BUILT_IN_STRCPY): Likewise.
10399 (BUILT_IN_STRLEN): Likewise.
10400 (BUILT_IN_MEMCPY_CHK): Likewise.
10401 (BUILT_IN_MEMMOVE_CHK): Likewise.
10402 (BUILT_IN_MEMPCPY_CHK): Likewise.
10403 (BUILT_IN_MEMSET_CHK): Likewise.
10404 (BUILT_IN_STPCPY_CHK): Likewise.
10405 (BUILT_IN_STRCAT_CHK): Likewise.
10406 (BUILT_IN_STRCPY_CHK): Likewise.
10407 * calls.c (store_bounds): Likewise.
10408 (emit_call_1): Likewise.
10409 (special_function_p): Likewise.
10410 (maybe_warn_nonstring_arg): Likewise.
10411 (initialize_argument_information): Likewise.
10412 (finalize_must_preallocate): Likewise.
10413 (compute_argument_addresses): Likewise.
10414 (expand_call): Likewise.
10415 * cfgexpand.c (expand_call_stmt): Likewise.
10416 (expand_return): Likewise.
10417 (expand_gimple_stmt_1): Likewise.
10418 (pass_expand::execute): Likewise.
10419 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
10420 (cgraph_node::remove): Likewise.
10421 (cgraph_node::dump): Likewise.
10422 (cgraph_node::verify_node): Likewise.
10423 * cgraph.h (chkp_function_instrumented_p): Likewise.
10424 (symtab_node::get_alias_target): Likewise.
10425 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
10426 (cgraph_local_p): Likewise.
10427 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
10428 (cgraph_edge::rebuild_references): Likewise.
10429 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
10430 (walk_polymorphic_call_targets): Likewise.
10431 (cgraph_node::expand_thunk): Likewise.
10432 (symbol_table::output_weakrefs): Likewise.
10433 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
10434 (ix86_handle_option): Likewise.
10435 * config/i386/constraints.md: Likewise.
10436 * config/i386/i386-builtin-types.def (BND): Likewise.
10440 * config/i386/i386-builtin.def (BDESC_END): Likewise.
10441 (BDESC_FIRST): Likewise.
10443 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
10444 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
10445 * config/i386/i386.c (enum reg_class): Likewise.
10446 (ix86_target_string): Likewise.
10447 (ix86_option_override_internal): Likewise.
10448 (ix86_conditional_register_usage): Likewise.
10449 (ix86_valid_target_attribute_inner_p): Likewise.
10450 (ix86_set_indirect_branch_type): Likewise.
10451 (ix86_set_current_function): Likewise.
10452 (ix86_function_arg_regno_p): Likewise.
10453 (init_cumulative_args): Likewise.
10454 (ix86_function_arg_advance): Likewise.
10455 (ix86_function_arg): Likewise.
10456 (ix86_pass_by_reference): Likewise.
10457 (ix86_function_value_regno_p): Likewise.
10458 (ix86_function_value_1): Likewise.
10459 (ix86_function_value_bounds): Likewise.
10460 (ix86_return_in_memory): Likewise.
10461 (ix86_setup_incoming_vararg_bounds): Likewise.
10462 (ix86_va_start): Likewise.
10463 (indirect_thunk_need_prefix): Likewise.
10464 (print_reg): Likewise.
10465 (ix86_print_operand): Likewise.
10466 (ix86_expand_call): Likewise.
10467 (ix86_output_function_return): Likewise.
10468 (reg_encoded_number): Likewise.
10469 (BDESC_VERIFYS): Likewise.
10470 (ix86_init_mpx_builtins): Likewise.
10471 (ix86_init_builtins): Likewise.
10472 (ix86_emit_cmove): Likewise.
10473 (ix86_emit_move_max): Likewise.
10474 (ix86_expand_builtin): Likewise.
10475 (ix86_builtin_mpx_function): Likewise.
10476 (ix86_get_arg_address_for_bt): Likewise.
10477 (ix86_load_bounds): Likewise.
10478 (ix86_store_bounds): Likewise.
10479 (ix86_load_returned_bounds): Likewise.
10480 (ix86_store_returned_bounds): Likewise.
10481 (ix86_class_likely_spilled_p): Likewise.
10482 (ix86_hard_regno_mode_ok): Likewise.
10483 (x86_order_regs_for_local_alloc): Likewise.
10484 (ix86_mitigate_rop): Likewise.
10485 (ix86_bnd_prefixed_insn_p): Likewise.
10486 (ix86_mpx_bound_mode): Likewise.
10487 (ix86_make_bounds_constant): Likewise.
10488 (ix86_initialize_bounds): Likewise.
10489 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
10490 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
10491 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
10492 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
10493 (TARGET_CHKP_BOUND_MODE): Likewise.
10494 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
10495 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
10496 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
10497 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
10498 * config/i386/i386.h (TARGET_MPX): Likewise.
10499 (TARGET_MPX_P): Likewise.
10500 (VALID_BND_REG_MODE): Likewise.
10501 (FIRST_BND_REG): Likewise.
10502 (LAST_BND_REG): Likewise.
10503 (enum reg_class): Likewise.
10504 (BND_REG_P): Likewise.
10505 (BND_REGNO_P): Likewise.
10506 (BNDmode): Likewise.
10507 (ADJUST_INSN_LENGTH): Likewise.
10508 * config/i386/i386.md: Likewise.
10509 * config/i386/i386.opt: Likewise.
10510 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
10511 (defined): Likewise.
10512 (LINK_MPX): Likewise.
10513 (MPX_SPEC): Likewise.
10514 (LIBMPX_SPEC): Likewise.
10515 (LIBMPXWRAPPERS_SPEC): Likewise.
10516 (CHKP_SPEC): Likewise.
10517 * config/i386/predicates.md: Likewise.
10518 * dbxout.c (dbxout_type): Likewise.
10519 * doc/extend.texi: Likewise.
10520 * doc/invoke.texi: Likewise.
10521 * doc/md.texi: Likewise.
10522 * doc/tm.texi: Likewise.
10523 * doc/tm.texi.in: Likewise.
10524 * dwarf2out.c (is_base_type): Likewise.
10525 (gen_formal_types_die): Likewise.
10526 (gen_subprogram_die): Likewise.
10527 (gen_type_die_with_usage): Likewise.
10528 (gen_decl_die): Likewise.
10529 (dwarf2out_late_global_decl): Likewise.
10530 * expr.c (expand_assignment): Likewise.
10531 (emit_storent_insn): Likewise.
10532 (store_expr_with_bounds): Likewise.
10533 (store_expr): Likewise.
10534 (expand_expr_real_1): Likewise.
10535 * expr.h (store_expr_with_bounds): Likewise.
10536 * function.c (use_register_for_decl): Likewise.
10537 (struct bounds_parm_data): Likewise.
10538 (assign_parms_augmented_arg_list): Likewise.
10539 (assign_parm_find_entry_rtl): Likewise.
10540 (assign_parm_is_stack_parm): Likewise.
10541 (assign_parm_load_bounds): Likewise.
10542 (assign_bounds): Likewise.
10543 (assign_parms): Likewise.
10544 (expand_function_start): Likewise.
10545 * gcc.c (CHKP_SPEC): Likewise.
10546 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10547 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
10548 (wrestrict_dom_walker::check_call): Likewise.
10549 * gimple.c (gimple_build_call_from_tree): Likewise.
10550 * gimple.h (enum gf_mask): Likewise.
10551 (gimple_call_with_bounds_p): Likewise.
10552 (gimple_call_set_with_bounds): Likewise.
10553 * gimplify.c (gimplify_init_constructor): Likewise.
10554 * ipa-cp.c (initialize_node_lattices): Likewise.
10555 (propagate_constants_across_call): Likewise.
10556 (find_more_scalar_values_for_callers_subset): Likewise.
10557 * ipa-hsa.c (process_hsa_functions): Likewise.
10558 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
10559 * ipa-icf.c (sem_function::merge): Likewise.
10560 * ipa-inline.c (early_inliner): Likewise.
10561 * ipa-pure-const.c (warn_function_noreturn): Likewise.
10562 (warn_function_cold): Likewise.
10563 (propagate_pure_const): Likewise.
10564 * ipa-ref.h (enum GTY): Likewise.
10565 * ipa-split.c (find_retbnd): Likewise.
10566 (consider_split): Likewise.
10567 (split_function): Likewise.
10568 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
10569 * ipa.c (walk_polymorphic_call_targets): Likewise.
10570 (symbol_table::remove_unreachable_nodes): Likewise.
10571 (process_references): Likewise.
10572 (cgraph_build_static_cdtor_1): Likewise.
10573 * lto-cgraph.c (lto_output_node): Likewise.
10574 (output_refs): Likewise.
10575 (compute_ltrans_boundary): Likewise.
10576 (input_overwrite_node): Likewise.
10577 (input_node): Likewise.
10578 (input_cgraph_1): Likewise.
10579 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
10580 * passes.c (pass_manager::execute_early_local_passes): Likewise.
10581 (class pass_chkp_instrumentation_passes): Likewise.
10582 (make_pass_chkp_instrumentation_passes): Likewise.
10583 * passes.def: Likewise.
10584 * rtl.h (struct GTY): Likewise.
10585 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
10586 * stor-layout.c (layout_type): Likewise.
10587 * symtab.c: Likewise.
10588 * target.def: Likewise.
10589 * targhooks.c (default_chkp_bound_type): Likewise.
10590 (default_chkp_bound_mode): Likewise.
10591 (default_builtin_chkp_function): Likewise.
10592 (default_chkp_function_value_bounds): Likewise.
10593 (default_chkp_make_bounds_constant): Likewise.
10594 (default_chkp_initialize_bounds): Likewise.
10595 * targhooks.h (default_chkp_bound_type): Likewise.
10596 (default_chkp_bound_mode): Likewise.
10597 (default_builtin_chkp_function): Likewise.
10598 (default_chkp_function_value_bounds): Likewise.
10599 (default_chkp_make_bounds_constant): Likewise.
10600 (default_chkp_initialize_bounds): Likewise.
10601 * toplev.c (compile_file): Likewise.
10602 (process_options): Likewise.
10603 * tree-core.h (DEF_BUILTIN): Likewise.
10604 (DEF_BUILTIN_CHKP): Likewise.
10605 * tree-inline.c (declare_return_variable): Likewise.
10606 (remap_gimple_stmt): Likewise.
10607 (copy_bb): Likewise.
10608 (initialize_inlined_parameters): Likewise.
10609 (expand_call_inline): Likewise.
10610 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
10611 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
10612 (make_pass_ipa_chkp_produce_thunks): Likewise.
10613 (make_pass_chkp): Likewise.
10614 (make_pass_chkp_opt): Likewise.
10615 (make_pass_chkp_instrumentation_passes): Likewise.
10616 * tree-pretty-print.c (dump_generic_node): Likewise.
10617 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
10618 * tree-ssa-dce.c (propagate_necessity): Likewise.
10619 (eliminate_unnecessary_stmts): Likewise.
10620 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
10621 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
10622 * tree-ssa-sccvn.h: Likewise.
10623 * tree-ssa-strlen.c (get_string_length): Likewise.
10624 (valid_builtin_call): Likewise.
10625 (adjust_last_stmt): Likewise.
10626 (handle_builtin_strchr): Likewise.
10627 (handle_builtin_strcpy): Likewise.
10628 (handle_builtin_stxncpy): Likewise.
10629 (handle_builtin_memcpy): Likewise.
10630 (handle_builtin_strcat): Likewise.
10631 (strlen_check_and_optimize_stmt): Likewise.
10632 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
10633 * tree-streamer-in.c: Likewise.
10634 * tree-streamer.c (record_common_node): Likewise.
10635 * tree.c (tree_code_size): Likewise.
10636 (wide_int_to_tree_1): Likewise.
10637 (type_contains_placeholder_1): Likewise.
10638 (build_common_tree_nodes): Likewise.
10639 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
10640 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
10641 (POINTER_BOUNDS_P): Likewise.
10642 (BOUNDED_TYPE_P): Likewise.
10643 (BOUNDED_P): Likewise.
10644 (CALL_WITH_BOUNDS_P): Likewise.
10645 (pointer_bounds_type_node): Likewise.
10646 * value-prof.c (gimple_ic): Likewise.
10647 * var-tracking.c (vt_add_function_parameters): Likewise.
10648 * varasm.c (make_decl_rtl): Likewise.
10649 (assemble_start_function): Likewise.
10650 (output_constant): Likewise.
10651 (maybe_assemble_visibility): Likewise.
10652 * varpool.c (ctor_for_folding): Likewise.
10653 * chkp-builtins.def: Remove.
10654 * ipa-chkp.c: Remove.
10655 * ipa-chkp.h: Remove.
10656 * rtl-chkp.c: Remove.
10657 * rtl-chkp.h: Remove.
10658 * tree-chkp-opt.c: Remove.
10659 * tree-chkp.c: Remove.
10660 * tree-chkp.h: Remove.
10662 2018-06-07 Carl Love <cel@us.ibm.com>
10664 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10665 vextract_fp_from_shortl): Add BE support.
10667 2018-06-07 Paul Koning <ni1d@arrl.net>
10669 * compare-elim.c (try_merge_compare): Don't merge compare if
10670 address contains a side effect.
10671 (try_eliminate_compare): Likewise.
10673 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
10675 * config.gcc: Support "tremont".
10676 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
10677 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10679 * config/i386/i386.c (m_TREMONT): Define.
10680 (processor_target_table): Add "tremont".
10681 (PTA_TREMONT): Define.
10682 (ix86_lea_outperforms): Add TARGET_TREMONT.
10683 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
10684 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
10685 and M_INTEL_GOLDMONT_PLUS.
10686 (fold_builtin_cpu): Add "tremont".
10687 (ix86_add_stmt_cost): Add TARGET_TREMONT.
10688 (ix86_option_override_internal): Add "tremont".
10689 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
10690 (processor_type): Add PROCESSOR_TREMONT.
10691 * config/i386/x86-tune.def: Add m_TREMONT.
10692 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
10694 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10696 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
10697 symbol defined for msp430i* devices to be lower case.
10699 2018-06-07 Richard Biener <rguenther@suse.de>
10701 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
10702 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
10703 Properly wrap signed arithmetic if overflow wraps.
10705 2018-06-07 Jakub Jelinek <jakub@redhat.com>
10707 PR tree-optimization/69615
10708 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
10709 of a cast from a same precision integral SSA_NAME in a bb dominated
10710 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
10711 cast to utype if rhs2 has already a compatible type.
10713 2018-06-07 Richard Biener <rguenther@suse.de>
10715 PR tree-optimization/85935
10716 * graphite-scop-detection.c (find_params_in_bb): Analyze
10717 condition operands with respect to the correct loop. Assert
10718 the analysis doesn't fail.
10720 2018-06-04 Carl Love <cel@us.ibm.com>
10722 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
10723 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
10724 as it is slightly cheaper.
10725 (first_match_or_eos_index_<mode>):
10726 Calculate index using natural element order.
10727 (first_match_index_<mode>):
10728 Calculate index using natural element order.
10729 (first_match_or_eos_index_<mode>):
10730 Calculate index using natural order.
10731 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
10732 for BE and LE modes.
10733 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
10734 P9V_BUILTIN_VCLZLSBB_V16QI.
10735 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
10738 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10740 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
10741 indentation and line wrap for many prototypes. Add missing
10742 @smallexample directives around block of prototypes for vec_xl and
10745 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
10747 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
10748 track if we pass or return IEEE 128-bit floating point.
10749 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
10750 C++ mangling that is compatible with GCC 8.1.
10751 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
10752 (init_cumulative_args): Note if we pass or return IEEE 128-bit
10753 floating point types.
10754 (rs6000_function_arg_advance_1): Likewise.
10755 (rs6000_mangle_type): Optionally generate mangled names that match
10756 what GCC 8.1 generated for IEEE 128-bit floating point types.
10757 (rs6000_globalize_decl_name): If we have an external function that
10758 passes or returns IEEE 128-bit types, generate a weak reference
10759 from the mangled name used in GCC 8.1 to the current mangled
10761 (rs6000_init_builtins): Make __ibm128 use the long double type if
10762 long double is IBM extended double. Make __float128 use the long
10763 double type if long double is IEEE 128-bit.
10766 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
10767 macro for __ibm128 built-in functions.
10768 (PACK_IF): Add __ibm128 pack/unpack functions.
10769 (UNPACK_IF): Likewise.
10770 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
10771 enable long double built-in functions if long double is IEEE
10772 128-bit floating point.
10773 (rs6000_invalid_builtin): Update long double built-in function
10775 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
10776 functions, adjust the built-in function to use the long double
10777 built-in function if __ibm128 and long double are the same type.
10778 * doc/extend.texi (PowerPC builtins): Update documention for
10779 __builtin_{,un}pack_longdouble. Add documentation for
10780 __builtin_{,un}pack_ibm128.
10782 2018-06-06 Jim Wilson <jimw@sifive.com>
10784 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
10785 (struct machine_function): New field interrupt_mode.
10786 (riscv_handle_type_attribute): New function. Add forward declaration.
10787 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
10788 (riscv_expand_epilogue): Check interrupt_mode field.
10789 (riscv_set_current_function): Check interrupt attribute args and
10790 set interrupt_mode field.
10791 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
10792 (riscv_sret, riscv_uret): New.
10793 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
10794 new arguments to interrupt attribute.
10796 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
10799 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
10800 Don't handle -mcpu=power8 if -mpower9-vector is also used.
10802 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10804 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
10805 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
10806 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10807 several redundant entries.
10809 2018-06-06 David Malcolm <dmalcolm@redhat.com>
10811 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
10812 type from "rtx" to "rtx_insn *".
10813 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
10814 for local "call_insn", removing cast.
10815 (ix86_expand_call): Likewise, introducing a "call_insn" local.
10817 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
10819 PR tree-optimization/86066
10820 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
10821 for BIT_INSERT_EXPR stores.
10823 2018-06-06 Richard Biener <rguenther@suse.de>
10825 PR tree-optimization/86062
10826 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
10827 component refs ontop
10828 of to be offsetted base.
10830 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10832 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
10833 to be static and remove check on interrupt attribute name.
10835 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
10837 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10838 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
10840 2018-06-05 Steve Ellcey <sellcey@cavium.com>
10843 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
10845 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
10846 Remove second argument, change how error is called.
10847 (aarch64_layout_arg): Remove second argument from
10848 aarch64_err_no_fpadvsimd call.
10849 (aarch64_init_cumulative_args): Ditto.
10850 (aarch64_gimplify_va_arg_expr): Ditto.
10851 * config/aarch64/aarch64.md (mov<mode>): Ditto.
10853 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
10855 * config/i386/i386.md (simple_return_indirect_internal): New expander.
10856 (*simple_return_indirect_internal<mode>): Rename from
10857 simple_return_indirect_internal. Use W mode iterator.
10858 (rstorssp): New expander.
10859 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
10860 (clrssbsy): New expander.
10861 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
10863 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
10865 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
10867 (cmse_check_pointed_object): Likewise.
10869 2018-06-05 Martin Liska <mliska@suse.cz>
10871 PR gcov-profile/47618
10872 * doc/invoke.texi: Document how -fprofile-dir format
10875 2018-06-05 Richard Biener <rguenther@suse.de>
10877 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
10878 removal pretend DOM info isn't available so we do not update
10879 it and only remove edges, not dominated blocks. Actually free
10880 DOM info in case we removed something. Remove unreachable blocks.
10881 (mfb_keep_latches): Work with either DOM info or marked backedges.
10882 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
10883 first. Mark backedges if DOM info isn't available.
10884 (Re-)compute DOM info after cleanup_control_flow_pre.
10886 2018-06-05 Richard Biener <rguenther@suse.de>
10888 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
10889 (locus_discrim_hasher::hash): Adjust.
10890 (locus_discrim_hasher::equal): Likewise.
10891 (next_discriminator_for_locus): Work on line directly.
10892 (same_line_p): Pass in expanded locus1 as well.
10893 (assign_discriminators): Avoid redundant location expansions.
10895 2018-06-05 Richard Biener <rguenther@suse.de>
10897 PR tree-optimization/86046
10898 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
10899 if required after clearing TREE_ADDRESSABLE.
10901 2018-06-05 Richard Biener <rguenther@suse.de>
10903 PR tree-optimization/86047
10904 * tree-ssa-loop.c (for_each_index): Glob handling of all
10905 decls and constants and really handle all of them.
10907 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10910 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
10911 qualifier_void_pointer and qualifier_const_void_pointer.
10912 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
10913 (arm_init_builtins): Handle the above.
10914 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
10915 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
10916 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
10919 2018-06-05 Martin Liska <mliska@suse.cz>
10921 * auto-profile.c (read_autofdo_file): Do not use
10922 gcov_ctr_summary struct.
10923 (afdo_callsite_hot_enough_for_early_inline): Likewise.
10924 * coverage.c (struct counts_entry): Likewise.
10925 (read_counts_file): Read just single summary entry.
10926 (get_coverage_counts): Use gcov_summary struct.
10927 * coverage.h (get_coverage_counts): Likewise.
10928 * gcov-dump.c (dump_working_sets): Likewise.
10929 (tag_summary): Dump just single summary.
10930 * gcov-io.c (gcov_write_summary): Write just histogram
10932 (gcov_read_summary): Read just single summary.
10933 (compute_working_sets): Use gcov_summary struct.
10934 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
10935 of GCOV_COUNTERS_SUMMABLE.
10936 (GCOV_COUNTERS_SUMMABLE): Remove.
10937 (GCOV_FIRST_VALUE_COUNTER): Replace with
10938 GCOV_COUNTER_V_INTERVAL.
10939 (struct gcov_ctr_summary): Remove.
10940 (struct gcov_summary): Directly use fields of former
10942 (compute_working_sets): Use gcov_summary struct.
10943 * gcov.c (read_count_file): Do not use ctrs fields.
10944 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
10946 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
10948 * profile.c: Likewise.
10949 * profile.h: Likewise.
10951 2018-06-05 Martin Liska <mliska@suse.cz>
10953 PR gcov-profile/84846
10954 * gcov.c (output_lines): Print working directory only
10955 in intermediate format.
10957 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
10959 * config/s390/s390-builtin-types.def: Add void function type.
10960 * config/s390/s390-builtins.def: Use the function type for the
10963 2018-06-04 Jim Wilson <jimw@sifive.com>
10965 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
10967 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
10968 and maybe_eh_return. Change regno to unsigned int. Use new args to
10969 handle EH_RETURN_DATA_REGNO registers properly.
10970 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
10971 (riscv_expand_epilogue): Update comment. Change argument name and
10972 type. Update code to use new name and type. Pass new args to
10973 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
10975 * config/riscv/riscv.md (NORMAL_RETURN): New.
10976 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
10977 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
10978 (eh_return): Call gen_eh_return_internal and emit barrier.
10979 (eh_return_internal): Call riscv_expand_epilogue.
10981 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
10983 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
10984 bit_insertion field and declare can_be_merged_into method.
10985 (merged_store_group::can_be_merged_into): New method.
10986 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
10987 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
10988 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
10990 2018-06-04 Richard Biener <rguenther@suse.de>
10992 PR tree-optimization/85955
10993 * builtins.c (fold_builtin_sincos): Convert pointers to
10994 destination to appropriate type before dereferencing.
10996 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
10998 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
11000 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
11002 * expr.c (expand_expr_real_1): Force the operand into memory if
11003 its TYPE_MODE is BLKmode and if there is no integer mode for
11004 the number of bits being extracted.
11006 2018-06-04 Jakub Jelinek <jakub@redhat.com>
11010 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
11011 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
11013 2018-06-04 Richard Biener <rguenther@suse.de>
11015 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
11016 (cleanup_tree_cfg_noloop): ... single caller. Do
11017 start_recording_case_labels later.
11019 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
11021 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
11022 to _IMMINTRIN_H_INCLUDED.
11023 * config/i386/pconfigintrin.h: Ditto.
11024 * config/i386/waitpkgintrin.h: Ditto.
11025 * config/i386/immintrin.h: Add includes for sgxintrin.h,
11026 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
11027 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
11028 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
11029 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
11030 waitpkgintrin.h and cldemoteintrin.h.
11032 2018-06-04 Richard Biener <rguenther@suse.de>
11034 PR tree-optimization/86038
11035 * tracer.c (find_best_successor): Check probability for
11036 being initialized, bail out if not.
11038 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
11041 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
11042 of bits to ignore when comparing architectures.
11044 2018-06-04 Jakub Jelinek <jakub@redhat.com>
11046 PR tree-optimization/69615
11047 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
11048 maximum or minimum of the type, try to merge it also as if
11049 range1 is + [-, x - 1] or + [x + 1, -].
11052 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
11054 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
11056 PR tree-optimization/86034
11057 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
11058 the unsigned bitfield type in a bit insertion sequence if it does not
11059 have a larger precision than the bitfield size.
11060 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
11062 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
11064 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
11066 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
11068 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
11069 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
11070 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
11071 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
11073 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
11075 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
11076 Disable -fdelete-null-pointer-checks for ELF toolchain.
11078 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
11079 Kito Cheng <kito.cheng@gmail.com>
11081 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
11082 (nds32le-*-*, nds32be-*-*): Integrate checking process.
11083 (nds32*-*-*): Add glibc and uclibc conditions.
11084 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
11085 (TARGET_EXCEPT_UNWIND_INFO): Define.
11086 * config/nds32/elf.h: New file.
11087 * config/nds32/linux.h: New file.
11088 * config/nds32/nds32-elf.opt: New file.
11089 * config/nds32/nds32-linux.opt: New file.
11090 * config/nds32/nds32-fp-as-gp.c
11091 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
11092 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
11094 (nds32_asm_file_end): Ditto.
11095 (nds32_print_operand): Ditto.
11096 (nds32_insert_attributes): Ditto.
11097 (nds32_init_libfuncs): New function.
11098 (TARGET_HAVE_TLS): Define.
11099 (TARGET_INIT_LIBFUNCS): Define.
11100 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
11102 (TARGET_ELF): Apply different mcmodel setting.
11103 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
11104 been migrated into elf.h and linux.h files.
11105 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
11106 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
11107 (mcmodel): The content has been migrated into nds32-elf.opt and
11108 nds32-linux.opt files.
11109 * config/nds32/t-elf: New file.
11110 * config/nds32/t-linux: New file.
11112 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
11113 Shiva Chen <shiva0217@gmail.com>
11115 * config/nds32/constants.md (unspec_volatile_element): Add
11116 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
11117 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
11119 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
11120 (make_pass_nds32_fp_as_gp): Declare.
11121 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
11123 (nds32_asm_function_end_prologue): Remove unused asm output.
11124 (nds32_asm_function_begin_epilogue): Remove unused asm output.
11125 (nds32_asm_file_start): Output necessary fp_as_gp information.
11126 (nds32_option_override): Adjust register usage.
11127 (nds32_expand_prologue): Consider fp_as_gp situation.
11128 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
11129 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
11132 (simple_return): Ditto.
11133 (omit_fp_begin): Output special directive for fp_as_gp.
11134 (omit_fp_end): Output special directive for fp_as_gp.
11135 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
11136 mforbid-fp-as-gp): New options.
11138 2018-06-01 Mark Wielaard <mark@klomp.org>
11140 * dwarf2out.c (dwarf2out_finish): Remove generation of
11141 DW_AT_loclists_base.
11143 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
11145 * gimple-ssa-store-merging.c: Include gimple-fold.h.
11146 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
11147 (struct merged_store_group): Add bit_insertion field.
11148 (dump_char_array): Use standard hexadecimal format.
11149 (merged_store_group::merged_store_group): Set bit_insertion to false.
11150 (merged_store_group::apply_stores): Use optimal buffer size. Deal
11151 with BIT_INSERT_EXPR stores. Move up code updating the mask and
11152 also print the mask in the dump file.
11153 (pass_store_merging::gate): Minor tweak.
11154 (imm_store_chain_info::coalesce_immediate): Fix wrong association
11155 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
11156 stores with INTEGER_CST stores.
11157 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
11158 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
11159 and use it throughout. Generate bit insertion sequences if need be.
11160 (pass_store_merging::process_store): Remove redundant condition.
11161 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
11163 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
11165 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
11166 the 128-bit floating point types. Fix function comment.
11168 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11170 * config/aarch64/aarch64-simd.md
11171 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
11173 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
11176 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
11178 PR tree-optimization/85989
11179 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
11181 (backprop::intersect_uses): Check it when deciding whether this
11182 is a backedge reference.
11183 (backprop::process_block): Add each phi to m_visited_phis
11184 after visiting it, then clear it at the end.
11186 2018-06-01 Richard Biener <rguenther@suse.de>
11188 * tree-vectorizer.h (vect_dr_stmt): New function.
11189 (vect_get_load_cost): Adjust.
11190 (vect_get_store_cost): Likewise.
11191 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
11192 Use vect_dr_stmt instead of DR_SMTT.
11193 (vect_record_base_alignments): Likewise.
11194 (vect_calculate_target_alignment): Likewise.
11195 (vect_compute_data_ref_alignment): Likewise and make static.
11196 (vect_update_misalignment_for_peel): Likewise.
11197 (vect_verify_datarefs_alignment): Likewise.
11198 (vector_alignment_reachable_p): Likewise.
11199 (vect_get_data_access_cost): Likewise. Pass down
11200 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
11201 (vect_get_peeling_costs_all_drs): Likewise.
11202 (vect_peeling_hash_get_lowest_cost): Likewise.
11203 (vect_enhance_data_refs_alignment): Likewise.
11204 (vect_find_same_alignment_drs): Likewise.
11205 (vect_analyze_data_refs_alignment): Likewise.
11206 (vect_analyze_group_access_1): Likewise.
11207 (vect_analyze_group_access): Likewise.
11208 (vect_analyze_data_ref_access): Likewise.
11209 (vect_analyze_data_ref_accesses): Likewise.
11210 (vect_vfa_segment_size): Likewise.
11211 (vect_small_gap_p): Likewise.
11212 (vectorizable_with_step_bound_p): Likewise.
11213 (vect_prune_runtime_alias_test_list): Likewise.
11214 (vect_analyze_data_refs): Likewise.
11215 (vect_supportable_dr_alignment): Likewise.
11216 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
11217 (vect_gen_prolog_loop_niters): Likewise.
11218 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11219 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
11221 (vect_recog_mask_conversion_pattern): Likewise.
11222 (vect_try_gather_scatter_pattern): Likewise.
11223 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
11224 to vect_get_store_cost.
11225 (vect_get_store_cost): Get stmt_info instead of DR.
11226 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
11227 (vect_get_load_cost): Get stmt_info instead of DR.
11229 2018-06-01 Richard Biener <rguenther@suse.de>
11231 PR middle-end/86017
11232 * gimple-fold.c (var_decl_component_p): Also allow offsetted
11233 vars wrapped in MEM_REFs.
11235 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
11237 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11238 Fix subreg tests so that we only return a choice between
11239 GENERAL_REGS and FP_REGS if the original classes included both.
11241 2018-06-01 Richard Biener <rguenther@suse.de>
11244 * tree-ssa-structalias.c (get_function_part_constraint):
11245 Handle NULL fi->decl.
11246 (find_func_aliases_for_call): Properly handle indirect
11247 fi from direct call.
11248 (find_func_clobbers): Likewise.
11249 (ipa_pta_execute): Likewise.
11250 (create_variable_info_for): For functions that are ifunc_resolver
11251 resolve to a varinfo that contains the result of the resolver call.
11252 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
11255 2018-05-31 Michael Collison <michael.collison@arm.com>
11257 * config/aarch64/aarch64.md:
11258 (*fix_to_zero_extenddfdi2): New pattern.
11259 * gcc.target/aarch64/fix_extend1.c: New testcase.
11261 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
11263 PR middle-end/78809
11264 PR middle-end/83026
11265 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
11266 and BUILT_IN_STRNCMP_EQ.
11267 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
11268 BUILT_IN_STRNCMP_EQ.
11269 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
11270 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
11271 (gimple_fold_builtin): Likewise.
11272 * tree-ssa-strlen.c (compute_string_length): New function.
11273 (determine_min_obsize): New function.
11274 (handle_builtin_string_cmp): New function to handle calls to
11275 string compare functions.
11276 (strlen_optimize_stmt): Add handling to builtin string compare
11278 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11279 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
11280 * tree.c (build_common_builtin_nodes): Add new defines of
11281 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
11283 2018-05-31 Jakub Jelinek <jakub@redhat.com>
11286 * bb-reorder.c (pass_partition_blocks::gate): Return false for
11287 functions with naked attribute.
11289 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
11291 * config/i386/sse.md (avx_vec_concat<mode>):
11292 Substitute concat_tg_mode mode attribute with xtg_mode.
11293 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
11294 (concat_tg_mode): Remove mode attribute.
11296 2018-05-31 Martin Sebor <msebor@redhat.com>
11299 * calls.c (alloc_max_size): Correct a logic error/typo.
11300 Treat excessive arguments as infinite. Warn for invalid arguments.
11301 * doc/invoke.texi (-Walloc-size-larger-than): Update.
11303 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
11306 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
11307 and movx for Haswell.
11309 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
11310 Cesar Philippidis <cesar@codesourcery.com>
11312 PR middle-end/85879
11313 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
11314 when emitting error on private/firstprivate reductions.
11315 * omp-low.c (lower_omp_target): Avoid reference-type processing
11316 on pointers for firstprivate clause.
11318 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
11320 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
11323 * config/aarch64/aarch64-simd.md
11324 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
11325 (aarch64_ld1_x3_<mode>): Likewise
11326 (aarch64_st1x2<VALLDIF:mode>): Likewise
11327 (aarch64_st1_x2_<mode>): Likewise
11328 (aarch64_st1x3<VALLDIF:mode>): Likewise
11329 (aarch64_st1_x3_<mode>): Likewise
11330 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
11331 (vld1_s8_x3): Likewise.
11332 (vld1_u16_x3): Likewise.
11333 (vld1_s16_x3): Likewise.
11334 (vld1_u32_x3): Likewise.
11335 (vld1_s32_x3): Likewise.
11336 (vld1_u64_x3): Likewise.
11337 (vld1_s64_x3): Likewise.
11338 (vld1_f16_x3): Likewise.
11339 (vld1_f32_x3): Likewise.
11340 (vld1_f64_x3): Likewise.
11341 (vld1_p8_x3): Likewise.
11342 (vld1_p16_x3): Likewise.
11343 (vld1_p64_x3): Likewise.
11344 (vld1q_u8_x3): Likewise.
11345 (vld1q_s8_x3): Likewise.
11346 (vld1q_u16_x3): Likewise.
11347 (vld1q_s16_x3): Likewise.
11348 (vld1q_u32_x3): Likewise.
11349 (vld1q_s32_x3): Likewise.
11350 (vld1q_u64_x3): Likewise.
11351 (vld1q_s64_x3): Likewise.
11352 (vld1q_f16_x3): Likewise.
11353 (vld1q_f32_x3): Likewise.
11354 (vld1q_f64_x3): Likewise.
11355 (vld1q_p8_x3): Likewise.
11356 (vld1q_p16_x3): Likewise.
11357 (vld1q_p64_x3): Likewise.
11358 (vst1_s64_x2): Likewise.
11359 (vst1_u64_x2): Likewise.
11360 (vst1_f64_x2): Likewise.
11361 (vst1_s8_x2): Likewise.
11362 (vst1_p8_x2): Likewise.
11363 (vst1_s16_x2): Likewise.
11364 (vst1_p16_x2): Likewise.
11365 (vst1_s32_x2): Likewise.
11366 (vst1_u8_x2): Likewise.
11367 (vst1_u16_x2): Likewise.
11368 (vst1_u32_x2): Likewise.
11369 (vst1_f16_x2): Likewise.
11370 (vst1_f32_x2): Likewise.
11371 (vst1_p64_x2): Likewise.
11372 (vst1q_s8_x2): Likewise.
11373 (vst1q_p8_x2): Likewise.
11374 (vst1q_s16_x2): Likewise.
11375 (vst1q_p16_x2): Likewise.
11376 (vst1q_s32_x2): Likewise.
11377 (vst1q_s64_x2): Likewise.
11378 (vst1q_u8_x2): Likewise.
11379 (vst1q_u16_x2): Likewise.
11380 (vst1q_u32_x2): Likewise.
11381 (vst1q_u64_x2): Likewise.
11382 (vst1q_f16_x2): Likewise.
11383 (vst1q_f32_x2): Likewise.
11384 (vst1q_f64_x2): Likewise.
11385 (vst1q_p64_x2): Likewise.
11386 (vst1_s64_x3): Likewise.
11387 (vst1_u64_x3): Likewise.
11388 (vst1_f64_x3): Likewise.
11389 (vst1_s8_x3): Likewise.
11390 (vst1_p8_x3): Likewise.
11391 (vst1_s16_x3): Likewise.
11392 (vst1_p16_x3): Likewise.
11393 (vst1_s32_x3): Likewise.
11394 (vst1_u8_x3): Likewise.
11395 (vst1_u16_x3): Likewise.
11396 (vst1_u32_x3): Likewise.
11397 (vst1_f16_x3): Likewise.
11398 (vst1_f32_x3): Likewise.
11399 (vst1_p64_x3): Likewise.
11400 (vst1q_s8_x3): Likewise.
11401 (vst1q_p8_x3): Likewise.
11402 (vst1q_s16_x3): Likewise.
11403 (vst1q_p16_x3): Likewise.
11404 (vst1q_s32_x3): Likewise.
11405 (vst1q_s64_x3): Likewise.
11406 (vst1q_u8_x3): Likewise.
11407 (vst1q_u16_x3): Likewise.
11408 (vst1q_u32_x3): Likewise.
11409 (vst1q_u64_x3): Likewise.
11410 (vst1q_f16_x3): Likewise.
11411 (vst1q_f32_x3): Likewise.
11412 (vst1q_f64_x3): Likewise.
11413 (vst1q_p64_x3): Likewise.
11415 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11417 * config/msp430/msp430.c (msp430_output_labelref): Prepend
11418 user_label_prefix to name.
11420 * tree-core.h: Update comment about the format of NAME string
11421 passed to handler in attribute_spec.
11423 * config/msp430/msp430.md: Remove erroneous subreg expression from
11424 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
11425 zero_extend{q,h}isi2.
11427 2018-05-30 Borislav Petkov <bp@suse.de>
11429 * doc/extend.texi: Document some architecture specific
11430 constraints and sort entries.
11432 2018-05-30 Martin Sebor <msebor@redhat.com>
11434 PR middle-end/85369
11435 * builtins.c (expand_builtin_stpcpy_1): New function.
11436 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
11437 only if the former succeeds.
11439 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
11441 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
11444 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11446 * doc/invoke.texi (-flinker-output): Document
11448 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11450 * passes.c (ipa_write_summaries): Only modify statements if body
11452 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
11453 incrementally linking.
11454 (ipa_passes): Likewise.
11455 * lto-cgraph.c (lto_output_node): When incrementally linking do not
11456 pass down resolution info.
11457 * common.opt (flag_incremental_link): Update info.
11458 * gcc.c (plugin specs): Turn flinker-output=* to
11459 -plugin-opt=-linker-output-known
11460 * toplev.c (compile_file): Also cut compilation when doing incremental
11462 * flag-types. (enum lto_partition_model): Add
11463 LTO_LINKER_OUTPUT_NOLTOREL.
11464 (invoke.texi): Add -flinker-output docs.
11465 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
11466 link same way as WPA; do not stream in dead initializers.
11468 * dwarf2out.c (dwarf2out_die_ref_for_decl,
11469 darf2out_register_external_decl): Support incremental link.
11471 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11473 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
11475 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11477 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
11478 it down to simple_object_copy_lto_debug_sections.
11479 (run_gcc): Determine incremental LTO link time and configure
11480 lto1 into non-wpa mode, disable renaming of debug sections.
11482 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
11484 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
11485 descriptions of various incorrectly documented functions.
11487 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
11490 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11491 address check not strict.
11493 2018-05-30 Richard Biener <rguenther@suse.de>
11495 PR tree-optimization/85964
11496 * tracer.c (better_p): Drop initialized count check, we only
11497 call the function with initialized counts now.
11498 (find_best_successor): Do find a best edge if one
11499 has uninitialized count.
11500 (find_best_predecessor): Likewise. Do BB frequency check only
11501 if count is initialized.
11503 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
11505 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
11506 (aarch64_ldrstr_offset_compare): New.
11507 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
11508 load/store orderings.
11509 (aarch64_gen_adjusted_ldpstp): Likewise.
11511 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11513 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11514 Check for subset of GENERAL_REGS and FP_REGS.
11515 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
11518 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
11520 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
11521 and wi::to_poly_offset. Add the current offset and then check
11522 whether the sum fits, rather than using an unchecked addition of
11523 a checked term. Check for a shwi rather than a uhwi.
11524 * expr.c (get_bit_range): Use tree_to_poly_uint64.
11525 (store_constructor): Use poly_int_tree_p.
11526 (expand_expr_real_1): Likewise.
11527 * function.c (assign_temp): Likewise.
11528 * fold-const.c (const_binop): Use poly_int_tree_p and
11529 wi::to_poly_offset.
11530 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
11532 * ipa-icf-gimple.c (func_checker::compare_operand): Use
11533 to_poly_offset for MEM offsets.
11534 * ipa-icf.c (sem_variable::equals): Likewise.
11535 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
11536 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
11537 wi::to_poly_offset for BIT_FIELD_REF offsets.
11538 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
11539 wi::to_poly_offset.
11540 * var-tracking.c (emit_note_insn_var_location): Use
11541 tree_to_poly_uint64.
11543 2018-05-29 Jim Wilson <jimw@sifive.com>
11545 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
11547 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
11550 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
11551 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
11553 (sse4_1_round<mode>2): Use nonimmediate_operand
11554 for operand 1 predicate.
11556 2018-05-29 Martin Sebor <msebor@redhat.com>
11557 Richard Biener <rguenther@suse.de>
11560 * calls.c (get_size_range): Call determine_value_range instead
11561 of get_value_range..
11562 * tree-vrp.h (determine_value_range): Declared new function.
11563 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
11565 2018-05-29 Richard Biener <rguenther@suse.de>
11567 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
11568 sure to use non-pattern stmts for get_earlier_stmt arguments.
11569 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
11570 called on pattern stmts.
11571 (get_later_stmt): Likewise.
11573 2018-05-29 Martin Liska <mliska@suse.cz>
11575 PR gcov-profile/85759
11576 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
11579 2018-05-29 Jakub Jelinek <jakub@redhat.com>
11581 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
11583 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
11587 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
11588 VEC_PACK_FLOAT_EXPR): New tree codes.
11589 * tree-pretty-print.c (op_code_prio): Handle
11590 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
11591 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
11592 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
11593 * tree-inline.c (estimate_operator_cost): Likewise.
11594 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
11595 * fold-const.c (const_binop): Likewise.
11596 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
11597 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
11598 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
11599 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
11600 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
11601 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
11602 * expr.c (expand_expr_real_2): Likewise.
11603 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
11604 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
11605 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
11607 * optabs.c (expand_widen_pattern_expr): For
11608 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
11609 sign from result type rather than operand's type.
11610 (expand_binop_directly): For vec_packu_float_optab and
11611 vec_packs_float_optab allow result type to be different from operand's
11613 * optabs-tree.c (optab_for_tree_code): Handle
11614 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
11615 VEC_PACK_FLOAT_EXPR. Formatting fixes.
11616 * tree-vect-generic.c (expand_vector_operations_1): Handle
11617 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
11618 VEC_PACK_FLOAT_EXPR.
11619 * tree-vect-stmts.c (supportable_widening_operation): Handle
11621 (supportable_narrowing_operation): Handle FLOAT_EXPR.
11622 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
11623 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
11624 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
11625 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
11627 (vec_pack<floatprefix>_float_<mode>): New expander.
11628 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
11630 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
11631 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
11632 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
11633 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
11634 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
11636 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
11637 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
11638 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
11639 VEC_PACK_FLOAT_EXPR): Document.
11641 2018-05-29 Richard Biener <rguenther@suse.de>
11643 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
11645 (stmt_vec_info_vec): Make pointer.
11646 (init_stmt_vec_info_vec): Remove.
11647 (free_stmt_vec_info_vec): Likewise.
11648 (set_stmt_vec_info_vec): New function.
11649 (free_stmt_vec_infos): Likewise.
11650 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
11651 (set_vinfo_for_stmt): Likewise.
11652 (get_earlier_stmt): Likewise.
11653 (get_later_stmt): Likewise.
11654 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
11655 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
11656 (vec_info::~vec_info): Free stmt_vec_infos.
11657 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
11658 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
11659 (pass_slp_vectorize::execute): Likewise.
11660 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
11661 (free_stmt_vec_info_vec): Likewise.
11662 (set_stmt_vec_info_vec): New function.
11663 (free_stmt_vec_infos): Likewise.
11664 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
11665 the global stmt_vec_info_vec.
11666 * tree-parloops.c (gather_scalar_reductions): Use
11667 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
11670 2018-05-29 Richard Biener <rguenther@suse.de>
11672 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
11674 2018-05-29 Martin Liska <mliska@suse.cz>
11675 David Malcolm <dmalcolm@redhat.com>
11677 * vec.c (test_reverse): New.
11678 (vec_c_tests): Add new test.
11679 * vec.h (vl_ptr>::reverse): New function.
11681 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
11683 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
11685 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
11688 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
11690 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
11692 2018-05-28 Richard Biener <rguenther@suse.de>
11694 PR tree-optimization/85933
11695 * tree-vect-data-refs.c (vect_record_base_alignments): Only
11696 look at stmts marked as vectorizable.
11698 2018-05-28 Richard Biener <rguenther@suse.de>
11700 PR tree-optimization/85934
11701 * tree-vect-generic.c (expand_vector_operations_1): Hoist
11702 vector boolean check before scalar optimization.
11704 2018-05-28 Jakub Jelinek <jakub@redhat.com>
11706 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
11709 2018-05-28 Mark Wielaard <mark@klomp.org>
11711 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
11712 if it is an expression containing a minus sign.
11714 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
11716 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
11718 2018-05-27 Paul Koning <ni1d@arrl.net>
11720 * config/pdp11/pdp11.md (truncsihi2): Remove.
11722 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
11723 Chung-Ju Wu <jasonwucj@gmail.com>
11725 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
11727 (unaligned_store_dw): Ditto.
11728 * config/nds32/nds32-memory-manipulation.c
11729 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
11730 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
11731 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
11732 (emit_setmem_word_loop): Rename to ...
11733 (emit_setmem_doubleword_loop): ... this.
11734 (nds32_gen_dup_4_byte_to_word_value): New function.
11735 (nds32_gen_dup_8_byte_to_double_word_value): New function.
11736 (nds32_expand_setmem_loop): Refine implementation.
11737 (nds32_expand_setmem_loop_v3m): Ditto.
11738 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
11741 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11743 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
11745 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11747 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
11748 (nds32_init_machine_status): Initialize machine->attr_naked_p and
11749 machine->attr_no_prologue_p.
11750 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
11751 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
11752 (nds32_expand_epilogue): Consider attr_naked_p.
11753 (nds32_expand_epilogue_v3pop): Likewise.
11754 (nds32_can_use_return_insn): Likewise.
11755 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
11756 attr_no_prologue_p fields.
11757 * config/nds32/nds32.opt (mret-in-naked-func): New option.
11759 2018-05-27 Jakub Jelinek <jakub@redhat.com>
11762 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
11764 * config/i386/sse.md
11765 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
11767 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
11769 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
11771 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
11773 (*<floatsuffix>floatv2div2sf2): Rename to ...
11774 (*float<floatunssuffix>v2div2sf2): ... this.
11775 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
11776 (float<floatunssuffix>v2div2sf2_mask): ... this.
11777 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
11778 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
11779 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
11781 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
11783 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11785 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11787 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11789 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11791 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
11792 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
11793 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
11794 gen_ufix_truncv8dfv8si2.
11795 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
11796 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
11797 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
11798 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
11799 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
11800 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
11801 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
11802 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
11804 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11808 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
11810 2018-05-25 Jim Wilson <jimw@sifive.com>
11812 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
11813 * config/riscv/riscv.c (struct machine_function): Add
11814 interrupt_handler_p and attribute_checked_p fields.
11815 (riscv_attribute_table): Add interrupt.
11816 (riscv_interrupt_type_p): New.
11817 (riscv_save_reg_p): Save extra regs for interrupt handler.
11818 (riscv_use_save_libcall): Return false for interrupt handler.
11819 (riscv_first_stack_step): Add forward declaration.
11820 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
11821 for interrupt handler with large frame. Use it for saved reg list.
11822 (riscv_expand_prologue): Move flag_stack_usage_info support to
11823 eliminate duplication.
11824 (riscv_expand_epilogue): Generate mret for interrupt handler.
11825 (riscv_epilogue_uses): New.
11826 (riscv_can_use_return_insn): Return false for interrupt handler.
11827 (riscv_function_ok_for_sibcall): Likewise.
11828 (riscv_set_current_function): Add interrupt handler support.
11829 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
11830 * config/riscv/riscv.md (UNSPECV_MRET): New.
11832 (riscv_frflags, riscv_fsflags): Use tab after opcode.
11834 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
11836 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
11838 PR tree-optimization/85712
11839 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
11840 this candidate has already been replaced in-situ by a copy.
11842 2018-05-25 Jason Merrill <jason@redhat.com>
11844 PR c++/80485 - inline function non-zero address.
11845 * symtab.c (nonzero_address): Check DECL_COMDAT.
11847 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
11850 * config/alpha/alpha.md (ashlsi3): New insn pattern.
11851 (*ashlsi_se): Rename from *ashldi_se. Define as sign
11852 extension of SImode operation. Use const123_operand predicate.
11853 (*saddsi_1): Remove.
11854 (*saddl_se_1): Ditto.
11855 (*ssubsi_1): Ditto.
11856 (*ssubl_se_1): Ditto.
11857 * config/alpha/predicates.md (const123_operand): New predicate.
11858 * config/alpha/constraints.md (P): Use IN_RANGE.
11860 2018-05-25 Richard Biener <rguenther@suse.de>
11862 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
11864 (ref_maybe_used_by_stmt_p): Likewise.
11865 (stmt_may_clobber_ref_p): Likewise.
11866 (stmt_may_clobber_ref_p_1): Likewise.
11867 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
11869 (ref_maybe_used_by_stmt_p): Likewise.
11870 (stmt_may_clobber_ref_p): Likewise.
11871 (stmt_may_clobber_ref_p_1): Likewise.
11872 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
11873 the alias oracle to disambiguate DRs with stmts DR analysis
11875 (vect_analyze_data_refs): Do not give up on not analyzable
11876 DRs for BB vectorization. Remove code truncating the dataref
11879 2018-05-25 Jakub Jelinek <jakub@redhat.com>
11882 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
11883 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
11884 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
11886 2018-05-25 Richard Biener <rguenther@suse.de>
11888 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
11889 function, combining stmt data ref gathering and fatal analysis
11891 (vect_analyze_data_refs): Remove now redudnant code and simplify.
11892 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
11893 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
11894 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
11895 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
11897 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11899 PR tree-optimization/85720
11900 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
11901 SCC if all partitions are builtins.
11902 (version_loop_by_alias_check): New parameter. Generate cancelable
11903 runtime alias check if all partitions are builtins.
11904 (distribute_loop): Update call to above function.
11906 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11908 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
11909 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
11910 (parm_default_def_partition_arg): Ditto.
11911 (set_parm_default_def_partition): Ditto.
11912 (get_parm_default_def_partitions): Ditto and make it static.
11913 (get_undefined_value_partitions): Ditto and make it static.
11914 (remove_ssa_form): Refactor call to init_var_map here.
11915 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
11916 computation for loop region.
11917 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
11918 (register_default_def): Delete.
11919 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
11920 (parm_default_def_partition_arg): Ditto.
11921 (set_parm_default_def_partition): Ditto.
11922 (get_parm_default_def_partitions): Ditto and make it static.
11923 (get_undefined_value_partitions): Ditto and make it static.
11924 (coalesce_with_default, coalesce_with_default): Update comment.
11925 (create_coalesce_list_for_region): New func factored out from
11926 create_outofssa_var_map.
11927 (populate_coalesce_list_for_outofssa): New func factored out from
11928 create_outofssa_var_map and coalesce_ssa_name.
11929 (create_outofssa_var_map): Delete.
11930 (coalesce_ssa_name): Refactor to support live range computation.
11931 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
11932 (get_parm_default_def_partitions): Delete.
11933 (get_undefined_value_partitions): Ditto.
11934 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
11935 computation for loop region.
11936 (new_tree_live_info, loe_visit_block): Ditto.
11937 (live_worklist, set_var_live_on_entry): Ditto.
11938 (calculate_live_on_exit, verify_live_on_entry): Ditto.
11939 * tree-ssa-live.h (struct _var_map): New fields.
11940 (init_var_map): Change decl.
11941 (region_contains_p): New.
11943 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11945 * tree-ssa-live.h (live_merge_and_clear): Delete.
11947 2018-05-25 Richard Biener <rguenther@suse.de>
11950 * tree-dump.c (dequeue_and_dump): Remove access to removed
11951 operand 2 of a SWITCH_EXPR.
11953 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11955 * doc/sourcebuild.texi (vect_double_cond_arith): Include
11956 multiplication and division.
11957 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
11958 (cond_udiv@var{m}, cond_umod@var{m}): Document.
11959 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
11960 (cond_udiv_optab, cond_umod_optab): New optabs.
11961 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
11962 (IFN_COND_RDIV): New internal functions.
11963 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
11964 TRUNC_MOD_EXPR and RDIV_EXPR.
11965 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
11966 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
11968 (SVE_INT_BINARY): Include mult.
11969 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
11970 (optab, sve_int_op): Handle mult.
11971 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
11973 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
11974 for SVE_INT_BINARY_SD.
11976 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11978 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
11979 (optab, sve_int_op): Handle div and udiv.
11980 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
11981 for SVE_INT_BINARY_SD.
11982 (*<optab><mode>3): New insn for the same.
11984 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11986 * tree-vect-patterns.c: Include predict.h.
11987 (vect_recog_divmod_pattern): Restrict check for division support
11988 to when optimizing for size.
11990 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11992 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
11993 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
11994 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
11995 (gimple_match_op::set_op): Likewise.
11996 (gimple_resimplify4): Declare.
11997 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
11998 (expr::gen_transform): Likewise.
11999 (decision_tree::gen): Generate a simplification routine for 4 operands.
12000 * gimple-match-head.c (gimple_simplify): Add an overload for
12001 4 operands. In the top-level function, handle up to 4 call
12002 arguments and call gimple_resimplify4.
12003 (gimple_resimplify4): New function.
12004 (build_call_internal): Pass a fourth operand.
12005 (maybe_push_to_seq): Likewise.
12006 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
12007 Fold VEC_COND_EXPRs of an operation and a default value into
12008 an IFN_COND_* function if possible.
12009 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
12011 (SVE_COND_FP_BINARY): Include them.
12012 (optab, sve_fp_op): Handle them.
12013 (SVE_INT_BINARY_REV): New code iterator.
12014 (SVE_COND_FP_BINARY_REV): New int iterator.
12015 (commutative): New int attribute.
12016 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
12018 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
12020 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
12021 (*cond_<optab><mode>): New patterns for reversed operands.
12023 2018-05-25 Richard Biener <rguenther@suse.de>
12025 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
12026 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
12027 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
12028 (STMT_VINFO_GROUPED_ACCESS): Adjust.
12029 * tree-vect-data-refs.c (everywhere): Adjust users.
12030 * tree-vect-loop.c (everywhere): Likewise.
12031 * tree-vect-slp.c (everywhere): Likewise.
12032 * tree-vect-stmts.c (everywhere): Likewise.
12033 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
12035 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12037 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
12039 (gcc_cv_as_section_exclude): ... this.
12040 Try Solaris as #exclude syntax.
12041 * configure: Regenerate.
12042 * config.in: Regenerate.
12043 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
12045 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
12046 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
12048 * varasm.c (default_elf_asm_named_section): Don't check if
12049 HAVE_GAS_SECTION_EXCLUDE is defined.
12051 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12053 * doc/md.texi: Update the documentation of the cond_* optabs
12054 to mention the new final operand. Fix GET_MODE_NUNITS call.
12055 Describe the scalar case too.
12056 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
12057 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
12059 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
12060 (get_conditional_internal_fn): Update comment.
12061 * tree-vect-loop.c (vectorizable_reduction): Pass the original
12062 accumulator value as a final argument to conditional functions.
12063 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
12064 a define_expand and add an "else" operand. Assert for now that
12065 the else operand is equal to operand 2. Use SVE_INT_BINARY and
12066 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
12067 (*cond_<optab><mode>): New patterns.
12068 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
12069 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
12070 (UNSPEC_COND_EOR): Delete.
12071 (optab): Remove associated mappings.
12072 (SVE_INT_BINARY): New code iterator.
12073 (sve_int_op): Remove int attribute and add "minus" to the code
12075 (SVE_COND_INT_OP): Delete.
12076 (SVE_COND_FP_OP): Rename to...
12077 (SVE_COND_FP_BINARY): ...this.
12079 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12081 * optabs.c (can_reuse_operands_p): New function.
12082 (maybe_legitimize_operands): Try to reuse the results for
12085 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
12087 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
12088 Add {q} suffix to insn mnemonic.
12090 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12092 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
12093 (msp430_warn_func_return): New.
12095 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
12097 * fold-const.c (tree_nonzero_bits): New function.
12098 * fold-const.h (tree_nonzero_bits): Likewise.
12099 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
12100 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
12102 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
12106 * varasm.c (assemble_alias): Check ifunc_resolver only on
12109 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
12112 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
12113 when memory input operand is handled.
12115 2018-05-24 Luis Machado <luis.machado@linaro.org>
12117 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
12119 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
12121 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
12123 * match.pd: Delay FMA folds until after vectorization.
12125 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
12128 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
12129 address check not strict.
12131 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
12133 * gimple-match.h (gimple_match_op): New class.
12134 (mprts_hook): Replace parameters with a gimple_match_op *.
12135 (maybe_build_generic_op): Likewise.
12136 (gimple_simplified_result_is_gimple_val): Replace parameters with
12137 a const gimple_match_op *.
12138 (gimple_simplify): Replace code_helper * and tree * parameters with
12139 a gimple_match_op * parameter.
12140 (gimple_resimplify1): Replace code_helper *, tree and tree *
12141 parameters with a gimple_match_op * parameter.
12142 (gimple_resimplify2): Likewise.
12143 (gimple_resimplify3): Likewise.
12144 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
12145 parameters with a gimple_match_op * parameter.
12146 * gimple-match-head.c (gimple_simplify): Change prototypes of
12147 auto-generated functions to take a gimple_match_op * instead of
12148 separate code_helper * and tree * parameters. Make the same
12149 change in the top-level overload and update calls to the
12150 gimple_resimplify routines. Update calls to the auto-generated
12151 functions and to maybe_push_res_to_seq in the publicly-facing
12152 operation-specific gimple_simplify overloads.
12153 (gimple_match_op::MAX_NUM_OPS): Define.
12154 (gimple_resimplify1): Replace rcode and ops with a single res_op
12155 parameter. Update call to gimple_simplify.
12156 (gimple_resimplify2): Likewise.
12157 (gimple_resimplify3): Likewise.
12158 (mprts_hook): Replace parameters with a gimple_match_op *.
12159 (maybe_build_generic_op): Likewise.
12160 (build_call_internal): Replace type, nargs and ops with
12161 a gimple_match_op *.
12162 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
12163 with a single gimple_match_op *. Update calls to mprts_hook,
12164 build_call_internal and gimple_simplified_result_is_gimple_val.
12165 Factor out code that is common to the tree_code and combined_fn cases.
12166 * genmatch.c (expr::gen_transform): Replace tem_code and
12167 tem_ops with a gimple_match_op called tem_op. Update calls
12168 to the gimple_resimplify functions and maybe_push_res_to_seq.
12169 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
12170 res_ops. Update call to the gimple_resimplify functions.
12171 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
12172 (decision_tree::gen): Make the functions take a gimple_match_op *
12173 called res_op instead of separate res_code and res_ops parameters.
12174 Update call accordingly.
12175 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
12176 and ops with a single res_op parameter. Update calls to
12177 maybe_build_generic_op and maybe_push_res_to_seq.
12178 (fold_stmt_1): Update calls to gimple_simplify and
12179 replace_stmt_with_simplification.
12180 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
12181 and gimple_simplified_result_is_gimple_val.
12182 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
12184 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
12185 with a gimple_match_op *.
12186 (vn_nary_build_or_lookup): Likewise. Update call to
12187 vn_nary_build_or_lookup_1.
12188 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
12189 gimple_match_op *. Update calls to the gimple_resimplify routines
12190 and to gimple_simplified_result_is_gimple_val.
12191 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
12192 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
12193 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
12194 (visit_nary_op): Likewise.
12195 (visit_reference_op_load): Likewise.
12197 2018-05-23 Luis Machado <luis.machado@linaro.org>
12199 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
12200 modifier for printing the step amount.
12202 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12205 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
12208 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
12210 * doc/sourcebuild.texi (Endianness): New subsubsection.
12212 2018-05-23 Luis Machado <luis.machado@linaro.org>
12214 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12215 <prefetch_dynamic_strides>: New const bool field.
12216 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12217 prefetch_dynamic_strides.
12218 (exynosm1_prefetch_tune): Likewise.
12219 (thunderxt88_prefetch_tune): Likewise.
12220 (thunderx_prefetch_tune): Likewise.
12221 (thunderx2t99_prefetch_tune): Likewise.
12222 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
12224 (aarch64_override_options_internal): Update to set
12225 PARAM_PREFETCH_DYNAMIC_STRIDES.
12226 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
12227 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
12228 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
12229 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
12230 prefetch-dynamic-strides setting.
12232 2018-05-23 Luis Machado <luis.machado@linaro.org>
12234 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12235 <minimum_stride>: New const int field.
12236 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12237 minimum_stride field defaulting to -1.
12238 (exynosm1_prefetch_tune): Likewise.
12239 (thunderxt88_prefetch_tune): Likewise.
12240 (thunderx_prefetch_tune): Likewise.
12241 (thunderx2t99_prefetch_tune): Likewise.
12242 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
12243 <default_opt_level>: Set to 3.
12244 (aarch64_override_options_internal): Update to set
12245 PARAM_PREFETCH_MINIMUM_STRIDE.
12246 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
12247 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
12248 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
12249 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
12250 stride is constant and is below the minimum stride threshold.
12252 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12254 * config/arm/arm-cpus.in (mode26): Delete.
12255 (armv4): Delete mode26 reference.
12256 * config/arm/arm.c (arm_configure_build_target): Delete use of
12259 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
12261 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
12263 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
12264 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
12266 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
12267 (floatunsdidf2): Ditto.
12269 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
12271 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
12272 (fixuns_trunc<mode>si2_avx512f): Ditto.
12273 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
12274 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
12275 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
12277 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
12279 PR rtl-optimization/79985
12280 * df-scan.c (df_insn_refs_collect): Remove special case for
12281 global registers and asm statements.
12283 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
12285 * extend.texi (Global Register Variables): Rewrite the bullet list.
12286 Note that the register is available for allocation. Note that access
12287 via inline asm must use constraints. Add note about async-signal
12288 handlers. Remove paragraph about automagic register selection.
12290 2018-05-23 Richard Biener <rguenther@suse.de>
12292 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
12293 of fixed offset from memset VN.
12295 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
12297 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
12298 first_interp field.
12299 (alloc_cand_and_find_basis): Initialize first_interp field.
12300 (slsr_process_mul): Modify first_interp field.
12301 (slsr_process_add): Likewise.
12302 (slsr_process_cast): Modify first_interp field for each new
12304 (slsr_process_copy): Likewise.
12305 (dump_candidate): Dump first_interp field.
12306 (replace_mult_candidate): Process all interpretations, not just
12308 (replace_rhs_if_not_dup): Likewise.
12309 (replace_one_candidate): Likewise.
12311 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
12313 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
12315 (aarch64_needs_frame_chain): New function.
12316 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
12318 2018-05-23 Sudakshina Das <sudi.das@arm.com>
12321 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
12322 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
12323 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
12324 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
12325 as true for strict-align.
12326 (aarch64_can_inline_p): Perform checks even when callee has no
12327 attributes to check for strict alignment.
12328 * doc/extend.texi (AArch64 Function Attributes): Document
12330 * doc/invoke.texi: (AArch64 Options): Likewise.
12332 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
12334 PR tree-optimization/85853
12335 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
12336 the handling of the root of the node to...
12337 (vect_slp_analyze_node_operations_1): ...this new function,
12338 and run the whole thing with the child nodes' def types
12339 set according to their SLP node's def type.
12341 2018-05-23 Richard Biener <rguenther@suse.de>
12343 PR middle-end/85874
12344 * tree-data-ref.c (create_runtime_alias_checks): Defer
12345 and ignore overflow warnings.
12347 2018-05-23 Yury Gribov <tetra2005@gmail.com>
12349 PR tree-optimization/85822
12350 * tree-vrp.c (is_masked_range_test): Fix handling of negative
12353 2018-05-23 Richard Biener <rguenther@suse.de>
12355 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
12356 memset constants via native_interpret_expr.
12358 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
12361 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
12363 (cgraph_node::create_alias): Likewise.
12364 (cgraph_node::get_availability): Check ifunc_resolver instead
12365 of looking up ifunc attribute.
12366 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
12367 * varasm.c (do_assemble_alias): Likewise.
12368 (assemble_alias): Likewise.
12369 (default_binds_local_p_3): Likewise.
12370 * cgraph.h (cgraph_node): Add ifunc_resolver.
12371 (cgraph_node::only_called_directly_or_aliased_p): Return false
12372 for IFUNC resolver.
12373 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
12375 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
12376 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
12377 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
12378 instead of looking up ifunc attribute.
12380 2018-05-22 Luis Machado <luis.machado@linaro.org>
12382 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
12384 2018-05-22 Martin Sebor <msebor@redhat.com>
12386 PR middle-end/85359
12387 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
12388 only when expasion succeeds.
12389 (expand_builtin_strcmp): Same.
12390 (expand_builtin_strncmp): Same.
12392 2018-05-22 Martin Sebor <msebor@redhat.com>
12394 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
12396 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
12397 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12399 * config/aarch64/aarch64-ldpstp.md: Replace uses of
12400 aarch64_mem_pair_operand with memory_operand and delete operand swapping
12402 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
12403 Add check for legitimate_address.
12404 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
12405 (aarch64_swap_ldrstr_operands): New.
12406 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
12409 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
12410 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12412 * config/aarch64/aarch64.md: New patterns to generate stp
12414 (store_pair_sw, store_pair_dw): New patterns to generate stp for
12415 single words and double words.
12416 (load_pair_sw, load_pair_dw): Likewise.
12417 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
12419 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
12421 * config/aarch64/aarch64-ldpstp.md: Modify peephole
12422 for different mode ldpstp and add peephole for merged zero stores.
12423 Likewise for loads.
12424 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
12426 (aarch64_gen_store_pair): Rename calls to match new patterns.
12427 (aarch64_gen_load_pair): Rename calls to match new patterns.
12428 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
12429 (load_pair<DREG:mode><DREG2:mode>): ... This.
12430 (store_pair<mode>): Rename to...
12431 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
12432 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
12433 New mode iterators.
12434 (V_INT_EQUIV): Handle SImode.
12435 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
12438 2018-05-22 Martin Sebor <msebor@redhat.com>
12441 * calls.c (maybe_warn_nonstring_arg): Use string length to set
12442 or ajust the presumed bound on an operation to avoid unnecessary
12445 2018-05-22 Martin Sebor <msebor@redhat.com>
12447 PR tree-optimization/85826
12448 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
12449 assuming that a DECL necesarily has a constant size.
12451 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
12453 PR middle-end/85862
12454 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
12456 2018-05-22 Richard Biener <rguenther@suse.de>
12458 PR tree-optimization/85834
12459 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
12460 non-constant and non-zero memset arguments.
12462 2018-05-22 Martin Liska <mliska@suse.cz>
12465 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
12467 2018-05-22 Richard Biener <rguenther@suse.de>
12469 PR tree-optimization/85863
12470 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
12471 comparisons when vectype is specified.
12472 (vectorizable_condition): Do not specify vectype for
12473 vect_is_simple_cond when SLP vectorizing.
12475 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
12478 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
12479 define __ibm128 as long double.
12480 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
12481 as a distinct type when IEEE 128-bit support is enabled.
12482 (init_float128_ieee): Fix up conversions between IFmode and IEEE
12483 128-bit types to use the correct functions.
12484 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
12485 convert between 128-bit floating point types that have different
12486 modes but the same representation, instead of using gen_lowpart to
12488 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
12490 (IFKF_reg): New attributes to give the register constraints for
12492 (extend<mode>tf2_internal): New insns to mark an explicit
12493 conversion between 128-bit floating point types that have a
12494 different mode but share the same representation.
12496 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
12498 PR tree-optimization/85814
12499 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
12500 a null return from get_strinfo when unsharing the next
12501 strinfo in the chain.
12503 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
12506 * varasm.c (weak_finish): Clean up weak_decls.
12508 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12510 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
12511 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
12512 UNSPEC_UADALP values.
12513 * config/aarch64/iterators.md (ABAL): New int iterator.
12516 (sur): Add mappings for the above.
12517 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
12519 (aarch64_<sur>abal<mode>_4): Likewise.
12520 (aarch64_<sur>adalp<mode>_3): Likewise.
12521 (<sur>sadv16qi): New define_expand.
12523 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
12525 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
12526 (*movdf_internal): Ditto.
12527 (*rcpsf2_sse): Ditto.
12528 (*rsqrtsf2_sse): Ditto.
12529 (*sqrt<mode>2_sse): Ditto.
12531 2018-05-21 Tamar Christina <tamar.christina@arm.com>
12533 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
12535 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
12536 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
12537 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
12538 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
12540 * config/aarch64/arm_neon.h: Likewise.
12541 * config/aarch64/iterators.md (VQ_I): New.
12543 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
12545 * config.gcc: Add arc/t-multilib-linux to tmake_file for
12547 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
12550 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
12552 * config/nds32/constraints.md (S): New constraint.
12553 * config/nds32/nds32.md (call_internal): Use constraint S.
12554 (call_value_internal): Likewise.
12555 (sibcall_internal): Likewise.
12556 (sibcall_value_internal): Likewise.
12558 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
12559 Chung-Ju Wu <jasonwucj@gmail.com>
12561 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
12562 into consideration.
12564 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
12565 Chung-Ju Wu <jasonwucj@gmail.com>
12567 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
12568 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
12569 (nds32_rtx_costs_impl): Simplify.
12570 (nds32_address_cost_impl): Simplify.
12571 (nds32_init_rtx_costs): New function.
12572 (nds32_rtx_costs_speed_prefer): Likewise.
12573 (nds32_rtx_costs_size_prefer): Likewise.
12574 (nds32_address_cost_speed_prefer): Likewise.
12575 (nds32_address_cost_speed_fwprop): Likewise.
12576 (nds32_address_cost_size_prefer): Likewise.
12577 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
12578 * config/nds32/nds32.c (nds32_option_override): Use
12579 nds32_init_rtx_costs function.
12581 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
12583 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
12584 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
12585 (TARGET_PIPELINE_N8): Likewise.
12586 (TARGET_PIPELINE_N10): Likewise.
12587 (TARGET_PIPELINE_N13): Likewise.
12588 (TARGET_PIPELINE_GRAYWOLF): Likewise.
12590 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
12592 * config/nds32/nds32-fpu.md: Update copyright year.
12594 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12596 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
12598 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12600 * config/nds32/nds32.c
12601 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
12602 * config/nds32/nds32.opt (minline-asm-r15): New option.
12604 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12606 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
12608 * config/nds32/nds32.md (abssi2): New pattern.
12610 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
12612 * config/i386/i386.md (rex64namesuffix): New mode attribute.
12613 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
12614 Merge insn pattern from sse_cvtsi2ss<round_name> and
12615 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
12616 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
12617 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
12618 using SWI48 mode iterator.
12619 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
12620 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
12621 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
12622 pattern from sse_cvttss2si<round_saeonly_name>
12623 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
12624 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
12625 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
12626 using SWI48 mode iterator.
12627 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
12628 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
12629 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
12630 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
12631 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
12632 using SWI48 mode iterator.
12633 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
12634 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
12635 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
12636 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
12637 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
12638 SWI48 mode iterator.
12639 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
12640 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
12641 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
12642 pattern from sse_cvttsd2si<round_saeonly_name>
12643 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
12645 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12647 * config/nds32/nds32-md-auxiliary.c
12648 (nds32_valid_smw_lwm_base_p): Refine.
12649 (nds32_output_smw_single_word): Refine.
12650 (nds32_output_smw_double_word): New.
12651 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
12653 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12655 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
12656 (nds32_output_stack_pop): Refine.
12657 (nds32_expand_unaligned_load): Refine.
12658 (nds32_expand_unaligned_store): Refine.
12660 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12661 Chung-Ju Wu <jasonwucj@gmail.com>
12663 * config/nds32/constants.md: Add TP_REGNUM constant.
12664 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
12665 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
12667 * config/nds32/nds32-doubleword.md: Consider flag_pic.
12668 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
12669 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
12670 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
12671 and PIC code generation.
12672 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
12674 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
12676 * config/nds32/nds32.md: Support TLS and PIC.
12677 * config/nds32/nds32.c: Support TLS and PIC.
12678 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
12679 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
12682 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12684 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
12685 mode with E_ prefix.
12687 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12688 Chung-Ju Wu <jasonwucj@gmail.com>
12690 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
12691 * config/nds32/nds32-md-auxiliary.c
12692 (symbolic_reference_mentioned_p): New.
12693 (nds32_legitimize_ict_address): New.
12694 (nds32_expand_ict_move): New.
12695 (nds32_indirect_call_referenced_p): New.
12696 (nds32_symbol_binds_local_p): Delete.
12697 (nds32_long_call_p): Modify.
12698 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
12699 * config/nds32/nds32-protos.h
12700 (symbolic_reference_mentioned_p): Declare.
12701 (nds32_legitimize_ict_address): Declare.
12702 (nds32_expand_ict_move): Declare.
12703 (nds32_indirect_call_referenced_p): Declare.
12704 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
12705 (nds32_relax_group): Use nds32_ict_const_p as condition.
12706 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
12707 (nds32_asm_file_start): Output ict_model directive in asm code.
12708 (nds32_legitimate_address_p): Consider indirect call.
12709 (nds32_print_operand): Consider indirect call.
12710 (nds32_print_operand_address): Consider indirect call.
12711 (nds32_insert_attributes): Handle "indirect_call" attribute.
12712 (TARGET_LEGITIMATE_ADDRESS_P): Define.
12713 (TARGET_LEGITIMATE_CONSTANT_P): Define.
12714 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
12715 (TARGET_DELEGITIMIZE_ADDRESS): Define.
12716 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
12717 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
12718 (TARGET_ICT_MODEL_SMALL): Define.
12719 (TARGET_ICT_MODEL_LARGE): Define.
12720 * config/nds32/nds32.md (movsi): Consider ict model.
12721 (call, call_value): Consider ict model.
12722 (sibcall, sibcall_value): Consider ict model.
12723 * config/nds32/nds32.opt (mict-model): New option.
12724 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
12727 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
12728 Monk Chiang <sh.chiang04@gmail.com>
12729 Jim Wilson <jimw@sifive.com>
12731 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
12732 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
12733 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
12734 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
12735 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
12736 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
12737 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
12738 compute save_libcall_adjustment properly.
12739 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
12740 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
12741 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
12742 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
12743 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
12744 (ABI_SPEC): Handle mabi=ilp32e.
12745 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
12746 (RVE): Add RVE mask.
12747 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
12748 <-march>: Add rv32e as an example.
12750 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
12753 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
12754 (intra_create_variable_infos): Handle C++ constructors.
12756 2018-05-18 Martin Liska <mliska@suse.cz>
12758 * passes.def: Remove a redundant pass.
12760 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
12763 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
12765 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12767 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
12769 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
12771 (armv2, armv2a, armv3, armv3m): Delete architectures.
12772 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
12773 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
12774 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
12776 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
12777 (*mulsidi3adddi): Likewise.
12778 (mulsidi3): Likewise.
12779 (*mulsidi3_nov6): Likewise.
12780 (umulsidi3): Likewise.
12781 (umulsidi3_nov6): Likewise.
12782 (umaddsidi4): Likewise.
12783 (*umulsidi3adddi): Likewise.
12784 (smulsi3_highpart): Likewise.
12785 (*smulsi3_highpart_nov6): Likewise.
12786 (umulsi3_highpart): Likewise.
12787 (*umulsi3_highpart_nov6): Likewise.
12788 * config/arm/arm.h (arm_arch3m): Delete.
12789 * config/arm/arm.c (arm_arch3m): Delete.
12790 (arm_option_override_internal): Update armv3-related comment.
12791 (arm_configure_build_target): Delete use of isa_bit_mode32.
12792 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
12793 (arm_rtx_costs_internal): Delete check of arm_arch3m.
12794 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
12795 (mulsa3): Likewise.
12796 (mulusa3): Likewise.
12797 * config/arm/arm-protos.h (arm_arch3m): Delete.
12798 * config/arm/arm-tables.opt: Regenerate.
12799 * config/arm/arm-tune.md: Likewise.
12800 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
12801 deleted architectures.
12803 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12805 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
12806 (armv5t, armv5te): New features.
12807 (ARMv5, ARMv5e): Delete fgroups.
12808 (ARMv5t, ARMv5te): Adjust for above changes.
12809 (ARMv6m): Likewise.
12810 (armv5, armv5e): Delete arches.
12811 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
12813 (*call_reg_arm): Likewise.
12814 (*call_value_reg_armv5): Likewise.
12815 (*call_value_reg_arm): Likewise.
12816 (*call_symbol): Likewise.
12817 (*call_value_symbol): Likewise.
12818 (*sibcall_insn): Likewise.
12819 (*sibcall_value_insn): Likewise.
12820 (clzsi2): Likewise.
12821 (prefetch): Likewise.
12822 (define_split and define_peephole2 dependent on arm_arch5):
12824 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
12826 (TARGET_ARM_QBIT): Likewise.
12827 (TARGET_DSP_MULTIPLY): Likewise.
12828 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
12829 (arm_arch5, arm_arch5e): Delete.
12830 (arm_arch5t, arm_arch5te): Declare.
12831 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
12832 (arm_arch5t): Declare.
12833 (arm_option_reconfigure_globals): Update for the above.
12834 (arm_options_perform_arch_sanity_checks): Update comment, replace
12835 use of arm_arch5 with arm_arch5t.
12836 (use_return_insn): Likewise.
12837 (arm_emit_call_insn): Likewise.
12838 (output_return_instruction): Likewise.
12839 (arm_final_prescan_insn): Likewise.
12840 (arm_coproc_builtin_available): Likewise.
12841 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
12842 arm_arch5e with arm_arch5t and arm_arch5te.
12843 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
12844 (arm_arch5t, arm_arch5te): Declare.
12845 * config/arm/arm-tables.opt: Regenerate.
12846 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
12847 * config/arm/t-multilib: Likewise.
12848 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
12849 instead of arm_arch5.
12850 (*call_reg_thumb1): Likewise.
12851 (*call_value_reg_thumb1_v5): Likewise.
12852 (*call_value_reg_thumb1): Likewise.
12853 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
12855 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
12857 2018-05-18 Martin Liska <mliska@suse.cz>
12859 PR gcov-profile/84846
12860 * doc/gcov.texi: Document -t option of gcov tool.
12862 2018-05-18 Martin Liska <mliska@suse.cz>
12864 PR gcov-profile/84846
12865 * gcov.c (print_usage): Add new -t option.
12866 (process_args): Handle the option.
12867 (generate_results): Use stdout as output when requested by
12870 2018-05-18 Martin Liska <mliska@suse.cz>
12872 PR gcov-profile/84846
12873 * coverage.c (coverage_init): Write PWD to .gcno file.
12874 * doc/gcov.texi: Document how working directory is printed.
12875 * gcov-dump.c (dump_gcov_file): Print PWD.
12876 * gcov.c (output_intermediate_file): Likewise.
12877 (read_graph_file): Read PWD string.
12878 (output_lines): Print PWD.
12880 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12882 PR middle-end/85817
12883 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
12884 for retval and return false if all args to phi are zero.
12886 2018-05-18 Richard Biener <rguenther@suse.de>
12888 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
12890 (evrp_dom_walker::before_dom_children): Call it.
12892 2018-05-18 Richard Biener <rguenther@suse.de>
12894 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
12895 results when processing array refs with variable index.
12897 2018-05-18 Toon Moene <toon@moene.org>
12899 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
12900 directly after that of -floop-interchange. Indicate that both
12901 options are enabled by default when specifying -O3.
12903 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12905 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
12906 iterator. Delete separate integer-mode vec_set<mode> expander.
12907 (aarch64_simd_vec_setv2di): Delete.
12908 (vec_setv2di): Delete.
12909 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
12910 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
12911 the "w, r" alternative.
12913 2018-05-18 Martin Liska <mliska@suse.cz>
12915 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
12916 * tree-pass.h (make_pass_lower_switch_O0): New function.
12917 * tree-switch-conversion.c (node_has_low_bound): Remove.
12918 (node_has_high_bound): Likewise.
12919 (node_is_bounded): Likewise.
12920 (class pass_lower_switch): Make it a template type and create
12922 (pass_lower_switch::execute): Add template argument.
12923 (make_pass_lower_switch): New function.
12924 (make_pass_lower_switch_O0): New function.
12925 (do_jump_if_equal): Remove.
12926 (emit_case_nodes): Simplify to just handle all 3 cases and leave
12927 all the hard work to tree optimization passes.
12929 2018-05-18 Martin Liska <mliska@suse.cz>
12931 * dbgcnt.c (limit_low): Renamed from limit.
12932 (limit_high): New variable.
12933 (dbg_cnt_is_enabled): Check for upper limit.
12934 (dbg_cnt): Adjust dumping.
12935 (dbg_cnt_set_limit_by_index): Add new argument for high
12937 (dbg_cnt_set_limit_by_name): Likewise.
12938 (dbg_cnt_process_single_pair): Parse new format.
12939 (dbg_cnt_process_opt): Use strtok.
12940 (dbg_cnt_list_all_counters): Remove 'value' and add
12942 * doc/invoke.texi: Document changes.
12944 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
12946 * doc/sourcebuild.texi (scalar_all_fma): Document.
12947 * tree.def (FMA_EXPR): Delete.
12948 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
12949 * internal-fn.c (ternary_direct): New macro.
12950 (expand_ternary_optab_fn): Likewise.
12951 (direct_ternary_optab_supported_p): Likewise.
12952 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
12953 * builtins.c (fold_builtin_fma): Delete.
12954 (fold_builtin_3): Don't call it.
12955 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
12956 * expr.c (expand_expr_real_2): Likewise.
12957 * fold-const.c (operand_equal_p): Likewise.
12958 (fold_ternary_loc): Likewise.
12959 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
12960 * gimple.c (DEFTREECODE): Likewise.
12961 * gimplify.c (gimplify_expr): Likewise.
12962 * optabs-tree.c (optab_for_tree_code): Likewise.
12963 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12964 * tree-eh.c (operation_could_trap_p): Likewise.
12965 (stmt_could_throw_1_p): Likewise.
12966 * tree-inline.c (estimate_operator_cost): Likewise.
12967 * tree-pretty-print.c (dump_generic_node): Likewise.
12968 (op_code_prio): Likewise.
12969 * tree-ssa-loop-im.c (stmt_cost): Likewise.
12970 * tree-ssa-operands.c (get_expr_operands): Likewise.
12971 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
12972 * fold-const-call.h (fold_fma): Delete.
12973 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
12974 CFN_FNMA and CFN_FNMS.
12975 (fold_fma): Delete.
12976 * genmatch.c (combined_fn): New enum.
12977 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
12978 (commutative_op): New function.
12979 (commutate): Use it. Handle more than 2 operands.
12980 (dt_operand::gen_gimple_expr): Use commutative_op.
12981 (parser::parse_expr): Allow :c to be used with non-binary
12982 operators if the commutative operand is known.
12983 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
12984 CFN_FMS, CFN_FNMA and CFN_FNMS.
12985 (backprop::process_assign_use): Remove FMA_EXPR handling.
12986 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
12987 (gen_hsa_fma): New function.
12988 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
12989 IFN_FNMA and IFN_FNMS.
12990 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
12991 * gimple-fold.h (follow_all_ssa_edges): Declare.
12992 * gimple-fold.c (follow_all_ssa_edges): New function.
12993 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
12994 gimple_build interface and use follow_all_ssa_edges to fold the result.
12995 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
12996 instead of checking for optabs directly.
12997 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
12998 rather than FMA_EXPRs.
12999 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
13000 call to IFN_FMA instead of an FMA_EXPR.
13002 2018-05-17 Jim Wilson <jimw@sifive.com>
13004 * expr.c (do_tablejump): When converting index to Pmode, if we have a
13005 sign extended promoted subreg, and the range does not have the sign bit
13006 set, then do a sign extend.
13008 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
13009 test, check for sign extended subreg and/or constant operands, and
13010 do a sign extend in that case.
13012 2018-05-17 Steve Ellcey <sellcey@cavium.com>
13014 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
13015 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
13017 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
13018 Change logics_shift_reg to logics_shift_imm.
13019 (thunderx2t99_fp_loadpair_basic): Delete.
13020 (thunderx2t99_fp_storepair_basic): Delete.
13021 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
13022 (thunderx2t99_asimd_polynomial): Delete.
13023 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
13024 and neon_fp_mul_d_scalar_q.
13025 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
13026 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
13027 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
13028 (thunderx2t99_asimd_lut): Add missing tbl types.
13029 (thunderx2t99_asimd_ext): Delete.
13030 (thunderx2t99_asimd_load1_1_mult): Delete.
13031 (thunderx2t99_asimd_load1_2_mult): Delete.
13032 (thunderx2t99_asimd_load1_ldp): New.
13033 (thunderx2t99_asimd_load1): New.
13034 (thunderx2t99_asimd_load2): Add missing *load2* types.
13035 (thunderx2t99_asimd_load3): New.
13036 (thunderx2t99_asimd_load4): New.
13037 (thunderx2t99_asimd_store1_1_mult): Delete.
13038 (thunderx2t99_asimd_store1_2_mult): Delete.
13039 (thunderx2t99_asimd_store2_mult): Delete.
13040 (thunderx2t99_asimd_store2_onelane): Delete.
13041 (thunderx2t99_asimd_store_stp): New.
13042 (thunderx2t99_asimd_store1): New.
13043 (thunderx2t99_asimd_store2): New.
13044 (thunderx2t99_asimd_store3): New.
13045 (thunderx2t99_asimd_store4): New.
13047 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
13049 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
13050 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
13052 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
13053 Segher Boessenkool <segher@kernel.crashing.org>
13056 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
13059 2018-05-17 Richard Biener <rguenther@suse.de>
13061 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
13062 for pruning loop and prune defs feeding only already visited PHIs.
13064 2018-05-17 Richard Biener <rguenther@suse.de>
13066 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
13068 2018-05-17 Bin Cheng <bin.cheng@arm.com>
13069 Richard Biener <rguenther@suse.de>
13071 PR tree-optimization/85793
13072 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
13073 for VMAT_ELEMENTWISE.
13075 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
13077 * internal-fn.h (lookup_internal_fn): Declare
13078 * internal-fn.c (lookup_internal_fn): New function.
13079 * gimple.c (gimple_build_call_from_tree): Handle calls to
13080 internal functions.
13081 * gimple-pretty-print.c (dump_gimple_call): Print "." before
13082 internal function names.
13083 * tree-pretty-print.c (dump_generic_node): Likewise.
13084 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
13086 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
13088 * gimple-fold.h (gimple_build): Make the function forms take
13089 combined_fn rather than built_in_function.
13090 (gimple_simplify): Likewise.
13091 * gimple-match-head.c (gimple_simplify): Likewise.
13092 * gimple-fold.c (gimple_build): Likewise.
13093 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
13094 rather than gimple_build_call_internal.
13095 (get_initial_defs_for_reduction): Likewise.
13096 (vect_create_epilog_for_reduction): Likewise.
13097 (vectorizable_live_operation): Likewise.
13099 2018-05-17 Martin Liska <mliska@suse.cz>
13101 * gimple-ssa-sprintf.c (format_directive): Do not use
13102 space in between 'G_' and '('.
13104 2018-05-17 Jakub Jelinek <jakub@redhat.com>
13107 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
13108 even if the mask is not all ones.
13111 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
13113 (ix86_gimple_fold_builtin): Likewise.
13116 * config/i386/i386.c: Include tree-vector-builder.h.
13117 (ix86_vector_shift_count): New function.
13118 (ix86_fold_builtin): Fold shift builtins by scalar count.
13119 (ix86_gimple_fold_builtin): Likewise.
13121 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
13122 _mm512_setzero): New intrinsics.
13124 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
13125 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13127 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
13128 code generation for cases where splatting a value is not useful.
13129 * simplify-rtx.c (simplify_ternary_operation): Simplify
13130 vec_merge across a vec_duplicate and a paradoxical subreg forming
13131 a vector mode to a vec_concat.
13133 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
13135 * config.gcc: Support "goldmont-plus".
13136 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
13138 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13139 PROCESSOR_GOLDMONT_PLUS.
13140 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
13141 (processor_target_table): Add "goldmont-plus".
13142 (PTA_GOLDMONT_PLUS): Define.
13143 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
13144 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
13145 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
13146 (fold_builtin_cpu): Add "goldmont-plus".
13147 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
13148 (ix86_option_override_internal): Add "goldmont-plus".
13149 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
13150 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
13151 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
13152 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
13154 2018-05-17 Richard Biener <rguenther@suse.de>
13156 PR tree-optimization/85757
13157 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
13158 remove defs that only feed that PHI from further processing.
13160 2018-05-16 Jim Wilson <jimw@sifive.com>
13162 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
13164 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
13166 2018-05-16 Mark Wielaard <mark@klomp.org>
13168 * dwarf2out.c (count_index_strings): New function.
13169 (output_indirect_strings): Call count_index_strings and generate
13170 header for dwarf_version >= 5.
13172 2018-05-16 Mark Wielaard <mark@klomp.org>
13174 * dwarf2out.c (dwarf_FORM): New function.
13175 (set_indirect_string): Use dwarf_FORM.
13176 (reset_indirect_string): Likewise.
13177 (size_of_die): Likewise.
13178 (value_format): Likewise.
13179 (output_die): Likewise.
13180 (add_skeleton_AT_string): Likewise.
13181 (output_macinfo_op): Likewise.
13182 (index_string): Likewise.
13183 (output_index_string_offset): Likewise.
13184 (output_index_string): Likewise.
13185 (count_index_strings): Likewise.
13187 2018-05-16 Carl Love <cel@us.ibm.com>
13189 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
13190 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
13192 2018-05-16 Martin Jambor <mjambor@suse.cz>
13194 * ipa-prop.c (ipa_free_all_edge_args): Remove.
13195 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
13197 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
13199 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
13200 (fnma<mode>4): Likewise.
13201 (fms<mode>4): Likewise.
13202 (fnms<mode>4): Likewise.
13203 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
13204 (aarch64_fnma<mode>4): Likewise.
13205 (aarch64_fms<mode>4): Likewise.
13206 (aarch64_fnms<mode>4): Likewise.
13207 (aarch64_fnmadd<mode>4): Likewise.
13209 2018-05-16 Jason Merrill <jason@redhat.com>
13211 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
13213 2018-05-16 Richard Biener <rguenther@suse.de>
13215 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
13216 (dump_stmt_cost): Declare.
13217 (add_stmt_cost): Dump cost we add.
13218 (add_stmt_costs): New function.
13219 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
13220 No longer exported.
13221 (vect_analyze_stmt): Adjust prototype.
13222 (vectorizable_condition): Likewise.
13223 (vectorizable_live_operation): Likewise.
13224 (vectorizable_reduction): Likewise.
13225 (vectorizable_induction): Likewise.
13226 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
13227 cost vector to pass to vectorizable_ and record afterwards.
13228 (vect_model_reduction_cost): Take cost vector argument and adjust.
13229 (vect_model_induction_cost): Likewise.
13230 (vectorizable_reduction): Likewise.
13231 (vectorizable_induction): Likewise.
13232 (vectorizable_live_operation): Likewise.
13233 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
13234 SLP_TREE_NUMBER_OF_VEC_STMTS.
13235 (vect_analyze_slp_cost_1): Remove.
13236 (vect_analyze_slp_cost): Likewise.
13237 (vect_slp_analyze_node_operations): Take visited args and
13238 a target cost vector. Avoid processing already visited stmt sets.
13239 (vect_slp_analyze_operations): Use a local cost vector to gather
13240 costs and register those of non-discarded instances.
13241 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
13242 (vect_schedule_slp_instance): Remove copying of
13243 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
13245 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
13246 adding cost. Record cost entry location.
13247 (vect_prologue_cost_for_slp_op): Function to compute cost of
13248 a constant or invariant generated for SLP vect in the prologue,
13249 split out from vect_analyze_slp_cost_1.
13250 (vect_model_simple_cost): Make static. Adjust for SLP costing.
13251 (vect_model_promotion_demotion_cost): Likewise.
13252 (vect_model_store_cost): Likewise, make static.
13253 (vect_model_load_cost): Likewise.
13254 (vectorizable_bswap): Add cost vector arg and adjust.
13255 (vectorizable_call): Likewise.
13256 (vectorizable_simd_clone_call): Likewise.
13257 (vectorizable_conversion): Likewise.
13258 (vectorizable_assignment): Likewise.
13259 (vectorizable_shift): Likewise.
13260 (vectorizable_operation): Likewise.
13261 (vectorizable_store): Likewise.
13262 (vectorizable_load): Likewise.
13263 (vectorizable_condition): Likewise.
13264 (vectorizable_comparison): Likewise.
13265 (can_vectorize_live_stmts): Likewise.
13266 (vect_analyze_stmt): Likewise.
13267 (vect_transform_stmt): Adjust calls to vectorizable_*.
13268 * tree-vectorizer.c: Include gimple-pretty-print.h.
13269 (dump_stmt_cost): New function.
13271 2018-05-16 Richard Biener <rguenther@suse.de>
13273 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
13274 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
13275 * tree-ssa-dse.c: Include tree-ssa-loop.h.
13276 (check_name): New callback.
13277 (dse_classify_store): Track cycles via a visited bitmap of PHI
13278 defs and simplify handling of in-loop and across loop dead stores
13279 and properly fail for loop-variant refs. Handle byte-tracking with
13280 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
13283 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
13285 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
13286 (vect_get_mask_type_for_stmt): Likewise.
13287 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
13289 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
13290 to determine the statement's vector type and the vector type that
13291 should be used for calculating nunits. Deal with cases in which
13292 the type has to be deferred.
13293 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
13294 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
13295 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
13296 (vect_determine_vf_for_stmt): New functions, split out from...
13297 (vect_determine_vectorization_factor): ...here.
13298 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
13299 (vect_get_mask_type_for_stmt): New functions, split out from
13300 vect_determine_vectorization_factor.
13302 2018-05-16 Richard Biener <rguenther@suse.de>
13304 * tree-cfg.c (verify_gimple_assign_ternary): Properly
13305 verify the [VEC_]COND_EXPR embedded comparison.
13307 2018-05-15 Martin Sebor <msebor@redhat.com>
13309 PR tree-optimization/85753
13310 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
13311 RECORD_TYPE in addition to ARRAY_TYPE.
13313 2018-05-15 Martin Sebor <msebor@redhat.com>
13315 PR middle-end/85643
13316 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
13318 2018-05-15 Richard Biener <rguenther@suse.de>
13320 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
13321 add by_clobber_p one. Change algorithm to collect all defs
13322 representing uses we need to walk and try reducing them to
13323 a single one before failing.
13324 (dse_dom_walker::dse_optimize_stmt): Adjust.
13326 2018-05-13 Mark Wielaard <mark@klomp.org>
13328 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
13329 (size_of_loc_descr): Likewise.
13330 (output_loc_operands): Likewise.
13331 (output_loc_operands_raw): Likewise.
13332 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
13333 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
13334 (hash_loc_operands): Likewise.
13335 (compare_loc_operands): Likewise.
13337 2018-05-14 Mark Wielaard <mark@klomp.org>
13339 * dwarf2out.c (count_index_addrs): New function.
13340 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
13342 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13344 PR tree-optimization/83648
13345 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
13346 return value as malloc candidate.
13348 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13351 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
13352 param as true in call to suggest_attribute.
13354 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
13356 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
13357 -mreadonly-in-sdata.
13359 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13361 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
13363 (aarch64_crypto_aesd_fused): Likewise.
13365 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
13367 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
13368 (movsi_aarch64): Likewise.
13369 (load_pairsi): Likewise.
13370 (load_pairdi): Likewise.
13371 (store_pairsi): Likewise.
13372 (store_pairdi): Likewise.
13373 (load_pairsf): Likewise.
13374 (load_pairdf): Likewise.
13375 (store_pairsf): Likewise.
13376 (store_pairdf): Likewise.
13377 (zero_extend): Likewise.
13378 (trunc): Swap alternatives.
13379 (fcvt_target): Add '?' to prefer w over r.
13381 2018-05-14 Jakub Jelinek <jakub@redhat.com>
13384 * config/i386/i386.md: Disallow non-commutative arithmetics in
13385 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
13386 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
13387 in the peephole2 before it.
13389 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
13391 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
13392 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
13393 (ix86_handle_option): Handle -mcldemote.
13394 * config.gcc: New header.
13395 * config/i386/cldemoteintrin.h: New file.
13396 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
13397 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
13399 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13400 OPTION_MASK_ISA_CLDEMOTE.
13401 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
13402 (ix86_valid_target_attribute_inner_p): Ditto.
13403 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
13404 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
13405 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
13406 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
13407 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
13409 * config/i386/i386.opt: Add -mcldemote.
13410 * config/i386/x86intrin.h: New header.
13411 * doc/invoke.texi: Add -mcldemote.
13413 2018-05-14 Richard Biener <rguenther@suse.de>
13415 * doc/match-and-simplify.texi: Adjust :s documentation.
13417 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
13419 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
13420 intended memcpy size.
13421 (REORDER_45): Likewise.
13423 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
13425 * sort.cc: New file.
13426 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
13427 * vec.c (qsort_chk): Use gcc_qsort.
13428 * Makefile.in (OBJS-libcommon): Add sort.o.
13429 (build/sort.o): New target. Use it...
13430 (BUILD_RTL): ... here, and...
13431 (build/gencfn-macros): ... here, and...
13432 (build/genmatch): ... here.
13434 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13435 Chung-Ju Wu <jasonwucj@gmail.com>
13437 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
13438 * config/nds32/nds32-graywolf.md: New file.
13439 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
13440 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
13442 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
13443 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
13444 * config/nds32/nds32.md (pipeline_model): Add graywolf.
13445 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
13446 * config/nds32/pipelines.md: Include n15 settings.
13448 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13449 Chung-Ju Wu <jasonwucj@gmail.com>
13451 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
13452 * config/nds32/nds32-n13.md: New file.
13453 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
13454 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
13456 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
13457 * config/nds32/nds32.md (pipeline_model): Add n13.
13458 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
13459 * config/nds32/pipelines.md: Include n13 settings.
13461 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13462 Chung-Ju Wu <jasonwucj@gmail.com>
13464 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
13465 * config/nds32/nds32-n10.md: New file.
13466 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
13467 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
13469 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
13470 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
13471 * config/nds32/nds32.md (pipeline_model): Add n10.
13472 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
13473 * config/nds32/pipelines.md: Include n10 settings.
13475 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
13476 Kito Cheng <kito.cheng@gmail.com>
13477 Chung-Ju Wu <jasonwucj@gmail.com>
13479 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
13480 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
13481 Add enum values for DSP extension instructions.
13482 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
13484 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
13485 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
13486 New code iterators.
13487 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
13488 * config/nds32/nds32-dspext.md: New file for DSP implementation.
13489 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
13490 * config/nds32/nds32-intrinsic.md: Likewise.
13491 * config/nds32/nds32_intrinsic.h: Likewise.
13492 * config/nds32/nds32-md-auxiliary.c: Likewise.
13493 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
13494 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
13495 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
13496 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
13497 * config/nds32/nds32-protos.h: New declarations for DSP extension.
13498 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
13499 TYPE_DMAC in switch statement.
13500 * config/nds32/nds32.c: New checking and implementation for DSP
13501 extension instructions.
13502 * config/nds32/nds32.h: Likewise.
13503 * config/nds32/nds32.md: Likewise.
13504 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
13505 * config/nds32/predicates.md: Implement new predicates for DSP
13508 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13510 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
13511 Reformat alternatives and attributes so it is easier to identify
13512 which constraints/attributes go with which instruction.
13513 (mov<mode>_hardfloat32, FMOVE64): Likewise.
13514 (mov<mode>_softfloat32, FMOVE64): Likewise.
13515 (mov<mode>_hardfloat64, FMOVE64): Likewise.
13516 (mov<mode>_softfloat64, FMOVE64): Likewise.
13518 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13520 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13522 (Basic PowerPC Built-in Functions): The new name of the
13523 subsection previously known as "PowerPC Built-in Functions".
13524 (Basic PowerPC Built-in Functions Available on all Configurations):
13526 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
13527 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
13528 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
13529 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
13531 2018-05-11 Martin Jambor <mjambor@suse.cz>
13534 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
13537 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
13540 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
13542 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
13544 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
13545 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
13546 (ix86_handle_option): Handle -mwaitpkg.
13547 * config.gcc: New header.
13548 * config/i386/cpuid.h (bit_WAITPKG): New bit.
13549 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
13550 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
13552 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13553 OPTION_MASK_ISA_WAITPKG.
13554 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
13555 (ix86_option_override_internal): Add PTA_WAITPKG.
13556 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
13557 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
13558 IX86_BUILTIN_TPAUSE.
13559 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
13560 __builtin_ia32_umwait and __builtin_ia32_tpause.
13561 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
13562 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
13563 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
13564 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
13565 UNSPECV_TPAUSE): New.
13566 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
13567 * config/i386/i386.opt: Add -mwaitpkg.
13568 * config/i386/waitpkgintrin.h: New file.
13569 * config/i386/x86intrin.h: New header.
13570 * doc/invoke.texi: Add -mwaitpkg.
13572 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
13575 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
13577 (cortex-m0): Use armv6s-m isa.
13578 (cortex-m0plus): Likewise.
13579 (cortex-m1): Likewise.
13580 (cortex-m0.small-multiply): Likewise.
13581 (cortex-m0plus.small-multiply): Likewise.
13582 (cortex-m1.small-multiply): Likewise.
13584 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
13585 Jakub Jelinek <jakub@redhat.com>
13587 PR tree-optimization/85692
13588 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
13589 source permute as well.
13591 2018-05-11 Martin Liska <mliska@suse.cz>
13594 * doc/extend.texi: Document LLVM style format for no_sanitize
13597 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
13599 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
13600 mode_supports_vsx_dform_quad to mode_supports_dq_form.
13601 (mode_supports_vsx_dform_quad): Likewise.
13602 (mode_supports_vmx_dform): Move these functions to be next to the
13603 other mode_supports functions.
13604 (mode_supports_dq_form): Likewise.
13605 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
13606 mode_supports_dq_form.
13607 (reg_offset_addressing_ok_p): Likewise.
13608 (offsettable_ok_by_alignment): Likewise.
13609 (rs6000_legitimate_offset_address_p): Likewise.
13610 (legitimate_lo_sum_address_p): Likewise.
13611 (rs6000_legitimize_address): Likewise.
13612 (rs6000_legitimize_reload_address): Likewise.
13613 (rs6000_secondary_reload_inner): Likewise.
13614 (rs6000_preferred_reload_class): Likewise.
13615 (rs6000_output_move_128bit): Likewise.
13617 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
13619 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
13620 Generate SImode target register for null target.
13621 <case IX86_BUILTIN_XGETBV>: Ditto.
13622 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
13623 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
13625 2018-05-10 Carl Love <cel@us.ibm.com>
13627 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
13628 dcbtt and dcbtstt if operands[2] is 0.
13630 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
13633 * config/i386/sse.md (usadv64qi): New expander.
13635 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
13637 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
13638 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
13639 -maltivec=be support.
13640 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
13641 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
13642 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13643 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
13644 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
13645 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
13646 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
13647 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
13648 altivec_vsumsws): Adjust.
13649 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
13650 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
13651 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
13653 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
13654 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
13655 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
13656 (altivec_lve<VI_char>x): Delete expand.
13657 (*altivec_lve<VI_char>x_internal): Rename to...
13658 (altivec_lve<VI_char>x): ... this.
13659 (altivec_lvxl_<mode>): Delete expand.
13660 (*altivec_lvxl_<mode>_internal): Rename to ...
13661 (altivec_lvxl_<mode>): ... this.
13662 (altivec_stvxl_<mode>): Delete expand.
13663 (*altivec_stvxl_<mode>_internal): Rename to ...
13664 (altivec_stvxl_<mode>): ... this.
13665 (altivec_stve<VI_char>x): Delete expand.
13666 (*altivec_stve<VI_char>x_internal): Rename to ...
13667 (altivec_stve<VI_char>x): ... this.
13668 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
13669 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
13670 reduc_plus_scal_<mode>): Adjust.
13671 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
13673 (rs6000_cpu_cpp_builtins): Adjust.
13674 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
13675 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
13676 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
13677 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
13678 -maltivec=be support.
13679 (rs6000_split_vec_extract_var): Adjust.
13680 (rs6000_split_v4si_init): Adjust.
13681 (swap_selector_for_mode): Delete.
13682 (altivec_expand_lvx_be, altivec_expand_stvx_be,
13683 altivec_expand_stvex_be): Delete.
13684 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
13685 -maltivec=be support.
13686 (rs6000_gimple_fold_builtin): Ditto.
13687 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
13689 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
13690 (TARGET_DIRECT_MOVE_64BIT): Adjust.
13691 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
13692 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
13693 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
13694 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
13695 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
13696 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
13697 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
13698 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
13699 anonymous split): Adjust.
13700 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
13701 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
13703 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
13705 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
13706 when --with-gxx-include-dir is also specified.
13707 * configure: Regenerate.
13709 2018-05-09 Jim Wilson <jimw@sifive.com>
13712 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
13713 * config/riscv/t-withmultilib: New.
13714 * config/riscv/withmultilib.h: New.
13715 * doc/install.texi: Document RISC-V --with-multilib-list support.
13717 2018-05-09 Richard Biener <rguenther@suse.de>
13719 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
13721 (vect_bb_vectorization_profitable_p): Adjust. Compute
13722 actual scalar cost using the cost vector and the add_stmt_cost
13725 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13727 PR rtl-optimization/85645
13728 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
13729 in the REG_CFA_REGISTER note for LR, don't leave it empty.
13731 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13733 PR rtl-optimization/85645
13734 * shrink-wrap.c (spread_components): Return a boolean saying if
13735 anything was changed.
13736 (try_shrink_wrapping_separate): Iterate spread_components until
13737 nothing changes anymore.
13739 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13741 PR rtl-optimization/85645
13742 * regrename.c (build_def_use): Also kill the chains that include the
13743 destination of a REG_CFA_REGISTER note.
13745 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13747 PR rtl-optimization/85645
13748 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
13749 insn that has a REG_CFA_REGISTER note.
13751 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
13753 * cfgexpand.c (expand_clobber): New function.
13754 (expand_gimple_stmt_1): Use it.
13755 * tree-vect-stmts.c (vect_clobber_variable): New function,
13757 (vectorizable_simd_clone_call): ...here.
13758 (vectorizable_store): Emit a clobber either side of an
13759 IFN_STORE_LANES sequence.
13760 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
13762 2018-05-09 Tom de Vries <tom@codesourcery.com>
13765 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
13766 (define_insn "trap_if_false"): Add exit after trap.
13768 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
13770 PR rtl-optimization/85638
13771 * bb-reorder.c: Include common/common-target.h.
13772 (create_forwarder_block): New function extracted from...
13773 (fix_up_crossing_landing_pad): ...here. Rename into...
13774 (dw2_fix_up_crossing_landing_pad): ...this.
13775 (sjlj_fix_up_crossing_landing_pad): New function.
13776 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
13777 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
13778 from both partitions and exit the loop after one iteration.
13780 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13783 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13785 (Basic PowerPC Built-in Functions): The new name of the
13786 subsection previously known as "PowerPC Built-in Functions".
13787 (Basic PowerPC Built-in Functions Available on all Configurations):
13789 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13791 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13793 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13795 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13798 2018-05-08 Jim Wilson <jimw@sifive.com>
13800 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
13801 (LD_EMUL_SUFFIX): New.
13802 (LINK_SPEC): Use it.
13804 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13806 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13808 (Basic PowerPC Built-in Functions): The new name of the
13809 subsection previously known as "PowerPC Built-in Functions".
13810 (Basic PowerPC Built-in Functions Available on all Configurations):
13812 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13814 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13816 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13818 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13821 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13824 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
13825 after cmpelim optimization.
13827 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
13829 * config.gcc: Support "goldmont".
13830 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
13831 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13832 PROCESSOR_GOLDMONT.
13833 * config/i386/i386.c (m_GOLDMONT): Define.
13834 (processor_target_table): Add "goldmont".
13835 (PTA_GOLDMONT): Define.
13836 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
13837 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
13838 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
13839 (fold_builtin_cpu): Add "goldmont".
13840 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
13841 (ix86_option_override_internal): Add "goldmont".
13842 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
13843 (processor_type): Add PROCESSOR_GOLDMONT.
13844 * config/i386/i386.md: Add CPU "glm".
13845 * config/i386/glm.md: New file.
13846 * config/i386/x86-tune.def: Add m_GOLDMONT.
13847 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
13849 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13852 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
13854 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
13855 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
13856 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
13859 * config/i386/i386.c (ix86_fold_builtin): Handle
13860 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
13863 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
13864 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
13866 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
13869 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
13870 (check_arch): Likewise.
13871 (check_fpu): Return the result rather than printing it.
13872 (end arch): Fix operator precedence.
13873 (end cpu): Likewise.
13874 (END): Print the result from check_fpu.
13876 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13877 Alan Hayward <alan.hayward@arm.com>
13878 David Sherwood <david.sherwood@arm.com>
13880 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
13881 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
13882 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
13883 (*fcmuo<mode>_and): New patterns.
13885 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13887 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
13888 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
13889 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
13890 (cmp_op, sve_imm_con): New code attributes.
13891 (SVE_COND_INT_CMP, imm_con): Delete.
13892 (cmp_op): Remove above unspecs from int attribute.
13893 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
13895 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
13896 comparison-specific unspecs.
13897 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
13898 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
13899 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
13900 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
13901 (*vec_fcm<cmp_op><mode>): Rename to...
13902 (*fcm<cmp_op><mode>): ...this and adjust likewise.
13903 (*vec_fcmuo<mode>): Rename to...
13904 (*fcmuo<mode>): ...this and adjust likewise.
13905 (*pred_fcm<cmp_op><mode>): New pattern.
13906 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
13907 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
13909 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
13911 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
13912 (aarch64_emit_sve_predicated_cond): New function.
13913 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
13914 (aarch64_emit_unspec_cond_or): Replace with...
13915 (aarch64_emit_sve_or_conds): ...this new function. Use
13916 aarch64_emit_sve_ptrue_op for the individual comparisons and
13917 aarch64_emit_binop to OR them together.
13918 (aarch64_emit_inverted_unspec_cond): Replace with...
13919 (aarch64_emit_sve_inverted_cond): ...this new function. Use
13920 aarch64_emit_sve_ptrue_op for the comparison and
13921 aarch64_emit_unop to invert the result.
13922 (aarch64_expand_sve_vec_cmp_float): Update after the above
13923 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
13925 2018-05-07 Nathan Sidwell <nathan@acm.org>
13927 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
13928 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
13929 (Backwards Compatibility): Likewise.
13931 2018-05-07 Luis Machado <luis.machado@linaro.org>
13935 2018-05-07 Luis Machado <luis.machado@linaro.org>
13937 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13938 <prefetch_dynamic_strides>: New const bool field.
13939 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13940 prefetch_dynamic_strides.
13941 (exynosm1_prefetch_tune): Likewise.
13942 (thunderxt88_prefetch_tune): Likewise.
13943 (thunderx_prefetch_tune): Likewise.
13944 (thunderx2t99_prefetch_tune): Likewise.
13945 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13947 (aarch64_override_options_internal): Update to set
13948 PARAM_PREFETCH_DYNAMIC_STRIDES.
13949 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13950 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13951 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13952 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13953 prefetch-dynamic-strides setting.
13955 2018-05-07 Luis Machado <luis.machado@linaro.org>
13957 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13958 <minimum_stride>: New const int field.
13959 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13960 minimum_stride field.
13961 (exynosm1_prefetch_tune): Likewise.
13962 (thunderxt88_prefetch_tune): Likewise.
13963 (thunderx_prefetch_tune): Likewise.
13964 (thunderx2t99_prefetch_tune): Likewise.
13965 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13966 (aarch64_override_options_internal): Update to set
13967 PARAM_PREFETCH_MINIMUM_STRIDE.
13968 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13969 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13970 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13971 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13972 stride is constant and is below the minimum stride threshold.
13974 2018-05-07 Luis Machado <luis.machado@linaro.org>
13976 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
13979 2018-05-07 Luis Machado <luis.machado@linaro.org>
13981 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13982 <prefetch_dynamic_strides>: New const bool field.
13983 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13984 prefetch_dynamic_strides.
13985 (exynosm1_prefetch_tune): Likewise.
13986 (thunderxt88_prefetch_tune): Likewise.
13987 (thunderx_prefetch_tune): Likewise.
13988 (thunderx2t99_prefetch_tune): Likewise.
13989 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13991 (aarch64_override_options_internal): Update to set
13992 PARAM_PREFETCH_DYNAMIC_STRIDES.
13993 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13994 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13995 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13996 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13997 prefetch-dynamic-strides setting.
13999 2018-05-07 Luis Machado <luis.machado@linaro.org>
14001 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14002 <minimum_stride>: New const int field.
14003 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14004 minimum_stride field.
14005 (exynosm1_prefetch_tune): Likewise.
14006 (thunderxt88_prefetch_tune): Likewise.
14007 (thunderx_prefetch_tune): Likewise.
14008 (thunderx2t99_prefetch_tune): Likewise.
14009 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
14010 (aarch64_override_options_internal): Update to set
14011 PARAM_PREFETCH_MINIMUM_STRIDE.
14012 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
14013 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
14014 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
14015 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
14016 stride is constant and is below the minimum stride threshold.
14018 2018-05-06 Jakub Jelinek <jakub@redhat.com>
14021 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
14022 the type is addressable. Don't force op into register if it has
14025 2018-05-05 Roland McGrath <mcgrathr@google.com>
14028 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
14029 any section for which we don't know a specific type it should have,
14030 regardless of name. Previously this was done only for the exact
14031 names ".init_array", ".fini_array", and ".preinit_array".
14032 (default_elf_asm_named_section): Add comment about
14033 relationship with default_section_type_flags and SECTION_NOTYPE.
14034 (get_section): Don't consider it a type conflict if one side has
14035 SECTION_NOTYPE and the other doesn't, as long as neither has the
14036 SECTION_BSS et al used in the default_section_type_flags logic.
14038 2018-05-05 Tom de Vries <tom@codesourcery.com>
14041 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
14042 (workaround_barsyncs): New function.
14043 (nvptx_reorg): Use workaround_barsyncs.
14044 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
14045 (define_expand "nvptx_membar_cta"): New define_expand.
14046 (define_insn "*nvptx_membar_cta"): New insn.
14048 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14050 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
14051 To improve optimization opportunities.
14052 * builtin-types.def: The new needed builtin types for the above.
14054 2018-05-04 Richard Biener <rguenther@suse.de>
14056 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
14057 * gimple-ssa-store-merging.c
14058 (imm_store_chain_info::output_merged_store): Remove redundant create,
14059 release split_store vector contents on failure.
14060 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
14061 scalar stmt vector on cache hit.
14063 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
14065 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14067 * config.gcc (powerpc-xilinx-eabi*): Remove.
14068 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
14070 (fusion_addis_mem_combo_load): Ditto.
14071 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
14073 (rs6000_cpu_cpp_builtins): Ditto.
14074 * config/rs6000/rs6000-linux.c
14075 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
14076 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
14077 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
14079 (rs6000_setup_reg_addr_masks): Ditto.
14080 (rs6000_init_hard_regno_mode_ok): Ditto.
14081 (rs6000_option_override_internal): Ditto.
14082 (legitimate_lo_sum_address_p): Ditto.
14083 (rs6000_legitimize_address): Ditto.
14084 (rs6000_legitimize_reload_address): Ditto.
14085 (rs6000_legitimate_address_p): Ditto.
14086 (abi_v4_pass_in_fpr): Ditto.
14087 (setup_incoming_varargs): Ditto.
14088 (rs6000_gimplify_va_arg): Ditto.
14089 (rs6000_split_multireg_move): Ditto.
14090 (rs6000_savres_strategy): Ditto.
14091 (rs6000_emit_prologue_components): Ditto.
14092 (rs6000_emit_epilogue_components): Ditto.
14093 (rs6000_emit_prologue): Ditto.
14094 (rs6000_emit_epilogue): Ditto.
14095 (rs6000_elf_file_end): Ditto.
14096 (rs6000_function_value): Ditto.
14097 (rs6000_libcall_value): Ditto.
14098 * config/rs6000/rs6000.h: Ditto.
14099 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
14100 (TARGET_MINMAX): ... this. New.
14101 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
14102 * config/rs6000/rs6000.md: Remove Xilinx FP support.
14103 (*movsi_internal1_single): Delete.
14104 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
14105 mfpu=, mxilinx-fpu): Delete.
14106 * config/rs6000/singlefp.h: Delete.
14107 * config/rs6000/sysv4.h: Remove Xilinx FP support.
14108 * config/rs6000/t-rs6000: Ditto.
14109 * config/rs6000/t-xilinx: Delete.
14110 * config/rs6000/titan.md: Adjust for fp_type removal.
14111 * config/rs6000/vsx.md: Remove Xilinx FP support.
14112 (VStype_simple): Delete.
14113 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
14114 * config/rs6000/xfpu.h: Delete.
14115 * config/rs6000/xfpu.md: Delete.
14116 * config/rs6000/xilinx.h: Delete.
14117 * config/rs6000/xilinx.opt: Delete.
14118 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
14119 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
14121 2018-05-04 Tom de Vries <tom@codesourcery.com>
14124 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
14127 2018-05-04 Richard Biener <rguenther@suse.de>
14129 PR middle-end/85627
14130 * tree-complex.c (update_complex_assignment): We are always in SSA form.
14131 (expand_complex_div_wide): Likewise.
14132 (expand_complex_operations_1): Likewise.
14133 (expand_complex_libcall): Preserve EH info of the original stmt.
14134 (tree_lower_complex): Handle removed blocks.
14135 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
14136 on complex multiplication and division libcall builtins.
14138 2018-05-04 Richard Biener <rguenther@suse.de>
14140 PR middle-end/85574
14141 * fold-const.c (negate_expr_p): Restrict negation of operand
14142 zero of a division to when we know that can happen without
14144 (fold_negate_expr_1): Likewise.
14146 2018-05-04 Jakub Jelinek <jakub@redhat.com>
14149 * real.h (real_nextafter): Declare.
14150 * real.c (real_nextafter): New function.
14151 * fold-const-call.c (fold_const_nextafter): New function.
14152 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
14153 CASE_CFN_NEXTTOWARD.
14154 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
14155 even when arg1_mode is different from arg0_mode.
14157 2018-05-03 Nathan Sidwell <nathan@acm.org>
14159 * doc/extend.texi (Deprecated Features): Remove
14160 -ffriend-injection.
14161 (Backwards Compatibility): Likewise.
14162 * doc/invoke.texi (C++ Language Options): Likewise.
14163 (C++ Dialect Options): Likewise.
14165 2018-05-03 Jakub Jelinek <jakub@redhat.com>
14168 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
14169 _mm512_mask_mullox_epi64): New intrinsics.
14171 2018-05-03 Tom de Vries <tom@codesourcery.com>
14174 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14175 dump files): Add offload-tree.
14177 2018-05-03 Richard Biener <rguenther@suse.de>
14179 PR tree-optimization/85615
14180 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
14181 to loops not nested in BBs loop father to avoid creating multi-entry
14184 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14186 PR tree-optimization/70291
14187 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
14188 arguments. Change return type to tree. Emit libcall as a new
14189 statement rather than replacing existing one when inplace_p is true.
14190 (expand_complex_multiplication_components): New function.
14191 (expand_complex_multiplication): Expand floating-point complex
14192 multiplication using the above.
14193 (expand_complex_division): Rename inner_type parameter to type.
14194 Update expand_complex_libcall call-site.
14195 (expand_complex_operations_1): Update expand_complex_multiplication
14196 and expand_complex_division call-sites.
14198 2018-05-02 Jakub Jelinek <jakub@redhat.com>
14201 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
14202 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
14203 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
14204 the highest significant bit of the shift count mask is clear. In
14205 check whether and[sq]i3 is needed verify that all significant bits
14206 of the shift count other than the highest are set.
14208 2018-05-02 Tom de Vries <tom@codesourcery.com>
14211 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
14212 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
14213 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
14214 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
14215 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
14216 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
14217 __builtin_goacc_parlevel_size.
14219 2018-05-02 Richard Biener <rguenther@suse.de>
14221 PR tree-optimization/85597
14222 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
14223 do not use split vect_get_vec_defs call but call vect_get_slp_defs
14226 2018-05-02 Tom de Vries <tom@codesourcery.com>
14229 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14230 dump files): Add ltrans-tree.
14232 2018-05-02 Tom de Vries <tom@codesourcery.com>
14235 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14236 dump files): Add wpa-ipa.
14238 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
14240 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
14241 powerpc*-*-linux*paired* target.
14242 * config/rs6000/750cl.h: Delete.
14243 * config/rs6000/paired.h: Delete.
14244 * config/rs6000/paired.md: Delete.
14245 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
14247 * config/rs6000/rs6000-builtin.def: Remove paired float support.
14248 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14249 comment. Remove paired float support.
14250 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
14251 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
14253 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
14254 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
14256 * config/rs6000/rs6000.c: Remove paired float support.
14257 (paired_expand_vector_init, paired_expand_vector_move,
14258 paired_emit_vector_compare, paired_emit_vector_cond_expr,
14259 (paired_expand_lv_builtin, paired_expand_stv_builtin,
14260 paired_expand_builtin, paired_expand_predicate_builtin,
14261 paired_init_builtins): Delete.
14262 * config/rs6000/rs6000.h: Remove paired float support.
14263 * config/rs6000/rs6000.md: Remove paired float support.
14264 (move_from_CR_ov_bit): Delete.
14265 * config/rs6000/rs6000.opt (mpaired): Delete.
14266 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
14267 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
14269 2018-05-02 Richard Biener <rguenther@suse.de>
14271 PR middle-end/85567
14272 * gimplify.c (gimplify_save_expr): When in SSA form allow
14273 SAVE_EXPRs to compute to SSA vars.
14275 2018-05-02 Jakub Jelinek <jakub@redhat.com>
14278 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
14279 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
14280 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
14281 clobber operands[2], instead use a new pseudo. Formatting fixes.
14283 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
14285 PR tree-optimization/85586
14286 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
14287 exit early for statements in the same group if the accesses are
14290 2018-05-02 Tom de Vries <tom@codesourcery.com>
14293 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
14296 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
14298 PR tree-optimization/85143
14299 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
14301 2018-05-01 Tom de Vries <tom@codesourcery.com>
14304 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
14305 not found" error message.
14307 2018-05-01 Tom de Vries <tom@codesourcery.com>
14310 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
14311 * vec.c (test_ordered_remove_if): New function.
14312 (vec_c_tests): Call test_ordered_remove_if.
14313 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
14314 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
14315 * tree-vect-patterns.c (vect_pattern_recog_1): Use
14316 VEC_ORDERED_REMOVE_IF.
14318 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14320 PR tree-optimization/82665
14321 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
14322 pointer subtraction where arguments come from a memchr call.
14324 2018-05-01 Jakub Jelinek <jakub@redhat.com>
14326 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
14327 --push-state --as-needed and --pop-state instead of --as-needed and
14328 --no-as-needed if ld supports it.
14329 * configure: Regenerated.
14332 * doc/install.texi2html: Replace _002d with - and _002a with * in
14333 generated html files using sed.
14335 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14338 * gcc-rich-location.c (blank_line_before_p): New function.
14339 (use_new_line): New function.
14340 (gcc_rich_location::add_fixit_insert_formatted): New function.
14341 * gcc-rich-location.h
14342 (gcc_rich_location::add_fixit_insert_formatted): New function.
14344 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14346 * selftest.c (assert_streq): Rename "expected" and "actual" to
14347 "val1" and "val2". Extend NULL-handling to cover both inputs
14348 symmetrically, while still requiring both to be non-NULL for a pass.
14349 * selftest.h (assert_streq): Rename "expected" and "actual" to
14351 (ASSERT_EQ): Likewise.
14352 (ASSERT_EQ_AT): Likewise.
14353 (ASSERT_KNOWN_EQ): Likewise.
14354 (ASSERT_KNOWN_EQ_AT): Likewise.
14355 (ASSERT_NE): Likewise.
14356 (ASSERT_MAYBE_NE): Likewise.
14357 (ASSERT_MAYBE_NE_AT): Likewise.
14358 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
14359 the assertion to pass.
14360 (ASSERT_STREQ_AT): Likewise.
14362 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
14364 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
14365 interaction with -pie.
14367 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14369 * selftest.h: Fix alphabetization of per-source-file selftest
14372 2018-04-30 Jason Merrill <jason@redhat.com>
14374 PR c++/61982 - dead stores to destroyed objects.
14375 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
14378 2018-04-30 Jason Merrill <jason@redhat.com>
14380 * tree.c (build_clobber): New.
14381 * tree.h: Declare it.
14382 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
14384 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14386 * diagnostic-show-locus.c (layout::layout): Update for
14387 location_get_source_line returning a char_span.
14388 (struct char_span): Move to input.h.
14389 (struct correction): Update for fields in char_span becoming
14391 (struct source_line): Update for location_get_source_line
14392 returning a char_span.
14393 (layout::print_line): Likewise.
14394 * edit-context.c (edited_file::print_content): Likewise.
14395 (edited_file::print_diff_hunk): Likewise.
14396 (edited_file::print_run_of_changed_lines): Likewise.
14397 (edited_file::get_num_lines): Likewise.
14398 (edited_line::edited_line): Likewise.
14399 * final.c (asm_show_source): Likewise.
14400 * input.c (location_get_source_line): Convert return type
14401 from const char * to char_span, losing the final "line_len"
14403 (dump_location_info): Update for the above.
14404 (get_substring_ranges_for_loc): Likewise. Use a char_span
14405 when handling the literal within the line.
14406 (test_reading_source_line): Update for location_get_source_line
14407 returning a char_span.
14408 * input.h (class char_span): Move here from
14409 diagnostic-show-locus.c, converting from a struct to a class.
14410 Make data members private.
14411 (char_span::operator bool): New.
14412 (char_span::length): New.
14413 (char_span::get_buffer): New.
14414 (char_span::operator[]): New.
14415 (char_span::subspan): Make const.
14416 (char_span::xstrdup): New.
14417 (location_get_source_line): Convert return type from const char *
14418 to char_span, losing the final "line_size" param.
14420 2018-04-30 Jan Hubicka <jh@suse.cz>
14422 * lto-wrapper.c (ltrans_priorities): New static var.
14423 (cmp_priority): New.
14424 (run_gcc): Read priorities and if doing parallel build order
14425 the Makefile by them.
14427 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14429 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
14431 2018-04-30 Richard Biener <rguenther@suse.de>
14433 * tree-cfg.c (verify_address): Remove base argument, add
14434 flag whether to check TREE_ADDRESSABLE and do that.
14435 (verify_expr): Remove.
14436 (verify_types_in_gimple_reference): Add pieces from verify_expr.
14437 (verify_gimple_assign_single): Likewise.
14438 (verify_gimple_switch): Likewise.
14439 (verify_expr_location_1): Dereference tp once. Add (disabled)
14440 piece from verify_expr.
14441 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
14443 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14445 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
14447 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14449 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
14450 (small_data_pattern): Likewise.
14451 (arc_rewrite_small_data): Likewise.
14452 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
14453 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
14454 (get_symbol_alignment): New function.
14455 (legitimate_small_data_address_p): Likewise.
14456 (legitimate_scaled_address): Update, call
14457 legitimate_small_data_address_p.
14458 (output_sdata): New static variable.
14459 (arc_print_operand): Update how we handle small data operands.
14460 (arc_print_operand_address): Likewise.
14461 (arc_legitimate_address_p): Update, use
14462 legitimate_small_data_address_p.
14463 (arc_rewrite_small_data_p): Remove.
14464 (arc_rewrite_small_data_1): Likewise.
14465 (arc_rewrite_small_data): Likewise.
14466 (small_data_pattern): Likewise.
14467 (compact_sda_memory_operand): Update to use
14468 legitimate_small_data_address_p and get_symbol_alignment.
14469 (prepare_move_operands): Don't rewite sdata pattern.
14470 (prepare_extend_operands): Remove.
14471 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
14473 (zero_extendqisi2): Likewise.
14474 (zero_extendhisi2): Likewise.
14475 (extendqihi2): Likewise.
14476 (extendqisi2): Likewise.
14477 (extendhisi2): Likewise.
14478 (addsi3): Likewise.
14479 (subsi3): Likewise.
14480 (andsi3): Likewise.
14481 * config/arc/constraints.md (Usd): Change it to memory constraint.
14483 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14485 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
14486 as source of std instructions.
14487 * config/arc/arc.md (movsi_insn): Update pattern predicate to
14488 allow 6-bit constants as source for store instructions.
14489 (movdi_insn): Update instruction pattern to allow 6-bit constants
14490 as source for store instructions.
14492 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
14494 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
14496 2018-04-30 Nathan Sidwell <nathan@acm.org>
14497 Sandra Loosemore <sandra@codesourcery.com>
14499 * dumpfile.c (dump_open): Allow '-' for stdout.
14500 * doc/invoke.texi (Developer Options): Document dump filename
14501 determination early. Document stdin/stdout selection.
14503 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
14505 Microblaze Target: PIC data text relative
14507 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
14508 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
14510 * config/microblaze/microblaze.h (microblaze_constant_address_p):
14511 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
14512 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
14513 New addressing mode for data-text relative position indepenedent code.
14514 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
14515 'ADDRESS_SYMBOLIC_TXT_REL'.
14516 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
14517 (microblaze_legitimate_pic_operand): Exclude function calls from
14518 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
14519 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
14521 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
14522 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
14523 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
14524 for 'address + offset'.
14525 (microblaze_expand_prologue): Add new function prologue call for
14527 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
14528 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
14529 table in case of TARGET_PIC_DATA_TEXT_REL.
14530 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
14531 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
14532 Add new macros 'UNSPEC_TEXT',
14533 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
14534 + exclude function calls from 'UNSPEC_PLT' in case of data text
14536 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
14537 new target hook for generating address diff vector tables in case of
14539 * doc/tm.texi : Regenerate.
14540 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
14541 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
14542 of addr diff vector generation.
14543 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
14544 target hook definition.
14545 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
14546 Add default function for generate_pic_addr_diff_vec -> flag_pic.
14547 * doc/invoke.texi (Add new pic option): Add new microblaze pic
14548 option for data text relative.
14550 2018-04-30 Richard Biener <rguenther@suse.de>
14552 * tree-chrec.h (evolution_function_is_constant_p): Remove
14554 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
14556 2018-04-30 Richard Biener <rguenther@suse.de>
14559 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
14561 2018-04-30 Richard Biener <rguenther@suse.de>
14563 PR tree-optimization/28364
14564 PR tree-optimization/85275
14565 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
14566 copying first exit test.
14568 2018-04-28 Mark Wielaard <mark@klomp.org>
14570 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
14571 dwarf_version >= 5.
14572 (dwarf_AT): Handle DW_AT_addr_base.
14573 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
14575 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
14578 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
14579 (*ashl<dwi>3_doubleword_mask_1): Ditto.
14580 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
14581 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
14583 2018-04-28 Richard Biener <rguenther@suse.de>
14585 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
14586 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
14587 to reflect use. Only add interesting stmts.
14589 2018-04-27 Martin Jambor <mjambor@suse.cz>
14592 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
14593 the jump function allows for passing through aggregate values.
14595 2018-04-27 David Malcolm <dmalcolm@redhat.com>
14597 * input.h (in_system_header_at): Convert from macro to inline
14599 (from_macro_expansion_at): Likewise.
14600 (from_macro_definition_at): Likewise.
14602 2018-04-27 Jeff Law <law@redhat.com>
14604 * config.gcc: Mark tile* targets as deprecated/obsolete.
14606 2018-04-27 Richard Biener <rguenther@suse.de>
14608 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
14611 2018-04-27 Richard Biener <rguenther@suse.de>
14613 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
14615 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
14617 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
14618 with Yd constraint. Set "preferred_for_speed" attribute from
14619 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
14620 with Yd constraint.
14621 (*movdi_internal): Ditto.
14622 (movti_interunit splitters): Remove
14623 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
14624 (movdi_interunit splitters): Ditto.
14625 * config/i386/constraints.md (Ye): Remove.
14626 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
14628 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14631 * config/aarch64/constraints.md (Usg): Limit to 31.
14632 (Usj): Limit to 63.
14634 2018-04-27 Jakub Jelinek <jakub@redhat.com>
14636 PR tree-optimization/85529
14637 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
14638 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
14639 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
14640 zero extension or masking of the MSB bit.
14641 (optimize_range_tests): Add FIRST_BB argument, pass it through
14642 to optimize_range_tests_var_bound.
14643 (maybe_optimize_range_tests, reassociate_bb): Adjust
14644 optimize_range_tests callers.
14646 2018-04-26 Richard Biener <rguenther@suse.de>
14647 Jakub Jelinek <jakub@redhat.com>
14649 * cgraph.h (symbol_table): Just declare debug method here.
14650 * symtab.c (symbol_table::debug): Define.
14652 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
14654 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
14656 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
14658 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
14659 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
14660 (*movdi_internal): Substitute Yi and Yj constraint with x
14661 and Ym and Yn constraint with y constraint. Update "isa"
14662 attribute and set "preferred_for_speed" attribute from
14663 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
14664 (*movsi_internal): Ditto.
14665 (*movdf_internal): Ditto.
14666 (*movsf_internal): Ditto.
14667 (*zero_extendsidi2): Ditto.
14668 * config/i386/sse.md (vec_set<mode>_0): Ditto.
14669 (sse2_loadld): Ditto.
14670 (*vec_extract<ssevecmodelower>_0): Ditto.
14671 (*vec_extractv4si_0_zext_sse4): Ditto.
14672 (vec_concatv2di): Ditto.
14673 (*vec_dup<mode>): Ditto.
14674 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
14675 * config/i386/constraints.md (Yi): Remove.
14680 2018-04-26 Nathan Sidwell <nathan@acm.org>
14682 * dumpfile.c (dump_open): New.
14683 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
14684 (dump_finish): Detect stdio/stderr by value not name.
14686 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
14688 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
14690 2018-04-26 Tom de Vries <tom@codesourcery.com>
14693 * config/nvptx/nvptx.c (verify_neutering_jumps)
14694 (verify_neutering_labels): New function
14695 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
14697 2018-04-26 Tom de Vries <tom@codesourcery.com>
14700 * config/nvptx/nvptx.c (needs_neutering_p): New function.
14701 (nvptx_single): Use needs_neutering_p to skip over insns that do not
14704 2018-04-26 Richard Biener <rguenther@suse.de>
14705 Tom de Vries <tom@codesourcery.com>
14708 * lto-streamer-out.c (output_function): Fixup loops if required to match
14709 discovery done in the reader.
14711 2018-04-26 Richard Biener <rguenther@suse.de>
14713 PR tree-optimization/85116
14714 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
14715 have a loop exit from the single latch predecessor. Remove
14716 case of header with just condition.
14717 (ch_base::copy_headers): Exclude infinite loops from any
14719 (pass_ch::execute): Record exits.
14721 2018-04-26 Richard Biener <rguenther@suse.de>
14723 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
14724 prologue cost vector and pass it to vect_get_load_cost.
14725 (vect_get_peeling_costs_all_drs): Likewise.
14726 (vect_peeling_hash_get_lowest_cost): Likewise.
14727 (vect_enhance_data_refs_alignment): Likewise.
14729 2018-04-26 Richard Biener <rguenther@suse.de>
14731 PR middle-end/85450
14732 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
14733 checking of integer<->pointer conversions.
14734 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
14735 sign-/zero-extending pointer types.
14736 (expand_omp_for_static_chunk): Likewise.
14738 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
14739 Jean Lee <xiaoyur347@gmail.com>
14741 * config/mips/mips.c (mips_asan_shadow_offset): New function.
14742 (TARGET_ASAN_SHADOW_OFFSET): Define.
14743 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
14744 true for -fsanitize=address.
14746 2018-04-25 Mark Wielaard <mark@klomp.org>
14748 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
14751 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14753 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
14754 than "alu", remove explicit "memory" and "imm_disp" attributes.
14755 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
14757 PR middle-end/85414
14758 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
14759 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
14760 gen_lowpart_no_emit.
14762 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
14765 * config/i386/i386.c (ix86_expand_builtin): Change memory
14766 operand to XI, extend p0 to Pmode.
14767 * config/i386/i386.md: Change unspec volatile and operand
14768 1 mode to XI, change operand 0 mode to P.
14770 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14772 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
14773 GET_MODE_MASK before any checking.
14774 (nds32_can_use_bset_p): Likewise.
14775 (nds32_can_use_btgl_p): Likewise.
14777 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14779 * config/nds32/nds32-doubleword.md: New define_split pattern for
14780 illegal register number.
14782 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14784 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
14786 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14788 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
14790 2018-04-25 Richard Biener <rguenther@suse.de>
14792 * lto-streamer.h (LTO_major_version): Bump to 8.
14794 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14796 * BASE-VER: Set to 9.0.0.
14798 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
14800 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
14801 in __abskf2 and __powikf2.
14803 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14806 * config/aarch64/constraints.md (Usg, Usj): New constraints.
14807 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
14808 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
14809 Use the above on operand 2. Reindent.
14810 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
14812 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14815 * common/config/i386/i386-common.c (ix86_handle_option): Don't
14817 * config/i386/i386.opt (mcet): Removed.
14818 * doc/install.texi: Remove -mcet documentation.
14819 * doc/invoke.texi: Likewise.
14821 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14824 * doc/install.texi: Remove -mcet from bootstrap-cet.
14826 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14829 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
14830 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
14834 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
14835 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
14836 containing a CONST_VECTOR.
14838 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
14840 * doc/install.texi: Update newlib dependency for nvptx.
14842 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14845 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
14846 instead of INTVAL when shifting x left.
14848 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
14850 PR tree-optimization/85478
14851 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
14852 vect_grouped_store_supported for single element vectors.
14854 2018-04-24 Richard Biener <rguenther@suse.de>
14857 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
14858 load cost increase to the case of non-constant step.
14860 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14863 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
14864 destination if any_malformed_asm.
14866 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
14868 PR middle-end/85496
14869 * expr.c (store_field): In the bitfield case, if the value comes from
14870 a function call and is returned in registers by means of a PARALLEL,
14871 do not change the mode of the temporary unless BLKmode and VOIDmode.
14873 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
14875 PR rtl-optimization/85423
14876 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
14877 dependencies to debug insns when the previous insn is non-debug.
14879 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14881 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
14882 enums into a single definition.
14883 (fls): Fix predicates and printing.
14886 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14888 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
14889 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
14890 and short u6 immediate.
14891 (check_if_valid_sleep_operand): Remove.
14892 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
14894 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14896 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
14897 flag_always_save_lp condition.
14898 * config/nds32/nds32.opt (malways-save-lp): New option.
14900 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14902 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
14903 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
14904 * config/nds32/nds32.h
14905 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
14906 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
14908 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14910 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
14911 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
14913 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14914 Chung-Ju Wu <jasonwucj@gmail.com>
14916 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
14918 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
14919 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
14921 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14923 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
14925 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14927 * config/nds32/nds32-protos.h (nds32_data_alignment,
14928 nds32_local_alignment): Declare.
14929 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
14930 nds32_local_alignment): New functions.
14931 (TARGET_CONSTANT_ALIGNMENT): Define.
14932 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
14934 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14936 * config/nds32/nds32.c
14937 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
14938 (TARGET_MODES_TIEABLE_P): Likewise.
14940 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14942 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
14943 level Ofast and Og.
14945 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
14946 Chung-Ju Wu <jasonwucj@gmail.com>
14948 * config/nds32/constants.md (unspec_volatile_element): Add enum values
14949 for unaligned access.
14950 * config/nds32/nds32-intrinsic.c: Implementation of expanding
14952 * config/nds32/nds32-intrinsic.md: Likewise.
14953 * config/nds32/nds32_intrinsic.h: Likewise.
14954 * config/nds32/nds32.h (nds32_builtins): Likewise.
14955 * config/nds32/nds32.opt (munaligned-access): New option.
14956 * config/nds32/nds32.c (nds32_asm_file_start): Display
14957 flag_unaligned_access status.
14959 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14961 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
14962 -mno-relax is present.
14963 * config/riscv/linux.h (LINK_SPEC): Ditto.
14965 2018-04-20 Martin Sebor <msebor@redhat.com>
14968 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
14970 (gimple_fold_builtin_stxcpy_chk): Same.
14971 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
14973 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
14976 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
14977 __powikf2 when long double is IEEE 128-bit.
14979 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14981 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
14982 step to make sure stack always aligned.
14984 2018-04-20 Carl Love <cel@us.ibm.com>
14987 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
14988 size check for arg0.
14990 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
14991 Tom de Vries <tom@codesourcery.com>
14994 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
14995 Emit insns for calls too.
14996 (nvptx_find_par): Always look for worker-level predecessor insn.
14997 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
14999 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
15000 (nvptx_process_pars): Propagate frames for calls.
15002 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
15005 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
15007 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
15008 (ix86_handle_option): Don't handle OPT_mibt.
15009 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
15011 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
15013 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
15014 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
15015 (ix86_target_macros): Define __CET__ with flag_cf_protection
15016 for -fcf-protection.
15017 * config/i386/i386.c (isa2_opts): Remove -mibt.
15018 * config/i386/i386.h (TARGET_IBT): Removed.
15019 (TARGET_IBT_P): Likewise.
15020 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
15021 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
15022 * config/i386/i386.opt (mcet): Update help message.
15023 (mshstk): Likewise.
15025 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
15026 -mcet as an alias for -mshstk.
15028 2018-04-20 Richard Biener <rguenther@suse.de>
15030 PR middle-end/85475
15031 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
15032 complexity by forcing a single use of the multiply operand.
15034 2018-04-20 Martin Jambor <mjambor@suse.cz>
15037 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
15038 recursion dependency to only apply to non-clones.
15040 2018-04-20 Martin Jambor <mjambor@suse.cz>
15043 * ipa-cp.c (create_specialized_node): Check that clones of
15044 self-recursive edges exist during IPA-CP.
15046 2018-04-19 Toon Moene <toon@moene.org>
15048 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
15051 2018-04-19 Jakub Jelinek <jakub@redhat.com>
15053 PR tree-optimization/85467
15054 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
15055 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
15056 VECTOR_CST element to type.
15058 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
15061 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
15062 * config/i386/i386.md (builtin_setjmp_setup): Removed.
15063 (builtin_longjmp): Likewise.
15064 (save_stack_nonlocal): New pattern.
15065 (restore_stack_nonlocal): Likewise.
15067 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
15070 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
15071 Replace ASM_OUTPUT_LABEL with fprintf.
15073 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
15076 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
15077 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
15078 * config/i386/i386-c.c (ix86_target_macros_internal): Also
15079 define __IBT__ and __SHSTK__ for -fcf-protection.
15080 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
15082 (ix86_trampoline_init): Likewise.
15083 (x86_output_mi_thunk): Likewise.
15084 (ix86_notrack_prefixed_insn_p): Likewise.
15085 (ix86_option_override_internal): Don't disallow -fcf-protection.
15086 * config/i386/i386.md (rdssp<mode>): Also enable for
15088 (incssp<mode>): Likewise.
15089 (nop_endbr): Likewise.
15090 * config/i386/i386.opt (mcet): Change help message to built-in
15093 (mshstk): Likewise.
15094 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
15095 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
15096 enable CET built-in functions.
15098 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
15100 * common/config/i386/i386-common.c
15101 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
15102 OPTION_MASK_ISA_MOVDIRI_UNSET,
15103 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
15104 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
15105 * config.gcc (movdirintrin.h): New header.
15106 * config/i386/cpuid.h (bit_MOVDIRI,
15107 bit_MOVDIR64B): New bits.
15108 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
15110 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
15111 (VOID, PVOID, PCVOID)): New function types.
15112 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
15113 __builtin_ia32_directstoreu_u64,
15114 __builtin_ia32_movdir64b): New builtins.
15115 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
15116 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
15118 (ix86_valid_target_attribute_inner_p): Ditto.
15119 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
15120 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
15121 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
15122 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
15123 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
15124 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
15125 (movdiri<mode>, movdir64b_<mode>): New.
15126 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
15127 * config/i386/immintrin.h: Include movdirintrin.h.
15128 * config/i386/movdirintrin.h: New file.
15129 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
15131 2018-04-19 Richard Biener <rguenther@suse.de>
15133 PR middle-end/85455
15134 * cfg.c (clear_bb_flags): When loop state says we have
15135 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
15137 2018-04-19 Richard Biener <rguenther@suse.de>
15139 PR tree-optimization/84737
15140 * tree-vect-data-refs.c (vect_copy_ref_info): New function
15141 copying restrict info.
15142 (vect_setup_realignment): Use it.
15143 * tree-vectorizer.h (vect_copy_ref_info): Declare.
15144 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
15145 the first DR to all generated stores.
15146 (vectorizable_load): Likewise for loads.
15148 2018-04-19 Jakub Jelinek <jakub@redhat.com>
15150 PR tree-optimization/85446
15151 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
15152 the integral and pointer types to have the same precision.
15154 * doc/install.texi: Document --disable-cet being the default and
15157 2018-04-18 Martin Liska <mliska@suse.cz>
15159 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
15162 2018-04-18 Martin Liska <mliska@suse.cz>
15165 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
15168 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
15169 arguments if they are comparable.
15171 2018-04-18 Martin Liska <mliska@suse.cz>
15174 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
15177 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
15180 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
15183 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
15184 ENDBR after calling __morestack.
15186 2018-04-18 David Malcolm <dmalcolm@redhat.com>
15189 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
15190 by using gcc_base_ver to generate a gcc_driver_version, and use
15191 it when generating GCC_DRIVER_NAME.
15192 * configure: Regenerate.
15194 2018-04-18 Jakub Jelinek <jakub@redhat.com>
15197 * config.gcc: Obsolete powerpc*-*-*spe*.
15199 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15202 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
15203 (stabstr_D): Change type of unum from unsigned int to
15204 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
15207 2018-04-17 Jim Wilson <jimw@sifive.com>
15210 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
15211 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
15212 Set arg_pointer_offset after using pretend_args_size.
15214 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15216 PR rtl-optimization/85431
15217 * dse.c (record_store): Ignore zero width stores.
15220 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
15221 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
15222 __builtin_stack_restore rather than after it.
15223 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
15224 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
15225 argument instead of virtual_dynamic_stack_rtx.
15227 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
15229 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
15231 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15232 Add note to error message to explain internal mapping of overloaded
15233 built-in function name to non-overloaded built-in function name.
15234 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
15237 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
15240 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
15241 where the inputs overlap with the output.
15243 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15245 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
15246 (=v, v) alternative and explicit "memory" attribute.
15247 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
15248 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15250 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
15251 "sselog1" type instead of "sselog".
15252 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
15253 "sselog". Remove explicit "memory" attribute.
15254 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
15255 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15257 (vec_extract_hi_v32hi): Merge all alternatives into one, use
15258 "sselog1" type instead of "sselog". Remove explicit "memory"
15260 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
15261 use "sselog1" type instead of "sselog". Remove explicit "memory"
15263 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
15264 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15266 (vec_extract_hi_v64qi): Merge all alternatives into one, use
15267 "sselog1" type instead of "sselog". Remove explicit "memory"
15269 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
15270 use "sselog1" type instead of "sselog". Remove explicit "memory"
15274 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
15276 PR middle-end/85414
15277 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
15280 2018-04-17 Martin Jambor <mjambor@suse.cz>
15283 * ipa-cp.c (create_specialized_node): Call
15284 expand_all_artificial_thunks if necessary.
15286 2018-04-17 Martin Liska <mliska@suse.cz>
15289 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
15290 in message, remote space in between '_G' and '('.
15292 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15295 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
15296 avx512f_vmcmp<mode>3<round_saeonly_name>,
15297 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
15298 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
15299 avx512f_rndscale<mode><round_saeonly_name>,
15300 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
15301 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
15302 Use %<iptr>2 instead of %2 for -masm=intel.
15303 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
15304 avx512f_vcvttss2usi<round_saeonly_name>,
15305 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
15307 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
15308 avx512f_vcvttsd2usi<round_saeonly_name>,
15309 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
15310 Use %q1 instead of %1 for -masm=intel.
15311 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
15312 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
15313 of %3 for -masm=intel.
15314 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
15316 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
15318 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
15320 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
15321 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
15323 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
15325 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
15326 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
15327 %g1 and one with %0 and %1.
15328 (avx512er_vmrcp28<mode><round_saeonly_name>,
15329 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
15330 %1 for -masm=intel.
15331 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
15332 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
15333 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
15334 of %0 and %{%4%} for -masm=intel.
15335 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
15336 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
15337 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
15338 order of %0 and %{%5%}%{z%} for -masm=intel.
15340 2018-04-17 Jan Hubicka <jh@suse.cz>
15343 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
15345 2018-04-17 Martin Liska <mliska@suse.cz>
15348 * multiple_target.c (create_dispatcher_calls): Set apostrophes
15349 for target_clone error message. Make default implementation
15350 clone to be a local declaration.
15351 (separate_attrs): Add new argument and check for an empty
15353 (expand_target_clones): Handle it.
15354 (ipa_target_clone): Make redirection just for target_clones
15357 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
15358 Tom de Vries <tom@codesourcery.com>
15360 PR middle-end/84955
15361 * omp-expand.c (expand_oacc_for): Add dummy false branch for
15362 tiled basic blocks without omp continue statements.
15364 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
15367 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
15368 vec_extract expression as having side effects to make sure it gets
15371 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
15374 * config/i386/i386.c (get_builtin_code_for_version): Check
15377 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
15380 * config.gcc: Support "skylake".
15381 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15383 * config/i386/i386.c (m_SKYLAKE): Define.
15384 (processor_target_table): Add "skylake".
15385 (ix86_option_override_internal): Add "skylake".
15386 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
15387 PROCESSOR_CANNONLAKE.
15388 (get_builtin_code_for_version): Fix priority for
15389 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
15390 PROCESSOR_SKYLAKE-AVX512.
15391 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
15392 (processor_type): Add PROCESSOR_SKYLAKE.
15394 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
15395 Jason Merrill <jason@redhat.com>
15398 * convert.c (convert_to_integer_1): Use direct recursion for
15399 enumeral types and types with a precision less than the number
15400 of bits in their mode.
15402 2018-04-16 Julia Koval <julia.koval@intel.com>
15405 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
15406 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
15408 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
15411 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
15412 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
15413 and -mno-direct-move.
15415 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
15418 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
15419 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
15420 Ensure negative shifts result in {0}.
15422 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
15424 PR rtl-optimization/79916
15425 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
15426 regs (if any) to define how to gnerate SD moves when LRA is in
15429 2018-04-13 Jakub Jelinek <jakub@redhat.com>
15431 PR rtl-optimization/85393
15432 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
15433 * except.c (expand_dw2_landing_pad_for_region): Make static.
15434 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
15435 a label and unconditional jump to old_bb, rather than
15436 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
15439 PR rtl-optimization/85376
15440 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
15441 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
15442 instead of a specific value.
15444 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
15445 Bin Cheng <bin.cheng@arm.com>
15447 PR tree-optimization/82965
15448 PR tree-optimization/83991
15449 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
15450 by_profile_only parameter.
15451 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
15452 information if the loop was predicted to iterate too many times.
15453 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
15455 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
15458 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
15461 2018-04-13 Martin Liska <mliska@suse.cz>
15462 Jakub Jelinek <jakub@redhat.com>
15464 PR middle-end/81657
15465 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
15466 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
15467 * builtins.c (expand_builtin_memory_copy_args): Use
15468 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
15469 handle dest_addr == pc_rtx.
15471 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15474 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
15475 asked to not generate direct moves.
15476 (fix_trunc<mode>si2_stfiwx): Similar.
15477 (fix_trunc<mode>si2_internal): Similar.
15479 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15482 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
15483 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
15484 lookup if dest in some wider mode is known to be const0_rtx and
15485 if so, record permanent equivalence for it to be ZERO_EXTEND of
15486 the narrower mode destination.
15488 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
15490 * lto-streamer-out.c (output_function): Revert 259346.
15491 * omp-expand.c (expand_oacc_for): Likewise.
15493 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
15495 PR rtl-optimization/85354
15496 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
15497 * sel-sched.c (sel_global_init): ... here.
15499 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
15502 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
15503 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
15504 mode for PE-COFF targets.
15505 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
15506 (i386_pe_asm_lto_end): Likewise.
15507 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
15508 (TARGET_ASM_LTO_END): Likewise.
15509 * config/i386/winnt.c (saved_debug_info_level): New static variable.
15510 (i386_pe_asm_lto_start): New function.
15511 (i386_pe_asm_lto_end): Likewise.
15513 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
15514 Richard Biener <rguenther@suse.de>
15516 PR middle-end/84955
15517 * lto-streamer-out.c (output_function): Fix CFG loop state before
15519 * omp-expand.c (expand_oacc_for): Handle calls to internal
15520 functions like regular functions.
15522 2018-04-12 Richard Biener <rguenther@suse.de>
15525 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
15526 for the early LTO debug to properly generate references to it
15527 during DIE emission. Do not re-use that for the skeleton for
15529 (dwarf2out_early_finish): Likewise.
15531 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15534 * config/i386/sse.md
15535 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
15536 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
15537 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
15538 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
15539 and output is a reg, avoid creating invalid lowpart subreg, but
15540 instead split into a 512-bit move. Don't split if not AVX512VL,
15541 input is xmm16+ reg and output is a mem.
15542 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
15543 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
15544 xmm16+ reg and output is a mem.
15546 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15548 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
15549 also for flag_dwarf2_cfi_asm.
15551 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15553 PR rtl-optimization/85342
15554 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
15555 a bool scalar var inside of the loop instead. Don't try to update
15556 recog_data.operand after failed apply_change_group.
15558 2018-04-12 Tom de Vries <tom@codesourcery.com>
15561 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
15562 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
15563 array with flexible array member as array without given dimension.
15564 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
15565 argument for undefined param to true.
15567 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
15570 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
15571 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
15572 from PowerPC section.
15573 * config/rs6000/sysv4.opt (mcall-): Improve help text.
15574 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
15575 help text that is too long.
15576 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
15577 help text that is too long.
15578 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
15579 help text that is too long.
15581 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
15583 * config/alpha/alpha.md (stack_probe_internal): Rename
15584 from "probe_stack". Update all callers.
15586 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15588 PR rtl-optimization/84566
15589 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
15590 sched_macro_fuse_insns.
15592 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15595 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
15596 (compute_block_dependences): ... from here.
15598 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15600 PR tree-optimization/85331
15601 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
15602 from int to HOST_WIDE_INT.
15604 2018-04-11 Martin Jambor <mjambor@suse.cz>
15607 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
15608 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
15609 not the same as the source val.
15610 (cgraph_edge_brings_value_p): New parameter.
15611 (gather_edges_for_value): Pass destination value to
15612 cgraph_edge_brings_value_p.
15613 (perhaps_add_new_callers): Likewise.
15614 (get_info_about_necessary_edges): Likewise and exclude values brought
15615 only by self-recursive edges.
15616 (create_specialized_node): Redirect only clones of self-calling edges.
15617 (+self_recursive_pass_through_p): New function.
15618 (find_more_scalar_values_for_callers_subset): Use it.
15619 (find_aggregate_values_for_callers_subset): Likewise.
15620 (known_aggs_to_agg_replacement_list): Removed.
15621 (decide_whether_version_node): Re-calculate known constants for all
15622 remaining context clones.
15624 2018-04-11 Richard Biener <rguenther@suse.de>
15627 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
15628 from early DWARF output.
15629 (dwarf2out_early_finish): Output line info unconditionally into
15630 early DWARF and add reference to it.
15632 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15635 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
15636 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
15637 other than V2DFmode using iptr mode attribute.
15638 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
15640 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15642 PR rtl-optimization/84659
15643 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
15645 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15648 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
15650 (output_loc_list): Pass address of a dummy size variable even in the
15651 locview handling loop.
15652 (index_location_lists): Add comment on why skip_loc_list_entry can't
15655 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
15658 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
15661 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
15664 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
15665 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
15666 and -mstring-compare-inline-limit.
15668 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15671 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
15672 for stack clash protection in a register whenever we need it to be in
15675 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15677 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
15678 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
15680 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15683 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
15685 (mlong-double-): Ditto.
15686 * config/rs6000/sysv4.opt (msdata=): Ditto.
15687 (mtls-size=): Ditto.
15689 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15691 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
15692 erroneous entries for
15693 "vector int vec_ldl (int, long int *)", and
15694 "vector unsigned int vec_ldl (int, unsigned long int *)".
15695 Add comments and entries for
15696 "vector bool char vec_ldl (int, bool char *)",
15697 "vector bool short vec_ldl (int, bool short *)",
15698 "vector bool int vec_ldl (int, bool int *)",
15699 "vector bool long long vec_ldl (int, bool long long *)",
15700 "vector pixel vec_ldl (int, pixel *)",
15701 "vector long long vec_ldl (int, long long *)",
15702 "vector unsigned long long vec_ldl (int, unsigned long long *)".
15703 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
15704 type tree bool_long_long_type_node and correct definition of
15705 bool_V2DI_type_node to make reference to this new type tree.
15706 (rs6000_mangle_type): Replace erroneous reference to
15707 bool_long_type_node with bool_long_long_type_node.
15708 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
15709 comments to emphasize sign distinctions for char and int types and
15710 replace RS6000_BTI_bool_long constant with
15711 RS6000_BTI_bool_long_long constant. Also add comment to restrict
15712 use of RS6000_BTI_pixel.
15713 (bool_long_type_node): Remove this macro definition.
15714 (bool_long_long_type_node): New macro definition
15716 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15718 PR rtl-optimization/85300
15719 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
15720 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
15721 simplify_unary_operation fails.
15723 2018-04-10 Martin Liska <mliska@suse.cz>
15725 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
15726 cgraph_edge and ipa_ref.
15728 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15732 * config/i386/sse.md
15733 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
15734 computation of the VEC_MERGE selector from mask.
15735 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
15736 Fix decoding of the VEC_MERGE selector into mask.
15738 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15740 PR tree-optimization/85286
15741 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
15743 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15745 * final.c (final_1): Set insn_last_address as well as
15746 insn_current_address.
15748 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15751 * explow.c (emit_stack_probe): Call validize_mem on memory location
15752 before passing it to gen_probe_stack. Create address operand and
15753 legitimize it for the probe_stack_address case.
15755 2018-04-09 Jan Hubicka <jh@suse.cz>
15758 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
15759 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
15760 * tree.c (free_lang_data_in_type): Fix handling of binfos;
15762 (free_lang_data): Rebuild type inheritance graph.
15764 2018-04-09 Martin Sebor <msebor@redhat.com>
15766 * invoke.texi (-finline-small-functions): Mention other optimization
15768 (-findirect-inlining, -fpartial-inlining): Same.
15769 (-finline-functions-called-once): Same.
15770 (-freorder-blocks-and-partition): Same.
15772 2018-04-09 Jan Hubicka <jh@suse.cz>
15775 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
15776 jumps; choose last target that matches the criteria (i.e.
15777 no partition changes for non-crossing jumps).
15778 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
15779 support for redirecting crossing jumps to non-crossing.
15781 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
15783 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
15784 also for naked functions.
15786 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
15788 * config/arc/arc.md (add_shift): New pattern.
15789 (add_shift2): Likewise.
15790 (sub_shift): Likewise.
15791 (sub_shift_cmp0_noout): Likewise.
15792 (compare_si_ashiftsi): Likewise.
15793 (xbfu_cmp0_noout): New combine pattern.
15794 (xbfu_cmp0"): Likewise.
15795 (movsi_set_cc_insn): Place the predicable variant first.
15796 (commutative_binary_cmp0_noout): Remove clobber.
15797 (commutative_binary_cmp0): New pattern.
15798 (noncommutative_binary_cmp0): Likewise.
15799 (noncommutative_binary_cmp0_noout): Likewise.
15800 (noncommutative_binary_comparison_result_used): Removed.
15801 (rsub_cmp0): New pattern.
15802 (rsub_cmp0_noout): Likewise.
15803 (extzvsi): Changed, keep only meaningful variants.
15804 (SQH, SEZ): New iterators.
15805 (SQH_postfix): New mode attribute.
15806 (SEZ_prefix): New code attribute.
15807 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
15808 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
15809 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
15810 of numerical value.
15811 (noncommutative_operator): Check the availability of barrel
15814 2018-04-09 Richard Biener <rguenther@suse.de>
15816 PR tree-optimization/85284
15817 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
15818 Only use the niter constraining form of simple_iv when the exit
15819 is always executed.
15821 2018-04-09 Tom de Vries <tom@codesourcery.com>
15824 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
15825 (define_expand "*memory_barrier"): New define_expand.
15826 (define_insn "memory_barrier"): New insn.
15828 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15830 PR rtl-optimization/80463
15831 PR rtl-optimization/83972
15832 PR rtl-optimization/83480
15834 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
15835 correct producer for the insn.
15836 (tidy_control_flow): Fixup seqnos in case of debug insns.
15838 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15840 PR rtl-optimization/83913
15842 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
15843 different sched-times when merging exprs.
15845 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15847 PR rtl-optimization/83962
15849 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
15850 tidy_fallthru_edge and tidy_control_flow.
15852 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15854 PR rtl-optimization/83530
15856 * sel-sched.c (force_next_insn): New global variable.
15857 (remove_insn_for_debug): When force_next_insn is true, also leave only
15858 next insn in the ready list.
15859 (sel_sched_region): When the region wasn't scheduled, make another pass
15860 over it with force_next_insn set to 1.
15862 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15864 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
15866 * config/nds32/constants.md (unspec_volatile_element): Add enum values
15867 for interrupt control.
15868 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15869 functions for interrupt control.
15870 * config/nds32/nds32-intrinsic.md: Likewise.
15871 * config/nds32/nds32_intrinsic.h: Likewise.
15872 * config/nds32/nds32.h (nds32_builtins): Likewise.
15874 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15876 * config/nds32/nds32.c (nds32_init_machine_status,
15877 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
15878 strict_aligned_p field.
15879 (nds32_expand_to_rtl_hook): New function.
15880 (TARGET_EXPAND_TO_RTL_HOOK): Define.
15881 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
15883 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15884 Chung-Ju Wu <jasonwucj@gmail.com>
15886 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
15887 * config/nds32/nds32-n7.md: New file.
15888 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
15889 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
15891 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
15892 * config/nds32/nds32.md (pipeline_model): Add n7.
15893 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
15894 * config/nds32/pipelines.md: Include n7 settings.
15896 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15897 Chung-Ju Wu <jasonwucj@gmail.com>
15899 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
15900 * config/nds32/nds32-e8.md: New file.
15901 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
15902 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
15904 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
15905 * config/nds32/nds32.md (pipeline_model): Add e8.
15906 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
15907 * config/nds32/pipelines.md: Include e8 settings.
15909 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15910 Chung-Ju Wu <jasonwucj@gmail.com>
15912 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
15913 * config/nds32/nds32-n8.md: New file.
15914 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
15915 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
15917 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
15918 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
15919 * config/nds32/nds32.md (pipeline_model): Add n8.
15920 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
15921 * config/nds32/pipelines.md: Include n8 settings.
15923 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15924 Chung-Ju Wu <jasonwucj@gmail.com>
15926 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
15927 * config/nds32/nds32-n9-2r1w.md: New file.
15928 * config/nds32/nds32-n9-3r2w.md: New file.
15929 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
15930 nds32_register_ports): New or modify for cpu n9.
15931 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
15933 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
15934 * config/nds32/nds32-utils.c: New file.
15935 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
15936 TARGET_MUL_SLOW): Define.
15937 * config/nds32/nds32.md (pipeline_model): New attribute.
15938 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
15939 New options that support cpu n9.
15940 * config/nds32/pipelines.md: Include n9 settings.
15941 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
15943 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15945 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
15946 information if necessary.
15947 (output_cond_branch_compare_zero): Likewise.
15948 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
15949 (nds32_target_alignment): Refine for alignment.
15950 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
15951 (FUNCTION_BOUNDARY): Modify.
15952 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
15954 * config/nds32/nds32.opt (malways-align, malign-functions): New.
15956 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15958 * config/nds32/constants.md (unspec_volatile_element): Add values for
15959 TLB operation and data prefetch.
15960 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15961 functions for TLB operation and data prefetch.
15962 * config/nds32/nds32-intrinsic.md: Likewise.
15963 * config/nds32/nds32_intrinsic.h: Likewise.
15964 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
15965 (nds32_print_operand): Likewise.
15966 * config/nds32/nds32.h (nds32_builtins): Likewise.
15968 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15969 Andrew Pinski <pinsika@gcc.gnu.org>
15971 PR middle-end/82976
15972 * match.pd: Use constant_boolean_node of correct type instead of
15973 boolean_true_node or boolean_false_node for simplifying
15974 pointer comparisons to zero.
15976 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15978 PR tree-optimization/80021
15979 * tree.c (verify_type_variant): Make error call in verify_variant_match
15980 translatable and remove final full stop.
15982 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15984 * config/nds32/constants.md (unspec_volatile_element): Add
15985 UNSPEC_VOLATILE_EH_RETURN.
15986 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
15987 nds32_output_stack_pop): Support dwarf exception handling process.
15988 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
15989 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
15990 exception handling process.
15991 (nds32_compute_stack_frame): Likewise.
15992 (nds32_return_addr_rtx): Likewise.
15993 (nds32_initial_elimination_offset): Likewise.
15994 (nds32_expand_prologue): Likewise.
15995 (nds32_expand_epilogue): Likewise.
15996 (nds32_dynamic_chain_address): New function.
15997 * config/nds32/nds32.h (machine_function): Add fields for dwarf
15998 exception handling.
15999 (DYNAMIC_CHAIN_ADDRESS): Define.
16000 (EH_RETURN_DATA_REGNO): Define.
16001 (EH_RETURN_STACKADJ_RTX): Define.
16002 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
16003 patterns for dwarf exception handling.
16005 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16007 * config/nds32/nds32.h: Clean up obsolete macros.
16009 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16011 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
16012 Add enum values for particular instructions.
16013 * config/nds32/nds32-intrinsic.c: Implementation of expanding
16014 particular intrinsic functions.
16015 * config/nds32/nds32-intrinsic.md: Likewise.
16016 * config/nds32/nds32_intrinsic.h: Likewise.
16017 * config/nds32/nds32.h (nds32_builtins): Likewise.
16018 * config/nds32/nds32.md (type): Add pbsad and pbsada.
16019 (btst, ave): New patterns for particular instructions.
16021 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16023 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
16024 Add enum values for atomic load/store and memory sync.
16025 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
16027 * config/nds32/nds32-intrinsic.md: Likewise.
16028 * config/nds32/nds32_intrinsic.h: Likewise.
16029 * config/nds32/nds32.h (nds32_builtins): Likewise.
16031 2018-04-07 Jakub Jelinek <jakub@redhat.com>
16033 PR tree-optimization/85257
16034 * fold-const.c (native_encode_vector): If not all elts could fit
16035 and off is -1, return 0 rather than offset.
16036 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
16037 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
16038 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
16039 adjust buffer in native_interpret_expr call.
16041 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16043 * config/nds32/constants.md (unspec_volatile_element): Add cache
16044 control enum values.
16045 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
16046 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
16047 * config/nds32/nds32.c (nds32_cctl_names): New.
16048 (nds32_print_operand): Handle cache control register names.
16049 * config/nds32/nds32.h (nds32_builtins): New enum values.
16050 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
16052 * config/nds32/nds32.md (type): Add mmu.
16053 * config/nds32/pipelines.md (simple_insn): Add mmu.
16055 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16057 * config/nds32/nds32.md (type): Remove call.
16058 * config/nds32/pipelines.md (simple_insn): Likewise.
16060 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16062 * config/nds32/constants.md (unspec_volatile_element): Add
16063 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
16064 UNSPEC_VOLATILE_FMFCFG.
16065 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
16066 description for fmfcfg and fmfcsr.
16067 (bdesc_1arg): Add fmtcsr.
16068 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
16069 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
16070 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
16071 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
16072 unspec_fmfcfg): New patterns.
16073 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
16074 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
16075 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
16076 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
16077 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
16078 __nds32__fmfcfg): Define.
16080 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16082 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
16083 intrinsic register names.
16084 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
16085 intrinsic register enum values and macros.
16087 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16089 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
16090 for load/store addressing form.
16091 (nds32_print_operand_address): Likewise.
16093 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
16096 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
16097 based on LABEL_REF. Remove useless assertion.
16098 (pic_address_needs_scratch): Fix formatting.
16099 (sparc_legitimize_pic_address): Minor tweaks.
16100 (sparc_delegitimize_address): Adjust assertion accordingly.
16101 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
16102 into symbolic_operand.
16103 (movsi_high_pic_label_ref): Likewise.
16104 (movsi_lo_sum_pic_label_ref): Likewise.
16105 (movdi_pic_label_ref): Likewise.
16106 (movdi_high_pic_label_ref): Likewise.
16107 (movdi_lo_sum_pic_label_ref): Likewise.
16109 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
16111 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
16112 custom LIB_SPEC setup.
16114 2018-04-06 Ruslan Bukin <br@bsdpad.com>
16115 Kito Cheng <kito.cheng@gmail.com>
16117 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
16118 * config/riscv/freebsd.h: New.
16120 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
16122 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
16123 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
16126 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
16127 Kito Cheng <kito.cheng@gmail.com>
16129 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
16130 nds32_output_call, nds32_symbol_binds_local_p): New functions.
16131 * config/nds32/nds32-protos.h (nds32_output_call,
16132 nds32_output_return): Declare.
16133 * config/nds32/nds32.md: Refine all the call and return patterns.
16135 2018-04-06 Jakub Jelinek <jakub@redhat.com>
16138 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
16139 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
16141 PR rtl-optimization/84872
16142 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
16143 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
16144 EDGE_CROSSING edge.
16146 2018-04-06 Tamar Christina <tamar.christina@arm.com>
16148 * expr.c (copy_blkmode_to_reg): Revert 254862.
16149 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
16151 2018-04-06 Richard Biener <rguenther@suse.de>
16153 PR middle-end/85244
16154 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
16155 after seeing a component reference with an adjacent field. Treat
16156 refs to arrays at struct end of external decls similar to
16157 refs to unconstrained commons.
16159 2018-04-06 Jakub Jelinek <jakub@redhat.com>
16162 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
16163 look through SAVE_EXPRs with non-side-effects argument. Adjust
16165 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
16168 2018-04-06 Richard Biener <rguenther@suse.de>
16170 PR middle-end/85180
16171 * alias.c (find_base_term): New wrapper around find_base_term
16172 unwinding CSELIB_VAL_PTR changes.
16173 (find_base_term): Do not restore CSELIB_VAL_PTR during the
16176 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16178 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
16180 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
16181 constant definitions.
16182 ("nop"): lr 0,0 -> nopr r0
16183 ("nop_lr0", "nop_lr1"): New insn definitions.
16185 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
16187 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
16188 NDS32_V3PUSH_AVAILABLE_P macro.
16190 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
16191 Chung-Ju Wu <jasonwucj@gmail.com>
16193 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
16194 (nds32*-*-*): Add float and fpu_config into supported_defaults.
16195 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
16196 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
16197 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
16198 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
16199 * config/nds32/constraints.md: New constraints and checking for hard
16200 float configuration.
16201 * config/nds32/iterators.md: New mode iterator and attribute for hard
16202 float configuration.
16203 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
16205 * config/nds32/nds32-fpu.md: New file.
16206 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
16207 deal with hard float code generation.
16208 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
16210 (abi_type, float_reg_number): New enum type.
16211 * config/nds32/nds32-predicates.c: New predicates for hard float.
16212 * config/nds32/nds32-protos.h: Declare functions for hard float.
16213 * config/nds32/nds32.c: Implementation for hard float configuration.
16214 * config/nds32/nds32.h: Definitions for hard float configuration.
16215 * config/nds32/nds32.md: Include hard float machine description and
16216 modify patterns for hard float configuration.
16217 * config/nds32/nds32.opt: New options for hard float configuration.
16218 * config/nds32/predicates.md: New predicates for hard float
16221 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16223 * common/config/nds32/nds32-common.c
16224 (nds32_option_optimization_table): Enable -mreleax-hint by default.
16226 2018-04-05 Jakub Jelinek <jakub@redhat.com>
16228 PR middle-end/85195
16229 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
16230 CONSTRUCTOR_ELT (ctor, ...)->value.
16232 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
16235 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
16237 2018-04-05 Tom de Vries <tom@codesourcery.com>
16240 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
16243 2018-04-05 Shiva Chen <shiva0217@gmail.com>
16244 Kito Cheng <kito.cheng@gmail.com>
16246 * config/nds32/constraints.md (U33): Fine-tune checking condition.
16247 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
16248 * config/nds32/nds32.h (nds32_16bit_address_type): Add
16249 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
16251 2018-04-05 Shiva Chen <shiva0217@gmail.com>
16252 Kito Cheng <kito.cheng@gmail.com>
16254 * config/nds32/constraints.md (Ufe): New memory constraint.
16255 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
16256 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
16257 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
16259 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
16260 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
16262 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16264 * config/nds32/nds32.md: Use optimize_size in the condition for
16265 alu-shift instructions.
16267 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16269 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
16271 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16273 * config/nds32/nds32.md (negsi2): Refine pattern.
16275 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
16276 Chung-Ju Wu <jasonwucj@gmail.com>
16278 * config/nds32/iterators.md (shift_rotate): New code iterator.
16279 (shift): New code attribute.
16280 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
16281 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
16282 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
16283 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
16284 bit-wise operations.
16285 (andsi3, *andsi3): Ditto.
16286 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
16287 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
16288 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
16289 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
16290 nds32_ior_operand, nds32_xor_operand): New predicates.
16292 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16294 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
16295 (addsi3, subsi3): ... this.
16297 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16299 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
16301 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16303 * config/nds32/nds32.md: Adjust indention.
16305 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
16307 * config/nds32/nds32.md (feature): New attribute.
16309 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16311 * config/nds32/nds32.md (subtype): New attribute.
16313 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16316 * config/arm/arm-builtins.c (arm_expand_builtin): Change
16317 expansion to perform a bitwise AND of the argument followed by a
16318 boolean negation of the result.
16320 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
16322 PR rtl-optimization/84878
16323 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
16324 the basic block. Assert the use reference is not artificial and that
16325 it has an associated insn.
16327 2018-04-04 Michael Matz <matz@suse.de>
16329 * builtins.c (compute_objsize): Pass correct operand
16330 to array_at_struct_end_p.
16332 2018-04-04 Richard Biener <rguenther@suse.de>
16335 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
16336 from contexts for DINFO_LEVEL_TERSE and below.
16338 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16340 * config/nds32/nds32-doubleword.md (move_<mode>): Require
16341 resiter_operand condition.
16342 * config/nds32/nds32.md (*move<mode>): Ditto.
16344 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16345 Monk Chiang <sh.chiang04@gmail.com>
16347 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
16349 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16351 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
16353 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16354 Kito Cheng <kito.cheng@gmail.com>
16356 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
16357 nds32_cond_code_str, output_cond_branch,
16358 output_cond_branch_compare_zero, nds32_expand_cbranch,
16359 nds32_expand_cstore, nds32_expand_movcc,
16360 nds32_output_cbranchsi4_equality_zero,
16361 nds32_output_cbranchsi4_equality_reg,
16362 nds32_output_cbranchsi4_equality_reg_or_const_int,
16363 nds32_output_cbranchsi4_greater_less_zero: New functions.
16364 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
16365 nds32_expand_cstore, nds32_expand_movcc,
16366 nds32_output_cbranchsi4_equality_zero,
16367 nds32_output_cbranchsi4_equality_reg,
16368 nds32_output_cbranchsi4_equality_reg_or_const_int,
16369 nds32_output_cbranchsi4_greater_less_zero): Declare.
16370 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
16371 nds32_rimm11s_operand): New predicates.
16372 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
16373 * config/nds32/nds32.md: Rewrite all the branch and conditional move
16376 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16378 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
16379 * config/nds32/nds32.md: Ditto.
16380 * config/nds32/pipelines.md: Ditto.
16382 2018-04-04 Richard Biener <rguenther@suse.de>
16384 PR tree-optimization/85168
16385 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
16386 propagating abnormals.
16388 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16390 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
16392 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16393 Kito Cheng <kito.cheng@gmail.com>
16395 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
16396 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
16397 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
16398 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
16399 * config/nds32/nds32.md (sibcall_internal): New.
16400 (sibcall_register): Remove.
16401 (sibcall_immediate): Remove.
16402 (sibcall_value_internal): New.
16403 (sibcall_value_register): Remove.
16404 (sibcall_value_immediate): Remove.
16405 * config/nds32/predicates.md (nds32_general_register_operand): New.
16406 (nds32_call_address_operand): New.
16408 2018-04-03 Jakub Jelinek <jakub@redhat.com>
16410 PR rtl-optimization/85167
16411 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
16412 bb_defs if *split_p, instead preinitialize it to NULL.
16414 PR tree-optimization/85156
16415 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
16416 evaluating the argument multiple times.
16418 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
16420 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
16422 (_mm_cvtpd_ps): Likewise.
16423 (_mm_cvttpd_epi32): Likewise.
16424 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
16425 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
16426 vector, pixel, and bool following altivec.h include.
16428 2018-04-03 Martin Sebor <msebor@redhat.com>
16430 * doc/extend.texi (Common Function Attributes): Clarify.
16431 (const attribute): Likewise.
16432 (pure attribute): Likewise.
16434 2018-04-03 Jakub Jelinek <jakub@redhat.com>
16437 * config/i386/i386.c (ix86_expand_vector_set): Use
16438 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
16440 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
16442 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
16443 instructions when changing rounding bits to preserve precision bits
16444 in the x87 control word.
16446 2018-04-03 Martin Liska <mliska@suse.cz>
16448 PR tree-optimization/82491
16449 * rtl.h (strip_offset_and_add): Replace += suboffset with
16450 poly_uint64 () + suboffset.
16452 2018-03-29 Martin Liska <mliska@suse.cz>
16453 Martin Jambor <mjambor@suse.cz>
16456 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
16457 param_type is not an integral or pointer type.
16459 2018-04-03 Richard Biener <rguenther@suse.de>
16461 * sese.h (recompute_all_dominators): Remove.
16463 2018-04-02 Martin Sebor <msebor@redhat.com>
16465 * doc/invoke.texi (-Wrestrict): Fix typos.
16467 2018-04-02 Jim Wilson <jimw@sifive.com>
16469 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
16470 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
16471 (<optab>di3, <optab>si3_extend): Likewise.
16472 (<optab>si3_mask, <optab>si3_mask_1): New.
16473 (<optab>di3_mask, <optab>di3_mask_1): New.
16474 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
16475 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
16476 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
16478 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
16480 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
16483 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
16485 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
16486 (nds32_canonicalize_comparison): New function.
16488 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16489 Kito Cheng <kito.cheng@gmail.com>
16490 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16492 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
16493 * config/nds32/constants.md (unspec_volatile_element): Add
16494 UNSPEC_VOLATILE_RELAX_GROUP.
16495 * config/nds32/nds32-relax-opt.c: New file.
16496 * config/nds32/nds32-predicates.c
16497 (nds32_symbol_load_store_p): New function.
16498 * config/nds32/nds32-protos.h
16499 (nds32_symbol_load_store_p): Declare function.
16500 (make_pass_nds32_relax_opt): Declare new rtl pass function.
16501 * config/nds32/nds32.c
16502 (nds32_register_pass): New function to register pass.
16503 (nds32_register_passes): New function to register passes.
16504 * config/nds32/nds32.md (relax_group): New pattern.
16505 * config/nds32/nds32.opt (mrelax-hint): New option.
16506 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
16508 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
16510 * config/nds32/t-nds32: Modify files dependency.
16512 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16514 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
16515 (PROFILE_HOOK): Define its implementation.
16517 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16519 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
16520 type and 32-bit size.
16522 2018-04-01 Jakub Jelinek <jakub@redhat.com>
16524 PR middle-end/85090
16525 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
16526 (V_128_256): New mode iterator.
16527 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
16528 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
16529 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
16531 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
16532 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
16534 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16537 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
16538 NaN inputs correctly.
16540 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
16543 * config/rs6000/vsx.md (??r): New mode attribute.
16544 (*vsx_mov<mode>_64bit): Use it.
16545 (*vsx_mov<mode>_32bit): Likewise.
16547 2018-03-30 Martin Sebor <msebor@redhat.com>
16549 PR tree-optimization/84818
16550 * builtins.c (check_access): Use warning_n.
16552 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16555 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
16557 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
16560 2018-03-30 Julia Koval <julia.koval@intel.com>
16563 * x86-tune.def (movx, partial_reg_dependency): Enable for
16566 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
16568 PR inline-asm/84985
16569 * lra-constraints.c (process_alt_operands): Move setting
16570 this_alternative_matches below.
16572 2018-03-29 Martin Liska <mliska@suse.cz>
16575 * doc/invoke.texi: Document how LTO works with debug info.
16576 Describe auto-load support of binutils. Mention 'x86-64'
16577 as valid option value of -march option.
16579 2018-03-29 Jakub Jelinek <jakub@redhat.com>
16581 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
16584 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
16585 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
16586 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
16589 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
16592 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
16593 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
16594 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
16595 for RS6000_BTM_POWERPC64.
16596 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
16597 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
16598 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
16601 (DIVDEU): Likewise.
16603 2018-03-28 Carl Love <cel@us.ibm.com>
16606 2017-09-27 Carl Love <cel@us.ibm.com>
16608 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
16609 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
16610 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
16613 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16615 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
16616 instead of __vector bool.
16617 (_mm_max_pu8): Likewise.
16618 (_mm_min_pi16): Likewise.
16620 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
16623 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
16624 (DIVWEUO): Likewise.
16625 (DIVDEO): Likewise.
16626 (DIVDEUO): Likewise.
16627 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
16628 DIVWEUO and DIVDEUO.
16629 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
16630 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
16631 (div_extend): Likewise.
16632 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
16634 (__builtin_divweuo): Likewise.
16635 (__builtin_divdeo): Likewise.
16636 (__builtin_divdeuo): Likewise.
16638 2018-03-28 Jakub Jelinek <jakub@redhat.com>
16641 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
16642 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
16644 PR tree-optimization/82004
16645 * gimple-match-head.c (optimize_pow_to_exp): New function.
16646 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
16647 Don't fold to exp if optimize_pow_to_exp is false.
16649 2018-03-28 Martin Liska <mliska@suse.cz>
16652 * calls.c (initialize_argument_information): Fix trailing space.
16653 * common.opt: Fix typo and provide better explanation for
16654 -fsanitize-coverage option.
16655 * config/i386/i386.opt: Fix typo.
16657 2018-03-28 Jakub Jelinek <jakub@redhat.com>
16658 Martin Liska <mliska@suse.cz>
16661 * gimplify.c (asan_poison_variable): Don't do the check for
16662 gimplify_omp_ctxp here.
16663 (gimplify_decl_expr): Do it here.
16664 (gimplify_target_expr): Likewise.
16666 2018-03-28 Martin Liska <mliska@suse.cz>
16669 * config/i386/i386.c (ix86_function_arg_advance): Do not call
16670 chkp_type_bounds_count if MPX is not enabled.
16672 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
16674 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
16676 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16679 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
16680 function to create the function decl for complex long double
16681 multiply and divide for -mabi=ieeelongdouble.
16682 (init_float128_ieee): Call it.
16684 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
16687 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
16688 -fcf-protection=branch -mibt.
16689 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
16691 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16694 * config/arm/arm.c (arm_valid_symbolic_address): Handle
16695 arm_word_relocations.
16697 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
16700 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
16701 extern array declarations.
16703 2018-03-27 Richard Biener <rguenther@suse.de>
16705 PR middle-end/84067
16706 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
16707 explicit single_use checks.
16709 2018-03-27 Richard Biener <rguenther@suse.de>
16711 PR tree-optimization/85082
16712 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
16715 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16717 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
16718 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
16719 Turn on fasynchronous-unwind-tables and funwind-tables.
16721 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
16724 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
16725 (*bmi_blsr_<mode>_ccz): Ditto.
16727 2018-03-26 Tom de Vries <tom@codesourcery.com>
16729 PR tree-optimization/85063
16730 * omp-general.c (offloading_function_p): New function. Factor out
16732 * omp-offload.c (pass_omp_target_link::gate): ... here.
16733 * omp-general.h (offloading_function_p): Declare.
16734 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
16735 with attribute omp declare target for offloading functions.
16737 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
16739 PR tree-optimization/84005
16740 * tree-data-ref.h (get_base_for_alignment): Declare.
16741 * tree-data-ref.c (get_base_for_alignment_1): New function.
16742 (get_base_for_alignment): Likewise.
16743 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16744 get_base_for_alignment to find a suitable base object, instead
16745 of always using drb->base_address.
16747 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16749 PR inline-asm/85022
16750 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
16751 known size by default.
16753 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
16755 PR inline-asm/85030
16756 * lra-constraints.c (process_alt_operands): Don't match BLKmode
16757 and non BLKmode operands.
16759 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16762 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
16763 Clean up attributes.
16765 2018-03-23 Richard Biener <rguenther@suse.de>
16768 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
16769 we are going to emit early debug for LTO.
16771 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16773 PR inline-asm/85034
16774 * function.c (match_asm_constraints_1): Don't optimize if input
16775 doesn't satisfy general_operand predicate for output's mode.
16777 PR inline-asm/85022
16778 * alias.c (write_dependence_p): Don't require for x_canonicalized
16779 non-VOIDmode if x has VOIDmode.
16782 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
16783 just don't try to optimize it rather than assert it never happens.
16785 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16787 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
16788 macro expansions for definition of ST_INTERNAL_<mode> and
16789 LD_INTERNAL_<mode> builtins.
16790 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
16792 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
16794 (altivec_expand_st_builtin): Likewise.
16795 (altivec_expand_builtin): Remove calls to deleted functions.
16796 (rs6000_address_for_altivec): Delete this function.
16797 * config/rs6000/vector.md: Remove expands for
16798 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
16800 2018-03-22 Sudakshina Das <sudi.das@arm.com>
16803 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
16804 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
16805 re-computing once computed.
16806 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
16807 (arm_init_machine_status): Initialize
16808 machine->static_chain_stack_bytes.
16810 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16813 * doc/extend.texi: Add four new prototypes for vec_ld.
16814 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
16815 definitions for more logical presentation.
16816 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
16817 entries for V1TI variants of __builtin_altivec_ld builtin.
16818 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
16819 handling of V1TI variant of LVX icode pattern.
16820 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
16821 (rs6000_gimple_fold_builtin): Likewise.
16822 (altivec_init_builtins): Add code to define
16823 __builtin_altivec_lvx_v1ti function.
16825 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16827 PR inline-asm/84941
16828 * function.c (match_asm_constraints_1): Don't do the optimization
16829 if input isn't a REG, SUBREG, MEM or constant.
16831 2018-03-22 Tom de Vries <tom@codesourcery.com>
16833 PR tree-optimization/84956
16834 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
16835 bb_has_abnormal_pred.
16837 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16840 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
16841 DECL_INITIAL (decl) to decl at the end.
16842 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
16843 adjust the comment.
16845 2018-03-21 Joseph Myers <joseph@codesourcery.com>
16847 * doc/extend.texi (__builtin_tgmath): Document when complex
16848 integer types are treated as _Complex _Float64.
16850 2018-03-21 Tom de Vries <tom@codesourcery.com>
16852 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
16854 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16856 PR tree-optimization/84960
16857 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
16858 if it is ENTRY block, move them into single succ of ENTRY in that case.
16860 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
16862 PR tree-optimization/84811
16863 * poly-int.h (poly_span_traits): Remove the T3 parameter and
16864 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
16865 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
16866 (known_subrange_p): Update accordingly. Cast each value involved
16867 in the size comparison, rather than casting the result of the
16870 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16872 PR tree-optimization/84982
16873 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
16874 by flipping the least significant bit rather than all bits from
16875 bitpos to bitpos + bitsize - 1.
16877 2018-03-21 Nathan Sidwell <nathan@acm.org>
16879 * doc/extend.texi (Deprecated Features): Remove mention of
16880 long-deleted deprecations.
16882 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16885 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
16886 * configure: Regenerate.
16888 2018-03-21 Tom de Vries <tom@codesourcery.com>
16890 PR tree-optimization/83126
16891 * tree-parloops.c (num_phis): New function.
16892 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
16894 2018-03-21 Nathan Sidwell <nathan@acm.org>
16896 * doc/extend.texi (Deprecated Features): Update deprecated flags,
16897 mention anon-struct/union members and trailing attributes.
16899 2018-03-21 Bin Cheng <bin.cheng@arm.com>
16901 PR tree-optimization/84969
16902 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
16903 builtin memset partitions if they set different rhs values.
16905 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16907 PR rtl-optimization/84989
16908 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
16909 VEC_DUPLICATE with scalar result mode.
16911 2018-03-21 Martin Liska <mliska@suse.cz>
16914 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
16915 not intended return statement.
16917 2018-03-21 Martin Liska <mliska@suse.cz>
16920 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
16921 (chkp_find_bound_slots_1): Limit number of iterations.
16923 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
16926 * Minor grammar fixes for x86 options.
16928 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16931 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
16932 holding REG_CFA_RESTORE notes, instead turn them into a USE.
16934 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
16937 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
16938 (altivec_lvx_<mode>_1op): Likewise.
16939 (altivec_stvx_<mode>_2op): Likewise.
16940 (altivec_stvx_<mode>_1op): Likewise.
16941 (altivec_lvx_<VM2:mode>): New define_expand.
16942 (altivec_stvx_<VM2:mode>): Likewise.
16943 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
16944 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16945 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
16946 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16947 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
16948 (rs6000_gen_lvx): Likewise.
16949 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
16950 (altivec_expand_stv_builtin): Likewise.
16951 (altivec_expand_builtin): Likewise.
16952 * config/rs6000/vector.md: Likewise.
16954 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16957 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
16960 2018-03-20 Richard Biener <rguenther@suse.de>
16963 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
16964 sign-conversions as zero, fall back to standard scalar_stmt
16967 2018-03-20 Martin Liska <mliska@suse.cz>
16970 * predict.c (rebuild_frequencies): Handle case when we have
16971 PROFILE_ABSENT, but flag_guess_branch_prob is false.
16973 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16976 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
16977 flag_section_anchors.
16978 * varasm.c (use_blocks_for_decl_p): Remove hack for
16979 dw2_force_const_mem.
16982 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
16984 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
16985 be created, use lowpart_subreg of operands[0] rather than operands[0]
16987 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
16988 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
16989 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
16990 and n constraint instead of aarch64_shift_imm_di and Usd.
16991 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
16992 (*aarch64_<optab>_reg_minus<mode>3): ... this.
16994 2018-03-20 Sudakshina Das <sudi.das@arm.com>
16997 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
16998 to favor GPR over NEON registers.
16999 (<shift>di3_neon): Likewise.
17001 2018-03-20 Tom de Vries <tom@codesourcery.com>
17004 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
17005 (nvptx_process_pars): Emit bar.sync asap and alap.
17007 2018-03-20 Tom de Vries <tom@codesourcery.com>
17010 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
17011 seen_label if seen_label is already set.
17013 2018-03-20 Jakub Jelinek <jakub@redhat.com>
17016 * config/i386/i386.c (fold_builtin_cpu): For features above 31
17017 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
17018 Use 1U instead of 1. Formatting fixes.
17021 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
17022 instead of TREE_TYPE (s1) for the return value.
17024 2018-03-19 Jakub Jelinek <jakub@redhat.com>
17026 PR tree-optimization/84946
17027 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
17028 bitsize + bitsize in poly_uint64 rather than poly_int64.
17031 * dwarf2asm.c: Include fold-const.c.
17032 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
17033 of decl rather than decl itself.
17035 PR rtl-optimization/84643
17036 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
17038 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
17041 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
17042 calling assemble_variable.
17044 2018-03-19 Sudakshina Das <sudi.das@arm.com>
17047 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
17048 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
17050 2018-03-19 Jim Wilson <jimw@sifive.com>
17053 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
17054 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
17055 (riscv_first_stack_step): Likewise.
17056 (riscv_option_override): Use STACK_BOUNDARY instead of
17057 MIN_STACK_BOUNDARY.
17058 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
17059 MIN_STACK_BOUNDARY.
17060 (BIGGEST_ALIGNMENT): Set to 128.
17061 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
17062 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
17065 2018-03-19 Richard Biener <rguenther@suse.de>
17067 PR tree-optimization/84933
17068 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
17069 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
17071 2018-03-19 Richard Biener <rguenther@suse.de>
17073 PR tree-optimization/84859
17074 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
17075 (cond_if_else_store_replacement): Perform sinking operation on
17076 single-store BBs regardless of MAX_STORES_TO_SINK setting.
17077 Generalize what a BB with a single eligible store is.
17079 2018-03-19 Richard Biener <rguenther@suse.de>
17081 PR tree-optimization/84929
17082 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
17083 chrec_is_positive against non-chrec arg.
17085 2018-03-19 Tamar Christina <tamar.christina@arm.com>
17088 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
17090 2018-03-18 Martin Liska <mliska@suse.cz>
17092 PR rtl-optimization/84635
17093 * regrename.c (build_def_use): Use matches_mode only when
17096 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
17098 PR tree-optimization/84913
17099 * tree-vect-loop.c (vectorizable_reduction): Don't try to
17100 vectorize chains of COND_EXPRs.
17102 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
17104 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
17106 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
17108 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
17110 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
17112 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
17114 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
17115 Kito Cheng <kito.cheng@gmail.com>
17117 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
17118 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
17119 (nds32_adjust_reg_alloc_order): New function.
17120 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
17122 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
17124 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
17125 nds32_print_operand, nds32_print_operand_address): Use
17126 HOST_WIDE_INT_PRINT_DEC instead.
17128 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
17130 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
17132 2018-03-17 Jakub Jelinek <jakub@redhat.com>
17135 * config/i386/i386.c (initial_ix86_tune_features,
17136 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
17137 unsigned long long.
17138 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
17139 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
17140 rather than 1u << ix86_tune. Formatting fix.
17141 (ix86_option_override_internal): Change ix86_arch_mask from
17142 unsigned int to unsigned HOST_WIDE_INT, initialize to
17143 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
17144 (ix86_function_specific_restore): Likewise.
17146 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17149 * postreload.c (reload_combine_recognize_pattern): Perform
17150 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
17151 truncate_int_for_mode the result for the destination's mode.
17154 * hsa-gen.c (mem_type_for_type): Fix comment typo.
17155 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
17157 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
17160 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
17163 * lra-assigns.c (lra_split_hard_reg_for): Don't use
17164 regno_allocno_class_array and sorted_pseudos.
17165 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
17166 insns where regno is used.
17168 2018-03-16 Martin Liska <mliska@suse.cz>
17171 * multiple_target.c (create_dispatcher_calls): Redirect
17172 reference in the symbol table.
17174 2018-03-16 Martin Liska <mliska@suse.cz>
17177 * multiple_target.c (create_dispatcher_calls): Redirect also
17180 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17184 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
17185 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
17186 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
17188 2018-03-16 Julia Koval <julia.koval@intel.com>
17190 * doc/invoke.texi (Skylake Server): Add CLWB.
17191 Cannonlake): Remove CLWB.
17193 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17195 PR tree-optimization/84841
17196 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
17198 (FLOAT_ONE_CONST_TYPE): Define.
17199 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
17200 (sort_by_operand_rank): Put entries with higher constant_type last
17201 rather than first to match comments.
17203 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
17205 * config/nios2/nios2.md (movsi_internal): Fix thinko in
17208 2018-03-15 Jakub Jelinek <jakub@redhat.com>
17211 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
17212 check and use address of target always.
17214 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
17217 * config/i386/i386.c (indirect_thunk_needed): Update comments.
17218 (indirect_thunk_bnd_needed): Likewise.
17219 (indirect_thunks_used): Likewise.
17220 (indirect_thunks_bnd_used): Likewise.
17221 (indirect_return_needed): New.
17222 (indirect_return_bnd_needed): Likewise.
17223 (output_indirect_thunk_function): Add a bool argument for
17225 (output_indirect_thunk_function): Don't generate alias for
17226 function return thunk.
17227 (ix86_code_end): Call output_indirect_thunk_function to generate
17228 function return thunks.
17229 (ix86_output_function_return): Set indirect_return_bnd_needed
17230 and indirect_return_needed instead of indirect_thunk_bnd_needed
17231 and indirect_thunk_needed.
17233 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
17235 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
17236 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
17237 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
17239 2018-03-15 David Malcolm <dmalcolm@redhat.com>
17240 Paul Hua <paul.hua.gm@gmail.com>
17243 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
17245 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
17247 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
17248 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
17249 resp. SFmode cases.
17251 2018-03-15 Tamar Christina <tamar.christina@arm.com>
17254 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
17255 instead of GET_MODE_SIZE when comparing Units.
17257 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
17260 * varasm.c (hash_section): Return an unchangeble hash value
17261 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
17262 Return !aarch64_can_use_per_function_literal_pools_p ().
17264 2018-03-15 Jakub Jelinek <jakub@redhat.com>
17267 * optabs.c (emit_conditional_move): Pass address of cmode's copy
17268 rather than address of cmode as last argument to prepare_cmp_insn.
17270 2018-03-15 Julia Koval <julia.koval@intel.com>
17272 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
17273 F_AVX512VNNI, F_AVX512BITALG): New.
17275 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
17278 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
17279 insn for floating-point loads and stores.
17281 2018-03-14 Carl Love <cel@us.ibm.com>
17283 * config/rs6000/rs6000-c.c: Add macro definitions for
17284 ALTIVEC_BUILTIN_VEC_PERMXOR.
17285 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
17286 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
17287 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
17288 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
17290 * config/doc/extend.texi: Add prototypes for vec_permxor.
17292 2018-03-14 David Malcolm <dmalcolm@redhat.com>
17295 * diagnostic-show-locus.c (class layout_point): Convert m_line
17296 from int to linenum_type.
17297 (line_span::comparator): Use linenum "compare" function when
17298 comparing line numbers.
17299 (test_line_span): New function.
17300 (layout_range::contains_point): Convert param "row" from int to
17302 (layout_range::intersects_line_p): Likewise.
17303 (layout::will_show_line_p): Likewise.
17304 (layout::print_source_line): Likewise.
17305 (layout::should_print_annotation_line_p): Likewise.
17306 (layout::print_annotation_line): Likewise.
17307 (layout::print_leading_fixits): Likewise.
17308 (layout::annotation_line_showed_range_p): Likewise.
17309 (struct line_corrections): Likewise for field m_row.
17310 (line_corrections::line_corrections): Likewise for param "row".
17311 (layout::print_trailing_fixits): Likewise.
17312 (layout::get_state_at_point): Likewise.
17313 (layout::get_x_bound_for_row): Likewise.
17314 (layout::print_line): Likewise.
17315 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
17316 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
17317 * input.c (selftest::test_linenum_comparisons): New function.
17318 (selftest::input_c_tests): Call it.
17319 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
17320 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
17321 * selftest.h (ASSERT_GT): New macro.
17322 (ASSERT_GT_AT): New macro.
17323 (ASSERT_LT): New macro.
17324 (ASSERT_LT_AT): New macro.
17326 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
17328 PR rtl-optimization/84780
17329 * combine.c (distribute_links): Don't make a link based on pc_rtx.
17331 2018-03-14 Martin Liska <mliska@suse.cz>
17333 * tree.c (record_node_allocation_statistics): Use
17334 get_stats_node_kind.
17335 (get_stats_node_kind): New function extracted from
17336 record_node_allocation_statistics.
17337 (free_node): Use get_stats_node_kind.
17339 2018-03-14 Richard Biener <rguenther@suse.de>
17341 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
17342 that the value-set of ANTIC_IN doesn't grow.
17345 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
17347 (BB_VISITED_WITH_VISITED_SUCCS): New define.
17348 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
17350 2018-03-14 Julia Koval <julia.koval@intel.com>
17352 * config.gcc (icelake-client, icelake-server): New.
17354 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
17355 (initial_ix86_arch_features): Ditto.
17356 (PTA_SKYLAKE): Add SGX.
17357 (PTA_ICELAKE): Remove.
17358 (PTA_ICELAKE_CLIENT): New.
17359 (PTA_ICELAKE_SERVER): New.
17360 (ix86_option_override_internal): Split up icelake on icelake client and
17362 (get_builtin_code_for_version): Ditto.
17363 (fold_builtin_cpu): Ditto.
17364 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
17365 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
17366 * config/i386/i386.h (processor_type): Ditto.
17367 * doc/invoke.texi: Ditto.
17369 2018-03-14 Jakub Jelinek <jakub@redhat.com>
17372 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
17373 INTEGER_CST offset, add it together with bitpos / 8 and
17374 sign extend based on POINTER_SIZE.
17378 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17381 * config/i386/constraints.md (Yc): New register constraint.
17382 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17383 Use Yc constraint for alternative 2 of operand 0. Remove
17384 preferred_for_speed attribute.
17386 2018-03-14 Richard Biener <rguenther@suse.de>
17388 PR tree-optimization/84830
17389 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
17390 with the old one to avoid oscillations.
17392 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
17395 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
17397 (assign_by_spills): Return a flag of reload assignment failure.
17398 Do not process the reload assignment failures. Do not spill other
17399 reload pseudos if they has the same reg class. Update n if
17401 (lra_assign): Add a return arg. Set up from the result of
17402 assign_by_spills call.
17403 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17404 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17405 usage_insns if it is not NULL.
17406 (spill_hard_reg_in_range): New function.
17407 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17408 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17409 function prototypes.
17410 (lra_assign): Change prototype.
17411 * lra.c (lra): Add code to deal with fails by splitting hard reg
17414 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
17416 * config/riscv/riscv.opt (mrelax): New option.
17417 * config/riscv/riscv.c (riscv_file_start): Emit ".option
17418 "norelax" when riscv_mrelax is disabled.
17419 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
17421 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17424 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
17425 reassociation for int modes.
17427 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17429 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
17430 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
17432 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
17433 * config/aarch64/aarch64-sve.md
17434 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
17435 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
17436 (*extend<mode><Vwide>2): Rename to...
17437 (aarch64_sve_extend<mode><Vwide>2): ...this.
17438 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
17439 renaming the old pattern to...
17440 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
17442 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
17443 define_expand, renaming the old pattern to...
17444 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
17445 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
17446 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
17447 account when deciding which SVE instruction the optab should use.
17448 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
17450 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17452 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
17453 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
17454 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
17455 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
17456 (tlsdesc_small_<mode>): Turn a define_expand and use
17457 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
17458 (tlsdesc_small_advsimd_<mode>): ...this.
17459 (tlsdesc_small_sve_<mode>): New pattern.
17461 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17463 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
17464 (UNSPEC_UMUL_HIGHPART): New constants.
17465 (MUL_HIGHPART): New int iteraor.
17466 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
17467 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
17469 (*<su>mul<mode>3_highpart): New define_insn.
17471 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
17474 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
17477 2018-03-13 Martin Liska <mliska@suse.cz>
17480 * (sem_item_optimizer::sem_item_optimizer): Initialize new
17482 (sem_item_optimizer::~sem_item_optimizer): Release it.
17483 (sem_item_optimizer::merge_classes): Register variable aliases.
17484 (sem_item_optimizer::fixup_pt_set): New function.
17485 (sem_item_optimizer::fixup_points_to_sets): Likewise.
17486 * ipa-icf.h: Declare new variables and functions.
17488 2018-03-13 Jakub Jelinek <jakub@redhat.com>
17490 PR middle-end/84834
17491 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
17492 integer_pow2p@2 and test integer_pow2p in condition.
17493 (A < 0 ? C : 0): Similarly for @1.
17495 PR middle-end/84831
17496 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
17497 characters starting at p contain '\0' character, don't look beyond
17501 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
17502 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
17505 * reg-stack.c (change_stack): Change update_end var from int to
17506 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
17507 also call set_block_for_insn on the newly added insns and rescan.
17510 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
17511 on the last operand.
17514 * tree.c (stabilize_reference_1): Return save_expr (e) for
17515 STATEMENT_LIST even if it doesn't have side-effects.
17517 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
17519 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
17521 2018-03-12 Renlin Li <renlin.li@arm.com>
17523 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
17524 aarch64_output_scalar_simd_mov_immediate.
17526 2018-03-12 Martin Sebor <msebor@redhat.com>
17528 PR tree-optimization/83456
17529 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
17530 for perfectly overlapping calls to memcpy.
17531 (gimple_fold_builtin_memory_chk): Same.
17532 (gimple_fold_builtin_strcpy): Handle no-warning.
17533 (gimple_fold_builtin_stxcpy_chk): Same.
17534 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
17536 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
17538 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
17539 parameter. Use it for SFmode.
17540 (rs6000_function_arg_advance_1): Adjust.
17541 (rs6000_function_arg): Adjust.
17542 (rs6000_gimplify_va_arg): Pass false for that new parameter.
17544 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
17546 PR rtl-optimization/84169
17547 PR rtl-optimization/84780
17548 * combine.c (can_combine_p): Check for a 2-insn combination whether
17549 the destination register is used between the two insns, too.
17551 2018-03-12 Richard Biener <rguenther@suse.de>
17553 PR tree-optimization/84803
17554 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
17555 for refs DR analysis didn't process.
17557 2018-03-12 Richard Biener <rguenther@suse.de>
17559 PR tree-optimization/84777
17560 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
17561 force-vectorize loops ignore whether we are optimizing for size.
17563 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
17565 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
17566 (TARGET_MD_ASM_ADJUST): Define.
17568 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
17569 Kito Cheng <kito.cheng@gmail.com>
17570 Chung-Ju Wu <jasonwucj@gmail.com>
17572 * config/nds32/nds32.c (nds32_compute_stack_frame,
17573 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
17574 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
17575 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
17576 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
17577 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
17578 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
17579 * config/nds32/nds32.md (prologue, epilogue): Use macro
17580 NDS32_V3PUSH_AVAILABLE_P to do checking.
17582 2018-03-11 Jakub Jelinek <jakub@redhat.com>
17585 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
17586 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
17587 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
17588 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
17589 addition of most attributes on !orig_type_die or the attribute not
17590 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
17592 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17593 Chung-Ju Wu <jasonwucj@gmail.com>
17595 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
17596 __NDS32_VH__ macro.
17597 * config/nds32/nds32.opt (mvh): New option.
17599 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17600 Chung-Ju Wu <jasonwucj@gmail.com>
17602 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
17604 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
17605 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
17608 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17609 Chung-Ju Wu <jasonwucj@gmail.com>
17611 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
17613 * config/nds32/nds32-multiple.md (strlensi): New pattern.
17614 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
17616 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
17617 Kito Cheng <kito.cheng@gmail.com>
17618 Chung-Ju Wu <jasonwucj@gmail.com>
17620 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
17621 UNSPEC_FFMISM and UNSPEC_FLMISM.
17622 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
17623 for ffb, ffmism and flmism.
17624 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
17625 (unspec_ffmism): Ditto.
17626 (unspec_flmism): Ditto.
17627 (nds32_expand_builtin_impl): Check if string extension is available.
17628 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
17629 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
17631 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
17634 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17637 * lra-assigns.c (assign_by_spills): Return a flag of reload
17638 assignment failure. Do not process the reload assignment
17639 failures. Do not spill other reload pseudos if they has the same
17641 (lra_assign): Add a return arg. Set up from the result of
17642 assign_by_spills call.
17643 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17644 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17645 usage_insns if it is not NULL.
17646 (spill_hard_reg_in_range): New function.
17647 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17648 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17649 function prototypes.
17650 (lra_assign): Change prototype.
17651 * lra.c (lra): Add code to deal with fails by splitting hard reg
17654 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
17657 * config/i386/i386.opt: Replace Enforcment with Enforcement.
17659 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
17662 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
17663 (dw_val_node): Add val_symbolic_view.
17664 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
17665 (symview_upper_bound): New.
17666 (new_line_info_table): Initialize symviews_since_reset.
17667 (dwarf2out_source_line): Count symviews_since_reset and set
17668 symview_upper_bound.
17669 (dw_val_equal_p): Handle symview.
17670 (add_AT_symview): New.
17671 (print_dw_val): Handle symview.
17672 (attr_checksum, attr_checksum_ordered): Likewise.
17673 (same_dw_val_p, size_of_die): Likewise.
17674 (value_format, output_die): Likewise.
17675 (add_high_low_attributes): Use add_AT_symview for entry_view.
17676 (dwarf2out_finish): Reset symview_upper_bound, clear
17679 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
17682 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
17683 Add strict argument and use it.
17684 (rs6000_split_multireg_move): Update for new strict argument.
17685 (mem_operand_gpr): Disallow all non-offsettable addresses.
17686 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
17688 2018-03-09 Jakub Jelinek <jakub@redhat.com>
17691 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
17692 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
17693 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
17696 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
17697 decl, use remap_type if we want to use the type.
17699 2018-03-09 Martin Sebor <msebor@redhat.com>
17701 PR tree-optimization/84526
17702 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
17704 (builtin_access::generic_overlap): Be prepared to handle non-array
17707 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
17709 PR rtl-optimization/84682
17710 * lra-constraints.c (process_address_1): Check is_address flag
17711 for address constraints.
17712 (process_alt_operands): Likewise.
17713 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
17714 preprocess_constraints.
17715 * recog.h (preprocess_constraints): Add oploc parameter.
17717 * recog.c (preprocess_constraints): Test address_operand for
17718 CT_ADDRESS constraints.
17720 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17723 * lra-assigns.c (assign_by_spills): Return a flag of reload
17724 assignment failure. Do not process the reload assignment
17725 failures. Do not spill other reload pseudos if they has the same
17727 (lra_assign): Add a return arg. Set up from the result of
17728 assign_by_spills call.
17729 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17730 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17731 usage_insns if it is not NULL.
17732 (spill_hard_reg_in_range): New function.
17733 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17734 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17735 function prototypes.
17736 (lra_assign): Change prototype.
17737 * lra.c (lra): Add code to deal with fails by splitting hard reg
17740 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17743 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
17744 Accept complain bool parameter. Only emit errors if it is true.
17745 (arm_parse_cpu_option_name): Likewise.
17746 (arm_target_thumb_only): Adjust callers of the above.
17747 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
17748 prototype to take a default true bool parameter.
17749 (arm_parse_arch_option_name): Likewise.
17751 2018-03-09 David Malcolm <dmalcolm@redhat.com>
17752 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
17756 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
17757 * configure: Regenerate.
17758 * configure.ac ("linker --version-script option"): New.
17759 ("linker soname option"): New.
17761 2018-03-09 Richard Biener <rguenther@suse.de>
17763 PR tree-optimization/84775
17764 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
17765 immediate uses of predicate stmts and mark them modified.
17768 PR tree-optimization/84178
17769 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17771 (version_loop_for_if_conversion): Delay update_ssa call.
17772 (tree_if_conversion): Delay update_ssa until after predicate
17775 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
17778 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
17779 when the function accesses prior frames.
17781 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17784 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
17785 gen_llsym, otherwise call maybe_gen_llsym.
17787 PR inline-asm/84742
17788 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
17789 has ',' character inside of it.
17791 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17794 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
17795 as clobbering CC_REGNUM.
17797 2018-03-08 Richard Biener <rguenther@suse.de>
17799 PR middle-end/84552
17800 * tree-scalar-evolution.c: Include tree-into-ssa.h.
17801 (follow_copies_to_constant): Do not follow SSA names registered
17804 2018-03-08 Richard Biener <rguenther@suse.de>
17806 PR tree-optimization/84178
17807 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17809 (version_loop_for_if_conversion): Delay update_ssa call.
17810 (tree_if_conversion): Delay update_ssa until after predicate
17813 2018-03-08 David Malcolm <dmalcolm@redhat.com>
17815 PR tree-optimization/84178
17816 * tree-if-conv.c (release_bb_predicate): Remove the
17817 the assertion that the stmts have NULL use_ops.
17818 Discard the statements, asserting that they haven't
17819 yet been added to a BB.
17821 2018-03-08 Richard Biener <rguenther@suse.de>
17823 PR tree-optimization/84746
17824 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
17825 (phi_translate): Pass in destination ANTIC_OUT set.
17826 (phi_translate_1): Likewise. For a simplified result lookup
17827 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
17828 (phi_translate_set): Adjust.
17829 (do_pre_regular_insertion): Likewise.
17830 (do_pre_partial_partial_insertion): Likewise.
17832 2018-03-08 Martin Liska <mliska@suse.cz>
17834 PR gcov-profile/84735
17835 * doc/gcov.texi: Document usage of profile files.
17836 * gcov-io.h: Document changes in the format.
17838 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
17842 * dwarf2out.c (struct dw_line_info_table): Update comments for
17844 (FORCE_RESET_NEXT_VIEW): New.
17845 (FORCE_RESETTING_VIEW_P): New.
17846 (RESETTING_VIEW_P): Check for -1 too.
17847 (ZERO_VIEW_P): Likewise.
17848 (new_line_info_table): Force-reset next view.
17849 (dwarf2out_begin_function): Likewise.
17850 (dwarf2out_source_line): Simplify zero_view_p initialization.
17851 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
17852 view directly. Omit view when omitting .loc at line 0.
17854 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17856 PR tree-optimization/84740
17857 * tree-switch-conversion.c (process_switch): Call build_constructors
17858 only if info.phi_count is non-zero.
17860 PR tree-optimization/84739
17861 * tree-tailcall.c (find_tail_calls): Check call arguments against
17862 DECL_ARGUMENTS (current_function_decl) rather than
17863 DECL_ARGUMENTS (func) when checking for tail recursion.
17865 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17867 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
17868 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
17869 Volker Reichelt's entry and add entries for people that perform
17870 GCC fuzzy testing and report numerous bugs.
17872 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17875 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
17876 readonly data in sdata, if that is disabled.
17877 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
17878 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
17879 -mreadonly-in-sdata option.
17881 2018-03-07 Martin Sebor <msebor@redhat.com>
17883 PR tree-optimization/84468
17884 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
17885 basic block when looking for nul assignment.
17887 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
17890 * except.h (output_function_exception_table): Adjust prototype.
17891 * except.c (output_function_exception_table): Remove FNNAME parameter
17892 and add SECTION parameter. Ouput one part of the table at a time.
17893 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
17894 the first part of the exception table and emit unwind directives.
17895 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
17896 (i386_pe_seh_cold_init): Likewise.
17897 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
17898 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
17899 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
17900 (ix86_output_call_insn): Emit a nop in one more case for SEH.
17901 * config/i386/winnt.c: Include except.h.
17902 (struct seh_frame_state): Add reg_offset, after_prologue and
17903 in_cold_section fields.
17904 (i386_pe_seh_end_prologue): Set seh->after_prologue.
17905 (i386_pe_seh_cold_init): New function.
17906 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
17907 to seh->in_cold_section.
17908 (seh_emit_push): Record the offset of the push.
17909 (seh_emit_save): Record the offet of the save.
17910 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
17911 Test seh->after_prologue to disregard the epilogue.
17912 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
17913 (i386_pe_end_cold_function): New function.
17915 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17918 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
17919 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
17922 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
17924 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
17925 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
17927 PR middle-end/84723
17928 * multiple_target.c: Include tree-inline.h and intl.h.
17929 (expand_target_clones): Diagnose and fail if node->definition and
17930 !tree_versionable_function_p (node->decl).
17932 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
17934 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
17936 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
17937 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17938 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
17940 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17943 * combine.c (try_combine): Use reg_or_subregno instead of handling
17944 just paradoxical SUBREGs and REGs.
17946 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
17948 * config/arc/arc.c (arc_finalize_pic): Remove function.
17949 (arc_must_save_register): We use single base PIC register, remove
17950 checks to save/restore the PIC register.
17951 (arc_expand_prologue): Likewise.
17952 * config/arc/arc-protos.h (arc_set_default_type_attributes):
17954 (arc_verify_short): Likewise.
17955 (arc_attr_type): Likewise.
17956 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
17957 (walk_stores): Likewise.
17958 (arc_address_cost): Make it static.
17959 (arc_verify_short): Likewise.
17960 (branch_dest): Likewise.
17961 (arc_attr_type): Likewise.
17962 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
17963 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
17964 (arc_final_prescan_insn): Remove inserting the nops due to
17965 hardware hazards. It is done in reorg step.
17966 (insn_length_variant_t): Remove.
17967 (insn_length_parameters_t): Likewise.
17968 (arc_insn_length_parameters): Likewise.
17969 (arc_get_insn_variants): Likewise.
17970 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
17972 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17974 PR inline-asm/84683
17975 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
17978 PR tree-optimization/84687
17979 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
17981 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
17983 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17985 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
17986 Rename to ppc_speculation_barrier.
17987 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
17988 __builtin_ppc_speculation_barrier.
17990 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17993 * combine.c (combine_simplify_rtx): Don't try to simplify if
17994 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
17997 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
17999 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
18000 to 32 bytes when compiling for POWER9.
18002 2018-03-05 Jakub Jelinek <jakub@redhat.com>
18005 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
18006 regparm >= 3 with no arg reg available also for calls with
18007 flag_force_indirect_call. Pass decl to ix86_function_regparm.
18010 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
18012 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
18014 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
18017 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
18019 2018-03-05 Richard Biener <rguenther@suse.de>
18021 PR tree-optimization/84486
18022 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
18023 When inserting a __builtin_assume_aligned call set the LHS
18024 SSA name alignment info accordingly.
18026 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
18028 PR tree-optimization/84114
18029 * config/aarch64/aarch64.c (aarch64_reassociation_width)
18030 Avoid reassociation of FLOAT_MODE addition.
18032 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
18034 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
18035 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
18036 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
18037 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
18038 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
18039 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
18040 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
18042 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
18043 __builtin_ia32_wbinvd): New builtins.
18044 (SPECIAL_ARGS2): New.
18045 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
18046 (SPECIAL_ARGS2): New.
18047 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
18048 (ix86_valid_target_attribute_inner_p): Ditto.
18049 (ix86_init_mmx_sse_builtins): Add special_args2.
18050 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
18051 TARGET_WBNOINVD_P): New.
18052 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
18053 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
18054 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
18055 * config/i386/immintrin.h (_wbinvd): New intrinsic.
18056 * config/i386/pconfigintrin.h: New file.
18057 * config/i386/wbnoinvdintrin.h: Ditto.
18058 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
18060 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
18062 2018-03-05 Richard Biener <rguenther@suse.de>
18064 PR tree-optimization/84670
18065 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
18067 (BB_VISITED_WITH_VISITED_SUCCS): New define.
18068 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
18069 (compute_antic_aux): Only assert the number of values in ANTIC_IN
18070 doesn't grow if all successors (recursively) were visited at least
18073 2018-03-05 Richard Biener <rguenther@suse.de>
18075 PR tree-optimization/84650
18076 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
18077 if executed in the loop pipeline.
18079 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
18081 * doc/configfiles.texi (Configuration Files): Move info about
18082 conditionalizing $target-protos.h to...
18083 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
18084 differs from $target-protos.h.
18086 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
18087 Chung-Ju Wu <jasonwucj@gmail.com>
18089 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
18090 * config/nds32/nds32-multiple.md (setmemsi): Define.
18091 * config/nds32/nds32-memory-manipulation.c
18092 (nds32_gen_dup_4_byte_to_word_value): New.
18093 (emit_setmem_word_loop): New.
18094 (emit_setmem_byte_loop): New.
18095 (nds32_expand_setmem_loop): New.
18096 (nds32_expand_setmem_loop_v3m): New.
18097 (nds32_expand_setmem_unroll): New.
18098 (nds32_expand_setmem): New.
18100 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
18101 Chung-Ju Wu <jasonwucj@gmail.com>
18103 * config/nds32/nds32-memory-manipulation.c
18104 (nds32_emit_load_store): New.
18105 (nds32_emit_post_inc_load_store): New.
18106 (nds32_emit_mem_move): New.
18107 (nds32_emit_mem_move_block): New.
18108 (nds32_expand_movmemsi_loop_unknown_size): New.
18109 (nds32_expand_movmemsi_loop_known_size): New.
18110 (nds32_expand_movmemsi_loop): New.
18111 (nds32_expand_movmemsi_unroll): New.
18112 (nds32_expand_movmemqi): Rename ...
18113 (nds32_expand_movmemsi): ... to this.
18114 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
18115 (movmemsi): ... to this.
18116 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
18117 (nds32_expand_movmemsi): ... to this.
18119 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
18120 Monk Chiang <sh.chiang04@gmail.com>
18121 Chung-Ju Wu <jasonwucj@gmail.com>
18123 * config/nds32/nds32-protos.h
18124 (nds32_expand_load_multiple): New arguments.
18125 (nds32_expand_store_multiple): Ditto.
18126 (nds32_valid_multiple_load_store): Rename ...
18127 (nds32_valid_multiple_load_store_p): ... to this.
18128 * config/nds32/nds32-memory-manipulation.c
18129 (nds32_expand_load_multiple): Refine implementation.
18130 (nds32_expand_store_multiple): Ditto.
18131 * config/nds32/nds32-multiple.md
18132 (load_multiple): Update nds32_expand_load_multiple interface.
18133 (store_multiple): Update nds32_expand_store_multiple interface.
18134 * config/nds32/nds32-predicates.c
18135 (nds32_valid_multiple_load_store): Rename ...
18136 (nds32_valid_multiple_load_store_p): ... to this and refine
18138 * config/nds32/predicates.md
18139 (nds32_load_multiple_and_update_address_operation): New predicate.
18140 (nds32_store_multiple_and_update_address_operation): New predicate.
18142 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
18143 Chung-Ju Wu <jasonwucj@gmail.com>
18145 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
18146 (combo): New attribute.
18147 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
18149 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
18151 * config/nds32/nds32.opt: Change -mcmodel= default value.
18153 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
18154 Monk Chiang <sh.chiang04@gmail.com>
18155 Chung-Ju Wu <jasonwucj@gmail.com>
18157 * config/nds32/constants.md (unspec_element): New enum.
18158 * config/nds32/constraints.md (Umw): New constraint.
18159 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
18160 * config/nds32/nds32-intrinsic.md: Likewise.
18161 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
18162 (nds32_valid_smw_lwm_base_p): New.
18163 (nds32_output_smw_single_word): New.
18164 (nds32_output_lmw_single_word): New.
18165 (nds32_expand_unaligned_load): New.
18166 (nds32_expand_unaligned_store): New.
18167 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
18168 (nds32_output_smw_single_word): Declare.
18169 (nds32_output_lmw_single_word): Declare.
18170 (nds32_expand_unaligned_load): Declare.
18171 (nds32_expand_unaligned_store): Declare.
18172 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
18173 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
18174 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
18175 NDS32_BUILTIN_UASTORE_DW.
18176 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
18179 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
18180 Kito Cheng <kito.cheng@gmail.com>
18181 Chung-Ju Wu <jasonwucj@gmail.com>
18183 * config/nds32/nds32-intrinsic.c
18184 (nds32_expand_builtin_null_ftype_reg): Delete.
18185 (nds32_expand_builtin_reg_ftype_imm): Ditto.
18186 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
18187 (nds32_read_argument): New.
18188 (nds32_legitimize_target): Ditto.
18189 (nds32_legitimize_argument): Ditto.
18190 (nds32_check_constant_argument): Ditto.
18191 (nds32_expand_unop_builtin): Ditto.
18192 (nds32_expand_unopimm_builtin): Ditto.
18193 (nds32_expand_binop_builtin): Ditto.
18194 (nds32_builtin_decl_impl): Ditto.
18195 (builtin_description): Ditto.
18196 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
18197 (nds32_init_builtins_impl): Ditto.
18198 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
18199 (nds32_builtin_decl): New.
18200 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
18201 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
18203 2018-03-02 Jeff Law <law@redhat.com>
18205 * reorg.c (stop_search_p): Handle DEBUG_INSN.
18206 (redundant_insn, fill_simple_delay_slots): Likewise.
18207 (fill_slots_from_thread): Likewise.
18208 * resource.c (mark_referenced_resources): Likewise.
18209 (mark_set_resources, find_dead_or_set_registers): Likewise.
18211 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18213 * substring-locations.h (format_warning_va): Formatting fix for
18214 ATTRIBUTE_GCC_DIAG.
18215 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
18217 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
18218 * substring-locations.c: Include intl.h.
18219 (format_warning_va): Turned into small wrapper around
18220 format_warning_n_va, renamed to ...
18221 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
18222 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
18224 (format_warning_at_substring_n): New function.
18225 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
18226 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
18227 format_warning_at_substring with just a shorter name instead of
18228 const function pointer.
18229 (fmtwarn_n): New function.
18230 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
18231 appropriate, get rid of all the fmtstr temporaries, move conditionals
18232 with G_() wrapped string literals directly into fmtwarn arguments,
18233 cast dir.len to (int), formatting fixes.
18235 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
18237 * doc/invoke.texi: Remove "Cilk Plus" references.
18239 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18240 Richard Biener <rguenther@suse.de>
18243 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
18244 for error or warning attributes if CALL_FROM_THUNK_P is set.
18247 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18250 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
18251 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
18254 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
18255 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
18257 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
18258 instead of -1U in last predictors element's probability member.
18260 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
18263 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
18264 arguments if they are comparable.
18266 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
18268 PR tree-optimization/84634
18269 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
18270 masks and masked_loop_p with a single loop_masks, making sure it's
18271 null for bb vectorization.
18273 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
18275 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
18276 (vect_analyze_data_ref_access): Use loop->safe_len rather than
18277 loop->force_vectorize to check whether there is no alias.
18279 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18282 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
18284 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
18286 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
18287 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
18288 instead of a loop around prev_real_insn.
18289 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
18292 PR inline-asm/84625
18293 * config/i386/i386.c (ix86_print_operand): Use conditional
18294 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
18297 2018-03-02 Richard Biener <rguenther@suse.de>
18299 PR tree-optimization/84427
18300 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
18301 (bitmap_set_subtract_values): Rewrite to handle multiple
18304 (prune_clobbered_mems): Likewise.
18305 (phi_translate): Take edge instead of pred/phiblock.
18306 (phi_translate_1): Likewise.
18307 (phi_translate_set): Likewise. Insert all translated
18308 exprs for a value into the set, keeping possibly multiple
18309 expressions per value.
18310 (compute_antic_aux): Adjust for phi_translate changes.
18311 When intersecting union the expressions and prune those
18312 not in the final value set, keeping possibly multiple
18313 expressions per value. Do not use value-insertion
18314 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
18315 all expressions. Add verification that the value-sets
18316 only shrink during iteration.
18317 (compute_partial_antic_aux): Adjust for the phi_translate changes.
18318 (do_pre_regular_insertion): Likewise.
18319 (do_pre_partial_partial_insertion): Likewise.
18321 2018-03-02 Richard Biener <rguenther@suse.de>
18324 * config/darwin.c (saved_debug_info_level): New static global.
18325 (darwin_asm_lto_start): Disable debug info generation for LTO out.
18326 (darwin_asm_lto_end): Restore debug info generation settings.
18328 2018-03-01 Martin Liska <mliska@suse.cz>
18331 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
18332 volatile arguments.
18334 2018-03-01 Richard Biener <rguenther@suse.de>
18337 * dwarf2out.c (gen_variable_die): Properly handle late VLA
18338 type annotation with LTO when debug was disabled at compile-time.
18340 2018-03-01 Matthew Fortune <mfortune@gmail.com>
18342 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
18344 (mips_final_postscan_insn): Likewise.
18346 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
18348 PR rtl-optimization/84528
18349 * alias.c (init_alias_target): Add commentary.
18350 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
18351 a unique base value if the frame pointer is not eliminated
18352 to the stack pointer.
18354 2018-03-01 Tom de Vries <tom@codesourcery.com>
18356 PR rtl-optimization/83327
18357 * lra-int.h (hard_regs_spilled_into): Declare.
18358 * lra.c (hard_regs_spilled_into): Define.
18359 (init_reg_info): Init hard_regs_spilled_into.
18360 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
18361 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
18362 (process_bb_lives): Handle hard_regs_spilled_into.
18363 (lra_create_live_ranges_1): Before doing liveness propagation, clear
18364 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
18366 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
18368 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
18369 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
18370 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
18371 * config/rs6000/aix72.h: New file.
18373 2018-02-28 Jakub Jelinek <jakub@redhat.com>
18375 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
18376 instead of warning_at with conditional singular and plural messages
18380 * stor-layout.c (update_alignment_for_field): For
18381 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
18382 && !DECL_PACKED (field), do the alignment update, just use
18383 only desired_align instead of MAX (type_align, desired_align)
18385 (place_field): Don't do known_align < desired_align handling
18386 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
18387 is non-NULL, instead do it after rli->prev_field handling and
18388 only if not within a bitfield word. For DECL_PACKED (field)
18389 use type_align of BITS_PER_UNIT.
18391 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
18393 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
18394 superfluous parentheses and trailing spaces.
18396 2018-02-28 Richard Biener <rguenther@suse.de>
18398 PR tree-optimization/84584
18399 * graphite-scop-detection.c (scop_detection::add_scop): Discard
18400 SCoPs with fake exit edge.
18402 2018-02-28 Martin Liska <mliska@suse.cz>
18405 * timevar.c (timer::print): Fix format to properly print 100%
18408 2018-02-28 Richard Biener <rguenther@suse.de>
18410 PR middle-end/84607
18411 * genmatch.c (capture_info::walk_match): Do not mark
18412 captured expressions without operands as expr_p given
18413 they act more like predicates and should be subject to
18414 "lost tail" side-effect preserving.
18416 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
18418 PR rtl-optimization/81611
18419 * auto-inc-dec.c (attempt_change): Move dead note from
18420 mem_insn if it's the next use of regno
18421 (find_address): Take address use of reg holding
18422 non-incremented value. Add parm to limit search to the named
18424 (merge_in_block): Attempt to use a mem insn that is the next
18425 use of the original regno.
18427 2018-02-27 Martin Sebor <msebor@redhat.com>
18430 * doc/invoke.texi (-Wmissing-attributes): New option.
18431 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
18433 2018-02-27 Martin Sebor <msebor@redhat.com>
18435 PR translation/84207
18436 * diagnostic-core.h (warning_n, error_n, inform_n): Change
18437 n argument to unsigned HOST_WIDE_INT.
18438 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
18439 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
18440 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
18441 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
18443 2018-02-27 Richard Biener <rguenther@suse.de>
18445 PR tree-optimization/84512
18446 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
18447 Do not use the estimate returned from record_stmt_cost for
18448 the scalar iteration cost but sum properly using add_stmt_cost.
18450 2018-02-27 Richard Biener <rguenther@suse.de>
18452 PR tree-optimization/84466
18453 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
18454 Adjust last change to less strictly validate use operands.
18456 2018-02-27 Martin Liska <mliska@suse.cz>
18458 PR gcov-profile/84548
18459 * gcov.c (process_file): Allow partial overlap and consider it
18460 also as group functions.
18461 (output_lines): Properly calculate range of lines for a group.
18463 2018-02-27 Martin Liska <mliska@suse.cz>
18465 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
18466 'ggc' suffixes. Change first column width.
18467 (timer::print): Fix formatting of the column.
18469 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
18471 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
18472 preserve inline entry blocks for the sake of debug inline
18473 entry point markers alone.
18474 (remove_unused_locals): Suggest in comments a better place to
18475 force the preservation of inline entry blocks that are
18476 otherwise unused, but do not preserve them.
18478 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18480 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
18482 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18485 * config/i386/constraints.md (Bs): Replace
18486 ix86_indirect_branch_register with
18487 TARGET_INDIRECT_BRANCH_REGISTER.
18489 * config/i386/i386.md (indirect_jump): Likewise.
18490 (tablejump): Likewise.
18491 (*sibcall_memory): Likewise.
18492 (*sibcall_value_memory): Likewise.
18493 Peepholes of indirect call and jump via memory: Likewise.
18494 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
18495 (*sibcall_value_GOT_32): Likewise.
18496 * config/i386/predicates.md (indirect_branch_operand): Likewise.
18497 (GOT_memory_operand): Likewise.
18498 (call_insn_operand): Likewise.
18499 (sibcall_insn_operand): Likewise.
18500 (GOT32_symbol_operand): Likewise.
18501 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
18503 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
18505 PR rtl-optimization/83496
18506 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
18507 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
18508 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
18509 redundant insn, if any.
18510 (relax_delay_slots): Likewise.
18511 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
18513 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
18515 PR tree-optimization/83965
18516 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
18517 that grouped statements are part of a reduction chain. Return
18518 true if the statement is not marked as a reduction itself but
18519 is part of a group.
18520 (vect_recog_dot_prod_pattern): Don't check whether the statement
18521 is part of a group here.
18522 (vect_recog_sad_pattern): Likewise.
18523 (vect_recog_widen_sum_pattern): Likewise.
18525 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
18528 * final.c (rest_of_clean_state): Also look for calls inside sequences.
18530 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18533 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
18535 (ix86_output_indirect_function_return): New prototype.
18536 (ix86_split_simple_return_pop_internal): Likewise.
18537 * config/i386/i386.c (indirect_return_via_cx): New.
18538 (indirect_return_via_cx_bnd): Likewise.
18539 (indirect_thunk_name): Handle return va CX_REG.
18540 (output_indirect_thunk_function): Create alias for
18541 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
18542 (ix86_output_indirect_jmp): Remove the bool argument.
18543 (ix86_output_indirect_function_return): New function.
18544 (ix86_split_simple_return_pop_internal): Likewise.
18545 * config/i386/i386.md (*indirect_jump): Don't pass false
18546 to ix86_output_indirect_jmp.
18547 (*tablejump_1): Likewise.
18548 (simple_return_pop_internal): Change it to define_insn_and_split.
18549 Call ix86_split_simple_return_pop_internal to split it for
18550 -mfunction-return=.
18551 (simple_return_indirect_internal): Call
18552 ix86_output_indirect_function_return instead of
18553 ix86_output_indirect_jmp.
18555 2018-02-26 Jakub Jelinek <jakub@redhat.com>
18558 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
18559 memset and value initialization afterwards.
18561 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
18563 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
18565 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18568 * common/config/aarch64/aarch64-common.c
18569 (aarch_option_optimization_table[]): Switch
18570 off fomit-frame-pointer
18572 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
18573 Chung-Ju Wu <jasonwucj@gmail.com>
18575 * config/nds32/nds32-multiple.md (load_multiple): Disallow
18577 (store_multiple): Ditto.
18579 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
18581 * config.gcc: Add --with-cpu support for nds32 target.
18582 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
18583 * config/nds32/nds32.opt: Add -mcpu= option.
18585 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
18587 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
18588 isel=yes): Warn for these deprecated options.
18590 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
18592 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
18593 ISA_2_5_MASKS_EMBEDDED.
18595 2018-02-23 Jakub Jelinek <jakub@redhat.com>
18597 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
18598 p->max as pointers rather than using iterative_hash_expr.
18600 2018-02-23 Carl Love <cel@us.ibm.com>
18602 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
18603 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
18605 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
18606 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
18607 P8V_BUILTIN_VEC_VUNSIGNED2.
18609 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
18612 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
18613 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
18615 * lra-constraints.c (curr_insn_transform): Set up
18616 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
18618 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
18619 * lra-eliminations.c (spill_pseudos): Ditto.
18620 (process_insn_for_elimination): Ditto.
18621 * lra-lives.c (reg_early_clobber_p): Use the new macros.
18622 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
18623 LRA_NON_CLOBBERED_ALT.
18625 2018-02-22 Martin Sebor <msebor@redhat.com>
18627 PR tree-optimization/84480
18628 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
18629 to maybe_diag_stxncpy_trunc. Call it.
18630 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
18631 from gimple_fold_builtin_strcpy. Print inlining stack.
18632 (handle_builtin_stxncpy): Print inlining stack.
18633 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
18635 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
18638 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
18639 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
18640 and -fcheck-pointer-bounds are used together.
18641 (indirect_thunk_prefix): New enum.
18642 (indirect_thunk_need_prefix): New function.
18643 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
18644 "_nt" instead of "_bnd" for NOTRACK prefix.
18645 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
18646 (output_indirect_thunk_function): Likewise.
18648 (ix86_code_end): Update output_indirect_thunk_function calls.
18649 (ix86_output_indirect_branch_via_reg): Replace
18650 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
18651 (ix86_output_indirect_branch_via_push): Likewise.
18652 (ix86_output_function_return): Likewise.
18653 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
18654 incompatible with -fcf-protection=branch and
18655 -fcheck-pointer-bounds.
18657 2018-02-22 Steve Ellcey <sellcey@cavium.com>
18660 * config/aarch64/aarch64.c (aarch64_print_address_internal):
18661 Change gcc_assert call to output_operand_lossage.
18663 2018-02-22 Steve Ellcey <sellcey@cavium.com>
18665 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
18667 2018-02-22 DJ Delorie <dj@redhat.com>
18668 Sebastian Perta <sebastian.perta@renesas.com>
18669 Oleg Endo <olegendo@gcc.gnu.org>
18671 * config/rx/rx.c (rx_rtx_costs): New function.
18672 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
18674 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18676 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
18678 2018-02-22 Martin Liska <mliska@suse.cz>
18681 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
18682 Add "native" as a possible value.
18684 2018-02-22 Martin Liska <mliska@suse.cz>
18687 * config/i386/i386.c (ix86_option_override_internal):
18688 Add "native" as a possible value for -march and -mtune.
18690 2018-02-22 Jakub Jelinek <jakub@redhat.com>
18693 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
18694 to all type variants.
18696 PR tree-optimization/84503
18697 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
18698 width as info->bitpos + info->bitsize - start.
18699 (merged_store_group::merge_overlapping): Simplify width computation.
18700 (check_no_overlap): New function.
18701 (imm_store_chain_info::try_coalesce_bswap): Compute expected
18702 start + width and last_order of the group, fail if check_no_overlap
18704 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
18705 to group if check_no_overlap fails.
18707 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18709 * config/rs6000/altivec.md: Delete contraint arguments to
18710 define_expand, define_split, and define_peephole2, and in
18711 define_insn_and_split if always unused.
18712 * config/rs6000/darwin.md: Ditto.
18713 * config/rs6000/dfp.md: Ditto.
18714 * config/rs6000/rs6000.md: Ditto.
18715 * config/rs6000/sync.md: Ditto.
18716 * config/rs6000/vector.md: Ditto.
18717 * config/rs6000/vsx.md: Ditto.
18719 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18721 * config/rs6000/altivec.md: Write output control strings as braced
18722 blocks instead of double-quoted strings.
18723 * config/rs6000/darwin.md: Ditto.
18724 * config/rs6000/rs6000.md: Ditto.
18725 * config/rs6000/vector.md: Ditto.
18726 * config/rs6000/vsx.md: Ditto.
18728 2018-02-21 Jason Merrill <jason@redhat.com>
18730 PR c++/84314 - ICE with templates and fastcall attribute.
18731 * attribs.c (build_type_attribute_qual_variant): Remove assert.
18733 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18735 * ipa-cp.c (determine_versionability): Fix comment typos.
18737 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18740 * ipa-cp.c (determine_versionability): Do not version functions caling
18743 2018-02-21 Martin Liska <mliska@suse.cz>
18746 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
18747 Add "native" as a possible value.
18748 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
18749 the macro when native cpu detection is available.
18751 2018-02-21 Martin Liska <mliska@suse.cz>
18754 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
18755 Add "native" as a possible value.
18756 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
18757 when native cpu detection is available.
18759 2018-02-21 Jakub Jelinek <jakub@redhat.com>
18760 Martin Sebor <msebor@redhat.com>
18762 PR tree-optimization/84478
18763 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
18765 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
18766 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
18767 support which is conservatively correct, for 2 only stay conservative
18768 for maxlen. Formatting and comment capitalization fixes. Add STRICT
18769 argument to the 2 argument get_range_strlen, adjust 6 arg
18770 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
18772 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
18773 (gimple_fold_builtin_strlen): Pass true as last argument to
18776 2018-02-20 Martin Sebor <msebor@redhat.com>
18778 PR middle-end/84095
18779 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
18780 (builtin_memref::set_base_and_offset): Same. Handle inner references.
18781 (builtin_memref::builtin_memref): Factor out parts into
18782 set_base_and_offset and call it.
18784 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18786 PR middle-end/84406
18787 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
18788 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
18789 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
18790 search at the associated MODE_INT.
18792 2018-02-20 Jeff Law <law@redhat.com>
18794 PR middle-end/82123
18795 PR tree-optimization/81592
18796 PR middle-end/79257
18797 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
18798 for range data rather than using global data.
18799 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
18800 range data rather than using global data.
18801 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
18802 pass it to children as needed.
18803 (struct directive::fmtresult): Similarly.
18804 (struct directive::set_width): Similarly.
18805 (struct directive::set_precision): Similarly.
18806 (format_integer, format_directive, parse_directive): Similarly.
18807 (format_none): Accept unnamed vr_values parameter.
18808 (format_percent, format_floating, format_character): Similarly.
18809 (format_string, format_plain): Similarly.
18810 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
18811 the EVRP range analyzer for range data rather than using global data.
18812 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
18813 gimple-ssa-evrp-analyze.h
18814 (class sprintf_dom_walker): Add after_dom_children member function.
18815 Add evrp_range_analyzer member.
18816 (sprintf_dom_walker::before_dom_children): Call into the EVRP
18817 range analyzer as needed.
18818 (sprintf_dom_walker::after_dom_children): New member function.
18819 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
18821 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
18822 (evrp_range_analyzer::pop_to_marker): Likewise.
18824 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18826 PR tree-optimization/84419
18827 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
18828 with the required type if its current type is compatible but
18831 2018-02-20 Jakub Jelinek <jakub@redhat.com>
18833 PR middle-end/82004
18834 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
18835 after vectorization.
18837 2018-02-20 Martin Liska <mliska@suse.cz>
18840 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
18841 possible values if we don't have a hint.
18843 2018-02-20 Martin Liska <mliska@suse.cz>
18847 * final.c (shorten_branches): Build align_tab array with one
18849 * opts.c (finish_options): Add alignment option limit check.
18850 (MAX_CODE_ALIGN): Likewise.
18851 (MAX_CODE_ALIGN_VALUE): Likewise.
18852 * doc/invoke.texi: Document maximum allowed option value for
18853 all -falign-* options.
18855 2018-02-19 Jakub Jelinek <jakub@redhat.com>
18858 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
18859 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
18860 * var-tracking.c (emit_note_insn_var_location): Remove all references
18861 to NOTE_INSN_CALL_ARG_LOCATION.
18862 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
18863 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
18864 Use copy_rtx_if_shared.
18865 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
18866 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
18867 (dwarf2out_var_location): Remove handling of
18868 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
18870 * final.c (final_scan_insn): Remove all references to
18871 NOTE_INSN_CALL_ARG_LOCATION.
18872 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
18873 before dumping final insns.
18874 * except.c (emit_note_eh_region_end): Remove all references to
18875 NOTE_INSN_CALL_ARG_LOCATION.
18876 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
18877 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
18878 * config/arc/arc.c (hwloop_optimize): Likewise.
18879 * config/arm/arm.c (create_fix_barrier): Likewise.
18880 * config/s390/s390.c (s390_chunkify_start): Likewise.
18881 * config/sh/sh.c (find_barrier): Likewise.
18882 * config/i386/i386.c (rest_of_insert_endbranch,
18883 ix86_seh_fixup_eh_fallthru): Likewise.
18884 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
18885 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
18886 * config/frv/frv.c (frv_function_prologue): Likewise.
18887 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
18889 (note_outside_basic_block_p): Remove all references to
18890 NOTE_INSN_CALL_ARG_LOCATION.
18891 * gengtype.c (adjust_field_rtx_def): Likewise.
18892 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
18894 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
18895 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
18898 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
18901 PR tree-optimization/84452
18902 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
18903 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
18906 2018-02-19 Martin Liska <mliska@suse.cz>
18909 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
18911 2018-02-19 Martin Liska <mliska@suse.cz>
18912 Richard Sandiford <richard.sandiford@linaro.org>
18914 PR tree-optimization/82491
18915 * gimple-fold.c (get_base_constructor): Make earlier bail out
18918 2018-02-19 Carl Love <cel@us.ibm.com>
18920 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
18921 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
18923 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
18924 P8V_BUILTIN_VEC_NEG.
18926 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
18928 * config/rl78/rl78.md (movdf): New define expand.
18930 2018-02-19 Martin Liska <mliska@suse.cz>
18933 * doc/invoke.texi: Fix typo.
18934 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
18936 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
18938 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
18939 handle rs6000_single_float and rs6000_double_float specially for
18942 2018-02-16 Jeff Law <law@redhat.com>
18944 * config/rx/rx.c (add_pop_cfi_notes): New function.;
18945 (pop_regs): Use it.
18947 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18950 * ipa-inline.c (inline_small_functions): Fix a typo.
18952 2018-02-16 Nathan Sidwell <nathan@acm.org>
18954 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
18956 2018-02-16 Carl Love <cel@us.ibm.com>
18958 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
18959 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
18960 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
18961 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
18962 expansion to P8V_BUILTIN_VEC_FLOAT2.
18964 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
18966 PR rtl-optimization/70023
18967 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
18968 src_regno into account.
18970 2018-02-16 Carl Love <cel@us.ibm.com>
18972 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
18973 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
18974 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
18975 * config/rs6000/rs6000.c: Remove case statements for
18976 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
18977 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
18978 and P9V_BUILTIN_VEC_VINSERT4B.
18979 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
18980 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
18981 * config/rs6000/vsx.md:
18982 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
18985 2018-02-16 Carl Love <cel@us.ibm.com>
18987 * config/rs6000/altivec.h: Add builtin names vec_extract4b
18989 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
18991 * config/rs6000/rs6000-c.c: Add the definitions for
18992 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
18993 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
18994 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
18995 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
18996 definition for insert4b and define insn *insert3b_internal.
18997 * doc/extend.texi: Add documentation for vec_extract4b.
18999 2018-02-16 Nathan Sidwell <nathan@acm.org>
19001 * doc/extend.texi (Backwards Compatibility): Mention friend
19002 injection. Note for-scope is deprecated.
19003 * doc/invoke.texi (-ffriend-injection): Deprecate.
19005 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
19007 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
19008 that moved to I2, also allow destinations that are a paradoxical
19009 subreg (instead of a normal reg).
19011 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
19014 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
19017 2018-02-16 Richard Biener <rguenther@suse.de>
19019 PR tree-optimization/84037
19020 PR tree-optimization/84016
19022 * config/i386/i386.c (ix86_builtin_vectorization_cost):
19023 Adjust vec_construct for the fact we need additional higher latency
19024 128bit inserts for AVX256 and AVX512 vector builds.
19025 (ix86_add_stmt_cost): Scale vector construction cost for
19028 2018-02-16 Richard Biener <rguenther@suse.de>
19030 PR tree-optimization/84417
19031 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
19032 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
19033 (non_rewritable_lvalue_p): Likewise, use poly-ints.
19035 2018-02-16 Martin Liska <mliska@suse.cz>
19038 * internal-fn.def (ASAN_CHECK): Set proper flags.
19039 (ASAN_MARK): Likewise.
19041 2018-02-16 Julia Koval <julia.koval@intel.com>
19043 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
19044 from PTA_CANNONLAKE.
19046 2018-02-16 Jakub Jelinek <jakub@redhat.com>
19049 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
19050 Use ++iter rather than iter++ for std::list iterators.
19051 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
19052 defer deleting them until all nodes in the forest are processed. Do
19053 free even leaf nodes. Change to_process into auto_vec.
19056 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
19057 * vec.h (vec_default_construct): Use memset instead of placement new
19058 if BROKEN_VALUE_INITIALIZATION is defined.
19059 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
19060 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
19063 PR rtl-optimization/83723
19064 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
19065 * lra.c (lra_substitute_pseudo): Likewise. If true, use
19066 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
19068 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
19070 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
19072 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
19074 PR rtl-optimization/81443
19075 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
19076 from inner REGs to paradoxical SUBREGs.
19078 2018-02-16 Richard Biener <rguenther@suse.de>
19080 PR tree-optimization/84399
19081 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
19082 For operands we can analyze at their definition make sure we can
19083 analyze them at each use as well.
19085 2018-02-16 Richard Biener <rguenther@suse.de>
19087 PR tree-optimization/84190
19088 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
19089 volatile accesses if the decl isn't volatile.
19091 2018-02-15 Jason Merrill <jason@redhat.com>
19093 PR c++/84314 - ICE with templates and fastcall attribute.
19094 * attribs.c (build_type_attribute_qual_variant): Don't clobber
19095 TYPE_CANONICAL on an existing type.
19097 2018-02-15 Jakub Jelinek <jakub@redhat.com>
19099 PR tree-optimization/84383
19100 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
19101 dstoff nor call operand_equal_p if dstbase is NULL.
19103 PR tree-optimization/84334
19104 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
19105 also a CONSTANT_CLASS_P, punt.
19107 2018-02-14 Jim Wilson <jimw@sifive.com>
19109 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
19110 first SMALL_OPERAND check. New local min_second_step. Move assert
19111 to where locals are set. Add TARGET_RVC support.
19112 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
19114 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
19116 * doc/invoke.texi: Correct -Wformat-overflow code sample.
19118 2018-02-14 Martin Sebor <msebor@redhat.com>
19120 PR tree-optimization/83698
19121 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
19122 arrays constrain the offset range to their bounds.
19123 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
19124 (builtin_access::overlap): Avoid setting the size of overlap if it's
19126 (maybe_diag_overlap): Also consider arrays when deciding what values
19127 of offsets to include in diagnostics.
19129 2018-02-14 Martin Sebor <msebor@redhat.com>
19132 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
19133 that correspond to the kind of a declaration.
19135 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
19138 * config/pa/pa.md: Load address of PIC label using the linkage table
19139 if the label is nonlocal.
19141 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19143 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
19144 warning message if user requests -maltivec=be.
19145 * doc/invoke.texi: Document deprecation of -maltivec=be.
19147 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
19150 * config/rs6000/rs6000-c.c: Update definitions for
19151 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
19152 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
19154 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19157 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
19158 add _get_ssp intrinsics. Remove argument from
19159 __builtin_ia32_rdssp[d|q].
19160 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
19161 * config/i386/i386-builtin.def: Remove argument from
19162 __builtin_ia32_rdssp[d|q].
19163 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
19164 ix86_expand_special_args_builtin for _rdssp[d|q].
19165 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
19166 Clear register before usage.
19167 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
19168 Add documentation for new _get_ssp and _inc_ssp intrinsics.
19170 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
19172 PR tree-optimization/84357
19173 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
19174 operand 1 of an ARRAY_REF too.
19176 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
19179 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
19180 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
19182 (set_of_reg): New struct.
19183 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
19184 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
19185 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
19187 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
19188 Split into bitclr, bitset, bitinvert patterns if appropriate.
19189 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
19190 use rx_fuse_in_memory_bitop.
19191 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
19192 to named insn, correct maximum insn length.
19194 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
19197 * machmode.def: Define a complex mode for PARTIAL_INT.
19198 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
19200 * doc/rtl.texi: Document CSPImode.
19201 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
19203 (msp430_hard_regno_nregs_with_padding): Likewise.
19205 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
19208 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
19210 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19212 PR rtl-optimization/84169
19213 * combine.c (try_combine): New variable split_i2i3. Set it to true if
19214 we generated a parallel as new i3 and we split that to new i2 and i3
19215 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
19216 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
19217 those to i2, not i1. Partially rewrite this scan code.
19219 2018-02-13 Jakub Jelinek <jakub@redhat.com>
19222 * stor-layout.c (place_field): For variable length fields, adjust
19223 offset_align afterwards not just based on the field's alignment,
19224 but also on the size.
19226 PR middle-end/84309
19227 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
19228 of exps and logs in the use_exp2 case.
19230 2018-02-13 Jeff Law <law@redhat.com>
19232 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
19233 entry for "vector".
19235 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
19238 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
19242 * common.opt (gas-loc-support, gas-locview-support): New.
19243 (ginline-points, ginternal-reset-location-views): New.
19244 * doc/invoke.texi: Document them. Use @itemx where intended.
19245 (gvariable-location-views): Adjust.
19246 * target.def (reset_location_view): New.
19247 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
19248 (TARGET_RESET_LOCATION_VIEW): New.
19249 * doc/tm.texi: Rebuilt.
19250 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
19251 (dwarf2out_default_as_locview_support): New.
19252 (output_asm_line_debug_info): Use option variables.
19253 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
19254 (output_loc_list): Likewise.
19255 (add_high_low_attributes): Check option variables.
19256 Don't output entry view attribute in strict mode.
19257 (gen_inlined_subroutine_die): Check option variables.
19258 (dwarf2out_inline_entry): Likewise.
19259 (init_sections_and_labels): Likewise.
19260 (dwarf2out_early_finish): Likewise.
19261 (maybe_reset_location_view): New, from...
19262 (dwarf2out_var_location): ... here. Call it.
19263 * debug.h (dwarf2out_default_as_loc_support): Declare.
19264 (dwarf2out_default_as_locview_support): Declare.
19265 * hooks.c (hook_int_rtx_insn_0): New.
19266 * hooks.h (hook_int_rtx_insn_0): Declare.
19267 * toplev.c (process_options): Take -gas-loc-support and
19268 -gas-locview-support from dwarf2out. Enable
19269 -gvariable-location-views by default only with locview
19270 assembler support. Enable -ginternal-reset-location-views by
19271 default only if the target defines the corresponding hook.
19272 Enable -ginline-points by default if location views are
19273 enabled; force it disabled if statement frontiers are
19275 * tree-inline.c (expand_call_inline): Check option variables.
19276 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
19278 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
19280 PR tree-optimization/84321
19281 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
19282 handling. Also check whether the anti-range contains any values
19283 that satisfy the mask; switch to a VR_RANGE if not.
19285 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
19288 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
19290 2018-02-13 Martin Jambor <mjambor@suse.cz>
19293 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
19294 of call statements, also set location of a load to a temporary.
19296 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
19298 * config/rl78/rl78.c (add_vector_labels): New function.
19299 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
19300 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
19301 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
19302 which checks that no arguments are passed.
19303 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
19304 * doc/extend.texi: Documentation for the new attribute.
19306 2018-02-13 Andreas Schwab <schwab@suse.de>
19308 * config/riscv/linux.h (CPP_SPEC): Define.
19310 2018-02-13 Jakub Jelinek <jakub@redhat.com>
19313 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
19314 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
19315 OPTION_MASK_ISA_AES as first argument to def_builtin_const
19316 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
19317 instead of OPTION_MASK_ISA_PCLMUL as first argument to
19318 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
19319 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
19320 temporarily for AES and PCLMUL builtins.
19322 PR tree-optimization/84339
19323 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
19324 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
19327 PR middle-end/84309
19328 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
19329 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
19330 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
19332 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
19334 * omp-simd-clone.h: New file.
19335 * omp-simd-clone.c: Include omp-simd-clone.h.
19336 (expand_simd_clones): No longer static.
19337 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
19338 cgraph.h and omp-simd-clone.h.
19339 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
19340 (vect_recog_widen_shift_pattern): Formatting fix.
19341 (vect_pattern_recog_1): Don't check optab for calls.
19344 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
19345 operands[2] into a REG before using gen_lowpart on it.
19347 2018-02-12 Jeff Law <law@redhat.com>
19350 * config/sh/sh.c (find_barrier): Consider a sibling call
19353 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
19354 successfully back substituting a reg.
19356 2018-02-12 Richard Biener <rguenther@suse.de>
19358 PR tree-optimization/84037
19359 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
19360 parameter, move visited init to caller.
19361 (vect_slp_analyze_operations): Separate cost from validity
19362 check, initialize visited once for all instances.
19363 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
19365 * tree-vect-stmts.c (vect_model_simple_cost): Make early
19367 (vect_model_promotion_demotion_cost): Likewise.
19368 (vectorizable_bswap): Guard cost modeling with !slp_node
19369 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
19371 (vectorizable_call): Likewise.
19372 (vectorizable_conversion): Likewise.
19373 (vectorizable_assignment): Likewise.
19374 (vectorizable_shift): Likewise.
19375 (vectorizable_operation): Likewise.
19376 (vectorizable_store): Likewise.
19377 (vectorizable_load): Likewise.
19378 (vectorizable_condition): Likewise.
19379 (vectorizable_comparison): Likewise.
19381 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
19384 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
19385 (ASAN_MARK): Fix fnspec to account for return value, change pointer
19386 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
19388 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19390 PR middle-end/83665
19391 * params.def (inline-min-speedup): Increase from 8 to 15.
19392 (max-inline-insns-auto): Decrease from 40 to 30.
19393 * ipa-split.c (consider_split): Add some buffer for function to
19394 be considered inlining candidate.
19395 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
19398 2018-02-12 Richard Biener <rguenther@suse.de>
19400 PR tree-optimization/84037
19401 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
19402 matched stmts if we cannot swap the non-matched ones.
19404 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
19406 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
19407 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
19408 _mm_maskz_scalef_round_ss): New intrinsics.
19409 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
19410 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
19411 __builtin_ia32_scalefss_round): Remove.
19412 (__builtin_ia32_scalefsd_mask_round,
19413 __builtin_ia32_scalefss_mask_round): New intrinsics.
19414 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
19415 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
19416 ((match_operand:VF_128 2 "<round_nimm_predicate>"
19417 "<round_constraint>")): Changed to ...
19418 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
19419 "<round_scalar_constraint>")): ... this.
19420 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
19421 %0, %1, %2<round_op3>}"): Changed to ...
19422 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
19423 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
19424 %2<round_scalar_mask_op3>}"): ... this.
19425 * config/i386/subst.md (round_scalar_nimm_predicate): New.
19427 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
19429 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
19430 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
19431 (_mm_maskz_sqrt_round_ss): New intrinsics.
19432 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
19433 (__builtin_ia32_sqrtsd_mask_round)
19434 (__builtin_ia32_sqrtss_mask_round): New builtins.
19435 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
19436 (__builtin_ia32_sqrtss_round): Remove.
19437 (__builtin_ia32_sqrtsd_mask_round)
19438 (__builtin_ia32_sqrtss_mask_round): New builtins.
19439 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
19440 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
19441 ((match_operand:VF_128 1 "vector_operand"
19442 "xBm,<round_constraint>")): Changed to ...
19443 ((match_operand:VF_128 1 "vector_operand"
19444 "xBm,<round_scalar_constraint>")): ... this.
19445 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
19446 %0, %2, %<iptr>1<round_op3>}): Changed to ...
19447 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
19448 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
19449 %<iptr>1<round_scalar_mask_op3>}): ... this.
19450 ((set_attr "prefix" "<round_prefix>")): Changed to ...
19451 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
19453 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
19456 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
19457 Cast vec_cmpeq result to correct type.
19458 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
19459 Cast vec_cmpgt result to correct type.
19461 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
19463 * final.c (final_scan_insn_1): Renamed from...
19464 (final_scan_insn): ... this. New wrapper, to recover
19465 seen from the outermost call in recursive ones.
19466 * config/sparc/sparc.c (output_return): Drop seen from call.
19467 (output_sibcall): Likewise.
19468 * config/visium/visium.c (output_branch): Likewise.
19470 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
19472 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
19475 2018-02-10 Alan Modra <amodra@gmail.com>
19478 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
19479 Specify LR as an input.
19481 2018-02-10 Jakub Jelinek <jakub@redhat.com>
19484 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
19485 remove_member_access_dummy_vars): New functions.
19486 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
19487 lower_omp_1, execute_lower_omp): Use them.
19489 PR rtl-optimization/84308
19490 * shrink-wrap.c (spread_components): Release todo vector.
19492 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
19494 PR rtl-optimization/57193
19495 * ira-color.c (struct allocno_color_data): Add member
19496 conflict_allocno_hard_prefs.
19497 (update_conflict_allocno_hard_prefs): New.
19498 (bucket_allocno_compare_func): Add a preference based on
19499 conflict_allocno_hard_prefs.
19500 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
19501 (color_allocnos): Remove a dead code. Initiate
19502 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
19504 2018-02-09 Jakub Jelinek <jakub@redhat.com>
19507 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
19508 constraint from =wa to wa. Avoid a subreg on the output operand,
19509 instead use a pseudo and subreg it in a move.
19510 (p9_xxbrd_<mode>): Changed to ...
19511 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
19512 (p9_xxbrd_v2df): New expander.
19513 (p9_xxbrw_<mode>): Changed to ...
19514 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
19515 (p9_xxbrw_v4sf): New expander.
19517 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
19519 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
19521 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
19524 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
19525 multiply in 32-bit mode.
19526 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
19527 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
19530 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
19532 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
19533 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
19534 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
19535 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
19537 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
19540 * dwarf2out.c (is_trivial_indirect_ref): New function.
19541 (dwarf2out_late_global_decl): Do not generate a location
19542 attribute for variables that have a non-trivial DECL_VALUE_EXPR
19543 and that are not defined in the current unit.
19545 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
19547 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
19548 instead of a libcall for UNORDERED.
19550 2018-02-09 Tamar Christina <tamar.christina@arm.com>
19553 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
19554 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
19556 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19559 * config/s390/s390.c (s390_set_current_function): Invoke
19560 s390_indirect_branch_settings also if fndecl didn't change.
19562 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
19564 * config/rs6000/rs6000.md (blockage): Set length to zero.
19566 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
19568 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
19570 2018-02-09 Jakub Jelinek <jakub@redhat.com>
19573 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
19574 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
19578 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
19579 PARALLEL incoming that failed vt_get_decl_and_offset check.
19581 PR middle-end/84237
19582 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
19583 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
19585 (get_variable_section): For decls in named .bss* sections pass true as
19586 second argument to bss_initializer_p.
19588 2018-02-09 Marek Polacek <polacek@redhat.com>
19589 Jakub Jelinek <jakub@redhat.com>
19592 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
19593 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
19594 Sync some changes from cxx_fold_indirect_ref.
19596 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
19598 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
19600 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
19601 (BLOCK_INLINE_ENTRY_LABEL): New.
19602 (dwarf2out_var_location): Disregard inline entry markers.
19603 (inline_entry_data): New struct.
19604 (inline_entry_data_hasher): New hashtable type.
19605 (inline_entry_data_hasher::hash): New.
19606 (inline_entry_data_hasher::equal): New.
19607 (inline_entry_data_table): New variable.
19608 (add_high_low_attributes): Add DW_AT_entry_pc and
19609 DW_AT_GNU_entry_view attributes if a pending entry is found
19610 in inline_entry_data_table. Add old entry_pc attribute only
19611 if debug nonbinding markers are disabled.
19612 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
19613 markers are enabled.
19614 (block_within_block_p, dwarf2out_inline_entry): New.
19615 (dwarf2out_finish): Check that no entries remained in
19616 inline_entry_data_table.
19617 * final.c (reemit_insn_block_notes): Handle inline entry notes.
19618 (final_scan_insn, notice_source_line): Likewise.
19619 (rest_of_clean_state): Skip inline entry markers.
19620 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
19622 * gimple.c (gimple_build_debug_inline_entry): New.
19623 * gimple.h (enum gimple_debug_subcode): Add
19624 GIMPLE_DEBUG_INLINE_ENTRY.
19625 (gimple_build_debug_inline_entry): Declare.
19626 (gimple_debug_inline_entry_p): New.
19627 (gimple_debug_nonbind_marker_p): Adjust.
19628 * insn-notes.def (INLINE_ENTRY): New.
19629 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
19630 inline entry marker notes.
19631 (print_insn): Likewise.
19632 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
19633 (INSN_DEBUG_MARKER_KIND): Likewise.
19634 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
19635 * tree-inline.c (expand_call_inline): Build and insert
19636 debug_inline_entry stmt.
19637 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
19638 inline entry blocks early, if nonbind markers are enabled.
19639 (dump_scope_block): Dump fragment info.
19640 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
19641 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
19642 (gimple_build_debug_inline_entry): New.
19643 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
19644 Enable/disable inline entry points too.
19645 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
19646 (DEBUG_INSN): Describe inline entry markers.
19648 * common.opt (gvariable-location-views): New.
19649 (gvariable-location-views=incompat5): New.
19650 * config.in: Rebuilt.
19651 * configure: Rebuilt.
19652 * configure.ac: Test assembler for view support.
19653 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
19654 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
19655 * dwarf2out.c (var_loc_view): New typedef.
19656 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
19657 (dwarf2out_locviews_in_attribute): New.
19658 (dwarf2out_locviews_in_loclist): New.
19659 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
19660 (enum dw_line_info_opcode): Add LI_adv_address.
19661 (struct dw_line_info_table): Add view.
19662 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
19663 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
19664 (zero_view_p): New variable.
19665 (ZERO_VIEW_P): New macro.
19666 (output_asm_line_debug_info): New.
19667 (struct var_loc_node): Add view.
19668 (add_AT_view_list, AT_loc_list): New.
19669 (add_var_loc_to_decl): Add view param. Test it against last.
19670 (new_loc_list): Add view params. Record them.
19671 (AT_loc_list_ptr): Handle loc and view lists.
19672 (view_list_to_loc_list_val_node): New.
19673 (print_dw_val): Handle dw_val_class_view_list.
19674 (size_of_die): Likewise.
19675 (value_format): Likewise.
19676 (loc_list_has_views): New.
19677 (gen_llsym): Set vl_symbol too.
19678 (maybe_gen_llsym, skip_loc_list_entry): New.
19679 (dwarf2out_maybe_output_loclist_view_pair): New.
19680 (output_loc_list): Output view list or entries too.
19681 (output_view_list_offset): New.
19682 (output_die): Handle dw_val_class_view_list.
19683 (output_dwarf_version): New.
19684 (output_compilation_unit_header): Use it.
19685 (output_skeleton_debug_sections): Likewise.
19686 (output_rnglists, output_line_info): Likewise.
19687 (output_pubnames, output_aranges): Update version comments.
19688 (output_one_line_info_table): Output view numbers in asm comments.
19689 (dw_loc_list): Determine current endview, pass it to new_loc_list.
19690 Call maybe_gen_llsym.
19691 (loc_list_from_tree_1): Adjust.
19692 (add_AT_location_description): Create view list attribute if
19693 needed, check it's absent otherwise.
19694 (convert_cfa_to_fb_loc_list): Adjust.
19695 (maybe_emit_file): Call output_asm_line_debug_info for test.
19696 (dwarf2out_var_location): Reset views as needed. Precompute
19697 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
19698 attribute. Set view.
19699 (new_line_info_table): Reset next view.
19700 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
19701 (dwarf2out_source_line): Likewise. Output view resets and labels to
19702 the assembler, or select appropriate line info opcodes.
19703 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
19704 (optimize_string_length): Catch it. Adjust.
19705 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
19706 dw_val_class_view_list, and remove it if no longer needed.
19707 (hash_loc_list): Hash view numbers.
19708 (loc_list_hasher::equal): Compare them.
19709 (optimize_location_lists): Check whether a view list symbol is
19710 needed, and whether the locview attribute is present, and
19711 whether they match. Remove the locview attribute if no longer
19713 (index_location_lists): Call skip_loc_list_entry for test.
19714 (dwarf2out_finish): Call output_asm_line_debug_info for test.
19715 Use output_dwarf_version.
19716 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
19717 (struct dw_val_node): Add val_view_list.
19718 * final.c (SEEN_NEXT_VIEW): New.
19719 (set_next_view_needed): New.
19720 (clear_next_view_needed): New.
19721 (maybe_output_next_view): New.
19722 (final_start_function): Rename to...
19723 (final_start_function_1): ... this. Take pointer to FIRST,
19724 add SEEN parameter. Emit param bindings in the initial view.
19725 (final_start_function): Reintroduce SEEN-less interface.
19726 (final): Rename to...
19727 (final_1): ... this. Take SEEN parameter. Output final pending
19728 next view at the end.
19729 (final): Reintroduce seen-less interface.
19730 (final_scan_insn): Output pending next view before switching
19731 sections or ending a block. Mark the next view as needed when
19732 outputting variable locations. Notify debug backend of section
19733 changes, and of location view changes.
19734 (rest_of_handle_final): Adjust.
19735 * toplev.c (process_options): Autodetect value for debug variable
19736 location views option. Warn on incompat5 without -gdwarf-5.
19737 * doc/invoke.texi (gvariable-location-views): New.
19738 (gvariable-location-views=incompat5): New.
19739 (gno-variable-location-views): New.
19741 2018-02-08 David Malcolm <dmalcolm@redhat.com>
19743 PR tree-optimization/84136
19744 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
19745 that the result of find_edge is non-NULL.
19747 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19750 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
19751 storing integer register in SImode. Fix cost of 256 and 512
19752 byte aligned SSE register store.
19754 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19756 * config/i386/i386.c (ix86_multiplication_cost): Fix
19757 multiplication cost for TARGET_AVX512DQ.
19759 2018-02-08 Marek Polacek <polacek@redhat.com>
19761 PR tree-optimization/84238
19762 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
19765 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19767 PR tree-optimization/84265
19768 * tree-vect-stmts.c (vectorizable_store): Don't treat
19769 VMAT_CONTIGUOUS accesses as grouped.
19770 (vectorizable_load): Likewise.
19772 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19774 PR tree-optimization/81635
19775 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
19776 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
19777 (test_round_for_mask): New functions.
19778 (wide_int_cc_tests): Call test_round_for_mask.
19779 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
19780 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
19781 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
19782 range returned by get_range_info.
19784 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19787 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
19788 * symtab.c: Include builtins.h
19789 (symtab_node::output_to_lto_symbol_table_p): Move here
19790 from lto-streamer-out.c:output_symbol_p.
19791 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
19792 (output_symbol_p): Move all logic to symtab.c
19793 (produce_symtab): Update.
19795 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19797 * config/s390/s390-opts.h (enum indirect_branch): Define.
19798 * config/s390/s390-protos.h (s390_return_addr_from_memory)
19799 (s390_indirect_branch_via_thunk)
19800 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
19801 (enum s390_indirect_branch_type): Define.
19802 * config/s390/s390.c (struct s390_frame_layout, struct
19803 machine_function): Remove.
19804 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
19805 (indirect_branch_table_label_no, indirect_branch_table_name):
19807 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
19808 (enum s390_indirect_branch_option): Define.
19809 (s390_return_addr_from_memory): New function.
19810 (s390_handle_string_attribute): New function.
19811 (s390_attribute_table): Add new attribute handler.
19812 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
19813 (s390_indirect_branch_via_thunk): New function.
19814 (s390_indirect_branch_via_inline_thunk): New function.
19815 (s390_function_ok_for_sibcall): When jumping via thunk disallow
19816 sibling call optimization for non z10 compiles.
19817 (s390_emit_call): Force indirect branch target to be a single
19818 register. Add r1 clobber for non-z10 compiles.
19819 (s390_emit_epilogue): Emit return jump via return_use expander.
19820 (s390_reorg): Handle JUMP_INSNs as execute targets.
19821 (s390_option_override_internal): Perform validity checks for the
19822 new command line options.
19823 (s390_indirect_branch_attrvalue): New function.
19824 (s390_indirect_branch_settings): New function.
19825 (s390_set_current_function): Invoke s390_indirect_branch_settings.
19826 (s390_output_indirect_thunk_function): New function.
19827 (s390_code_end): Implement target hook.
19828 (s390_case_values_threshold): Implement target hook.
19829 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
19831 * config/s390/s390.h (struct s390_frame_layout)
19832 (struct machine_function): Move here from s390.c.
19833 (TARGET_INDIRECT_BRANCH_NOBP_RET)
19834 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
19835 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
19836 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
19837 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
19838 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
19839 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
19840 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
19841 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
19842 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
19843 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
19844 (mnemonic attribute): Add values which aren't recognized
19846 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
19847 pattern for branch conversion. Fix mnemonic attribute.
19848 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
19849 indirect branch via thunk if requested.
19850 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
19851 ("*indirect_jump"): Disable for branch conversion using out of
19853 ("indirect_jump_via_thunk<mode>_z10")
19854 ("indirect_jump_via_thunk<mode>")
19855 ("indirect_jump_via_inlinethunk<mode>_z10")
19856 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
19857 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
19858 ("casesi_jump_via_inlinethunk<mode>_z10")
19859 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
19860 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
19861 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
19862 ("*indirect2_jump"): Disable for branch conversion.
19863 ("casesi_jump"): Turn into expander and expand patterns for branch
19865 ("return_use"): New expander.
19866 ("*return"): Emit return via thunk and rename it to ...
19867 ("*return<mode>"): ... this one.
19868 * config/s390/s390.opt: Add new options and and enum for the
19871 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19873 * lra-constraints.c (match_reload): Unconditionally use
19874 gen_lowpart_SUBREG, rather than selecting between that
19875 and equivalent gen_rtx_SUBREG code.
19877 2018-02-08 Richard Biener <rguenther@suse.de>
19879 PR tree-optimization/84233
19880 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
19881 changed flag instead of boguously re-using phi_inserted.
19883 2018-02-08 Martin Jambor <mjambor@suse.cz>
19885 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
19886 static local variables.
19888 2018-02-08 Richard Biener <rguenther@suse.de>
19890 PR tree-optimization/84278
19891 * tree-vect-stmts.c (vectorizable_store): When looking for
19892 smaller vector types to perform grouped strided loads/stores
19893 make sure the mode is supported by the target.
19894 (vectorizable_load): Likewise.
19896 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19898 * config/aarch64/aarch64.c (aarch64_components_for_bb):
19899 Increase LDP/STP opportunities by adding adjacent callee-saves.
19901 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19903 PR rtl-optimization/84068
19904 PR rtl-optimization/83459
19905 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
19907 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
19909 PR tree-optimization/84224
19910 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
19911 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
19912 non-zero arguments.
19914 2018-02-07 Iain Sandoe <iain@codesourcery.com>
19917 * config/rs6000/altivec.md (*restore_world): Remove LR use.
19918 * config/rs6000/predicates.md (restore_world_operation): Adjust op
19919 count, remove one USE.
19921 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19923 * doc/install.texi (Configuration): Document the
19924 --with-long-double-format={ibm,ieee} PowerPC configuration
19928 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
19929 Convert from define_expand to be define_insn_and_split. Rework
19930 float/double/_Float128 conversions to QI/HI/SImode to work with
19931 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
19932 conversions to QI/HImode types did a store and then a load to
19933 truncate the value. For conversions to VSX registers, don't split
19934 the insn, instead emit the code directly. Use the code iterator
19935 any_fix to combine signed and unsigned conversions.
19936 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
19937 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
19938 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19939 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
19940 (fix_<mode>di2_hw): Likewise.
19941 (fixuns_<mode>di2_hw): Likewise.
19942 (fix_<mode>si2_hw): Likewise.
19943 (fixuns_<mode>si2_hw): Likewise.
19944 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
19945 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19946 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
19947 fix<uns>_trunc<SFDF:mode>si2_p8.
19948 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
19949 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19950 (fix<uns>_<mode>_mem): Likewise.
19951 (fctiw<u>z_<mode>_mem): Likewise.
19952 (fix<uns>_<mode>_mem): Likewise.
19953 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
19954 the register allocator from doing a direct move to the GPRs to do
19955 a store, and instead use the ISA 3.0 store byte/half-word from
19956 vector register instruction. For IEEE 128-bit floating point,
19957 also optimize stores of 32-bit ints.
19958 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
19960 2018-02-07 Alan Hayward <alan.hayward@arm.com>
19962 * genextract.c (push_pathstr_operand): New function to support
19964 (walk_rtx): Call push_pathstr_operand.
19965 (print_path): Support [a-zA-Z].
19967 2018-02-07 Richard Biener <rguenther@suse.de>
19969 PR tree-optimization/84037
19970 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
19971 (cse_and_gimplify_to_preheader): Declare.
19972 (vect_get_place_in_interleaving_chain): Likewise.
19973 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
19975 (_loop_vec_info::~_loop_vec_info): Delete it.
19976 (cse_and_gimplify_to_preheader): New function.
19977 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
19978 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
19979 (vectorizable_load): Likewise. For grouped stores always base
19980 the IV on the first element.
19981 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
19982 condition before gimplifying.
19984 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19986 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
19987 *DIV_EXPR and *MOD_EXPR.
19989 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
19992 * config/i386/i386.c (ix86_option_override_internal): Mask out
19993 the CF_SET bit when checking -fcf-protection.
19995 2018-02-07 Tom de Vries <tom@codesourcery.com>
19998 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
20001 2018-02-07 Richard Biener <rguenther@suse.de>
20003 PR tree-optimization/84204
20004 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
20007 PR tree-optimization/84205
20008 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
20009 special-case isl_ast_op_zdiv_r.
20011 PR tree-optimization/84223
20012 * graphite-scop-detection.c (gather_bbs::before_dom_children):
20013 Only add conditions from within the region.
20014 (gather_bbs::after_dom_children): Adjust.
20016 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
20019 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
20020 * config/avr/avr.md: Only post-reload split REG-REG moves if
20021 either register is GENERAL_REG_P.
20023 2018-02-07 Jakub Jelinek <jakub@redhat.com>
20025 PR tree-optimization/84235
20026 * tree-ssa-scopedtables.c
20027 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
20028 if the subtraction is performed in floating point type where NaNs are
20029 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
20030 build 1. Formatting fix.
20032 2018-02-06 Jakub Jelinek <jakub@redhat.com>
20035 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
20036 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
20037 and skip it regardless of bb boundaries. Use CALL_P macro,
20038 don't test INSN_P (insn) together with CALL_P or JUMP_P check
20039 unnecessarily, formatting fix.
20041 2018-02-06 Michael Collison <michael.collison@arm.com>
20043 * config/arm/thumb2.md:
20044 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
20045 (*thumb_mov_notscc): Ditto.
20047 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20050 * config/rs6000/rs6000.md (su code attribute): Use "u" for
20051 unsigned_fix, not "s".
20053 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20055 * configure.ac (gcc_fn_eh_frame_ro): New function.
20056 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
20057 correct .eh_frame permissions.
20058 * configure: Regenerate.
20060 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
20062 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
20063 irrelevant options.
20065 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20067 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20068 Display warning message for -mno-speculate-indirect-jumps.
20070 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
20072 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
20074 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
20076 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
20078 PR tree-optimization/84225
20079 * tree-eh.c (find_trapping_overflow): Only call
20080 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
20082 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
20085 * config/i386/i386.c: Reimplement the check of possible options
20086 -mibt/-mshstk conbination. Change error messages.
20087 * doc/invoke.texi: Fix a typo: remove extra '='.
20089 2018-02-06 Marek Polacek <polacek@redhat.com>
20091 PR tree-optimization/84228
20092 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
20094 2018-02-06 Tamar Christina <tamar.christina@arm.com>
20097 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
20098 emitted arch directives.
20099 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
20100 __ARM_FEATURE_COPROC before changing architectures.
20102 2018-02-06 Richard Biener <rguenther@suse.de>
20104 * config/i386/i386.c (print_reg): Fix typo.
20105 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
20107 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
20109 * configure: Regenerate.
20111 2018-02-05 Martin Sebor <msebor@redhat.com>
20113 PR tree-optimization/83369
20114 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
20117 2018-02-05 Martin Liska <mliska@suse.cz>
20119 * doc/invoke.texi: Cherry-pick upstream r323995.
20121 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
20123 * ira.c (ira_init_register_move_cost): Adjust comment.
20125 2018-02-05 Martin Liska <mliska@suse.cz>
20127 PR gcov-profile/84137
20128 * doc/gcov.texi: Fix typo in documentation.
20130 2018-02-05 Martin Liska <mliska@suse.cz>
20132 PR gcov-profile/83879
20133 * doc/gcov.texi: Document necessity of --dynamic-list-data when
20134 using dlopen functionality.
20136 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
20138 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
20139 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
20140 _mm_maskz_range_ss, _mm_mask_range_round_ss,
20141 _mm_maskz_range_round_ss): New intrinsics.
20142 (__builtin_ia32_rangesd128_round)
20143 (__builtin_ia32_rangess128_round): Remove.
20144 (__builtin_ia32_rangesd128_mask_round,
20145 __builtin_ia32_rangess128_mask_round): New builtins.
20146 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
20147 __builtin_ia32_rangess128_round): Remove.
20148 (__builtin_ia32_rangesd128_mask_round,
20149 __builtin_ia32_rangess128_mask_round): New builtins.
20150 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
20151 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
20152 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
20153 "<round_saeonly_constraint>")): Changed to ...
20154 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
20155 "<round_saeonly_scalar_constraint>")): ... this.
20156 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20157 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
20158 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
20159 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
20160 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
20162 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
20164 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
20166 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
20167 Remove all values except native, 8540 and 8548.
20169 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
20171 * config/i386/i386.c (ix86_output_function_return): Pass
20172 INVALID_REGNUM, instead of -1, as invalid register number to
20173 indirect_thunk_name and output_indirect_thunk.
20175 2018-02-02 Julia Koval <julia.koval@intel.com>
20177 * config.gcc: Add -march=icelake.
20178 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
20179 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
20180 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
20181 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
20182 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
20183 (processor_target_table): Add icelake.
20184 (ix86_option_override_internal): Handle new PTAs.
20185 (get_builtin_code_for_version): Handle icelake.
20186 (M_INTEL_COREI7_ICELAKE): New.
20187 (fold_builtin_cpu): Handle icelake.
20188 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
20189 * doc/invoke.texi: Add -march=icelake.
20191 2018-02-02 Julia Koval <julia.koval@intel.com>
20193 * config/i386/i386.c (ix86_option_override_internal): Change flags type
20194 to wide_int_bitmask.
20195 * wide-int-bitmask.h: New.
20197 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
20200 * config/i386/i386.md: Replace Pmode with word_mode in
20201 builtin_setjmp_setup and builtin_longjmp to support x32.
20203 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
20207 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
20209 (rs6000_supported_cpu_names): New static variable.
20210 (linux_cpu_translation_table): Likewise.
20211 (elf_platform) <cpu>: Define new static variable and use it.
20212 Translate kernel AT_PLATFORM name to canonical name if needed.
20213 Error if platform name is unknown.
20215 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
20218 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
20220 2018-02-01 Jeff Law <law@redhat.com>
20223 * config/i386/i386.c (release_scratch_register_on_entry): Add new
20224 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
20225 the scratch if RELEASE_VIA_POP is false.
20226 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
20227 If we have to save a temporary register, decrement SIZE appropriately.
20228 Pass new arguments to release_scratch_register_on_entry.
20229 (ix86_adjust_stack_and_probe): Likewise.
20230 (ix86_emit_probe_stack_range): Pass new arguments to
20231 release_scratch_register_on_entry.
20233 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
20235 PR rtl-optimization/84157
20236 * combine.c (change_zero_ext): Use REG_P predicate in
20237 front of HARD_REGISTER_P predicate.
20239 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
20241 * config/avr/avr.c (avr_option_override): Move disabling of
20242 -fdelete-null-pointer-checks to...
20243 * common/config/avr/avr-common.c (avr_option_optimization_table):
20246 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20248 PR tree-optimization/81635
20249 * tree-data-ref.c (split_constant_offset_1): For types that
20250 wrap on overflow, try to use range info to prove that wrapping
20253 2018-02-01 Renlin Li <renlin.li@arm.com>
20256 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
20257 TAILCALL_ADDR_REGS.
20258 (aarch64_register_move_cost): Likewise.
20259 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
20260 TAILCALL_ADDR_REGS.
20261 (REG_CLASS_NAMES): Likewise.
20262 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
20263 TAILCALL_ADDR_REGS. Remove IP registers.
20264 * config/aarch64/aarch64.md (Ucs): Update register constraint.
20266 2018-02-01 Richard Biener <rguenther@suse.de>
20268 * domwalk.h (dom_walker::dom_walker): Add additional constructor
20269 for specifying RPO order and allow NULL for that.
20270 * domwalk.c (dom_walker::dom_walker): Likewise.
20271 (dom_walker::walk): Handle NULL RPO order.
20272 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
20274 (rewrite_update_dom_walker): Likewise.
20275 (mark_def_dom_walker): Likewise.
20277 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20279 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
20280 (aarch64_maybe_expand_sve_subreg_move): Declare.
20281 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
20282 * config/aarch64/predicates.md (aarch64_any_register_operand): New
20284 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
20285 that are semantically a reverse operation.
20286 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
20287 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
20288 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
20290 (aarch64_can_change_mode_class): For big-endian, forbid changes
20291 between two SVE modes if they have different element sizes.
20293 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20295 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
20296 the TImode handling for big-endian targets.
20298 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20300 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
20301 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
20303 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
20304 Remove BSWAP handing for big-endian targets and use the form of
20305 LD1RQ appropariate for the mode.
20307 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20309 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
20310 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
20311 duplicated element.
20313 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20316 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
20317 check for operands that need to go through aarch64_sve_reload_be.
20319 2018-02-01 Jakub Jelinek <jakub@redhat.com>
20321 PR tree-optimization/81661
20322 PR tree-optimization/84117
20323 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
20324 * tree-eh.c: Include gimplify.h.
20325 (find_trapping_overflow, replace_trapping_overflow,
20326 rewrite_to_non_trapping_overflow): New functions.
20327 * tree-vect-loop.c: Include tree-eh.h.
20328 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
20329 * tree-data-ref.c: Include tree-eh.h.
20330 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
20332 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
20334 PR rtl-optimization/84123
20335 * combine.c (change_zero_ext): Check if hard register satisfies
20336 can_change_dest_mode before calling gen_lowpart_SUBREG.
20338 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
20341 * ira.c (ira_init_register_move_cost): Remove assert.
20343 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20345 PR rtl-optimization/84071
20346 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
20347 * doc/tm.texi: Regenerate.
20349 2018-01-31 Richard Biener <rguenther@suse.de>
20351 PR tree-optimization/84132
20352 * tree-data-ref.c (analyze_miv_subscript): Properly
20353 check whether evolution_function_is_affine_multivariate_p
20354 before calling gcd_of_steps_may_divide_p.
20356 2018-01-31 Julia Koval <julia.koval@intel.com>
20359 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
20360 * config/i386/i386.md (rdpid_rex64) New.
20361 (rdpid): Make 32bit only.
20363 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
20366 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
20367 an IDENTIFIER_NODE for FUNCTION_TYPE's.
20369 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20372 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
20374 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
20376 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20378 PR rtl-optimization/84071
20379 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
20380 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
20382 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
20384 * config/arc/arc.c (arc_handle_aux_attribute): New function.
20385 (arc_attribute_table): Add 'aux' attribute.
20386 (arc_in_small_data_p): Consider aux like variables.
20387 (arc_is_aux_reg_p): New function.
20388 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
20389 (arc_get_aux_arg): New function.
20390 (prepare_move_operands): Handle aux-register access.
20391 (arc_handle_aux_attribute): New function.
20392 * doc/extend.texi (ARC Variable attributes): Add subsection.
20394 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
20396 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
20397 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
20398 (arc_attribute_table): Add 'uncached' attribute.
20399 (arc_print_operand): Print '.di' flag for uncached memory
20401 (arc_in_small_data_p): Do not consider for small data the uncached
20403 (arc_is_uncached_mem_p): New function.
20404 * config/arc/predicates.md (compact_store_memory_operand): Check
20405 for uncached memory accesses.
20406 (nonvol_nonimm_operand): Likewise.
20407 * doc/extend.texi (ARC Type Attribute): New subsection.
20409 2018-01-31 Jakub Jelinek <jakub@redhat.com>
20412 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
20413 falign-loops=): Add Optimization flag.
20415 2018-01-30 Jeff Law <law@redhat.com>
20418 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
20419 INT_REGISTERS_SAVED. Check it prior to calling
20420 get_scratch_register_on_entry.
20421 (ix86_adjust_stack_and_probe): Similarly.
20422 (ix86_emit_probe_stack_range): Similarly.
20423 (ix86_expand_prologue): Corresponding changes.
20425 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20428 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
20429 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
20431 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
20434 * lra-constraints.c (curr_insn_transform): Process AND in the
20437 2018-01-30 Jakub Jelinek <jakub@redhat.com>
20439 PR rtl-optimization/83986
20440 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
20441 dependence against last_pending_memory_flush in addition to
20442 pending_jump_insns.
20444 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
20446 PR tree-optimization/81611
20447 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
20450 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20453 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
20456 2018-01-30 Richard Biener <rguenther@suse.de>
20457 Jakub Jelinek <jakub@redhat.com>
20459 PR tree-optimization/84111
20460 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
20461 inner loops added during recursion, as they don't have up-to-date
20464 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
20467 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
20468 (can_inline_edge_by_limits_p): ... here.
20469 (can_early_inline_edge_p, check_callers,
20470 update_caller_keys, update_callee_keys, recursive_inlining,
20471 add_new_edges_to_heap, speculation_useful_p,
20472 inline_small_functions,
20473 inline_small_functions, flatten_function,
20474 inline_to_all_callers_1): Update.
20476 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
20478 * profile-count.c (profile_count::combine_with_ipa_count): Handle
20481 2018-01-30 Richard Biener <rguenther@suse.de>
20483 PR tree-optimization/83008
20484 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
20485 invariant and constant vector uses in stmts when they need
20486 more than one stmt.
20488 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20491 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
20492 * configure: Regenerate.
20494 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20496 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
20498 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
20499 Use gen_rtx_REG rather than gen_lowpart.
20501 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20503 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
20504 rather than 0 when creating partial subregs.
20506 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20508 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
20511 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20514 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
20515 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
20516 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
20517 flags. This restores the settings used before the 2017-07-24.
20518 Turning off pre increment/decrement/modify allows IVOPTS to
20519 optimize DF/SF loops where the index is an int.
20521 2018-01-29 Richard Biener <rguenther@suse.de>
20522 Kelvin Nilsen <kelvin@gcc.gnu.org>
20525 * tree-vect-stmts.c (vectorizable_call): Don't call
20526 targetm.vectorize_builtin_md_vectorized_function if callee is
20529 2018-01-22 Carl Love <cel@us.ibm.com>
20531 * doc/extend.tex: Fix typo in second arg in
20532 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
20534 2018-01-29 Richard Biener <rguenther@suse.de>
20536 PR tree-optimization/84086
20537 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
20538 (flush_ssaname_freelist): When SSA names were released reset
20539 the SCEV hash table.
20541 2018-01-29 Richard Biener <rguenther@suse.de>
20543 PR tree-optimization/84057
20544 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
20545 removed paths when removing edges.
20547 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
20549 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
20550 -mfunction-return=@var{choice}.
20552 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
20554 PR diagnostic/84034
20555 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
20556 Handle CR like TAB.
20557 (layout::print_source_line): Likewise.
20558 (test_get_line_width_without_trailing_whitespace): Add test cases.
20560 2018-01-27 Jakub Jelinek <jakub@redhat.com>
20562 PR middle-end/84040
20563 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
20566 2018-01-26 Jim Wilson <jimw@sifive.com>
20568 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
20570 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
20573 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20575 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
20576 and CMP + SUB-immediate -> SUBS.
20578 2018-01-26 Martin Sebor <msebor@redhat.com>
20580 PR tree-optimization/83896
20581 * tree-ssa-strlen.c (get_string_len): Rename...
20582 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
20583 Avoid assuming length is constant.
20584 (handle_char_store): Use HOST_WIDE_INT for string length.
20586 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
20589 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
20590 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
20592 2018-01-26 Richard Biener <rguenther@suse.de>
20594 PR rtl-optimization/84003
20595 * dse.c (record_store): Only record redundant stores when
20596 the earlier store aliases at least all accesses the later one does.
20598 2018-01-26 Jakub Jelinek <jakub@redhat.com>
20600 PR rtl-optimization/83985
20601 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
20602 REG_CFA_RESTORE insns.
20603 (delete_unmarked_insns): Don't ignore separate shrink wrapping
20604 REG_CFA_RESTORE insns here.
20607 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
20608 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
20610 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20612 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
20613 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
20614 (arc_init): Likewise.
20615 (arc_override_options): Likewise.
20616 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
20618 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
20620 * config/arc/arc.h (TARGET_DBNZ): Define.
20621 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
20622 properly set the tune attribute.
20623 (dbnz): Use TARGET_DBNZ guard.
20624 * config/arc/arc.opt (mtune): Add core3 option.
20626 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20628 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
20629 recognize new pic like addresses.
20630 (arc_delegitimize_address): Clean up.
20632 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20634 * config/arc/arc-arches.def: Option mrf16 valid for all
20636 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
20637 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
20638 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
20639 * config/arc/arc-tables.opt: Regenerate.
20640 * config/arc/arc.c (arc_conditional_register_usage): Handle
20641 reduced register file case.
20642 (arc_file_start): Set must have build attributes.
20643 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
20644 mrf16 option value.
20645 * config/arc/arc.opt (mrf16): Add new option.
20646 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
20647 * config/arc/genmultilib.awk: Handle new mrf16 option.
20648 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
20649 * config/arc/t-multilib: Regenerate.
20650 * doc/invoke.texi (ARC Options): Document mrf16 option.
20652 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20654 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
20655 * config/arc/arc.c (arc_handle_secure_attribute): New function.
20656 (arc_attribute_table): Add 'secure_call' attribute.
20657 (arc_print_operand): Print secure call operand.
20658 (arc_function_ok_for_sibcall): Don't optimize tail calls when
20660 (arc_is_secure_call_p): New function. * config/arc/arc.md
20661 (call_i): Add support for sjli instruction.
20662 (call_value_i): Likewise.
20663 * config/arc/constraints.md (Csc): New constraint.
20665 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20666 John Eric Martin <John.Martin@emmicro-us.com>
20668 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
20669 * config/arc/arc.c (_arc_jli_section): New struct.
20670 (arc_jli_section): New type.
20671 (rc_jli_sections): New static variable.
20672 (arc_handle_jli_attribute): New function.
20673 (arc_attribute_table): Add jli_always and jli_fixed attribute.
20674 (arc_file_end): New function.
20675 (TARGET_ASM_FILE_END): Define.
20676 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
20677 (arc_add_jli_section): New function.
20678 (jli_call_scan): Likewise.
20679 (arc_reorg): Call jli_call_scan.
20680 (arc_output_addsi): Remove 'S' from printing asm operand.
20681 (arc_is_jli_call_p): New function.
20682 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
20684 (movhi_insn): Likewise.
20685 (movsi_insn): Likewise.
20686 (movsi_set_cc_insn): Likewise.
20687 (loadqi_update): Likewise.
20688 (load_zeroextendqisi_update): Likewise.
20689 (load_signextendqisi_update): Likewise.
20690 (loadhi_update): Likewise.
20691 (load_zeroextendhisi_update): Likewise.
20692 (load_signextendhisi_update): Likewise.
20693 (loadsi_update): Likewise.
20694 (loadsf_update): Likewise.
20695 (movsicc_insn): Likewise.
20696 (bset_insn): Likewise.
20697 (bxor_insn): Likewise.
20698 (bclr_insn): Likewise.
20699 (bmsk_insn): Likewise.
20700 (bicsi3_insn): Likewise.
20701 (cmpsi_cc_c_insn): Likewise.
20702 (movsi_ne): Likewise.
20703 (movsi_cond_exec): Likewise.
20704 (clrsbsi2): Likewise.
20705 (norm_f): Likewise.
20715 (call_i): Remove 'S' asm letter, add jli instruction.
20716 (call_value_i): Likewise.
20717 * config/arc/arc.op (mjli-always): New option.
20718 * config/arc/constraints.md (Cji): New constraint.
20719 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
20721 (subsf3_fpx): Likewise.
20722 (mulsf3_fpx): Likewise.
20723 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
20725 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
20726 function attrbutes.
20727 * doc/invoke.texi (ARC): Document mjli-always option.
20729 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
20731 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
20732 avoid addition with 0 and use incw and decw where possible.
20734 2018-01-26 Richard Biener <rguenther@suse.de>
20736 PR tree-optimization/81082
20737 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
20738 association if it requires casting to unsigned.
20739 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
20740 from fold_plusminus_mult_expr to catch important cases late when
20741 range info is available.
20743 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20745 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
20746 * configure.ac (hidden_linkonce): New test.
20747 * configure: Regenerate.
20748 * config.in: Regenerate.
20750 2018-01-26 Julia Koval <julia.koval@intel.com>
20752 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
20753 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
20754 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
20755 _mm_mask_bitshuffle_epi64_mask): Fix type.
20756 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
20757 USI_FTYPE_V4DI_V4DI_USI): Remove.
20758 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
20759 __builtin_ia32_vpshufbitqmb256_mask,
20760 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
20761 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
20762 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
20764 2018-01-26 Alan Modra <amodra@gmail.com>
20767 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
20768 UNSPEC_VBPERMQ. Sort other unspecs.
20770 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
20772 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
20774 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
20776 PR middle-end/83055
20777 * predict.c (drop_profile): Do not push/pop cfun; update also
20779 (handle_missing_profiles): Fix logic looking for zero profiles.
20781 2018-01-25 Jakub Jelinek <jakub@redhat.com>
20783 PR middle-end/83977
20784 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
20785 on functions with #pragma omp declare simd or functions with simd
20787 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
20788 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
20789 Remove trailing \n from warning_at calls.
20791 2018-01-25 Tom de Vries <tom@codesourcery.com>
20794 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20795 for neutered workers.
20797 2018-01-24 Joseph Myers <joseph@codesourcery.com>
20800 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
20801 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
20803 2018-01-24 Jeff Law <law@redhat.com>
20806 * i386.c (get_probe_interval): Move to earlier point.
20807 (ix86_compute_frame_layout): If -fstack-clash-protection and
20808 the frame is larger than the probe interval, then use pushes
20809 to save registers rather than reg->mem moves.
20810 (ix86_expand_prologue): Remove conditional for int_registers_saved
20813 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
20816 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
20817 min/max for never referenced object.
20819 2018-01-24 Jakub Jelinek <jakub@redhat.com>
20821 PR middle-end/83977
20822 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
20824 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
20825 attributes from DECL_ATTRIBUTES (decl) without affecting
20826 DECL_ATTRIBUTES (current_function_decl).
20827 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
20828 functions with non-NULL DECL_ABSTRACT_ORIGIN.
20830 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
20832 PR tree-optimization/83979
20833 * fold-const.c (fold_comparison): Use constant_boolean_node
20834 instead of boolean_{true,false}_node.
20836 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
20838 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
20841 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20843 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20844 Simplify the clause that sets the length attribute.
20845 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20846 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
20847 clause that sets the length attribute.
20848 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20850 2018-01-24 Tom de Vries <tom@codesourcery.com>
20853 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
20854 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
20855 Add strict parameter.
20856 (prevent_branch_around_nothing): Insert dummy insn between branch to
20857 label and label with no ptx insn inbetween.
20858 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
20860 2018-01-24 Tom de Vries <tom@codesourcery.com>
20863 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20864 for neutered threads in warp.
20865 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
20867 2018-01-24 Richard Biener <rguenther@suse.de>
20869 PR tree-optimization/83176
20870 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
20873 2018-01-24 Richard Biener <rguenther@suse.de>
20875 PR tree-optimization/82819
20876 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
20877 code generating pluses that are no-ops in the target precision.
20879 2018-01-24 Richard Biener <rguenther@suse.de>
20881 PR middle-end/84000
20882 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
20884 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20886 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
20887 to merge probabilities.
20888 * predict.c (probably_never_executed): Also mark as cold functions
20889 with global 0 profile and guessed local profile.
20890 * profile-count.c (profile_probability::combine_with_count): New
20892 * profile-count.h (profile_probability::operator*,
20893 profile_probability::operator*=, profile_probability::operator/,
20894 profile_probability::operator/=): Reduce precision to adjusted
20895 and set value to guessed on contradictory divisions.
20896 (profile_probability::combine_with_freq): Remove.
20897 (profile_probability::combine_wiht_count): Declare.
20898 (profile_count::force_nonzero):: Set to adjusted.
20899 (profile_count::probability_in):: Set quality to adjusted.
20900 * tree-ssa-tail-merge.c (replace_block_by): Use
20901 combine_with_count.
20903 2018-01-23 Andrew Waterman <andrew@sifive.com>
20904 Jim Wilson <jimw@sifive.com>
20906 * config/riscv/riscv.c (riscv_stack_boundary): New.
20907 (riscv_option_override): Set riscv_stack_boundary. Handle
20908 riscv_preferred_stack_boundary_arg.
20909 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
20910 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
20911 (STACK_BOUNDARY): Set to riscv_stack_boundary.
20912 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
20913 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
20914 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
20916 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
20919 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
20920 of struct ix86_frame.
20921 (ix86_expand_epilogue): Likewise. Add a local variable for
20922 the reg_save_offset field in struct ix86_frame.
20924 2018-01-23 Bin Cheng <bin.cheng@arm.com>
20926 PR tree-optimization/82604
20927 * tree-loop-distribution.c (enum partition_kind): New enum item
20928 PKIND_PARTIAL_MEMSET.
20929 (partition_builtin_p): Support above new enum item.
20930 (generate_code_for_partition): Ditto.
20931 (compute_access_range): Differentiate cases that equality can be
20932 proven at all loops, the innermost loops or no loops.
20933 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
20934 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
20935 (finalize_partitions, distribute_loop): Don't fuse partition of
20936 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
20937 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
20938 parloop is enabled.
20940 2018-01-23 Martin Liska <mliska@suse.cz>
20942 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
20943 order to ignore the predictor.
20944 (PRED_POLYMORPHIC_CALL): Likewise.
20945 (PRED_RECURSIVE_CALL): Likewise.
20947 2018-01-23 Martin Liska <mliska@suse.cz>
20949 * tree-profile.c (tree_profiling): Print function header to
20950 aware reader which function we are working on.
20951 * value-prof.c (gimple_find_values_to_profile): Do not print
20952 not interesting value histograms.
20954 2018-01-23 Martin Liska <mliska@suse.cz>
20956 * profile-count.h (enum profile_quality): Add
20957 profile_uninitialized as the first value. Do not number values
20958 as they are zero based.
20959 (profile_count::verify): Update sanity check.
20960 (profile_probability::verify): Likewise.
20962 2018-01-23 Nathan Sidwell <nathan@acm.org>
20964 * doc/invoke.texi (ffor-scope): Deprecate.
20966 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20968 PR tree-optimization/83510
20969 * domwalk.c (set_all_edges_as_executable): New function.
20970 (dom_walker::dom_walker): Convert bool param
20971 "skip_unreachable_blocks" to enum reachability. Move setup of
20972 edge flags to set_all_edges_as_executable and only do it when
20973 reachability is REACHABLE_BLOCKS.
20974 * domwalk.h (enum dom_walker::reachability): New enum.
20975 (dom_walker::dom_walker): Convert bool param
20976 "skip_unreachable_blocks" to enum reachability.
20977 (set_all_edges_as_executable): New decl.
20978 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
20979 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
20981 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
20982 but converting true to REACHABLE_BLOCKS.
20983 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
20985 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
20986 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
20987 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
20989 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
20990 if check_all_array_refs will be called.
20992 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20994 * tree.c (selftest::test_location_wrappers): Add more test
20997 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20999 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
21000 (selftest::test_bit_in_range): Likewise.
21002 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
21005 * doc/sourcebuild.texi (vect_float): Say that the selector
21006 only describes the situation when -funsafe-math-optimizations is on.
21007 (vect_float_strict): Document.
21009 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
21011 PR tree-optimization/83965
21012 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
21013 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
21014 instead of checking only for a reduction.
21015 (vect_recog_widen_sum_pattern): Likewise.
21017 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
21019 * predict.c (probably_never_executed): Only use precise profile info.
21020 (compute_function_frequency): Skip after inlining hack since we now
21021 have quality checking.
21023 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
21025 * profile-count.h (profile_probability::very_unlikely,
21026 profile_probability::unlikely, profile_probability::even): Set
21027 precision to guessed.
21029 2018-01-23 Richard Biener <rguenther@suse.de>
21031 PR tree-optimization/83963
21032 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21033 Properly terminate dominator walk when crossing the exit edge not
21034 when visiting its source block.
21036 2018-01-23 Jakub Jelinek <jakub@redhat.com>
21039 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
21040 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
21042 2018-01-22 Jakub Jelinek <jakub@redhat.com>
21044 PR tree-optimization/83957
21045 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
21046 semicolon after for body surrounded by braces.
21048 PR tree-optimization/83081
21049 * profile-count.h (profile_probability::split): New method.
21050 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
21051 Use profile_probability::split.
21052 (do_compare_rtx_and_jump): Fix adjustment of probabilities
21053 when splitting a single conditional jump into 2.
21055 2018-01-22 David Malcolm <dmalcolm@redhat.com>
21057 PR tree-optimization/69452
21058 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
21061 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21063 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
21064 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
21065 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
21067 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21069 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
21071 * config/rl78/rl78.md (movdi): New define_expand.
21072 * config/rl78/rl78.c (rl78_split_movdi): New function.
21074 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21077 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
21079 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
21080 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
21081 128-bit to produce an UNSPEC move to get the double word with the
21082 signbit and then a shift directly to do signbit.
21083 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
21084 implementation with a new version that just does either a direct
21085 move or a regular move. Move memory interface to separate insns.
21086 Move insns so they are next to the expander.
21087 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
21088 with signbit move. Split big and little endian case.
21089 (signbit<mode>2_dm_mem_le): Likewise.
21090 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
21091 (signbit<mode>2_dm2): Likewise.
21093 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21095 * config/rl78/rl78.md (anddi3): New define_expand.
21097 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21099 * config/rl78/rl78.md (umindi3): New define_expand.
21101 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21103 * config/rl78/rl78.md (smindi3): New define_expand.
21105 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21107 * config/rl78/rl78.md (smaxdi3): New define_expand.
21109 2018-01-22 Carl Love <cel@us.ibm.com>
21111 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
21112 LVX_V1TI): Add macro expansion.
21113 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
21114 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
21115 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
21116 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
21117 Change check to determine if the instruction is a byte reversing
21118 entry. Fix typo in comment.
21119 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
21120 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
21121 Add def_builtin calls for new builtins.
21122 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
21123 Add define_insn expansion.
21125 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21127 * config/rl78/rl78.md (umaxdi3): New define_expand.
21129 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21131 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
21132 for non-QImode registers.
21134 2018-01-22 Richard Biener <rguenther@suse.de>
21136 PR tree-optimization/83963
21137 * graphite-scop-detection.c (scop_detection::get_sese): Delay
21138 including the loop exit block.
21139 (scop_detection::merge_sese): Likewise.
21140 (scop_detection::add_scop): Do it here instead.
21142 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21144 * doc/sourcebuild.texi (arm_softfloat): Document.
21146 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
21149 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
21150 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
21151 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
21153 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21154 David Edelsohn <dje.gcc@gmail.com>
21157 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
21158 Change "crset eq" to "crset 2".
21159 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
21160 (*call_indirect_aix<mode>_nospec): Likewise.
21161 (*call_value_indirect_aix<mode>_nospec): Likewise.
21162 (*call_indirect_elfv2<mode>_nospec): Likewise.
21163 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
21164 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
21165 change assembly output from . to $.
21166 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
21167 (indirect_jump<mode>_nospec): Change assembly output from . to $.
21168 (*tablejump<mode>_internal1_nospec): Likewise.
21170 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
21173 * config/sh/sh_optimize_sett_clrt.cc:
21174 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
21176 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
21178 PR tree-optimization/83940
21179 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
21180 offset_dt to vect_constant_def rather than vect_unknown_def_type.
21181 (vect_check_load_store_mask): Add a mask_dt_out parameter and
21182 use it to pass back the definition type.
21183 (vect_check_store_rhs): Likewise rhs_dt_out.
21184 (vect_build_gather_load_calls): Add a mask_dt argument and use
21185 it instead of a call to vect_is_simple_use.
21186 (vectorizable_store): Update calls to vect_check_load_store_mask
21187 and vect_check_store_rhs. Use the dt returned by the latter instead
21188 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
21189 instead of calls to vect_is_simple_use. Pass the scalar rather
21190 than the vector operand to vect_is_simple_use when handling
21191 second and subsequent copies of an rhs value.
21192 (vectorizable_load): Update calls to vect_check_load_store_mask
21193 and vect_build_gather_load_calls. Use the cached mask_dt and
21194 gs_info.offset_dt instead of calls to vect_is_simple_use.
21196 2018-01-20 Jakub Jelinek <jakub@redhat.com>
21198 PR middle-end/83945
21199 * tree-emutls.c: Include gimplify.h.
21200 (lower_emutls_2): New function.
21201 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
21202 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
21203 it before further processing.
21206 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
21207 UINTVAL (trueop1) instead of INTVAL (op1).
21209 2018-01-19 Jakub Jelinek <jakub@redhat.com>
21213 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
21214 INCOMING_FRAME_SP_OFFSET if not defined.
21215 (scan_trace): Add ENTRY argument. If true and
21216 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
21217 emit a note to adjust the CFA offset.
21218 (create_cfi_notes): Adjust scan_trace callers.
21219 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
21220 INCOMING_FRAME_SP_OFFSET in the CIE.
21221 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
21222 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
21224 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
21225 * doc/tm.texi: Regenerated.
21227 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21229 PR rtl-optimization/83147
21230 * lra-constraints.c (remove_inheritance_pseudos): Use
21231 lra_substitute_pseudo_within_insn.
21233 2018-01-19 Tom de Vries <tom@codesourcery.com>
21234 Cesar Philippidis <cesar@codesourcery.com>
21237 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
21239 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
21242 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
21243 spaces for function labels.
21245 2018-01-19 Martin Liska <mliska@suse.cz>
21247 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
21248 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
21249 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
21250 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
21251 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
21252 (PRED_CONST_RETURN): Change from 69 to 65.
21253 (PRED_NULL_RETURN): Change from 91 to 71.
21254 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
21255 (PRED_LOOP_GUARD): Change from 66 to 73.
21257 2018-01-19 Martin Liska <mliska@suse.cz>
21259 * predict.c (predict_insn_def): Add new assert.
21260 (struct branch_predictor): Change type to signed integer.
21261 (test_prediction_value_range): Amend test to cover
21262 PROB_UNINITIALIZED.
21263 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
21264 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
21265 (PRED_LOOP_ITERATIONS_MAX): Likewise.
21266 (PRED_LOOP_IV_COMPARE): Likewise.
21267 * predict.h (PROB_UNINITIALIZED): Define new constant.
21269 2018-01-19 Martin Liska <mliska@suse.cz>
21271 * predict.c (dump_prediction): Add new format for
21272 analyze_brprob.py script which is enabled with -details
21274 * profile-count.h (precise_p): New function.
21276 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
21278 PR tree-optimization/83922
21279 * tree-vect-loop.c (vect_verify_full_masking): Return false if
21280 there are no statements that need masking.
21281 (vect_active_double_reduction_p): New function.
21282 (vect_analyze_loop_operations): Use it when handling phis that
21283 are not in the loop header.
21285 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
21287 PR tree-optimization/83914
21288 * tree-vect-loop.c (vectorizable_induction): Don't convert
21289 init_expr or apply the peeling adjustment for inductions
21290 that are nested within the vectorized loop.
21292 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21294 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
21297 2018-01-18 Jakub Jelinek <jakub@redhat.com>
21301 * function.h (gimplify_parameters): Add gimple_seq * argument.
21302 * function.c: Include gimple.h and options.h.
21303 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
21304 for the added local temporaries if needed.
21305 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
21306 if there are any parameter cleanups, wrap whole body into a
21307 try/finally with the cleanups.
21309 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21312 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
21313 Use GET_MODE_CLASS for scalar floating point.
21315 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
21319 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
21320 Fix call of call_cgraph_insertion_hooks.
21322 2018-01-18 Martin Sebor <msebor@redhat.com>
21324 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
21326 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
21329 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
21332 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
21335 * common.opt: (-ffile-prefix-map): New option.
21336 * opts.c (common_handle_option): Defer it.
21337 * opts-global.c (handle_common_deferred_options): Handle it.
21338 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
21339 * file-prefix-map.h: New file.
21340 (remap_debug_filename, add_debug_prefix_map): ...here.
21341 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
21342 * final.c (debug_prefix_map, add_debug_prefix_map
21343 remap_debug_filename): Move to...
21344 * file-prefix-map.c: New file.
21345 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
21346 generalize, get rid of alloca(), use strrchr() instead of strchr().
21347 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
21348 Implement in terms of add_prefix_map().
21349 (remap_macro_filename, remap_debug_filename): Implement in term of
21351 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
21352 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
21353 * dbxout.c: Include file-prefix-map.h.
21354 * varasm.c: Likewise.
21355 * vmsdbgout.c: Likewise.
21356 * xcoffout.c: Likewise.
21357 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
21358 * doc/cppopts.texi (-fmacro-prefix-map): Document.
21359 * doc/invoke.texi (-ffile-prefix-map): Document.
21360 (-fdebug-prefix-map): Update description.
21362 2018-01-18 Martin Liska <mliska@suse.cz>
21364 * config/i386/i386.c (indirect_thunk_name): Document that also
21366 (output_indirect_thunk): Document why both instructions
21367 (pause and lfence) are generated.
21369 2018-01-18 Richard Biener <rguenther@suse.de>
21371 PR tree-optimization/83887
21372 * graphite-scop-detection.c
21373 (scop_detection::get_nearest_dom_with_single_entry): Remove.
21374 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
21375 (scop_detection::merge_sese): Re-implement with a flood-fill
21376 algorithm that properly finds a SESE region if it exists.
21378 2018-01-18 Jakub Jelinek <jakub@redhat.com>
21381 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
21382 pointer_diff optimizations use view_convert instead of convert.
21384 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21386 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
21387 Generate different code for -mno-speculate-indirect-jumps.
21388 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
21389 (*call_indirect_aix<mode>): Disable for
21390 -mno-speculate-indirect-jumps.
21391 (*call_indirect_aix<mode>_nospec): New define_insn.
21392 (*call_value_indirect_aix<mode>): Disable for
21393 -mno-speculate-indirect-jumps.
21394 (*call_value_indirect_aix<mode>_nospec): New define_insn.
21395 (*sibcall_nonlocal_sysv<mode>): Generate different code for
21396 -mno-speculate-indirect-jumps.
21397 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
21399 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
21401 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
21402 long double type, set the flags for noting the default long double
21403 type, even if we don't pass or return a long double type.
21405 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
21408 * ipa-inline.c (flatten_function): Do not overwrite final inlining
21411 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
21413 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
21414 support for merge[hl].
21415 (fold_mergehl_helper): New helper function.
21416 (tree-vector-builder.h): New #include for tree_vector_builder usage.
21417 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
21418 (altivec_vmrglw_direct): Add xxmrglw insn.
21420 2018-01-17 Andrew Waterman <andrew@sifive.com>
21422 * config/riscv/riscv.c (riscv_conditional_register_usage): If
21423 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
21425 2018-01-17 David Malcolm <dmalcolm@redhat.com>
21428 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
21429 call the lto_location_cache before reading the
21430 DECL_SOURCE_LOCATION of the types.
21432 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21433 Richard Sandiford <richard.sandiford@linaro.org>
21435 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
21436 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
21437 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
21438 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
21439 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
21441 * config/aarch64/constraints.md (aarch64_movti_operand):
21443 * config/aarch64/predicates.md (Uti): Add new constraint.
21445 2018-01-17 Carl Love <cel@us.ibm.com>
21447 * config/rs6000/vsx.md (define_expand xl_len_r,
21448 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
21449 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
21451 (define_expand, define_insn): Move the shift left from the
21452 define_insn to the define_expand for lxvl and stxvl instructions.
21453 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
21454 and XL_LEN_R definitions to PURE.
21456 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
21458 * config/i386/i386.c (indirect_thunk_name): Declare regno
21459 as unsigned int. Compare regno with INVALID_REGNUM.
21460 (output_indirect_thunk): Ditto.
21461 (output_indirect_thunk_function): Ditto.
21462 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
21463 in the call to output_indirect_thunk_function.
21465 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
21467 PR middle-end/83884
21468 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
21469 rather than the size of inner_type to determine the stack slot size
21470 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
21472 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
21475 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
21478 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21480 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
21481 endian Linux systems to optionally enable multilibs for selecting
21482 the long double type if the user configured an explicit type.
21483 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
21484 have no long double multilibs if not defined.
21485 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
21486 warn if the user used -mabi={ieee,ibm}longdouble and we built
21487 multilibs for long double.
21488 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
21489 appropriate multilib option.
21490 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
21492 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
21493 for building long double multilibs.
21494 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
21496 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
21498 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
21501 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
21503 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
21506 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
21509 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
21512 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
21514 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
21515 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
21517 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
21519 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
21520 different rtl trees depending on TARGET_64BIT.
21521 (rs6000_gen_lvx): Likewise.
21523 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
21525 * config/visium/visium.md (nop): Tweak comment.
21526 (hazard_nop): Likewise.
21528 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21530 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
21531 -mspeculate-indirect-jumps.
21532 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
21533 for -mno-speculate-indirect-jumps.
21534 (*call_indirect_elfv2<mode>_nospec): New define_insn.
21535 (*call_value_indirect_elfv2<mode>): Disable for
21536 -mno-speculate-indirect-jumps.
21537 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
21538 (indirect_jump): Emit different RTL for
21539 -mno-speculate-indirect-jumps.
21540 (*indirect_jump<mode>): Disable for
21541 -mno-speculate-indirect-jumps.
21542 (*indirect_jump<mode>_nospec): New define_insn.
21543 (tablejump): Emit different RTL for
21544 -mno-speculate-indirect-jumps.
21545 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
21546 (tablejumpsi_nospec): New define_expand.
21547 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
21548 (tablejumpdi_nospec): New define_expand.
21549 (*tablejump<mode>_internal1): Disable for
21550 -mno-speculate-indirect-jumps.
21551 (*tablejump<mode>_internal1_nospec): New define_insn.
21552 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
21555 2018-01-16 Artyom Skrobov tyomitch@gmail.com
21557 * caller-save.c (insert_save): Drop unnecessary parameter. All
21560 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21561 Richard Biener <rguenth@suse.de>
21564 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
21565 return early, inline manually is_gimple_sizepos. Make sure if we
21566 call gimplify_expr we don't end up with a gimple constant.
21567 * tree.c (variably_modified_type_p): Don't return true for
21568 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
21569 * gimplify.h (is_gimple_sizepos): Remove.
21571 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21573 PR tree-optimization/83857
21574 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
21575 vectorizable_live_operation for pure SLP statements.
21576 (vectorizable_live_operation): Handle PHIs.
21578 2018-01-16 Richard Biener <rguenther@suse.de>
21580 PR tree-optimization/83867
21581 * tree-vect-stmts.c (vect_transform_stmt): Precompute
21582 nested_in_vect_loop_p since the scalar stmt may get invalidated.
21584 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21587 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
21588 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
21589 If off is not INTEGER_CST, issue a may not be aligned warning
21590 rather than isn't aligned. Use isn%'t rather than isn't.
21591 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
21593 <case MULT_EXPR>: Improve the case when bottom and one of the
21594 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
21595 operand, in that case check if the other operand is multiple of
21596 bottom divided by the INTEGER_CST operand.
21598 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21601 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
21602 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
21603 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
21604 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
21605 * config/pa/pa.c (pa_function_arg_advance): Likewise.
21606 (pa_function_arg, pa_arg_partial_bytes): Likewise.
21607 (pa_function_arg_size): New function.
21609 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21611 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
21612 in a separate statement.
21614 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21616 PR tree-optimization/83847
21617 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
21618 group gathers and scatters.
21620 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21622 PR rtl-optimization/86620
21623 * params.def (max-sched-ready-insns): Bump minimum value to 1.
21625 PR rtl-optimization/83213
21626 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
21627 to last if both are JUMP_INSNs.
21629 PR tree-optimization/83843
21630 * gimple-ssa-store-merging.c
21631 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
21632 store_immediate_info for bswap/nop orig_stores.
21634 2018-01-15 Andrew Waterman <andrew@sifive.com>
21636 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
21638 <UDIV>: Increase cost if !TARGET_DIV.
21640 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21642 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
21643 (define_attr "cr_logical_3op"): New.
21644 (cceq_ior_compare): Adjust.
21645 (cceq_ior_compare_complement): Adjust.
21646 (*cceq_rev_compare): Adjust.
21647 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
21648 (is_cracked_insn): Adjust.
21649 (insn_must_be_first_in_group): Adjust.
21650 * config/rs6000/40x.md: Adjust.
21651 * config/rs6000/440.md: Adjust.
21652 * config/rs6000/476.md: Adjust.
21653 * config/rs6000/601.md: Adjust.
21654 * config/rs6000/603.md: Adjust.
21655 * config/rs6000/6xx.md: Adjust.
21656 * config/rs6000/7450.md: Adjust.
21657 * config/rs6000/7xx.md: Adjust.
21658 * config/rs6000/8540.md: Adjust.
21659 * config/rs6000/cell.md: Adjust.
21660 * config/rs6000/e300c2c3.md: Adjust.
21661 * config/rs6000/e500mc.md: Adjust.
21662 * config/rs6000/e500mc64.md: Adjust.
21663 * config/rs6000/e5500.md: Adjust.
21664 * config/rs6000/e6500.md: Adjust.
21665 * config/rs6000/mpc.md: Adjust.
21666 * config/rs6000/power4.md: Adjust.
21667 * config/rs6000/power5.md: Adjust.
21668 * config/rs6000/power6.md: Adjust.
21669 * config/rs6000/power7.md: Adjust.
21670 * config/rs6000/power8.md: Adjust.
21671 * config/rs6000/power9.md: Adjust.
21672 * config/rs6000/rs64.md: Adjust.
21673 * config/rs6000/titan.md: Adjust.
21675 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21677 * config/i386/predicates.md (indirect_branch_operand): Rewrite
21678 ix86_indirect_branch_register logic.
21680 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21682 * config/i386/constraints.md (Bs): Update
21683 ix86_indirect_branch_register check. Don't check
21684 ix86_indirect_branch_register with GOT_memory_operand.
21686 * config/i386/predicates.md (GOT_memory_operand): Don't check
21687 ix86_indirect_branch_register here.
21688 (GOT32_symbol_operand): Likewise.
21690 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21692 * config/i386/predicates.md (constant_call_address_operand):
21693 Rewrite ix86_indirect_branch_register logic.
21694 (sibcall_insn_operand): Likewise.
21696 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21698 * config/i386/constraints.md (Bs): Replace
21699 ix86_indirect_branch_thunk_register with
21700 ix86_indirect_branch_register.
21702 * config/i386/i386.md (indirect_jump): Likewise.
21703 (tablejump): Likewise.
21704 (*sibcall_memory): Likewise.
21705 (*sibcall_value_memory): Likewise.
21706 Peepholes of indirect call and jump via memory: Likewise.
21707 * config/i386/i386.opt: Likewise.
21708 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21709 (GOT_memory_operand): Likewise.
21710 (call_insn_operand): Likewise.
21711 (sibcall_insn_operand): Likewise.
21712 (GOT32_symbol_operand): Likewise.
21714 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21716 PR middle-end/83837
21717 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
21718 type rather than type addr's type points to.
21719 (expand_omp_atomic_mutex): Likewise.
21720 (expand_omp_atomic): Likewise.
21722 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21725 * config/i386/i386.c (output_indirect_thunk_function): Use
21726 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
21727 for __x86_return_thunk.
21729 2018-01-15 Richard Biener <rguenther@suse.de>
21731 PR middle-end/83850
21732 * expmed.c (extract_bit_field_1): Fix typo.
21734 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21737 * config/arm/iterators.md (VF): New mode iterator.
21738 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
21739 Remove integer-related logic from pattern.
21740 (neon_vabd<mode>_3): Likewise.
21742 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21744 PR middle-end/82694
21745 * common.opt (fstrict-overflow): No longer an alias.
21746 (fwrapv-pointer): New option.
21747 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
21748 also for pointer types based on flag_wrapv_pointer.
21749 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
21750 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
21751 opts->x_flag_wrapv got set.
21752 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
21753 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
21754 POINTER_TYPE_OVERFLOW_UNDEFINED.
21755 * match.pd: Likewise in address comparison pattern.
21756 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
21758 2018-01-15 Richard Biener <rguenther@suse.de>
21761 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
21762 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
21763 Reset type names to their identifier if their TYPE_DECL doesn't
21764 have linkage (and thus is used for ODR and devirt).
21765 (save_debug_info_for_decl): Remove.
21766 (save_debug_info_for_type): Likewise.
21767 (add_tree_to_fld_list): Adjust.
21768 * tree-pretty-print.c (dump_generic_node): Make dumping of
21769 type names more robust.
21771 2018-01-15 Richard Biener <rguenther@suse.de>
21773 * BASE-VER: Bump to 8.0.1.
21775 2018-01-14 Martin Sebor <msebor@redhat.com>
21778 * builtins.c (check_access): Avoid warning when the no-warning bit
21781 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
21783 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
21784 * ira-color (allocno_hard_regs_compare): Likewise.
21786 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
21789 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
21790 Use .pushsection/.popsection.
21792 2018-01-14 Martin Sebor <msebor@redhat.com>
21795 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
21797 2018-01-14 Jakub Jelinek <jakub@redhat.com>
21799 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
21800 entry from extra_headers.
21801 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
21802 extra_headers, make the list bitwise identical to the i?86-*-* one.
21804 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21806 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
21807 -mcmodel=large with -mindirect-branch=thunk,
21808 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
21809 -mfunction-return=thunk-extern.
21810 * doc/invoke.texi: Document -mcmodel=large is incompatible with
21811 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
21812 -mfunction-return=thunk and -mfunction-return=thunk-extern.
21814 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21816 * config/i386/i386.c (print_reg): Print the name of the full
21817 integer register without '%'.
21818 (ix86_print_operand): Handle 'V'.
21819 * doc/extend.texi: Document 'V' modifier.
21821 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21823 * config/i386/constraints.md (Bs): Disallow memory operand for
21824 -mindirect-branch-register.
21826 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21827 (GOT_memory_operand): Likewise.
21828 (call_insn_operand): Likewise.
21829 (sibcall_insn_operand): Likewise.
21830 (GOT32_symbol_operand): Likewise.
21831 * config/i386/i386.md (indirect_jump): Call convert_memory_address
21832 for -mindirect-branch-register.
21833 (tablejump): Likewise.
21834 (*sibcall_memory): Likewise.
21835 (*sibcall_value_memory): Likewise.
21836 Disallow peepholes of indirect call and jump via memory for
21837 -mindirect-branch-register.
21838 (*call_pop): Replace m with Bw.
21839 (*call_value_pop): Likewise.
21840 (*sibcall_pop_memory): Replace m with Bs.
21841 * config/i386/i386.opt (mindirect-branch-register): New option.
21842 * doc/invoke.texi: Document -mindirect-branch-register option.
21844 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21846 * config/i386/i386-protos.h (ix86_output_function_return): New.
21847 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
21848 set function_return_type.
21849 (indirect_thunk_name): Add ret_p to indicate thunk for function
21851 (output_indirect_thunk_function): Pass false to
21852 indirect_thunk_name.
21853 (ix86_output_indirect_branch_via_reg): Likewise.
21854 (ix86_output_indirect_branch_via_push): Likewise.
21855 (output_indirect_thunk_function): Create alias for function
21856 return thunk if regno < 0.
21857 (ix86_output_function_return): New function.
21858 (ix86_handle_fndecl_attribute): Handle function_return.
21859 (ix86_attribute_table): Add function_return.
21860 * config/i386/i386.h (machine_function): Add
21861 function_return_type.
21862 * config/i386/i386.md (simple_return_internal): Use
21863 ix86_output_function_return.
21864 (simple_return_internal_long): Likewise.
21865 * config/i386/i386.opt (mfunction-return=): New option.
21866 (indirect_branch): Mention -mfunction-return=.
21867 * doc/extend.texi: Document function_return function attribute.
21868 * doc/invoke.texi: Document -mfunction-return= option.
21870 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21872 * config/i386/i386-opts.h (indirect_branch): New.
21873 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
21874 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
21875 with local indirect jump when converting indirect call and jump.
21876 (ix86_set_indirect_branch_type): New.
21877 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
21878 (indirectlabelno): New.
21879 (indirect_thunk_needed): Likewise.
21880 (indirect_thunk_bnd_needed): Likewise.
21881 (indirect_thunks_used): Likewise.
21882 (indirect_thunks_bnd_used): Likewise.
21883 (INDIRECT_LABEL): Likewise.
21884 (indirect_thunk_name): Likewise.
21885 (output_indirect_thunk): Likewise.
21886 (output_indirect_thunk_function): Likewise.
21887 (ix86_output_indirect_branch_via_reg): Likewise.
21888 (ix86_output_indirect_branch_via_push): Likewise.
21889 (ix86_output_indirect_branch): Likewise.
21890 (ix86_output_indirect_jmp): Likewise.
21891 (ix86_code_end): Call output_indirect_thunk_function if needed.
21892 (ix86_output_call_insn): Call ix86_output_indirect_branch if
21894 (ix86_handle_fndecl_attribute): Handle indirect_branch.
21895 (ix86_attribute_table): Add indirect_branch.
21896 * config/i386/i386.h (machine_function): Add indirect_branch_type
21897 and has_local_indirect_jump.
21898 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
21900 (tablejump): Likewise.
21901 (*indirect_jump): Use ix86_output_indirect_jmp.
21902 (*tablejump_1): Likewise.
21903 (simple_return_indirect_internal): Likewise.
21904 * config/i386/i386.opt (mindirect-branch=): New option.
21905 (indirect_branch): New.
21908 (thunk-inline): Likewise.
21909 (thunk-extern): Likewise.
21910 * doc/extend.texi: Document indirect_branch function attribute.
21911 * doc/invoke.texi: Document -mindirect-branch= option.
21913 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
21916 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
21918 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
21920 * ipa-inline.c (want_inline_small_function_p): Return false if
21921 inlining has already failed with CIF_FINAL_ERROR.
21922 (update_caller_keys): Call want_inline_small_function_p before
21924 (update_callee_keys): Likewise.
21926 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21928 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
21930 (rs6000_quadword_masked_address_p): Likewise.
21931 (quad_aligned_load_p): Likewise.
21932 (quad_aligned_store_p): Likewise.
21933 (const_load_sequence_p): Add comment to describe the outer-most loop.
21934 (mimic_memory_attributes_and_flags): New function.
21935 (rs6000_gen_stvx): Likewise.
21936 (replace_swapped_aligned_store): Likewise.
21937 (rs6000_gen_lvx): Likewise.
21938 (replace_swapped_aligned_load): Likewise.
21939 (replace_swapped_load_constant): Capitalize argument name in
21940 comment describing this function.
21941 (rs6000_analyze_swaps): Add a third pass to search for vector loads
21942 and stores that access quad-word aligned addresses and replace
21943 with stvx or lvx instructions when appropriate.
21944 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
21945 New function prototype.
21946 (rs6000_quadword_masked_address_p): Likewise.
21947 (rs6000_gen_lvx): Likewise.
21948 (rs6000_gen_stvx): Likewise.
21949 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
21950 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
21951 when memory address is aligned.
21952 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
21953 this split to select lvx instruction when memory address is aligned.
21954 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
21955 instruction when memory address is aligned.
21956 (*vsx_le_perm_load_v16qi): Likewise.
21957 (four unnamed splitters): Modify to select the stvx instruction
21958 when memory is aligned.
21960 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
21962 * predict.c (determine_unlikely_bbs): Handle correctly BBs
21963 which appears in the queue multiple times.
21965 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21966 Alan Hayward <alan.hayward@arm.com>
21967 David Sherwood <david.sherwood@arm.com>
21969 * tree-vectorizer.h (vec_lower_bound): New structure.
21970 (_loop_vec_info): Add check_nonzero and lower_bounds.
21971 (LOOP_VINFO_CHECK_NONZERO): New macro.
21972 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
21973 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
21974 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
21975 fields. Make seg_len the distance travelled, not including the
21977 (dr_direction_indicator): Declare.
21978 (dr_zero_step_indicator): Likewise.
21979 (dr_known_forward_stride_p): Likewise.
21980 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
21982 (runtime_alias_check_p): Allow runtime alias checks with
21984 (operator ==): Compare access_size and align.
21985 (prune_runtime_alias_test_list): Rework for new distinction between
21986 the access_size and seg_len.
21987 (create_intersect_range_checks_index): Likewise. Cope with polynomial
21989 (get_segment_min_max): New function.
21990 (create_intersect_range_checks): Use it.
21991 (dr_step_indicator): New function.
21992 (dr_direction_indicator): Likewise.
21993 (dr_zero_step_indicator): Likewise.
21994 (dr_known_forward_stride_p): Likewise.
21995 * tree-loop-distribution.c (data_ref_segment_size): Return
21996 DR_STEP * (niters - 1).
21997 (compute_alias_check_pairs): Update call to the dr_with_seg_len
21999 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
22000 (vect_preserves_scalar_order_p): New function, split out from...
22001 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
22002 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
22003 (vect_vfa_access_size): New function.
22004 (vect_vfa_align): Likewise.
22005 (vect_compile_time_alias): Take access_size_a and access_b arguments.
22006 (dump_lower_bound): New function.
22007 (vect_check_lower_bound): Likewise.
22008 (vect_small_gap_p): Likewise.
22009 (vectorizable_with_step_bound_p): Likewise.
22010 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
22011 depencies if the vectorization factor is 1. Convert the checks
22012 for nonzero steps into checks on the bounds of DR_STEP. Try using
22013 a bunds check for variable steps if the minimum required step is
22014 relatively small. Update calls to the dr_with_seg_len
22015 constructor and to vect_compile_time_alias.
22016 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
22018 (vect_loop_versioning): Call it.
22019 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
22021 (vect_estimate_min_profitable_iters): Account for any bounds checks.
22023 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22024 Alan Hayward <alan.hayward@arm.com>
22025 David Sherwood <david.sherwood@arm.com>
22027 * doc/sourcebuild.texi (vect_scatter_store): Document.
22028 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
22030 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
22032 * genopinit.c (main): Add supports_vec_scatter_store and
22033 supports_vec_scatter_store_cached to target_optabs.
22034 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
22035 IFN_MASK_SCATTER_STORE.
22036 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
22038 * internal-fn.h (internal_store_fn_p): Declare.
22039 (internal_fn_stored_value_index): Likewise.
22040 * internal-fn.c (scatter_store_direct): New macro.
22041 (expand_scatter_store_optab_fn): New function.
22042 (direct_scatter_store_optab_supported_p): New macro.
22043 (internal_store_fn_p): New function.
22044 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
22045 IFN_MASK_SCATTER_STORE.
22046 (internal_fn_mask_index): Likewise.
22047 (internal_fn_stored_value_index): New function.
22048 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
22049 for scatter stores.
22050 * optabs-query.h (supports_vec_scatter_store_p): Declare.
22051 * optabs-query.c (supports_vec_scatter_store_p): New function.
22052 * tree-vectorizer.h (vect_get_store_rhs): Declare.
22053 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
22054 true for scatter stores.
22055 (vect_gather_scatter_fn_p): Handle scatter stores too.
22056 (vect_check_gather_scatter): Consider using scatter stores if
22057 supports_vec_scatter_store_p.
22058 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
22059 scatter stores too.
22060 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
22061 internal_fn_stored_value_index.
22062 (check_load_store_masking): Handle scatter stores too.
22063 (vect_get_store_rhs): Make public.
22064 (vectorizable_call): Use internal_store_fn_p.
22065 (vectorizable_store): Handle scatter store internal functions.
22066 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
22067 when deciding whether the end of the group has been reached.
22068 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
22069 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
22070 (mask_scatter_store<mode>): New insns.
22072 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22073 Alan Hayward <alan.hayward@arm.com>
22074 David Sherwood <david.sherwood@arm.com>
22076 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
22077 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
22078 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
22080 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
22081 Use vect_truncate_gather_scatter_offset if we can't treat the
22082 operation as a normal gather load or scatter store.
22083 (get_group_load_store_type): Take the gather_scatter_info
22084 as argument. Try using a gather load or scatter store for
22085 single-element groups.
22086 (get_load_store_type): Update calls to get_group_load_store_type
22087 and vect_use_strided_gather_scatters_p.
22089 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22090 Alan Hayward <alan.hayward@arm.com>
22091 David Sherwood <david.sherwood@arm.com>
22093 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
22094 optional tree argument.
22095 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
22097 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
22098 but continue to use the current value as a fallback.
22099 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
22100 to compare the updates.
22101 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
22102 (get_load_store_type): Use it when handling a strided access.
22103 (vect_get_strided_load_store_ops): New function.
22104 (vect_get_data_ptr_increment): Likewise.
22105 (vectorizable_load): Handle strided gather loads. Always pass
22106 a step to vect_create_data_ref_ptr and bump_vector_ptr.
22108 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22109 Alan Hayward <alan.hayward@arm.com>
22110 David Sherwood <david.sherwood@arm.com>
22112 * doc/md.texi (gather_load@var{m}): Document.
22113 (mask_gather_load@var{m}): Likewise.
22114 * genopinit.c (main): Add supports_vec_gather_load and
22115 supports_vec_gather_load_cached to target_optabs.
22116 * optabs-tree.c (init_tree_optimization_optabs): Use
22117 ggc_cleared_alloc to allocate target_optabs.
22118 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
22119 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
22121 * internal-fn.h (internal_load_fn_p): Declare.
22122 (internal_gather_scatter_fn_p): Likewise.
22123 (internal_fn_mask_index): Likewise.
22124 (internal_gather_scatter_fn_supported_p): Likewise.
22125 * internal-fn.c (gather_load_direct): New macro.
22126 (expand_gather_load_optab_fn): New function.
22127 (direct_gather_load_optab_supported_p): New macro.
22128 (direct_internal_fn_optab): New function.
22129 (internal_load_fn_p): Likewise.
22130 (internal_gather_scatter_fn_p): Likewise.
22131 (internal_fn_mask_index): Likewise.
22132 (internal_gather_scatter_fn_supported_p): Likewise.
22133 * optabs-query.c (supports_at_least_one_mode_p): New function.
22134 (supports_vec_gather_load_p): Likewise.
22135 * optabs-query.h (supports_vec_gather_load_p): Declare.
22136 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
22137 and memory_type field.
22138 (NUM_PATTERNS): Bump to 15.
22139 * tree-vect-data-refs.c: Include internal-fn.h.
22140 (vect_gather_scatter_fn_p): New function.
22141 (vect_describe_gather_scatter_call): Likewise.
22142 (vect_check_gather_scatter): Try using internal functions for
22143 gather loads. Recognize existing calls to a gather load function.
22144 (vect_analyze_data_refs): Consider using gather loads if
22145 supports_vec_gather_load_p.
22146 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
22147 (vect_get_gather_scatter_offset_type): Likewise.
22148 (vect_convert_mask_for_vectype): Likewise.
22149 (vect_add_conversion_to_patterm): Likewise.
22150 (vect_try_gather_scatter_pattern): Likewise.
22151 (vect_recog_gather_scatter_pattern): New pattern recognizer.
22152 (vect_vect_recog_func_ptrs): Add it.
22153 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
22154 internal_fn_mask_index and internal_gather_scatter_fn_p.
22155 (check_load_store_masking): Take the gather_scatter_info as an
22156 argument and handle gather loads.
22157 (vect_get_gather_scatter_ops): New function.
22158 (vectorizable_call): Check internal_load_fn_p.
22159 (vectorizable_load): Likewise. Handle gather load internal
22161 (vectorizable_store): Update call to check_load_store_masking.
22162 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
22163 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
22164 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
22165 (aarch64_gather_scale_operand_d): New predicates.
22166 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
22167 (mask_gather_load<mode>): New insns.
22169 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22170 Alan Hayward <alan.hayward@arm.com>
22171 David Sherwood <david.sherwood@arm.com>
22173 * optabs.def (fold_left_plus_optab): New optab.
22174 * doc/md.texi (fold_left_plus_@var{m}): Document.
22175 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
22176 * internal-fn.c (fold_left_direct): Define.
22177 (expand_fold_left_optab_fn): Likewise.
22178 (direct_fold_left_optab_supported_p): Likewise.
22179 * fold-const-call.c (fold_const_fold_left): New function.
22180 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
22181 * tree-parloops.c (valid_reduction_p): New function.
22182 (gather_scalar_reductions): Use it.
22183 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
22184 (vect_finish_replace_stmt): Declare.
22185 * tree-vect-loop.c (fold_left_reduction_fn): New function.
22186 (needs_fold_left_reduction_p): New function, split out from...
22187 (vect_is_simple_reduction): ...here. Accept reductions that
22188 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
22189 (vect_force_simple_reduction): Also store the reduction type in
22190 the assignment's STMT_VINFO_REDUC_TYPE.
22191 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
22192 (merge_with_identity): New function.
22193 (vect_expand_fold_left): Likewise.
22194 (vectorize_fold_left_reduction): Likewise.
22195 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
22196 scalar phi in place for it. Check for target support and reject
22197 cases that would reassociate the operation. Defer the transform
22198 phase to vectorize_fold_left_reduction.
22199 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
22200 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
22201 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
22203 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22205 * tree-if-conv.c (predicate_mem_writes): Remove redundant
22206 call to ifc_temp_var.
22208 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22209 Alan Hayward <alan.hayward@arm.com>
22210 David Sherwood <david.sherwood@arm.com>
22212 * target.def (legitimize_address_displacement): Take the original
22213 offset as a poly_int.
22214 * targhooks.h (default_legitimize_address_displacement): Update
22216 * targhooks.c (default_legitimize_address_displacement): Likewise.
22217 * doc/tm.texi: Regenerate.
22218 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
22219 as an argument, moving assert of ad->disp == ad->disp_term to...
22220 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
22221 Try calling targetm.legitimize_address_displacement before expanding
22222 the address rather than afterwards, and adjust for the new interface.
22223 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
22224 Match the new hook interface. Handle SVE addresses.
22225 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
22226 new hook interface.
22228 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22230 * Makefile.in (OBJS): Add early-remat.o.
22231 * target.def (select_early_remat_modes): New hook.
22232 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
22233 * doc/tm.texi: Regenerate.
22234 * targhooks.h (default_select_early_remat_modes): Declare.
22235 * targhooks.c (default_select_early_remat_modes): New function.
22236 * timevar.def (TV_EARLY_REMAT): New timevar.
22237 * passes.def (pass_early_remat): New pass.
22238 * tree-pass.h (make_pass_early_remat): Declare.
22239 * early-remat.c: New file.
22240 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
22242 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
22244 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22245 Alan Hayward <alan.hayward@arm.com>
22246 David Sherwood <david.sherwood@arm.com>
22248 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
22249 vfm1 with a bound_epilog parameter.
22250 (vect_do_peeling): Update calls accordingly, and move the prologue
22251 call earlier in the function. Treat the base bound_epilog as 0 for
22252 fully-masked loops and retain vf - 1 for other loops. Add 1 to
22253 this base when peeling for gaps.
22254 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
22255 with fully-masked loops.
22256 (vect_estimate_min_profitable_iters): Handle the single peeled
22257 iteration in that case.
22259 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22260 Alan Hayward <alan.hayward@arm.com>
22261 David Sherwood <david.sherwood@arm.com>
22263 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
22264 single-element interleaving even if the size is not a power of 2.
22265 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
22266 accesses for single-element interleaving if the group size is
22269 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22270 Alan Hayward <alan.hayward@arm.com>
22271 David Sherwood <david.sherwood@arm.com>
22273 * doc/md.texi (fold_extract_last_@var{m}): Document.
22274 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
22275 * optabs.def (fold_extract_last_optab): New optab.
22276 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
22277 * internal-fn.c (fold_extract_direct): New macro.
22278 (expand_fold_extract_optab_fn): Likewise.
22279 (direct_fold_extract_optab_supported_p): Likewise.
22280 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
22281 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22282 EXTRACT_LAST_REDUCTION.
22283 (get_initial_def_for_reduction): Do not create an initial vector
22284 for EXTRACT_LAST_REDUCTION reductions.
22285 (vectorizable_reduction): Leave the scalar phi in place for
22286 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
22287 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
22288 epilogue code for EXTRACT_LAST_REDUCTION and defer the
22289 transform phase to vectorizable_condition.
22290 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
22292 (vect_finish_stmt_generation): ...here.
22293 (vect_finish_replace_stmt): New function.
22294 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
22295 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
22297 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
22299 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22300 Alan Hayward <alan.hayward@arm.com>
22301 David Sherwood <david.sherwood@arm.com>
22303 * doc/md.texi (extract_last_@var{m}): Document.
22304 * optabs.def (extract_last_optab): New optab.
22305 * internal-fn.def (EXTRACT_LAST): New internal function.
22306 * internal-fn.c (cond_unary_direct): New macro.
22307 (expand_cond_unary_optab_fn): Likewise.
22308 (direct_cond_unary_optab_supported_p): Likewise.
22309 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
22310 loops using EXTRACT_LAST.
22311 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
22312 (extract_last_<mode>): ...this optab.
22313 (vec_extract<mode><Vel>): Update accordingly.
22315 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22316 Alan Hayward <alan.hayward@arm.com>
22317 David Sherwood <david.sherwood@arm.com>
22319 * target.def (empty_mask_is_expensive): New hook.
22320 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
22321 * doc/tm.texi: Regenerate.
22322 * targhooks.h (default_empty_mask_is_expensive): Declare.
22323 * targhooks.c (default_empty_mask_is_expensive): New function.
22324 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
22325 if the target says that empty masks are expensive.
22326 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
22328 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
22330 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22331 Alan Hayward <alan.hayward@arm.com>
22332 David Sherwood <david.sherwood@arm.com>
22334 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
22335 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
22336 (vect_use_loop_mask_for_alignment_p): New function.
22337 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
22338 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
22339 niters_skip argument. Make sure that the first niters_skip elements
22340 of the first iteration are inactive.
22341 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
22342 Update call to vect_set_loop_masks_directly.
22343 (get_misalign_in_elems): New function, split out from...
22344 (vect_gen_prolog_loop_niters): ...here.
22345 (vect_update_init_of_dr): Take a code argument that specifies whether
22346 the adjustment should be added or subtracted.
22347 (vect_update_init_of_drs): Likewise.
22348 (vect_prepare_for_masked_peels): New function.
22349 (vect_do_peeling): Skip prologue peeling if we're using a mask
22350 instead. Update call to vect_update_inits_of_drs.
22351 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
22353 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
22354 alignment. Do not include the number of peeled iterations in
22355 the minimum threshold in that case.
22356 (vectorizable_induction): Adjust the start value down by
22357 LOOP_VINFO_MASK_SKIP_NITERS iterations.
22358 (vect_transform_loop): Call vect_prepare_for_masked_peels.
22359 Take the number of skipped iterations into account when calculating
22361 * tree-vect-stmts.c (vect_gen_while_not): New function.
22363 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22364 Alan Hayward <alan.hayward@arm.com>
22365 David Sherwood <david.sherwood@arm.com>
22367 * doc/sourcebuild.texi (vect_fully_masked): Document.
22368 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
22369 default value to 0.
22370 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
22372 (vect_analyze_loop_2): ...here. Don't check the vectorization
22373 factor against the number of loop iterations if the loop is
22376 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22377 Alan Hayward <alan.hayward@arm.com>
22378 David Sherwood <david.sherwood@arm.com>
22380 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
22381 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
22382 (dump_groups): Update accordingly.
22383 (iv_use::mem_type): New member variable.
22384 (address_p): New function.
22385 (record_use): Add a mem_type argument and initialize the new
22387 (record_group_use): Add a mem_type argument. Use address_p.
22388 Remove obsolete null checks of base_object. Update call to record_use.
22389 (find_interesting_uses_op): Update call to record_group_use.
22390 (find_interesting_uses_cond): Likewise.
22391 (find_interesting_uses_address): Likewise.
22392 (get_mem_type_for_internal_fn): New function.
22393 (find_address_like_use): Likewise.
22394 (find_interesting_uses_stmt): Try find_address_like_use before
22395 calling find_interesting_uses_op.
22396 (addr_offset_valid_p): Use the iv mem_type field as the type
22397 of the addressed memory.
22398 (add_autoinc_candidates): Likewise.
22399 (get_address_cost): Likewise.
22400 (split_small_address_groups_p): Use address_p.
22401 (split_address_groups): Likewise.
22402 (add_iv_candidate_for_use): Likewise.
22403 (autoinc_possible_for_pair): Likewise.
22404 (rewrite_groups): Likewise.
22405 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
22406 (determine_group_iv_cost): Update after split of USE_ADDRESS.
22407 (get_alias_ptr_type_for_ptr_address): New function.
22408 (rewrite_use_address): Rewrite address uses in calls that were
22409 identified by find_address_like_use.
22411 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22412 Alan Hayward <alan.hayward@arm.com>
22413 David Sherwood <david.sherwood@arm.com>
22415 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
22417 * gimple-expr.h (is_gimple_addressable: Likewise.
22418 * gimple-expr.c (is_gimple_address): Likewise.
22419 * internal-fn.c (expand_call_mem_ref): New function.
22420 (expand_mask_load_optab_fn): Use it.
22421 (expand_mask_store_optab_fn): Likewise.
22423 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22424 Alan Hayward <alan.hayward@arm.com>
22425 David Sherwood <david.sherwood@arm.com>
22427 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
22428 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
22429 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
22430 (cond_umax@var{mode}): Document.
22431 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
22432 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
22433 (cond_umin_optab, cond_umax_optab): New optabs.
22434 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
22435 (COND_IOR, COND_XOR): New internal functions.
22436 * internal-fn.h (get_conditional_internal_fn): Declare.
22437 * internal-fn.c (cond_binary_direct): New macro.
22438 (expand_cond_binary_optab_fn): Likewise.
22439 (direct_cond_binary_optab_supported_p): Likewise.
22440 (get_conditional_internal_fn): New function.
22441 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
22442 Cope with reduction statements that are vectorized as calls rather
22444 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
22445 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
22446 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
22447 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
22448 (UNSPEC_COND_EOR): New unspecs.
22449 (optab): Add mappings for them.
22450 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
22451 (sve_int_op, sve_fp_op): New int attributes.
22453 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22454 Alan Hayward <alan.hayward@arm.com>
22455 David Sherwood <david.sherwood@arm.com>
22457 * optabs.def (while_ult_optab): New optab.
22458 * doc/md.texi (while_ult@var{m}@var{n}): Document.
22459 * internal-fn.def (WHILE_ULT): New internal function.
22460 * internal-fn.h (direct_internal_fn_supported_p): New override
22461 that takes two types as argument.
22462 * internal-fn.c (while_direct): New macro.
22463 (expand_while_optab_fn): New function.
22464 (convert_optab_supported_p): Likewise.
22465 (direct_while_optab_supported_p): New macro.
22466 * wide-int.h (wi::udiv_ceil): New function.
22467 * tree-vectorizer.h (rgroup_masks): New structure.
22468 (vec_loop_masks): New typedef.
22469 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
22470 and fully_masked_p.
22471 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
22472 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
22473 (vect_max_vf): New function.
22474 (slpeel_make_loop_iterate_ntimes): Delete.
22475 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
22476 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
22477 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
22478 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
22479 internal-fn.h, stor-layout.h and optabs-query.h.
22480 (vect_set_loop_mask): New function.
22481 (add_preheader_seq): Likewise.
22482 (add_header_seq): Likewise.
22483 (interleave_supported_p): Likewise.
22484 (vect_maybe_permute_loop_masks): Likewise.
22485 (vect_set_loop_masks_directly): Likewise.
22486 (vect_set_loop_condition_masked): Likewise.
22487 (vect_set_loop_condition_unmasked): New function, split out from
22488 slpeel_make_loop_iterate_ntimes.
22489 (slpeel_make_loop_iterate_ntimes): Rename to..
22490 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
22491 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
22492 (vect_do_peeling): Update call accordingly.
22493 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
22495 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
22496 mask_compare_type, can_fully_mask_p and fully_masked_p.
22497 (release_vec_loop_masks): New function.
22498 (_loop_vec_info): Use it to free the loop masks.
22499 (can_produce_all_loop_masks_p): New function.
22500 (vect_get_max_nscalars_per_iter): Likewise.
22501 (vect_verify_full_masking): Likewise.
22502 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
22503 retries, and free the mask rgroups before retrying. Check loop-wide
22504 reasons for disallowing fully-masked loops. Make the final decision
22505 about whether use a fully-masked loop or not.
22506 (vect_estimate_min_profitable_iters): Do not assume that peeling
22507 for the number of iterations will be needed for fully-masked loops.
22508 (vectorizable_reduction): Disable fully-masked loops.
22509 (vectorizable_live_operation): Likewise.
22510 (vect_halve_mask_nunits): New function.
22511 (vect_double_mask_nunits): Likewise.
22512 (vect_record_loop_mask): Likewise.
22513 (vect_get_loop_mask): Likewise.
22514 (vect_transform_loop): Handle the case in which the final loop
22515 iteration might handle a partial vector. Call vect_set_loop_condition
22516 instead of slpeel_make_loop_iterate_ntimes.
22517 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
22518 (check_load_store_masking): New function.
22519 (prepare_load_store_mask): Likewise.
22520 (vectorizable_store): Handle fully-masked loops.
22521 (vectorizable_load): Likewise.
22522 (supportable_widening_operation): Use vect_halve_mask_nunits for
22524 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
22525 (vect_gen_while): New function.
22526 * config/aarch64/aarch64.md (umax<mode>3): New expander.
22527 (aarch64_uqdec<mode>): New insn.
22529 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22530 Alan Hayward <alan.hayward@arm.com>
22531 David Sherwood <david.sherwood@arm.com>
22533 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
22534 (reduc_xor_scal_optab): New optabs.
22535 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
22536 (reduc_xor_scal_@var{m}): Document.
22537 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
22538 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
22539 internal functions.
22540 * fold-const-call.c (fold_const_call): Handle them.
22541 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
22542 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
22543 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
22544 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
22545 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
22546 (UNSPEC_XORV): New unspecs.
22547 (optab): Add entries for them.
22548 (BITWISEV): New int iterator.
22549 (bit_reduc_op): New int attributes.
22551 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22552 Alan Hayward <alan.hayward@arm.com>
22553 David Sherwood <david.sherwood@arm.com>
22555 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
22556 * internal-fn.def (VEC_SHL_INSERT): New internal function.
22557 * optabs.def (vec_shl_insert_optab): New optab.
22558 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
22559 (duplicate_and_interleave): Likewise.
22560 * tree-vect-loop.c: Include internal-fn.h.
22561 (neutral_op_for_slp_reduction): New function, split out from
22562 get_initial_defs_for_reduction.
22563 (get_initial_def_for_reduction): Handle option 2 for variable-length
22564 vectors by loading the neutral value into a vector and then shifting
22565 the initial value into element 0.
22566 (get_initial_defs_for_reduction): Replace the code argument with
22567 the neutral value calculated by neutral_op_for_slp_reduction.
22568 Use gimple_build_vector for constant-length vectors.
22569 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
22570 but the first group_size elements have a neutral value.
22571 Use duplicate_and_interleave otherwise.
22572 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
22573 Update call to get_initial_defs_for_reduction. Handle SLP
22574 reductions for variable-length vectors by creating one vector
22575 result for each scalar result, with the elements associated
22576 with other scalar results stubbed out with the neutral value.
22577 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
22578 Require IFN_VEC_SHL_INSERT for double reductions on
22579 variable-length vectors, or SLP reductions that have
22580 a neutral value. Require can_duplicate_and_interleave_p
22581 support for variable-length unchained SLP reductions if there
22582 is no neutral value, such as for MIN/MAX reductions. Also require
22583 the number of vector elements to be a multiple of the number of
22584 SLP statements when doing variable-length unchained SLP reductions.
22585 Update call to vect_create_epilog_for_reduction.
22586 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
22587 and remove initial values.
22588 (duplicate_and_interleave): Make public.
22589 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
22590 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
22592 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22593 Alan Hayward <alan.hayward@arm.com>
22594 David Sherwood <david.sherwood@arm.com>
22596 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
22597 (can_duplicate_and_interleave_p): New function.
22598 (vect_get_and_check_slp_defs): Take the vector of statements
22599 rather than just the current one. Remove excess parentheses.
22600 Restriction rejectinon of vect_constant_def and vect_external_def
22601 for variable-length vectors to boolean types, or types for which
22602 can_duplicate_and_interleave_p is false.
22603 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
22604 (duplicate_and_interleave): New function.
22605 (vect_get_constant_vectors): Use gimple_build_vector for
22606 constant-length vectors and suitable variable-length constant
22607 vectors. Use duplicate_and_interleave for other variable-length
22608 vectors. Don't defer the update when inserting new statements.
22610 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22611 Alan Hayward <alan.hayward@arm.com>
22612 David Sherwood <david.sherwood@arm.com>
22614 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
22615 min_profitable_iters doesn't go negative.
22617 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22618 Alan Hayward <alan.hayward@arm.com>
22619 David Sherwood <david.sherwood@arm.com>
22621 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
22622 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
22623 * optabs.def (vec_mask_load_lanes_optab): New optab.
22624 (vec_mask_store_lanes_optab): Likewise.
22625 * internal-fn.def (MASK_LOAD_LANES): New internal function.
22626 (MASK_STORE_LANES): Likewise.
22627 * internal-fn.c (mask_load_lanes_direct): New macro.
22628 (mask_store_lanes_direct): Likewise.
22629 (expand_mask_load_optab_fn): Handle masked operations.
22630 (expand_mask_load_lanes_optab_fn): New macro.
22631 (expand_mask_store_optab_fn): Handle masked operations.
22632 (expand_mask_store_lanes_optab_fn): New macro.
22633 (direct_mask_load_lanes_optab_supported_p): Likewise.
22634 (direct_mask_store_lanes_optab_supported_p): Likewise.
22635 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
22637 (vect_load_lanes_supported): Likewise.
22638 * tree-vect-data-refs.c (strip_conversion): New function.
22639 (can_group_stmts_p): Likewise.
22640 (vect_analyze_data_ref_accesses): Use it instead of checking
22641 for a pair of assignments.
22642 (vect_store_lanes_supported): Take a masked_p parameter.
22643 (vect_load_lanes_supported): Likewise.
22644 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
22645 vect_store_lanes_supported and vect_load_lanes_supported.
22646 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
22647 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
22648 parameter. Don't allow gaps for masked accesses.
22649 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
22650 and vect_load_lanes_supported.
22651 (get_load_store_type): Take a masked_p parameter and update
22652 call to get_group_load_store_type.
22653 (vectorizable_store): Update call to get_load_store_type.
22654 Handle IFN_MASK_STORE_LANES.
22655 (vectorizable_load): Update call to get_load_store_type.
22656 Handle IFN_MASK_LOAD_LANES.
22658 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22659 Alan Hayward <alan.hayward@arm.com>
22660 David Sherwood <david.sherwood@arm.com>
22662 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
22664 * config/aarch64/aarch64-protos.h
22665 (aarch64_sve_struct_memory_operand_p): Declare.
22666 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
22667 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
22668 (VPRED, vpred): Handle SVE structure modes.
22669 * config/aarch64/constraints.md (Utx): New constraint.
22670 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
22671 (aarch64_sve_struct_nonimmediate_operand): New predicates.
22672 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
22673 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
22674 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
22675 structure modes. Split into pieces after RA.
22676 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
22677 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
22679 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
22680 SVE structure modes.
22681 (aarch64_classify_address): Likewise.
22682 (sizetochar): Move earlier in file.
22683 (aarch64_print_operand): Handle SVE register lists.
22684 (aarch64_array_mode): New function.
22685 (aarch64_sve_struct_memory_operand_p): Likewise.
22686 (TARGET_ARRAY_MODE): Redefine.
22688 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22689 Alan Hayward <alan.hayward@arm.com>
22690 David Sherwood <david.sherwood@arm.com>
22692 * target.def (array_mode): New target hook.
22693 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
22694 * doc/tm.texi: Regenerate.
22695 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
22696 * hooks.c (hook_optmode_mode_uhwi_none): New function.
22697 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
22698 targetm.array_mode.
22699 * stor-layout.c (mode_for_array): Likewise. Support polynomial
22702 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22703 Alan Hayward <alan.hayward@arm.com>
22704 David Sherwood <david.sherwood@arm.com>
22706 * fold-const.c (fold_binary_loc): Check the argument types
22707 rather than the result type when testing for a vector operation.
22709 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22711 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
22712 * doc/tm.texi: Regenerate.
22714 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22715 Alan Hayward <alan.hayward@arm.com>
22716 David Sherwood <david.sherwood@arm.com>
22718 * doc/invoke.texi (-msve-vector-bits=): Document new option.
22719 (sve): Document new AArch64 extension.
22720 * doc/md.texi (w): Extend the description of the AArch64
22721 constraint to include SVE vectors.
22722 (Upl, Upa): Document new AArch64 predicate constraints.
22723 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
22725 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
22726 (msve-vector-bits=): New option.
22727 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
22728 SVE when these are disabled.
22729 (sve): New extension.
22730 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
22731 modes. Adjust their number of units based on aarch64_sve_vg.
22732 (MAX_BITSIZE_MODE_ANY_MODE): Define.
22733 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
22734 aarch64_addr_query_type.
22735 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
22736 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
22737 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
22738 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
22739 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
22740 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
22741 (aarch64_simd_imm_zero_p): Delete.
22742 (aarch64_check_zero_based_sve_index_immediate): Declare.
22743 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22744 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22745 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22746 (aarch64_sve_float_mul_immediate_p): Likewise.
22747 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22748 rather than an rtx.
22749 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
22750 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
22751 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
22752 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
22753 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
22754 (aarch64_regmode_natural_size): Likewise.
22755 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
22756 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
22758 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
22759 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
22760 for VG and the SVE predicate registers.
22761 (V_ALIASES): Add a "z"-prefixed alias.
22762 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
22763 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
22764 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
22765 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
22766 (REG_CLASS_NAMES): Add entries for them.
22767 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
22768 and the predicate registers.
22769 (aarch64_sve_vg): Declare.
22770 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
22771 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
22772 (REGMODE_NATURAL_SIZE): Define.
22773 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
22775 * config/aarch64/aarch64.c: Include cfgrtl.h.
22776 (simd_immediate_info): Add a constructor for series vectors,
22777 and an associated step field.
22778 (aarch64_sve_vg): New variable.
22779 (aarch64_dbx_register_number): Handle VG and the predicate registers.
22780 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
22781 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
22782 (VEC_ANY_DATA, VEC_STRUCT): New constants.
22783 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
22784 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
22785 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
22786 (aarch64_get_mask_mode): New functions.
22787 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
22788 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
22789 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
22790 predicate modes and predicate registers. Explicitly restrict
22791 GPRs to modes of 16 bytes or smaller. Only allow FP registers
22792 to store a vector mode if it is recognized by
22793 aarch64_classify_vector_mode.
22794 (aarch64_regmode_natural_size): New function.
22795 (aarch64_hard_regno_caller_save_mode): Return the original mode
22797 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
22798 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
22799 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
22800 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
22802 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
22803 does not overlap dest if the function is frame-related. Handle
22805 (aarch64_split_add_offset): New function.
22806 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
22807 them aarch64_add_offset.
22808 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
22809 and update call to aarch64_sub_sp.
22810 (aarch64_add_cfa_expression): New function.
22811 (aarch64_expand_prologue): Pass extra temporary registers to the
22812 functions above. Handle the case in which we need to emit new
22813 DW_CFA_expressions for registers that were originally saved
22814 relative to the stack pointer, but now have to be expressed
22815 relative to the frame pointer.
22816 (aarch64_output_mi_thunk): Pass extra temporary registers to the
22818 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
22819 IP0 and IP1 values for SVE frames.
22820 (aarch64_expand_vec_series): New function.
22821 (aarch64_expand_sve_widened_duplicate): Likewise.
22822 (aarch64_expand_sve_const_vector): Likewise.
22823 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
22824 Handle SVE constants. Use emit_move_insn to move a force_const_mem
22825 into the register, rather than emitting a SET directly.
22826 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
22827 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
22828 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
22829 (offset_9bit_signed_scaled_p): New functions.
22830 (aarch64_replicate_bitmask_imm): New function.
22831 (aarch64_bitmask_imm): Use it.
22832 (aarch64_cannot_force_const_mem): Reject expressions involving
22833 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
22834 (aarch64_classify_index): Handle SVE indices, by requiring
22835 a plain register index with a scale that matches the element size.
22836 (aarch64_classify_address): Handle SVE addresses. Assert that
22837 the mode of the address is VOIDmode or an integer mode.
22838 Update call to aarch64_classify_symbol.
22839 (aarch64_classify_symbolic_expression): Update call to
22840 aarch64_classify_symbol.
22841 (aarch64_const_vec_all_in_range_p): New function.
22842 (aarch64_print_vector_float_operand): Likewise.
22843 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
22844 "vN" for FP registers with SVE modes. Handle (const ...) vectors
22845 and the FP immediates 1.0 and 0.5.
22846 (aarch64_print_address_internal): Handle SVE addresses.
22847 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
22848 (aarch64_regno_regclass): Handle predicate registers.
22849 (aarch64_secondary_reload): Handle big-endian reloads of SVE
22851 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
22852 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
22853 (aarch64_convert_sve_vector_bits): New function.
22854 (aarch64_override_options): Use it to handle -msve-vector-bits=.
22855 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22856 rather than an rtx.
22857 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
22858 Handle SVE vector and predicate modes. Accept VL-based constants
22859 that need only one temporary register, and VL offsets that require
22860 no temporary registers.
22861 (aarch64_conditional_register_usage): Mark the predicate registers
22862 as fixed if SVE isn't available.
22863 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
22864 Return true for SVE vector and predicate modes.
22865 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
22866 rather than an unsigned int. Handle SVE modes.
22867 (aarch64_preferred_simd_mode): Update call accordingly. Handle
22869 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
22871 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22872 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22873 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22874 (aarch64_sve_float_mul_immediate_p): New functions.
22875 (aarch64_sve_valid_immediate): New function.
22876 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
22877 Explicitly reject structure modes. Check for INDEX constants.
22878 Handle PTRUE and PFALSE constants.
22879 (aarch64_check_zero_based_sve_index_immediate): New function.
22880 (aarch64_simd_imm_zero_p): Delete.
22881 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
22882 vector modes. Accept constants in the range of CNT[BHWD].
22883 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
22884 ask for an Advanced SIMD mode.
22885 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
22886 (aarch64_simd_vector_alignment): Handle SVE predicates.
22887 (aarch64_vectorize_preferred_vector_alignment): New function.
22888 (aarch64_simd_vector_alignment_reachable): Use it instead of
22890 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
22891 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
22893 (MAX_VECT_LEN): Delete.
22894 (expand_vec_perm_d): Add a vec_flags field.
22895 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
22896 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
22897 (aarch64_evpc_ext): Don't apply a big-endian lane correction
22899 (aarch64_evpc_rev): Rename to...
22900 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
22901 (aarch64_evpc_rev_global): New function.
22902 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
22903 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
22905 (aarch64_evpc_sve_tbl): New function.
22906 (aarch64_expand_vec_perm_const_1): Update after rename of
22907 aarch64_evpc_rev. Handle SVE permutes too, trying
22908 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
22909 than aarch64_evpc_tbl.
22910 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
22911 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
22912 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
22913 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
22914 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
22915 (aarch64_expand_sve_vcond): New functions.
22916 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
22917 of aarch64_vector_mode_p.
22918 (aarch64_dwarf_poly_indeterminate_value): New function.
22919 (aarch64_compute_pressure_classes): Likewise.
22920 (aarch64_can_change_mode_class): Likewise.
22921 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
22922 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
22923 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
22924 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
22925 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
22926 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
22927 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
22928 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
22930 (Dn, Dl, Dr): Accept const as well as const_vector.
22931 (Dz): Likewise. Compare against CONST0_RTX.
22932 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
22933 of "vector" where appropriate.
22934 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
22935 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
22936 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
22937 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
22938 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
22939 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
22940 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
22941 (v_int_equiv): Extend to SVE modes.
22942 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
22944 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
22945 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
22946 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
22947 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
22948 (SVE_COND_FP_CMP): New int iterators.
22949 (perm_hilo): Handle the new unpack unspecs.
22950 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
22952 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
22953 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
22954 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
22955 (aarch64_equality_operator, aarch64_constant_vector_operand)
22956 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
22957 (aarch64_sve_nonimmediate_operand): Likewise.
22958 (aarch64_sve_general_operand): Likewise.
22959 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
22960 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
22961 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
22962 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
22963 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
22964 (aarch64_sve_float_arith_immediate): Likewise.
22965 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
22966 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
22967 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
22968 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
22969 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
22970 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
22971 (aarch64_sve_float_arith_operand): Likewise.
22972 (aarch64_sve_float_arith_with_sub_operand): Likewise.
22973 (aarch64_sve_float_mul_operand): Likewise.
22974 (aarch64_sve_vec_perm_operand): Likewise.
22975 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
22976 (aarch64_mov_operand): Accept const_poly_int and const_vector.
22977 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
22978 as well as const_vector.
22979 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
22980 in file. Use CONST0_RTX and CONSTM1_RTX.
22981 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
22982 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
22983 Use aarch64_simd_imm_zero.
22984 * config/aarch64/aarch64-sve.md: New file.
22985 * config/aarch64/aarch64.md: Include it.
22986 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
22987 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
22988 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
22989 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
22990 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
22991 (sve): New attribute.
22992 (enabled): Disable instructions with the sve attribute unless
22994 (movqi, movhi): Pass CONST_POLY_INT operaneds through
22995 aarch64_expand_mov_immediate.
22996 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
22997 CNT[BHSD] immediates.
22998 (movti): Split CONST_POLY_INT moves into two halves.
22999 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
23000 Split additions that need a temporary here if the destination
23001 is the stack pointer.
23002 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
23003 (*add<mode>3_poly_1): New instruction.
23004 (set_clobber_cc): New expander.
23006 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23008 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
23009 parameter and use it instead of GET_MODE_SIZE (innermode). Use
23010 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
23011 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
23012 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
23013 Change innermode from fixed_mode_size to machine_mode.
23014 (simplify_subreg): Update call accordingly. Handle a constant-sized
23015 subreg of a variable-length CONST_VECTOR.
23017 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23018 Alan Hayward <alan.hayward@arm.com>
23019 David Sherwood <david.sherwood@arm.com>
23021 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
23022 (add_offset_to_base): New function, split out from...
23023 (create_mem_ref): ...here. When handling a scale other than 1,
23024 check first whether the address is valid without the offset.
23025 Add it into the base if so, leaving the index and scale as-is.
23027 2018-01-12 Jakub Jelinek <jakub@redhat.com>
23030 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
23031 fold_for_warn before checking if arg2 is INTEGER_CST.
23033 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
23035 * config/rs6000/predicates.md (load_multiple_operation): Delete.
23036 (store_multiple_operation): Delete.
23037 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
23038 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
23039 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
23040 guarded by TARGET_STRING.
23041 (rs6000_output_load_multiple): Delete.
23042 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23043 OPTION_MASK_STRING / TARGET_STRING handling.
23044 (print_operand) <'N', 'O'>: Add comment that these are unused now.
23045 (const rs6000_opt_masks) <"string">: Change mask to 0.
23046 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
23047 (MASK_STRING): Delete.
23048 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
23050 (load_multiple): Delete.
23057 (store_multiple): Delete.
23064 (movmemsi_8reg): Delete.
23065 (corresponding unnamed define_insn): Delete.
23066 (movmemsi_6reg): Delete.
23067 (corresponding unnamed define_insn): Delete.
23068 (movmemsi_4reg): Delete.
23069 (corresponding unnamed define_insn): Delete.
23070 (movmemsi_2reg): Delete.
23071 (corresponding unnamed define_insn): Delete.
23072 (movmemsi_1reg): Delete.
23073 (corresponding unnamed define_insn): Delete.
23074 * config/rs6000/rs6000.opt (mno-string): New.
23075 (mstring): Replace by deprecation warning stub.
23076 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
23078 2018-01-12 Jakub Jelinek <jakub@redhat.com>
23080 * regrename.c (regrename_do_replace): If replacing the same
23081 reg multiple times, try to reuse last created gen_raw_REG.
23084 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
23085 main to workaround a bug in GDB.
23087 2018-01-12 Tom de Vries <tom@codesourcery.com>
23090 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
23092 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
23094 PR rtl-optimization/80481
23095 * ira-color.c (get_cap_member): New function.
23096 (allocnos_conflict_by_live_ranges_p): Use it.
23097 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
23098 (setup_slot_coalesced_allocno_live_ranges): Ditto.
23100 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
23103 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
23104 (*saddl_se_1): Ditto.
23105 (*ssubsi_1): Ditto.
23106 (*ssubl_se_1): Ditto.
23108 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
23110 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
23111 rather than wi::to_widest for DR_INITs.
23112 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
23113 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
23114 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
23116 (vect_analyze_group_access_1): Note that here.
23118 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
23120 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
23121 polynomial type sizes.
23123 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
23125 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
23126 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
23127 (gimple_add_tmp_var): Likewise.
23129 2018-01-12 Martin Liska <mliska@suse.cz>
23131 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
23132 (gimple_alloc_sizes): Likewise.
23133 (dump_gimple_statistics): Use PRIu64 in printf format.
23134 * gimple.h: Change uint64_t to int.
23136 2018-01-12 Martin Liska <mliska@suse.cz>
23138 * tree-core.h: Use uint64_t instead of int.
23139 * tree.c (tree_node_counts): Likewise.
23140 (tree_node_sizes): Likewise.
23141 (dump_tree_statistics): Use PRIu64 in printf format.
23143 2018-01-12 Martin Liska <mliska@suse.cz>
23145 * Makefile.in: As qsort_chk is implemented in vec.c, add
23146 vec.o to linkage of gencfn-macros.
23147 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
23148 passing the info to record_node_allocation_statistics.
23149 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
23151 * ggc-common.c (struct ggc_usage): Add operator== and use
23152 it in operator< and compare function.
23153 * mem-stats.h (struct mem_usage): Likewise.
23154 * vec.c (struct vec_usage): Remove operator< and compare
23155 function. Can be simply inherited.
23157 2018-01-12 Martin Jambor <mjambor@suse.cz>
23160 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
23161 * tree-ssa-math-opts.c: Include domwalk.h.
23162 (convert_mult_to_fma_1): New function.
23163 (fma_transformation_info): New type.
23164 (fma_deferring_state): Likewise.
23165 (cancel_fma_deferring): New function.
23166 (result_of_phi): Likewise.
23167 (last_fma_candidate_feeds_initial_phi): Likewise.
23168 (convert_mult_to_fma): Added deferring logic, split actual
23169 transformation to convert_mult_to_fma_1.
23170 (math_opts_dom_walker): New type.
23171 (math_opts_dom_walker::after_dom_children): New method, body moved
23172 here from pass_optimize_widening_mul::execute, added deferring logic
23174 (pass_optimize_widening_mul::execute): Moved most of code to
23175 math_opts_dom_walker::after_dom_children.
23176 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
23177 * config/i386/i386.c (ix86_option_override_internal): Added
23178 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
23180 2018-01-12 Richard Biener <rguenther@suse.de>
23183 * dwarf2out.c (gen_variable_die): Do not reset old_die for
23184 inline instance vars.
23186 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
23189 * config/rx/rx.c (rx_is_restricted_memory_address):
23190 Handle SUBREG case.
23192 2018-01-12 Richard Biener <rguenther@suse.de>
23194 PR tree-optimization/80846
23195 * target.def (split_reduction): New target hook.
23196 * targhooks.c (default_split_reduction): New function.
23197 * targhooks.h (default_split_reduction): Declare.
23198 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
23199 target requests first reduce vectors by combining low and high
23201 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
23202 (get_vectype_for_scalar_type_and_size): Export.
23203 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
23204 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
23205 * doc/tm.texi: Regenerate.
23206 * config/i386/i386.c (ix86_split_reduction): Implement
23207 TARGET_VECTORIZE_SPLIT_REDUCTION.
23209 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23212 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
23213 in PIC mode except for TARGET_VXWORKS_RTP.
23214 * config/sparc/sparc.c: Include cfgrtl.h.
23215 (TARGET_INIT_PIC_REG): Define.
23216 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
23217 (sparc_pic_register_p): New predicate.
23218 (sparc_legitimate_address_p): Use it.
23219 (sparc_legitimize_pic_address): Likewise.
23220 (sparc_delegitimize_address): Likewise.
23221 (sparc_mode_dependent_address_p): Likewise.
23222 (gen_load_pcrel_sym): Remove 4th parameter.
23223 (load_got_register): Adjust call to above. Remove obsolete stuff.
23224 (sparc_expand_prologue): Do not call load_got_register here.
23225 (sparc_flat_expand_prologue): Likewise.
23226 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
23227 (sparc_use_pseudo_pic_reg): New function.
23228 (sparc_init_pic_reg): Likewise.
23229 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
23230 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
23232 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
23234 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23235 Add item for branch_cost.
23237 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23239 PR rtl-optimization/83565
23240 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
23241 not extend the result to a larger mode for rotate operations.
23242 (num_sign_bit_copies1): Likewise.
23244 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23247 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
23249 Use values-Xc.o for -pedantic.
23250 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
23252 2018-01-12 Martin Liska <mliska@suse.cz>
23255 * ipa-devirt.c (final_warning_record::grow_type_warnings):
23257 (possible_polymorphic_call_targets): Use it.
23258 (ipa_devirt): Likewise.
23260 2018-01-12 Martin Liska <mliska@suse.cz>
23262 * profile-count.h (enum profile_quality): Use 0 as invalid
23263 enum value of profile_quality.
23265 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
23267 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
23268 -mext-string options.
23270 2018-01-12 Richard Biener <rguenther@suse.de>
23272 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
23273 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
23274 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
23276 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
23278 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23280 * configure.ac (--with-long-double-format): Add support for the
23281 configuration option to change the default long double format on
23283 * config.gcc (powerpc*-linux*-*): Likewise.
23284 * configure: Regenerate.
23285 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
23286 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
23287 used without modification.
23289 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23291 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
23292 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
23293 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
23294 MISC_BUILTIN_SPEC_BARRIER.
23295 (rs6000_init_builtins): Likewise.
23296 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
23298 (speculation_barrier): New define_insn.
23299 * doc/extend.texi: Document __builtin_speculation_barrier.
23301 2018-01-11 Jakub Jelinek <jakub@redhat.com>
23304 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
23305 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
23306 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
23308 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
23309 integral modes instead of "ss" and "sd".
23310 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
23311 vectors with 32-bit and 64-bit elements.
23312 (vecdupssescalarmodesuffix): New mode attribute.
23313 (vec_dup<mode>): Use it.
23315 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
23318 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
23319 frame if argument is passed on stack.
23321 2018-01-11 Jakub Jelinek <jakub@redhat.com>
23324 * ree.c (combine_reaching_defs): Optimize also
23325 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
23326 reg2=any_extend(exp); reg1=reg2;, formatting fix.
23328 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
23330 PR middle-end/83189
23331 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
23333 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
23335 PR middle-end/83718
23336 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
23337 after they are computed.
23339 2018-01-11 Bin Cheng <bin.cheng@arm.com>
23341 PR tree-optimization/83695
23342 * gimple-loop-linterchange.cc
23343 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
23344 reset cached scev information after interchange.
23345 (pass_linterchange::execute): Remove call to scev_reset_htab.
23347 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23349 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
23350 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
23351 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
23352 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
23353 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
23354 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
23355 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
23356 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
23357 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
23358 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
23359 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
23360 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
23361 (V_lane_reg): Likewise.
23362 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
23364 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
23365 (vfmal_lane_low<mode>_intrinsic,
23366 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
23367 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
23368 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
23369 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
23370 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
23371 vfmsl_lane_high<mode>_intrinsic): New define_insns.
23373 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23375 * config/arm/arm-cpus.in (fp16fml): New feature.
23376 (ALL_SIMD): Add fp16fml.
23377 (armv8.2-a): Add fp16fml as an option.
23378 (armv8.3-a): Likewise.
23379 (armv8.4-a): Add fp16fml as part of fp16.
23380 * config/arm/arm.h (TARGET_FP16FML): Define.
23381 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
23383 * config/arm/arm-modes.def (V2HF): Define.
23384 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
23385 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
23386 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
23387 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
23388 vfmsl_low, vfmsl_high): New set of builtins.
23389 * config/arm/iterators.md (PLUSMINUS): New code iterator.
23390 (vfml_op): New code attribute.
23391 (VFMLHALVES): New int iterator.
23392 (VFML, VFMLSEL): New mode attributes.
23393 (V_reg): Define mapping for V2HF.
23394 (V_hi, V_lo): New mode attributes.
23395 (VF_constraint): Likewise.
23396 (vfml_half, vfml_half_selector): New int attributes.
23397 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
23399 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
23400 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
23402 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
23403 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
23404 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
23405 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
23407 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
23408 Document new effective target and option set.
23410 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23412 * config/arm/arm-cpus.in (armv8_4): New feature.
23413 (ARMv8_4a): New fgroup.
23414 (armv8.4-a): New arch.
23415 * config/arm/arm-tables.opt: Regenerate.
23416 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
23417 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
23418 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
23419 Add matching rules for -march=armv8.4-a and extensions.
23420 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
23422 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
23425 * config/rx/rx.md (BW): New mode attribute.
23426 (sync_lock_test_and_setsi): Add mode suffix to insn output.
23428 2018-01-11 Richard Biener <rguenther@suse.de>
23430 PR tree-optimization/83435
23431 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
23432 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
23433 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
23435 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23436 Alan Hayward <alan.hayward@arm.com>
23437 David Sherwood <david.sherwood@arm.com>
23439 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
23441 (aarch64_classify_address): Initialize it. Track polynomial offsets.
23442 (aarch64_print_address_internal): Use it to check for a zero offset.
23444 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23445 Alan Hayward <alan.hayward@arm.com>
23446 David Sherwood <david.sherwood@arm.com>
23448 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
23449 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
23450 Return a poly_int64 rather than a HOST_WIDE_INT.
23451 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
23452 rather than a HOST_WIDE_INT.
23453 * config/aarch64/aarch64.h (aarch64_frame): Protect with
23454 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
23455 hard_fp_offset, frame_size, initial_adjust, callee_offset and
23456 final_offset from HOST_WIDE_INT to poly_int64.
23457 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
23458 to_constant when getting the number of units in an Advanced SIMD
23460 (aarch64_builtin_vectorized_function): Check for a constant number
23462 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
23464 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
23465 attribute instead of GET_MODE_NUNITS.
23466 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
23467 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
23468 GET_MODE_SIZE for fixed-size registers.
23469 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
23470 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
23471 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
23472 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
23473 (aarch64_print_operand, aarch64_print_address_internal)
23474 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
23475 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
23476 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
23477 Handle polynomial GET_MODE_SIZE.
23478 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
23479 wider than SImode without modification.
23480 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
23481 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
23482 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
23483 passing and returning SVE modes.
23484 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
23485 rather than GEN_INT.
23486 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
23487 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
23488 (aarch64_allocate_and_probe_stack_space): Likewise.
23489 (aarch64_layout_frame): Cope with polynomial offsets.
23490 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
23491 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
23492 polynomial offsets.
23493 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
23494 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
23495 poly_int64 rather than a HOST_WIDE_INT.
23496 (aarch64_get_separate_components, aarch64_process_components)
23497 (aarch64_expand_prologue, aarch64_expand_epilogue)
23498 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
23499 (aarch64_anchor_offset): New function, split out from...
23500 (aarch64_legitimize_address): ...here.
23501 (aarch64_builtin_vectorization_cost): Handle polynomial
23502 TYPE_VECTOR_SUBPARTS.
23503 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
23505 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
23506 number of elements from the PARALLEL rather than the mode.
23507 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
23508 rather than GET_MODE_BITSIZE.
23509 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
23510 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
23511 (aarch64_expand_vec_perm_const_1): Handle polynomial
23512 d->perm.length () and d->perm elements.
23513 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
23514 Apply to_constant to d->perm elements.
23515 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
23516 polynomial CONST_VECTOR_NUNITS.
23517 (aarch64_move_pointer): Take amount as a poly_int64 rather
23519 (aarch64_progress_pointer): Avoid temporary variable.
23520 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
23521 the mode attribute instead of GET_MODE.
23523 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23524 Alan Hayward <alan.hayward@arm.com>
23525 David Sherwood <david.sherwood@arm.com>
23527 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
23528 x exists before using it.
23529 (aarch64_add_constant_internal): Rename to...
23530 (aarch64_add_offset_1): ...this. Replace regnum with separate
23531 src and dest rtxes. Handle the case in which they're different,
23532 including when the offset is zero. Replace scratchreg with an rtx.
23533 Use 2 additions if there is no spare register into which we can
23534 move a 16-bit constant.
23535 (aarch64_add_constant): Delete.
23536 (aarch64_add_offset): Replace reg with separate src and dest
23537 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
23538 Use aarch64_add_offset_1.
23539 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
23540 an rtx rather than an int. Take the delta as a poly_int64
23541 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
23542 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
23543 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
23544 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
23545 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
23546 and aarch64_add_sp.
23547 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
23548 aarch64_add_constant.
23550 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23552 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
23553 Use scalar_float_mode.
23555 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23557 * config/aarch64/aarch64-simd.md
23558 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
23559 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
23560 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
23561 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
23562 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
23563 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
23564 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
23565 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
23566 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
23567 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
23569 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23572 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
23573 targ_options->x_arm_arch_string is non NULL.
23575 2018-01-11 Tamar Christina <tamar.christina@arm.com>
23577 * config/aarch64/aarch64.h
23578 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
23580 2018-01-11 Sudakshina Das <sudi.das@arm.com>
23583 * expmed.c (emit_store_flag_force): Swap if const op0
23584 and change VOIDmode to mode of op0.
23586 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23588 PR rtl-optimization/83761
23589 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
23590 than bytes to mode_for_size.
23592 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23594 PR middle-end/83189
23595 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
23596 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
23599 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23601 PR middle-end/83575
23602 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
23603 when in layout mode.
23604 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
23605 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
23608 2018-01-10 Michael Collison <michael.collison@arm.com>
23610 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
23611 * config/aarch64/aarch64-option-extension.def: Add
23612 AARCH64_OPT_EXTENSION of 'fp16fml'.
23613 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23614 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
23615 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
23616 * config/aarch64/constraints.md (Ui7): New constraint.
23617 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
23618 (VFMLA_SEL_W): Ditto.
23621 (VFMLA16_LOW): New int iterator.
23622 (VFMLA16_HIGH): Ditto.
23623 (UNSPEC_FMLAL): New unspec.
23624 (UNSPEC_FMLSL): Ditto.
23625 (UNSPEC_FMLAL2): Ditto.
23626 (UNSPEC_FMLSL2): Ditto.
23627 (f16mac): New code attribute.
23628 * config/aarch64/aarch64-simd-builtins.def
23629 (aarch64_fmlal_lowv2sf): Ditto.
23630 (aarch64_fmlsl_lowv2sf): Ditto.
23631 (aarch64_fmlalq_lowv4sf): Ditto.
23632 (aarch64_fmlslq_lowv4sf): Ditto.
23633 (aarch64_fmlal_highv2sf): Ditto.
23634 (aarch64_fmlsl_highv2sf): Ditto.
23635 (aarch64_fmlalq_highv4sf): Ditto.
23636 (aarch64_fmlslq_highv4sf): Ditto.
23637 (aarch64_fmlal_lane_lowv2sf): Ditto.
23638 (aarch64_fmlsl_lane_lowv2sf): Ditto.
23639 (aarch64_fmlal_laneq_lowv2sf): Ditto.
23640 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
23641 (aarch64_fmlalq_lane_lowv4sf): Ditto.
23642 (aarch64_fmlsl_lane_lowv4sf): Ditto.
23643 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
23644 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
23645 (aarch64_fmlal_lane_highv2sf): Ditto.
23646 (aarch64_fmlsl_lane_highv2sf): Ditto.
23647 (aarch64_fmlal_laneq_highv2sf): Ditto.
23648 (aarch64_fmlsl_laneq_highv2sf): Ditto.
23649 (aarch64_fmlalq_lane_highv4sf): Ditto.
23650 (aarch64_fmlsl_lane_highv4sf): Ditto.
23651 (aarch64_fmlalq_laneq_highv4sf): Ditto.
23652 (aarch64_fmlsl_laneq_highv4sf): Ditto.
23653 * config/aarch64/aarch64-simd.md:
23654 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
23655 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
23656 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
23657 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
23658 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
23659 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
23660 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
23661 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
23662 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
23663 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
23664 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
23665 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
23666 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
23667 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
23668 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
23669 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
23670 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
23671 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
23672 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
23673 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
23674 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
23675 (vfmlsl_low_u32): Ditto.
23676 (vfmlalq_low_u32): Ditto.
23677 (vfmlslq_low_u32): Ditto.
23678 (vfmlal_high_u32): Ditto.
23679 (vfmlsl_high_u32): Ditto.
23680 (vfmlalq_high_u32): Ditto.
23681 (vfmlslq_high_u32): Ditto.
23682 (vfmlal_lane_low_u32): Ditto.
23683 (vfmlsl_lane_low_u32): Ditto.
23684 (vfmlal_laneq_low_u32): Ditto.
23685 (vfmlsl_laneq_low_u32): Ditto.
23686 (vfmlalq_lane_low_u32): Ditto.
23687 (vfmlslq_lane_low_u32): Ditto.
23688 (vfmlalq_laneq_low_u32): Ditto.
23689 (vfmlslq_laneq_low_u32): Ditto.
23690 (vfmlal_lane_high_u32): Ditto.
23691 (vfmlsl_lane_high_u32): Ditto.
23692 (vfmlal_laneq_high_u32): Ditto.
23693 (vfmlsl_laneq_high_u32): Ditto.
23694 (vfmlalq_lane_high_u32): Ditto.
23695 (vfmlslq_lane_high_u32): Ditto.
23696 (vfmlalq_laneq_high_u32): Ditto.
23697 (vfmlslq_laneq_high_u32): Ditto.
23698 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
23699 (AARCH64_FL_FOR_ARCH8_4): New.
23700 (AARCH64_ISA_F16FML): New ISA flag.
23701 (TARGET_F16FML): New feature flag for fp16fml.
23702 (doc/invoke.texi): Document new fp16fml option.
23704 2018-01-10 Michael Collison <michael.collison@arm.com>
23706 * config/aarch64/aarch64-builtins.c:
23707 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
23708 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23709 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
23710 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
23711 (AARCH64_ISA_SHA3): New ISA flag.
23712 (TARGET_SHA3): New feature flag for sha3.
23713 * config/aarch64/iterators.md (sha512_op): New int attribute.
23714 (CRYPTO_SHA512): New int iterator.
23715 (UNSPEC_SHA512H): New unspec.
23716 (UNSPEC_SHA512H2): Ditto.
23717 (UNSPEC_SHA512SU0): Ditto.
23718 (UNSPEC_SHA512SU1): Ditto.
23719 * config/aarch64/aarch64-simd-builtins.def
23720 (aarch64_crypto_sha512hqv2di): New builtin.
23721 (aarch64_crypto_sha512h2qv2di): Ditto.
23722 (aarch64_crypto_sha512su0qv2di): Ditto.
23723 (aarch64_crypto_sha512su1qv2di): Ditto.
23724 (aarch64_eor3qv8hi): Ditto.
23725 (aarch64_rax1qv2di): Ditto.
23726 (aarch64_xarqv2di): Ditto.
23727 (aarch64_bcaxqv8hi): Ditto.
23728 * config/aarch64/aarch64-simd.md:
23729 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
23730 (aarch64_crypto_sha512su0qv2di): Ditto.
23731 (aarch64_crypto_sha512su1qv2di): Ditto.
23732 (aarch64_eor3qv8hi): Ditto.
23733 (aarch64_rax1qv2di): Ditto.
23734 (aarch64_xarqv2di): Ditto.
23735 (aarch64_bcaxqv8hi): Ditto.
23736 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
23737 (vsha512h2q_u64): Ditto.
23738 (vsha512su0q_u64): Ditto.
23739 (vsha512su1q_u64): Ditto.
23740 (veor3q_u16): Ditto.
23741 (vrax1q_u64): Ditto.
23742 (vxarq_u64): Ditto.
23743 (vbcaxq_u16): Ditto.
23744 * config/arm/types.md (crypto_sha512): New type attribute.
23745 (crypto_sha3): Ditto.
23746 (doc/invoke.texi): Document new sha3 option.
23748 2018-01-10 Michael Collison <michael.collison@arm.com>
23750 * config/aarch64/aarch64-builtins.c:
23751 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
23752 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23753 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
23754 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
23755 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
23756 (AARCH64_ISA_SM4): New ISA flag.
23757 (TARGET_SM4): New feature flag for sm4.
23758 * config/aarch64/aarch64-simd-builtins.def
23759 (aarch64_sm3ss1qv4si): Ditto.
23760 (aarch64_sm3tt1aq4si): Ditto.
23761 (aarch64_sm3tt1bq4si): Ditto.
23762 (aarch64_sm3tt2aq4si): Ditto.
23763 (aarch64_sm3tt2bq4si): Ditto.
23764 (aarch64_sm3partw1qv4si): Ditto.
23765 (aarch64_sm3partw2qv4si): Ditto.
23766 (aarch64_sm4eqv4si): Ditto.
23767 (aarch64_sm4ekeyqv4si): Ditto.
23768 * config/aarch64/aarch64-simd.md:
23769 (aarch64_sm3ss1qv4si): Ditto.
23770 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
23771 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
23772 (aarch64_sm4eqv4si): Ditto.
23773 (aarch64_sm4ekeyqv4si): Ditto.
23774 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
23775 (sm3part_op): Ditto.
23776 (CRYPTO_SM3TT): Ditto.
23777 (CRYPTO_SM3PART): Ditto.
23778 (UNSPEC_SM3SS1): New unspec.
23779 (UNSPEC_SM3TT1A): Ditto.
23780 (UNSPEC_SM3TT1B): Ditto.
23781 (UNSPEC_SM3TT2A): Ditto.
23782 (UNSPEC_SM3TT2B): Ditto.
23783 (UNSPEC_SM3PARTW1): Ditto.
23784 (UNSPEC_SM3PARTW2): Ditto.
23785 (UNSPEC_SM4E): Ditto.
23786 (UNSPEC_SM4EKEY): Ditto.
23787 * config/aarch64/constraints.md (Ui2): New constraint.
23788 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
23789 * config/arm/types.md (crypto_sm3): New type attribute.
23790 (crypto_sm4): Ditto.
23791 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
23792 (vsm3tt1aq_u32): Ditto.
23793 (vsm3tt1bq_u32): Ditto.
23794 (vsm3tt2aq_u32): Ditto.
23795 (vsm3tt2bq_u32): Ditto.
23796 (vsm3partw1q_u32): Ditto.
23797 (vsm3partw2q_u32): Ditto.
23798 (vsm4eq_u32): Ditto.
23799 (vsm4ekeyq_u32): Ditto.
23800 (doc/invoke.texi): Document new sm4 option.
23802 2018-01-10 Michael Collison <michael.collison@arm.com>
23804 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
23805 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
23806 (AARCH64_FL_FOR_ARCH8_4): New.
23807 (AARCH64_FL_V8_4): New flag.
23808 (doc/invoke.texi): Document new armv8.4-a option.
23810 2018-01-10 Michael Collison <michael.collison@arm.com>
23812 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23813 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
23814 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
23815 * config/aarch64/aarch64-option-extension.def: Add
23816 AARCH64_OPT_EXTENSION of 'sha2'.
23817 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
23818 (crypto): Disable sha2 and aes if crypto disabled.
23819 (crypto): Enable aes and sha2 if enabled.
23820 (simd): Disable sha2 and aes if simd disabled.
23821 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
23823 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
23824 (TARGET_SHA2): New feature flag for sha2.
23825 (TARGET_AES): New feature flag for aes.
23826 * config/aarch64/aarch64-simd.md:
23827 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
23828 conditional on TARGET_AES.
23829 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
23830 (aarch64_crypto_sha1hsi): Make pattern conditional
23832 (aarch64_crypto_sha1hv4si): Ditto.
23833 (aarch64_be_crypto_sha1hv4si): Ditto.
23834 (aarch64_crypto_sha1su1v4si): Ditto.
23835 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
23836 (aarch64_crypto_sha1su0v4si): Ditto.
23837 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
23838 (aarch64_crypto_sha256su0v4si): Ditto.
23839 (aarch64_crypto_sha256su1v4si): Ditto.
23840 (doc/invoke.texi): Document new aes and sha2 options.
23842 2018-01-10 Martin Sebor <msebor@redhat.com>
23844 PR tree-optimization/83781
23845 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
23848 2018-01-11 Martin Sebor <msebor@gmail.com>
23849 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23851 PR tree-optimization/83501
23852 PR tree-optimization/81703
23854 * tree-ssa-strlen.c (get_string_cst): Rename...
23855 (get_string_len): ...to this. Handle global constants.
23856 (handle_char_store): Adjust.
23858 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
23859 Jim Wilson <jimw@sifive.com>
23861 * config/riscv/riscv-protos.h (riscv_output_return): New.
23862 * config/riscv/riscv.c (struct machine_function): New naked_p field.
23863 (riscv_attribute_table, riscv_output_return),
23864 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
23865 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
23866 (riscv_compute_frame_info): Only compute frame->mask if not a naked
23868 (riscv_expand_prologue): Add early return for naked function.
23869 (riscv_expand_epilogue): Likewise.
23870 (riscv_function_ok_for_sibcall): Return false for naked function.
23871 (riscv_set_current_function): New.
23872 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
23873 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
23874 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
23875 * doc/extend.texi (RISC-V Function Attributes): New.
23877 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
23879 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
23880 check for 128-bit long double before checking TCmode.
23881 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
23882 128-bit long doubles before checking TFmode or TCmode.
23883 (FLOAT128_IBM_P): Likewise.
23885 2018-01-10 Martin Sebor <msebor@redhat.com>
23887 PR tree-optimization/83671
23888 * builtins.c (c_strlen): Unconditionally return zero for the empty
23890 Use -Warray-bounds for warnings.
23891 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
23892 for non-constant array indices with COMPONENT_REF, arrays of
23893 arrays, and pointers to arrays.
23894 (gimple_fold_builtin_strlen): Determine and set length range for
23895 non-constant character arrays.
23897 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
23899 PR middle-end/81897
23900 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
23903 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
23905 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
23907 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23910 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
23911 VECTOR_MEM_ALTIVEC_OR_VSX_P.
23912 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
23913 indexed_or_indirect_operand predicate.
23914 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
23915 (*vsx_le_perm_load_v8hi): Likewise.
23916 (*vsx_le_perm_load_v16qi): Likewise.
23917 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
23918 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
23919 (*vsx_le_perm_store_v8hi): Likewise.
23920 (*vsx_le_perm_store_v16qi): Likewise.
23921 (eight unnamed splitters): Likewise.
23923 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23925 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
23926 * config/rs6000/emmintrin.h: Likewise.
23927 * config/rs6000/mmintrin.h: Likewise.
23928 * config/rs6000/xmmintrin.h: Likewise.
23930 2018-01-10 David Malcolm <dmalcolm@redhat.com>
23933 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
23935 * tree.c (tree_nop_conversion): Return true for location wrapper
23937 (maybe_wrap_with_location): New function.
23938 (selftest::check_strip_nops): New function.
23939 (selftest::test_location_wrappers): New function.
23940 (selftest::tree_c_tests): Call it.
23941 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
23942 (maybe_wrap_with_location): New decl.
23943 (EXPR_LOCATION_WRAPPER_P): New macro.
23944 (location_wrapper_p): New inline function.
23945 (tree_strip_any_location_wrapper): New inline function.
23947 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
23950 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
23951 stack_realign_offset for the largest alignment of stack slot
23953 (ix86_find_max_used_stack_alignment): New function.
23954 (ix86_finalize_stack_frame_flags): Use it. Set
23955 max_used_stack_alignment if we don't realign stack.
23956 * config/i386/i386.h (machine_function): Add
23957 max_used_stack_alignment.
23959 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
23961 * config/arm/arm.opt (-mbranch-cost): New option.
23962 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
23965 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
23968 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
23969 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
23971 2018-01-10 Richard Biener <rguenther@suse.de>
23974 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
23975 early out so it also covers the case where we have a non-NULL
23978 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23980 PR tree-optimization/83753
23981 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
23982 for non-strided grouped accesses if the number of elements is 1.
23984 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23987 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
23988 * i386.h (TARGET_USE_GATHER): Define.
23989 * x86-tune.def (X86_TUNE_USE_GATHER): New.
23991 2018-01-10 Martin Liska <mliska@suse.cz>
23994 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
23995 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
23997 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
23998 CLEANUP_NO_PARTITIONING is not set.
24000 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
24002 * doc/rtl.texi: Remove documentation of (const ...) wrappers
24003 for vectors, as a partial revert of r254296.
24004 * rtl.h (const_vec_p): Delete.
24005 (const_vec_duplicate_p): Don't test for vector CONSTs.
24006 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
24007 * expmed.c (make_tree): Likewise.
24010 * common.md (E, F): Use CONSTANT_P instead of checking for
24012 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
24013 checking for CONST_VECTOR.
24015 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
24017 PR middle-end/83575
24018 * predict.c (force_edge_cold): Handle in more sane way edges
24019 with no prediction.
24021 2018-01-09 Carl Love <cel@us.ibm.com>
24023 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
24025 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
24026 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
24027 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
24028 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
24029 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
24030 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
24031 * config/rs6000/rs6000-protos.h: Add extern defition for
24032 rs6000_generate_float2_double_code.
24033 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
24035 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
24036 (float2_v2df): Add define_expand.
24038 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
24041 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
24042 op_mode in the force_to_mode call.
24044 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
24046 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
24047 instead of checking each element individually.
24048 (aarch64_evpc_uzp): Likewise.
24049 (aarch64_evpc_zip): Likewise.
24050 (aarch64_evpc_ext): Likewise.
24051 (aarch64_evpc_rev): Likewise.
24052 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
24053 instead of checking each element individually. Return true without
24055 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
24056 whether all selected elements come from the same input, instead of
24057 checking each element individually. Remove calls to gen_rtx_REG,
24058 start_sequence and end_sequence and instead assert that no rtl is
24061 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
24063 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
24064 order of HIGH and CONST checks.
24066 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
24068 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
24069 if the destination isn't an SSA_NAME.
24071 2018-01-09 Richard Biener <rguenther@suse.de>
24073 PR tree-optimization/83668
24074 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
24076 (canonicalize_loop_form): ... here, renamed from ...
24077 (canonicalize_loop_closed_ssa_form): ... this and amended to
24078 swap successor edges for loop exit blocks to make us use
24079 the RPO order we need for initial schedule generation.
24081 2018-01-09 Joseph Myers <joseph@codesourcery.com>
24083 PR tree-optimization/64811
24084 * match.pd: When optimizing comparisons with Inf, avoid
24085 introducing or losing exceptions from comparisons with NaN.
24087 2018-01-09 Martin Liska <mliska@suse.cz>
24090 * asan.c (shadow_mem_size): Add gcc_assert.
24092 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
24094 Don't save registers in main().
24097 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
24098 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
24099 * config/avr/avr.c (avr_set_current_function): Don't error if
24100 naked, OS_task or OS_main are specified at the same time.
24101 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
24103 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
24105 * common/config/avr/avr-common.c (avr_option_optimization_table):
24106 Switch on -mmain-is-OS_task for optimizing compilations.
24108 2018-01-09 Richard Biener <rguenther@suse.de>
24110 PR tree-optimization/83572
24111 * graphite.c: Include cfganal.h.
24112 (graphite_transform_loops): Connect infinite loops to exit
24113 and remove fake edges at the end.
24115 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
24117 * ipa-inline.c (edge_badness): Revert accidental checkin.
24119 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
24122 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
24123 symbols; not inline clones.
24125 2018-01-09 Jakub Jelinek <jakub@redhat.com>
24128 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
24129 hard registers. Formatting fixes.
24131 PR preprocessor/83722
24132 * gcc.c (try_generate_repro): Pass
24133 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
24134 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
24137 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
24138 Kito Cheng <kito.cheng@gmail.com>
24140 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
24141 (riscv_leaf_function_p): Delete.
24142 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
24144 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24146 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
24148 (do_ifelse): New function.
24149 (do_isel): New function.
24150 (do_sub3): New function.
24151 (do_add3): New function.
24152 (do_load_mask_compare): New function.
24153 (do_overlap_load_compare): New function.
24154 (expand_compare_loop): New function.
24155 (expand_block_compare): Call expand_compare_loop() when appropriate.
24156 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
24157 option description.
24158 (-mblock-compare-inline-loop-limit): New option.
24160 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24163 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
24164 Reverse order of second and third operands in first alternative.
24165 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
24166 of first and second elements in UNSPEC_VPERMR vector.
24167 (altivec_expand_vec_perm_le): Likewise.
24169 2018-01-08 Jeff Law <law@redhat.com>
24171 PR rtl-optimizatin/81308
24172 * tree-switch-conversion.c (cfg_altered): New file scoped static.
24173 (process_switch): If group_case_labels makes a change, then set
24175 (pass_convert_switch::execute): If a switch is converted, then
24176 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
24178 PR rtl-optimization/81308
24179 * recog.c (split_all_insns): Conditionally cleanup the CFG after
24182 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
24184 PR target/83663 - Revert r255946
24185 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
24186 generation for cases where splatting a value is not useful.
24187 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
24188 across a vec_duplicate and a paradoxical subreg forming a vector
24189 mode to a vec_concat.
24191 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24193 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
24194 -march=armv8.3-a variants.
24195 * config/arm/t-multilib: Likewise.
24196 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
24198 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24200 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
24202 (cceq_ior_compare_complement): Give it a name so I can use it, and
24203 change boolean_or_operator predicate to boolean_operator so it can
24204 be used to generate a crand.
24205 (eqne): New code iterator.
24206 (bd/bd_neg): New code_attrs.
24207 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
24208 a single define_insn.
24209 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
24210 decrement (bdnzt/bdnzf/bdzt/bdzf).
24211 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
24212 with the new names of the branch decrement patterns, and added the
24213 names of the branch decrement conditional patterns.
24215 2018-01-08 Richard Biener <rguenther@suse.de>
24217 PR tree-optimization/83563
24218 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
24221 2018-01-08 Richard Biener <rguenther@suse.de>
24223 PR middle-end/83713
24224 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
24226 2018-01-08 Richard Biener <rguenther@suse.de>
24228 PR tree-optimization/83685
24229 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
24230 references to abnormals.
24232 2018-01-08 Richard Biener <rguenther@suse.de>
24235 * dwarf2out.c (output_indirect_strings): Handle empty
24236 skeleton_debug_str_hash.
24237 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
24239 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
24241 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
24242 (emit_store_direct): Likewise.
24243 (arc_trampoline_adjust_address): Likewise.
24244 (arc_asm_trampoline_template): New function.
24245 (arc_initialize_trampoline): Use asm_trampoline_template.
24246 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
24247 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
24248 * config/arc/arc.md (flush_icache): Delete pattern.
24250 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
24252 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
24253 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
24256 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
24259 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
24260 by not USED_FOR_TARGET.
24261 (make_pass_resolve_sw_modes): Likewise.
24263 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
24265 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
24268 2018-01-08 Richard Biener <rguenther@suse.de>
24270 PR middle-end/83580
24271 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
24273 2018-01-08 Richard Biener <rguenther@suse.de>
24275 PR middle-end/83517
24276 * match.pd ((t * 2) / 2) -> t): Add missing :c.
24278 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
24280 PR middle-end/81897
24281 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
24282 basic blocks with a small number of successors.
24283 (convert_control_dep_chain_into_preds): Improve handling of
24285 (dump_predicates): Split apart into...
24286 (dump_pred_chain): ...here...
24287 (dump_pred_info): ...and here.
24288 (can_one_predicate_be_invalidated_p): Add debugging printfs.
24289 (can_chain_union_be_invalidated_p): Improve check for invalidation
24291 (uninit_uses_cannot_happen): Avoid unnecessary if
24292 convert_control_dep_chain_into_preds yielded nothing.
24294 2018-01-06 Martin Sebor <msebor@redhat.com>
24296 PR tree-optimization/83640
24297 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
24298 subtracting negative offset from size.
24299 (builtin_access::overlap): Adjust offset bounds of the access to fall
24300 within the size of the object if possible.
24302 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
24304 PR rtl-optimization/83699
24305 * expmed.c (extract_bit_field_1): Restrict the vector usage of
24306 extract_bit_field_as_subreg to cases in which the extracted
24307 value is also a vector.
24309 * lra-constraints.c (process_alt_operands): Test for the equivalence
24310 substitutions when detecting a possible reload cycle.
24312 2018-01-06 Jakub Jelinek <jakub@redhat.com>
24315 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
24316 by default if flag_selective_schedling{,2}. Formatting fixes.
24318 PR rtl-optimization/83682
24319 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
24320 if it has non-VECTOR_MODE element mode.
24321 (vec_duplicate_p): Likewise.
24323 PR middle-end/83694
24324 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
24325 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
24327 2018-01-05 Jakub Jelinek <jakub@redhat.com>
24330 * config/i386/i386-builtin.def
24331 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
24332 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
24333 Require also OPTION_MASK_ISA_AVX512F in addition to
24334 OPTION_MASK_ISA_GFNI.
24335 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
24336 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
24337 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
24338 to OPTION_MASK_ISA_GFNI.
24339 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
24340 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
24341 OPTION_MASK_ISA_AVX512BW.
24342 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
24343 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
24344 addition to OPTION_MASK_ISA_GFNI.
24345 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
24346 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
24347 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
24348 to OPTION_MASK_ISA_GFNI.
24349 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
24350 a requirement for all ISAs rather than any of them with a few
24352 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
24354 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
24355 bitmasks to be enabled with 3 exceptions, instead of requiring any
24356 enabled ISA with lots of exceptions.
24357 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
24358 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
24359 Change avx512bw in isa attribute to avx512f.
24360 * config/i386/sgxintrin.h: Add license boilerplate.
24361 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
24362 to __AVX512F__ and __AVX512VL to __AVX512VL__.
24363 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
24364 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
24366 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
24367 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
24368 temporarily sse2 rather than sse if not enabled already.
24371 * config/i386/sse.md (VI248_VLBW): Rename to ...
24372 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
24373 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
24374 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
24375 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
24376 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
24377 mode iterator instead of VI248_VLBW.
24379 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
24381 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
24382 (record_modified): Skip clobbers; add debug output.
24383 (param_change_prob): Use sreal frequencies.
24385 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
24387 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24388 punt for user-aligned variables.
24390 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
24392 * tree-chrec.c (chrec_contains_symbols): Return true for
24395 2018-01-05 Sudakshina Das <sudi.das@arm.com>
24398 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
24399 of (x|y) == x for BICS pattern.
24401 2018-01-05 Jakub Jelinek <jakub@redhat.com>
24403 PR tree-optimization/83605
24404 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
24405 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
24408 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
24410 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
24411 * config/epiphany/rtems.h: New file.
24413 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24414 Uros Bizjak <ubizjak@gmail.com>
24417 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
24418 QIreg_operand instead of register_operand predicate.
24419 * config/i386/i386.c (ix86_rop_should_change_byte_p,
24420 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
24421 comments instead of -fmitigate[-_]rop.
24423 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24426 * cgraphunit.c (symbol_table::compile): Switch to text_section
24427 before calling assembly_start debug hook.
24428 * run-rtl-passes.c (run_rtl_passes): Likewise.
24431 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24433 * tree-vrp.c (extract_range_from_binary_expr_1): Check
24434 range_int_cst_p rather than !symbolic_range_p before calling
24435 extract_range_from_multiplicative_op_1.
24437 2018-01-04 Jeff Law <law@redhat.com>
24439 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
24440 redundant test in assertion.
24442 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24444 * doc/rtl.texi: Document machine_mode wrapper classes.
24446 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24448 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
24449 using tree_to_uhwi.
24451 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24453 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
24454 the VEC_PERM_EXPR fold to fail.
24456 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24459 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
24460 to switched_sections.
24462 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24465 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
24466 test for d.testing.
24468 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
24471 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
24472 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
24474 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24477 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
24478 is BLKmode and bitpos not zero or mode change is needed.
24480 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24483 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
24486 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
24489 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
24490 instead of MULT rtx. Update all corresponding splitters.
24491 (*saddl_se): Ditto.
24492 (*ssub<modesuffix>): Ditto.
24493 (*ssubl_se): Ditto.
24494 (*cmp_sadd_di): Update split patterns.
24495 (*cmp_sadd_si): Ditto.
24496 (*cmp_sadd_sidi): Ditto.
24497 (*cmp_ssub_di): Ditto.
24498 (*cmp_ssub_si): Ditto.
24499 (*cmp_ssub_sidi): Ditto.
24500 * config/alpha/predicates.md (const23_operand): New predicate.
24501 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
24502 Look for ASHIFT, not MULT inner operand.
24503 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
24505 2018-01-04 Martin Liska <mliska@suse.cz>
24507 PR gcov-profile/83669
24508 * gcov.c (output_intermediate_file): Add version to intermediate
24510 * doc/gcov.texi: Document new field 'version' in intermediate
24511 file format. Fix location of '-k' option of gcov command.
24513 2018-01-04 Martin Liska <mliska@suse.cz>
24516 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
24518 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24520 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
24522 2018-01-03 Martin Sebor <msebor@redhat.com>
24524 PR tree-optimization/83655
24525 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
24526 checking calls with invalid arguments.
24528 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24530 * tree-vect-stmts.c (vect_get_store_rhs): New function.
24531 (vectorizable_mask_load_store): Delete.
24532 (vectorizable_call): Return false for masked loads and stores.
24533 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
24534 instead of gimple_assign_rhs1.
24535 (vectorizable_load): Handle IFN_MASK_LOAD.
24536 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
24538 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24540 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
24542 (vectorizable_mask_load_store): ...here.
24543 (vectorizable_load): ...and here.
24545 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24547 * tree-vect-stmts.c (vect_build_all_ones_mask)
24548 (vect_build_zero_merge_argument): New functions, split out from...
24549 (vectorizable_load): ...here.
24551 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24553 * tree-vect-stmts.c (vect_check_store_rhs): New function,
24555 (vectorizable_mask_load_store): ...here.
24556 (vectorizable_store): ...and here.
24558 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24560 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
24562 (vectorizable_mask_load_store): ...here.
24564 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24566 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
24567 (vect_model_store_cost): Take a vec_load_store_type instead of a
24569 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
24570 (vect_model_store_cost): Take a vec_load_store_type instead of a
24572 (vectorizable_mask_load_store): Update accordingly.
24573 (vectorizable_store): Likewise.
24574 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
24576 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24578 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
24579 IFN_MASK_LOAD calls here rather than...
24580 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
24582 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24583 Alan Hayward <alan.hayward@arm.com>
24584 David Sherwood <david.sherwood@arm.com>
24586 * expmed.c (extract_bit_field_1): For vector extracts,
24587 fall back to extract_bit_field_as_subreg if vec_extract
24590 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24591 Alan Hayward <alan.hayward@arm.com>
24592 David Sherwood <david.sherwood@arm.com>
24594 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
24595 they are variable or constant sized.
24596 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
24597 slots for constant-sized data.
24599 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24600 Alan Hayward <alan.hayward@arm.com>
24601 David Sherwood <david.sherwood@arm.com>
24603 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
24604 handling COND_EXPRs with boolean comparisons, try to find a better
24605 basis for the mask type than the boolean itself.
24607 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24609 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
24610 is calculated and how it can be overridden.
24611 * genmodes.c (max_bitsize_mode_any_mode): New variable.
24612 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
24614 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
24617 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24618 Alan Hayward <alan.hayward@arm.com>
24619 David Sherwood <david.sherwood@arm.com>
24621 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
24622 Remove the mode argument.
24623 (aarch64_simd_valid_immediate): Remove the mode and inverse
24625 * config/aarch64/iterators.md (bitsize): New iterator.
24626 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
24627 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
24628 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
24629 aarch64_simd_valid_immediate.
24630 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
24631 (aarch64_reg_or_bic_imm): Likewise.
24632 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
24633 with an insn_type enum and msl with a modifier_type enum.
24634 Replace element_width with a scalar_mode. Change the shift
24635 to unsigned int. Add constructors for scalar_float_mode and
24636 scalar_int_mode elements.
24637 (aarch64_vect_float_const_representable_p): Delete.
24638 (aarch64_can_const_movi_rtx_p)
24639 (aarch64_simd_scalar_immediate_valid_for_move)
24640 (aarch64_simd_make_constant): Update call to
24641 aarch64_simd_valid_immediate.
24642 (aarch64_advsimd_valid_immediate_hs): New function.
24643 (aarch64_advsimd_valid_immediate): Likewise.
24644 (aarch64_simd_valid_immediate): Remove mode and inverse
24645 arguments. Rewrite to use the above. Use const_vec_duplicate_p
24646 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
24647 and aarch64_float_const_representable_p on the result.
24648 (aarch64_output_simd_mov_immediate): Remove mode argument.
24649 Update call to aarch64_simd_valid_immediate and use of
24650 simd_immediate_info.
24651 (aarch64_output_scalar_simd_mov_immediate): Update call
24654 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24655 Alan Hayward <alan.hayward@arm.com>
24656 David Sherwood <david.sherwood@arm.com>
24658 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
24659 (mode_nunits): Likewise CONST_MODE_NUNITS.
24660 * machmode.def (ADJUST_NUNITS): Document.
24661 * genmodes.c (mode_data::need_nunits_adj): New field.
24662 (blank_mode): Update accordingly.
24663 (adj_nunits): New variable.
24664 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
24666 (emit_mode_size_inline): Set need_bytesize_adj for all modes
24667 listed in adj_nunits.
24668 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
24669 listed in adj_nunits. Don't emit case statements for such modes.
24670 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
24671 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
24672 nothing if adj_nunits is nonnull.
24673 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
24674 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
24675 (emit_mode_fbit): Update use of print_maybe_const_decl.
24676 (emit_move_size): Likewise. Treat the array as non-const
24678 (emit_mode_adjustments): Handle adj_nunits.
24680 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24682 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
24683 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
24684 (VECTOR_MODES): Use it.
24685 (make_vector_modes): Take the prefix as an argument.
24687 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24688 Alan Hayward <alan.hayward@arm.com>
24689 David Sherwood <david.sherwood@arm.com>
24691 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
24692 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
24693 for MODE_VECTOR_BOOL.
24694 * machmode.def (VECTOR_BOOL_MODE): Document.
24695 * genmodes.c (VECTOR_BOOL_MODE): New macro.
24696 (make_vector_bool_mode): New function.
24697 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
24699 * lto-streamer-in.c (lto_input_mode_table): Likewise.
24700 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
24702 * stor-layout.c (int_mode_for_mode): Likewise.
24703 * tree.c (build_vector_type_for_mode): Likewise.
24704 * varasm.c (output_constant_pool_2): Likewise.
24705 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
24706 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
24707 for MODE_VECTOR_BOOL.
24708 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
24709 of mode class checks.
24710 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
24711 instead of a list of mode class checks.
24712 (expand_vector_scalar_condition): Likewise.
24713 (type_for_widest_vector_mode): Handle BImode as an inner mode.
24715 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24716 Alan Hayward <alan.hayward@arm.com>
24717 David Sherwood <david.sherwood@arm.com>
24719 * machmode.h (mode_size): Change from unsigned short to
24721 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
24722 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24723 or if measurement_type is not polynomial.
24724 (fixed_size_mode::includes_p): Check for constant-sized modes.
24725 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
24726 return a poly_uint16 rather than an unsigned short.
24727 (emit_mode_size): Change the type of mode_size from unsigned short
24728 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
24729 (emit_mode_adjustments): Cope with polynomial vector sizes.
24730 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24732 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24734 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
24735 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
24736 * caller-save.c (setup_save_areas): Likewise.
24737 (replace_reg_with_saved_mem): Likewise.
24738 * calls.c (emit_library_call_value_1): Likewise.
24739 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
24740 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
24741 (gen_lowpart_for_combine): Likewise.
24742 * convert.c (convert_to_integer_1): Likewise.
24743 * cse.c (equiv_constant, cse_insn): Likewise.
24744 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
24745 (cselib_subst_to_values): Likewise.
24746 * dce.c (word_dce_process_block): Likewise.
24747 * df-problems.c (df_word_lr_mark_ref): Likewise.
24748 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
24749 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
24750 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
24751 (rtl_for_decl_location): Likewise.
24752 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
24753 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
24754 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
24755 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
24756 (expand_expr_real_1): Likewise.
24757 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
24758 (pad_below): Likewise.
24759 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24760 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
24761 * ira.c (get_subreg_tracking_sizes): Likewise.
24762 * ira-build.c (ira_create_allocno_objects): Likewise.
24763 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
24764 (ira_sort_regnos_for_alter_reg): Likewise.
24765 * ira-costs.c (record_operand_costs): Likewise.
24766 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
24767 (resolve_simple_move): Likewise.
24768 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
24769 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
24770 (lra_constraints): Likewise.
24771 (CONST_POOL_OK_P): Reject variable-sized modes.
24772 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
24773 (add_pseudo_to_slot, lra_spill): Likewise.
24774 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24775 * optabs-query.c (get_best_extraction_insn): Likewise.
24776 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24777 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
24778 (expand_mult_highpart, valid_multiword_target_p): Likewise.
24779 * recog.c (offsettable_address_addr_space_p): Likewise.
24780 * regcprop.c (maybe_mode_change): Likewise.
24781 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
24782 * regrename.c (build_def_use): Likewise.
24783 * regstat.c (dump_reg_info): Likewise.
24784 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
24785 (find_reloads, find_reloads_subreg_address): Likewise.
24786 * reload1.c (eliminate_regs_1): Likewise.
24787 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
24788 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
24789 (simplify_binary_operation_1, simplify_subreg): Likewise.
24790 * targhooks.c (default_function_arg_padding): Likewise.
24791 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
24792 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
24793 (verify_gimple_assign_ternary): Likewise.
24794 * tree-inline.c (estimate_move_cost): Likewise.
24795 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24796 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
24797 (get_address_cost_ainc): Likewise.
24798 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24799 (vect_supportable_dr_alignment): Likewise.
24800 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24801 (vectorizable_reduction): Likewise.
24802 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
24803 (vectorizable_operation, vectorizable_load): Likewise.
24804 * tree.c (build_same_sized_truth_vector_type): Likewise.
24805 * valtrack.c (cleanup_auto_inc_dec): Likewise.
24806 * var-tracking.c (emit_note_insn_var_location): Likewise.
24807 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
24808 (ADDR_VEC_ALIGN): Likewise.
24810 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24811 Alan Hayward <alan.hayward@arm.com>
24812 David Sherwood <david.sherwood@arm.com>
24814 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
24816 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24817 or if measurement_type is polynomial.
24818 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
24819 * combine.c (make_extraction): Likewise.
24820 * dse.c (find_shift_sequence): Likewise.
24821 * dwarf2out.c (mem_loc_descriptor): Likewise.
24822 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
24823 (extract_bit_field, extract_low_bits): Likewise.
24824 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
24825 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
24826 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
24827 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
24828 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24829 * reload.c (find_reloads): Likewise.
24830 * reload1.c (alter_reg): Likewise.
24831 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
24832 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
24833 * tree-if-conv.c (predicate_mem_writes): Likewise.
24834 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
24835 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
24836 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
24837 * valtrack.c (dead_debug_insert_temp): Likewise.
24838 * varasm.c (mergeable_constant_section): Likewise.
24839 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
24841 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24842 Alan Hayward <alan.hayward@arm.com>
24843 David Sherwood <david.sherwood@arm.com>
24845 * expr.c (expand_assignment): Cope with polynomial mode sizes
24846 when assigning to a CONCAT.
24848 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24849 Alan Hayward <alan.hayward@arm.com>
24850 David Sherwood <david.sherwood@arm.com>
24852 * machmode.h (mode_precision): Change from unsigned short to
24854 (mode_to_precision): Return a poly_uint16 rather than an unsigned
24856 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
24857 or if measurement_type is not polynomial.
24858 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
24859 in which the mode is already known to be a scalar_int_mode.
24860 * genmodes.c (emit_mode_precision): Change the type of mode_precision
24861 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
24863 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24864 for GET_MODE_PRECISION.
24865 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24866 for GET_MODE_PRECISION.
24867 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
24869 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
24870 (expand_field_assignment, make_extraction): Likewise.
24871 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
24872 (get_last_value): Likewise.
24873 * convert.c (convert_to_integer_1): Likewise.
24874 * cse.c (cse_insn): Likewise.
24875 * expr.c (expand_expr_real_1): Likewise.
24876 * lra-constraints.c (simplify_operand_subreg): Likewise.
24877 * optabs-query.c (can_atomic_load_p): Likewise.
24878 * optabs.c (expand_atomic_load): Likewise.
24879 (expand_atomic_store): Likewise.
24880 * ree.c (combine_reaching_defs): Likewise.
24881 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
24882 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
24883 * tree.h (type_has_mode_precision_p): Likewise.
24884 * ubsan.c (instrument_si_overflow): Likewise.
24886 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24887 Alan Hayward <alan.hayward@arm.com>
24888 David Sherwood <david.sherwood@arm.com>
24890 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
24891 polynomial numbers of units.
24892 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
24893 (valid_vector_subparts_p): New function.
24894 (build_vector_type): Remove temporary shim and take the number
24895 of units as a poly_uint64 rather than an int.
24896 (build_opaque_vector_type): Take the number of units as a
24897 poly_uint64 rather than an int.
24898 * tree.c (build_vector_from_ctor): Handle polynomial
24899 TYPE_VECTOR_SUBPARTS.
24900 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
24901 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
24902 (build_vector_from_val): If the number of units is variable,
24903 use build_vec_duplicate_cst for constant operands and
24904 VEC_DUPLICATE_EXPR otherwise.
24905 (make_vector_type): Remove temporary is_constant ().
24906 (build_vector_type, build_opaque_vector_type): Take the number of
24907 units as a poly_uint64 rather than an int.
24908 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
24910 * cfgexpand.c (expand_debug_expr): Likewise.
24911 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
24912 (store_constructor, expand_expr_real_1): Likewise.
24913 (const_scalar_mask_from_tree): Likewise.
24914 * fold-const-call.c (fold_const_reduction): Likewise.
24915 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
24916 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
24917 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
24918 (fold_relational_const): Likewise.
24919 (native_interpret_vector): Likewise. Change the size from an
24920 int to an unsigned int.
24921 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
24922 TYPE_VECTOR_SUBPARTS.
24923 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
24924 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
24925 duplicating a non-constant operand into a variable-length vector.
24926 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
24927 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
24928 * ipa-icf.c (sem_variable::equals): Likewise.
24929 * match.pd: Likewise.
24930 * omp-simd-clone.c (simd_clone_subparts): Likewise.
24931 * print-tree.c (print_node): Likewise.
24932 * stor-layout.c (layout_type): Likewise.
24933 * targhooks.c (default_builtin_vectorization_cost): Likewise.
24934 * tree-cfg.c (verify_gimple_comparison): Likewise.
24935 (verify_gimple_assign_binary): Likewise.
24936 (verify_gimple_assign_ternary): Likewise.
24937 (verify_gimple_assign_single): Likewise.
24938 * tree-pretty-print.c (dump_generic_node): Likewise.
24939 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24940 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
24941 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
24942 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
24943 (vect_shift_permute_load_chain): Likewise.
24944 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
24945 (expand_vector_condition, optimize_vector_constructor): Likewise.
24946 (lower_vec_perm, get_compute_type): Likewise.
24947 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24948 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
24949 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
24950 (vect_recog_mask_conversion_pattern): Likewise.
24951 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
24952 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
24953 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
24954 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
24955 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
24956 (vectorizable_shift, vectorizable_operation, vectorizable_store)
24957 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
24958 (supportable_widening_operation): Likewise.
24959 (supportable_narrowing_operation): Likewise.
24960 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
24962 * varasm.c (output_constant): Likewise.
24964 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24965 Alan Hayward <alan.hayward@arm.com>
24966 David Sherwood <david.sherwood@arm.com>
24968 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
24969 so that both the length == 3 and length != 3 cases set up their
24970 own permute vectors. Add comments explaining why we know the
24971 number of elements is constant.
24972 (vect_permute_load_chain): Likewise.
24974 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24975 Alan Hayward <alan.hayward@arm.com>
24976 David Sherwood <david.sherwood@arm.com>
24978 * machmode.h (mode_nunits): Change from unsigned char to
24980 (ONLY_FIXED_SIZE_MODES): New macro.
24981 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
24982 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
24983 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
24985 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
24986 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
24987 or if measurement_type is not polynomial.
24988 * genmodes.c (ZERO_COEFFS): New macro.
24989 (emit_mode_nunits_inline): Make mode_nunits_inline return a
24991 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
24992 Use ZERO_COEFFS when emitting initializers.
24993 * data-streamer.h (bp_pack_poly_value): New function.
24994 (bp_unpack_poly_value): Likewise.
24995 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24996 for GET_MODE_NUNITS.
24997 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24998 for GET_MODE_NUNITS.
24999 * tree.c (make_vector_type): Remove temporary shim and make
25000 the real function take the number of units as a poly_uint64
25001 rather than an int.
25002 (build_vector_type_for_mode): Handle polynomial nunits.
25003 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
25004 * emit-rtl.c (const_vec_series_p_1): Likewise.
25005 (gen_rtx_CONST_VECTOR): Likewise.
25006 * fold-const.c (test_vec_duplicate_folding): Likewise.
25007 * genrecog.c (validate_pattern): Likewise.
25008 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
25009 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
25010 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
25011 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
25012 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
25013 * rtlanal.c (subreg_get_info): Likewise.
25014 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25015 (vect_grouped_load_supported): Likewise.
25016 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
25017 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
25018 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
25019 (simplify_const_unary_operation, simplify_binary_operation_1)
25020 (simplify_const_binary_operation, simplify_ternary_operation)
25021 (test_vector_ops_duplicate, test_vector_ops): Likewise.
25022 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
25023 instead of CONST_VECTOR_NUNITS.
25024 * varasm.c (output_constant_pool_2): Likewise.
25025 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
25026 explicit-encoded elements in the XVEC for variable-length vectors.
25028 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25030 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
25032 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25033 Alan Hayward <alan.hayward@arm.com>
25034 David Sherwood <david.sherwood@arm.com>
25036 * coretypes.h (fixed_size_mode): Declare.
25037 (fixed_size_mode_pod): New typedef.
25038 * builtins.h (target_builtins::x_apply_args_mode)
25039 (target_builtins::x_apply_result_mode): Change type to
25040 fixed_size_mode_pod.
25041 * builtins.c (apply_args_size, apply_result_size, result_vector)
25042 (expand_builtin_apply_args_1, expand_builtin_apply)
25043 (expand_builtin_return): Update accordingly.
25045 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25047 * cse.c (hash_rtx_cb): Hash only the encoded elements.
25048 * cselib.c (cselib_hash_rtx): Likewise.
25049 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
25050 CONST_VECTOR encoding.
25052 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25053 Jeff Law <law@redhat.com>
25056 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
25057 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
25058 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
25059 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
25062 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
25063 explicitly probe *sp in a noreturn function if there were any callee
25064 register saves or frame pointer is needed.
25066 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25069 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
25070 BLKmode for ternary, binary or unary expressions.
25073 * var-tracking.c (delete_vta_debug_insn): New inline function.
25074 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
25075 insns from get_insns () to NULL instead of each bb separately.
25076 Use delete_vta_debug_insn. No longer static.
25077 (vt_debug_insns_local, variable_tracking_main_1): Adjust
25078 delete_vta_debug_insns callers.
25079 * rtl.h (delete_vta_debug_insns): Declare.
25080 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
25081 instead of variable_tracking_main.
25083 2018-01-03 Martin Sebor <msebor@redhat.com>
25085 PR tree-optimization/83603
25086 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
25087 arguments past the endof the argument list in functions declared
25088 without a prototype.
25089 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
25090 Avoid checking when arguments are null.
25092 2018-01-03 Martin Sebor <msebor@redhat.com>
25095 * doc/extend.texi (attribute const): Fix a typo.
25096 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
25097 issuing -Wsuggest-attribute for void functions.
25099 2018-01-03 Martin Sebor <msebor@redhat.com>
25101 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
25102 offset_int::from instead of wide_int::to_shwi.
25103 (maybe_diag_overlap): Remove assertion.
25104 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
25105 * gimple-ssa-sprintf.c (format_directive): Same.
25106 (parse_directive): Same.
25107 (sprintf_dom_walker::compute_format_length): Same.
25108 (try_substitute_return_value): Same.
25110 2018-01-03 Jeff Law <law@redhat.com>
25112 PR middle-end/83654
25113 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
25114 non-constant residual for zero at runtime and avoid probing in
25115 that case. Reorganize code for trailing problem to mirror handling
25118 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25120 PR tree-optimization/83501
25121 * tree-ssa-strlen.c (get_string_cst): New.
25122 (handle_char_store): Call get_string_cst.
25124 2018-01-03 Martin Liska <mliska@suse.cz>
25126 PR tree-optimization/83593
25127 * tree-ssa-strlen.c: Include tree-cfg.h.
25128 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
25129 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
25130 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
25132 (strlen_dom_walker::before_dom_children): Call
25133 gimple_purge_dead_eh_edges. Dump tranformation with details
25135 (strlen_dom_walker::before_dom_children): Update call by adding
25136 new argument cleanup_eh.
25137 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
25139 2018-01-03 Martin Liska <mliska@suse.cz>
25142 * cif-code.def (VARIADIC_THUNK): New enum value.
25143 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
25146 2018-01-03 Jan Beulich <jbeulich@suse.com>
25148 * sse.md (mov<mode>_internal): Tighten condition for when to use
25149 vmovdqu<ssescalarsize> for TI and OI modes.
25151 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25153 Update copyright years.
25155 2018-01-03 Martin Liska <mliska@suse.cz>
25158 * ipa-visibility.c (function_and_variable_visibility): Skip
25159 functions with noipa attribure.
25161 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25163 * gcc.c (process_command): Update copyright notice dates.
25164 * gcov-dump.c (print_version): Ditto.
25165 * gcov.c (print_version): Ditto.
25166 * gcov-tool.c (print_version): Ditto.
25167 * gengtype.c (create_file): Ditto.
25168 * doc/cpp.texi: Bump @copying's copyright year.
25169 * doc/cppinternals.texi: Ditto.
25170 * doc/gcc.texi: Ditto.
25171 * doc/gccint.texi: Ditto.
25172 * doc/gcov.texi: Ditto.
25173 * doc/install.texi: Ditto.
25174 * doc/invoke.texi: Ditto.
25176 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25178 * vector-builder.h (vector_builder::m_full_nelts): Change from
25179 unsigned int to poly_uint64.
25180 (vector_builder::full_nelts): Update prototype accordingly.
25181 (vector_builder::new_vector): Likewise.
25182 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
25183 (vector_builder::operator ==): Likewise.
25184 (vector_builder::finalize): Likewise.
25185 * int-vector-builder.h (int_vector_builder::int_vector_builder):
25186 Take the number of elements as a poly_uint64 rather than an
25188 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
25189 from unsigned int to poly_uint64.
25190 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
25191 (vec_perm_indices::new_vector): Likewise.
25192 (vec_perm_indices::length): Likewise.
25193 (vec_perm_indices::nelts_per_input): Likewise.
25194 (vec_perm_indices::input_nelts): Likewise.
25195 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
25196 number of elements per input as a poly_uint64 rather than an
25197 unsigned int. Use the original encoding for variable-length
25198 vectors, rather than clamping each individual element.
25199 For the second and subsequent elements in each pattern,
25200 clamp the step and base before clamping their sum.
25201 (vec_perm_indices::series_p): Handle polynomial element counts.
25202 (vec_perm_indices::all_in_range_p): Likewise.
25203 (vec_perm_indices_to_tree): Likewise.
25204 (vec_perm_indices_to_rtx): Likewise.
25205 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
25206 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
25207 (tree_vector_builder::new_binary_operation): Handle polynomial
25208 element counts. Return false if we need to know the number
25209 of elements at compile time.
25210 * fold-const.c (fold_vec_perm): Punt if the number of elements
25211 isn't known at compile time.
25213 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25215 * vec-perm-indices.h (vec_perm_builder): Change element type
25216 from HOST_WIDE_INT to poly_int64.
25217 (vec_perm_indices::element_type): Update accordingly.
25218 (vec_perm_indices::clamp): Handle polynomial element_types.
25219 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25220 (vec_perm_indices::all_in_range_p): Likewise.
25221 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
25223 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
25224 polynomial vec_perm_indices element types.
25225 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
25226 * fold-const.c (fold_vec_perm): Likewise.
25227 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
25228 * tree-vect-generic.c (lower_vec_perm): Likewise.
25229 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25230 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
25231 element type to HOST_WIDE_INT.
25233 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25234 Alan Hayward <alan.hayward@arm.com>
25235 David Sherwood <david.sherwood@arm.com>
25237 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
25238 rather than an int. Use plus_constant.
25239 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
25240 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
25242 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25243 Alan Hayward <alan.hayward@arm.com>
25244 David Sherwood <david.sherwood@arm.com>
25246 * calls.c (emit_call_1, expand_call): Change struct_value_size from
25247 a HOST_WIDE_INT to a poly_int64.
25249 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25250 Alan Hayward <alan.hayward@arm.com>
25251 David Sherwood <david.sherwood@arm.com>
25253 * calls.c (load_register_parameters): Cope with polynomial
25254 mode sizes. Require a constant size for BLKmode parameters
25255 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
25256 forces a parameter to be padded at the lsb end in order to
25257 fill a complete number of words, require the parameter size
25258 to be ordered wrt UNITS_PER_WORD.
25260 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25261 Alan Hayward <alan.hayward@arm.com>
25262 David Sherwood <david.sherwood@arm.com>
25264 * reload1.c (spill_stack_slot_width): Change element type
25265 from unsigned int to poly_uint64_pod.
25266 (alter_reg): Treat mode sizes as polynomial.
25268 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25269 Alan Hayward <alan.hayward@arm.com>
25270 David Sherwood <david.sherwood@arm.com>
25272 * reload.c (complex_word_subreg_p): New function.
25273 (reload_inner_reg_of_subreg, push_reload): Use it.
25275 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25276 Alan Hayward <alan.hayward@arm.com>
25277 David Sherwood <david.sherwood@arm.com>
25279 * lra-constraints.c (process_alt_operands): Reject matched
25280 operands whose sizes aren't ordered.
25281 (match_reload): Refer to this check here.
25283 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25284 Alan Hayward <alan.hayward@arm.com>
25285 David Sherwood <david.sherwood@arm.com>
25287 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
25288 that the mode size is in the set {1, 2, 4, 8, 16}.
25290 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25291 Alan Hayward <alan.hayward@arm.com>
25292 David Sherwood <david.sherwood@arm.com>
25294 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
25295 Use plus_constant instead of gen_rtx_PLUS.
25297 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25298 Alan Hayward <alan.hayward@arm.com>
25299 David Sherwood <david.sherwood@arm.com>
25301 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
25302 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
25303 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
25304 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
25305 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
25306 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
25307 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
25308 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
25309 * config/i386/i386.c (ix86_push_rounding): ...this new function.
25310 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
25312 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
25313 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
25314 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
25315 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
25316 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
25317 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
25318 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
25319 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
25320 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
25321 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
25323 * expr.c (emit_move_resolve_push): Treat the input and result
25324 of PUSH_ROUNDING as a poly_int64.
25325 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
25326 (emit_push_insn): Likewise.
25327 * lra-eliminations.c (mark_not_eliminable): Likewise.
25328 * recog.c (push_operand): Likewise.
25329 * reload1.c (elimination_effects): Likewise.
25330 * rtlanal.c (nonzero_bits1): Likewise.
25331 * calls.c (store_one_arg): Likewise. Require the padding to be
25332 known at compile time.
25334 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25335 Alan Hayward <alan.hayward@arm.com>
25336 David Sherwood <david.sherwood@arm.com>
25338 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
25339 Use plus_constant instead of gen_rtx_PLUS.
25341 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25342 Alan Hayward <alan.hayward@arm.com>
25343 David Sherwood <david.sherwood@arm.com>
25345 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
25346 rather than an int.
25348 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25349 Alan Hayward <alan.hayward@arm.com>
25350 David Sherwood <david.sherwood@arm.com>
25352 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
25353 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
25354 via stack temporaries. Treat the mode size as polynomial too.
25356 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25357 Alan Hayward <alan.hayward@arm.com>
25358 David Sherwood <david.sherwood@arm.com>
25360 * expr.c (expand_expr_real_2): When handling conversions involving
25361 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
25362 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
25363 as a poly_uint64 too.
25365 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25366 Alan Hayward <alan.hayward@arm.com>
25367 David Sherwood <david.sherwood@arm.com>
25369 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
25371 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25372 Alan Hayward <alan.hayward@arm.com>
25373 David Sherwood <david.sherwood@arm.com>
25375 * combine.c (can_change_dest_mode): Handle polynomial
25376 REGMODE_NATURAL_SIZE.
25377 * expmed.c (store_bit_field_1): Likewise.
25378 * expr.c (store_constructor): Likewise.
25379 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
25380 and polynomial REGMODE_NATURAL_SIZE.
25381 (gen_lowpart_common): Likewise.
25382 * reginfo.c (record_subregs_of_mode): Likewise.
25383 * rtlanal.c (read_modify_subreg_p): Likewise.
25385 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25386 Alan Hayward <alan.hayward@arm.com>
25387 David Sherwood <david.sherwood@arm.com>
25389 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
25390 numbers of elements.
25392 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25393 Alan Hayward <alan.hayward@arm.com>
25394 David Sherwood <david.sherwood@arm.com>
25396 * match.pd: Cope with polynomial numbers of vector elements.
25398 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25399 Alan Hayward <alan.hayward@arm.com>
25400 David Sherwood <david.sherwood@arm.com>
25402 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
25403 in a POINTER_PLUS_EXPR.
25405 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25406 Alan Hayward <alan.hayward@arm.com>
25407 David Sherwood <david.sherwood@arm.com>
25409 * omp-simd-clone.c (simd_clone_subparts): New function.
25410 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
25411 (ipa_simd_modify_function_body): Likewise.
25413 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25414 Alan Hayward <alan.hayward@arm.com>
25415 David Sherwood <david.sherwood@arm.com>
25417 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
25418 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
25419 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
25420 (expand_vector_condition, vector_element): Likewise.
25421 (subparts_gt): New function.
25422 (get_compute_type): Use subparts_gt.
25423 (count_type_subparts): Delete.
25424 (expand_vector_operations_1): Use subparts_gt instead of
25425 count_type_subparts.
25427 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25428 Alan Hayward <alan.hayward@arm.com>
25429 David Sherwood <david.sherwood@arm.com>
25431 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
25432 (vect_compile_time_alias): ...this new function. Do the calculation
25433 on poly_ints rather than trees.
25434 (vect_prune_runtime_alias_test_list): Update call accordingly.
25436 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25437 Alan Hayward <alan.hayward@arm.com>
25438 David Sherwood <david.sherwood@arm.com>
25440 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
25442 (vect_schedule_slp_instance): Likewise.
25444 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25445 Alan Hayward <alan.hayward@arm.com>
25446 David Sherwood <david.sherwood@arm.com>
25448 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
25449 constant and extern definitions for variable-length vectors.
25450 (vect_get_constant_vectors): Note that the number of units
25451 is known to be constant.
25453 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25454 Alan Hayward <alan.hayward@arm.com>
25455 David Sherwood <david.sherwood@arm.com>
25457 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
25458 of units as polynomial. Choose between WIDE and NARROW based
25461 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25462 Alan Hayward <alan.hayward@arm.com>
25463 David Sherwood <david.sherwood@arm.com>
25465 * tree-vect-stmts.c (simd_clone_subparts): New function.
25466 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
25468 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25469 Alan Hayward <alan.hayward@arm.com>
25470 David Sherwood <david.sherwood@arm.com>
25472 * tree-vect-stmts.c (vectorizable_call): Treat the number of
25473 vectors as polynomial. Use build_index_vector for
25474 IFN_GOMP_SIMD_LANE.
25476 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25477 Alan Hayward <alan.hayward@arm.com>
25478 David Sherwood <david.sherwood@arm.com>
25480 * tree-vect-stmts.c (get_load_store_type): Treat the number of
25481 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
25482 for variable-length vectors.
25483 (vectorizable_mask_load_store): Treat the number of units as
25484 polynomial, asserting that it is constant if the condition has
25485 already been enforced.
25486 (vectorizable_store, vectorizable_load): Likewise.
25488 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25489 Alan Hayward <alan.hayward@arm.com>
25490 David Sherwood <david.sherwood@arm.com>
25492 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
25493 of units as polynomial. Punt if we can't tell at compile time
25494 which vector contains the final result.
25496 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25497 Alan Hayward <alan.hayward@arm.com>
25498 David Sherwood <david.sherwood@arm.com>
25500 * tree-vect-loop.c (vectorizable_induction): Treat the number
25501 of units as polynomial. Punt on SLP inductions. Use an integer
25502 VEC_SERIES_EXPR for variable-length integer reductions. Use a
25503 cast of such a series for variable-length floating-point
25506 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25507 Alan Hayward <alan.hayward@arm.com>
25508 David Sherwood <david.sherwood@arm.com>
25510 * tree.h (build_index_vector): Declare.
25511 * tree.c (build_index_vector): New function.
25512 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
25513 of units as polynomial, forcibly converting it to a constant if
25514 vectorizable_reduction has already enforced the condition.
25515 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
25516 to create a {1,2,3,...} vector.
25517 (vectorizable_reduction): Treat the number of units as polynomial.
25518 Choose vectype_in based on the largest scalar element size rather
25519 than the smallest number of units. Enforce the restrictions
25522 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25523 Alan Hayward <alan.hayward@arm.com>
25524 David Sherwood <david.sherwood@arm.com>
25526 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
25527 number of units as polynomial.
25529 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25530 Alan Hayward <alan.hayward@arm.com>
25531 David Sherwood <david.sherwood@arm.com>
25533 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
25534 * target.def (autovectorize_vector_sizes): Return the vector sizes
25535 by pointer, using vector_sizes rather than a bitmask.
25536 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
25537 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
25538 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
25540 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
25541 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
25542 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
25543 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
25544 * omp-general.c (omp_max_vf): Likewise.
25545 * omp-low.c (omp_clause_aligned_alignment): Likewise.
25546 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
25547 * tree-vect-loop.c (vect_analyze_loop): Likewise.
25548 * tree-vect-slp.c (vect_slp_bb): Likewise.
25549 * doc/tm.texi: Regenerate.
25550 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
25552 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
25553 the vector size as a poly_uint64 rather than an unsigned int.
25554 (current_vector_size): Change from an unsigned int to a poly_uint64.
25555 (get_vectype_for_scalar_type): Update accordingly.
25556 * tree.h (build_truth_vector_type): Take the size and number of
25557 units as a poly_uint64 rather than an unsigned int.
25558 (build_vector_type): Add a temporary overload that takes
25559 the number of units as a poly_uint64 rather than an unsigned int.
25560 * tree.c (make_vector_type): Likewise.
25561 (build_truth_vector_type): Take the number of units as a poly_uint64
25562 rather than an unsigned int.
25564 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25565 Alan Hayward <alan.hayward@arm.com>
25566 David Sherwood <david.sherwood@arm.com>
25568 * target.def (get_mask_mode): Take the number of units and length
25569 as poly_uint64s rather than unsigned ints.
25570 * targhooks.h (default_get_mask_mode): Update accordingly.
25571 * targhooks.c (default_get_mask_mode): Likewise.
25572 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
25573 * doc/tm.texi: Regenerate.
25575 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25576 Alan Hayward <alan.hayward@arm.com>
25577 David Sherwood <david.sherwood@arm.com>
25579 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
25580 * omp-general.c (omp_max_vf): Likewise.
25581 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
25582 (expand_omp_simd): Handle polynomial safelen.
25583 * omp-low.c (omplow_simd_context): Add a default constructor.
25584 (omplow_simd_context::max_vf): Change from int to poly_uint64.
25585 (lower_rec_simd_input_clauses): Update accordingly.
25586 (lower_rec_input_clauses): Likewise.
25588 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25589 Alan Hayward <alan.hayward@arm.com>
25590 David Sherwood <david.sherwood@arm.com>
25592 * tree-vectorizer.h (vect_nunits_for_cost): New function.
25593 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
25594 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
25595 (vect_analyze_slp_cost): Likewise.
25596 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
25597 (vect_model_load_cost): Likewise.
25599 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25600 Alan Hayward <alan.hayward@arm.com>
25601 David Sherwood <david.sherwood@arm.com>
25603 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
25604 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
25605 from an unsigned int * to a poly_uint64_pod *.
25606 (calculate_unrolling_factor): New function.
25607 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
25609 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25610 Alan Hayward <alan.hayward@arm.com>
25611 David Sherwood <david.sherwood@arm.com>
25613 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
25614 from an unsigned int to a poly_uint64.
25615 (_loop_vec_info::slp_unrolling_factor): Likewise.
25616 (_loop_vec_info::vectorization_factor): Change from an int
25618 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
25619 (vect_get_num_vectors): New function.
25620 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
25621 (vect_get_num_copies): Use vect_get_num_vectors.
25622 (vect_analyze_data_ref_dependences): Change max_vf from an int *
25623 to an unsigned int *.
25624 (vect_analyze_data_refs): Change min_vf from an int * to a
25626 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
25627 than an unsigned HOST_WIDE_INT.
25628 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
25629 (vect_analyze_data_ref_dependence): Change max_vf from an int *
25630 to an unsigned int *.
25631 (vect_analyze_data_ref_dependences): Likewise.
25632 (vect_compute_data_ref_alignment): Handle polynomial vf.
25633 (vect_enhance_data_refs_alignment): Likewise.
25634 (vect_prune_runtime_alias_test_list): Likewise.
25635 (vect_shift_permute_load_chain): Likewise.
25636 (vect_supportable_dr_alignment): Likewise.
25637 (dependence_distance_ge_vf): Take the vectorization factor as a
25638 poly_uint64 rather than an unsigned HOST_WIDE_INT.
25639 (vect_analyze_data_refs): Change min_vf from an int * to a
25641 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
25642 vfm1 as a poly_uint64 rather than an int. Make the same change
25643 for the returned bound_scalar.
25644 (vect_gen_vector_loop_niters): Handle polynomial vf.
25645 (vect_do_peeling): Likewise. Update call to
25646 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
25647 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
25649 * tree-vect-loop.c (vect_determine_vectorization_factor)
25650 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
25651 (vect_get_known_peeling_cost): Likewise.
25652 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
25653 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
25654 (vect_transform_loop): Likewise. Use the lowest possible VF when
25655 updating the upper bounds of the loop.
25656 (vect_min_worthwhile_factor): Make static. Return an unsigned int
25657 rather than an int.
25658 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
25659 polynomial unroll factors.
25660 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
25661 (vect_make_slp_decision): Likewise.
25662 (vect_supported_load_permutation_p): Likewise, and polynomial
25664 (vect_analyze_slp_cost): Handle polynomial vf.
25665 (vect_slp_analyze_node_operations): Likewise.
25666 (vect_slp_analyze_bb_1): Likewise.
25667 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
25668 than an unsigned HOST_WIDE_INT.
25669 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
25670 (vectorizable_load): Handle polynomial vf.
25671 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
25673 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
25675 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25676 Alan Hayward <alan.hayward@arm.com>
25677 David Sherwood <david.sherwood@arm.com>
25679 * match.pd: Handle bit operations involving three constants
25680 and try to fold one pair.
25682 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25684 * tree-vect-loop-manip.c: Include gimple-fold.h.
25685 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
25686 niters_maybe_zero parameters. Handle other cases besides a step of 1.
25687 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
25688 Add a path that uses a step of VF instead of 1, but disable it
25690 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
25691 and niters_no_overflow parameters. Update calls to
25692 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
25693 Create a new SSA name if the latter choses to use a ste other
25694 than zero, and return it via niters_vector_mult_vf_var.
25695 * tree-vect-loop.c (vect_transform_loop): Update calls to
25696 vect_do_peeling, vect_gen_vector_loop_niters and
25697 slpeel_make_loop_iterate_ntimes.
25698 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
25699 (vect_gen_vector_loop_niters): Update declarations after above changes.
25701 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
25703 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
25704 128-bit round to integer instructions.
25705 (ceil<mode>2): Likewise.
25706 (btrunc<mode>2): Likewise.
25707 (round<mode>2): Likewise.
25709 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25711 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
25712 unaligned VSX load/store on P8/P9.
25713 (expand_block_clear): Allow the use of unaligned VSX
25714 load/store on P8/P9.
25716 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25718 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
25720 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
25721 swap associated with both a load and a store.
25723 2018-01-02 Andrew Waterman <andrew@sifive.com>
25725 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
25726 * config/riscv/riscv.md (clear_cache): Use it.
25728 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
25730 * web.c: Remove out-of-date comment.
25732 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25734 * expr.c (fixup_args_size_notes): Check that any existing
25735 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
25736 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
25737 (emit_single_push_insn): ...here.
25739 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25741 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
25742 (const_vector_encoded_nelts): New function.
25743 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
25744 (const_vector_int_elt, const_vector_elt): Declare.
25745 * emit-rtl.c (const_vector_int_elt_1): New function.
25746 (const_vector_elt): Likewise.
25747 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
25748 of CONST_VECTOR_ELT.
25750 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25752 * expr.c: Include rtx-vector-builder.h.
25753 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
25754 directly on the tree encoding.
25755 (const_vector_from_tree): Likewise.
25756 * optabs.c: Include rtx-vector-builder.h.
25757 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
25758 sequence of "u" values.
25759 * vec-perm-indices.c: Include rtx-vector-builder.h.
25760 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
25761 directly on the vec_perm_indices encoding.
25763 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25765 * doc/rtl.texi (const_vector): Describe new encoding scheme.
25766 * Makefile.in (OBJS): Add rtx-vector-builder.o.
25767 * rtx-vector-builder.h: New file.
25768 * rtx-vector-builder.c: Likewise.
25769 * rtl.h (rtx_def::u2): Add a const_vector field.
25770 (CONST_VECTOR_NPATTERNS): New macro.
25771 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
25772 (CONST_VECTOR_DUPLICATE_P): Likewise.
25773 (CONST_VECTOR_STEPPED_P): Likewise.
25774 (CONST_VECTOR_ENCODED_ELT): Likewise.
25775 (const_vec_duplicate_p): Check for a duplicated vector encoding.
25776 (unwrap_const_vec_duplicate): Likewise.
25777 (const_vec_series_p): Check for a non-duplicated vector encoding.
25778 Say that the function only returns true for integer vectors.
25779 * emit-rtl.c: Include rtx-vector-builder.h.
25780 (gen_const_vec_duplicate_1): Delete.
25781 (gen_const_vector): Call gen_const_vec_duplicate instead of
25782 gen_const_vec_duplicate_1.
25783 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
25784 (gen_const_vec_duplicate): Use rtx_vector_builder.
25785 (gen_const_vec_series): Likewise.
25786 (gen_rtx_CONST_VECTOR): Likewise.
25787 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
25788 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25789 Build a new vector rather than modifying a CONST_VECTOR in-place.
25790 (handle_special_swappables): Update call accordingly.
25791 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
25792 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25793 Build a new vector rather than modifying a CONST_VECTOR in-place.
25794 (handle_special_swappables): Update call accordingly.
25796 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25798 * simplify-rtx.c (simplify_const_binary_operation): Use
25799 CONST_VECTOR_ELT instead of XVECEXP.
25801 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25803 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
25804 the selector elements to be different from the data elements
25805 if the selector is a VECTOR_CST.
25806 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
25807 ssizetype for the selector.
25809 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25811 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
25812 before testing each element individually.
25813 * tree-vect-generic.c (lower_vec_perm): Likewise.
25815 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25817 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
25818 * selftest-run-tests.c (selftest::run_tests): Call it.
25819 * vector-builder.h (vector_builder::operator ==): New function.
25820 (vector_builder::operator !=): Likewise.
25821 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
25822 (vec_perm_indices::all_from_input_p): New function.
25823 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25824 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
25825 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
25826 instead of reading the VECTOR_CST directly. Detect whether both
25827 vector inputs are the same before constructing the vec_perm_indices,
25828 and update the number of inputs argument accordingly. Use the
25829 utility functions added above. Only construct sel2 if we need to.
25831 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25833 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
25834 the broadcast of the low byte.
25835 (expand_mult_highpart): Use an explicit encoding for the permutes.
25836 * optabs-query.c (can_mult_highpart_p): Likewise.
25837 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25838 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25839 (vectorizable_bswap): Likewise.
25840 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
25841 explicit encoding for the power-of-2 permutes.
25842 (vect_permute_store_chain): Likewise.
25843 (vect_grouped_load_supported): Likewise.
25844 (vect_permute_load_chain): Likewise.
25846 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25848 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
25849 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
25850 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
25851 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25852 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
25853 (vect_gen_perm_mask_any): Likewise.
25855 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25857 * int-vector-builder.h: New file.
25858 * vec-perm-indices.h: Include int-vector-builder.h.
25859 (vec_perm_indices): Redefine as an int_vector_builder.
25860 (auto_vec_perm_indices): Delete.
25861 (vec_perm_builder): Redefine as a stand-alone class.
25862 (vec_perm_indices::vec_perm_indices): New function.
25863 (vec_perm_indices::clamp): Likewise.
25864 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
25865 (vec_perm_indices::new_vector): New function.
25866 (vec_perm_indices::new_expanded_vector): Update for new
25867 vec_perm_indices class.
25868 (vec_perm_indices::rotate_inputs): New function.
25869 (vec_perm_indices::all_in_range_p): Operate directly on the
25870 encoded form, without computing elided elements.
25871 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
25872 encoding. Update for new vec_perm_indices class.
25873 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
25874 the given vec_perm_builder.
25875 (expand_vec_perm_var): Update vec_perm_builder constructor.
25876 (expand_mult_highpart): Use vec_perm_builder instead of
25877 auto_vec_perm_indices.
25878 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
25879 vec_perm_indices instead of auto_vec_perm_indices. Use a single
25880 or double series encoding as appropriate.
25881 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
25882 vec_perm_indices instead of auto_vec_perm_indices.
25883 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25884 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25885 (vect_permute_store_chain): Likewise.
25886 (vect_grouped_load_supported): Likewise.
25887 (vect_permute_load_chain): Likewise.
25888 (vect_shift_permute_load_chain): Likewise.
25889 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25890 (vect_transform_slp_perm_load): Likewise.
25891 (vect_schedule_slp_instance): Likewise.
25892 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25893 (vectorizable_mask_load_store): Likewise.
25894 (vectorizable_bswap): Likewise.
25895 (vectorizable_store): Likewise.
25896 (vectorizable_load): Likewise.
25897 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
25898 vec_perm_indices instead of auto_vec_perm_indices. Use
25899 tree_to_vec_perm_builder to read the vector from a tree.
25900 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
25901 vec_perm_builder instead of a vec_perm_indices.
25902 (have_whole_vector_shift): Use vec_perm_builder and
25903 vec_perm_indices instead of auto_vec_perm_indices. Leave the
25904 truncation to calc_vec_perm_mask_for_shift.
25905 (vect_create_epilog_for_reduction): Likewise.
25906 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
25907 from auto_vec_perm_indices to vec_perm_indices.
25908 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25909 instead of changing individual elements.
25910 (aarch64_vectorize_vec_perm_const): Use new_vector to install
25911 the vector in d.perm.
25912 * config/arm/arm.c (expand_vec_perm_d::perm): Change
25913 from auto_vec_perm_indices to vec_perm_indices.
25914 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25915 instead of changing individual elements.
25916 (arm_vectorize_vec_perm_const): Use new_vector to install
25917 the vector in d.perm.
25918 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
25919 Update vec_perm_builder constructor.
25920 (rs6000_expand_interleave): Likewise.
25921 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
25922 (rs6000_expand_interleave): Likewise.
25924 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25926 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
25927 to qimode could truncate the indices.
25928 * optabs.c (expand_vec_perm_var): Likewise.
25930 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25932 * Makefile.in (OBJS): Add vec-perm-indices.o.
25933 * vec-perm-indices.h: New file.
25934 * vec-perm-indices.c: Likewise.
25935 * target.h (vec_perm_indices): Replace with a forward class
25937 (auto_vec_perm_indices): Move to vec-perm-indices.h.
25938 * optabs.h: Include vec-perm-indices.h.
25939 (expand_vec_perm): Delete.
25940 (selector_fits_mode_p, expand_vec_perm_var): Declare.
25941 (expand_vec_perm_const): Declare.
25942 * target.def (vec_perm_const_ok): Replace with...
25943 (vec_perm_const): ...this new hook.
25944 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
25945 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
25946 * doc/tm.texi: Regenerate.
25947 * optabs.def (vec_perm_const): Delete.
25948 * doc/md.texi (vec_perm_const): Likewise.
25949 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
25950 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
25951 expand_vec_perm for constant permutation vectors. Assert that
25952 the mode of variable permutation vectors is the integer equivalent
25953 of the mode that is being permuted.
25954 * optabs-query.h (selector_fits_mode_p): Declare.
25955 * optabs-query.c: Include vec-perm-indices.h.
25956 (selector_fits_mode_p): New function.
25957 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
25958 is defined, instead of checking whether the vec_perm_const_optab
25959 exists. Use targetm.vectorize.vec_perm_const instead of
25960 targetm.vectorize.vec_perm_const_ok. Check whether the indices
25961 fit in the vector mode before using a variable permute.
25962 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
25963 vec_perm_indices instead of an rtx.
25964 (expand_vec_perm): Replace with...
25965 (expand_vec_perm_const): ...this new function. Take the selector
25966 as a vec_perm_indices rather than an rtx. Also take the mode of
25967 the selector. Update call to shift_amt_for_vec_perm_mask.
25968 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
25969 Use vec_perm_indices::new_expanded_vector to expand the original
25970 selector into bytes. Check whether the indices fit in the vector
25971 mode before using a variable permute.
25972 (expand_vec_perm_var): Make global.
25973 (expand_mult_highpart): Use expand_vec_perm_const.
25974 * fold-const.c: Includes vec-perm-indices.h.
25975 * tree-ssa-forwprop.c: Likewise.
25976 * tree-vect-data-refs.c: Likewise.
25977 * tree-vect-generic.c: Likewise.
25978 * tree-vect-loop.c: Likewise.
25979 * tree-vect-slp.c: Likewise.
25980 * tree-vect-stmts.c: Likewise.
25981 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
25983 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
25984 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
25985 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
25986 (aarch64_vectorize_vec_perm_const): ...this new function.
25987 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25988 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25989 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
25990 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
25991 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25992 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25993 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
25995 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
25996 check for NEON modes.
25997 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
25998 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
25999 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
26000 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
26002 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
26003 the old VEC_PERM_CONST conditions.
26004 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
26005 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
26006 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
26007 (ia64_vectorize_vec_perm_const_ok): Merge into...
26008 (ia64_vectorize_vec_perm_const): ...this new function.
26009 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
26010 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
26011 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
26012 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
26013 * config/mips/mips.c (mips_expand_vec_perm_const)
26014 (mips_vectorize_vec_perm_const_ok): Merge into...
26015 (mips_vectorize_vec_perm_const): ...this new function.
26016 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
26017 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
26018 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
26019 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
26020 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
26021 (rs6000_expand_vec_perm_const): Delete.
26022 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
26024 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26025 (altivec_expand_vec_perm_const_le): Take each operand individually.
26026 Operate on constant selectors rather than rtxes.
26027 (altivec_expand_vec_perm_const): Likewise. Update call to
26028 altivec_expand_vec_perm_const_le.
26029 (rs6000_expand_vec_perm_const): Delete.
26030 (rs6000_vectorize_vec_perm_const_ok): Delete.
26031 (rs6000_vectorize_vec_perm_const): New function.
26032 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
26033 an element count and rtx array.
26034 (rs6000_expand_extract_even): Update call accordingly.
26035 (rs6000_expand_interleave): Likewise.
26036 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
26037 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
26038 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
26039 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
26040 (rs6000_expand_vec_perm_const): Delete.
26041 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
26042 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26043 (altivec_expand_vec_perm_const_le): Take each operand individually.
26044 Operate on constant selectors rather than rtxes.
26045 (altivec_expand_vec_perm_const): Likewise. Update call to
26046 altivec_expand_vec_perm_const_le.
26047 (rs6000_expand_vec_perm_const): Delete.
26048 (rs6000_vectorize_vec_perm_const_ok): Delete.
26049 (rs6000_vectorize_vec_perm_const): New function. Remove stray
26050 reference to the SPE evmerge intructions.
26051 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
26052 an element count and rtx array.
26053 (rs6000_expand_extract_even): Update call accordingly.
26054 (rs6000_expand_interleave): Likewise.
26055 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
26056 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
26058 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26060 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26062 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
26063 vector mode and that that mode matches the mode of the data
26065 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
26066 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
26067 directly using expand_vec_perm_1 when forcing selectors into
26069 (expand_vec_perm_var): New function, split out from expand_vec_perm.
26071 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26073 * optabs-query.h (can_vec_perm_p): Delete.
26074 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
26075 * optabs-query.c (can_vec_perm_p): Split into...
26076 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
26077 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
26078 particular selector is valid.
26079 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
26080 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
26081 (vect_grouped_load_supported): Likewise.
26082 (vect_shift_permute_load_chain): Likewise.
26083 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
26084 (vect_transform_slp_perm_load): Likewise.
26085 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
26086 (vectorizable_bswap): Likewise.
26087 (vect_gen_perm_mask_checked): Likewise.
26088 * fold-const.c (fold_ternary_loc): Likewise. Don't take
26089 implementations of variable permutation vectors into account
26090 when deciding which selector to use.
26091 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
26092 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
26093 with a false third argument.
26094 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
26095 to test whether the constant selector is valid and can_vec_perm_var_p
26096 to test whether a variable selector is valid.
26098 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26100 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
26101 * optabs-query.c (can_vec_perm_p): Likewise.
26102 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
26103 instead of vec_perm_indices.
26104 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
26105 (vect_gen_perm_mask_checked): Likewise,
26106 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
26107 (vect_gen_perm_mask_checked): Likewise,
26109 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26111 * optabs-query.h (qimode_for_vec_perm): Declare.
26112 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
26113 (qimode_for_vec_perm): ...this new function.
26114 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
26116 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26118 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
26119 does not have a conditional at the top.
26121 2018-01-02 Richard Biener <rguenther@suse.de>
26123 * ipa-inline.c (big_speedup_p): Fix expression.
26125 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
26128 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
26131 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
26135 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
26136 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
26137 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
26138 cond_taken_branch_cost 3->4.
26140 2018-01-01 Jakub Jelinek <jakub@redhat.com>
26142 PR tree-optimization/83581
26143 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
26144 TODO_cleanup_cfg if any changes have been made.
26146 PR middle-end/83608
26147 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
26148 convert_modes if target mode has the right side, but different mode
26151 PR middle-end/83609
26152 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
26153 last argument when extracting from CONCAT. If either from_real or
26154 from_imag is NULL, use expansion through memory. If result is not
26155 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
26156 the parts directly to inner mode, if even that fails, use expansion
26159 PR middle-end/83623
26160 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
26161 check for bswap in mode rather than HImode and use that in expand_unop
26164 Copyright (C) 2018 Free Software Foundation, Inc.
26166 Copying and distribution of this file, with or without modification,
26167 are permitted in any medium without royalty provided the copyright
26168 notice and this notice are preserved.