1 2017-05-02 Bin Cheng <bin.cheng@arm.com>
3 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
4 (get_computation): Delete.
5 (get_computation_cost): Implement like get_computation_cost_at.
6 Use get_computation_at.
7 (get_computation_cost_at): Delete.
8 (rewrite_use_nonlinear_expr): Use get_computation_at.
9 (rewrite_use_compare, remove_unused_ivs): Ditto.
11 2017-05-02 Bin Cheng <bin.cheng@arm.com>
13 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
15 2017-05-02 Bin Cheng <bin.cheng@arm.com>
17 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
18 (ivopts_global_cost_for_size): Rename parameter and update uses.
19 (iv_ca_recount_cost): Update uses.
20 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21 candidates seperately in n_invs and n_cands.
22 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
27 (find_inv_vars_cb): New.
28 (find_depends): Renamed to ...
29 (find_inv_vars): ... this.
30 (add_candidate_1, force_var_cost): Call find_inv_vars.
31 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33 2017-05-02 Bin Cheng <bin.cheng@arm.com>
35 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
36 inv_vars. Add inv_exprs.
37 (struct iv_cand): Rename depends_on to inv_vars.
38 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
39 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
40 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
41 (dump_cand): Dump inv_vars.
42 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
43 (record_invariant, find_depends, add_candidate_1): Ditto.
44 (set_group_iv_cost, force_var_cost): Ditto.
45 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
46 (get_computation_cost_at, get_computation_cost): Ditto.
47 (determine_group_iv_cost_generic): Ditto.
48 (determine_group_iv_cost_address): Ditto.
49 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
50 (determine_group_iv_costs): Ditto.
51 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
52 (iv_ca_set_remove_invariants): Renamed to ...
53 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
54 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
55 (iv_ca_set_add_invariants): Renamed to ...
56 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
57 (iv_ca_set_cp): Use iv_ca_set_add_invs.
58 (iv_ca_has_deps): Support inv_vars and inv_exprs.
59 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
60 (create_new_ivs): Remove useless dump.
62 2017-05-02 Bin Cheng <bin.cheng@arm.com>
64 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
66 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
67 (iv_ca_set_no_cp, create_new_iv): Ditto.
69 2017-05-02 Bin Cheng <bin.cheng@arm.com>
71 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
73 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
75 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
78 2017-05-02 Martin Liska <mliska@suse.cz>
80 * doc/gcov.texi: Add missing preposition.
81 * gcov.c (function_info::function_info): Properly fill up
84 2017-05-02 Tamar Christina <tamar.christina@arm.com>
86 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
88 2017-05-02 Tamar Christina <tamar.christina@arm.com>
90 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
92 2017-05-02 Martin Liska <mliska@suse.cz>
95 * lto-streamer-in.c (lto_read_tree_1): Remove
97 * lto-streamer.c (struct tree_hash_entry): Likewise.
98 (struct tree_entry_hasher): Likewise.
99 (tree_entry_hasher::hash): Likewise.
100 (tree_entry_hasher::equal): Likewise.
101 (lto_streamer_init): Likewise.
102 (lto_orig_address_map): Likewise.
103 (lto_orig_address_get): Likewise.
104 (lto_orig_address_remove): Likewise.
105 * lto-streamer.h: Likewise.
106 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
107 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
109 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
111 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
112 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
113 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
114 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
115 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
116 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
117 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
118 (mm_maskz_sub_ss): New intrinsics.
119 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
120 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
121 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
122 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
123 (__builtin_ia32_subss_mask_round): New builtins.
124 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
125 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
126 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
128 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
129 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
131 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
134 2017-05-02 Martin Jambor <mjambor@suse.cz>
136 PR tree-optimization/78687
137 * tree-sra.c (access): New field parent.
138 (process_subtree_disqualification): New function.
139 (disqualify_candidate): Call it.
140 (build_accesses_from_assign): Reset write flag if creating an
142 (build_access_subtree): Fill in parent field and also prpagate
144 (create_artificial_child_access): New parameter set_grp_write, set
145 grp_write to its value.
146 (propagate_subaccesses_across_link): Also propagate grp_write flag
148 (propagate_all_subaccesses): Push the closest parent back to work
149 queue if add_access_to_work_queue returned true.
151 2017-05-02 Richard Biener <rguenther@suse.de>
153 * common.opt (fstrict-overflow): Alias negative to fwrapv.
154 * doc/invoke.texi (fstrict-overflow): Remove all traces of
155 -fstrict-overflow documentation.
156 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
157 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
158 flag_strict_overflow.
159 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
160 * lto-opts.c (lto_write_options): Do not stream it.
161 * lto-wrapper.c (merge_and_complain): Do not handle it.
162 * opts.c (default_options_table): Do not set -fstrict-overflow.
163 (finish_options): Likewise do not clear it when sanitizing.
164 * simplify-rtx.c (simplify_const_relational_operation): Do not
165 test flag_strict_overflow.
167 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
169 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
170 using enabled attribute.
171 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
172 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
173 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
174 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
175 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
176 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
177 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
178 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
179 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
180 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
182 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
184 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
186 2017-05-02 Richard Biener <rguenther@suse.de>
188 PR tree-optimization/80591
190 2017-04-10 Richard Biener <rguenther@suse.de>
192 * tree-ssa-structalias.c (find_func_aliases): Properly handle
195 2017-05-02 Richard Biener <rguenther@suse.de>
197 PR tree-optimization/80549
198 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
199 (cleanup_tree_cfg_noloop): Create forwarders to known loop
200 headers if they do not have a preheader.
202 2017-05-02 Martin Liska <mliska@suse.cz>
205 * common.opt: Fix typo.
206 * doc/invoke.texi: Likewise.
208 2017-05-01 Jan Beulich <jbeulich@suse.com>
210 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
211 swapping, add (x,x,m,x,n) alternative.
213 2017-05-01 Nathan Sidwell <nathan@acm.org>
215 * calls.c (combine_pending_stack_adjustment_and_call): Remove
216 unnecessary unadjusted_alignment check.
218 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
221 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
223 * gimplify.c (gimplify_cilk_detach): New function.
224 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
225 * tree-core.h: Document EXPR_CILK_SPAWN.
226 * tree.h (EXPR_CILK_SPAWN): Define.
228 2017-05-01 David Malcolm <dmalcolm@redhat.com>
230 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
231 to use new fixit_hint representation, using the "replace" logic.
232 (get_line_span_for_fixit_hint): Likewise.
233 (layout::print_any_fixits): Likewise.
234 (selftest::test_one_liner_many_fixits): Rename to...
235 (selftest::test_one_liner_many_fixits_1): ...this, and update
236 comment and expected output to reflect that the multiple fix-it
237 hints are now consolidated into one insertion.
238 (selftest::test_one_liner_many_fixits_2): New test.
239 (selftest::test_diagnostic_show_locus_one_liner): Update for
241 (selftest::test_fixit_consolidation): Update for fix-it API
243 * diagnostic.c (print_parseable_fixits): Likewise.
244 * edit-context.c (edited_line::m_line_events): Convert from
245 auto_vec <line_event *> to auto_vec <line_event>.
246 (class line_event): Convert from abstract base class to a concrete
247 class, taking over the role of replace_event.
248 (class insert_event): Delete.
249 (class replace_event): Rename to class line_event. Convert to
251 (edit_context::add_fixits): Reimplement.
252 (edit_context::apply_insert): Delete.
253 (edit_context::apply_replace): Rename to...
254 (edit_context::apply_fixit): ...this. Convert to half-open range.
255 (edited_file::apply_insert): Delete.
256 (edited_file::apply_replace): Rename to...
257 (edited_file::apply_fixit): ...this.
258 (edited_line::~edited_line): Drop deletion of events.
259 (edited_line::apply_insert): Delete.
260 (edited_line::apply_replace): Rename to...
261 (edited_line::apply_fixit): ...this. Convert to half-open range.
262 Update for change to type of m_line_events.
263 * edit-context.h (edit_context::apply_insert): Delete.
264 (edit_context::apply_replace): Rename to...
265 (edit_context::apply_fixit): ...this.
267 2017-05-01 Martin Sebor <msebor@redhat.com>
269 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
272 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
275 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
276 __get_cpuid_max returns 0.
277 (__get_cpuid_count): Ditto.
279 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
281 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
282 replacement expression is another instance of one of its arguments.
284 2017-05-01 Jakub Jelinek <jakub@redhat.com>
287 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
288 check for stack push/pop autoinc.
289 * config/i386/i386.c (ix86_agi_dependent): Return false
290 if the only reason why modified_in_p returned true is that
291 addr is SP based and set_insn is a push or pop.
293 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
296 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
297 (account_size_time): Use two predicates - exec_pred and
299 (evaluate_conditions_for_known_args): Compute both clause and
301 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
302 (inline_summary_t::duplicate): Update.
303 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
305 (compute_inline_parameters): Likewise.
306 (estimate_edge_size_and_time): Update caluclation of time.
307 (estimate_node_size_and_time): Compute both time and nonspecialized
309 (estimate_ipcp_clone_size_and_time): Update.
310 (inline_merge_summary): Update.
311 (do_estimate_edge_time): Update.
312 (do_estimate_edge_size): Update.
313 (do_estimate_edge_hints): Update.
314 (inline_read_section, inline_write_summary): Stream both new predicates.
315 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
317 (compute_inlined_call_time): Cleanup.
318 (big_speedup_p): Update.
319 (edge_badness): Update.
320 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
321 (size_time_entry): Replace predicate by exec_predicate and
323 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
324 (estimate_edge_time): Return also nonspec_time.
325 (reset_edge_growth_cache): Update.
327 2017-04-29 Jakub Jelinek <jakub@redhat.com>
329 PR rtl-optimization/80491
330 * ifcvt.c (noce_process_if_block): When looking for x setter
331 with missing else_bb, don't check only the insn right before
332 cond_earliest, but look for the last insn that x is modified in
335 PR rtl-optimization/80491
336 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
338 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
340 PR tree-optimization/80487
341 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
343 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
345 PR tree-optimization/79697
346 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
347 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
348 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
350 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
351 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
353 2017-04-28 Martin Sebor <msebor@redhat.com>
355 PR tree-optimization/80523
356 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
357 (init_target_to_host_charmap, target_to_host, target_strtol10): New
359 (maybe_warn, format_directive, parse_directive): Use new functions.
360 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
362 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
364 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
366 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
368 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
369 target_header_dir): Set correctly.
370 * configure: Regenerated.
371 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
372 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
373 instead of SYSTEM_HEADER_DIR.
375 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
377 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
378 (estimate_local_effects): Likewise.
379 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
380 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
381 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
382 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
383 do_estimate_edge_time, estimate_edge_time): Likewise.
384 * ipa-inline-analysis.c (estimate_node_size_and_time,
385 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
386 (estimate_time_after_inlining): Remove.
388 2017-04-28 Martin Liska <mliska@suse.cz>
390 * doc/gcov.texi: Enhance documentation of gcov.
392 2017-04-28 Martin Liska <mliska@suse.cz>
394 * doc/gcov.texi: Sort options in alphabetic order.
395 * doc/gcov-dump.texi: Likewise.
396 * doc/gcov-tool.texi: Likewise.
397 * gcov.c (print_usage): Likewise.
398 * gcov-dump.c (print_usage): Likewise.
399 * gcov-tool.c (print_merge_usage_message): Likewise.
400 (print_rewrite_usage_message): Likewise.
401 (print_overlap_usage_message): Likewise.
403 2017-04-28 Martin Liska <mliska@suse.cz>
405 PR gcov-profile/53915
406 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
408 2017-04-28 Martin Liska <mliska@suse.cz>
410 PR gcov-profile/79891
411 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
412 is marked by compiler as living on a line.
413 (get_cycles_count): Remove usage of the union.
414 (output_intermediate_file): Likewise.
415 (find_source): Fix GNU coding style.
416 (accumulate_line_counts): Remove old non-all block mode.
417 (output_lines): Remove usage of the union.
418 * profile.c (output_location): Include all BBs, even if
419 belonging to a same line (and file) as a previous BB.
421 2017-04-28 Martin Liska <mliska@suse.cz>
423 * gcov.c (process_args): Handle new argument 'w'.
424 (read_graph_file): Assign ID to BBs.
425 (output_branch_count): Display BB # if verbose flag is set.
426 (output_lines): Likewise for arcs.
427 (print_usage): Add '--verbose' option help.
428 * doc/gcov.texi: Document --verbose (-w) option.
430 2017-04-28 Martin Liska <mliska@suse.cz>
432 * gcov.c (struct block_location_info): New struct.
433 (process_file): Fill up the new structure.
434 (read_graph_file): Replace usage of encoding by the newly added
436 (add_line_counts): Likewise.
437 (accumulate_line_counts): Remove usage of the union.
438 (function_info::function_info): New function.
439 (function_info::~function_info): Likewise.
440 (process_file): Call delete instead of release_function.
441 (release_function): Release the function.
442 (release_structures): Call delete instead of release_function.
443 (solve_flow_graph): Replace usage of num_blocks.
444 (find_exception_blocks): Likewise.
445 (output_lines): Fix GNU coding style.
447 2017-04-28 Martin Liska <mliska@suse.cz>
450 * coverage.c (coverage_remove_note_file): New function.
451 * coverage.h: Declare the function.
452 * toplev.c (finalize): Clean if an error has been seen.
454 2017-04-28 Martin Liska <mliska@suse.cz>
456 PR gcov-profile/80031
457 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
458 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
459 * gcov.c (read_graph_file): Read just number of blocks.
460 * profile.c (branch_prob): Do not stream 0 flags per a basic
463 2017-04-28 Martin Liska <mliska@suse.cz>
465 * gcov-dump.c (tag_*): Add new argument to declarations.
466 (dump_gcov_file): Likewise.
467 (tag_blocks): Add and use new argument depth.
468 (tag_arcs): Likewise.
469 (tag_lines): Likewise.
470 (tag_counters): Likewise.
471 (tag_summary): Likewise.
472 (dump_working_sets): Use depth to do a proper indentation.
474 2017-04-28 Jakub Jelinek <jakub@redhat.com>
477 * cgraph.h (symtab_node::debug_symtab): No longer inline.
478 * symtab.c (symtab_node::debug_symtab): Move definition here.
480 2017-04-28 Richard Biener <rguenther@suse.de>
482 * lto-streamer.h (LTO_major_version): Bump to 7.
484 2017-04-28 Richard Biener <rguenther@suse.de>
486 * tree-vrp.c (assert_info): New struct.
487 (add_assert_info): New helper.
488 (register_edge_assert_for_2): Refactor to add asserts to a vector
490 (register_edge_assert_for_1): Likewise.
491 (register_edge_assert_for): Likewise.
492 (finish_register_edge_assert_for): New helper actually registering
493 asserts where live on edge.
494 (find_conditional_asserts): Adjust.
495 (find_switch_asserts): Likewise.
496 (evrp_dom_walker::try_find_new_range): Generalize.
497 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
499 2017-04-27 Marek Polacek <polacek@redhat.com>
502 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
503 arg10 and arg11 to itype.
505 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
507 * doc/extend.texi (Object Size Checking): Improve grammar.
509 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
512 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
513 that the logic for permitting reciprocal estimates matches that
516 2017-04-27 Jakub Jelinek <jakub@redhat.com>
519 * tree.c (type_cache_hasher::equal): Only compare
520 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
521 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
522 non-aggregate element types.
523 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
524 about the flag on ARRAY_TYPEs in the comment, formatting fix.
526 2017-04-27 Richard Biener <rguenther@suse.de>
529 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
530 stripping ARRAY_REFs from MEM_EXPR make sure we're not
531 keeping a reference to a trailing array.
533 2017-04-27 Richard Biener <rguenther@suse.de>
536 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
537 being in loop-closed SSA form conservatively.
538 (chrec_fold_multiply_poly_poly): Likewise.
540 2017-04-27 Tamar Christina <tamar.christina@arm.com>
543 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
544 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
546 2017-04-27 Jakub Jelinek <jakub@redhat.com>
549 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
550 (aarch64_function_arg_alignment): Return unsigned int again, but still
551 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
552 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
553 Don't emit -Wpsabi note.
554 (aarch64_function_arg_boundary): Likewise.
555 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
558 2017-04-26 Nathan Sidwell <nathan@acm.org>
560 * tree.h (crc32_unsigned_n): Declare.
561 (crc32_unsigned, crc32_unsigned): Make inline.
562 * tree.c (crc32_unsigned_bits): Replace with ...
563 (crc32_unsigned_n): ... this.
564 (crc32_unsigned, crc32_byte): Remove.
565 (crc32_string): Remove unnecessary braces.
567 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
569 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
570 * ipa-inline-analysis.c (MAX_TIME): Remove.
571 (account_size_time): Use sreal for time.
572 (dump_inline_summary): Update.
573 (estimate_function_body_sizes): Update.
574 (estimate_edge_size_and_time): Update.
575 (estimate_calls_size_and_time): Update.
576 (estimate_node_size_and_time): Update.
577 (inline_merge_summary): Update.
578 (inline_update_overall_summary): Update.
579 (estimate_time_after_inlining): Update.
580 (inline_read_section): Update.
581 (inline_write_summary): Update.
582 * ipa-inline.c (compute_uninlined_call_time): Update.
583 (compute_inlined_call_time): Update.
584 (recursive_inlining): Update.
585 (inline_small_functions): Update.
586 (dump_overall_stats): Update.
587 * ipa-inline.h: Include sreal.h.
588 (size_time_entry): Turn time to sreal.
589 (inline_summary): Turn self_time nad time to sreal.
591 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
593 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
595 (sreal::stream_out, sreal::stream_in): New.
596 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
598 2017-04-25 Jakub Jelinek <jakub@redhat.com>
600 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
603 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
606 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
608 (dimode_scalar_chain::compute_convert_gain): Ditto.
609 (dimode_scalar_chain::make_vector_copies): Ditto.
610 (dimode_scalar_chain::convert_reg): Ditto.
611 (dimode_scalar_chain::convert_insn): Ditto.
612 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
613 (VI248_AVX512BW_1): New mode iterator.
614 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
615 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
618 2017-04-25 Martin Sebor <msebor@redhat.com>
620 PR tree-optimization/80497
621 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
622 constants are representable in HOST_WIDE_INT.
623 (parse_directive): Ditto.
625 2017-04-25 Martin Sebor <msebor@redhat.com>
628 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
629 (new_zero_array): Adjust signature.
630 (dom_info::dom_init): Used unsigned rather that size_t.
631 (dom_info::dom_info): Same.
633 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
634 Jakub Jelinek <jakub@redhat.com>
637 * config/arm/arm.c: Include gimple.h.
638 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
639 returns negative, increment ncrn only if it returned positive.
640 (arm_needs_doubleword_align): Return int instead of bool,
641 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
642 members, but if there is any such non-FIELD_DECL
643 > PARM_BOUNDARY aligned decl, return -1 instead of false.
644 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
645 returns negative, increment nregs only if it returned positive.
646 (arm_setup_incoming_varargs): Likewise.
647 (arm_function_arg_boundary): Emit -Wpsabi note if
648 arm_needs_doubleword_align returns negative, return
649 DOUBLEWORD_ALIGNMENT only if it returned positive.
651 2017-04-25 Marek Polacek <polacek@redhat.com>
654 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
655 first argument to type.
657 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
660 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
661 type checks to test for compatibility instead of equality.
663 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
664 Jakub Jelinek <jakub@redhat.com>
667 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
669 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
670 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
671 the alignment computation, but return their maximum in warn_alignment.
672 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
673 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
675 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
676 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
679 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
681 * config/arc/simdext.md (dmpyh): Fix typo.
683 2017-04-25 Richard Biener <rguenther@suse.de>
685 PR tree-optimization/80492
686 * alias.c (compare_base_decls): Handle registers with asm
687 specification conservatively.
688 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
689 compare_base_decls returning dont-know properly.
691 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
693 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
694 (legitimate_offset_address_p): New function.
695 (arc_legitimate_address_p): Use above function.
697 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
699 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
701 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
703 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
704 ACCH registers whenever they are available.
706 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
708 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
709 double regs fix when not used.
711 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
713 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
715 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
716 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
718 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
720 * config/arc/arc.c (arc_output_addsi): Check for h-register class
721 when emitting short ADD instructions.
723 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
725 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
727 (cmpsi_cc_c_insn): Likewise.
728 (cbranchsi4_scratch): Compute proper instruction length using
729 compact_hreg_operand.
730 * config/arc/predicates.md (compact_hreg_operand): New predicate.
732 2017-04-25 Richard Biener <rguenther@suse.de>
735 * passes.c (pass_manager::pass_manager): Initialize
738 2017-04-25 Richard Biener <rguenther@suse.de>
740 PR tree-optimization/79201
741 * tree-ssa-sink.c (statement_sink_location): Handle calls.
743 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
746 * config/s390/vector.md: Split MEM->GPR vector moves for
747 non-s_operand addresses.
749 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
752 * config/s390/predicates.md (reload_const_wide_int_operand): New
754 * config/s390/s390.md ("movti"): Remove d/P alternative.
755 ("movti_bigconst"): New pattern definition.
757 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
760 * s390-protos.h (s390_expand_cs_hqi): Removed.
761 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
762 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
763 modes as well as CCZ1mode and CCZmode.
764 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
765 signature of s390_emit_compare_and_swap.
766 (s390_expand_cs_hqi): Likewise, make static.
767 (s390_expand_cs_tdsi): Generate an explicit compare before trying
768 compare-and-swap, in some cases.
769 (s390_expand_cs): Wrapper function.
770 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
772 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
773 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
774 patterns for small and large integers. Forbid symref memory operands.
775 Move expander to s390.c. Require cc register.
776 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
777 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
778 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
779 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
780 symref memory operands. Remove CC mode and call s390_match_ccmode
782 ("atomic_exchange<mode>"): Allow and implement all integer modes.
784 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
786 * config/s390/s390.md (define_peephole2): New peephole to help
787 combining the load-and-test pattern with volatile memory.
789 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
791 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
792 with CCZmode for TARGET_Z196.
794 2017-04-25 Jakub Jelinek <jakub@redhat.com>
796 PR rtl-optimization/80501
797 * combine.c (make_compound_operation_int): Set subreg_code to SET
798 even for AND with mask of the sign bit of mode.
800 PR rtl-optimization/80500
801 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
804 2017-04-25 Julian Brown <julian@codesourcery.com>
805 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
807 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
809 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
811 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
813 2017-04-25 Julian Brown <julian@codesourcery.com>
814 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
816 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
817 (thunderx2t99_sha): New Reservation.
819 2017-04-25 Julian Brown <julian@codesourcery.com>
820 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
822 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
823 type for 1-element load.
825 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
827 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
829 2017-04-24 Martin Jambor <mjambor@suse.cz>
831 PR tree-optimization/80293
832 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
833 char arrays not totally scalarizable if it is false.
834 (analyze_all_variable_accesses): Pass correct value in the new
835 parameter. Add a statistics counter.
837 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
840 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
842 2017-04-24 Richard Biener <rguenther@suse.de>
844 PR tree-optimization/80494
845 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
846 out for complex types.
848 2017-04-24 Richard Biener <rguenther@suse.de>
850 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
851 * tree-ssa-sccvn.c (print_scc): Print SCC size.
852 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
853 (DFS): Adjust and never fail.
854 (sccvn_dom_walker::fail): Remove.
855 (sccvn_dom_walker::before_dom_children): Adjust.
856 (run_scc_vn): Likewise and never fail.
857 * tree-ssa-pre.c (pass_pre::execute): Adjust.
858 (pass_fre::execute): Likewise.
860 2017-04-24 Richard Biener <rguenther@suse.de>
862 PR tree-optimization/79725
863 * tree-ssa-sink.c (statement_sink_location): Return whether
864 failure reason was zero uses. Move that check later.
865 (sink_code_in_bb): Deal with zero uses by removing the stmt
868 2017-04-24 Richard Biener <rguenther@suse.de>
871 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
872 pointer-based references.
874 2017-04-24 Richard Biener <rguenther@suse.de>
877 * pass_manager.h (pass_manager::operator new): Remove.
878 (pass_manager::operator delete): Likewise.
879 * passes.c (pass_manager::operator new): Remove.
880 (pass_manager::operator delete): Likewise.
881 (pass_manager::pass_manager): Zero individual pass members.
883 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
886 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
887 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
888 Check "XEXP (src, 1)" operand here.
889 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
890 Check "XEXP (src, 1)" operand here.
891 (dimode_scalar_chain::make_vector_copies): Detect count register
892 of a shift instruction. Zero extend count register from QImode
893 to DImode to satisfy vector shift pattern count operand predicate.
894 Substitute vector shift count operand with a DImode copy.
895 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
898 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
900 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
901 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
902 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
903 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
904 (UNSPEC_NOREX_MEM): Remove definition.
906 2017-04-21 Richard Biener <rguenther@suse.de>
908 PR tree-optimization/79547
909 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
910 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
911 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
912 without any constraints.
914 2017-04-21 Richard Biener <rguenther@suse.de>
916 PR tree-optimization/78847
917 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
919 2017-04-21 Richard Biener <rguenther@suse.de>
921 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
922 (build_distinct_type_copy): Likewise.
923 (build_variant_type_copy): Likewise.
924 * tree.c (build_qualified_type): Pass down mem-stat info.
925 (build_distinct_type_copy): Likewise.
926 (build_variant_type_copy): Likewise.
928 2017-04-21 Richard Biener <rguenther@suse.de>
930 PR tree-optimization/80237
931 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
933 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
934 for a simplified result.
936 2016-04-21 Richard Biener <rguenther@suse.de>
938 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
939 sth as strict as a simple_iv but a chrec without symbols and an
940 operand defined in the loop we are peeling (and not some subloop).
941 (propagate_constants_for_unrolling): Propagate all constants.
943 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
946 * config/i386/i386.c (print_reg): Remove assert for disalowed
947 regno values, call output_operand_lossage instead.
949 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
952 * config/i386/constraints.md (Yc): New register constraint.
953 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
954 Use Yc constraint for alternative 2 of operand 0. Remove
955 preferred_for_speed attribute.
957 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
959 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
960 lastprivate clauses in SIMT case.
962 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
964 * doc/invoke.texi (-Wextra-semi): Document new warning option.
966 2017-04-20 Richard Biener <rguenther@suse.de>
968 PR tree-optimization/57796
969 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
971 (vect_model_load_cost): Cost gathers as N scalar loads.
973 2017-04-20 Richard Biener <rguenther@suse.de>
975 * ggc-page.c (ggc_allocated_p): Rename to ...
976 (safe_lookup_page_table_entry): ... this and return the lookup
978 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
980 2017-04-20 Richard Biener <rguenther@suse.de>
982 PR tree-optimization/80453
983 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
984 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
986 (vn_phi_eq): Pass them down.
987 (vn_phi_lookup): Record them.
988 (vn_phi_insert): Likewise.
990 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
992 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
993 uninitialized variable warning to avoid buffer overrun.
995 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
998 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
999 is suppressed for '{ 0 }' in C.
1001 2017-04-20 Jakub Jelinek <jakub@redhat.com>
1003 * BASE-VER: Set to 8.0.0.
1005 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1007 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
1008 priority .init_array and .fini_array section with SECTION_NOTYPE
1011 2017-04-20 Jakub Jelinek <jakub@redhat.com>
1014 * tree.h (build_array_type): Add typeless_storage default argument.
1015 * tree.c (type_cache_hasher::equal): Also compare
1016 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
1017 (build_array_type): Add typeless_storage argument, set
1018 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
1020 (build_nonshared_array_type): Adjust build_array_type_1 caller.
1021 (build_array_type): Likewise. Add typeless_storage argument.
1023 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
1024 Jakub Jelinek <jakub@redhat.com>
1026 PR tree-optimization/80426
1027 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
1028 operation on symbolic operands, also compute the overflow for the
1029 invariant part when the operation degenerates into a negation.
1031 2017-04-19 Jakub Jelinek <jakub@redhat.com>
1034 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
1035 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
1038 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
1040 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
1043 * config/avr/avr.c (tree.h): Include it.
1044 (cgraph.h): Include it.
1045 (avr_encode_section_info): Don't warn for uninitialized progmem
1046 variable if it's just an alias.
1048 2017-04-19 Richard Biener <rguenther@suse.de>
1051 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
1052 when needed by AutoPGO.
1054 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
1057 * doc/lto.texi: Remove an extra 'that'.
1059 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
1061 PR rtl-optimization/80429
1062 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
1063 are only used in debug insns.
1065 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
1066 Vladimir Makarov <vmakarov@redhat.com>
1068 * config/sparc/predicates.md (input_operand): Add comment. Return
1069 true for any memory operand when LRA is in progress.
1070 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
1072 2017-04-18 Jeff Law <law@redhat.com>
1075 * mips.md ({return,simple_return}_internal): Do not overwrite
1078 2017-04-18 Jakub Jelinek <jakub@redhat.com>
1080 PR tree-optimization/80443
1081 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
1082 instead of adding 1, subtract -1 and similarly instead of subtracting
1085 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
1087 PR rtl-optimization/80357
1088 * haifa-sched.c (tmp_bitmap): New variable.
1089 (model_recompute): Handle duplicate use records.
1090 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
1091 (free_global_sched_pressure_data): Free it.
1093 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
1096 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1097 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1098 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1099 instead of SYSTEM_HEADER_DIR.
1101 2017-04-18 Jeff Law <law@redhat.com>
1104 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
1105 predecessors after walking up the insn chain.
1107 2017-04-18 Jakub Jelinek <jakub@redhat.com>
1110 * dwarf2out.c (modified_type_die): Try harder not to emit internal
1111 sizetype type into debug info.
1113 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1116 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
1117 unneeded test for TARGET_UPPER_REGS_SF.
1118 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
1120 2017-04-18 Jakub Jelinek <jakub@redhat.com>
1123 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
1124 instead of gsi_after_labels.
1126 2017-04-18 Jeff Law <law@redhat.com>
1128 * regcprop.c (maybe_mode_change): Avoid creating copies of the
1132 2017-04-13 Jeff Law <law@redhat.com>
1133 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
1134 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
1136 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
1139 * config/avr/avr.c (intl.h): Include it.
1140 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
1142 2017-04-18 Martin Liska <mliska@suse.cz>
1144 PR gcov-profile/78783
1145 * gcov-tool.c (gcov_output_files): Validate that destination
1146 file is either removed by the tool or by a user.
1148 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
1149 Guy Benyei <guybe@mellanox.com>
1151 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
1152 block, and do not negate it, the stored id is already negative.
1154 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
1156 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
1158 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1161 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
1162 masks of options that should be turned off if the VSX vector
1163 options are turned off.
1164 (OTHER_P8_VECTOR_MASKS): Likewise.
1165 (OTHER_VSX_VECTOR_MASKS): Likewise.
1166 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
1167 rs6000_disable_incompatible_switches to validate no type switches
1169 (rs6000_incompatible_switch): New function to disallow turning on
1170 other vector options if -mno-vsx, -mno-power8-vector, or
1171 -mno-power9-vector are specified.
1173 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
1175 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
1177 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
1179 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
1180 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
1181 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
1182 (ARG_POINTER_CFA_OFFSET): Likewise.
1184 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
1186 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
1187 conditions to take advantage of various optimizations.
1189 2017-04-13 Jeff Law <law@redhat.com>
1191 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
1192 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
1193 (zero_extendsidi2_dext): Likewise.
1195 2017-04-13 Jakub Jelinek <jakub@redhat.com>
1198 * fold-const.c (fold_ternary_loc): Revert
1199 use op0 instead of fold_convert_loc (loc, type, arg0) part of
1202 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
1204 PR rtl-optimization/80343
1205 * lra-remat.c (update_scratch_ops): Assign original hard reg to
1208 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
1211 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
1212 to ubsan_encode_value.
1214 2017-04-13 Jeff Law <law@redhat.com>
1216 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
1217 appearing in DEBUG_INSNs.
1219 2017-04-13 Martin Liska <mliska@suse.cz>
1221 PR gcov-profile/80413
1222 * gcov-io.c (gcov_write_string): Copy to buffer just when
1223 allocated size is greater than zero.
1225 2017-04-13 Jakub Jelinek <jakub@redhat.com>
1228 * dwarf2out.c (decls_for_scope): Ignore declarations of
1229 current_function_decl in BLOCK_NONLOCALIZED_VARS.
1231 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
1234 * ipa-visibility.c (non_local_p): Fix typos.
1235 (localize_node): When localizing symbol in same comdat group,
1236 dissolve the group only when we know external symbols are going
1238 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
1240 2017-04-12 Jakub Jelinek <jakub@redhat.com>
1242 PR tree-optimization/79390
1243 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
1244 order does not result in usable sequence, retry with reversed operand
1250 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
1251 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
1252 op0 instead of fold_convert_loc (loc, type, arg0).
1254 2017-04-12 Jeff Law <law@redhat.com>
1256 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
1257 has a delay slot in the generated code.
1259 * config/cris/cris.md (cris_preferred_reload_class): Return
1260 GENNONACR_REGS rather than GENERAL_REGS.
1262 2017-04-12 Jakub Jelinek <jakub@redhat.com>
1265 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
1266 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
1267 signedness of the result type.
1269 2017-04-12 Richard Biener <rguenther@suse.de>
1270 Jeff Law <law@redhat.com>
1272 PR tree-optimization/80359
1273 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
1274 trim stores to TARGET_MEM_REFs.
1276 2017-04-12 Richard Biener <rguenther@suse.de>
1278 PR tree-optimization/79390
1279 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
1280 threading case even more.
1282 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
1285 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
1286 for quad_address_p for TImode, instead of just not indexed_address.
1288 2017-04-12 Richard Biener <rguenther@suse.de>
1289 Bernd Edlinger <bernd.edlinger@hotmail.de>
1292 * alias.c (component_uses_parent_alias_set_from): Handle
1293 TYPE_TYPELESS_STORAGE.
1294 (get_alias_set): Likewise.
1295 * tree-core.h (tree_type_common): Add typeless_storage flag.
1296 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
1297 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
1298 for types containing members with TYPE_TYPELESS_STORAGE.
1299 (place_field): Likewise.
1300 (layout_type): Likewise for ARRAY_TYPE.
1301 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
1302 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
1303 TYPE_TYPELESS_STORAGE.
1304 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
1306 2017-04-12 Jakub Jelinek <jakub@redhat.com>
1309 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
1310 first argument to type.
1312 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1316 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
1317 CONST0_RTX (mode) rather than const0_rtx where appropriate.
1318 (rs6000_expand_binop_builtin): Likewise.
1319 (rs6000_expand_ternop_builtin): Likewise; also add missing
1320 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
1321 vshasigma built-ins.
1322 * doc/extend.texi: Document that vec_xxpermdi's third argument
1325 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
1327 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
1328 Use shift_const cost parameter when calculating gain of STV shifts.
1330 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
1332 PR rtl-optimization/70478
1333 * lra-constraints.c (process_alt_operands): Check memory for
1334 disfavoring memory insn operand.
1336 2017-04-11 Jakub Jelinek <jakub@redhat.com>
1339 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
1340 left shift in unsigned HOST_WIDE_INT type.
1342 PR rtl-optimization/80385
1343 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
1344 (not (neg X)) into (plus X -1) for complex or non-integral modes.
1347 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
1348 if they have any depend clauses.
1350 2017-04-11 Martin Liska <mliska@suse.cz>
1353 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
1354 * ipa-split.c (split_function): Create a local comdat symbol
1355 if caller is in a comdat group.
1357 2017-04-11 Martin Liska <mliska@suse.cz>
1360 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
1363 2017-04-11 Martin Sebor <msebor@redhat.com>
1366 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
1367 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
1369 (directive::set_width, directive::set_precision, format_character):
1371 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
1374 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
1377 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
1378 conflict, set target->arch_name instead of target->cpu_name.
1380 2017-04-11 Richard Biener <rguenther@suse.de>
1382 PR tree-optimization/80374
1383 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
1384 build_zero_cst, remove fold_convertible_p check again.
1386 2017-04-11 Martin Liska <mliska@suse.cz>
1389 * ubsan.c (instrument_object_size): Do not instrument register
1392 2017-04-11 Jakub Jelinek <jakub@redhat.com>
1395 * config/i386/i386-builtin-types.def
1396 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
1397 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
1398 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
1399 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
1400 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
1401 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
1402 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
1403 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
1404 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
1405 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
1406 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
1407 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
1408 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
1409 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
1410 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
1411 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
1412 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
1413 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
1414 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
1415 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
1416 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
1417 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
1418 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
1419 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
1420 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
1421 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
1422 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
1423 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
1424 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
1425 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
1426 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
1427 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
1428 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
1429 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
1430 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
1431 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
1432 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
1433 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
1434 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
1435 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
1436 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
1437 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
1438 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
1439 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
1440 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
1441 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
1443 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
1444 flag to second_arg_count, handle 4 argument function type _COUNT
1445 aliases, handle second_arg_count on second argument rather than last.
1447 2017-04-10 Jeff Law <law@redhat.com>
1449 PR tree-optimization/80374
1450 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
1451 record anything if we can not convert integer_zero_node to the
1454 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
1457 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1458 Enhance special handling given to the TARGET_P9_MINMAX option in
1459 relation to certain other options.
1461 2017-04-10 Bin Cheng <bin.cheng@arm.com>
1463 PR tree-optimization/80153
1464 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
1465 remove POINTER_PLUS_EXPR's base part directly, rather than through
1468 2017-04-10 Richard Biener <rguenther@suse.de>
1469 Bin Cheng <bin.cheng@arm.com>
1471 PR tree-optimization/80153
1472 * tree-affine.c (aff_combination_to_tree): Get base pointer from
1473 the first element of pointer type aff_tree. Build result expr in
1475 (add_elt_to_tree): Convert to type unconditionally. Remove other
1477 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
1478 (rewrite_use_nonlinear_expr): Check invariant using iv information.
1480 2017-04-10 Richard Biener <rguenther@suse.de>
1482 * tree-ssa-structalias.c (find_func_aliases): Properly handle
1485 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
1487 PR rtl-optimization/70478
1488 * lra-constraints.c (curr_small_class_check): New.
1489 (update_and_check_small_class_inputs): New.
1490 (process_alt_operands): Update curr_small_class_check. Disfavor
1491 alternative insn memory operands. Check available regs for small
1494 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
1497 * config/mips/mips.opt (-mvirt): Update description.
1498 * doc/invoke.texi (-mvirt): Likewise.
1500 2017-04-10 Richard Biener <rguenther@suse.de>
1503 * fold-const.c (fold_binary_loc): Look at unstripped ops when
1504 looking for NEGATE_EXPR in -A / -B to A / B folding.
1506 2017-04-10 Martin Liska <mliska@suse.cz>
1508 PR gcov-profile/80224
1509 * gcov.c (print_usage): Fix usage string.
1510 (get_gcov_intermediate_filename): Remove.
1511 (output_gcov_file): Use both for normal and intermediate format.
1512 (generate_results): Do not initialize special file for
1513 intermediate format.
1515 2017-04-10 Richard Biener <rguenther@suse.de>
1517 PR tree-optimization/80304
1518 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
1521 2017-04-10 Nathan Sidwell <nathan@acm.org>
1524 * config/rs6000/rs6000.c (rs6000_vector_type): New.
1525 (rs6000_init_builtins): Use it.
1527 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1529 * config/arm/arm.md (<mrc>): Add mode to SET source.
1532 2017-04-10 Richard Biener <rguenther@suse.de>
1535 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
1537 2017-04-10 Jakub Jelinek <jakub@redhat.com>
1540 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
1541 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
1542 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
1543 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
1544 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
1545 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
1546 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
1547 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
1548 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
1549 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
1550 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
1551 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
1552 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
1553 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
1554 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
1555 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
1556 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
1557 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
1558 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
1559 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
1560 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
1561 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
1562 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
1564 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
1566 PR rtl-optimization/70478
1567 * lra-constraints.c: Reverse the last patch.
1569 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
1571 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
1572 Add comment for WCHAR_T.
1574 2017-04-08 Martin Liska <mliska@suse.cz>
1577 2017-04-07 Martin Liska <mliska@suse.cz>
1580 * ipa-split.c (split_function): Add function part to a same comdat
1583 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1586 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
1588 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
1590 * rs6000/rs6000.c (vec_load_pendulum): Rename...
1591 (vec_pairing): ...to this.
1592 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
1593 (rs6000_sched_init): Adjust for name change.
1594 (struct rs6000_sched_context): Likewise.
1595 (rs6000_init_sched_context): Likewise.
1596 (rs6000_set_sched_context): Likewise.
1598 2017-04-07 Jakub Jelinek <jakub@redhat.com>
1604 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
1606 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
1607 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
1608 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
1610 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
1612 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
1614 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
1616 PR rtl-optimization/70703
1617 * ira-color.c (update_conflict_hard_regno_costs): Use
1618 int64_t instead of HOST_WIDE_INT.
1620 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
1622 PR rtl-optimization/70478
1623 * lra-constraints.c (process_alt_operands): Disfavor alternative
1624 insn memory operands.
1626 2017-04-07 Jeff Law <law@redhat.com>
1628 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
1629 CALL and NOTE_INSN_CALL_ARG_LOCATION.
1631 2017-04-07 Martin Liska <mliska@suse.cz>
1634 * config/aarch64/aarch64.c (aarch64_process_target_attr):
1635 Show error message instead of an ICE.
1637 2017-04-07 Martin Liska <mliska@suse.cz>
1640 * ipa-split.c (split_function): Add function part to a same comdat
1643 2017-04-07 Richard Biener <rguenther@suse.de>
1646 * tree.c (get_unwidened): Also handle ! for_type case for
1648 * convert.c (do_narrow): Split out from ...
1649 (convert_to_integer_1): ... here. Do not pass final truncation
1650 type to get_unwidened for TRUNC_DIV_EXPR.
1652 2017-04-07 Richard Biener <rguenther@suse.de>
1654 * tree-affine.c (wide_int_ext_for_comb): Take type rather
1656 (aff_combination_const): Adjust.
1657 (aff_combination_scale): Likewise.
1658 (aff_combination_add_elt): Likewise.
1659 (aff_combination_add_cst): Likewise.
1660 (aff_combination_convert): Likewise.
1661 (add_elt_to_tree): Likewise. Remove unused argument.
1662 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
1664 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
1666 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
1668 * config/arm/arm.c (arm_default_short_enums): Use
1669 ARM_DEFAULT_SHORT_ENUMS.
1670 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
1672 2017-04-06 Jakub Jelinek <jakub@redhat.com>
1675 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
1676 members with redundant out-of-class redeclaration.
1678 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
1681 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
1682 * config/i386/i386.md (*zero_extendsidi2):
1683 Add (?*x,*x) and (?*v,*v) alternatives.
1685 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
1688 * config/i386/i386.c (ix86_expand_builtin)
1689 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
1690 mode from insn data. Convert operands to insn operand mode.
1691 Copy operands that don't satisfy insn predicate to a register.
1693 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
1695 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
1698 2017-04-06 Richard Biener <rguenther@suse.de>
1700 PR tree-optimization/80334
1701 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
1702 preserve alignment of accesses.
1704 2017-04-06 Richard Biener <rguenther@suse.de>
1706 PR tree-optimization/80262
1707 * tree-sra.c (build_ref_for_offset): Preserve address-space
1709 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
1710 Drop useless address-space information on MEM_REF offsets.
1712 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
1714 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
1716 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
1718 PR rtl-optimization/70703
1719 * ira-color.c (update_conflict_hard_regno_costs): Use
1720 HOST_WIDE_INT instead of long.
1722 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
1725 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
1726 not defined for x86_64 target. Add -mmmx target option when __SSE2__
1728 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
1729 for x86_64 target. Handle -m3dnowa option.
1731 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
1733 PR rtl-optimization/70703
1734 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
1735 (update_conflict_hard_regno_costs): Use long instead of unsigned
1736 arithmetic for cost calculation.
1738 2017-04-05 Jakub Jelinek <jakub@redhat.com>
1739 Bernd Edlinger <bernd.edlinger@hotmail.de>
1742 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
1745 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
1748 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
1749 ptr_mode with Pmode throughout.
1750 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
1751 into probe_stack_range and use DImode.
1753 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
1756 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
1757 call_eh_return is true.
1759 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1761 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
1762 Initialize last_match_fntype_index.
1764 2017-04-05 Jakub Jelinek <jakub@redhat.com>
1767 * tree-nvr.c: Include internal-fn.h.
1768 (pass_return_slot::execute): Ignore internal calls without
1771 2017-04-04 Jakub Jelinek <jakub@redhat.com>
1772 Richard Biener <rguenther@suse.de>
1775 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
1776 captures used multiple times, except for the last use.
1777 * generic-match-head.c: Include gimplify.h.
1779 2017-04-04 Jakub Jelinek <jakub@redhat.com>
1781 PR tree-optimization/79390
1782 * target.h (struct noce_if_info): Declare.
1783 * targhooks.h (default_noce_conversion_profitable_p): Declare.
1784 * target.def (noce_conversion_profitable_p): New target hook.
1785 * ifcvt.h (struct noce_if_info): New type, moved from ...
1786 * ifcvt.c (struct noce_if_info): ... here.
1787 (noce_conversion_profitable_p): Renamed to ...
1788 (default_noce_conversion_profitable_p): ... this. No longer
1790 (noce_try_store_flag_constants, noce_try_addcc,
1791 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
1792 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
1793 instead of noce_conversion_profitable_p.
1794 * config/i386/i386.c: Include ifcvt.h.
1795 (ix86_option_override_internal): Don't override
1796 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
1797 (ix86_noce_conversion_profitable_p): New function.
1798 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
1799 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
1800 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
1801 * doc/tm.texi: Regenerated.
1803 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1805 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
1808 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
1811 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
1812 instructions for small multiply cores.
1814 2017-04-04 Jeff Law <law@redhat.com>
1816 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
1818 (mips_expand_vec_perm_const): Initialize elements in orig_perm
1819 that are not set by the loop over the elements.
1821 2017-04-04 Jakub Jelinek <jakub@redhat.com>
1824 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
1825 int mode, convert_modes it to mode as unsigned, otherwise use
1826 lowpart_subreg to mode rather than SImode.
1827 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
1828 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
1829 Use DImode instead of SImode for the shift count operand.
1830 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
1833 2017-04-04 Richard Biener <rguenther@suse.de>
1836 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
1837 arithmetic done for the negate or the plus. Simplify.
1838 (A - (-B) -> A + B): Likewise.
1839 * fold-const.c (split_tree): Make sure to not negate pointers.
1841 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
1843 PR rtl-optimization/60818
1844 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
1845 a compare of comparisons with the thing compared if this results
1846 in a different machine mode.
1848 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
1850 * alias.c (base_alias_check): Fix typo in comment.
1851 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
1852 * cgraphunit.c (symbol_table::compile): Likewise.
1853 * collect2.c (maybe_run_lto_and_relink): Likewise.
1854 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
1855 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
1856 * config/avr/avr.c (avr_map_op_t): Likewise.
1857 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
1858 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
1859 * config/epiphany/epiphany.md (movcc): Likewise.
1860 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
1861 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
1863 * config/mips/mips.c (mips_save_restore_reg): Likewise.
1864 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
1865 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
1866 * config/sh/sh.c (sh_rtx_costs): Likewise.
1867 * fold-const.c (fold_truth_andor): Likewise.
1868 * genautomata.c (collapse_flag): Likewise.
1869 * gengtype.h (struct type::u::s): Likewise.
1870 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
1871 * input.c (FORMAT_AMOUNT): Likewise.
1872 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
1873 (known_aggs_to_agg_replacement_list): Likewise.
1874 * ipa-inline-analysis.c: Likewise.
1875 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
1876 * ipa-polymorphic-call.c
1877 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
1878 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
1879 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
1881 * modulo-sched.c (apply_reg_moves): Likewise.
1882 * omp-expand.c (build_omp_regions_1): Likewise.
1883 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
1884 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
1885 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
1886 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
1887 * value-prof.c: Likewise.
1888 * var-tracking.c (val_reset): Likewise.
1890 2017-04-03 Richard Biener <rguenther@suse.de>
1892 PR tree-optimization/80275
1893 * fold-const.c (split_address_to_core_and_offset): Handle
1896 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
1898 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
1899 descriptors is at least equal to that of functions.
1901 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
1903 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
1905 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
1908 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
1909 (mov<IMOD4:mode>): New expander.
1910 (*mov<IMOD4:mode>_internal): New insn and split pattern.
1912 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
1914 PR rtl-optimization/79405
1915 * fwprop.c (propagations_left): New variable.
1916 (forward_propagate_into): Decrement it.
1917 (fwprop_init): Initialize it.
1918 (fw_prop): If the variable has reached zero, stop propagating.
1919 (fwprop_addr): Ditto.
1921 2017-03-31 Jakub Jelinek <jakub@redhat.com>
1924 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
1925 a FUNCTION_DECL, pass it as decl instead of origin to
1928 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
1930 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
1933 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
1936 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
1937 TARGET_VSX_SMALL_INTEGER.
1939 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1941 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1942 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
1944 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
1946 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
1947 extraction from odd-numbered MSA register.
1949 2017-03-31 Jakub Jelinek <jakub@redhat.com>
1952 * expmed.c (store_bit_field_1): Don't attempt to create
1953 a word subreg out of hard registers wider than word if they
1954 have HARD_REGNO_NREGS of 1 for their mode.
1957 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
1958 conversions to integer types wider than word and pointer.
1961 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
1962 (rtx_equal_for_cselib_p): Pass 0 to it.
1963 * cselib.c (cselib_hasher::equal): Likewise.
1964 (rtx_equal_for_cselib_1): Add depth argument. If depth
1965 is 128, don't look up VALUE locs and punt. Increment
1966 depth in recursive calls when walking VALUE locs.
1968 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
1970 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
1971 (make_gcov_file_name): Use the canonical path name for generating
1973 (read_line): Fix handling of files with ascii null bytes.
1975 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
1977 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
1978 to initialise a vector register instead
1979 of using a const_int.
1981 2017-03-30 Jakub Jelinek <jakub@redhat.com>
1983 PR translation/80189
1984 * gimplify.c (omp_default_clause): Use %qs instead of %s in
1985 diagnostic messages.
1987 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
1990 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
1991 (dfp_diex_<mode>): Update mode of operand 1.
1992 * doc/extend.texi (dxex, dxexq): Document change to return type.
1993 (diex, diexq): Document change to argument type.
1995 2017-03-30 Martin Jambor <mjambor@suse.cz>
1998 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
1999 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
2000 it reflects the signature changes performed at the callee side.
2001 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
2002 to cgraph_build_function_type_skip_args.
2003 (build_function_decl_skip_args): Adjust call to the above function.
2005 2017-03-30 Jakub Jelinek <jakub@redhat.com>
2008 * config/i386/sse.md
2009 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
2010 register as dest whenever it is a MEM not rtx_equal_p to the
2011 corresponding dup operand, and when forcing into reg move the
2012 reg into the memory afterwards.
2013 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
2014 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
2015 for the force_reg mode.
2016 (avx512vl_vextractf128<mode>): Use register as dest either
2017 always when a MEM, or when it is a MEM not rtx_equal_p to the
2018 corresponding dup operand, or even not when it is a CONST_VECTOR
2019 depending on the mode and lo vs. hi.
2020 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
2022 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
2023 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
2024 Likewise. Require that operands[2] is even.
2025 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
2026 Remove extraneous parens. Require that operands[2] is a multiple
2028 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
2029 operands[0] is a MEM if <mask_applied>, the predicates/constraints
2030 disallow memory then.
2032 2017-03-30 Richard Biener <rguenther@suse.de>
2034 PR tree-optimization/77498
2035 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
2036 to non-constants over backedges.
2038 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
2040 PR rtl-optimization/80233
2041 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
2042 as last_combined_insn. Do not test for BARRIER_P separately.
2044 2017-03-29 Andreas Schwab <schwab@suse.de>
2047 * calls.c (prepare_call_address): Convert funexp to Pmode before
2048 copying to temp reg.
2050 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2052 PR tree-optimization/80158
2053 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
2054 Handle possible future case of more than one alternate
2056 (replace_rhs_if_not_dup): Likewise.
2057 (replace_one_candidate): Likewise.
2059 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
2061 PR rtl-optimization/80193
2062 * ira.c (ira): Do not check allocation for LRA.
2064 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
2066 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
2067 (nvptx_output_simt_exit): Declare.
2068 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
2069 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
2070 (init_softstack_frame): Move initialization of crtl->is_leaf to...
2071 (nvptx_declare_function_name): ...here. Emit declaration of local
2072 memory space buffer for omp_simt_enter insn.
2073 (nvptx_output_unisimt_switch): New.
2074 (nvptx_output_softstack_switch): New.
2075 (nvptx_output_simt_enter): New.
2076 (nvptx_output_simt_exit): New.
2077 * config/nvptx/nvptx.h (struct machine_function): New fields
2078 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
2079 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
2080 (UNSPECV_SIMT_EXIT): Ditto.
2081 (omp_simt_enter_insn): New insn.
2082 (omp_simt_enter): New expansion.
2083 (omp_simt_exit): New insn.
2084 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
2086 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
2087 (expand_GOMP_SIMT_ENTER_ALLOC): New.
2088 (expand_GOMP_SIMT_EXIT): New.
2089 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
2090 (GOMP_SIMT_ENTER_ALLOC): Ditto.
2091 (GOMP_SIMT_EXIT): Ditto.
2092 * target-insns.def (omp_simt_enter): New insn.
2093 (omp_simt_exit): Ditto.
2094 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
2096 (lower_rec_simd_input_clauses): Implement SIMT privatization.
2097 (lower_rec_input_clauses): Likewise.
2098 (lower_lastprivate_clauses): Handle SIMT privatization.
2100 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
2101 (ompdevlow_adjust_simt_enter): New.
2102 (find_simtpriv_var_op): New.
2103 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
2104 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
2106 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
2107 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
2108 (copy_decl_for_dup_finish): Ditto.
2110 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
2112 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
2115 * config/i386/i386.c (ix86_option_override_internal): Always
2116 allow -mpreferred-stack-boundary=3 for 64-bit targets.
2118 2017-03-28 Bin Cheng <bin.cheng@arm.com>
2120 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
2122 2017-03-28 Bin Cheng <bin.cheng@arm.com>
2124 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
2125 mark new edge's irreducible flag accordign to it.
2126 (vect_do_peeling): Check loop preheader edge's irreducible flag
2127 and pass it to function slpeel_add_loop_guard.
2129 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
2131 PR tree-optimization/80218
2132 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
2133 Update block frequencies and counts.
2135 2017-03-28 Richard Biener <rguenther@suse.de>
2137 PR tree-optimization/78644
2138 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
2139 of a simplification result we may not use it at all.
2141 2017-03-28 Richard Biener <rguenther@suse.de>
2144 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
2145 without arguments, generate default definition of a SSA name.
2147 2017-03-28 Richard Biener <rguenther@suse.de>
2150 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
2151 TYPE_REF_CAN_ALIAS_ALL references.
2152 * fold-const.c (fold_indirect_ref_1): Likewise.
2154 2017-03-28 Martin Liska <mliska@suse.cz>
2157 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
2158 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
2160 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
2161 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
2163 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
2164 (EXTRA_SPECS): Define.
2165 (SUBTARGET_EXTRA_SPECS): Likewise.
2166 (SUBTARGET_CPP_SPEC): Likewise.
2167 * config/arc/elf.h (EXTRA_SPECS): Renamed to
2168 SUBTARGET_EXTRA_SPECS.
2169 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
2171 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
2173 * config/arc/simdext.md (vst64_insn): Update pattern.
2174 (vld32wh_insn): Likewise.
2175 (vld32wl_insn): Likewise.
2176 (vld64_insn): Likewise.
2177 (vld32_insn): Likewise.
2179 2017-03-28 Marek Polacek <polacek@redhat.com>
2182 * fold-const.c (fold_comparison): Use protected_set_expr_location
2183 instead of SET_EXPR_LOCATION.
2185 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
2187 * tree.c (add_expr): Avoid name lookup warning.
2189 2017-03-27 Jeff Law <law@redhat.com>
2191 PR tree-optimization/80216
2192 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
2193 function name. Limit recursion depth.
2194 (record_temporary_equivalences): Corresponding changes.
2196 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
2198 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
2201 2017-03-27 Jakub Jelinek <jakub@redhat.com>
2204 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
2206 * cfgcleanup.c (reg_note_cfa_p): New array.
2207 (insns_have_identical_cfa_notes): New function.
2208 (old_insns_match_p): Don't cross-jump in between /f
2209 and non-/f instructions. If both i1 and i2 are frame related,
2210 verify all CFA notes, their order and content.
2212 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2215 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
2216 HImode and SImode with zero extend to DImode to one insn.
2217 (bswap<mode>2_extenddi): Likewise.
2218 (bswapsi2_extenddi): Likewise.
2219 (bswaphi2_extendsi): Likewise.
2220 (bswaphi2): Combine bswap HImode and SImode into one insn.
2221 Separate memory insns from swapping register.
2222 (bswapsi2): Likewise.
2223 (bswap<mode>2): Likewise.
2224 (bswaphi2_internal): Delete, no longer used.
2225 (bswapsi2_internal): Likewise.
2226 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
2227 store, and gpr<-gpr swap insns.
2228 (bswap<mode>2_store): Likewise.
2229 (bswaphi2_reg): Register only splitter, combine with the splitter.
2230 (bswaphi2 splitter): Likewise.
2231 (bswapsi2_reg): Likewise.
2232 (bswapsi2 splitter): Likewise.
2233 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
2234 the insns into load, store, and register/register insns.
2235 (bswapdi2_ldbrx): Likewise.
2236 (bswapdi2_load): Likewise.
2237 (bswapdi2_store): Likewise.
2238 (bswapdi2_reg): Likewise.
2240 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
2242 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
2243 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
2245 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2248 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
2250 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2251 special handling for target option conflicts between dform
2252 options (-mpower9-dform, -mpower9-dform-vector,
2253 -mpower9-dform-scalar) and -mno-direct-move.
2255 2017-03-27 Richard Biener <rguenther@suse.de>
2257 PR tree-optimization/80181
2258 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
2260 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
2262 * config/arc/predicates.md (move_double_src_operand): Replace the
2263 call to move_double_src_operand with a call to address_operand.
2265 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
2267 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
2268 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
2269 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
2271 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
2273 * config/arc/predicates.md (long_immediate_loadstore_operand):
2274 Consider scaled addresses cases.
2276 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
2278 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
2279 restored when in interrupt.
2280 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
2281 doesn't have delay slot.
2283 2017-03-27 Richard Biener <rguenther@suse.de>
2286 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
2287 inlined thunk clones.
2289 2017-03-27 Jakub Jelinek <jakub@redhat.com>
2292 * asan.c (instrument_derefs): Copy over last operand from
2293 original COMPONENT_REF to the new COMPONENT_REF with
2294 DECL_BIT_FIELD_REPRESENTATIVE.
2295 * ubsan.c (instrument_object_size): Likewise.
2297 2017-03-27 Richard Biener <rguenther@suse.de>
2299 PR tree-optimization/80170
2300 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
2301 sure DR/SCEV didnt fold in constants we do not see when looking
2302 at the reference base alignment.
2304 2017-03-27 Richard Biener <rguenther@suse.de>
2307 * gimple-fold.c (fold_ctor_reference): Properly guard against
2308 NULL return value from canonicalize_constructor_val.
2310 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
2313 * config/i386/i386.c (ix86_expand_builtin)
2314 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
2315 flags reg setting and flags reg using instructions.
2316 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
2317 clobbering instructions to zero extend op2.
2319 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
2321 * doc/install.texi (Configuration) <--with-aix-soname>:
2322 Update link to AIX ld.
2324 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
2326 PR rtl-optimization/80160
2327 PR rtl-optimization/80159
2328 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
2329 reg_alternate_class into account.
2331 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
2334 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
2335 to consider in curr_insn_transform.
2337 2017-03-24 Jakub Jelinek <jakub@redhat.com>
2339 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
2340 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
2341 and emit_mode_inner.
2343 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2345 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
2346 argument to the overloaded builtin variants. Use the new flag to
2347 deprecate certain builtin variants.
2348 * config/s390/s390-builtin-types.def: Add new builtin types.
2349 * config/s390/s390-builtins.h: Support new flags field for
2350 overloaded builtins.
2351 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
2352 (s390_macro_to_expand): Enable vector float data type.
2353 (s390_cpu_cpp_builtins_internal): Indicate support of the new
2354 builtins by incrementing the __VEC__ version number.
2355 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
2357 (s390_resolve_overloaded_builtin): Emit error messages depending
2358 on the builtin flags.
2359 * config/s390/s390.c (s390_expand_builtin): Support additional
2360 flags argument. Change error message to match the messages
2361 emitted in s390-c.c.
2362 * config/s390/s390.md: New UNSPEC_* constants.
2363 (op_type): Add new instruction types.
2364 * config/s390/vecintrin.h: Add new builtins and test data class
2366 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
2367 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
2368 (VEC_INEXACT, VEC_NOINEXACT): New constants.
2369 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
2370 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
2371 ("vec_mergel<mode>"): V_HW -> VEC_HW.
2373 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
2374 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
2375 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
2376 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
2378 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
2379 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
2380 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
2381 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
2383 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
2384 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
2385 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
2386 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
2387 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
2388 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
2389 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
2391 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
2392 ("vec_scatter_element<V_HW_4:mode>_DI")
2393 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
2394 ("vec_fpint<mode>", "vflls")
2395 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
2396 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
2397 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
2398 ("*vec_cmphe<mode>_cc"): ... these.
2400 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
2401 mode constant instead of magic value.
2403 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2405 * config/s390/s390.c (s390_expand_vec_compare): Support other
2406 vector floating point modes than just V2DF.
2407 (s390_expand_vcond): Likewise.
2408 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
2409 (s390_cannot_change_mode_class): Prevent mode changes between TF
2410 and V1TF in vector registers.
2411 * config/s390/s390.md (DF, SF): New mode attributes.
2412 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
2413 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
2414 SFmode support for VRs.
2415 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
2417 (VFT, VF_HW): New mode iterators.
2418 (vw, sdx): New mode attributes.
2419 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
2420 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
2421 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
2422 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
2423 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
2424 also the new vector floating point modes. Renaming to ...
2426 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
2427 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
2428 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
2429 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
2430 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
2431 ("vec_unordered<mode>"): ... these.
2433 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
2434 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
2435 ("*vec_extendv2df"): New insn definitions.
2437 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2439 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
2440 ("mulditi3_2", "*muldi3_sign"): New patterns.
2441 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
2442 rename the pattern definition.
2444 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2446 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
2448 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
2450 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2452 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
2453 instruction if possible.
2454 * config/s390/vector.md (vec_halfnumelts): New mode
2456 ("*vec_vllezlf<mode>"): New pattern.
2458 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2460 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
2461 ("popcountv4si2", "popcountv2di2"): Rename to ...
2462 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
2463 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
2465 ("popcount<mode>2_vxe"): New pattern.
2467 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2469 * common/config/s390/s390-common.c (processor_flags_table): Add
2471 * config.gcc: Add arch12.
2472 * config/s390/driver-native.c (s390_host_detect_local_cpu):
2473 Default to arch12 for unknown CPU model numbers.
2474 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
2475 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
2476 PROCESSOR_max sanity check.
2477 * config/s390/s390-opts.h (enum processor_type): Add
2479 * config/s390/s390.c (processor_table): Add arch12.
2480 (s390_expand_builtin): Add check for B_VXE flag.
2481 (s390_issue_rate): Add PROCESSOR_ARCH12.
2482 (s390_get_sched_attrmask): Likewise.
2483 (s390_get_unit_mask): Likewise.
2484 (s390_sched_score): Enable z13 scheduling for arch12.
2485 (s390_sched_reorder): Likewise.
2486 (s390_sched_variable_issue): Likewise.
2487 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
2489 (s390_tune_attr): Use z13 scheduling also for arch12.
2490 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
2491 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
2492 (TARGET_VXE_P): New macros.
2493 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
2494 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
2495 * config/s390/s390.opt: Add arch12 as processor_type.
2497 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2499 * config/s390/s390.md
2500 ("fixuns_truncdddi2", "fixuns_trunctddi2")
2501 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
2502 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
2504 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
2505 Rename expanders to ...
2507 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
2508 ("fixuns_truncdddi2_emu"): ... these.
2510 ("fixuns_trunc<mode>si2_emu"): New expander.
2512 ("*fixuns_truncdfdi2_z13"): Rename to ...
2513 ("*fixuns_truncdfdi2_vx"): ... this.
2515 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2517 * config/s390/2964.md: Remove the single element vector compare
2518 instructions which are no longer used.
2519 * config/s390/s390.c (s390_select_ccmode): Remove handling of
2521 (s390_canonicalize_comparison): Remove handling of DFmode
2523 (s390_expand_vec_compare_scalar): Remove function.
2524 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
2525 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
2527 ("*cmp<mode>_ccs"): Add wfcdb instruction.
2529 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2531 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
2533 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
2534 will anyway by matched by mov<mode>_64dfp.
2536 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2538 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
2539 vlef/vstef. Add missing operand to vleif.
2541 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2543 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
2544 pair for all vector types with 64 bit elements.
2545 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
2546 * config/s390/vector.md (V_HW_64): ... here.
2547 (V_128_NOSINGLE): New mode iterator.
2548 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
2549 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
2550 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
2551 ("*vec_load_pairv2di"): Change to ...
2552 ("*vec_load_pair<mode>"): ... this one.
2554 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2556 * config/s390/constraints.md: Add comments.
2557 (jKK): Reject element sizes > 8 bytes.
2558 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
2560 * config/s390/s390.md: Add the s_operand checks formerly in
2561 s390_split_ok_p to various splitters where they are still
2563 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
2564 for 128 bit vectors. Plus two splitters.
2566 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2568 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
2571 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2574 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
2575 error if the boundary argument is not constant.
2577 2017-03-24 Jakub Jelinek <jakub@redhat.com>
2579 PR rtl-optimization/80112
2580 * loop-doloop.c (doloop_condition_get): Don't check condition
2581 if cmp isn't SET with IF_THEN_ELSE src.
2583 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2585 PR tree-optimization/80158
2586 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
2587 replacing a candidate statement, also replace it for the
2588 candidate's alternate interpretation.
2589 (replace_rhs_if_not_dup): Likewise.
2590 (replace_one_candidate): Likewise.
2592 2017-03-24 Richard Biener <rguenther@suse.de>
2594 PR tree-optimization/80167
2595 * graphite-isl-ast-to-gimple.c
2596 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
2598 (translate_isl_ast_to_gimple::get_rename): Likewise.
2600 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
2602 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
2603 handling of certain combinations of target options, including the
2604 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
2605 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
2607 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2610 * config/arm/arm.md (*load_multiple): Add reload_completed to
2613 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2614 Richard Biener <rguenth@suse.de>
2616 PR tree-optimization/79908
2617 PR tree-optimization/80136
2618 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
2619 been cast away, gimplify_and_add suffices.
2621 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
2623 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
2625 2017-03-23 Richard Biener <rguenther@suse.de>
2627 PR tree-optimization/80032
2628 * gimplify.c (gimple_push_cleanup): Forced unconditional
2629 cleanups still have to go to the conditional_cleanups
2632 2017-03-22 Jakub Jelinek <jakub@redhat.com>
2634 PR tree-optimization/80072
2635 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
2637 (next_operand_entry_id): Change type to unsigned int.
2638 (sort_by_operand_rank): Make sure to return the right return value
2639 even if unsigned fields are bigger than INT_MAX.
2640 (struct oecount): Change cnt and id type to unsigned int.
2641 (oecount_hasher::equal): Formatting fix.
2642 (oecount_cmp): Make sure to return the right return value
2643 even if unsigned fields are bigger than INT_MAX.
2644 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
2647 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
2648 TREE_READONLY on result if writing it more than once.
2651 * doc/invoke.texi (-fsanitize=thread): Document that with
2652 -fnon-call-exceptions atomics are not able to throw
2656 * tsan.c: Include tree-eh.h.
2657 (instrument_builtin_call): Call maybe_clean_eh_stmt or
2658 maybe_clean_or_replace_eh_stmt where needed.
2659 (instrument_memory_accesses): Add cfg_changed argument.
2660 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
2661 if it returned true.
2662 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
2664 PR rtl-optimization/63191
2665 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
2666 wrapper function, moved the whole old content into ...
2667 (ix86_delegitimize_address_1): ... this. New inline function.
2668 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
2669 true as last argument instead of ix86_delegitimize_address.
2671 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
2673 * config/aarch64/aarch64.c (generic_branch_cost): Copy
2674 cortexa57_branch_cost.
2676 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
2678 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
2680 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2683 * doc/md.texi (Constraints): Document wA constraint.
2684 * config/rs6000/constraints.md (wA): New.
2685 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
2686 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
2687 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
2688 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
2690 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
2693 * gimplify.c (is_oacc_declared): New function.
2694 (oacc_default_clause): Use it to set default flags for acc declared
2695 variables inside parallel regions.
2696 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
2698 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
2699 declare attribute to any decl as necessary.
2701 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2704 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
2705 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
2706 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
2707 (arm_arch_lpae): This.
2708 * config/arm/arm.c (arm_arch7ve): Rename into ...
2709 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
2710 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
2713 2017-03-22 Martin Liska <mliska@suse.cz>
2716 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
2717 error message instead of an ICE.
2719 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2721 * doc/extend.texi (6.11 Additional Floating Types): Revise.
2723 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
2725 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
2727 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2730 2017-03-21 Martin Sebor <msebor@redhat.com>
2732 * doc/extend.texi: Use "cannot" instead of "can't."
2733 * doc/hostconfig.texi: Same.
2734 * doc/install.texi: Same.
2735 * doc/invoke.texi: Same.
2736 * doc/loop.texi: Same.
2737 * doc/md.texi: Same.
2738 * doc/objc.texi: Same.
2739 * doc/rtl.texi: Same.
2740 * doc/tm.texi: Same.
2741 * doc/tm.texi.in: Same.
2742 * doc/trouble.texi: Same.
2744 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
2747 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
2748 (collect_checksum_attributes): Set it.
2749 (die_checksum_ordered): Use it.
2751 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2753 PR tree-optimization/79908
2754 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
2755 change: For a VA_ARG whose LHS has been cast away, use
2756 force_gimple_operand to construct the side effects.
2758 2017-03-21 David Malcolm <dmalcolm@redhat.com>
2760 PR translation/80001
2761 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
2762 more amenable to translation.
2763 (oacc_loop_auto_partitions): Likewise.
2765 2017-03-21 Marek Polacek <polacek@redhat.com>
2766 Martin Sebor <msebor@redhat.com>
2768 PR tree-optimization/80109
2769 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
2772 2017-03-21 Jakub Jelinek <jakub@redhat.com>
2773 Segher Boessenkool <segher@kernel.crashing.org>
2776 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
2777 check reg_used_between_p between insn and one of succ or succ2
2778 depending on if succ is artificial insn not inserted into insn
2781 2017-03-21 Martin Liska <mliska@suse.cz>
2783 PR gcov-profile/80081
2784 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
2785 * doc/gcc.texi: Include gcov-dump stuff.
2786 * doc/gcov-dump.texi: New file.
2788 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
2790 PR rtl-optimization/79150
2791 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
2792 conditional jump, if the jump is the last insn of the loop.
2794 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2795 Richard Biener <rguenth@suse.de>
2797 PR tree-optimization/79908
2798 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
2799 been cast away, use force_gimple_operand to construct the side
2802 2017-03-21 Martin Liska <mliska@suse.cz>
2805 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
2808 2017-03-21 Brad Spengler <spender@grsecurity.net>
2811 * plugin.c (htab_hash_plugin): New function.
2812 (add_new_plugin): Use it and adjust.
2813 (parse_plugin_arg_opt): Adjust.
2814 (init_one_plugin): Likewise.
2816 2017-03-21 Richard Biener <rguenther@suse.de>
2818 PR tree-optimization/80032
2819 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
2820 if set force the cleanup to happen unconditionally.
2821 (gimplify_target_expr): Push inserted clobbers with force_uncond
2822 to avoid them being removed by control-dependent DCE.
2824 2017-03-21 Richard Biener <rguenther@suse.de>
2826 PR tree-optimization/80122
2827 * tree-inline.c (copy_bb): Do not expans va-arg packs or
2828 va_arg_pack_len when the inlined call stmt requires pack
2830 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
2832 2017-03-21 Jakub Jelinek <jakub@redhat.com>
2835 * tsan.c (instrument_builtin_call): If the memory model argument
2836 is not a constant, assume it is valid.
2839 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
2842 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
2844 PR rtl-optimization/79910
2845 * combine.c (can_combine_p): Do not allow combining an I0 or I1
2846 if its dest is used by an insn before I2 (other than the combined
2847 insns themselves, which are properly handled already).
2849 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
2852 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
2854 * combine.c (record_used_regs): New static function.
2855 (try_combine): Handle situations where there is an additional
2856 instruction between I2 and I3 which needs to have a LOG_LINK
2860 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
2862 * combine.c (try_combine): Delete redundant i1 test. Call
2863 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
2865 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2868 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
2871 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2873 PR tree-optimization/80054
2874 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
2875 the optimization if a PHI or any of its arguments is not dominated
2876 by the candidate's basis. Use gphi* rather than gimple* as
2878 (replace_profitable_candidates): Clean up a gimple* variable that
2879 should be a gphi* variable.
2881 2017-03-20 Martin Sebor <msebor@redhat.com>
2884 * doc/extend.texi (attribute constructor): Document present limitation.
2886 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
2889 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
2890 __POWER9_VECTOR__ #ifdef control, change template definition to
2891 use Power9-specific built-in function.
2892 (vec_any_eq): Likewise.
2893 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
2894 to control outcomes from this test.
2895 (vector_ae_<mode>p): For VEC_F modes, likewise.
2897 2017-03-20 Ian Lance Taylor <iant@google.com>
2899 * config/i386/i386.c (ix86_function_regparm): Save an extra
2900 register for -fsplit-stack with DECL_STATIC_CHAIN.
2902 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
2905 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
2906 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
2908 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
2910 * config/riscv/riscv.c (riscv_print_operand): Use "fence
2912 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
2915 2017-03-20 Marek Polacek <polacek@redhat.com>
2918 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
2920 2017-03-20 Richard Biener <rguenther@suse.de>
2922 PR tree-optimization/80113
2923 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
2924 allocate extra SSA name for PHI def.
2925 (add_close_phis_to_outer_loops): Likewise.
2926 (add_close_phis_to_merge_points): Likewise.
2927 (copy_loop_close_phi_args): Likewise.
2928 (copy_cond_phi_nodes): Likewise.
2930 2017-03-20 Martin Liska <mliska@suse.cz>
2933 * tree-chkp.c (chkp_build_returned_bound): Do not build
2934 returned bounds for a LHS that's not a BOUNDED_P type.
2936 2017-03-20 Martin Liska <mliska@suse.cz>
2940 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
2941 COMPLEX_CST and VECTOR_CST.
2943 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2946 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
2947 target operand. A new splitter adds the clobber statement in case
2948 the target operand is dead anyway.
2950 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
2952 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
2953 to age-old versions of binutils and glibc.
2955 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
2957 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
2959 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
2961 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
2963 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
2965 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
2966 requirement for binutils 2.13.
2968 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
2970 * combine.c (try_combine): Delete redundant i1 test. Call
2971 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
2973 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
2975 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
2976 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
2978 <riscv64-*-elf>: Re-arrange section
2979 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
2980 <riscv32-*-linux>: Likewise.
2981 <riscv64-*-elf>: Likewise
2982 <riscv64-*-linux>: Likewise.
2984 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
2987 * aarch64.opt(verbose-cost-dump): Fix typo.
2989 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
2992 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
2993 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
2995 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
2997 * reload.c (find_reloads): When reloading a nonoffsettable address,
2998 use RELOAD_OTHER for it and its address reloads.
3000 PR rtl-optimization/79910
3001 * combine.c (record_used_regs): New static function.
3002 (try_combine): Handle situations where there is an additional
3003 instruction between I2 and I3 which needs to have a LOG_LINK
3006 2017-03-17 Jeff Law <law@redhat.com>
3008 PR tree-optimization/71437
3009 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
3010 conditional in the hash table first.
3011 (vrp_dom_walker::before_dom_children): Extract condition from
3012 ASSERT_EXPR. Record condition, its inverion and any implied
3015 2017-03-17 Marek Polacek <polacek@redhat.com>
3016 Markus Trippelsdorf <markus@trippelsdorf.de>
3018 PR tree-optimization/80079
3019 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
3022 2017-03-17 Richard Biener <rguenther@suse.de>
3025 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
3026 Properly verify the LHS before the RHS possibly claims to be
3028 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
3031 2017-03-17 Martin Jambor <mjambor@suse.cz>
3033 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
3034 (List of -O2 options): Likewise.
3035 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
3038 2017-03-17 Tom de Vries <tom@codesourcery.com>
3040 * gcov-dump.c (print_usage): Print bug_report_url.
3042 2017-03-17 Richard Biener <rguenther@suse.de>
3045 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
3046 (parser::peek): Likewise.
3048 2017-03-17 Richard Biener <rguenther@suse.de>
3050 PR tree-optimization/80048
3051 * sese.c (free_sese_info): Properly release rename_map and
3052 copied_bb_map elements.
3054 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
3056 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
3057 Add linked-list forward and backlinks. Insert on
3058 construction, remove on destruction.
3059 (class pass_store_merging): Add m_stores_head field.
3060 (pass_store_merging::terminate_and_process_all_chains):
3061 Iterate over m_stores_head list.
3062 (pass_store_merging::terminate_all_aliasing_chains):
3064 (pass_store_merging::execute): Check for debug stmts first.
3065 Push new chains onto the m_stores_head stack.
3067 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3070 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
3071 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
3072 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
3074 2017-03-16 Jeff Law <law@redhat.com>
3076 PR tree-optimization/71437
3077 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
3078 member function. Implementation moved into after_dom_children
3079 member function and into the threader's thread_outgoing_edges
3081 (dom_opt_dom_walker::after_dom_children): Simplify by moving
3082 some code into new thread_outgoing_edges.
3083 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
3084 definition. Simplify marker handling (do it here). Assume we always
3085 have the available expression and the const/copies tables.
3086 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
3088 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
3089 * tree-vrp.c (equiv_stack): No longer file scoped.
3090 (vrp_dom_walker): New class.
3091 (vrp_dom_walker::before_dom_children): New member function.
3092 (vrp_dom_walker::after_dom_children): Likewise.
3093 (identify_jump_threads): Setup domwalker. Use it rather than
3094 walking edges in a random order by hand. Simplify setup/finalization.
3095 (finalize_jump_threads): Remove.
3096 (vrp_finalize): Do not call identify_jump_threads here.
3097 (execute_vrp): Do it here instead and call thread_through_all_blocks
3100 PR tree-optimization/71437
3101 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
3103 (simplify_stmt_for_jump_threading): Add basic_block argument. All
3105 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
3106 (dom_opt_dom_walker::thread_across_edge): Remove
3107 handle_dominating_asserts argument. All callers changed.
3108 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
3109 changes. Remove calls to lhs_of_dominating_assert. Other
3110 uses of handle_dominating_asserts turn into unconditional code
3111 (simplify_control_stmt_condition_1): Likewise.
3112 (simplify_control_stmt_condition): Likewise.
3113 (thread_through_normal_block, thread_across_edge): Likewise.
3114 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
3115 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
3116 object if it is not an SSA_NAME.
3117 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
3118 before calling into the VRP specific simplifiers.
3119 (identify_jump_threads): Remove handle_dominating_asserts
3122 2017-03-16 Jakub Jelinek <jakub@redhat.com>
3125 * tree-diagnostic.c (default_tree_printer): No longer static.
3126 * tree-diagnostic.h (default_tree_printer): New prototype.
3128 2017-03-16 Tamar Christina <tamar.christina@arm.com>
3130 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
3131 Change ins into fmov.
3133 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3135 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
3136 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
3137 Use h_con constraint for operand 1.
3138 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
3139 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
3141 2017-03-15 Jeff Law <law@redhat.com>
3143 PR tree-optimization/71437
3144 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
3145 (record_temporary_equivalences): Use it.
3147 PR tree-optimization/71437
3148 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
3149 tree-ssa-scopedtables.
3150 (lookup_avail_expr, build_and_record_new_cond): Likewise.
3151 (record_conditions, record_cond, vuse_eq): Likewise.
3152 (record_edge_info): Adjust to API tweak of record_conditions.
3153 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
3154 (record_temporary_equivalences, optimize_stmt): Likewise.
3155 (eliminate_redundant_computations): Likewise.
3156 (record_equivalences_from_stmt): Likewise.
3157 * tree-ssa-scopedtables.c: Include options.h and params.h.
3158 (vuse_eq): New function, moved from tree-ssa-dom.c
3159 (build_and_record_new_cond): Likewise.
3160 (record_conditions): Likewise. Accept vector of conditions rather
3161 than edge_equivalence structure for first argument.
3162 for the first argument.
3163 (avail_exprs_stack::lookup_avail_expr): New member function, moved
3164 from tree-ssa-dom.c.
3165 (avail_exprs_stack::record_cond): Likewise.
3166 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
3167 from tree-ssa-dom.c.
3168 (avail_exprs_stack): Add new member functions lookup_avail_expr
3170 (record_conditions): Declare.
3172 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
3175 * lra-constraints.c (process_alt_operands): Increase reject for
3176 reloading an input/output operand.
3178 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
3181 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
3182 insns to convert from signed/unsigned char/short to IEEE 128-bit
3184 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
3186 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
3189 * config/i386/i386.c (ix86_vector_duplicate_value): Create
3190 subreg of inner mode for values already in registers.
3192 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
3194 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
3195 iteration reg is used after the loop.
3197 2017-03-14 Martin Sebor <msebor@redhat.com>
3199 PR tree-optimization/79800
3200 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
3201 precision in negative-positive range.
3202 (format_floating): Call non-const overload with adjusted precision.
3204 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3207 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
3210 2017-03-14 Martin Sebor <msebor@redhat.com>
3213 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
3214 * builtins.def (aligned_alloc): Use it.
3217 * Makefile.in (GTFILES): Add calls.c.
3218 * calls.c: Include "gt-calls.h".
3220 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
3222 PR rtl-optimization/79728
3223 * regs.h (struct target_regs): New field
3224 x_contains_allocatable_regs_of_mode.
3225 (contains_allocatable_regs_of_mode): New macro.
3226 * reginfo.c (init_reg_sets_1): Initialize it, and change
3227 contains_reg_of_mode so it includes global regs as well.
3228 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
3229 rather than contains_regs_of_mode.
3231 2017-03-14 Martin Liska <mliska@suse.cz>
3233 * doc/invoke.texi: Document options that can't be combined with
3234 -fcheck-pointer-bounds.
3236 2017-03-14 Martin Liska <mliska@suse.cz>
3239 * doc/invoke.texi (-Wchkp): Document the option.
3241 2017-03-14 Martin Liska <mliska@suse.cz>
3243 * Makefile.in: Install gcov-dump.
3245 2017-03-14 Martin Liska <mliska@suse.cz>
3247 * multiple_target.c (expand_target_clones): Bail out for
3248 an invalid attribute.
3250 2017-03-14 Richard Biener <rguenther@suse.de>
3252 * alias.c (struct alias_set_entry): Pack properly.
3253 * cfgloop.h (struct loop): Likewise.
3254 * cse.c (struct set): Likewise.
3255 * ipa-utils.c (struct searchc_env): Likewise.
3256 * loop-invariant.c (struct invariant): Likewise.
3257 * lra-remat.c (struct cand): Likewise.
3258 * recog.c (struct change_t): Likewise.
3259 * rtl.h (struct address_info): Likewise.
3260 * symbol-summary.h (function_summary): Likewise.
3261 * tree-loop-distribution.c (struct partition): Likewise.
3262 * tree-object-size.c (struct object_size_info): Likewise.
3263 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
3264 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
3265 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
3266 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
3267 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
3268 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
3269 (struct _stmt_vec_info): Likewise.
3271 2017-03-14 Martin Liska <mliska@suse.cz>
3274 * multiple_target.c (create_dispatcher_calls): Check that
3275 a target can create a function dispatcher.
3277 2017-03-14 Martin Liska <mliska@suse.cz>
3280 * multiple_target.c (expand_target_clones): Drop local.local
3281 flag for default implementation.
3283 2017-03-14 Richard Biener <rguenther@suse.de>
3285 PR tree-optimization/80030
3286 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
3288 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
3290 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
3291 gcc_fallthrough() instead of __attribute__((fallthrough));
3293 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
3295 * doc/gcc.texi: Remove "up" link to (DIR).
3296 * doc/gccint.texi: Ditto.
3298 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
3300 * doc/install.texi (Specific) <avr>: Remove reference to
3303 2017-03-13 Jeff Law <law@redhat.com>
3305 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
3306 attribute rather than comments.
3308 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
3309 match_scratch operand is highest.
3311 2017-03-13 Martin Liska <mliska@suse.cz>
3314 * ipa-pure-const.c (warn_function_noreturn): If the declarations
3315 is a CHKP clone, use original declaration.
3317 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
3319 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
3320 (arc_conditional_register_usage): Use a different allocation order
3321 when optimizing for size.
3322 * common/config/arc/arc-common.c (arc_option_optimization_table):
3323 Section anchors default on when optimizing for size.
3325 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
3327 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
3329 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
3331 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
3332 * config/arc/arc.md (cpu_facility): Add cd variant.
3333 (*movqi_insn): Add code density variant.
3334 (*movhi_insn): Likewise.
3335 (*movqi_insn): Likewise.
3336 (*addsi3_mixed): Likewise.
3337 (subsi3_insn): Likewise.
3339 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
3341 * config/arc/arc.md (movsi_cond_exec): Update constraint.
3343 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
3345 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
3346 expressions with MINUS and UNARY ops.
3348 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3351 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
3353 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
3354 between vec_select and vector argument.
3355 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
3356 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
3357 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
3358 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
3359 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
3360 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
3362 2017-03-13 Richard Biener <rguenther@suse.de>
3365 * params.def (vect-max-peeling-for-alignment): Fix typo.
3367 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
3369 * doc/install.texi (Specific) <mips-*-*>: Remove description of
3370 issue that only occurred with binutils below 2.18.
3372 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
3374 * doc/install.texi (Specific) <cris-axis-elf>: No longer
3375 refer to binutils 2.11/2.12 minimum.
3377 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
3379 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
3380 ftp.kernel.org and simplify binutils requirement.
3382 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
3384 * doc/invoke.texi (Warning Options): Fix spelling of link-time
3386 (Optimize Options): Ditto. Also remove redundancy.
3388 2017-03-10 David Malcolm <dmalcolm@redhat.com>
3390 PR translation/79848
3391 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
3393 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
3394 to G_ to avoid double translation.
3396 2017-03-10 David Malcolm <dmalcolm@redhat.com>
3398 PR translation/79923
3399 * auto-profile.c (get_combined_location): Convert leading
3400 character of diagnostics to lower case and remove trailing period.
3401 (read_profile): Likewise for various diagnostics.
3402 * config/arm/arm.c (arm_option_override): Remove trailing period
3403 from various diagnostics.
3404 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
3405 (msp430_expand_delay_cycles): Likewise.
3407 2017-03-10 David Malcolm <dmalcolm@redhat.com>
3410 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
3411 full command-line argument, rather than just "str".
3412 (aarch64_validate_march): Likewise.
3413 (aarch64_validate_mtune): Likewise.
3415 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
3417 PR rtl-optimization/78911
3418 * lra-assigns.c (must_not_spill_p): New function.
3419 (spill_for): Use it.
3421 2017-03-10 Jakub Jelinek <jakub@redhat.com>
3423 PR tree-optimization/79981
3424 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
3425 ATOMIC_COMPARE_EXCHANGE ifn result.
3426 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
3427 IFN_ATOMIC_COMPARE_EXCHANGE.
3429 2017-03-10 David Malcolm <dmalcolm@redhat.com>
3432 * opts.c (parse_sanitizer_options): Add missing question mark to
3433 "did you mean" message.
3435 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3437 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
3439 (VMULEUH_UNS): Likewise.
3440 (VMULOUB_UNS): Likewise.
3441 (VMULOUH_UNS): Likewise.
3442 * config/rs6000/rs6000.c (builtin_function_type): Remove
3443 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
3445 2017-03-10 David Malcolm <dmalcolm@redhat.com>
3448 * read-rtl-function.c (function_reader::read_rtx_operand): Update
3449 x with result of extra_parsing_for_operand_code_0.
3450 (function_reader::extra_parsing_for_operand_code_0): Convert
3451 return type from void to rtx, returning x. When reading
3452 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
3453 larger size containing struct block_symbol.
3455 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
3457 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
3458 -mfloat128-hardware without -m64.
3460 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
3463 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
3464 entries to the case statement that marks unsigned arguments to
3465 overloaded functions.
3467 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
3469 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
3470 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
3472 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
3475 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
3476 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
3478 2017-03-10 Martin Liska <mliska@suse.cz>
3482 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
3483 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
3486 2017-03-10 Olivier Hainque <hainque@adacore.com>
3488 * tree-switch-conversion (array_value_type): Start by resetting
3489 candidate type to it's main variant.
3491 2017-03-10 Jakub Jelinek <jakub@redhat.com>
3493 PR rtl-optimization/79909
3494 * combine.c (try_combine): Use simplify_replace_rtx on individual
3495 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
3496 of the whole CALL_INSN_FUNCTION_USAGE.
3498 PR tree-optimization/79972
3499 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
3500 get_range_info on SSA_NAMEs. Formatting fixes.
3502 2017-03-10 Richard Biener <rguenther@suse.de>
3503 Jakub Jelinek <jakub@redhat.com>
3505 PR tree-optimization/77975
3506 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
3507 edge to be constant.
3508 (get_val_for): For constant x return it. Formatting fix.
3509 (loop_niter_by_eval): Avoid pointless looping if the next iteration
3510 would use the same bases as the current one.
3512 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3514 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
3515 instead of vec_select for V1TImode.
3516 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
3518 (VSX_LE_128): Add V1TI to this mode iterator.
3519 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
3520 (*vsx_le_perm_store_<mode>): Likewise.
3521 (pre-reload splitter for VSX stores): Likewise.
3522 (post-reload splitter for VSX stores): Likewise.
3523 (*vsx_xxpermdi2_le_<mode>): Likewise.
3524 (*vsx_lxvd2x2_le_<mode>): Likewise.
3525 (*vsx_stxvd2x2_le_<mode>): Likewise.
3527 2017-03-09 Michael Eager <eager@eagercon.com>
3529 Correct failures with --enable-checking=yes,rtl.
3531 * config/microblaze/microblaze.c (microblaze_expand_shift):
3532 Replace GET_CODE test with CONST_INT_P and INTVAL test with
3533 test for const0_rtx.
3534 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
3535 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
3537 2017-03-09 Richard Biener <rguenther@suse.de>
3539 PR tree-optimization/79977
3540 * graphite-scop-detection.c (scop_detection::merge_sese):
3541 Handle the case of extra exits to blocks dominating the entry.
3543 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
3545 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3548 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
3550 PR rtl-optimization/79949
3551 * lra-constraints.c (process_alt_operands): Check memory when
3552 trying to predict a cycle. Print about the overall increase.
3554 2017-03-09 Richard Biener <rguenther@suse.de>
3557 * gimple-expr.c (useless_type_conversion_p): Preserve
3558 TYPE_SATURATING for fixed-point types.
3560 2017-03-09 Richard Biener <rguenther@suse.de>
3563 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
3564 alignment of BLKmode params.
3566 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3569 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
3570 (VALL_NO_V2Q): Likewise.
3572 * config/aarch64/aarch64-simd.md
3573 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
3575 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
3576 VALL_NO_V2Q mode iterator.
3577 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
3579 2017-03-09 Martin Liska <mliska@suse.cz>
3581 PR tree-optimization/79631
3582 * tree-chkp-opt.c (chkp_is_constant_addr): Call
3583 tree_int_cst_sign_bit just for INTEGER constants.
3585 2017-03-09 Martin Liska <mliska@suse.cz>
3589 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
3592 2017-03-09 Marek Polacek <polacek@redhat.com>
3595 * tree.c (inchash::add_expr): Handle TREE_VEC.
3597 2017-03-09 Martin Liska <mliska@suse.cz>
3600 (chkp_narrow_size_and_offset): New function.
3601 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
3602 (void chkp_parse_bit_field_ref): New function.
3603 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
3604 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
3606 2017-03-09 Martin Liska <mliska@suse.cz>
3609 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
3610 (chkp_find_bounds_1): Remove gcc_unreachable.
3612 2017-03-09 Jakub Jelinek <jakub@redhat.com>
3615 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
3616 BUILT_IN_SYNC*, determine the access type from the size suffix and
3617 always build a MEM_REF with that type. Handle forgotten
3618 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
3621 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
3622 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
3623 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
3624 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
3625 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
3626 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
3627 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
3628 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
3629 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
3630 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
3631 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
3632 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
3633 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
3634 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
3635 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
3636 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
3637 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
3638 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
3639 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
3640 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
3641 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
3642 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
3643 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
3644 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
3645 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
3646 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
3647 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
3648 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
3649 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
3650 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
3651 definitions outside of __OPTIMIZE__ guarded section.
3654 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
3655 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
3656 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
3657 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
3660 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3662 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
3663 ("vfenez<mode>"): Add missing constraints.
3665 2017-03-08 Martin Sebor <msebor@redhat.com>
3668 * config/nds32/nds32.c (nds32_option_override):
3669 Fix misspelled diagnostic.
3671 2017-03-08 Jakub Jelinek <jakub@redhat.com>
3674 * gimplify.c (gimplify_omp_for): Replace index var in outer
3675 taskloop statement with an artificial variable and add
3676 OMP_CLAUSE_PRIVATE clause for it.
3678 2017-03-08 Richard Biener <rguenther@suse.de>
3680 PR tree-optimization/79955
3681 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
3682 for accesses that are completely outside of the variable.
3684 2017-03-08 Andrew Haley <aph@redhat.com>
3686 PR tree-optimization/79943
3687 * tree-ssa-loop-split.c (compute_new_first_bound): When
3688 calculating the new upper bound, (END-BEG) should be added, not
3691 2017-03-08 Jakub Jelinek <jakub@redhat.com>
3693 * config/avr/avr.md (setmemhi): Make sure match_dup
3694 operand number comes before match_scratch.
3696 2017-03-08 Richard Biener <rguenther@suse.de>
3698 PR tree-optimization/79920
3699 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
3700 with ncopies == 1 to ...
3701 (vect_transform_slp_perm_load): ... here. Properly compute
3702 all element loads by iterating VF times over the group. Do
3703 not handle ncopies (computed in a broken way) in
3704 vect_create_mask_and_perm.
3706 2017-03-08 Jakub Jelinek <jakub@redhat.com>
3709 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
3710 is a uniform vector, use uniform_vector_p return value instead of
3711 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
3713 2017-03-07 Marek Polacek <polacek@redhat.com>
3716 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
3717 (alloca_call_type): Likewise.
3719 2017-03-07 Martin Liska <mliska@suse.cz>
3721 * gcov.c (process_args): Put comment to correct location.
3723 2017-03-07 Martin Liska <mliska@suse.cz>
3726 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
3727 Use array_at_struct_end_p instead of DECL_CHAIN (field).
3728 (chkp_narrow_bounds_for_field): Likewise.
3729 (chkp_parse_array_and_component_ref): Pass one more argument to
3732 2017-03-07 Richard Biener <rguenther@suse.de>
3734 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
3737 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
3739 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
3740 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
3742 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3745 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
3746 to end of description.
3747 (PARAM_MAX_STORES_TO_MERGE): Likewise.
3749 2017-03-07 Jakub Jelinek <jakub@redhat.com>
3751 PR rtl-optimization/79901
3752 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
3754 (*avx512f_<code><mode>3<mask_name>): ... this.
3755 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
3756 iterator instead of VI8_AVX2_AVX512BW.
3758 PR rtl-optimization/79901
3759 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
3760 min/max expander, expand it using expand_vec_cond_expr.
3763 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
3766 2017-03-06 Jakub Jelinek <jakub@redhat.com>
3769 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
3770 to void * for PCH reasons.
3771 * dwarf2out.c (output_loc_operands, output_die): Cast
3772 v.val_vec.array to unsigned char *.
3774 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
3777 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
3780 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
3782 PR rtl-optimization/79571
3783 * lra-constraints.c (process_alt_operands): Calculate static
3784 reject and subtract it from overall when only addresses will be
3787 2017-03-06 Julia Koval <julia.koval@intel.com>
3790 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
3791 incoming stack boundary to 128 for 64-bit targets.
3793 2017-03-06 Richard Biener <rguenther@suse.de>
3795 PR tree-optimization/79894
3796 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
3797 to NULL after folding it.
3799 2017-03-06 Richard Biener <rguenther@suse.de>
3801 PR tree-optimization/79824
3802 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
3803 check disabling peeling for gaps.
3805 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
3807 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
3808 attributes): Document gettimeofday.
3810 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
3812 * config/s390/s390.c (s390_option_override_internal): Set
3813 PARAM_MIN_VECT_LOOP_BOUND
3815 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
3817 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
3818 * config/s390/s390.md: Likewise.
3820 2017-03-06 Jakub Jelinek <jakub@redhat.com>
3823 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
3824 (<avx2_avx512>_perm<mode>): Rename to ...
3825 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
3827 (<avx512>_perm<mode>_mask): Rename to ...
3828 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
3830 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
3831 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
3832 instead of VI8F_256_512.
3833 (avx512f_perm<mode>): New define_expand.
3834 (avx512f_perm<mode>_mask): Likewise.
3835 (avx512f_perm<mode>_1<mask_name>): New define_insn.
3836 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
3838 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3840 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
3841 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
3843 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
3845 2017-03-06 Martin Liska <mliska@suse.cz>
3848 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
3849 when having a SSA NAME w/o VAR_DECL assigned to it.
3851 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3853 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
3854 msa_dpsub_<su>_d): Fix MODE for vec_select.
3856 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3858 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
3860 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
3862 2017-03-06 Richard Biener <rguenther@suse.de>
3864 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
3865 * plugin.c (register_plugin_info): Likewise.
3866 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
3868 2017-03-05 Jakub Jelinek <jakub@redhat.com>
3870 * config/i386/sse.md (sse_storehps, sse_storelps,
3871 avx_<castmode><avxsizesuffix>_<castmode>,
3872 avx512f_<castmode><avxsizesuffix>_<castmode>,
3873 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
3874 in condition that at least one operand is not a MEM.
3876 2017-03-03 Jakub Jelinek <jakub@redhat.com>
3879 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
3880 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
3882 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
3883 gimple_call_nothrow_p flag based on whether original builtin can throw.
3884 If it can, emit following stmts on the fallthrough edge.
3885 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
3886 don't create new bb if inserting just debug stmts on the edge, try to
3887 insert them on the fallthru bb or just reset debug stmts.
3889 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
3892 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
3893 restore recog_data (including the operand rtxes inside it) around
3894 the call to get_insn_template.
3896 2017-03-03 Martin Sebor <msebor@redhat.com>
3898 PR tree-optimization/79699
3899 * context.c (context::~context): Free MPFR caches to avoid
3900 a memory leak on program exit.
3902 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3904 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3905 Use wide_int::ulow () instead of .elt (0).
3907 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
3909 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
3910 (*pushxf): Limit oF constraint to 32bit targets and add oC
3911 constraint for 64bit targets.
3912 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
3913 (*pushdf): Change rmF constraint to rmC.
3915 2017-03-03 Martin Liska <mliska@suse.cz>
3917 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
3918 Remove unused variable.
3920 2017-03-03 Jakub Jelinek <jakub@redhat.com>
3923 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
3924 is a memory operand, increase num_memory.
3925 (ix86_expand_args_builtin): Likewise.
3927 2017-03-03 Jan Hubicka <jh@suse.cz>
3930 * ipa-devirt.c (maybe_record_node): Properly handle
3931 __cxa_pure_virtual visibility.
3933 2017-03-03 Martin Liska <mliska@suse.cz>
3935 PR tree-optimization/79803
3936 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
3938 (pass_loop_prefetch::execute): Disabled optimization if an
3939 assumption about L1 cache size is not met.
3941 2017-03-03 Martin Liska <mliska@suse.cz>
3943 PR rtl-optimization/79574
3944 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
3945 (hash_scan_set): Likewise.
3946 (dump_hash_table): Likewise.
3947 (hoist_code): Likewise.
3949 2017-03-03 Richard Biener <rguenther@suse.de>
3951 * fixed-value.c (fixed_from_string): Restore use of elt (1)
3952 in place of uhigh ().
3953 (fixed_convert_from_real): Likewise.
3955 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
3958 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
3960 2017-03-03 Richard Biener <rguenther@suse.de>
3963 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
3964 TYPE_OVERFLOW_UNDEFINED check.
3966 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3968 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
3970 (vector_ae_<mode>_p): Likewise.
3971 (vector_nez_<mode>_p): Likewise.
3972 (vector_ne_v2di_p): Likewise.
3973 (vector_ae_v2di_p): Likewise.
3974 (vector_ne_<mode>_p): Likewise.
3975 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
3977 (vsx_tsqrt<mode>2_fe): Likewise.
3979 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
3982 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
3984 2017-03-02 Jakub Jelinek <jakub@redhat.com>
3986 PR rtl-optimization/79780
3987 * cprop.c (one_cprop_pass): When second and further conditional trap
3988 in a single basic block is turned into an unconditional trap, turn it
3989 into a deleted note to avoid RTL verification failures.
3991 2017-03-02 Richard Biener <rguenther@suse.de>
3993 * fold-const.c (const_binop): Use ulow () instead of elt (0).
3995 2017-03-02 Richard Biener <rguenther@suse.de>
3997 PR tree-optimization/79345
3999 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
4000 param and abort the walk, returning -1 if it is hit.
4001 (walk_aliased_vdefs): Take a limit param and pass it on.
4002 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
4003 defaulting to 0 and return a signed int.
4004 * tree-ssa-uninit.c (struct check_defs_data): New struct.
4005 (check_defs): New helper.
4006 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
4007 about uninitialized memory.
4008 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
4009 bogus uninitialized warning.
4010 (fixed_convert_from_real): Likewise.
4012 2017-03-02 Bin Cheng <bin.cheng@arm.com>
4014 PR tree-optimization/66768
4015 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
4016 iv_use if base object can't be determined.
4018 2017-03-02 Jakub Jelinek <jakub@redhat.com>
4020 PR tree-optimization/79345
4021 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
4022 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
4023 (get_pattern_stats): Initialize it.
4024 * genemit.c (gen_expand): Verify match_scratch numbers come after
4025 match_operand/match_dup numbers.
4026 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
4027 match_scratch numbers.
4028 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
4030 * config/s390/s390.md (trunctdsd2): Likewise.
4032 2017-03-02 Richard Biener <rguenther@suse.de>
4034 * wide-int.h (wide_int_storage::operator=): Implement in terms
4037 2017-03-02 Richard Biener <rguenther@suse.de>
4039 PR tree-optimization/79777
4040 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
4041 the to insert expression to sth existing.
4043 2017-03-01 Martin Sebor <msebor@redhat.com>
4046 * gimple-ssa-sprintf.c
4047 (directive::known_width_and_precision): New function.
4048 (format_integer): Use it.
4049 (get_mpfr_format_length): Consider the full range of precision
4050 when computing %g output with the # flag. Set the likely byte
4051 count to 3 rather than 1 when precision is indeterminate.
4052 (format_floating): Correct the lower bound of precision.
4054 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4056 * doc/invoke.texi: Document default code model for 64-bit Linux.
4058 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4061 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
4062 udiv rather than div since input pattern is unsigned.
4064 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
4066 * config/i386/i386.c (print_reg): Warn for values of
4067 unsupported size in integer register.
4069 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4072 * config/rs6000/predicates.md (current_file_function_operand): Do
4073 not allow self calls to be local if the function is replaceable.
4075 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
4078 * config/rs6000/altivec.h (vec_ctz and others): Change the
4079 preprocessor macro that controls conditional compilation from
4080 _ARCH_PWR9 to __POWER9_VECTOR__.
4081 (vec_all_ne): Change parameterization of __altivec_scalar_pred
4082 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
4083 control (instead of _ARCH_PWR9 control) so that template
4084 definition uses power9-specific function.
4085 (vec_any_eq): Likewise.
4086 (vec_all_ne): Change macro definition to use a power9-specific
4087 expansion under #ifdef __POWER9_VECTOR__ control (instead of
4088 _ARCH_PWR9 control).
4089 (vec_any_eq) Likewise.
4090 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
4091 expansion for CMPNEF to remove support for xvcmpnesp instruction.
4092 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
4093 support for xvcmpnedp instruction.
4094 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
4095 macro expansion so that Power9 implementation of vec_all_ne does
4096 not use the AltiVec predicate framework.
4097 (VCMPNEH_P): Likewise.
4098 (VCMPNEW_P): Likewise.
4099 (VCMPNED_P): Likewise.
4100 (VCMPNEFP_P): Likewise.
4101 (VCMPNEDP_P): Likewise.
4102 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
4103 implementation of vec_any_eq to not use AltiVec predicate
4105 (VCMPAEH_P): Likewise.
4106 (VCMPAEW_P): Likewise.
4107 (VCMPAED_P): Likewise.
4108 (VCMPAEFP_P): Likewise.
4109 (VCMPAEDP_P): Likewise.
4110 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
4111 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
4112 not use the AltiVec predicate framework.
4113 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
4114 of vec_any_eq to not use AltiVec predicate framework.
4115 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
4116 support for predefined __POWER9_VECTOR__ macro to indicate that
4117 Power9 instruction selection is enabled.
4118 (altivec_overloaded_builtins): Remove extraneous
4119 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
4120 function argument types RS6000_BTI_bool_V16QI and
4121 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
4122 entry for overloaded function argument types RS6000_BTI_bool_V4SI
4123 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
4124 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
4125 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
4126 Power9 for implementations of vec_cmpne. Change the signature for
4127 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
4128 (representing vec_all_ne) to remove the previously described first
4129 argument of type RS6000_BTI_INTSI, as this was an artifact of
4130 reliance on the AltiVec predicate framework, which is no longer
4131 used in the implementation of these functions. Add
4132 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
4133 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
4134 since, unlike the AltiVec predicate framework implementation, we
4135 do not share function descriptors between vec_alle and vec_anyeq.
4136 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
4137 set of modes that receive special treatment even when
4138 TARGET_P9_VECTOR is true. The special treatment emits code that
4139 does not depend on Power9 instructions.
4140 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
4141 define_expand to not rely on AltiVec predicate framework.
4142 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
4144 (vector_ne_v2di_p): Change this define_expand to not rely on
4145 AltiVec predicate framework.
4146 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
4148 (vector_ne_<mode>_p): Change this define_expand to not rely on
4149 AltiVec predicate framework.
4150 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
4152 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
4153 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
4154 define_insn pattern.
4155 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
4156 define_insn pattern because the xvcmpne<VSs>. instruction is not
4158 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
4159 instruction is not supported.
4161 2017-03-01 Jakub Jelinek <jakub@redhat.com>
4163 * config/nvptx/nvptx.c: Include intl.h.
4165 2017-03-01 Martin Jambor <mjambor@suse.cz>
4168 * ipa-prop.h (ipa_bits): Removed field known.
4169 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
4170 to pointers. Adjusted their comments to warn about their sharing.
4171 (ipcp_transformation_summary): Change bits to a vector of pointers.
4172 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
4173 (ipa_get_ipa_bits_for_value): Declare.
4174 * tree-vrp.h (value_range): Mark as GTY((for_user)).
4175 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
4176 (ipa_bits_hash_table): Likewise.
4177 (ipa_vr_ggc_hash_traits): Likewise.
4178 (ipa_vr_hash_table): Likewise.
4179 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
4180 being pointers and vr_known being removed.
4181 (ipa_set_jf_unknown): Likewise.
4182 (ipa_get_ipa_bits_for_value): New function.
4183 (ipa_set_jfunc_bits): Likewise.
4184 (ipa_get_value_range): New overloaded functions.
4185 (ipa_set_jfunc_vr): Likewise.
4186 (ipa_compute_jump_functions_for_edge): Use the above functions to
4187 construct bits and vr parts of jump functions.
4188 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
4189 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
4191 (ipcp_grow_transformations_if_necessary): Also allocate
4192 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
4194 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
4195 them. Fix too long lines.
4196 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
4197 vr_known being removed.
4198 (ipa_read_jump_function): Use new setter functions to construct bits
4199 and vr parts of jump functions or set them to NULL.
4200 (write_ipcp_transformation_info): Adjust for bits being pointers.
4201 (read_ipcp_transformation_info): Likewise.
4202 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
4204 Include gt-ipa-prop.h.
4205 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
4207 (ipcp_store_bits_results): Likewise.
4208 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
4209 Do not write to existing jump functions but use a temporary instead.
4211 2017-03-01 Jakub Jelinek <jakub@redhat.com>
4214 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
4215 attempt to use its first operand as BIT_FIELD_REF base.
4217 2017-03-01 Richard Biener <rguenther@suse.de>
4220 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
4221 interpolating formula in wrapping arithmetic.
4222 (chrec_apply): Convert chrec_evaluate return value to wanted type.
4224 2017-03-01 Jakub Jelinek <jakub@redhat.com>
4226 PR tree-optimization/79734
4227 * tree-vect-generic.c (expand_vector_condition): Optimize
4228 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
4229 Handle VEC_COND_EXPR where comparison has different inner width from
4232 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
4234 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
4235 markup, and similar issues. Remove @opindex entries for things
4236 that aren't options. Add missing -mmpy-option entries.
4238 2017-02-28 Jakub Jelinek <jakub@redhat.com>
4240 PR tree-optimization/79737
4241 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
4242 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
4243 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
4244 instead of byte_size. Formatting fix.
4245 (shift_bytes_in_array_right): Formatting fix.
4247 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
4250 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
4251 condition on optimize for the leaf function test.
4253 2017-02-28 Martin Liska <mliska@suse.cz>
4256 * read-rtl-function.c (function_reader::handle_unknown_directive):
4257 Bail out when one uses -flto.
4259 2017-02-28 Martin Liska <mliska@suse.cz>
4261 * common.opt: Replace space with tabular for options of <number>
4263 * config/i386/i386.opt: Show <number> value for
4264 -mlarge-data-threshold.
4265 * opts.c (print_filtered_help): Do not display number in hexadecimal
4268 2017-02-28 Martin Liska <mliska@suse.cz>
4270 * common.opt: Fix --help=option -Q for options which are of
4273 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
4275 * config/i386/i386.c (print_reg): Error out for values
4276 of 8-bit size in invalid integer register.
4278 2017-02-28 Martin Sebor <msebor@redhat.com>
4280 PR tree-optimization/79691
4281 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
4283 2017-02-28 Jakub Jelinek <jakub@redhat.com>
4286 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
4287 gcc_unreachable with output_operand_lossage.
4289 2017-02-28 Richard Biener <rguenther@suse.de>
4291 PR tree-optimization/79740
4292 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
4294 (visit_nary_op): Insert the nary into the hashtable if we
4295 pattern-matched sth.
4296 * tree-ssa-pre.c (eliminate_insert): Robustify.
4298 2017-02-28 Richard Biener <rguenther@suse.de>
4301 * fold-const.c (decode_field_reference): Reject out-of-bound
4304 2017-02-28 Jakub Jelinek <jakub@redhat.com>
4306 * config/i386/i386.c: Include intl.h.
4307 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
4308 instead of just cond ? "..." : "...".
4309 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
4310 * coverage.c (read_counts_file): Likewise.
4311 * omp-offload.c: Include intl.h.
4312 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
4313 of just cond ? "..." : "...".
4314 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
4315 of just cond ? "..." : "...".
4317 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
4320 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
4322 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
4323 'tune for' CPU name.
4324 * config/arm/arm-cpu-data.h: Regenerated.
4326 2017-02-28 Richard Biener <rguenther@suse.de>
4328 PR tree-optimization/79732
4329 * tree-inline.c (expand_call_inline): Do not shadow var.
4331 2017-02-28 Richard Biener <rguenther@suse.de>
4333 PR tree-optimization/79723
4334 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
4335 address-space properly.
4337 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
4339 * doc/optinfo.texi (Optimization groups): Fix option used for
4341 * doc/invoke.texi (-fopt-info): Document "omp".
4342 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
4343 (OPTGROUP_ALL): Add OPTGROUP_OMP.
4344 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
4345 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
4346 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
4348 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
4350 * dumpfile.c (optgroup_options): Instead of "openmp", associate
4351 OPTGROUP_OMP with "omp".
4353 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
4356 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
4357 for arithmetic shift of unsigned V2DI.
4359 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
4361 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
4362 arc/linux.h headers.
4363 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
4364 (LINK_SPEC): Likewise.
4365 (ARC_TLS_EXTRA_START_SPEC): Likewise.
4366 (EXTRA_SPECS): Likewise.
4367 (STARTFILE_SPEC): Likewise.
4368 (ENDFILE_SPEC): Likewise.
4369 (LIB_SPEC): Likewise.
4370 (TARGET_SDATA_DEFAULT): Likewise.
4371 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
4372 (MULTILIB_DEFAULTS): Likewise.
4373 (DWARF2_UNWIND_INFO): Likewise.
4374 * config/arc/big.h: New file.
4375 * config/arc/elf.h: Likewise.
4376 * config/arc/linux.h: Likewise.
4377 * config/arc/t-uClibc: Remove.
4379 2017-02-27 Bin Cheng <bin.cheng@arm.com>
4381 PR tree-optimization/77536
4382 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
4383 (tree_transform_and_unroll_loop): Use above function to compute the
4384 estimated niter of unrolled loop and use it when scaling profile.
4385 Also use count info rather than frequency if it's non-zero.
4386 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
4387 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
4388 (vect_transform_loop): Call above function.
4390 2017-02-27 Richard Biener <rguenther@suse.de>
4392 PR tree-optimization/45397
4393 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
4394 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
4395 (visit_nary_op): Add pattern matching for CSEing sign-changed
4396 or truncated operations with wider ones.
4398 2017-02-27 Richard Biener <rguenther@suse.de>
4400 PR tree-optimization/79690
4401 * tree-vect-stmts.c (vectorizable_store): Use vector type
4402 built from the DR with address-space.
4404 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
4406 * doc/invoke.texi (Optimize Options): Refine the description
4407 of asan-use-after-return.
4409 2017-02-25 Alan Modra <amodra@gmail.com>
4411 PR rtl-optimization/79584
4412 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
4413 base, not ad->base_term, the reg within base. Remove assertion
4414 that ad->base == ad->base_term. Replace gen_int_mode using
4415 bogus mode with const0_rtx.
4417 2017-02-25 Jakub Jelinek <jakub@redhat.com>
4420 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
4421 FMA_EXPR like tcc_binary or tcc_unary.
4423 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
4426 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
4428 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
4429 (output_loc_operands): Handle DW_OP_call_ref and
4430 DW_OP_GNU_variable_value.
4431 (struct variable_value_struct): New type.
4432 (struct variable_value_hasher): Likewise.
4433 (variable_value_hash): New variable.
4434 (string_types): Remove.
4435 (copy_loc_descr): New function.
4436 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
4437 (prepend_loc_descr_to_each): New function.
4438 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
4439 instead of add_loc_descr_to_each if the first argument is single
4440 location list and the second has multiple.
4441 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
4442 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
4443 when looking for variable value which doesn't have other location info.
4444 (loc_list_from_tree): Formatting fix.
4445 (gen_array_type_die): Simplify DW_AT_string_length handling.
4446 (adjust_string_types): Remove.
4447 (gen_subprogram_die): Don't call adjust_string_types nor test/set
4448 string_types. Call resolve_variable_values.
4449 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
4450 (resolve_addr_in_expr): Likewise. Add A argument.
4451 (copy_deref_exprloc): Remove deref argument. Adjust for the
4452 original expression being DW_OP_GNU_variable_value with optionally
4453 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
4454 optionally after it.
4455 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
4456 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
4457 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
4458 (variable_value_hasher::hash, variable_value_hasher::equal): New
4460 (resolve_variable_value_in_expr, resolve_variable_value,
4461 resolve_variable_values, note_variable_value_in_expr,
4462 note_variable_value): New functions.
4463 (dwarf2out_early_finish): Call note_variable_value on all toplevel
4466 2017-02-24 Jakub Jelinek <jakub@redhat.com>
4469 * opts.h (handle_generated_option): Add GENERATED_P argument.
4470 * opts-common.c (handle_option): Adjust function comment.
4471 (handle_generated_option): Add GENERATED_P argument, pass it to
4473 (control_warning_option): Pass false to handle_generated_option
4475 * opts.c (maybe_default_option): Pass true to handle_generated_option
4477 * optc-gen.awk: Likewise.
4479 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4481 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
4482 a REG, look at the REG it is a SUBREG of.
4483 (splitter for cmpeqsi_t): Ditto.
4485 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4487 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
4488 the special USEs with the pattern of the insn, not the insn itself.
4490 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
4493 * doc/invoke.texi: Document -mload-store-pairs.
4495 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4496 Sandra Loosemore <sandra@codesourcery.com>
4498 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
4499 argument isn't a CONST_INT.
4500 (nios2_alternate_compare_const): Assert op is a CONST_INT.
4501 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
4502 (nios2_validate_compare): Bypass alternate compare logic if *op2
4504 (ldstwm_operation_p): Return false if first_base is not a REG or
4505 if first_offset is not a CONST_INT.
4507 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4509 * config/cris/cris.md: Use correct operand in a define_peephole2.
4511 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4513 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
4515 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4517 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
4518 this_insn if it is an INSN or JUMP_INSN.
4519 (force_offsettable): Look at base, not at addr.
4520 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
4521 on things that aren't necessarily CONST_INTs.
4523 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
4525 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
4526 -mfpmath=sse is the default also for x86-32 targets with SSE2
4527 instruction set when @option{-ffast-math} is enabled
4529 2017-02-24 Jeff Law <law@redhat.com>
4531 PR rtl-optimizatoin/79286
4532 * ira.c (update_equiv_regs): Drop may_trap_p exception to
4535 2017-02-24 Richard Biener <rguenther@suse.de>
4537 PR tree-optimization/79389
4538 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
4541 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
4543 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
4544 function comment to reflect reality.
4545 (loop_exits_before_overflow): Fix typo in function description.
4547 2017-02-24 Richard Biener <rguenther@suse.de>
4549 PR tree-optimization/79389
4550 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
4551 properly that a threading opportunity exists. Detect conditional
4552 copy/constant propagation opportunities.
4554 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
4556 * config/visium/visium.md (type): Add trap.
4557 (b): New mode attribute.
4558 (*btst): Rename into...
4559 (*btst<mode>): ...this and adjust.
4560 (*cbranchsi4_btst_insn): Rename into...
4561 (*cbranch<mode>4_btst_insn): ...this and adjust.
4562 (trap): New define_insn.
4564 2017-02-23 Jakub Jelinek <jakub@redhat.com>
4566 PR tree-optimization/79389
4567 * ifcvt.c (struct noce_if_info): Add rev_cond field.
4568 (noce_reversed_cond_code): New function.
4569 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
4570 reversed_comparison_code. Formatting fix.
4571 (noce_try_store_flag): Test rev_cond != NULL in addition to
4572 reversed_comparison_code.
4573 (noce_try_store_flag_constants): Likewise.
4574 (noce_try_store_flag_mask): Likewise.
4575 (noce_try_addcc): Use rev_cond if non-NULL instead of
4576 reversed_comparison_code.
4577 (noce_try_cmove_arith): Likewise. Formatting fixes.
4578 (noce_try_minmax, noce_try_abs): Clear rev_cond.
4579 (noce_find_if_block): Initialize rev_cond.
4580 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
4581 instead of false as last argument never attempt to reverse it
4584 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4586 PR tree-optimization/79663
4587 * tree-predcom.c (combine_chains): Process refs in reverse order
4588 only for ZERO length chains, and add explaining comment.
4590 2017-02-23 Jeff Law <law@redhat.com>
4592 PR tree-optimization/79578
4593 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
4594 in call to operand_equal_p.
4596 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
4599 * config/i386/cpuid.h: Fix another undefined behavior.
4601 2017-02-23 Richard Biener <rguenther@suse.de>
4603 PR tree-optimization/79683
4604 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
4605 vector types for data-refs.
4607 2017-02-23 Martin Liska <mliska@suse.cz>
4609 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
4611 2017-02-23 Jakub Jelinek <jakub@redhat.com>
4614 * internal-fn.c (get_range_pos_neg): Moved to ...
4615 * tree.c (get_range_pos_neg): ... here. No longer static.
4616 * tree.h (get_range_pos_neg): New prototype.
4617 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
4618 are known to be in between 0 and signed maximum inclusive, try to
4619 expand both unsigned and signed divmod and use the cheaper one from
4622 2017-02-22 Jeff Law <law@redhat.com>
4624 PR tree-optimization/79578
4625 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
4626 to compare base operands.
4628 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
4631 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
4632 gpc_reg_operand instead of fpr_reg_operand.
4634 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
4636 * config/mips/mips.c (mips_return_in_memory): Force FP
4637 vector types to be returned in memory for o32 ABI.
4639 2017-02-22 Jakub Jelinek <jakub@redhat.com>
4641 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
4642 instead of DW_TAG_member for static data member declarations and don't
4643 set no_linkage_name for static inline data members.
4644 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
4647 2017-02-22 Martin Liska <mliska@suse.cz>
4649 * doc/invoke.texi: Replace inequality signs with square brackets
4652 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4654 PR tree-optimization/68644
4655 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
4657 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
4660 * lra-constraints.c (simplify_operand_subreg): Handle
4661 WORD_REGISTER_OPERATIONS targets.
4663 2017-02-22 Jakub Jelinek <jakub@redhat.com>
4666 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
4667 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
4668 elimination by swapping fld*.
4670 2017-02-22 Richard Biener <rguenther@suse.de>
4672 PR tree-optimization/79673
4673 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
4674 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
4675 irrelevant address-space qualifiers and avoiding a
4676 ADDR_SPACE_CONVERT_EXPR from fold_convert.
4678 2017-02-22 Richard Biener <rguenther@suse.de>
4680 PR tree-optimization/79666
4681 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
4682 to not symbolically negate if that may introduce undefined
4685 2017-02-22 Martin Liska <mliska@suse.cz>
4688 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
4689 * data-streamer-out.c (streamer_write_gcov_count_stream):
4691 * value-prof.c (stream_out_histogram_value): Make assert more
4692 precise based on type of counter.
4694 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
4697 * config/i386/i386.md (standard_x87sse_constant_load splitter):
4698 Use nonimmediate_operand instead of memory_operand for operand 1.
4699 (float-extend standard_x87sse_constant_load splitter): Ditto.
4701 2017-02-21 Jeff Law <law@redhat.com>
4703 PR tree-optimization/79621
4704 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
4705 blocks with edges to themselves.
4707 2017-02-21 Jakub Jelinek <jakub@redhat.com>
4710 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
4711 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
4712 Use gimple_call_builtin_p.
4715 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
4716 on temporarily removed DEBUG_INSNs.
4718 PR tree-optimization/79649
4719 * tree-loop-distribution.c (classify_partition): Give up on
4720 non-generic address space loads/stores.
4722 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
4724 * doc/loop.texi (Loop manipulation): Remove nonexistent
4725 tree_ssa_loop_version from the documentation.
4726 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
4728 2017-02-21 Jakub Jelinek <jakub@redhat.com>
4731 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
4732 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
4733 * config/rs6000/rs6000.c: Include except.h.
4734 (rs6000_expand_split_stack_prologue): Call
4735 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
4737 2017-02-21 Martin Jambor <mjambor@suse.cz>
4740 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
4743 2017-02-21 Martin Jambor <mjambor@suse.cz>
4745 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
4746 for backward compatibility only.
4747 * doc/invoke.texi (Option Summary): Remove all references to
4750 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
4754 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4756 * lra-constraints.c (curr_insn_transform): Handle
4757 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
4759 2017-02-21 Martin Liska <mliska@suse.cz>
4761 * config/i386/i386.opt: Replace -masm-dialect with -masm.
4763 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
4765 PR translation/79638
4766 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
4768 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
4771 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
4772 (arm_function_ok_for_sibcall): Return false for an indirect call by
4773 descriptor if all the argument registers are used.
4774 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
4775 alignment of the function.
4777 2017-02-21 Jakub Jelinek <jakub@redhat.com>
4779 PR tree-optimization/61441
4780 * simplify-rtx.c (simplify_const_unary_operation): For
4781 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
4782 the sNaN unmodified.
4784 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
4786 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
4787 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
4788 instead of SYSTEM_HEADER_DIR.
4790 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
4791 Martin Liška <mliska@suse.cz>
4793 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
4794 Fix typos and grammar, use active voice, and clarify.
4796 2017-02-20 Marek Polacek <polacek@redhat.com>
4799 * gimplify.c (gimplify_expr): Handle unused *&&L;.
4802 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
4804 2017-02-20 Jakub Jelinek <jakub@redhat.com>
4807 * config/i386/i386.c (ix86_expand_builtin): Handle
4808 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
4809 ix86_builtins_isa[fcode].isa as a requirement of those
4810 flags and any other flag in the bitmask.
4811 (ix86_init_mmx_sse_builtins): Use 0 instead of
4812 ~OPTION_MASK_ISA_64BIT as mask.
4813 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
4814 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
4815 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
4816 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
4818 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4821 * lra-constraints.c (split_reg): Check requested split mode
4822 is supported by the register.
4824 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4826 * lra-constraints.c (simplify_operand_subreg): Remove early
4829 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4832 * lra-constraints.c (curr_insn_transform): Tighten condition
4833 for converting SUBREG reloads from OP_OUT to OP_INOUT.
4835 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4838 * lra-constraints.c (curr_insn_transform): Handle
4839 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
4841 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
4844 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
4846 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
4848 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
4851 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
4854 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4856 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
4859 2017-02-18 Jakub Jelinek <jakub@redhat.com>
4861 * final.c (last_columnnum, override_columnnum): New variables.
4862 (final_start_function): Set last_columnnum, pass it to begin_prologue
4863 hook and pass 0 to dwarf2out_begin_prologue.
4864 (final_scan_insn): Update override_columnnum. Pass last_columnnum
4865 to source_line debug hook.
4866 (notice_source_line): Compute last_columnnum and for debug_column_info
4867 return true on column changes.
4868 * debug.h (struct gcc_debug_hooks): Add column argument to
4869 source_line and begin_prologue hooks.
4870 (debug_nothing_int_charstar_int_bool): Remove prototype.
4871 (debug_nothing_int_int_charstar,
4872 debug_nothing_int_int_charstar_int_bool): New prototypes.
4873 (dwarf2out_begin_prologue): Add column argument.
4874 * debug.c (do_nothing_debug_hooks): Adjust source_line and
4875 begin_prologue hooks.
4876 (debug_nothing_int_charstar_int_bool): Remove.
4877 (debug_nothing_int_int_charstar,
4878 debug_nothing_int_int_charstar_int_bool): New functions.
4879 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
4880 through to dwarf2out_source_line.
4881 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
4882 (dwarf2out_source_line): Add column argument, emit it if requested.
4883 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
4885 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
4886 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
4887 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
4888 through to dwarf2out_begin_prologue.
4889 (vmsdbgout_source_line): Add column argument, pass it through to
4890 dwarf2out_source_line.
4891 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
4892 dbxout_source_line caller.
4893 (dbxout_source_line): Add column argument.
4895 * common.opt (gno-column-info, gcolumn-info): New options.
4896 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
4897 (check_die): Also test for multiple DW_AT_decl_column attributes.
4898 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
4899 DW_AT_decl_column if requested.
4900 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
4902 (gen_variable_die): Likewise.
4903 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
4904 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
4907 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
4908 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
4909 (ix86_handle_option): Handle OPT_m3dnowa.
4910 * doc/invoke.texi (-m3dnowa): Document.
4911 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
4912 -m3dnowa instead of -m3dnow -march=athlon.
4915 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
4916 instead of gcc_assert for K, r and R code checks. Formatting fixes.
4918 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4921 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
4922 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
4923 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
4924 generator for vsx_xxpermdi_<mode>_be.
4925 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
4926 force big-endian semantics.
4927 (vsx_xxpermdi_<mode>_be): New define_expand with same
4928 implementation as previous version of vsx_xxpermdi_<mode>.
4930 2017-02-17 Jakub Jelinek <jakub@redhat.com>
4932 PR tree-optimization/79327
4933 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
4934 variable, its initialization and use.
4936 2017-02-17 Julia Koval <julia.koval@intel.com>
4938 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
4939 (OPTION_MASK_ISA_PKU_UNSET): New.
4940 (ix86_handle_option): Handle -mrdpid.
4941 * config/i386/cpuid.h (bit_RDPID): New.
4942 * config/i386/driver-i386.c (host_detect_local_cpu):
4943 Detect RDPID feature.
4944 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
4945 * config/i386/i386-c.c (ix86_target_macros_internal):
4947 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
4948 (ix86_valid_target_attribute_inner_p): Add "rdpid".
4949 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
4950 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
4951 * config/i386/i386.md (define_insn "rdpid"): New.
4952 * config/i386/i386.opt Add -mrdpid.
4953 * config/i386/immintrin.h (_rdpid_u32): New.
4955 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
4957 PR rtl-optimization/79541
4958 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
4959 instead of transforming it into USE.
4961 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
4963 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
4964 If HONOR_SNANS (SFmode) force the input to a register.
4965 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
4966 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
4967 an frsp or similar insn.
4969 2017-02-17 Martin Liska <mliska@suse.cz>
4971 PR rtl-optimization/79577
4972 * params.def (selsched-max-sched-times): Increase minimum to 1.
4974 2017-02-17 Martin Liska <mliska@suse.cz>
4976 PR rtl-optimization/79574
4977 * gcse.c (want_to_gcse_p): Prevent integer overflow.
4979 2017-02-17 Martin Liska <mliska@suse.cz>
4981 PR tree-optimization/79529
4982 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
4983 ssa_defined_default_def_p to handle cases which are implicitly
4985 * tree-ssa.c (ssa_defined_default_def_p): New function.
4986 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
4987 which are implicitly defined.
4988 * tree-ssa.h (ssa_defined_default_def_p): Declare.
4990 2017-02-17 Richard Biener <rguenther@suse.de>
4993 * params.def (max-ssa-name-query-depth): Limit to 10.
4995 2017-02-17 Richard Biener <rguenther@suse.de>
4997 PR tree-optimization/79552
4998 * tree-ssa-structalias.c (visit_loadstore): Properly verify
5001 2017-02-17 Richard Biener <rguenther@suse.de>
5004 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
5006 2017-02-17 Marek Polacek <polacek@redhat.com>
5009 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
5010 (fold_negate_expr): New wrapper.
5012 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
5014 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
5015 Correct terminology and de-emphasize pre-standard behavior.
5017 2017-02-16 Alan Modra <amodra@gmail.com>
5019 PR rtl-optimization/79286
5020 * ira.c (def_dominates_uses): New function.
5021 (update_equiv_regs): Don't create an equivalence for insns that
5022 may trap where the register def does not dominate the use.
5024 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
5026 PR rtl-optimization/78127
5027 * lra.c (lra): Call lra_eliminate before finish the loop after
5030 2017-02-16 Richard Biener <rguenther@suse.de>
5032 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
5034 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
5035 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
5036 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
5037 (isl_val_int_from_wi): New function.
5038 (extract_affine_gmp): Rename to ...
5039 (extract_affine_wi): ... this, take a widest_int.
5040 (extract_affine_int): Just wrap extract_affine_wi.
5041 (add_param_constraints): Use isl_val_int_from_wi.
5042 (add_loop_constraints): Likewise, and extract_affine_wi.
5044 2017-02-15 Jeff Law <law@redhat.com>
5047 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
5048 ira_init_register_move_cost_if_necessary.
5050 2017-02-15 Martin Sebor <msebor@redhat.com>
5053 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
5054 removed in a prior commit.
5056 2017-02-15 Bin Cheng <bin.cheng@arm.com>
5058 PR tree-optimization/79347
5059 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
5060 counters during peeling.
5062 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
5064 * Makefile.in (site.exp): Remove "set ISLVER".
5066 2017-02-15 Jakub Jelinek <jakub@redhat.com>
5069 * real.c (real_from_integer): Call real_convert even for decimal.
5071 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
5074 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
5076 2017-02-14 Andrew Pinski <apinski@cavium.com>
5078 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
5079 cores and change the partno/implementer to be correct.
5080 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
5081 the 'B" as the implementer.
5082 * config/aarch64/aarch64-tune.md: Regenerate.
5084 2017-02-14 Carl Love <cel@us.ibm.com>
5086 * config/rs6000/rs6000.c: Add case statement entry to make the
5087 xvcvuxdsp built-in argument unsigned.
5088 * config/rs6000/vsx.md: Fix the source and return operand types so they
5089 match the instruction definitions from the ISA document. Fix typo
5090 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
5093 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
5096 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
5097 member early_clobber_alts.
5098 * lra-lives.c (reg_early_clobber_p): New.
5099 (process_bb_lives): Use it.
5100 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
5101 (debug_operand_data): Initialize early_clobber_alts.
5102 (setup_operand_alternative): Set up early_clobber_alts.
5103 (collect_non_operand_hard_regs): Ditto. Pass early clobber
5104 alternatives to new_insn_reg.
5105 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
5107 (lra_update_insn_regno_info): Pass the new arg.
5109 2017-02-14 Jakub Jelinek <jakub@redhat.com>
5112 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
5113 (new_oacc_loop_raw): Don't clear already cleared fields.
5116 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
5117 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
5118 _mm512_prefetch_i64gather_ps): New inline functions and macros.
5120 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
5123 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
5125 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
5128 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
5129 the extra instruction to the right place to store 128-bit constant
5132 2017-02-14 Martin Sebor <msebor@redhat.com>
5135 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
5136 warning for strings of unknown length.
5138 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
5140 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
5142 2017-02-14 Jeff Law <law@redhat.com>
5145 * ira-costs.c (scan_one_insn): Initialize register move costs
5146 for pseudos seen in USE/CLOBBER insns.
5148 PR tree-optimization/79095
5149 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
5150 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
5151 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
5152 if the operands are known to be not equal, then the resulting range
5154 (intersect_ranges): If the new range is ~[0,0] and the old range is
5155 wide, then prefer ~[0,0].
5156 * tree-vrp.c (overflow_comparison_p_1): New function.
5157 (overflow_comparison_p): New function.
5158 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
5159 if NAME is used in an overflow test.
5160 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
5161 overflow check that can be expressed as an equality test, then adjust
5162 ops to be that equality test.
5164 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5166 * config/s390/s390-builtin-types.def: Remove flags argument.
5167 * config/s390/s390.c (s390_init_builtins): Likewise.
5169 2017-02-14 Martin Liska <mliska@suse.cz>
5171 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
5172 vector. Fix trailing white spaces.
5174 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
5176 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
5179 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5181 PR rtl-optimization/68664
5182 * config/arm/arm.c (arm_sched_can_speculate_insn):
5183 New function. Declare prototype.
5184 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
5186 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5188 PR rtl-optimization/68664
5189 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
5191 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
5193 2017-02-14 Amit Pawar <amit.pawar@amd.com>
5195 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
5196 max skip bytes for function, loop and jump.
5198 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5200 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
5201 ABS_EXPR for gimple dump.
5203 2017-02-14 Jakub Jelinek <jakub@redhat.com>
5206 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
5208 PR tree-optimization/79408
5209 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
5210 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
5211 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
5212 also if rhs1 is INTEGER_CST.
5214 2017-02-14 Richard Biener <rguenther@suse.de>
5217 * tree-into-ssa.c (insert_phi_nodes): When the function can
5218 have abnormal edges rewrite SSA names with broken use-def
5219 dominance out of SSA and register them for PHI insertion.
5221 2017-02-13 Martin Sebor <msebor@redhat.com>
5224 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
5225 clearing info.nowrite flag when snprintf size argument is a range.
5227 2017-02-13 Jakub Jelinek <jakub@redhat.com>
5229 * cprop.c (cprop_jump): Add missing space in string literal.
5230 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
5231 (get_constraint_for_component_ref): Likewise.
5232 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
5233 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
5234 * lra-constraints.c (process_alt_operands): Likewise.
5235 * ipa-inline.c (inline_small_functions): Likewise.
5236 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
5237 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
5238 * trans-mem.c (diagnose_tm_1_op): Likewise.
5239 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
5240 (grid_parallel_clauses_gridifiable): Likewise.
5242 * config/nvptx/mkoffload.c (process): Add space in between
5245 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
5246 "MOD4_SSE_REGS" and "ALL_REGS".
5248 * spellcheck.c (test_data): Add , in between "foo" and "food".
5250 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5253 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
5254 boundary crossing check and subsequent code generation agree.
5256 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5258 * config/aarch64/aarch64.c (has_memory_op): Delete.
5259 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
5262 2017-02-13 Jakub Jelinek <jakub@redhat.com>
5264 PR rtl-optimization/79388
5265 PR rtl-optimization/79450
5266 * combine.c (distribute_notes): When removing TEM_INSN for which
5267 corresponding dest has last value recorded, invalidate that last
5270 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5272 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
5273 of explicit '@'. Add missing assembly comment marker on branch costs
5276 2017-02-13 Nathan Sidwell <nathan@acm.org>
5278 * gengtype-lex.l (<in_struct>): Add '/'.
5280 2017-02-13 Martin Liska <mliska@suse.cz>
5283 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
5285 2017-02-13 Richard Biener <rguenther@suse.de>
5287 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
5289 * configure: Re-generate.
5290 * config.in: Likewise.
5291 * graphite-dependences.c: Simplify as if
5292 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
5293 * graphite-isl-ast-to-gimple.c: Likewise.
5294 * graphite-optimize-isl.c: Likewise.
5295 * graphite-poly.c: Likewise.
5296 * graphite-sese-to-poly.c: Likewise.
5297 * graphite.h: Likewise.
5298 * toplev.c: Include isl/version.h and use isl_version () for
5299 printing the ISL version.
5300 * doc/install.texi: Update ISL requirement.
5302 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
5304 * doc/standards.texi (Standards): Update reference to
5307 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
5309 * doc/extend.texi (Named Address Spaces): sourceware.org now
5311 * doc/install.texi (Binaries): Ditto.
5314 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
5316 * doc/cpp.texi: Replace "stringify"/"stringification" with C
5317 standard terminology "stringize"/"stringizing" throughout.
5318 * doc/cppinternals.texi: Likewise.
5320 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
5322 * doc/extend.texi: Fix some spelling mistakes and typos.
5323 * doc/invoke.texi: Likewise.
5325 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
5328 * params.def (inline-min-speedup) Change from 10 to 8.
5330 2017-02-11 Jakub Jelinek <jakub@redhat.com>
5332 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
5335 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
5338 * ipa-inline-analysis.c (get_minimal_bb): New function.
5339 (record_modified): Use it.
5340 (remap_edge_change_prob): Handle also ancestor functions.
5342 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
5344 * doc/contrib.texi (Contributors): Remove broken link into
5345 the Mauve CVS repository.
5347 2017-02-11 Jakub Jelinek <jakub@redhat.com>
5350 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
5351 result computation whenever lhs doesn't have vector mode, not
5352 just when it has BLKmode.
5354 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
5356 * doc/makefile.texi (profiledbootstrap): Refer to the
5357 installation instructions only in textual form.
5359 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5362 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
5364 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
5366 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
5367 (Specific): Update mingw-w64 reference.
5369 (Specific): Remove broken link to Renesas RX processor.
5371 2017-02-10 Richard Biener <rguenther@suse.de>
5373 * toplev.c (process_options): Do not mention obsolete graphite
5374 options when printing sorry message about missing graphite support.
5375 Mention -floop-nest-optimize.
5377 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
5379 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
5380 (vtst_p16): Likewise.
5381 (vtstq_p8): Likewise.
5382 (vtstq_p16): Likewise.
5384 (vtstq_p64): Likewise.
5385 * config/arm/arm_neon.h (vgetq_lane_p64): New.
5386 (vset_lane_p64): New.
5387 (vsetq_lane_p64): New.
5389 2017-02-10 Jakub Jelinek <jakub@redhat.com>
5391 PR tree-optimization/79411
5392 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
5393 stmt operands are SSA_NAMEs used in abnormal phis.
5394 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
5397 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
5400 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
5403 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
5405 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
5407 2017-02-09 Jakub Jelinek <jakub@redhat.com>
5409 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
5413 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
5414 not arbitrary TREE_CONSTANT.
5417 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
5418 "omp declare target link" attribute unless is_global_var.
5419 * omp-offload.c (find_link_var_op): Likewise.
5421 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
5422 Chung-Lin Tang <cltang@codesourcery.com>
5424 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
5426 (gimplify_adjust_omp_clauses): Don't delete TILE.
5427 (gimplify_omp_for): Deal with TILE.
5428 * internal-fn.c (expand_GOACC_TILE): New function.
5429 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
5431 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
5432 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
5434 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
5435 avoid DIV for outermost collapse var.
5436 (expand_oacc_for): Insert tile element loop as needed. Adjust.
5437 Remove out of date comments, fix whitespace.
5438 * omp-general.c (omp_extract_for_data): Deal with tiling.
5439 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
5440 adjust OLF_DIM_BASE value.
5441 (struct omp_for_data): Add tiling field.
5442 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
5443 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
5444 for auto loops. Remove default auto determining, moved to
5445 oacc_loop_fixed_partitions.
5446 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
5447 stmts, add e_mask field.
5448 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
5449 (oacc_thread_numbers): Use oacc_dim_call.
5450 (oacc_xform_tile): New.
5451 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
5452 (finish_oacc_loop): Adjust for ifns vector.
5453 (oacc_loop_discover_walk): Append loop abstraction sites to list,
5454 add case for GOACC_TILE fns.
5455 (oacc_loop_xform_loop): Delete.
5456 (oacc_loop_process): Iterate over call list directly, and add
5457 handling for GOACC_TILE fns.
5458 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
5460 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
5461 vector partitioning to outer loops. Assign 2 partitions to loops
5462 when available. Add TILE handling.
5463 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
5464 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
5465 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
5466 * tree.c (omp_clause_num_ops): Adjust TILE ops.
5467 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
5469 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
5471 * configure.ac (ACX_BUGURL): Update.
5472 * configure: Regenerate.
5474 2017-02-09 Richard Biener <rguenther@suse.de>
5476 PR tree-optimization/69823
5477 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
5478 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
5480 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
5482 * config/arc/arc-c.def: Add __NPS400__ definition.
5483 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
5484 (TARGET_NPS400): Define.
5486 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
5488 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
5490 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
5492 (arc_cpu_types): Fill the arch_info field with a pointer into the
5493 arc_arch_types table.
5494 (arc_selected_cpu): Declare.
5495 * config/arc/arc.c (arc_selected_cpu): Make global.
5496 (arc_selected_arch): Delete.
5497 (arc_base_cpu): Delete.
5498 (arc_override_options): Remove references to deleted variables,
5499 update access to arch information.
5500 (ARC_OPT): Update access to arch information.
5501 (ARC_OPTX): Likewise.
5502 * config/arc/arc.h (arc_base_cpu): Remove declaration.
5503 (TARGET_ARC600): Update access to arch information.
5504 (TARGET_ARC601): Likewise.
5505 (TARGET_ARC700): Likewise.
5506 (TARGET_EM): Likewise.
5507 (TARGET_HS): Likewise.
5508 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
5511 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
5514 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
5515 condition/operands for integer GE/LE/GEU/LEU operations.
5517 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
5519 PR translation/79397
5520 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
5523 2017-02-08 Martin Jambor <mjambor@suse.cz>
5526 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
5527 whether allocation happened.
5528 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
5529 nothing was allocated.
5531 2017-02-08 Jakub Jelinek <jakub@redhat.com>
5533 PR tree-optimization/79408
5534 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
5535 constant, but SSA_NAME with a known integer range, use the minimum
5536 of that range instead of op1 to determine if modulo can be replaced
5537 with its first operand.
5539 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5541 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
5543 2017-02-08 Richard Biener <rguenther@suse.de>
5545 PR tree-optimization/71824
5546 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
5547 Check all loops contained in the merged region.
5549 2017-02-07 Andrew Pinski <apinski@cavium.com>
5551 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
5553 2017-02-07 Andrew Pinski <apinski@cavium.com>
5555 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
5556 (thunderxt88): Likewise.
5557 (thunderxt81): Disable LSE and change v8.1 to v8.
5558 (thunderxt83): Likewise.
5560 2017-02-07 Jakub Jelinek <jakub@redhat.com>
5561 Richard Biener <rguenther@suse.de>
5564 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
5565 type from int to size_t.
5566 * ira-costs.c (struct_costs_size): Change type from int to size_t.
5568 2017-02-07 Jakub Jelinek <jakub@redhat.com>
5570 PR rtl-optimization/79386
5571 * cprop.c (bypass_conditional_jumps): Initialize
5572 bypass_last_basic_block already before splitting bbs after
5573 unconditional traps...
5574 (bypass_conditional_jumps): ... rather than here.
5577 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
5578 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
5579 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
5580 fix -masm=intel patterns.
5582 2017-02-07 Richard Biener <rguenther@suse.de>
5584 PR tree-optimization/79256
5586 * builtins.c (get_object_alignment_2): Use min_align_of_type
5587 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
5588 and ADJUST_FIELD_ALIGN.
5590 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
5592 * doc/tm.texi: Regenerate.
5593 * stor-layout.c (layout_decl): Adjust.
5594 (update_alignment_for_field): Likewise.
5595 (place_field): Likewise.
5596 (min_align_of_type): Likewise.
5597 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
5598 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
5599 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
5600 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
5601 * config/frv/frv.c (frv_adjust_field_align): Likewise.
5602 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
5603 * config/i386/i386.c (x86_field_alignment): Likewise.
5604 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
5605 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
5606 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
5607 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
5608 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
5609 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
5613 2017-01-30 Richard Biener <rguenther@suse.de>
5615 PR tree-optimization/79256
5616 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
5617 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
5620 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
5622 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
5623 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
5624 builtins to SImode and emit a zero-extend, if necessary.
5626 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
5628 * docs/invoke.texi (RISC-V Options): Alphabetize.
5630 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
5632 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
5635 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
5637 * config/riscv/riscv.c: New file.
5638 * gcc/common/config/riscv/riscv-common.c: Likewise.
5639 * config.gcc: Likewise.
5640 * config/riscv/constraints.md: Likewise.
5641 * config/riscv/elf.h: Likewise.
5642 * config/riscv/generic.md: Likewise.
5643 * config/riscv/linux.h: Likewise.
5644 * config/riscv/multilib-generator: Likewise.
5645 * config/riscv/peephole.md: Likewise.
5646 * config/riscv/pic.md: Likewise.
5647 * config/riscv/predicates.md: Likewise.
5648 * config/riscv/riscv-builtins.c: Likewise.
5649 * config/riscv/riscv-c.c: Likewise.
5650 * config/riscv/riscv-ftypes.def: Likewise.
5651 * config/riscv/riscv-modes.def: Likewise.
5652 * config/riscv/riscv-opts.h: Likewise.
5653 * config/riscv/riscv-protos.h: Likewise.
5654 * config/riscv/riscv.h: Likewise.
5655 * config/riscv/riscv.md: Likewise.
5656 * config/riscv/riscv.opt: Likewise.
5657 * config/riscv/sync.md: Likewise.
5658 * config/riscv/t-elf-multilib: Likewise.
5659 * config/riscv/t-linux: Likewise.
5660 * config/riscv/t-linux-multilib: Likewise.
5661 * config/riscv/t-riscv: Likewise.
5662 * configure.ac: Likewise.
5663 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
5664 Waterman as RISC-V maintainers.
5665 * doc/install.texi: Add RISC-V entries.
5666 * doc/invoke.texi: Add RISC-V options section.
5667 * doc/md.texi: Add RISC-V constraints section.
5668 * configure: Regenerated.
5670 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5673 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
5674 false values to be constant vectors with all 0 or all 1 bits set.
5675 (vcondu<mode><mode>): Likewise.
5676 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
5678 (fpmask_comparison_operator): Update comment.
5679 (vecint_comparison_operator): New predicate.
5680 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
5681 vector conditionals when the true and false values are constant
5682 vectors with all 0 bits or all 1 bits set.
5684 2017-02-06 Martin Sebor <msebor@redhat.com>
5686 PR tree-optimization/79376
5687 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
5689 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
5691 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
5692 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
5693 to simplify split condition.
5695 2017-02-06 Jakub Jelinek <jakub@redhat.com>
5697 * omp-expand.c (oxpand_omp_atomic_fetch_op,
5698 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
5701 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
5703 PR rtl-optimization/68664
5704 * target.def (can_speculate_insn): New hook.
5705 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
5706 * doc/tm.texi: Regenerate.
5707 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
5708 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
5709 (rs6000_sched_can_speculate_insn): New function.
5711 2017-02-06 Jakub Jelinek <jakub@redhat.com>
5713 PR tree-optimization/79284
5714 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
5715 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
5716 vectorizable_mask_load_store, vectorizable_operation,
5717 vect_is_simple_cond, get_same_sized_vectype): Use it instead
5718 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
5719 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
5720 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
5721 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5722 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
5723 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
5724 is_gimple_assign (stmt). Replace another such test with
5725 is_gimple_assign (stmt).
5727 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
5730 * config/avr/avr.c (rtl-iter.h): Include it.
5731 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
5732 (avr_legitimate_combined_insn): ...and implementation.
5734 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
5736 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
5737 * config/s390/s390.c (s390_const_operand_ok)
5738 (s390_canonicalize_comparison, s390_extract_part)
5739 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
5740 (s390_contiguous_bitmask_p, s390_rtx_costs)
5741 (legitimize_pic_address): Likewise.
5742 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
5743 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
5744 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
5745 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
5746 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
5748 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
5750 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
5751 REGNO($0) == REGNO($1).
5753 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5755 * config/s390/linux.h(SIZE_TYPE): Add comment.
5757 2017-02-06 Julian Brown <julian@codesourcery.com>
5758 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5759 Virendra Pathak <virendra.pathak@broadcom.com>
5761 * config/aarch64/aarch64-cores.def: Change the scheduler
5763 * config/aarch64/aarch64.md: Include thunderx2t99.md.
5764 * config/aarch64/thunderx2t99.md: New file.
5766 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5768 * doc/standards.texi (Go Language): Update link to language
5771 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
5773 * tree-eh.c (lower_resx): Sanitize profile.
5774 (cleanup_empty_eh_move_lp): Likewise.
5776 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
5779 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
5781 * cfgloopmanip.h (loop_version): Update prototype.
5782 * modulo-sched.c (sms_schedule): Update call of loop_version.
5783 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
5784 * tree-parloops.c (gen_parallel_loop): Likewise.
5785 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
5786 * tree-ssa-loop-split.c (split_loop): Likewise.
5787 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
5788 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
5790 2017-02-05 Martin Liska <mliska@suse.cz>
5793 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
5795 (print_operand_address): Initialize a struct to zero.
5797 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5799 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
5800 garbage collector only in textual form.
5802 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5804 * doc/extend.texi (x86 specific memory model extensions for
5805 transactional memory): Simplify a phrase.
5807 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
5810 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
5811 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
5812 (atomic_storedi_1): Likewise.
5814 2017-02-04 Jakub Jelinek <jakub@redhat.com>
5816 PR tree-optimization/79338
5817 * tree-parloops.c (gather_scalar_reductions): Don't call
5818 vect_analyze_loop_form for loop->inner before destroying loop's
5821 2017-02-03 Martin Sebor <msebor@redhat.com>
5823 PR tree-optimization/79327
5824 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
5825 when precision has resulted in leading zeros.
5826 (format_integer): Adjust the likely counter to assume an unknown
5827 argument that may be zero is non-zero.
5829 2017-02-03 Jason Merrill <jason@redhat.com>
5832 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
5833 avoid copying non-taken branch.
5835 2017-02-03 Jakub Jelinek <jakub@redhat.com>
5837 PR tree-optimization/79340
5838 * tree-vect-loop.c (vectorizable_reduction): Release
5839 vec_defs elements after safe_splicing them into other vectors.
5842 PR tree-optimization/79327
5843 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
5844 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
5846 (format_integer): Use wide_int_to_tree instead of build_int_cst
5847 + to_?hwi. If argmin is NULL, just set argmin and argmax to
5848 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
5849 of shortest and longest sequence.
5851 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
5853 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
5854 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
5856 2017-02-03 Walter Lee <walt@tilera.com>
5859 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
5860 after initial stackframe link reg save.
5861 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
5863 2017-02-03 Jakub Jelinek <jakub@redhat.com>
5866 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
5867 wu for stxssp alternative.
5869 2017-02-03 Martin Sebor <msebor@redhat.com>
5871 PR tree-optimization/79352
5872 * gimple-fold.c (get_range_strlen): Add argument.
5873 (get_range_strlen): Change return type to bool.
5874 (get_maxval_strlen): Pass in a dummy argument.
5875 * gimple-fold.h (get_range_strlen): Change return type to bool.
5876 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
5877 * tree.h (array_at_struct_end_p): Add argument.
5878 * tree.c (array_at_struct_end_p): Handle it.
5880 2017-02-03 Martin Liska <mliska@suse.cz>
5883 * multiple_target.c (create_dispatcher_calls): Redirect edge
5884 from a caller of a dispatcher.
5885 (expand_target_clones): Make the clones local.
5886 (ipa_target_clone): Do both target clones and resolvers.
5887 (ipa_dispatcher_calls): Remove the pass.
5888 (pass_dispatcher_calls::gate): Likewise.
5889 (make_pass_dispatcher_calls): Likewise.
5890 * passes.def (pass_target_clone): Put as very first IPA early
5893 2017-02-03 Martin Liska <mliska@suse.cz>
5895 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
5896 in case of a function with ifunc attribute.
5898 2017-02-03 Martin Liska <mliska@suse.cz>
5900 * cgraph.c (cgraph_node::dump): Dump function version info.
5901 * symtab.c (symtab_node::dump_base): Add missing new line.
5903 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
5905 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
5906 (ifcombine_ifandif): Use it.
5908 2017-02-03 Martin Liska <mliska@suse.cz>
5910 * doc/invoke.texi: Document default value for
5911 use-after-scope-direct-emission-threshold.
5913 2017-02-03 Martin Liska <mliska@suse.cz>
5915 PR tree-optimization/79339
5916 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
5917 (format_floating): Likewise.
5919 2017-02-03 Martin Liska <mliska@suse.cz>
5922 * ipa-prop.c (ipa_node_params_t::insert): Remove current
5924 (ipa_node_params_t::remove): Likewise.
5925 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
5926 initialization from removed ipa_node_params_t::insert.
5927 (ipa_node_params::~ipa_node_params): Move from removed
5928 ipa_node_params_t::release.
5929 * symbol-summary.h (symbol_summary::m_released): New member.
5930 Do not release a summary twice. Do not allow to call finalizer
5931 for types of a summary that live in GGC memory.
5933 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5935 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
5938 2017-02-02 Martin Sebor <msebor@redhat.com>
5941 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
5942 (format_string): Tighten up the range of output for non-constant
5943 strings and correct the expected range for wide non-constant strings.
5945 2017-02-02 Martin Sebor <msebor@redhat.com>
5947 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
5950 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
5952 (-fdump-tree-@var): Add to index and document how to come up
5953 with pass-specific option and dump file names.
5954 (-fdump-passes): Clarify where to look for output.
5956 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
5959 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
5960 statistics of the analyzed path; allow threading for speed when
5961 any of BBs along the path are optimized for speed.
5963 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
5966 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
5967 settings of the virtual registers.
5970 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
5972 * explow.c (get_dynamic_stack_size): Take known alignment of stack
5973 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
5976 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5978 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
5979 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
5981 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5983 * config/s390/s390.md: Add missing comments with the expanded
5985 * config/s390/vector.md: Likewise.
5986 * config/s390/vx-builtins.md: Likewise.
5988 2017-02-02 Jakub Jelinek <jakub@redhat.com>
5991 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
5992 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
5993 conditions on a single line.
5995 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5997 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
5998 __S390_VX__ to __VX__.
6000 2017-02-01 Andrew Pinski <apinski@cavium.com>
6002 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
6003 stmt_info to record_stmt_cost.
6004 (vect_get_known_peeling_cost): Pass stmt_info if known to
6006 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
6007 cpu_vector_cost field into
6008 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
6009 field into vec_int_stmt_cost and vec_fp_stmt_cost.
6010 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
6011 splitting of scalar_stmt_cost and vec_stmt_cost.
6012 (thunderx_vector_cost): Likewise.
6013 (cortexa57_vector_cost): LIkewise.
6014 (exynosm1_vector_cost): Likewise.
6015 (xgene1_vector_cost): Likewise.
6016 (thunderx2t99_vector_cost): Improve after the splitting of the two
6018 (aarch64_builtin_vectorization_cost): Update for the splitting of
6019 scalar_stmt_cost and vec_stmt_cost.
6021 2017-02-01 Torvald Riegel <triegel@redhat.com>
6022 Richard Henderson <rth@redhat.com>
6024 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
6025 conditional on existance of a fast atomic load.
6026 * optabs-query.c (can_atomic_load_p): New function.
6027 * optabs-query.h (can_atomic_load_p): Declare it.
6028 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
6029 no fast atomic load is available for the particular size of access.
6030 (expand_atomic_compare_and_swap): Likewise.
6031 (expand_atomic_load): Likewise.
6032 (expand_atomic_store): Likewise.
6033 (expand_atomic_fetch_op): Likewise.
6034 * testsuite/lib/target-supports.exp
6035 (check_effective_target_sync_int_128): Remove x86 because it provides
6036 no fast atomic load.
6037 (check_effective_target_sync_int_128_runtime): Likewise.
6039 2017-02-01 Richard Biener <rguenther@suse.de>
6041 * graphite.c: Include tree-vectorizer.h for find_loop_location.
6042 (graphite_transform_loops): Provide opt-info for optimized nests.
6043 * tree-parloop.c (parallelize_loops): Provide opt-info for
6046 2017-02-01 Richard Biener <rguenther@suse.de>
6049 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
6052 2017-02-01 Richard Biener <rguenther@suse.de>
6054 PR tree-optimization/71824
6055 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
6056 Verify the loops are valid in the merged SESE region.
6057 (scop_detection::can_represent_loop_1): Check analyzing the
6058 evolution of the number of iterations in the region succeeds.
6060 2017-01-31 Ian Lance Taylor <iant@golang.org>
6062 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
6063 REG_ARGS_SIZE note to 32-bit push insns and call insn.
6065 2017-01-31 David Malcolm <dmalcolm@redhat.com>
6067 PR preprocessor/79210
6068 * input.c (get_substring_ranges_for_loc): Replace line_width
6069 assertion with error-handling.
6071 2017-01-31 Richard Biener <rguenther@suse.de>
6073 PR tree-optimization/77318
6074 * graphite-sese-to-poly.c (extract_affine): Fix assert.
6075 (create_pw_aff_from_tree): Take loop parameter.
6076 (add_condition_to_pbb): Pass loop of the condition to
6077 create_pw_aff_from_tree.
6079 2017-01-31 Jakub Jelinek <jakub@redhat.com>
6081 * config/s390/s390.c (s390_asan_shadow_offset): New function.
6082 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
6084 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6088 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
6090 (convert_int_to_float128): Likewise.
6091 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
6092 (convert_int_to_float128): Likewise.
6093 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
6094 (UNSPEC_IEEE128_CONVERT): Likewise.
6095 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
6096 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
6097 Use local variables for IBM extended format.
6098 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
6099 (fix_trunc<mode>si2_fprs): Likewise.
6100 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
6101 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
6102 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
6103 to know that we can now have integers of all sizes in vector
6105 (fix<uns>_<mode>di2_hw): Likewise.
6106 (float<uns>_<mode>si2_hw): Likewise.
6107 (fix_<mode>si2_hw): Likewise.
6108 (fixuns_<mode>si2_hw): Likewise.
6109 (float<uns>_<mode>di2_hw): Likewise.
6110 (float_<mode>di2_hw): Likewise.
6111 (float_<mode>si2_hw): Likewise.
6112 (floatuns_<mode>di2_hw): Likewise.
6113 (floatuns_<mode>si2_hw): Likewise.
6114 (xscvqp<su>wz_<mode>): Delete, no longer used.
6115 (xscvqp<su>dz_<mode>): Likewise.
6116 (xscv<su>dqp_<mode>): Likewise.
6117 (ieee128_mfvsrd_64bit): Likewise.
6118 (ieee128_mfvsrd_32bit): Likewise.
6119 (ieee128_mfvsrwz): Likewise.
6120 (ieee128_mtvsrw): Likewise.
6121 (ieee128_mtvsrd_64bit): Likewise.
6122 (ieee128_mtvsrd_32bit): Likewise.
6124 2017-01-31 Martin Liska <mliska@suse.cz>
6127 * ipa-prop.c (ipa_free_all_node_params): Call release method
6128 instead of ~sumbol_summary to not to trigger double times
6131 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
6133 PR tree-optimization/71691
6134 * bitmap.h (class auto_bitmap): New.
6135 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
6136 is_maybe_undefined instead of ssa_undefined_value_p.
6138 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6140 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
6141 __S390_ARCH_LEVEL__ to __ARCH__.
6143 2017-01-31 Jakub Jelinek <jakub@redhat.com>
6145 PR tree-optimization/79267
6146 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
6147 if should_remove_lhs_p is true.
6149 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
6152 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
6153 (add_alignment_attribute): New.
6154 (base_type_die): Add alignment attribute.
6155 (subrange_type_die): Likewise.
6156 (modified_type_die): Likewise.
6157 (gen_array_type_die): Likewise.
6158 (gen_descr_array_type_die: Likewise.
6159 (gen_enumeration_type_die): Likewise.
6160 (gen_subprogram_die): Likewise.
6161 (gen_variable_die): Likewise.
6162 (gen_field_die): Likewise.
6163 (gen_ptr_to_mbr_type_die): Likewise.
6164 (gen_struct_or_union_type_die): Likewise.
6165 (gen_subroutine_type_die): Likewise.
6166 (gen_typedef_die): Likewise.
6167 (base_type_cmp): Compare alignment attribute.
6169 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6172 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
6173 (setb_unsigned) New pattern for setb with CCUNS.
6174 * config/rs6000/rs6000.c (expand_block_compare): Use a different
6175 subfc./subfe sequence to avoid overflow problems. Generate a
6176 shorter sequence with cmpld/setb for power9.
6177 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
6178 for generating subfc. instruction.
6179 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
6180 now uses this instruction.
6182 2017-01-30 Ian Lance Taylor <iant@google.com>
6185 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
6186 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
6188 2017-01-30 Martin Sebor <msebor@redhat.com>
6190 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
6191 Move constant to the right of a relational operator.
6192 (get_mpfr_format_length, format_character, format_string): Ditto.
6193 (should_warn_p, maybe_warn): Same.
6195 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
6197 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
6200 * asan.c (get_translation_unit_decl): Remove function.
6201 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
6203 2017-01-30 Martin Liska <mliska@suse.cz>
6205 PR gcov-profile/79259
6206 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
6209 2017-01-30 Martin Liska <mliska@suse.cz>
6212 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
6213 Initialize variables with NULL value.
6215 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
6218 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
6220 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
6222 2017-01-30 Richard Biener <rguenther@suse.de>
6224 PR tree-optimization/79276
6225 * tree-vrp.c (process_assert_insertions): Properly adjust common
6226 when removing a duplicate.
6228 * gcc.dg/torture/pr79276.c: New testcase.
6230 2017-01-30 Richard Biener <rguenther@suse.de>
6232 PR tree-optimization/79256
6233 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
6234 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
6236 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
6238 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
6241 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
6242 ("*r<noxa>sbg_<mode>_sll_bitmask")
6243 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
6244 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
6245 Use contiguous_bitmask_nowrap_operand.
6247 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6250 * config/rs6000/altivec.h (vec_xl): Revise #define.
6251 (vec_xst): Likewise.
6253 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
6255 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
6257 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
6259 PR rtl-optimization/79194
6260 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
6261 traps before call to bypass_conditional_jumps.
6263 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
6265 PR tree-optimization/71374
6266 * lra-constraints.c (check_conflict_input_operands): New.
6267 (match_reload): Use it.
6269 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
6272 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
6273 account to calculate conflict_set.
6275 2017-01-27 Bin Cheng <bin.cheng@arm.com>
6277 PR rtl-optimization/78559
6278 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
6279 other_insn in combine.
6281 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
6283 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
6284 uint16_type_node for BT_UINT16.
6286 2017-01-27 David Malcolm <dmalcolm@redhat.com>
6288 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
6289 "RTL Tests" to menu.
6290 (GIMPLE Tests): New node.
6291 (RTL Tests): New node.
6293 2017-01-27 Richard Biener <rguenther@suse.de>
6295 PR tree-optimization/79245
6296 * tree-loop-distribution.c (distribute_loop): Apply cost
6297 modeling also to detected patterns.
6299 2017-01-27 Richard Biener <rguenther@suse.de>
6301 PR tree-optimization/71433
6302 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
6303 (compare_assert_loc): New function.
6304 (process_assert_insertions): Sort and optimize assert locations
6305 to remove duplicates and push down identical assertions on
6306 edges to their destination block.
6308 2017-01-27 Richard Biener <rguenther@suse.de>
6310 PR tree-optimization/79244
6311 * tree-vrp.c (remove_range_assertions): Forcefully propagate
6312 out SSA names even if abnormal.
6314 2017-01-27 Jakub Jelinek <jakub@redhat.com>
6316 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
6317 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
6318 instead of MPFR_RNDN.
6320 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
6323 * arm.c (arm_option_override): Don't call build_target_option_node
6324 until after doing all option overrides.
6325 (arm_valid_target_attribute_tree): Likewise.
6327 2017-01-27 Martin Liska <mliska@suse.cz>
6329 * doc/invoke.texi (-fprofile-arcs): Document profiling support
6330 for {cd}tors and C++ {cd}tors.
6332 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
6334 * config/s390/s390.md ("*setmem_long_and")
6335 ("*setmem_long_and_31z"): Use zero_extend instead of and.
6337 2017-01-26 Martin Sebor <msebor@redhat.com>
6339 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
6342 2017-01-26 Martin Sebor <msebor@redhat.com>
6344 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
6345 HAVE_DFmode before using XFmode or DFmode.
6346 (parse_directive): Avoid using the z length modifier to avoid
6347 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
6350 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
6351 to accept adjustment as an array.
6352 (get_int_range): New function.
6353 (struct directive): Make width and prec arrays.
6354 (directive::set_width, directive::set_precision): Call get_int_range.
6355 (format_integer, format_floating): Handle width and precision ranges.
6356 (format_string, parse_directive): Same.
6358 2017-01-26 Jakub Jelinek <jakub@redhat.com>
6361 * dwarf2out.c (generate_skeleton_bottom_up): For children with
6362 comdat_type_p set, just clone them, but keep the children in the
6366 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
6367 which have direct callers with -fvar-tracking-assignments enabled
6369 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
6370 inside of type units.
6372 2017-01-26 Martin Sebor <msebor@redhat.com>
6375 * gimple-ssa-sprintf.c (struct result_range): Add likely and
6377 (struct format_result): Replace number_chars, number_chars_min,
6378 and number_chars_max with a single member of struct result_range.
6380 (format_result::operator+=): Adjust.
6381 (struct fmtresult): Remove bounded. Handle likely and unlikely
6383 (fmtresult::adjust_for_width_or_precision): New function.
6384 (fmtresult:type_max_digits): New function.
6385 (bytes_remaining): Handle likely and unlikely counters.
6386 (min_bytes_remaining): Remove.
6387 (format_percent): Simplify.
6388 (format_integer, format_floating): Set likely and unlikely counters.
6389 (get_string_length, format_character, format_string): Same.
6390 (format_plain, should_warn_p): New function.
6391 (maybe_warn): Call should_warn_p. Update diagnostic messages
6392 and handle those for all directives, including plain strings.
6393 (format_directive): Handle likely and unlikely counters.
6394 Remove unnecessary quoting from diagnostics. Add an informational
6396 (add_bytes): Remove.
6397 (pass_sprintf_length::compute_format_length): Simplify.
6398 (try_substitute_return_value): Handle likely and unlikely counters.
6400 2017-01-26 Carl Love <cel@us.ibm.com>
6402 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
6403 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
6405 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
6408 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
6409 endianess for subregs into account.
6410 * lra-constraints.c (lra_constraints): Do risky transformations
6411 always on the first iteration.
6412 * lra-lives.c (check_pseudos_live_through_calls): Add arg
6413 last_call_used_reg_set.
6414 (process_bb_lives): Define and use last_call_used_reg_set.
6415 * lra.c (lra): Always continue after lra_constraints on the first
6418 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
6420 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
6422 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
6424 2017-01-26 Jakub Jelinek <jakub@redhat.com>
6426 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
6427 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
6428 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
6429 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
6430 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
6431 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
6432 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
6433 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
6434 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
6436 2017-01-26 Marek Polacek <polacek@redhat.com>
6439 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
6440 for the third operand.
6442 2017-01-26 Jakub Jelinek <jakub@redhat.com>
6445 * omp-low.c (struct omp_context): Add simt_stmt field.
6446 (scan_omp_for): Return omp_context *.
6447 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
6448 context to the _simt_ SIMD stmt.
6449 (lower_omp_for): For combined SIMD with sibling _simt_
6450 SIMD, make sure to use the same decls in _looptemp_
6451 clauses as in the sibling.
6453 2017-01-26 David Sherwood <david.sherwood@arm.com>
6456 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
6459 2017-01-26 Jakub Jelinek <jakub@redhat.com>
6462 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
6463 emit fld b; fld a; if possible.
6465 * brig-builtins.def: Update copyright years.
6466 * config/arm/arm_acle_builtins.def: Update copyright years.
6468 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6471 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
6472 constraint instead of o for the stxsd instruction.
6474 2017-01-25 Carl Love <cel@us.ibm.com>
6476 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
6477 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
6479 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
6481 * doc/invoke.texi (C++ Dialect Options): Fix typo.
6483 2017-01-25 Richard Biener <rguenther@suse.de>
6485 PR tree-optimization/69264
6486 * target.def (vector_alignment_reachable): Improve documentation.
6487 * doc/tm.texi: Regenerate.
6488 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
6490 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
6491 earlier changes with respect to TYPE_USER_ALIGN.
6492 (vector_alignment_reachable_p): Likewise. Improve dumping.
6494 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6497 * config/arm/arm.md (xordi3): Force constant operand into a register
6500 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6502 * doc/invoke.texi (-fstore-merging): Correct default optimization
6503 levels at which it is enabled.
6504 (-O): Move -fstore-merging from list to...
6507 2017-01-25 Richard Biener <rguenther@suse.de>
6510 * omp-expand.c: Include debug.h.
6511 (expand_omp_taskreg): Make sure to generate early debug before
6512 outlining anything from a function.
6513 (expand_omp_target): Likewise.
6514 (grid_expand_target_grid_body): Likewise.
6516 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
6519 * asan.c (get_translation_unit_decl): New function.
6520 (asan_add_global): Extract modules file name from globals
6521 TRANSLATION_UNIT_DECL name.
6523 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
6526 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
6527 for long calls with APCS frame and VFP.
6529 2017-01-24 David Malcolm <dmalcolm@redhat.com>
6531 * cfg.c (original_copy_tables_initialized_p): New function.
6532 * cfg.h (original_copy_tables_initialized_p): New decl.
6533 * cfgrtl.c (relink_block_chain): Guard the call to
6534 free_original_copy_tables with a call to
6535 original_copy_tables_initialized_p.
6536 * cgraph.h (symtab_node::native_rtl_p): New decl.
6537 * cgraphunit.c (symtab_node::native_rtl_p): New function.
6538 (symtab_node::needed_p): Don't assert for early assembly output
6539 for __RTL functions.
6540 (cgraph_node::finalize_function): Set "force_output" for __RTL
6542 (cgraph_node::analyze): Bail out early for __RTL functions.
6543 (analyze_functions): Update assertion to support __RTL functions.
6544 (cgraph_node::expand): Bail out early for __RTL functions.
6545 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
6547 * function.h (struct function): Update comment for field
6549 * gimple-expr.c: Include "tree-pass.h".
6550 (gimple_has_body_p): Return false for __RTL functions.
6551 * Makefile.in (OBJS): Add run-rtl-passes.o.
6552 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
6554 (gcc::pass_manager::get_clean_slate): New accessor.
6555 * passes.c: Include "insn-addr.h".
6556 (should_skip_pass_p): Add logging. Update logic for running
6557 "expand" to be compatible with both __GIMPLE and __RTL. Guard
6558 property-provider override so it is only done for gimple passes.
6560 (skip_pass): New function.
6561 (execute_one_pass): Call skip_pass when skipping passes.
6562 * read-md.c (md_reader::read_char): Support filtering
6563 the input to a subset of line numbers.
6564 (md_reader::md_reader): Initialize fields
6565 m_first_line and m_last_line.
6566 (md_reader::read_file_fragment): New function.
6567 * read-md.h (md_reader::read_file_fragment): New decl.
6568 (md_reader::m_first_line): New field.
6569 (md_reader::m_last_line): New field.
6570 * read-rtl-function.c (function_reader::create_function): Only
6571 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
6572 curr_properties. Set DECL_INITIAL to a dummy block.
6573 (read_rtl_function_body_from_file_range): New function.
6574 * read-rtl-function.h (read_rtl_function_body_from_file_range):
6576 * run-rtl-passes.c: New file.
6577 * run-rtl-passes.h: New file.
6579 2017-01-24 Jeff Law <law@redhat.com>
6581 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
6584 2017-01-24 Bin Cheng <bin.cheng@arm.com>
6586 PR tree-optimization/79159
6587 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
6588 (record_nonwrapping_iv): Improve boundary using above function if no
6589 value range information.
6591 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
6592 Martin Jambor <mjambor@suse.cz>
6594 * brig-builtins.def: New file.
6595 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
6596 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
6597 (DEF_HSAIL_SAT_BUILTIN): Likewise.
6598 (DEF_HSAIL_INTR_BUILTIN): Likewise.
6599 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
6600 * builtin-types.def (BT_INT8): New.
6601 (BT_INT16): Likewise.
6602 (BT_UINT8): Likewise.
6603 (BT_UINT16): Likewise.
6604 (BT_FN_ULONG): Likewise.
6605 (BT_FN_UINT_INT): Likewise.
6606 (BT_FN_UINT_ULONG): Likewise.
6607 (BT_FN_UINT_LONG): Likewise.
6608 (BT_FN_UINT_PTR): Likewise.
6609 (BT_FN_ULONG_PTR): Likewise.
6610 (BT_FN_INT8_FLOAT): Likewise.
6611 (BT_FN_INT16_FLOAT): Likewise.
6612 (BT_FN_UINT32_FLOAT): Likewise.
6613 (BT_FN_UINT16_FLOAT): Likewise.
6614 (BT_FN_UINT8_FLOAT): Likewise.
6615 (BT_FN_UINT64_FLOAT): Likewise.
6616 (BT_FN_UINT16_UINT32): Likewise.
6617 (BT_FN_UINT32_UINT16): Likewise.
6618 (BT_FN_UINT16_UINT16_UINT16): Likewise.
6619 (BT_FN_INT_PTR_INT): Likewise.
6620 (BT_FN_UINT_PTR_UINT): Likewise.
6621 (BT_FN_LONG_PTR_LONG): Likewise.
6622 (BT_FN_ULONG_PTR_ULONG): Likewise.
6623 (BT_FN_VOID_UINT64_UINT64): Likewise.
6624 (BT_FN_UINT8_UINT8_UINT8): Likewise.
6625 (BT_FN_INT8_INT8_INT8): Likewise.
6626 (BT_FN_INT16_INT16_INT16): Likewise.
6627 (BT_FN_INT_INT_INT): Likewise.
6628 (BT_FN_UINT_FLOAT_UINT): Likewise.
6629 (BT_FN_FLOAT_UINT_UINT): Likewise.
6630 (BT_FN_ULONG_UINT_UINT): Likewise.
6631 (BT_FN_ULONG_UINT_PTR): Likewise.
6632 (BT_FN_ULONG_ULONG_ULONG): Likewise.
6633 (BT_FN_UINT_UINT_UINT): Likewise.
6634 (BT_FN_VOID_UINT_PTR): Likewise.
6635 (BT_FN_UINT_UINT_PTR: Likewise.
6636 (BT_FN_UINT32_UINT64_PTR): Likewise.
6637 (BT_FN_INT_INT_UINT_UINT): Likewise.
6638 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
6639 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
6640 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
6641 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
6642 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
6643 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
6644 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
6645 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
6646 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
6647 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
6648 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
6649 * doc/frontends.texi: List BRIG FE.
6650 * doc/install.texi (Testing): Add BRIG tesring requirements.
6651 * doc/invoke.texi (Overall Options): Mention BRIG.
6652 * doc/standards.texi (Standards): Doucment BRIG HSA version.
6654 2017-01-24 Richard Biener <rguenther@suse.de>
6656 PR translation/79208
6657 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
6659 2017-01-24 Martin Jambor <mjambor@suse.cz>
6662 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
6663 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
6666 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
6669 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
6670 casts from signed to unsigned really don't have a range.
6672 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
6674 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
6675 GMP_RNDx for compatiblity.
6677 2017-01-24 Martin Liska <mliska@suse.cz>
6680 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
6681 that would prevent us to call alloca with -1 as argument.
6683 2017-01-24 Jakub Jelinek <jakub@redhat.com>
6685 * dwarf2out.c (output_compilation_unit_header, output_file_names):
6686 Avoid -Wformat-security warning.
6688 2017-01-23 Andrew Pinski <apinski@cavium.com>
6690 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
6693 2017-01-23 Martin Sebor <msebor@redhat.com>
6696 * gimple-ssa-sprintf.c (warn_level): New global.
6697 (format_integer): Use it here and throughout the rest of the file.
6698 Use the same switch to compute sign as base.
6699 (maybe_warn): New function.
6700 (format_directive): Factor out warnings into maybe_warn.
6701 Add debugging output. Use warn_level.
6702 (add_bytes): Use warn_level.
6703 (pass_sprintf_length::compute_format_length): Add debugging output.
6704 (try_substitute_return_value): Same.
6705 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
6708 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
6709 (struct fmtresult, format_integer, format_floating): Adjust.
6710 (fmtresult::fmtresult): Set max correctly in two argument ctor.
6711 (get_string_length, format_string,format_directive): Same.
6712 (pass_sprintf_length::compute_format_length): Same.
6713 (try_substitute_return_value): Simplify slightly.
6716 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
6717 (fmtresult::operator+=): Outlined.
6718 (struct fmtresult): Add ctors.
6719 (struct conversion_spec): Rename...
6720 (struct directive): ...to this. Add and remove data members.
6721 (directive::set_width, directive::set_precision): New functions.
6722 (format_percent): Use fmtresult ctor.
6723 (get_width_and_precision): Remove.
6724 (format_integer): Make naming changes. Avoid computing width and
6726 (format_floating): Same. Adjust indentation.
6727 (format_character, format_none): New functions.
6728 (format_string): Moved character handling to format_character.
6729 (format_directive): Remove arguments, change return type.
6730 (parse_directive): New function.
6731 (pass_sprintf_length::compute_format_length): Move directive
6732 parsing to parse_directive.
6734 2017-01-23 Jakub Jelinek <jakub@redhat.com>
6736 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
6737 (assign_assembler_name_if_needed): ... this.
6738 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
6739 (assign_assembler_name_if_needed): ... this.
6740 (free_lang_data_in_cgraph): Adjust callers.
6741 * cgraphunit.c (cgraph_node::analyze): Likewise.
6742 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
6745 2017-01-23 Richard Biener <rguenther@suse.de>
6747 PR tree-optimization/79088
6748 PR tree-optimization/79188
6749 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
6750 resetting loop bounds after last path deletion. Reset loop
6751 bounds of the target loop, make code match the comments.
6752 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
6753 Make sure loops need no fixups.
6755 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
6757 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
6758 exponent support with double type for first argument.
6759 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
6760 type returned by __builtin_vec_extract_sig,
6761 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
6762 functions from "vector int" to "vector unsigned int" or from
6763 "vector long long int" to "vector unsigned long long int".
6764 Changed type returned by __builtin_vec_extract_exp,
6765 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
6766 functions from "vector int" to "vector unsigned int" or from
6767 "vector long long int" to "vector unsigned long long int".
6768 Changed return type of __builtin_vec_test_data_class,
6769 __builtin_vec_test_data_class_sp, and
6770 __builtin_vec_test_data_class_dp from "vector int" to
6771 "vector bool int" or from "vector long long int" to "vector bool
6772 long long int" and changed second argument type from "unsigned
6773 int" to "int". Added new overloaded function forms "vector float
6774 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
6775 "vector float __builtin_vec_insert_exp_sp (vector float, vector
6776 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
6777 double, vector unsigned long long int)" and "vector double
6778 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
6779 long int)". Changed return type of
6780 __builtin_scalar_test_data_class and
6781 __builtin_scalar_test_data_class_sp and
6782 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
6783 int" and changed second argument from "unsigned int" to "int".
6784 Changed type returned by __builtin_scalar_test_neg,
6785 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
6786 from "int" to "bool int". Added new overloaded function form
6787 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
6788 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
6789 exponent double-precision with floating point first argument.
6790 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
6791 documentation of scalar_test_data_class, scalar_test_neg,
6792 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
6793 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
6794 vec_test_data_class built-in functions to reflect refinements in
6795 their type signatures.
6797 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
6799 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
6801 (aarch64_elf_asm_destructor): Likewise.
6803 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
6805 PR rtl-optimization/78634
6806 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
6807 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
6808 * ifcvt.c (noce_try_cmove): Add missing cost check.
6810 PR rtl-optimization/71724
6811 * combine.c (if_then_else_cond): Look for situations where it is
6812 beneficial to undo the work of one of the recursive calls.
6814 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6816 PR tree-optimization/70754
6817 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6818 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6819 combined stmt before it if not NULL.
6820 (combine_chains): Process refs reversely and compute dominance point
6823 2017-01-23 Martin Liska <mliska@suse.cz>
6825 PR tree-optimization/79196
6826 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
6827 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
6829 (strlen_optimize_stmt): Call the renamed function.
6831 2017-01-23 Michael Matz <matz@suse.de>
6833 PR tree-optimization/78384
6834 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
6836 2017-01-23 Richard Biener <rguenther@suse.de>
6838 PR tree-optimization/79186
6839 * tree-vrp.c (register_new_assert_for): Make sure we've seen
6840 both incoming edges before moving an assert.
6842 2017-01-23 Martin Jambor <mjambor@suse.cz>
6844 * ipa-prop.c (load_from_param_1): Removed.
6845 (load_from_unmodified_param): Bits from load_from_param_1 put back
6847 (load_from_param): Removed.
6848 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
6849 with stmt. Reverted back to use of load_from_unmodified_param.
6851 2017-01-23 Martin Jambor <mjambor@suse.cz>
6854 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
6855 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
6856 field a pointer to garbage collected vector, mark lattices and
6857 ipcp_orig_node with GTY((skip)).
6858 (ipa_get_param_count): Adjust to descriptors being a pointer.
6859 (ipa_get_param): Likewise.
6860 (ipa_get_type): Likewise.
6861 (ipa_get_param_move_cost): Likewise.
6862 (ipa_set_param_used): Likewise.
6863 (ipa_get_controlled_uses): Likewise.
6864 (ipa_set_controlled_uses): Likewise.
6865 (ipa_is_param_used): Likewise.
6866 (ipa_node_params_t): Move into garbage collector. New methods insert
6868 (ipa_node_params_sum): Annotate wth GTY(()).
6869 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
6871 (ipa_load_from_parm_agg): Adjust declaration.
6872 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
6873 * ipa-profile.c (ipa_profile): Likewise.
6874 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
6875 (ipa_populate_param_decls): Make descriptors parameter garbage
6877 (ipa_dump_param): Adjust to descriptors being a pointer.
6878 (ipa_alloc_node_params): Likewise.
6879 (ipa_initialize_node_params): Likewise.
6880 (load_from_param_1): Make descriptors parameter garbage collected.
6881 (load_from_unmodified_param): Likewise.
6882 (load_from_param): Likewise.
6883 (ipa_load_from_parm_agg): Likewise.
6884 (ipa_node_params::~ipa_node_params): Removed.
6885 (ipa_free_all_node_params): Remove call to delete operator.
6886 (ipa_node_params_t::insert): New.
6887 (ipa_node_params_t::remove): Likewise.
6888 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
6889 copy known_csts and known_contexts vectors.
6890 (ipa_read_node_info): Adjust to descriptors being a pointer.
6891 (ipcp_modif_dom_walker): Make m_descriptors field garbage
6893 (ipcp_transform_function): Make descriptors variable garbage
6896 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
6898 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
6899 * config/i386/avx512dqintrin.h: Ditto.
6900 * config/i386/avx512fintrin.h: Ditto.
6901 * gcc/config/i386/i386.c: Handle new builtins.
6902 * config/i386/i386-builtin.def: Add new builtins.
6903 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
6904 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
6906 2017-01-23 Jakub Jelinek <jakub@redhat.com>
6907 Martin Liska <mliska@suse.cz>
6909 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
6910 * asan.c (asan_expand_poison_ifn): Support stores and use
6911 appropriate ASAN report function.
6912 * internal-fn.c (expand_ASAN_POISON_USE): New function.
6913 * internal-fn.def (ASAN_POISON_USE): Declare.
6914 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
6915 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
6916 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
6917 ASAN_POISON calls w/o LHS.
6918 * tree-ssa.c (execute_update_addresses_taken): Create clobber
6919 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
6920 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
6921 * gimplify.c (asan_poison_variables): Add attribute
6922 use_after_scope_memory to variables that really needs to live
6924 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
6925 having the attribute.
6927 2017-01-23 Martin Liska <mliska@suse.cz>
6929 * asan.c (create_asan_shadow_var): New function.
6930 (asan_expand_poison_ifn): Likewise.
6931 * asan.h (asan_expand_poison_ifn): New declaration.
6932 * internal-fn.c (expand_ASAN_POISON): Likewise.
6933 * internal-fn.def (ASAN_POISON): New builtin.
6934 * sanopt.c (pass_sanopt::execute): Expand
6935 asan_expand_poison_ifn.
6936 * tree-inline.c (copy_decl_for_dup_finish): Make function
6938 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
6939 * tree-ssa.c (is_asan_mark_p): New function.
6940 (execute_update_addresses_taken): Rewrite local variables
6941 (identified just by use-after-scope as addressable) into SSA.
6943 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
6945 * doc/install.texi (Specific): opensource.apple.com uses https
6946 now. Remove trailing slash.
6948 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
6950 * README.Portability: Remove note on an Irix compatibility issue.
6952 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
6954 * gcov.c (INCLUDE_ALGORITHM): Define.
6955 (INCLUDE_VECTOR): Define.
6956 No longer include <vector> and <algorithm> directly.
6958 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
6960 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
6962 * doc/invoke.texi (Code Gen Options): Ditto.
6964 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
6967 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
6969 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
6971 rtl-optimization/79125
6972 * cprop.c (local_cprop_pass): Handle cases where we make an
6975 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
6979 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
6980 read from, for big endian.
6982 2017-01-20 Jiong Wang <jiong.wang@arm.com>
6984 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
6985 register pauth builtins for LP64 only.
6987 2017-01-20 Marek Polacek <polacek@redhat.com>
6990 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
6993 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
6995 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
6997 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
6998 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
6999 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
7001 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7004 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
7005 in big-endian mode when they are not a single duplicated value.
7007 2017-01-20 Richard Biener <rguenther@suse.de>
7009 * BASE-VER: Bump to 7.0.1.
7011 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
7013 * omp-low.c (omplow_simd_context): New struct. Use it...
7014 (lower_rec_simd_input_clauses): ...here and...
7015 (lower_rec_input_clauses): ...here to hold common data. Adjust all
7016 references to idx, lane, max_vf, is_simt.
7018 2017-01-20 Graham Markall <graham.markall@embecosm.com>
7020 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
7023 2017-01-20 Martin Jambor <mjambor@suse.cz>
7025 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
7026 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
7028 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
7029 (GTFILES): Rename hsa.c to hsa-common.c.
7030 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
7031 * hsa-dump.c: Likewise.
7032 * hsa-gen.c: Likewise.
7033 * hsa-regalloc.c: Likewise.
7034 * ipa-hsa.c: Likewise.
7035 * omp-expand.c: Likewise.
7036 * omp-low.c: Likewise.
7037 * toplev.c: Likewise.
7039 2017-01-20 Marek Polacek <polacek@redhat.com>
7042 * doc/invoke.texi: Document -Wduplicated-branches.
7043 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
7044 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
7045 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
7046 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
7047 return 0 only when not OEP_LEXICOGRAPHIC.
7048 (fold_build_cleanup_point_expr): Use the expression
7049 location when building CLEANUP_POINT_EXPR.
7050 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
7051 * tree.c (add_expr): Handle error_mark_node.
7053 2017-01-20 Martin Liska <mliska@suse.cz>
7056 * tree-profile.c (init_ic_make_global_vars): Do not call
7058 (gimple_init_gcov_profiler): Likewise.
7060 2017-01-20 Martin Liska <mliska@suse.cz>
7063 * cgraph.h (maybe_create_reference): Remove argument and
7065 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
7067 * ipa-cp.c (create_specialized_node): Likewise.
7068 * symtab.c (symtab_node::maybe_create_reference): Handle
7069 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
7071 2017-01-20 Martin Liska <mliska@suse.cz>
7073 * read-rtl-function.c (function_reader::create_function): Use
7074 build_decl instread of build_decl_stat.
7076 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
7078 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
7079 * config/i386/avx512dqintrin.h: Ditto.
7080 * config/i386/avx512fintrin.h: Ditto.
7081 * config/i386/i386-builtin-types.def: Add new types.
7082 * gcc/config/i386/i386.c: Handle new types.
7083 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
7084 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
7085 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
7086 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
7087 (__builtin_ia32_kshiftridi): New.
7088 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
7090 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
7094 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
7095 define to rs6000_init_stack_protect_guard.
7096 (rs6000_init_stack_protect_guard): New function.
7098 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
7099 Yunqiang Su <yunqiang.su@imgtec.com>
7101 * config.gcc (supported_defaults): Add madd4.
7102 (with_madd4): Add validation.
7103 (all_defaults): Add madd4.
7104 * config/mips/mips.opt (mmadd4): New option.
7105 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
7107 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
7109 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
7110 (ISA_HAS_FUSED_MADD4): Likewise.
7111 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
7112 * gcc/doc/install.texi (--with-madd4): Document the new option.
7114 2017-01-19 Jiong Wang <jiong.wang@arm.com>
7116 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
7117 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
7118 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
7119 (aarch64_init_pauth_hint_builtins): New.
7120 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
7121 (aarch64_expand_builtin): Expand new builtins.
7123 2017-01-19 Jiong Wang <jiong.wang@arm.com>
7125 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
7126 * combine-stack-adj.c (no_unhandled_cfa): Handle
7127 REG_CFA_TOGGLE_RA_MANGLE.
7128 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
7129 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
7130 info for return address signing.
7131 (aarch64_expand_epilogue): Likewise.
7133 2017-01-19 Jiong Wang <jiong.wang@arm.com>
7135 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
7136 * config/aarch64/aarch64-protos.h
7137 (aarch64_return_address_signing_enabled): New declaration.
7138 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
7140 (aarch64_expand_prologue): Sign return address before it's pushed onto
7142 (aarch64_expand_epilogue): Authenticate return address fetched from
7144 (aarch64_override_options): Sanity check for ILP32 and ISA level.
7145 (aarch64_attributes): New function attributes for "sign-return-address".
7146 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
7147 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
7148 ("*do_return"): Generate combined instructions according to key index.
7149 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
7150 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
7152 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
7153 * config/aarch64/aarch64.opt (msign-return-address=): New.
7154 * doc/extend.texi (AArch64 Function Attributes): Documents
7155 "sign-return-address=".
7156 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
7158 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
7160 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
7161 overall option summary.
7163 2017-01-19 Jiong Wang <jiong.wang@arm.com>
7165 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
7166 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
7167 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
7168 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
7170 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
7172 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
7173 -mpower9-minmax by default for -mcpu=power9.
7174 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
7175 128-bit floating point.
7177 2017-01-20 Alan Modra <amodra@gmail.com>
7179 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
7180 optimizing for size.
7182 2017-01-20 Alan Modra <amodra@gmail.com>
7185 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
7186 for strcmp and strncmp from corresponding builtin decl.
7188 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
7190 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
7191 instead of i386/rtems-64.h.
7192 * config/i386/rtems-64.h: Remove.
7194 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
7198 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
7200 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
7202 2017-01-19 Tamar Christina <tamar.christina@arm.com>
7204 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
7205 Change int to HOST_WIDE_INT.
7206 * config/aarch64/aarch64-protos.h
7207 (aarch64_simd_gen_const_vector_dup): Likewise.
7208 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
7210 2017-01-19 David Malcolm <dmalcolm@redhat.com>
7212 * langhooks-def.h (lhd_type_for_size): New decl.
7213 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
7214 * langhooks.c (lhd_type_for_size): New function, taken from
7217 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
7219 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
7220 define_bypass for CR latency.
7221 (power9-cracked-alu): Update bypass latency and remove power9-branch.
7222 (power9-alu2): Add define_bypass for CR latency.
7224 (power9-mul): Update insn latency.
7225 (power9-mul-compare): Update insn latency, bypass latency and remove
7228 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7230 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
7232 * config/aarch64/aarch64.md
7233 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
7234 aarch64_nopcrelative_literal_loads.
7235 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
7237 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
7239 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
7241 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
7243 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
7246 * config.gcc (supported_defaults): Add lxc1-sxc1.
7247 (with_lxc1_sxc1): Add validation.
7248 (all_defaults): Add lxc1-sxc1.
7249 * config/mips/mips.opt (mlxc1-sxc1): New option.
7250 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
7252 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
7253 __mips_no_lxc1_sxc1.
7254 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
7255 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
7256 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
7258 2017-01-19 Richard Biener <rguenther@suse.de>
7260 PR tree-optimization/72488
7261 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
7262 sure to restore SSA info.
7263 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
7265 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
7267 PR rtl-optimization/79121
7268 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
7269 of the inner type when shifting an extended value.
7271 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
7274 * symtab.c (symtab_node::equal_address_to): Fix comparing of
7275 interposable aliases.
7277 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
7280 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
7281 Use the evmergelohi instruction.
7282 (mov_si<mode>_e500_subreg4_2_le): Likewise.
7283 (mov_sitf_e500_subreg8_2_be): Likewise.
7284 (mov_sitf_e500_subreg12_2_le): Likewise.
7285 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
7286 (mov_si<mode>_e500_subreg4_2_be): Likewise.
7287 (mov_sitf_e500_subreg8_2_le): Likewise.
7288 (mov_sitf_e500_subreg12_2_be): Likewise.
7290 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7292 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
7293 attribute from vecsimple to vecperm.
7294 (altivec_vbpermq2): Likewise.
7296 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7299 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
7301 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7302 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
7303 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
7304 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
7305 case where N arg is SIZE_MAX.
7306 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
7307 (cmpstrsi): Add pattern.
7309 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7311 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7312 __builtin_vec_revb builtins.
7313 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
7314 built-in functions to support generation of the ISA 3.0 XXBR<x>
7315 vector byte reverse instructions.
7316 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
7317 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
7318 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
7319 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
7320 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
7321 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
7322 (P9V_BUILTIN_VEC_REVB): Likewise.
7323 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
7324 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
7325 (p9_xxbrq_v16qi): Likewise.
7326 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
7327 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
7328 (p9_xxbrh_v8hi): Likewise.
7329 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
7330 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
7331 vec_revb built-in functions.
7333 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
7335 PR rtl-optimization/78952
7336 * config/i386/i386.md (any_extract): New code iterator.
7337 (*insvqi_2): Use any_extract for source operand.
7338 (*insvqi_3): Use any_shiftrt for source operand.
7340 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
7342 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
7344 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
7346 2017-01-18 Matthias Klose <doko@ubuntu.com>
7348 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
7350 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7352 * config/rs6000/altivec.h (vec_bperm): Change #define.
7353 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
7354 (altivec_vbpermq2): New define_insn.
7355 (altivec_vbpermd): Likewise.
7356 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
7358 (VBPERMD): Likewise.
7359 (VBPERM): New polymorphic function interface.
7360 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
7361 Add entries for P9V_BUILTIN_VEC_VBPERM.
7362 * doc/extend.texi: Add interfaces for vec_bperm.
7364 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7366 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
7367 first letter of error messages.
7368 (s390_resolve_overloaded_builtin): Likewise.
7369 * config/s390/s390.c (s390_expand_builtin): Likewise.
7370 (s390_invalid_arg_for_unprototyped_fn): Likewise.
7371 (s390_valid_target_attribute_inner_p): Likewise.
7372 * config/s390/s390.md ("tabort"): Likewise.
7374 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
7376 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
7377 (ISA_AVOID_DIV_HILO): New macro.
7378 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
7379 (ISA_HAS_DDIV): Likewise.
7381 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
7383 * doc/invoke.texi (fabi-version): Correct number of occurrences.
7385 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
7387 * doc/invoke.texi (fabi-version): Spelling fix.
7389 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
7392 * doc/invoke.texi (fabi-version): Mention mangling fix for
7395 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
7398 * doc/invoke.texi (fabi-version): Document discriminator mangling.
7400 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
7403 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
7404 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
7406 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
7408 (stack_protect_test): Ditto.
7409 * config/rs6000/rs6000.opt (mstack-protector-guard=,
7410 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
7412 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
7413 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
7414 -mstack-protector-guard-offset=.
7415 (RS/6000 and PowerPC Options): Ditto.
7417 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
7419 * config/i386/i386.h (MASK_CLASS_P): New define.
7420 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
7421 there are no registers from different register sets also when
7422 mask registers are used. Update function comment.
7423 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
7424 to (*k/*r) and (*k/*km) alternatives.
7426 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
7428 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
7429 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
7430 (EH_RETURN_HANDLER_RTX): New define.
7431 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
7432 Force frame pointer in EH return functions.
7433 (aarch64_expand_epilogue): Add barrier for eh_return.
7434 (aarch64_final_eh_return_addr): Remove.
7435 (aarch64_eh_return_handler_rtx): New function.
7436 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
7438 (aarch64_eh_return_handler_rtx): New prototype.
7440 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7442 * config/rs6000/altivec.h (vec_rlmi): New #define.
7443 (vec_vrlnm): Likewise.
7444 (vec_rlnm): Likewise.
7445 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
7446 (UNSPEC_VRLNM): Likewise.
7447 (VIlong): New mode iterator.
7448 (altivec_vrl<VI_char>mi): New define_insn.
7449 (altivec_vrl<VI_char>nm): Likewise.
7450 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
7453 (RLNM): New polymorphic function entry.
7454 (VRLWMI): New monomorphic function entry.
7456 (RLMI): New polymorphic function entry.
7457 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
7458 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
7459 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
7462 2017-01-17 Jakub Jelinek <jakub@redhat.com>
7465 * dwarf2out.c (field_byte_offset): Restore the
7466 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
7467 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
7468 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
7471 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
7474 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
7476 2017-01-17 Jakub Jelinek <jakub@redhat.com>
7479 * dwarf2out.c (add_data_member_location_attribute): For constant
7480 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
7481 instead of DW_AT_data_member_location, DW_AT_bit_offset and
7482 DW_AT_byte_size attributes.
7484 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
7486 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
7487 after forcing to constant memory when the code model is medium.
7489 2017-01-17 Julia Koval <julia.koval@intel.com>
7492 * config/i386/avx512fintrin.h
7493 (_mm512_i32gather_ps): Change __addr type to void const*.
7494 (_mm512_mask_i32gather_ps): Ditto.
7495 (_mm512_i32gather_pd): Ditto.
7496 (_mm512_mask_i32gather_pd): Ditto.
7497 (_mm512_i64gather_ps): Ditto.
7498 (_mm512_mask_i64gather_ps): Ditto.
7499 (_mm512_i64gather_pd): Ditto.
7500 (_mm512_mask_i64gather_pd): Ditto.
7501 (_mm512_i32gather_epi32): Ditto.
7502 (_mm512_mask_i32gather_epi32): Ditto.
7503 (_mm512_i32gather_epi64): Ditto.
7504 (_mm512_mask_i32gather_epi64): Ditto.
7505 (_mm512_i64gather_epi32): Ditto.
7506 (_mm512_mask_i64gather_epi32): Ditto.
7507 (_mm512_i64gather_epi64): Ditto.
7508 (_mm512_mask_i64gather_epi64): Ditto.
7509 (_mm512_i32scatter_ps): Change __addr type to void*.
7510 (_mm512_mask_i32scatter_ps): Ditto.
7511 (_mm512_i32scatter_pd): Ditto.
7512 (_mm512_mask_i32scatter_pd): Ditto.
7513 (_mm512_i64scatter_ps): Ditto.
7514 (_mm512_mask_i64scatter_ps): Ditto.
7515 (_mm512_i64scatter_pd): Ditto.
7516 (_mm512_mask_i64scatter_pd): Ditto.
7517 (_mm512_i32scatter_epi32): Ditto.
7518 (_mm512_mask_i32scatter_epi32): Ditto.
7519 (_mm512_i32scatter_epi64): Ditto.
7520 (_mm512_mask_i32scatter_epi64): Ditto.
7521 (_mm512_i64scatter_epi32): Ditto.
7522 (_mm512_mask_i64scatter_epi32): Ditto.
7523 (_mm512_i64scatter_epi64): Ditto.
7524 (_mm512_mask_i64scatter_epi64): Ditto.
7525 * config/i386/avx512pfintrin.h
7526 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
7527 (_mm512_mask_prefetch_i32gather_ps): Ditto.
7528 (_mm512_mask_prefetch_i64gather_pd): Ditto.
7529 (_mm512_mask_prefetch_i64gather_ps): Ditto.
7530 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
7531 (_mm512_prefetch_i32scatter_ps): Ditto.
7532 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
7533 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
7534 (_mm512_prefetch_i64scatter_pd): Ditto.
7535 (_mm512_prefetch_i64scatter_ps): Ditto.
7536 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
7537 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
7538 * config/i386/avx512vlintrin.h
7539 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
7540 (_mm_mmask_i32gather_ps): Ditto.
7541 (_mm256_mmask_i32gather_pd): Ditto.
7542 (_mm_mmask_i32gather_pd): Ditto.
7543 (_mm256_mmask_i64gather_ps): Ditto.
7544 (_mm_mmask_i64gather_ps): Ditto.
7545 (_mm256_mmask_i64gather_pd): Ditto.
7546 (_mm_mmask_i64gather_pd): Ditto.
7547 (_mm256_mmask_i32gather_epi32): Ditto.
7548 (_mm_mmask_i32gather_epi32): Ditto.
7549 (_mm256_mmask_i32gather_epi64): Ditto.
7550 (_mm_mmask_i32gather_epi64): Ditto.
7551 (_mm256_mmask_i64gather_epi32): Ditto.
7552 (_mm_mmask_i64gather_epi32): Ditto.
7553 (_mm256_mmask_i64gather_epi64): Ditto.
7554 (_mm_mmask_i64gather_epi64): Ditto.
7555 (_mm256_i32scatter_ps): Change __addr type to void*.
7556 (_mm256_mask_i32scatter_ps): Ditto.
7557 (_mm_i32scatter_ps): Ditto.
7558 (_mm_mask_i32scatter_ps): Ditto.
7559 (_mm256_i32scatter_pd): Ditto.
7560 (_mm256_mask_i32scatter_pd): Ditto.
7561 (_mm_i32scatter_pd): Ditto.
7562 (_mm_mask_i32scatter_pd): Ditto.
7563 (_mm256_i64scatter_ps): Ditto.
7564 (_mm256_mask_i64scatter_ps): Ditto.
7565 (_mm_i64scatter_ps): Ditto.
7566 (_mm_mask_i64scatter_ps): Ditto.
7567 (_mm256_i64scatter_pd): Ditto.
7568 (_mm256_mask_i64scatter_pd): Ditto.
7569 (_mm_i64scatter_pd): Ditto.
7570 (_mm_mask_i64scatter_pd): Ditto.
7571 (_mm256_i32scatter_epi32): Ditto.
7572 (_mm256_mask_i32scatter_epi32): Ditto.
7573 (_mm_i32scatter_epi32): Ditto.
7574 (_mm_mask_i32scatter_epi32): Ditto.
7575 (_mm256_i32scatter_epi64): Ditto.
7576 (_mm256_mask_i32scatter_epi64): Ditto.
7577 (_mm_i32scatter_epi64): Ditto.
7578 (_mm_mask_i32scatter_epi64): Ditto.
7579 (_mm256_i64scatter_epi32): Ditto.
7580 (_mm256_mask_i64scatter_epi32): Ditto.
7581 (_mm_i64scatter_epi32): Ditto.
7582 (_mm_mask_i64scatter_epi32): Ditto.
7583 (_mm256_i64scatter_epi64): Ditto.
7584 (_mm256_mask_i64scatter_epi64): Ditto.
7585 (_mm_i64scatter_epi64): Ditto.
7586 (_mm_mask_i64scatter_epi64): Ditto.
7587 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
7588 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
7589 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
7590 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
7591 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
7592 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
7593 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
7594 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
7595 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
7596 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
7597 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
7598 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
7599 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
7600 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
7601 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
7602 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
7603 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
7604 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
7605 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
7606 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
7607 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
7608 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
7609 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
7610 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
7611 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
7612 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
7613 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
7614 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
7615 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
7616 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
7617 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
7618 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
7619 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
7620 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
7621 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
7622 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
7623 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
7624 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
7625 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
7626 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
7627 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
7628 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
7629 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
7630 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
7631 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
7632 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
7633 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
7634 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
7635 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
7636 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
7637 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
7638 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
7639 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
7640 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
7641 definitions accordingly.
7643 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
7644 Kuan-Lin Chen <kuanlinchentw@gmail.com>
7647 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
7650 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
7653 * ira-conflicts.c (ira_build_conflicts): Update total conflict
7654 hard regs for inner regno.
7656 2017-01-17 Martin Liska <mliska@suse.cz>
7659 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
7660 assumption and add comment.
7662 2017-01-17 Nathan Sidwell <nathan@acm.org>
7664 * ipa-visibility.c (localize_node): New function, broken out of ...
7665 (function_and_variable_visibility): ... here. Call it.
7667 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
7670 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
7671 correctly set frequency of oudgoing edge.
7672 (duplicate_thread_path): Fix profile updating.
7674 2017-01-17 Jakub Jelinek <jakub@redhat.com>
7677 * configure.ac: Add GCC_BASE_VER.
7678 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
7679 version from BASE-VER file.
7680 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
7681 (gcc.o): Depend on $(BASEVER).
7682 * common.opt (dumpfullversion): New option.
7683 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
7684 * doc/invoke.texi: Document -dumpfullversion.
7685 * doc/install.texi: Document --with-gcc-major-version-only.
7686 * configure: Regenerated.
7688 2017-01-17 Richard Biener <rguenther@suse.de>
7690 PR tree-optimization/71433
7691 * tree-vrp.c (register_new_assert_for): Merge same asserts
7692 on all incoming edges.
7693 (process_assert_insertions_for): Handle insertions at the
7696 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
7698 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
7699 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
7701 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
7704 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
7707 2017-01-17 Alan Modra <amodra@gmail.com>
7710 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
7711 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
7712 symbolic stack limit when pic.
7714 2017-01-16 Martin Sebor <msebor@redhat.com>
7716 PR tree-optimization/78608
7717 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
7719 2017-01-16 Jeff Law <law@redhat.com>
7722 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
7723 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
7724 for several include directories that may be relative to sysroot.
7725 * config/i386/x-mingw32 (gplus_includedir): Define.
7726 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
7727 (native_system_includedir): Likewise.
7728 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
7729 override if TARGET_SYSTEM_ROOT is defined.
7730 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
7732 PR tree-optimization/79090
7733 PR tree-optimization/33562
7734 PR tree-optimization/61912
7735 PR tree-optimization/77485
7736 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
7737 and computed trims into the dump file.
7739 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
7741 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
7743 2017-01-16 Jakub Jelinek <jakub@redhat.com>
7746 * gimplify.c (gimplify_init_constructor): If want_value and
7747 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
7751 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
7752 sequence. Formatting fixes.
7753 (doloop_optimize): Formatting fixes.
7756 * gcc.c (debug_level_greater_than_spec_func): New function.
7757 (static_spec_functions): Add debug-level-gt spec function.
7758 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
7760 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
7761 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
7762 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
7763 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
7764 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
7765 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
7767 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
7769 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
7770 QImode fixups to general and mask registers only.
7772 2017-01-16 Carl Love <cel@us.ibm.com>
7774 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
7775 for built-in functions
7776 vector signed char vec_nabs (vector signed char)
7777 vector signed short vec_nabs (vector signed short)
7778 vector signed int vec_nabs (vector signed int)
7779 vector signed long long vec_nabs (vector signed long long)
7780 vector float vec_nabs (vector float)
7781 vector double vec_nabs (vector double)
7782 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
7784 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
7785 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
7786 * doc/extend.texi: Update the documentation file for the new built-in
7789 2017-01-16 Martin Sebor <msebor@redhat.com>
7791 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
7794 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7796 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
7797 UNSPEC_VSX__XXSPLTD to require special splat handling.
7799 2017-01-16 David Malcolm <dmalcolm@redhat.com>
7802 * system.h: Poison strndup.
7804 2017-01-16 Alan Modra <amodra@gmail.com>
7807 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
7810 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
7812 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
7814 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
7816 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
7817 call recog here. Assert that INSN_CODE (insn) is non-negative.
7819 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
7822 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
7824 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
7825 in the currently scheduled RTL fragment.
7827 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
7829 PR rtl-optimization/78751
7830 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
7833 2017-01-14 Jeff Law <law@redhat.com>
7835 PR tree-optimization/79090
7836 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
7837 variable length stores.
7838 (compute_trims): Delete dead assignment to *trim_tail.
7839 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
7842 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
7844 PR rtl-optimization/78626
7845 PR rtl-optimization/78727
7846 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
7847 of a block, and split such blocks after everything else is finished.
7849 2017-01-14 Alan Modra <amodra@gmail.com>
7852 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
7853 target legitimate_combined_insn.
7854 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
7855 (rs6000_legitimate_combined_insn): New function.
7856 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
7858 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
7859 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
7860 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
7862 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
7864 * doc/frontends.texi (G++ and GCC): Remove references to Java.
7866 2017-01-13 Jeff Law <law@redhat.com>
7868 PR tree-optimization/33562
7869 PR tree-optimization/61912
7870 PR tree-optimization/77485
7871 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
7873 (delete_dead_assignment): Likewise.
7874 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
7875 statement to delete_dead_call and delete_dead_assignment.
7877 2017-01-13 David Malcolm <dmalcolm@redhat.com>
7880 * substring-locations.c (format_warning_va): Strengthen case 1 so
7881 that both endpoints of the substring must be within the format
7882 range for just the substring to be printed.
7884 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
7886 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
7887 * config/i386/i386.c (ix86_target_string): Add missing options
7888 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
7889 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
7890 flags_other and ix86_target_other to flags2_other. Display unknown
7892 (ix86_valid_target_attribute_inner_p): Add missing options and
7893 reorder options by implied ISAs, as in ix86_target_string.
7895 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
7897 * hash-table.h (hash_table::too_empty_p): New function.
7898 (hash_table::expand): Use it.
7899 (hash_table::traverse): Likewise.
7900 (hash_table::empty_slot): Use sizeof (value_type) instead of
7901 sizeof (PTR) to convert bytes to elements. Shrink the table
7902 if the current size is excessive for the current number of
7905 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
7907 * ira-costs.c (record_reg_classes): Break from the inner loop
7908 early once alt_fail is known to be true. Update outer loop
7909 handling accordingly.
7911 2017-01-13 Jeff Law <law@redhat.com>
7913 * tree-ssa-dse.c (decrement_count): New function.
7914 (increment_start_addr, maybe_trim_memstar_call): Likewise.
7915 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
7916 when we know the partially dead statement is a mem* function.
7918 PR tree-optimization/61912
7919 PR tree-optimization/77485
7920 * tree-ssa-dse.c: Include expr.h.
7921 (maybe_trim_constructor_store): New function.
7922 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
7924 PR tree-optimization/33562
7925 PR tree-optimization/61912
7926 PR tree-optimization/77485
7927 * doc/invoke.texi: Document new dse-max-object-size param.
7928 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
7929 * tree-ssa-dse.c: Include params.h.
7930 (dse_store_status): New enum.
7931 (initialize_ao_ref_for_dse): New, partially extracted from
7933 (valid_ao_ref_for_dse, normalize_ref): New.
7934 (setup_live_bytes_from_ref, compute_trims): Likewise.
7935 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
7936 (maybe_trim_partially_dead_store): Likewise.
7937 (maybe_trim_complex_store): Likewise.
7938 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
7939 Track what bytes live from the original store. Return tri-state
7940 for dead, partially dead or live.
7941 (dse_dom_walker): Add constructor, destructor and new private members.
7942 (delete_dead_call, delete_dead_assignment): New extracted from
7944 (dse_optimize_stmt): Make a member of dse_dom_walker.
7945 Use initialize_ao_ref_for_dse.
7947 PR tree-optimization/33562
7948 PR tree-optimization/61912
7949 PR tree-optimization/77485
7950 * sbitmap.h (bitmap_count_bits): Prototype.
7951 (bitmap_clear_range, bitmap_set_range): Likewise.
7952 * sbitmap.c (bitmap_clear_range): New function.
7953 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
7955 2017-01-13 Martin Liska <mliska@suse.cz>
7958 * function.c (set_cfun): Add new argument force.
7959 * function.h (set_cfun): Likewise.
7960 * ipa-inline-transform.c (inline_call): Use the function when
7961 strict alising from is dropped for function we inline to.
7963 2017-01-13 Richard Biener <rguenther@suse.de>
7965 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
7966 for dumping GIMPLE INTEGER_CSTs.
7968 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7970 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
7971 to 201112L since C++17.
7973 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7976 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
7977 if -fsanitize=kernel-address is present.
7979 2017-01-13 Richard Biener <rguenther@suse.de>
7981 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
7982 as _Literal ( type ) number in case usual suffixes do not
7983 preserve all information.
7985 2017-01-13 Richard Biener <rguenther@suse.de>
7987 PR tree-optimization/77283
7988 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
7989 and ssa-iterators.h.
7990 (is_feasible_trace): Implement a cost model based on joiner
7993 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7996 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
7997 char or short to __float128/_Float128 directly.
7999 2017-01-12 Martin Sebor <msebor@redhat.com>
8001 to -Wformat-overflow.
8002 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
8003 (min_bytes_remaining): Same.
8004 (get_string_length): Same.
8005 (format_string): Same.
8006 (format_directive): Same.
8008 (pass_sprintf_length::handle_gimple_call): Same.
8010 2017-01-12 Jakub Jelinek <jakub@redhat.com>
8012 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
8013 info.nowrite calls with no lhs that can't throw. Return bool
8014 whether gsi_remove has been called or not.
8015 (pass_sprintf_length::handle_gimple_call): Return bool whether
8016 try_substitute_return_value called gsi_remove. Formatting fix.
8017 (pass_sprintf_length::execute): Don't use gsi_remove if
8018 handle_gimple_call returned true.
8021 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
8022 be removed due to side-effects, don't remove following barrier nor
8023 turn the successor edge into fallthru edge.
8025 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8028 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
8029 element-reversing loads and stores as not swappable.
8031 2017-01-12 Nathan Sidwell <nathan@acm.org>
8032 Nicolai Stange <nicstange@gmail.com>
8034 * combine.c (try_combine): Don't ignore result of overlap checking
8035 loop. Combine overlap & asm check into single loop.
8037 2017-01-12 Richard Biener <rguenther@suse.de>
8039 * tree-pretty-print.c (dump_generic_node): Provide -gimple
8040 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
8042 2017-01-12 Richard Biener <rguenther@suse.de>
8044 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
8045 and TS_TARGET_OPTION directly derive from TS_BASE.
8046 * tree-core.h (tree_optimization_option): Derive from tree_base.
8047 (tree_target_option): Likewise.
8049 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
8051 * config/i386/i386.c (memory_address_length): Increase len
8052 only when rip_relative_addr_p returns false.
8054 2017-01-11 Julia Koval <julia.koval@intel.com>
8056 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
8057 (OPTION_MASK_ISA_SGX_SET): New.
8058 (ix86_handle_option): Handle OPT_msgx.
8059 * config.gcc: Added sgxintrin.h.
8060 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
8061 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
8062 * config/i386/i386.c (ix86_target_string): Add -msgx.
8064 (ix86_option_override_internal): Handle new options.
8065 (ix86_valid_target_attribute_inner_p): Add sgx.
8066 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
8067 * config/i386/i386.opt: Add msgx.
8068 * config/i386/sgxintrin.h: New file.
8069 * config/i386/x86intrin.h: Add sgxintrin.h.
8071 2017-01-11 Jakub Jelinek <jakub@redhat.com>
8074 * fold-const.c (maybe_nonzero_address): Return 1 for function
8076 (tree_single_nonzero_warnv_p): Don't handle function local objects
8080 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
8083 2017-01-11 David Malcolm <dmalcolm@redhat.com>
8086 * opts.c: Include "spellcheck.h"
8087 (struct string_fragment): New struct.
8088 (struct edit_distance_traits<const string_fragment &>): New
8090 (get_closest_sanitizer_option): New function.
8091 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
8093 2017-01-11 Jakub Jelinek <jakub@redhat.com>
8095 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
8097 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
8098 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
8099 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
8100 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
8101 for initial die_offset if dwarf_split_debug_info.
8102 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
8103 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
8105 (output_skeleton_debug_sections): Formatting fix. Use
8106 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
8107 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
8109 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
8111 * config/arm/cortex-a53.md: Add bypasses for
8113 (cortex_a53_r2f): Only use for transfers.
8114 (cortex_a53_f2r): Likewise.
8115 (cortex_a53_r2f_cvt): Add reservation for conversions.
8116 (cortex_a53_f2r_cvt): Likewise.
8118 2017-01-11 Tamar Christina <tamar.christina@arm.com>
8120 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
8121 to all inlined functions, change static to extern.
8123 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
8126 * config/arm/arm.c (legitimize_pic_address): Handle reference to
8128 (arm_assemble_integer): Likewise.
8130 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
8132 * config.gcc: Use new awk script to check CPU, FPU and architecture
8133 parameters for --with-... options.
8134 * config/arm/parsecpu.awk: New file
8135 * config/arm/arm-cpus.in: New file.
8136 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
8138 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
8140 * config/arm/t-arm: Update dependency rules.
8141 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
8142 of processing .def files.
8143 * config/arm/genopt.sh: Deleted.
8144 * config/arm/gentune.sh: Deleted.
8145 * config/arm/arm-cores.def: Deleted.
8146 * config/arm/arm-arches.def: Deleted.
8147 * config/arm/arm-fpus.def: Deleted.
8148 * config/arm/arm-tune.md: Regenerated.
8149 * config/arm/arm-tables.opt: Regenerated.
8150 * config/arm/arm-cpu.h: New generated file.
8151 * config/arm/arm-cpu-data.h: New generated file.
8152 * config/arm/arm-cpu-cdata.h: New generated file.
8154 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
8157 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
8159 (input_varpool_node): Unpack dynamically_initialized bit.
8161 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
8163 PR rtl-optimization/79032
8164 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
8165 the alignment of the adjusted memory reference against that of MODE,
8166 instead of the alignment of the original memory reference.
8168 2017-01-11 Martin Jambor <mjambor@suse.cz>
8170 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
8172 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
8173 decorated functions.
8175 2017-01-11 Richard Biener <rguenther@suse.de>
8177 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
8178 set range/nonnull info for PHI results. Do not set it on
8179 stmts marked for removal.
8181 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
8183 * expr.c (store_field): In the bitfield case, fetch the return value
8184 from the registers before applying a single big-endian adjustment.
8185 Always do a final load for a BLKmode value not larger than a word.
8187 2017-01-10 David Malcolm <dmalcolm@redhat.com>
8190 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
8191 that we correctly handle column numbers greater than
8192 LINE_MAP_MAX_COLUMN_NUMBER.
8194 2017-01-10 Martin Sebor <msebor@redhat.com>
8197 * gimple-ssa-sprintf.c (get_destination_size): Call
8198 {init,fini}object_sizes.
8199 * tree-object-size.c (addr_object_size): Adjust.
8200 (pass_through_call): Adjust.
8201 (pass_object_sizes::execute): Adjust.
8202 * tree-object-size.h (fini_object_sizes): Declare.
8204 2017-01-10 Martin Sebor <msebor@redhat.com>
8206 PR tree-optimization/78775
8207 * builtins.c (get_size_range): Move...
8208 * calls.c: ...to here.
8209 (alloc_max_size): Accept zero argument.
8210 (operand_signed_p): Remove.
8211 (maybe_warn_alloc_args_overflow): Call get_size_range.
8212 * calls.h (get_size_range): Declare.
8214 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
8216 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
8217 from TI's devices.csv file as of September 2016.
8218 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
8220 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
8222 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
8223 * doc/invoke.texi: Likewise.
8224 * doc/md.texi: Likewise.
8225 * doc/objc.texi: Likewise.
8227 2017-01-10 Joshua Conner <joshconner@google.com>
8229 * config/arm/fuchsia-elf.h: New file.
8230 * config/fuchsia.h: New file.
8231 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
8232 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
8234 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
8236 2016-01-10 Richard Biener <rguenther@suse.de>
8238 PR tree-optimization/79034
8239 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
8240 Propagate out degenerate PHIs in the joiner.
8242 2017-01-10 Martin Liska <mliska@suse.cz>
8244 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
8245 (sort_congruence_classes_by_decl_uid): Likewise.
8246 (sort_congruence_class_groups_by_decl_uid): Likewise.
8247 (sem_item_optimizer::merge_classes): Sort class, groups in these
8248 classes and members in the groups by DECL_UID of declarations.
8249 This would make merge operations stable.
8251 2017-01-10 Martin Liska <mliska@suse.cz>
8253 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
8254 usage of m_classes_vec.
8255 (sem_item_optimizer::~sem_item_optimizer): Likewise.
8256 (sem_item_optimizer::get_group_by_hash): Likewise.
8257 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
8258 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
8259 (sem_item_optimizer::verify_classes): Likewise.
8260 (sem_item_optimizer::process_cong_reduction): Likewise.
8261 (sem_item_optimizer::dump_cong_classes): Likewise.
8262 (sem_item_optimizer::merge_classes): Likewise.
8263 * ipa-icf.h (congruence_class_hash): Rename from
8264 congruence_class_group_hash. Remove declaration of m_classes_vec.
8266 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
8268 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
8269 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
8270 * config.gcc: Add avx512vpopcntdqintrin.h.
8271 * config/i386/avx512vpopcntdqintrin.h: New.
8272 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
8273 * config/i386/i386-builtin-types.def: Add new types.
8274 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
8275 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
8276 __builtin_ia32_vpopcountq_v8di_mask): New.
8277 * config/i386/i386-c.c (ix86_target_macros_internal): Define
8278 __AVX512VPOPCNTDQ__.
8279 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
8280 (PTA_AVX512VPOPCNTDQ): Define.
8281 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
8282 TARGET_AVX512VPOPCNTDQ_P): Define.
8283 * config/i386/i386.opt: Add mavx512vpopcntdq.
8284 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
8285 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
8287 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8290 * predict.def (PRED_CALL): Set to 67.
8292 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
8294 * expr.c (store_field): In the bitfield case, if the value comes from
8295 a function call and is of an aggregate type returned in registers, do
8296 not modify the field mode; extract the value in all cases if the mode
8297 is BLKmode and the size is not larger than a word.
8299 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
8302 * config/i386/cpuid.h: Fix undefined behavior.
8304 2017-01-04 Jeff Law <law@redhat.com>
8306 PR tree-optimization/79007
8307 PR tree-optimization/67955
8308 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
8309 conservative for pt.null when flag_non_call_exceptions is on.
8311 2017-01-09 Jakub Jelinek <jakub@redhat.com>
8313 PR translation/79019
8314 PR translation/79020
8315 * params.def (PARAM_INLINE_MIN_SPEEDUP,
8316 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
8317 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
8319 * config/avr/avr.opt (maccumulate-args): Likewise.
8320 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
8321 * common.opt (freport-bug): Likewise.
8322 * cif-code.def (CIF_FINAL_ERROR): Likewise.
8323 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
8324 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
8325 translatable string.
8326 * config/i386/i386.c (function_value_32): Likewise.
8327 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
8328 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
8330 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
8331 * common/config/msp430/msp430-common.c (msp430_handle_option):
8333 * symtab.c (symtab_node::verify_base): Likewise.
8334 * opts.c (set_debug_level): Likewise.
8335 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
8336 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
8337 missing whitespace to translatable strings.
8338 * config/avr/avr.md (bswapsi2): Fix typo in comment.
8339 * config/sh/superh.h: Likewise.
8340 * config/i386/xopintrin.h: Likewise.
8341 * config/i386/znver1.md: Likewise.
8342 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
8343 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
8344 * double-int.h (struct double_int): Likewise.
8345 * double-int.c (div_and_round_double): Likewise.
8346 * wide-int.cc: Likewise.
8347 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
8348 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
8349 * cfgcleanup.c (crossjumps_occured): Renamed to ...
8350 (crossjumps_occurred): ... this.
8351 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
8354 PR tree-optimization/78899
8355 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
8356 returning bool return struct loop *, NULL for failure and the new
8358 (versionable_outer_loop_p): Don't version outer loop if it has
8359 dont_vectorized bit set.
8360 (tree_if_conversion): When versioning outer loop, ensure
8361 tree_if_conversion is performed also on the inner loop of the
8362 non-vectorizable outer loop copy.
8363 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
8364 LOOP_VECTORIZED in inner loop of the scalar outer loop and
8365 prevent vectorization of it.
8366 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
8367 the outer loop vectorization of the non-scalar version is attempted
8368 before vectorization of the inner loop in scalar version. If
8369 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
8370 vectorization of its inner loop.
8371 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
8372 has 2 inner loops, rename also on edges from bb whose single pred
8373 is outer_loop->header. Fix typo in function comment.
8375 2017-01-09 Martin Sebor <msebor@redhat.com>
8378 * asan.c (asan_emit_stack_protection): Increase local buffer size
8379 to avoid snprintf truncation warning.
8381 2017-01-09 Andrew Pinski <apinski@cavium.com>
8383 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
8384 to reference thunderx2t99 for the tuning structure
8385 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
8387 (thunderx2t99_extra_costs): This.
8388 * config/aarch64/aarch64-tune.md: Regenerate.
8389 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
8390 (vulcan_addrcost_table): This.
8391 (vulcan_regmove_cost): Rename to ...
8392 (thunderx2t99_regmove_cost): This.
8393 (vulcan_vector_cost): Rename to ...
8394 (thunderx2t99_vector_cost): this.
8395 (vulcan_branch_cost): Rename to ...
8396 (thunderx2t99_branch_cost): This.
8397 (vulcan_tunings): Rename to ...
8398 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
8399 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
8401 2017-01-09 Martin Jambor <mjambor@suse.cz>
8405 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
8406 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
8407 (propagate_vr_accross_jump_function): Use the above function for all
8408 value range computations for pass-through jump functions and type
8409 converasion from explicit value range values.
8410 (ipcp_propagate_stage): Do not attempt to deduce types of formal
8411 parameters from TYPE_ARG_TYPES.
8412 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
8413 (ipa_write_node_info): Stream type of the actual argument.
8414 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
8416 2017-01-09 Martin Liska <mliska@suse.cz>
8419 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
8420 (lookup_compiler): Do not show error message with have_E.
8422 2017-01-09 Jakub Jelinek <jakub@redhat.com>
8424 PR tree-optimization/78938
8425 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
8426 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
8427 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
8428 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
8431 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8433 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
8436 2017-01-09 Richard Biener <rguenther@suse.de>
8438 PR tree-optimization/78997
8439 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
8440 name condition properly.
8442 2017-01-09 Richard Biener <rguenther@suse.de>
8445 * dwarf2out.c (is_cxx): New overload with context.
8446 (is_naming_typedef_decl): Use it.
8448 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
8450 * invoke.texi (Option Summary): Correct spacing in option lists
8451 and add line breaks to fix over-long lines.
8453 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
8457 * extend.texi (Common Variable Attributes): Add xref to GCC
8458 Internals manual to explain mode attribute keywords.
8460 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
8463 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
8464 and Preprocessor Options.
8465 (Options for Linking): Document -pthread here....
8466 (RS/6000 and PowerPC Options): ...not here.
8467 (Solaris 2 Options): ...or here.
8468 * doc/cppopts.texi: Document -pthread.
8470 2017-01-08 Martin Sebor <msebor@redhat.com>
8473 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
8474 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
8475 New member functions.
8476 (format_directive): Used them.
8478 (pass_sprintf_length::handle_gimple_call): Same.
8479 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
8480 to avoid truncation for any argument.
8481 (extract_affine_mul): Same.
8482 * tree.c (get_file_function_name): Same.
8484 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8487 * predict.def (PRED_INDIR_CALL): Set to 86.
8489 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
8491 PR preprocessor/54124
8492 * doc/cppopts.texi: Reformat -d subtable to list the full name
8493 of the options. Add cross-reference to the docs for the general
8494 compiler -d options.
8495 * doc/invoke.texi (Developer Options): Add cross-reference to the
8496 preprocessor-specific -d option documentation.
8498 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
8500 PR preprocessor/13498
8501 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
8502 redudant material, and reflect new command-line options.
8503 (System Headers): Likewise.
8505 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
8507 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
8508 -isystem, and -idirafter. Copy-edit.
8509 * doc/cppopts.texi: Copy-edit. Remove contradiction about
8510 default for -ftrack-macro-expansion. Delete obsolete and
8511 badly-formatted implementation details about -fdebug-cpp output.
8512 * doc/cppwarnopts.texi: Copy-edit.
8514 2017-01-07 David Malcolm <dmalcolm@redhat.com>
8517 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
8518 that the transition from a max line width >= 1<<10 to narrower
8519 lines works correctly.
8521 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
8523 * doc/options.texi (PerFunction): New.
8524 * opt-functions.awk (switch_flags): Map both Optimization and
8525 PerFunction to CL_OPTIMIZATION.
8526 * opth-gen.awk: Test for PerFunction flag along with
8528 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
8529 it only when the latter is present. Skip those that don't in
8530 the hash function generator.
8531 * common.opt (fvar-tracking): Mark as PerFunction instead of
8533 (fvar-tracking-assignments): Likewise.
8534 (fvar-tracking-assignments-toggle): Likewise.
8535 (fvar-tracking-uninit): Likewise.
8537 2017-01-07 Jakub Jelinek <jakub@redhat.com>
8539 PR translation/79018
8540 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
8543 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
8546 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
8547 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
8548 TARGET_LONG_JUMP_TABLE_OFFSETS.
8549 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
8550 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
8551 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
8552 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
8553 * config/m68k/m68k.md (tablejump expander): Likewise.
8554 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
8555 TARGET_LONG_JUMP_TABLE_OFFSETS.
8556 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
8557 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
8559 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
8560 David Holsgrove <david.holsgrove@xilinx.com>
8562 * common/config/microblaze/microblaze-common.c
8563 (TARGET_EXCEPT_UNWIND_INFO): Remove.
8564 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
8566 * config/microblaze/microblaze.c (microblaze_must_save_register)
8567 (microblaze_expand_epilogue, microblaze_return_addr): Handle
8569 (microblaze_eh_return): New function.
8570 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
8571 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
8572 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
8573 * config/microblaze/microblaze.md (eh_return): New pattern.
8575 2017-01-06 Jakub Jelinek <jakub@redhat.com>
8577 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
8578 GCC_DIAGNOSTIC_STRINGIFY): Define.
8580 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
8582 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8584 * config/arm/arm.md (<mcrr>): New.
8586 * config/arm/arm.c (arm_arch5te): New.
8587 (arm_option_override): Set arm_arch5te.
8588 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
8590 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
8591 (arm_mcrr_qualifiers): ... this. New.
8592 (MRRC_QUALIFIERS): Define to...
8593 (arm_mrrc_qualifiers): ... this. New.
8594 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
8596 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
8597 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
8598 (MRRCI, mrrc, MRRC): New.
8599 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
8600 VUNSPEC_MRRC2): New.
8602 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8604 * config/arm/arm.md (<mcr>): New.
8606 * config/arm/arm.c (arm_coproc_builtin_available): Add
8607 support for mcr, mrc, mcr2 and mrc2.
8608 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
8609 (arm_mcr_qualifiers): ... this. New.
8610 (MRC_QUALIFIERS): Define to ...
8611 (arm_mrc_qualifiers): ... this. New.
8612 (MCR_QUALIFIERS): Define to ...
8613 (arm_mcr_qualifiers): ... this. New.
8614 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
8616 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
8617 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
8618 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
8621 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8623 * config/arm/arm.md (*ldc): New.
8627 * config/arm/arm.c (arm_coproc_builtin_available): Add
8628 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
8629 (arm_coproc_ldc_stc_legitimate_address): New.
8630 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
8631 'qualifier_const_pointer'.
8632 (LDC_QUALIFIERS): Define to...
8633 (arm_ldc_qualifiers): ... this. New.
8634 (STC_QUALIFIERS): Define to...
8635 (arm_stc_qualifiers): ... this. New.
8636 * config/arm/arm-protos.h
8637 (arm_coproc_ldc_stc_legitimate_address): New.
8638 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
8639 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
8640 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
8641 stc2, stcl, stc2l): New.
8642 * config/arm/constraints.md (Uz): New.
8643 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
8644 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
8645 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
8646 VUNSPEC_STC2L): New.
8648 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8650 * config/arm/arm.md (<cdp>): New.
8651 * config/arm/arm.c (neon_const_bounds): Rename this ...
8652 (arm_const_bounds): ... this.
8653 (arm_coproc_builtin_available): New.
8654 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
8655 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
8656 (CDP_QUALIFIERS): Define to...
8657 (arm_cdp_qualifiers): ... this. New.
8659 (arm_expand_builtin_args): Add case for 6 arguments.
8660 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
8661 (arm_const_bounds): ... this.
8662 (arm_coproc_builtin_available): New.
8663 * config/arm/arm_acle.h (__arm_cdp): New.
8665 * config/arm/arm_acle_builtins.def (cdp): New.
8667 * config/arm/iterators.md (CDPI,CDP,cdp): New.
8668 * config/arm/neon.md: Rename all 'neon_const_bounds' to
8670 * config/arm/types.md (coproc): New.
8671 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
8672 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
8673 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
8674 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
8676 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8678 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
8679 (UBINOP_QUALIFIERS): New.
8681 (acle_builtin_data): New. Change comment.
8682 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
8683 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
8684 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
8685 arm_acle_builtins.def.
8686 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
8687 (arm_init_acle_builtins): New.
8688 (CRC32_BUILTIN): Remove.
8689 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
8690 crc32cb, crc32ch and crc32cw.
8691 (arm_init_crc32_builtins): Remove.
8692 (arm_init_builtins): Use arm_init_acle_builtins rather
8693 than arm_init_crc32_builtins.
8694 (arm_expand_acle_builtin): New.
8695 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
8696 * config/arm/arm_acle_builtins.def: New.
8698 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8700 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
8701 (arm_builtin_datum): ... this.
8702 (arm_init_neon_builtin): Rename to ...
8703 (arm_init_builtin): ... this. Add a new parameters PREFIX
8704 and USE_SIG_IN_NAME.
8705 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
8706 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
8707 'arm_builtin_datum'.
8708 (arm_init_vfp_builtins): Likewise.
8709 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
8710 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
8711 (arm_expand_neon_args): Rename to ...
8712 (arm_expand_builtin_args): ... this. Rename builtin_arg
8713 enum values and differentiate between ARG_BUILTIN_MEMORY
8714 and ARG_BUILTIN_NEON_MEMORY.
8715 (arm_expand_neon_builtin_1): Rename to ...
8716 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
8717 values, arm_expand_builtin_args and add bool parameter NEON.
8718 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
8719 (arm_expand_vfp_builtin): Likewise.
8720 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
8722 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8725 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
8726 * predict.c (tree_estimate_probability_bb): Reverse direction of
8727 polymorphic call predictor.
8729 2017-01-06 David Malcolm <dmalcolm@redhat.com>
8731 * passes.c (execute_one_pass): Split out pass-skipping logic into...
8732 (determine_pass_name_match): ...this new function and...
8733 (should_skip_pass_p): ...this new function.
8735 2017-01-06 Nathan Sidwell <nathan@acm.org>
8737 * ipa-visibility.c (function_and_variable_visibility): Reformat
8738 comments and long lines. Remove extrneous if.
8739 * symtab.c (symtab_node::make_decl_local): Fix code format.
8740 (symtab_node::set_section_for_node): Fix comment typo.
8742 2017-01-06 Martin Liska <mliska@suse.cz>
8745 * lra-constraints.c: Rename invariant to lra_invariant.
8746 * predict.c (set_even_probabilities): Initialize e to NULL.
8748 2017-01-05 Martin Sebor <msebor@redhat.com>
8750 PR tree-optimization/78910
8751 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
8752 (format_integer): Correct off-by-one error in the handling
8753 of precision with negative numbers in signed conversions..
8755 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
8757 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
8759 2017-01-05 Jakub Jelinek <jakub@redhat.com>
8761 PR tree-optimization/71016
8762 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
8763 factor_out_conditional_conversion. Formatting fix.
8764 (factor_out_conditional_conversion): Add cond_stmt argument.
8765 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
8766 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
8769 2017-01-05 David Malcolm <dmalcolm@redhat.com>
8771 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
8772 read-rtl-function.o, and selftest-rtl.o.
8773 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
8774 (selftest::aarch64_test_loading_full_dump): New function.
8775 (selftest::aarch64_run_selftests): New function.
8776 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
8777 selftest::aarch64_run_selftests.
8778 * config/i386/i386.c
8779 (selftest::ix86_test_loading_dump_fragment_1): New function.
8780 (selftest::ix86_test_loading_call_insn): New function.
8781 (selftest::ix86_test_loading_full_dump): New function.
8782 (selftest::ix86_test_loading_unspec): New function.
8783 (selftest::ix86_run_selftests): Call the new functions.
8784 * emit-rtl.c (maybe_set_max_label_num): New function.
8785 * emit-rtl.h (maybe_set_max_label_num): New decl.
8786 * function.c (instantiate_decls): Guard call to
8787 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
8788 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
8790 * gensupport.c (gen_reader::gen_reader): Pass "false"
8791 for new "compact" param of rtx_reader.
8792 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
8793 rather than an empty string for NULL strings.
8794 * read-md.c: Potentially include config.h rather than bconfig.h.
8795 Wrap include of errors.h with #ifdef GENERATOR_FILE.
8796 (have_error): New global, copied from errors.c.
8797 (md_reader::read_name): Rename to...
8798 (md_reader::read_name_1): ...this, adding "out_loc" param,
8799 and converting "missing name or number" to returning false, rather
8801 (md_reader::read_name): Reimplement in terms of read_name_1.
8802 (md_reader::read_name_or_nil): New function.
8803 (md_reader::read_string): Handle "(nil)" by returning NULL.
8804 (md_reader::md_reader): Add new param "compact".
8805 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
8806 (md_reader::read_file): New method.
8807 * read-md.h (md_reader::md_reader): Add new param "compact".
8808 (md_reader::read_file): New method.
8809 (md_reader::is_compact): New accessor.
8810 (md_reader::read_name): Convert return type from void to file_location.
8811 (md_reader::read_name_or_nil): New decl.
8812 (md_reader::read_name_1): New decl.
8813 (md_reader::m_compact): New field.
8814 (noop_reader::noop_reader): Pass "false" for new "compact" param
8816 (rtx_reader::rtx_reader): Add new "compact" param.
8817 (rtx_reader::read_rtx_operand): Make virtual and convert return
8818 type from void to rtx.
8819 (rtx_reader::read_until): New decl.
8820 (rtx_reader::handle_any_trailing_information): New virtual function.
8821 (rtx_reader::postprocess): New virtual function.
8822 (rtx_reader::finalize_string): New virtual function.
8823 (rtx_reader::m_in_call_function_usage): New field.
8824 (rtx_reader::m_reuse_rtx_by_id): New field.
8825 * read-rtl-function.c: New file.
8826 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
8827 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
8828 (selftest::verify_three_block_rtl_cfg): New decl.
8829 * read-rtl-function.h: New file.
8830 * read-rtl.c: Potentially include config.h rather than bconfig.h.
8831 For host, include function.h, memmodel.h, and emit-rtl.h.
8832 (one_time_initialization): New function.
8833 (struct compact_insn_name): New struct.
8834 (compact_insn_names): New array.
8835 (find_code): Handle insn codes in compact dumps.
8836 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
8837 (bind_subst_iter_and_attr): Likewise.
8838 (add_condition_to_string): Likewise.
8839 (add_condition_to_rtx): Likewise.
8840 (apply_attribute_uses): Likewise.
8841 (add_current_iterators): Likewise.
8842 (apply_iterators): Likewise.
8843 (initialize_iterators): Guard usage of apply_subst_iterator with
8844 #ifdef GENERATOR_FILE.
8845 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
8846 (md_reader::read_mapping): Likewise.
8847 (add_define_attr_for_define_subst): Likewise.
8848 (add_define_subst_attr): Likewise.
8849 (read_subst_mapping): Likewise.
8850 (check_code_iterator): Likewise.
8851 (rtx_reader::read_rtx): Likewise. Move one-time initialization
8853 (one_time_initialization): New function.
8854 (rtx_reader::read_until): New method.
8855 (read_flags): New function.
8856 (parse_reg_note_name): New function.
8857 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
8858 Handle reuse_rtx ids.
8859 Wrap iterator lookup within #ifdef GENERATOR_FILE.
8860 Add parsing support for RTL dumps, mirroring the special-cases in
8861 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
8862 values, and calling handle_any_trailing_information.
8863 (rtx_reader::read_rtx_operand): Convert return type from void
8864 to rtx, returning return_rtx. Handle case 'e'. Call
8865 finalize_string on XSTR and XTMPL fields.
8866 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
8867 "(nil)" values were omitted. Call the postprocess vfunc on the
8869 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
8870 class ctor. Initialize m_in_call_function_usage. Call
8871 one_time_initialization.
8872 * rtl-tests.c (selftest::test_uncond_jump): Call
8873 set_new_first_and_last_insn.
8874 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
8875 * selftest-rtl.c: New file.
8876 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
8877 (selftest::get_insn_by_uid): New decl.
8878 * selftest-run-tests.c (selftest::run_tests): Call
8879 read_rtl_function_c_tests.
8880 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
8881 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
8884 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
8886 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
8887 operands in a special way. Assert that pos+len <= mode precision.
8889 2017-01-05 Jakub Jelinek <jakub@redhat.com>
8891 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
8892 3 argument Alias with unlimited for the negative form.
8893 (fno-vect-cost-model): Removed.
8895 2017-01-05 Martin Liska <mliska@suse.cz>
8897 * hsa-gen.c (gen_hsa_divmod): New function.
8898 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
8900 2017-01-05 Martin Liska <mliska@suse.cz>
8903 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
8906 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8908 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
8909 small constant length operands.
8911 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8913 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
8914 between loop iterations.
8916 2017-01-05 Martin Liska <mliska@suse.cz>
8919 * gimplify.c (gimplify_decl_expr): Compare to
8920 asan_poisoned_variables instread of checking flags.
8921 (gimplify_target_expr): Likewise.
8922 (gimplify_expr): Likewise.
8923 (gimplify_function_tree): Conditionally initialize
8924 asan_poisoned_variables.
8926 2017-01-04 Jeff Law <law@redhat.com>
8928 PR tree-optimizatin/78812
8929 * rtl.h (contains_mem_rtx_p): Prototype.
8930 * ifcvt.c (containts_mem_rtx_p): Move from here to...
8931 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
8932 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
8933 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
8934 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
8936 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8938 * input.c (assert_char_at_range): Default-initialize actual_range.
8940 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8942 * df-scan.c (df_ref_create_structure): Make regno unsigned,
8943 to match the caller.
8945 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8947 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
8948 insns after final jump in test to emit dummy move.
8950 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8952 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
8953 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
8955 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8957 * multiple_target.c (create_dispatcher_calls): Init e_next.
8958 * tree-ssa-loop-split.c (split_loop): Init border.
8959 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
8962 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
8967 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
8968 (altivec_register_operand): Do not return true if the operand
8969 contains a SUBREG mixing SImode and SFmode.
8970 (vsx_register_operand): Likewise.
8971 (vsx_reg_sfsubreg_ok): New predicate.
8972 (vfloat_operand): Do not return true if the operand contains a
8973 SUBREG mixing SImode and SFmode.
8974 (vint_operand): Likewise.
8975 (vlogical_operand): Likewise.
8976 (gpc_reg_operand): Likewise.
8977 (int_reg_operand): Likewise.
8978 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
8979 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
8980 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
8982 (rs6000_emit_move_si_sf_subreg): New helper function.
8983 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
8984 fixup SUBREGs involving SImode and SFmode.
8985 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
8986 numbers for the new peephole2 optimization.
8987 (peephole2 for SFmode unions): New peephole2 to optimize cases in
8988 the GLIBC math library that do AND/IOR/XOR operations on single
8989 precision floating point.
8990 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
8991 target macros to say whether we need to avoid SUBREGs mixing
8993 (TARGET_ALLOW_SF_SUBREG): Likewise.
8994 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
8995 (UNSPEC_SI_FROM_SF): Likewise.
8996 (iorxor): Change spacing.
8997 (and_ior_xor): New iterator for AND, IOR, and XOR.
8998 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
8999 (movdi_from_sf_zero_ext): Likewise.
9000 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
9001 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
9002 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
9003 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
9004 (fms<mode>4): Likewise.
9005 (fnma<mode>4): Likewise.
9006 (fnms<mode>4): Likewise.
9007 (nfma<mode>4): Likewise.
9008 (nfms<mode>4): Likewise.
9010 2017-01-04 Marek Polacek <polacek@redhat.com>
9013 * doc/invoke.texi: Document -Wpointer-compare.
9015 2017-01-04 Jakub Jelinek <jakub@redhat.com>
9017 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
9020 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
9021 descriptions for -gdwarf-5 and emit them as uleb128 instead of
9024 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
9027 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
9028 documentation of the powerpc_popcntb_ok attribute.
9029 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
9030 code to issue warning messages if a requested CPU configuration is
9031 not supported by the binary (assembler and loader) toolchain.
9032 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
9033 made to define a built-in function that has been disabled.
9034 (paired_init_builtins): Add assertion to prevent ICE if attempt is
9035 made to define a built-in function that has been disabled.
9036 (altivec_init_builtins): Add comment explaining why definition
9037 of the DST built-in functions is not preceded by an assertion
9038 check. Add assertions to prevent ICE if attempts are made to
9039 define an altivec predicate or an abs* built-in function that has
9041 (htm_init_builtins): Add comment explaining why definition of the
9042 htm built-in functions is not preceded by an assertion check.
9044 2017-01-04 Jeff Law <law@redhat.com>
9046 PR tree-optimizatin/67955
9047 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
9048 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
9049 the points-to solution does not include pt_null. Use DECL_PT_UID
9052 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
9054 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
9055 Use gen_int_mode instead of gen_lopwart for const_int operands.
9057 2017-01-04 Jakub Jelinek <jakub@redhat.com>
9059 PR tree-optimization/71563
9060 * match.pd: Simplify X << Y into X if Y is known to be 0 or
9061 out of range value - has low bits known to be zero.
9063 2017-01-04 Alan Modra <amodra@gmail.com>
9065 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
9066 * configure: Regenerate.
9067 * config.in: Regenerate.
9069 2017-01-04 Jakub Jelinek <jakub@redhat.com>
9072 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
9073 a substring of the message, but strcmp with the whole message. Ifdef
9074 ENABLE_NLS, translate the message first using dgettext.
9076 2017-01-03 Jeff Law <law@redhat.com>
9078 PR tree-optimizatin/78856
9079 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
9080 (mark_threaded_blocks): Remove code to truncate thread paths that
9081 cross multiple loop headers. Instead invalidate the cached loop
9082 iteration information and handle case of a thread path walking
9083 into an irreducible region.
9085 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9088 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
9089 assertions. Add support for doing the signbit if the IEEE 128-bit
9090 floating point value is in a GPR.
9091 * config/rs6000/rs6000.md (Fsignbit): Delete.
9092 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
9093 Update the length attribute if the value is in a GPR.
9094 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
9095 the sign or zero extension instruction, since the value is always 0/1.
9096 (signbit<mode>2_dm2): Delete using <Fsignbit>.
9099 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
9100 extracting SImode to a GPR register so that we can generate a
9101 store, limit the vector to be in a traditional Altivec register
9102 for the vextuwrx instruction.
9104 2017-01-03 Ian Lance Taylor <iant@google.com>
9106 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
9108 2017-01-03 Martin Sebor <msebor@redhat.com>
9110 PR tree-optimization/78696
9111 * gimple-ssa-sprintf.c (format_floating): Correct handling of
9112 precision. Use MPFR for %f for greater fidelity. Correct handling
9114 (pass_sprintf_length::compute_format_length): Set width and precision
9115 specified by asrerisk to void_node for vararg functions.
9116 (try_substitute_return_value): Adjust dump output.
9118 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
9120 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
9122 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
9124 * doc/invoke.texi (SPARC options): Document -mlra as the default.
9125 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
9126 -mlra/-mno-lra was passed to the compiler.
9128 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
9130 PR rtl-optimization/65618
9131 * emit-rtl.c (try_split): Move initialization of "before" and
9132 "after" to just before the call to emit_insn_after_setloc.
9134 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
9136 * doc/md.texi (Standard Names): Remove reference to Java frontend.
9138 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
9140 * dwarf2out.c (gen_enumeration_type_die): When
9141 -gno-strict-dwarf, add a DW_AT_encoding attribute.
9143 2017-01-03 Jakub Jelinek <jakub@redhat.com>
9145 PR tree-optimization/78965
9146 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
9147 Change first argument from const call_info & to call_info &. For %n
9148 set info.nowrite to false.
9151 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
9152 possibly throwing calls.
9154 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
9155 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
9156 and fns handling, rather than in a separate case SSA_NAME.
9158 2017-01-02 Jeff Law <law@redhat.com>
9160 * config/darwin-driver.c (darwin_driver_init): Const-correctness
9161 fixes for first_period and second_period variables.
9163 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
9166 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
9167 (*insvqi_1): New insn pattern.
9168 (*insvqi_1_mem_rex64): Ditto.
9170 (*insvqi_3): Rename from *insvqi.
9172 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
9174 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
9176 * doc/cfg.texi (Edges): Remove reference to Java.
9177 (Maintaining the CFG): Ditto.
9179 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
9182 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
9183 transparent aliases.
9185 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
9188 * predict.def (PRED_CALL): Update hitrate.
9189 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
9190 * predict.c (tree_estimate_probability_bb): Split CALL predictor
9191 into direct/indirect/polymorphic variants.
9193 2017-01-01 Jakub Jelinek <jakub@redhat.com>
9195 Update copyright years.
9197 * gcc.c (process_command): Update copyright notice dates.
9198 * gcov-dump.c (print_version): Ditto.
9199 * gcov.c (print_version): Ditto.
9200 * gcov-tool.c (print_version): Ditto.
9201 * gengtype.c (create_file): Ditto.
9202 * doc/cpp.texi: Bump @copying's copyright year.
9203 * doc/cppinternals.texi: Ditto.
9204 * doc/gcc.texi: Ditto.
9205 * doc/gccint.texi: Ditto.
9206 * doc/gcov.texi: Ditto.
9207 * doc/install.texi: Ditto.
9208 * doc/invoke.texi: Ditto.
9210 Copyright (C) 2017 Free Software Foundation, Inc.
9212 Copying and distribution of this file, with or without modification,
9213 are permitted in any medium without royalty provided the copyright
9214 notice and this notice are preserved.