1 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
3 * config/s390/predicates.md: Fix typo.
4 * config/s390/s390.md: Allow immediates for load on condition.
6 2018-11-05 Martin Liska <mliska@suse.cz>
8 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
9 * bitmap.h (struct bitmap_usage): Likewise.
10 * ggc-common.c (SCALE): Remove.
12 (struct ggc_usage): Use SIZE_AMOUNT. And update
14 * ggc-page.c (SCALE): Remove.
16 (ggc_print_statistics): Use SIZE_AMOUNT.
17 * gimple.h (SCALE): Remove.
19 * input.c (ONE_K): Remove.
22 (STAT_LABEL): Likewise.
23 (FORMAT_AMOUNT): Likewise.
24 (dump_line_table_statistics): Use SIZE_AMOUNT.
25 * mem-stats.h (struct mem_usage): Likewise.
26 * rtl.c (dump_rtx_statistics): Likewise.
27 (rtx_alloc_counts): Change type to size_t.
28 (rtx_alloc_sizes): Likewise.
30 (dump_rtx_statistics): Sort first based on counts.
31 * tree.c (tree_nodes_cmp): New.
32 (tree_codes_cmp): New.
33 (dump_tree_statistics): Sort first based on counts.
34 * system.h (ONE_K): New.
36 (SIZE_SCALE): Likewise.
37 (SIZE_LABEL): Likewise.
38 (SIZE_AMOUNT): Likewise.
39 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
40 * tree-dfa.c (dump_dfa_stats): Likewise.
41 * tree-phinodes.c (phinodes_print_statistics): Likewise.
42 * tree-ssanames.c (ssanames_print_statistics): Likewise.
43 * tree.c (dump_tree_statistics): Likewise.
44 * vec.c (struct vec_usage): Likewise.
45 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
46 trigger a -Werror=format-overflow with
47 --enable-gather-detailed-stats.
49 2018-11-05 Martin Liska <mliska@suse.cz>
51 * mem-stats.h (mem_alloc_description::release_instance_overhead):
53 * vec.c (struct vec_usage): Register m_element_size.
54 (vec_prefix::register_overhead): New arguments: elements and
56 (vec_prefix::release_overhead): Subtract elements.
57 * vec.h (struct vec_prefix): Change signature.
58 (va_heap::reserve): Pass proper arguments.
59 (va_heap::release): Likewise.
61 2018-11-05 Martin Liska <mliska@suse.cz>
63 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
67 2018-11-05 Richard Biener <rguenther@suse.de>
69 * tree-scalar-evolution.h (final_value_replacement_loop): Update
71 * tree-scalar-evolution.c (final_value_replacement_loop): Return
72 whether anything was done.
73 (scev_const_prop): Remove constant propagation part, fold
75 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
76 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
79 2018-11-05 Jakub Jelinek <jakub@redhat.com>
81 PR tree-optimization/87859
82 * gimple-ssa-store-merging.c (struct merged_store_group): Add
83 only_constants and first_nonmergeable_order members.
84 (merged_store_group::merged_store_group): Initialize them.
85 (merged_store_group::do_merge): Clear only_constants member if
86 adding something other than INTEGER_CST store.
87 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
88 stores with order >= first_nonmergeable_order. Use
89 merged_store->only_constants instead of always recomputing it.
90 Set merged_store->first_nonmergeable_order if we've skipped any
91 stores. Attempt to merge overlapping INTEGER_CST stores that
92 we would otherwise skip.
95 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
97 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
100 * config/i386/emmintrin.h (__v16qs): New to cope with option
102 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
103 (_mm_cmplt_epi8): Likewise.
104 (_mm_cmpgt_epi8): Likewise.
106 2018-11-05 Richard Biener <rguenther@suse.de>
108 PR rtl-optimization/87852
109 * fwprop.c (use_killed_between): Only consider single-defs of the
110 use whose definition statement dominates the use.
112 2018-11-05 Martin Liska <mliska@suse.cz>
115 * doc/invoke.texi: Remove options that are
116 not disabled with -Os.
118 2018-11-05 Martin Liska <mliska@suse.cz>
121 * doc/extend.texi: Update constrain about the last argument
122 of __builtin_expect_with_probability.
124 2018-11-05 Martin Liska <mliska@suse.cz>
127 * predict.c (expr_expected_value_1): Verify
128 that last argument is a real constants and emit
131 2018-11-05 Martin Liska <mliska@suse.cz>
133 PR gcov-profile/77698
134 * ipa-profile.c (ipa_profile): Adjust hotness threshold
137 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
139 PR tree-optimization/86572
140 * builtins.c (c_strlen): Handle negative offsets in a safe way.
142 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
144 PR tree-optimization/87672
145 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
146 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
148 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
151 * cfgexpand.c (pass_expand::execute): Move the call to
152 finish_eh_generation in front of the call to expand_stack_alignment.
154 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
156 * common/config/i386/i386-common.c (processor_alias_table): Add
158 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
159 (case ${target}): Add znver2.
160 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
161 -march=native recognize znver2 processors.
162 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
163 * config/i386/i386.c (m_znver2): New definition.
164 (m_ZNVER): New definition.
165 (m_AMD_MULTIPLE): Includes m_znver2.
166 (processor_cost_table): Add znver2 entry.
167 (processor_target_table): Add znver2 entry.
168 (get_builtin_code_for_version): Set priority for
170 (processor_model): Add M_AMDFAM17H_ZNVER2.
171 (arch_names_table): Ditto.
172 (ix86_reassociation_width): Include znver2.
173 * config/i386/i386.h (TARGET_znver2): New definition.
174 (struct ix86_size_cost): Add TARGET_ZNVER2.
175 (enum processor_type): Add PROCESSOR_ZNVER2.
176 * config/i386/i386.md (define_attr "cpu"): Add znver2.
177 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
178 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
179 (ix86_adjust_cost): Add znver2.
180 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
181 * gcc/doc/extend.texi: Add details about znver2.
182 * gcc/doc/invoke.texi: Add details about znver2.
184 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
188 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
191 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
193 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
194 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
196 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
198 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
199 shrink-wrapping when -mtrack-speculation.
201 2018-11-02 Richard Biener <rguenther@suse.de>
203 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
204 (pop_cost_one_pair): Do not free pair.
205 (pop_best_coalesce): Likewise.
206 (create_coalesce_list): Initialize obstack.
207 (delete_coalesce_list): Free obstack.
208 (find_coalesce_pair): Obstack-allocate coalesce pairs.
209 (add_cost_one_coalesce): Likewise.
210 (struct live_track): Remove bitmap pointer indirections.
211 (new_live_track): Adjust.
212 (delete_live_track): Likewise.
213 (live_track_remove_partition): Likewise.
214 (live_track_add_partition): Likewise.
215 (live_track_live_p): Likewise.
216 (live_track_process_def): Likewise.
217 (live_track_clear_base_vars): Likewise.
219 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
221 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
222 * configure: Regenerate.
223 * config.in: Regenerate.
224 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
225 HAVE_AS_REGISTER_PSEUDO_OP guard.
226 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
227 !HAVE_AS_REGISTER_PSEUDO_OP support.
228 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
231 2018-11-02 Richard Biener <rguenther@suse.de>
233 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
236 2018-11-02 Richard Biener <rguenther@suse.de>
238 PR tree-optimization/87776
239 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
240 executable when iterating but running into rpo-vn-max-loop-depth
241 and not eliding the iteration.
243 2018-11-30 Jan Hubicka <jh@suse.cz>
245 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
248 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
250 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
252 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
253 Combine with rs6000_address_for_fpconvert.
254 (rs6000_address_for_fpconvert): Combine with
255 rs6000_force_indexed_or_indirect_mem.
256 (rs6000_expand_vector_init): Change function call from
257 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
258 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
259 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
260 (floatsi<mode>2_lfiwax_mem): Ditto.
261 (floatunssi<mode>2_lfiwzx): Ditto.
262 (floatunssi<mode>2_lfiwzx_mem): Ditto.
263 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
264 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
265 (fix_trunc<mode>si2_stfiwx): Ditto.
266 (fixuns_trunc<mode>si2_stfiwx): Ditto.
267 (float_<mode>si2_hw): Ditto.
268 (floatuns_<mode>si2_hw): Ditto.
269 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
270 (vsx_splat_<mode>): Ditto.
272 2018-11-01 Joseph Myers <joseph@codesourcery.com>
274 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
275 at end of assembler input text.
276 * configure: Regenerate.
278 2018-11-01 Jakub Jelinek <jakub@redhat.com>
280 PR tree-optimization/87826
281 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
282 negative or larger or equal to type's precision.
284 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
286 * opts.c (default_options_table): Do not enable
287 OPT_fdelayed_branch at -Og.
288 * doc/invoke.texi (-fdelayed-branch): Document it.
290 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
292 * optabs-libfuncs.c (build_libfunc_function_visibility):
293 New, split out from...
294 (build_libfunc_function): ... here.
295 (init_one_libfunc_visibility): New, split out from ...
296 (init_one_libfunc): ... here.
298 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
299 scratch register need not be early-clobber. Document the reason
300 why we cannot use ST<OP>.
302 2018-10-31 Joseph Myers <joseph@codesourcery.com>
305 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
306 line for second argument of AC_DEFINE_UNQUOTED.
307 * doc/install.texi (Tools/packages necessary for modifying GCC):
308 Update to autoconf 2.69 and automake 1.15.1.
309 * aclocal.m4, config.in, configure: Regenerate.
311 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
313 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
316 2018-10-31 Martin Liska <mliska@suse.cz>
319 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
320 Add new argument invalid_extension.
321 (aarch64_get_all_extension_candidates): New function.
322 (aarch64_rewrite_selected_cpu): Add NULL to function call.
323 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
325 (aarch64_get_all_extension_candidates): New function.
326 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
327 argument invalid_extension.
328 (aarch64_parse_cpu): Likewise.
329 (aarch64_print_hint_for_extensions): New function.
330 (aarch64_validate_mcpu): Provide hint about invalid extension.
331 (aarch64_validate_march): Likewise.
332 (aarch64_handle_attr_arch): Pass new argument.
333 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
334 (aarch64_handle_attr_isa_flags): Likewise.
336 2018-10-31 Richard Biener <rguenther@suse.de>
340 * tree-outof-ssa.c (insert_backedge_copies): Restrict
341 copy generation to useful cases. Place the copy before
342 the definition of the backedge value when possible.
344 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
346 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
347 * config/arc/arc.c (arc_active_insn): New function.
348 (check_store_cacheline_hazard): Likewise.
349 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
350 (arc_override_options): Disable delay slot scheduler for older
352 (arc_store_addr_hazard_p): New implementation, old one renamed to
354 (arc_store_addr_hazard_internal_p): Renamed.
355 (arc_reorg): Don't combine into brcc instructions which are part
356 of hardware hazard solution.
357 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
358 (tune_arc700): Likewise.
359 * config/arc/arc.opt (arc7xx): New tune value.
360 * config/arc/arc700.md: Improve A7 scheduler.
362 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
364 * config/arc/arc.c (arc_override_options): Remove
365 TARGET_COMPACT_CASESI.
366 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
367 (CASE_VECTOR_MODE): Likewise.
368 (CASE_VECTOR_PC_RELATIVE): Likewise.
369 (CASE_VECTOR_SHORTEN_MODE): Likewise.
370 (CASE_VECTOR_SHORTEN_MODE1): Delete.
371 (ADDR_VEC_ALIGN): Update.
372 (ASM_OUTPUT_CASE_LABEL): Undefine.
373 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
374 (TARGET_BI_BIH): Define.
375 (DEFAULT_BRANCH_INDEX): Likewise.
376 * config/arc/arc.md (casesi): Rework to accept BI/BIH
377 instructions, remove compact_casesi use case.
378 (casesi_compact_jump): Remove.
379 (casesi_dispatch): New pattern.
380 * config/arc/arc.opt: Add mbranch-index option. Deprecate
381 compact_casesi option.
382 * doc/invoke.texi: Document mbranch-index option.
384 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
386 * config/arc/arc.c (arc_get_tp): Remove function.
387 (arc_emit_call_tls_get_addr): Likewise.
388 (arc_call_tls_get_addr): New function.
389 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
390 * config/arc/arc.md (tls_load_tp_soft): Remove.
391 (tls_gd_get_addr): Likewise.
393 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
395 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
396 (mulsi3_600_lib): Remove pattern.
397 (umulsi3_highpart_600_lib_le): Likewise.
398 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
399 (umulsidi3): Remove call to umulsidi3_600_lib.
400 (umulsidi3_600_lib): Remove pattern.
401 (peephole2): Remove peephole using the above deprecated patterns.
403 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
406 * config/arm/arm.c (arm_option_check_internal): Disable the combined
407 use of -mslow-flash-data and -mword-relocations.
408 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
409 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
411 * doc/invoke.texi (-mword-relocations): Mention conflict with
413 (-mslow-flash-data): Reciprocally.
415 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
417 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
418 16-byte modes held in GP registers to use an even regno.
420 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
421 (aarch64_atomic_ldop_supported_p): Remove.
422 (aarch64_gen_atomic_ldop): Remove.
423 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
424 Fully expand LSE operations here.
425 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
426 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
427 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
428 and use ATOMIC_LDOP instead; use register_operand for the input;
429 drop the split and emit insns directly.
430 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
431 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
432 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
434 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
435 (aarch64_gen_atomic_ldop): Don't call it.
436 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
437 Use aarch64_reg_or_zero.
438 (aarch64_atomic_exchange<ALLI>): Likewise.
439 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
440 operand 0; use aarch64_reg_or_zero for input; merge ...
441 (@aarch64_atomic_swp<ALLI>): ... this and remove.
443 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
444 (aarch64_split_compare_and_swap): Use it.
445 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
446 test oldval against the proper predicate.
447 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
448 Use nonmemory_operand for expected.
449 (cas_short_expected_pred): New.
450 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
451 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
452 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
453 (aarch64_plushi_operand): New.
455 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
456 Force oldval into the rval register for TARGET_LSE; emit the compare
457 during initial expansion so that it may be deleted if unused.
458 (aarch64_gen_atomic_cas): Remove.
459 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
460 Change =&r to +r for operand 0; use match_dup for operand 2;
461 remove is_weak and mod_f operands as unused. Drop the split
463 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
464 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
465 (@aarch64_atomic_cas<GPI>): Similarly.
467 2018-10-31 Richard Biener <rguenther@suse.de>
469 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
472 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
474 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
475 * config/aarch64/aarch64.md: Include saphira.md
476 * config/aarch64/saphira.md: New file for pipeline description.
478 2018-10-30 Martin Sebor <msebor@redhat.com>
481 * gimple-ssa-sprintf.c (format_directive): Use %G to include
483 (sprintf_dom_walker::compute_format_length):
484 Avoid setting POSUNDER4K here.
485 (get_destination_size): Handle null argument values.
486 (get_user_idx_format): New function.
487 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
488 functions, including user-defined with attribute format printf.
489 Use %G to include inlining context.
492 2018-10-30 Jan Hubicka <jh@suse.cz>
494 * params.def (lto-partitions): Bump from 32 to 128.
496 2018-10-30 Jan Hubicka <jh@suse.cz>
499 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
501 (free_lang_data_in_type): Forward declare.
502 (fld_type_variant_equal_p): New function.
503 (fld_type_variant): New function
504 (fld_incomplete_types): New hash.
505 (fld_incomplete_type_of): New function
506 (fld_simplfied-type): New function.
507 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
508 (free_lang_data): Allocate and free fld_incomplete_type; update call
509 of free_lang_data_in_decl.
511 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
513 * gcov.c (output_lines): Remove duplicate line.
515 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
517 * config/rs6000/rs6000.md (bswapdi2): Force address into register
518 if not in indexed or indirect form.
519 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
520 (bswapdi2_store): Ditto.
521 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
523 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
524 Prototype for helper function.
526 2018-10-30 Martin Sebor <msebor@redhat.com>
528 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
529 (target, pragma GCC optimize, pragma GCC target): Ditto.
531 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
533 * doc/extend.texi: Fix prototype and description of
534 __builtin_expect_with_probability.
536 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
538 * cgraph.h (clone_function_name_1): Replaced by new
539 clone_function_name_numbered that takes name as string; for
540 privatize_symbol_name_1 use only.
541 (clone_function_name): Renamed to
542 clone_function_name_numbered to be explicit about numbering.
543 (clone_function_name): New two-argument function that does
544 not number its output.
545 (clone_function_name): New three-argument function that
546 takes a number to append to its output.
547 * cgraphclones.c (duplicate_thunk_for_node):
548 (clone_function_name_1): Renamed.
549 (clone_function_name_numbered): Two new functions.
550 (clone_function_name): Improved documentation.
551 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
552 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
553 * final.c (final_scan_insn_1): Use the new clone_function_name
555 * multiple_target.c (create_dispatcher_calls): Ditto.
556 (create_target_clone): Ditto.
557 * omp-expand.c (grid_expand_target_grid_body): Ditto.
558 * omp-low.c (create_omp_child_function_name): Ditto.
559 * omp-simd-clone.c (simd_clone_create): Ditto.
560 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
561 new clone_function_name without numbering.
563 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
565 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
566 Assert that the allocation size is not zero.
568 2018-10-30 Richard Biener <rguenther@suse.de>
570 PR tree-optimization/87800
571 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
572 non-induction or reduction PHIs.
574 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
576 * config/aarch64/falkor-tag-collision-avoidance.c
577 (execute_tag_collision_avoidance): Call df_note_add_problem.
579 2018-10-30 Martin Liska <mliska@suse.cz>
581 * doc/extend.texi: Fix typo in documentation
582 of __builtin_expect_with_probability.
584 2018-10-29 David Malcolm <dmalcolm@redhat.com>
587 * input.c (get_substring_ranges_for_loc): Detect if
588 linemap_resolve_location gives us a NULL map, and reject
591 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
593 * config.gcc (xstormy16-*-elf): Set tm_d_file.
595 2018-10-29 David Malcolm <dmalcolm@redhat.com>
596 Martin Sebor <msebor@redhat.com>
597 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
599 * doc/ux.texi (Quoting): New subsection, adapted from material at
600 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
601 MartinSebor and ManuelLopezIbanez.
602 (Fix-it hints): Note that fix-it hints shouldn't be marked for
605 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
608 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
611 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
613 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
615 2018-10-29 Paul Koning <ni1d@arrl.net>
617 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
619 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
621 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
622 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
623 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
624 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
625 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
626 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
627 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
628 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
629 Change 'vector' to '__vector'.
630 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
631 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
632 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
633 _mm_avg_pu16): Likewise. And, whitespace corrections.
635 2018-10-29 Richard Biener <rguenther@suse.de>
637 PR tree-optimization/87785
638 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
641 2018-10-29 Olivier Hainque <hainque@adacore.com>
643 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
645 2018-10-29 Olivier Hainque <hainque@adacore.com>
647 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
648 ports configuration macro, defaults to "gnu".
649 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
650 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
652 2018-10-29 Olivier Hainque <hainque@adacore.com>
654 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
656 2018-10-29 Richard Biener <rguenther@suse.de>
658 PR tree-optimization/87790
659 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
660 (vect_make_slp_decision): Adjust.
661 (vect_slp_analyze_bb_1): Likewise.
662 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
665 2018-10-29 Richard Biener <rguenther@suse.de>
667 PR tree-optimization/87785
668 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
670 (vect_build_slp_tree): Likewise.
671 (vect_gather_slp_loads): New function.
672 (vect_analyze_slp_instance): Gather loads separately from the
675 2018-10-29 Martin Liska <mliska@suse.cz>
677 * Makefile.in: Make dependency to json.o.
678 * doc/gcov.texi: Document new JSON format, remove
679 old intermediate format documentation.
680 * gcov.c (struct function_info): Come up with m_name and
682 (function_info::function_info): Initialize it.
683 (function_info::~function_info): Release it.
684 (main): Rename flag_intermediate_format to flag_json_format.
685 (print_usage): Describe --json-format.
686 (process_args): Set flag_json_format.
687 (output_intermediate_line): Remove.
688 (output_intermediate_json_line): Likewise.
689 (get_gcov_intermediate_filename): Return new extension
691 (output_intermediate_file): Implement JSON emission.
692 (output_json_intermediate_file): Implement JSON emission.
693 (generate_results): Use ::get_name for function name.
694 Handle JSON output file.
695 (read_graph_file): Use ::get_name instead of cplus_demangle.
696 (read_count_file): Likewise.
697 (solve_flow_graph): Likewise.
698 (add_line_counts): Likewise.
699 (accumulate_line_counts): Use new flag_json_format.
700 (output_function_details): Use ::get_name instead of cplus_demangle.
701 (output_lines): Likewise.
702 * json.cc (test_writing_literals): Add new tests.
703 * json.h (class literal): Add new boolean constructor.
705 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
707 PR rtl-optimization/87701
708 PR rtl-optimization/87780
709 * combine.c (make_more_copies): Rewrite.
711 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
713 * doc/generic.texi (ABSU_EXPR): Document.
714 * match.pd (absu(x)*absu(x) -> x*x): Handle.
715 (absu(absu(X)) -> absu(X)): Likewise.
716 (absu(-X) -> absu(X)): Likewise.
717 (absu(X) where X is nonnegative -> X): Likewise.
719 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
721 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
722 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
723 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
724 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
725 (s-tm-texi): Also check timestamp on d-target.def.
726 (generated_files): Add TM_D_H and d-target-hooks-def.h.
727 (build/genhooks.o): Also depend on D_TARGET_DEF.
728 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
730 * config/aarch64/aarch64-d.c: New file.
731 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
733 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
735 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
736 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
737 * config/arm/arm-d.c: New file.
738 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
739 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
740 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
741 * config/arm/t-arm (arm-d.o): New rule.
742 * config/default-d.c: New file.
743 * config/glibc-d.c: New file.
744 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
745 * config/i386/i386-d.c: New file.
746 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
747 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
748 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
749 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
750 * config/i386/t-i386 (i386-d.o): New rule.
751 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
752 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
753 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
754 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
755 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
756 * config/mips/mips-d.c: New file.
757 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
758 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
759 * config/mips/t-mips (mips-d.o): New rule.
760 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
761 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
762 * config/powerpcspe/powerpcspe-d.c: New file.
763 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
765 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
766 Support GNU D by using 0 as the language type.
767 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
768 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
769 * config/riscv/riscv-d.c: New file.
770 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
772 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
773 * config/riscv/t-riscv (riscv-d.o): New rule.
774 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
775 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
776 * config/rs6000/rs6000-d.c: New file.
777 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
779 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
780 Support GNU D by using 0 as the language type.
781 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
782 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
783 * config/s390/s390-d.c: New file.
784 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
785 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
786 * config/s390/t-s390 (s390-d.o): New rule.
787 * config/sparc/sparc-d.c: New file.
788 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
790 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
791 * config/sparc/t-sparc (sparc-d.o): New rule.
792 * config/t-glibc (glibc-d.o): New rule.
793 * configure: Regenerated.
794 * configure.ac (tm_d_file): New variable.
795 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
796 * doc/contrib.texi (Contributors): Add self for the D frontend.
797 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
798 * doc/install.texi (Configuration): Mention libphobos as an option for
799 --enable-shared. Mention d as an option for --enable-languages.
800 (Testing): Mention check-d as a target.
801 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
802 name suffixes. Mention d as a -x option.
803 * doc/sourcebuild.texi (Top Level): Mention libphobos.
804 * doc/standards.texi (Standards): Add section on D language.
805 * doc/tm.texi: Regenerated.
806 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
807 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
808 * dwarf2out.c (is_dlang): New function.
809 (gen_compile_unit_die): Use DW_LANG_D for D.
810 (declare_in_namespace): Return module die for D, instead of adding
811 extra declarations into the namespace.
812 (gen_namespace_die): Generate DW_TAG_module for D.
813 (gen_decl_die): Handle CONST_DECLSs for D.
814 (dwarf2out_decl): Likewise.
815 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
816 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
818 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
819 * genhooks.c: Include d/d-target.def.
821 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
824 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
825 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
826 (STACK_DYNAMIC_OFFSET): Likewise.
828 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
831 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
834 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
836 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
837 return type and other typos.
839 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
841 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
842 a shorter sequence with fewer branches.
843 (emit_final_str_compare_gpr): Ditto.
845 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
847 * config/rs6000/tmmintrin.h: New file.
848 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
850 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
852 * config/rs6000/mmintrin.h: Enable 32bit compilation.
853 * config/rs6000/xmmintrin.h: Likewise.
855 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
857 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
859 2018-10-26 Richard Biener <rguenther@suse.de>
861 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
863 (vect_mark_slp_stmts_relevant): Likewise.
864 (vect_detect_hybrid_slp_stmts): Likewise.
865 (vect_bb_slp_scalar_cost): Likewise.
866 (vect_remove_slp_scalar_calls): Likewise.
868 2018-10-26 Jan Hubicka <jh@suse.cz>
870 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
871 (warn_types_mismatch): Fix walk of DECL_NAME.
872 (odr_types_equivalent_p): Fix overactive assert.
874 2018-10-26 Richard Biener <rguenther@suse.de>
876 PR tree-optimization/87105
877 * tree-vectorizer.h (_slp_tree::refcnt): New member.
878 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
880 (vect_create_new_slp_node): Initialize refcnt to one.
882 (scalar_stmts_set_t, bst_fail): Remove.
883 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
884 (vect_build_slp_tree): Add bst_map argument and lookup
885 already created SLP nodes.
886 (vect_print_slp_tree): Handle a SLP graph, print SLP node
888 (vect_slp_rearrange_stmts): Handle a SLP graph.
889 (vect_analyze_slp_instance): Adjust and free SLP nodes from
890 the CSE map. Fix indenting.
891 (vect_schedule_slp_instance): Add short-cut.
893 2018-10-26 Martin Liska <mliska@suse.cz>
896 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
897 addr_expr and not with pointers.
899 2018-10-26 Jan Hubicka <jh@suse.cz>
901 * tree.c (free_lang_data_in_type): Only check main variants.
902 * ipa-devirt.c (warn_odr): Make static.
903 (types_same_for_odr): Drop strict variant.
904 (types_odr_comparable): Likewise.
905 (odr_or_derived_type_p): Look for main variants.
906 (odr_name_hasher::equal): Cleanup comment.
907 (odr_subtypes_equivalent): Add warn and warned arguments; check main
909 (type_variants_equivalent_p): break out from ...
910 (odr_types_equivalent): ... here; go for main variants where needed.
911 (warn_odr): ... here; turn static.
912 (warn_types_mismatch): Compare mangled names of main variants.
913 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
914 (type_with_linkage_p): Sanity check that we look at main variant.
915 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
916 * tree.h (types_same_for_odr): Drop strict argument.
918 2018-10-26 Richard Biener <rguenther@suse.de>
920 PR tree-optimization/87746
921 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
922 Simplify and fix WRT strided store groups with size not
923 equal to step in element count.
924 (vect_analyze_group_access_1): Dump the whole group.
926 2018-10-25 Carl Love <cel@us.ibm.com>
928 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
929 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
930 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
931 precicion entry for each overloaded builtin.
932 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
933 VSCEDPUO): Rename overloaded name.
934 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
935 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
936 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
937 define_expand for xscmpexqp instruction.
938 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
940 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
941 Jinsong Ji <jji@us.ibm.com>
943 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
944 function with vec_sl.
945 (_mm_slli_epi32): Likewise.
946 (_mm_slli_epi64): Likewise.
947 (_mm_srai_epi16): Replace deprecated function with vec_sra.
948 (_mm_srai_epi32): Likewise.
949 (_mm_srli_epi16): Replace deprecated function with vec_sr.
950 (_mm_srli_epi32): Likewise.
951 (_mm_srli_epi64): Likewise.
952 (_mm_sll_epi16): Replace deprecated function with vec_sl.
953 (_mm_sll_epi32): Likewise.
954 (_mm_sll_epi64): Likewise.
955 (_mm_sra_epi16): Replace deprecated function with vec_sra.
956 (_mm_sra_epi32): Likewise.
957 (_mm_srl_epi16): Replace deprecated function with vec_sr.
958 (_mm_srl_epi32): Likewise.
959 (_mm_srl_epi64): Likewise.
961 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
962 Jinsong Ji <jji@us.ibm.com>
964 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
965 comparison operators with vec_cmp* for compatibility due to
966 unfortunate history; clean up formatting and use types more
968 (_mm_sll_epi32): Likewise.
969 (_mm_sll_epi64): Likewise.
970 (_mm_srl_epi16): Likewise.
971 (_mm_srl_epi32): Likewise.
972 (_mm_srl_epi64): Likewise.
974 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
975 Jinsong Ji <jji@us.ibm.com>
977 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
978 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
979 __vector __bool int. Use vec_cmpgt in preference to deprecated
980 function vec_vcmpgtfp.
981 (_mm_max_ps): Likewise.
983 2018-10-25 Jeff Law <law@redhat.com>
985 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
988 2018-10-25 Martin Sebor <msebor@redhat.com>
990 * doc/extend.texi (aligned): Expand attribute description.
991 (Alignment): Rename section. Discuss function arguments.
993 2018-10-25 Jan Hubicka <jh@suse.cz>
995 * ipa-devirt.c (main_odr_variant): Remove.
996 (hash_odr_name, types_same_for_odr, types_odr_comparable,
997 odr_name_hasher::equal, odr_subtypes_equivalent_p):
998 Drop use of main_odr_variant.
999 (add_type_duplicate): Silence confused warnings on integer types.
1000 (get_odr_type): Always look for main variant.
1001 (register_odr_type): Simplify.
1003 2018-10-25 Richard Biener <rguenther@suse.de>
1005 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
1006 Initialize ng to silence error with release checking bootstrap.
1008 2018-10-25 Richard Biener <rguenther@suse.de>
1010 * tree-if-conv.c: Include tree-ssa-sccvn.h.
1011 (tree_if_conversion): Run CSE on the if-converted loop body.
1013 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
1015 * config/s390/constraints.md (ZL): New constraint.
1016 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
1018 * config/s390/s390.md (movdi_larl): Remove.
1019 (movdi_64): Add the LARL alternative.
1021 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
1024 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
1025 (RTX_CODE_HWINT_P): New macro.
1026 (rtx_code_size): Use RTX_CODE_HWINT_P ().
1028 2018-10-25 Jan Hubicka <jh@suse.cz>
1030 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
1033 2018-10-25 Richard Biener <rguenther@suse.de>
1035 PR tree-optimization/87665
1036 PR tree-optimization/87745
1037 * tree-vectorizer.h (get_earlier_stmt): Remove.
1038 (get_later_stmt): Pick up UID from the original non-pattern stmt.
1040 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
1042 * options.texi (Deprecated): Move list to Var section.
1044 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
1045 Jinsong Ji <jji@us.ibm.com>
1047 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
1048 __vector long to __vector long long.
1049 (_mm_cvtpd_ps): Likewise.
1050 (_mm_cvttpd_epi32): Likewise.
1051 (_mm_cvtpi32_pd): Likewise.
1052 (_mm_unpackhi_epi64): Likewise.
1053 (_mm_unpacklo_epi64): Likewise.
1055 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
1057 PR rtl-optimization/87720
1058 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
1060 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
1062 * gimple-ssa-isolate-paths.c
1063 (find_implicit_erroneous_behavior): Do not change code if the
1064 pass is running for warnings only.
1065 (find_explicit_erroneous_behavior): Likewise.
1067 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
1069 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
1070 Define as rs6000_mangle_decl_assembler_name.
1071 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
1072 long double to IEEE long double, switch the names of the long
1073 double built-in functions to be <func>f128 instead of <func>l.
1075 2018-10-24 Martin Sebor <msebor@redhat.com>
1077 * doc/extend.texi (nonnull): List no-argument form. Reference
1078 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
1080 2018-10-24 Richard Biener <rguenther@suse.de>
1082 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
1084 2018-10-24 Martin Liska <mliska@suse.cz>
1086 PR tree-optimization/84436
1087 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
1089 (switch_conversion::contains_linear_function_p): New.
1090 (switch_conversion::build_one_array): Support linear
1091 transformation on input.
1092 * tree-switch-conversion.h (struct switch_conversion): Add
1093 contains_linear_function_p declaration.
1095 2018-10-24 Richard Biener <rguenther@suse.de>
1097 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
1098 if its argument is CONSTANT_CLASS_P.
1100 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
1102 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
1103 it is wrong for forward declarations.
1105 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
1107 * config/s390/s390.c (s390_check_qrst_address): Add the missing
1108 SYMBOL_REF_P () check.
1110 2018-10-24 Richard Biener <rguenther@suse.de>
1112 PR tree-optimization/87105
1113 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
1114 dump classification.
1115 (vect_analyze_data_ref_accesses): Handle duplicate loads and
1116 stores by splitting the affected group after the fact.
1117 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
1118 fail the SLP build because of size constraints.
1120 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1122 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
1123 * configure: Regenerate.
1124 * config.in: Regenerate.
1125 * varasm.c (mergeable_string_section): Use readonly_data_section
1126 if linker doesn't support SHF_MERGE with alignment > 8.
1127 (mergeable_constant_section): Likewise.
1129 2018-10-24 Richard Biener <rguenther@suse.de>
1131 PR tree-optimization/84013
1132 * tree-ssa-structalias.c (struct msdi_data): New struct for
1133 marshalling data to walk_stmt_load_store_ops.
1134 (maybe_set_dependence_info): Refactor as callback for
1135 walk_stmt_load_store_ops.
1136 (compute_dependence_clique): Set restrict info on all stmt kinds.
1138 2018-10-24 Martin Liska <mliska@suse.cz>
1140 * cgraph.c (cgraph_node::dump):
1141 Remove reduntant dumps and make tp_first_run dump more compact.
1143 2018-10-24 Richard Biener <rguenther@suse.de>
1145 PR tree-optimization/87665
1146 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
1149 2018-10-12 Jeff Law <law@redhat.com>
1151 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
1154 2018-10-23 Richard Biener <rguenther@suse.de>
1156 * tree-vrp.c (add_assert_info): Guard dump_printf with
1158 * gimple-ssa-evrp-analyze.c
1159 (evrp_range_analyzer::record_ranges_from_incoming_edge):
1160 Use value_range::ignore_equivs_equal_p.
1162 2018-10-23 Richard Biener <rguenther@suse.de>
1164 PR tree-optimization/87105
1165 PR tree-optimization/87608
1166 * passes.def (pass_all_early_optimizations): Add early phi-opt
1168 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
1169 addition to debug stmts.
1170 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
1171 and abs replacement early.
1172 * tree-cfg.c (gimple_empty_block_p): Likewise.
1174 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
1177 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
1178 specified to configure.
1179 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
1181 2018-10-23 Richard Biener <rguenther@suse.de>
1183 PR tree-optimization/87700
1184 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
1186 2018-10-23 Jakub Jelinek <jakub@redhat.com>
1189 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
1190 second argument from __mmask16 to __mmask8.
1191 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
1192 _mm_mask_packs_epi32): Likewise.
1193 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
1195 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
1197 2018-10-23 Richard Biener <rguenther@suse.de>
1199 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
1201 2018-10-23 Richard Biener <rguenther@suse.de>
1203 PR tree-optimization/86144
1204 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
1205 over simd attribute.
1207 2018-10-23 Richard Biener <rguenther@suse.de>
1209 PR tree-optimization/87693
1210 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
1211 the case we do not find the taken edge.
1213 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
1214 Jinsong Ji <jji@us.ibm.com>
1216 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
1217 (_mm_store_pd): Use unaligned vector type for pointer cast.
1218 (_mm_maskmoveu_si128): Likewise.
1219 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
1220 (_mm_store_ps): Use unaligned vector type for pointer cast.
1222 2018-10-22 Paul Koning <ni1d@arrl.net>
1224 * symtab.c (symtab_node::increase_alignment): Correct max
1227 2018-10-22 Yury Gribov <tetra2005@gmail.com>
1229 PR tree-optimization/87633
1230 * match.pd: Do not generate unordered integer comparisons.
1232 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
1234 PR rtl-optimization/87600
1235 * combine.c: Add include of expr.h.
1236 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
1237 register to a pseudo.
1238 (make_more_copies): New function, add a copy to a new pseudo after
1239 the moves from hard registers into pseudos.
1240 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
1241 later. Call make_more_copies.
1243 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
1245 * lra-constraints.c (process_alt_operands): New local array,
1246 matching_early_clobber. Check matching_early_clobber before
1247 decrementing reject, and set matching_early_clobber after.
1249 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
1252 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
1253 call output_addr_const and hope for the best.
1255 2018-10-22 Richard Biener <rguenther@suse.de>
1257 * gimple-ssa-evrp-analyze.c
1258 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
1259 smarter about what ranges to use.
1260 * tree-vrp.c (add_assert_info): Dump here.
1261 (register_edge_assert_for_2): Instead of here at multiple but
1264 * gcc.dg/tree-ssa/evrp12.c: New testcase.
1265 * gcc.dg/predict-6.c: Adjust.
1266 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
1267 * gcc.dg/tree-ssa/vrp02.c: Likewise.
1268 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
1270 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
1271 Richard Biener <rguenther@suse.de>
1273 * bitmap.h: Update data structure documentation, including a
1274 description of bitmap views as either linked-lists or splay trees.
1275 (struct bitmap_element_def): Update comments for splay tree bitmaps.
1276 (struct bitmap_head_def): Likewise.
1277 (bitmap_list_view, bitmap_tree_view): New prototypes.
1278 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
1280 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
1281 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
1282 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
1283 released bitmap element here.
1284 (bitmap_element_free): Remove.
1285 (bitmap_elt_clear_from): Work on splay tree bitmaps.
1286 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
1287 this function similar ones such that linked-list bitmap implementation
1288 functions are grouped.
1289 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
1290 and moved for grouping.
1291 (bitmap_list_insert_element_after): Renamed from
1292 bitmap_elt_insert_after, and moved for grouping.
1293 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
1294 (bitmap_tree_link_left, bitmap_tree_link_right,
1295 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
1296 bitmap_tree_link_element, bitmap_tree_unlink_element,
1297 bitmap_tree_find_element): New functions for splay-tree bitmap
1299 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
1300 Renamed and moved, see above entries.
1301 (bitmap_tree_listify_from): New function to convert part of a splay
1302 tree bitmap to a linked-list bitmap.
1303 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
1304 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
1305 (bitmap_find_bit): Remove.
1306 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
1307 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
1308 Handle splay tree bitmaps.
1309 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
1310 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
1311 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
1312 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
1313 bitmap_intersect_compl_p, bitmap_ior_and_compl,
1314 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
1315 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
1316 corresponding changes to use linked-list specific bitmap_element
1317 manipulation functions as applicable for efficiency.
1318 (bitmap_tree_to_vec): New function.
1319 (debug_bitmap_elt_file): New function split out from ...
1320 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
1321 (bitmap_print): Likewise.
1323 PR tree-optimization/63155
1324 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
1326 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
1329 2018-10-22 Martin Liska <mliska@suse.cz>
1331 PR tree-optimization/87686
1333 2018-08-29 Martin Liska <mliska@suse.cz>
1335 * tree-switch-conversion.c (switch_conversion::expand):
1336 Strenghten assumption about gswitch statements.
1338 2018-10-22 Martin Liska <mliska@suse.cz>
1340 * ipa-icf.c (sem_item::compare_attributes): Remove.
1341 (sem_item::compare_referenced_symbol_properties): Use
1342 attribute_list_equal instead.
1343 (sem_function::equals_wpa): Likewise.
1344 * ipa-icf.h: Remove compare_attributes.
1346 2018-10-22 Richard Biener <rguenther@suse.de>
1349 * mem-stats.h (mem_usage::operator==): Fix pasto.
1351 2018-10-22 Richard Biener <rguenther@suse.de>
1353 PR tree-optimization/87640
1354 * tree-vrp.c (set_value_range_with_overflow): Decompose
1356 (extract_range_from_binary_expr_1): Adjust.
1358 2018-10-22 Martin Jambor <mjambor@suse.cz>
1360 * tree-eh.h (stmt_could_throw_p): Add function parameter.
1361 (stmt_can_throw_external): Likewise.
1362 (stmt_can_throw_internal): Likewise.
1363 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
1364 (lower_eh_constructs_2): Likewise.
1365 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
1366 (stmt_can_throw_external): Likewise.
1367 (stmt_can_throw_internal): Likewise.
1368 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
1369 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
1370 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
1371 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
1372 (pass_lower_eh_dispatch::execute): Pass cfun to
1373 stmt_can_throw_external.
1374 (cleanup_empty_eh): Likewise.
1375 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
1376 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
1377 stmt_can_throw_external instead of pushing it to cfun.
1378 (symbol_table::create_edge): Likewise.
1379 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
1380 stmt_can_throw_internal.
1381 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
1382 to stmt_could_throw_p.
1383 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
1384 stmt_can_throw_internal.
1385 (pass_store_merging::execute): Likewise.
1386 * gimple-ssa-strength-reduction.c
1387 (find_candidates_dom_walker::before_dom_children): Pass cfun to
1389 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
1390 stmt_can_throw_internal.
1391 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
1392 to stmt_can_throw_external.
1393 (check_stmt): Pass cfun to stmt_could_throw_p.
1394 (check_stmt): Pass cfun to stmt_can_throw_external.
1395 (pass_nothrow::execute): Likewise.
1396 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
1397 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
1398 stmt_can_throw_internal.
1399 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
1400 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
1401 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
1402 * tree-complex.c (expand_complex_libcall): Pass cfun to
1403 stmt_could_throw_p and to stmt_can_throw_internal.
1404 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
1405 * tree-inline.c (copy_edges_for_bb): Likewise.
1406 (maybe_move_debug_stmts_to_successors): Likewise.
1407 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
1409 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
1410 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
1411 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
1412 stmt_can_throw_internal.
1413 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
1414 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
1416 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
1417 stmt_can_throw_internal.
1418 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
1419 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
1421 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
1422 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
1423 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1424 (convert_mult_to_fma_1): Likewise.
1425 (convert_to_divmod): Likewise.
1426 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
1427 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
1428 * tree-ssa-propagate.c
1429 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
1430 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
1431 (maybe_optimize_range_tests): Likewise.
1432 (linearize_expr_tree): Likewise.
1433 (reassociate_bb): Likewise.
1434 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
1435 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
1436 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
1437 (handle_char_store): Likewise.
1438 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
1439 stmt_can_throw_internal.
1440 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
1442 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
1443 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
1444 (vectorizable_simd_clone_call): Likewise.
1445 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
1446 (gimple_stringop_fixed_value): Likewise.
1448 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
1450 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
1451 literal pool references.
1452 (s390_check_qrst_address): Adapt to the new behavior of
1453 s390_loadrelative_operand_p ().
1455 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1458 * config/i386/sse.md (*andnot<mode>3_bcst): New.
1460 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1463 * config/i386/sse.md (*<code><mode>3_bcst): New.
1465 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1468 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
1469 V4DI, V16SI and V8DI.
1470 (*sub<mode>3<mask_name>_bcst): New.
1471 (*add<mode>3<mask_name>_bcst): Likewise.
1473 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
1474 Jinsong Ji <jji@us.ibm.com>
1476 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
1477 __m64 with __vector unsigned long long for compatibility.
1478 (_mm_movemask_epi8): Likewise.
1479 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
1480 (_mm_cvttps_pi32): Likewise.
1481 (_mm_cvtpi32_ps): Likewise.
1482 (_mm_cvtps_pi16): Likewise.
1483 (_mm_loadh_pi): Likewise.
1484 (_mm_storeh_pi): Likewise.
1485 (_mm_movehl_ps): Likewise.
1486 (_mm_movelh_ps): Likewise.
1487 (_mm_loadl_pi): Likewise.
1488 (_mm_storel_pi): Likewise.
1489 (_mm_movemask_ps): Likewise.
1490 (_mm_shuffle_pi16): Likewise.
1492 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1495 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
1496 __builtin_ia32_vfnmsubpd512_mask.
1497 (_mm512_mask_fnmsub_round_pd): Likewise.
1498 (_mm512_fnmsub_pd): Likewise.
1499 (_mm512_mask_fnmsub_pd): Likewise.
1500 (_mm512_maskz_fnmsub_round_pd): Use
1501 __builtin_ia32_vfnmsubpd512_maskz.
1502 (_mm512_maskz_fnmsub_pd): Likewise.
1503 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
1504 (_mm512_mask_fnmsub_round_ps): Likewise.
1505 (_mm512_fnmsub_ps): Likewise.
1506 (_mm512_mask_fnmsub_ps): Likewise.
1507 (_mm512_maskz_fnmsub_round_ps): Use
1508 __builtin_ia32_vfnmsubps512_maskz.
1509 (_mm512_maskz_fnmsub_ps): Likewise.
1510 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
1511 __builtin_ia32_vfnmsubpd256_mask.
1512 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
1513 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1514 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
1515 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1516 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1517 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
1518 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
1519 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
1520 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
1521 __builtin_ia32_vfnmsubpd.
1522 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
1523 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
1524 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
1525 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
1526 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
1527 * config/i386/i386-builtin.def: Add
1528 __builtin_ia32_vfnmsubpd256_mask,
1529 __builtin_ia32_vfnmsubpd256_maskz,
1530 __builtin_ia32_vfnmsubpd128_mask,
1531 __builtin_ia32_vfnmsubpd128_maskz,
1532 __builtin_ia32_vfnmsubps256_mask,
1533 __builtin_ia32_vfnmsubps256_maskz,
1534 __builtin_ia32_vfnmsubps128_mask,
1535 __builtin_ia32_vfnmsubps128_maskz,
1536 __builtin_ia32_vfnmsubpd512_mask,
1537 __builtin_ia32_vfnmsubpd512_maskz,
1538 __builtin_ia32_vfnmsubps512_mask,
1539 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
1540 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
1541 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
1542 __builtin_ia32_vfnmsubpd256.
1543 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
1544 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
1545 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
1547 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
1549 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
1551 (fmai_vmfnmsub_<mode><round_name>): Likewise.
1553 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1556 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
1557 __builtin_ia32_vfnmaddpd512_mask.
1558 (_mm512_mask_fnmadd_round_pd): Likewise.
1559 (_mm512_fnmadd_pd): Likewise.
1560 (_mm512_mask_fnmadd_pd): Likewise.
1561 (_mm512_maskz_fnmadd_round_pd): Use
1562 __builtin_ia32_vfnmaddpd512_maskz.
1563 (_mm512_maskz_fnmadd_pd): Likewise.
1564 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
1565 (_mm512_mask_fnmadd_round_ps): Likewise.
1566 (_mm512_fnmadd_ps): Likewise.
1567 (_mm512_mask_fnmadd_ps): Likewise.
1568 (_mm512_maskz_fnmadd_round_ps): Use
1569 __builtin_ia32_vfnmaddps512_maskz.
1570 (_mm512_maskz_fnmadd_ps): Likewise.
1571 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
1572 __builtin_ia32_vfnmaddpd256_mask.
1573 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
1574 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
1575 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
1576 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1577 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1578 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
1579 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
1580 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
1581 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
1582 __builtin_ia32_vfnmaddpd.
1583 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
1584 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
1585 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
1586 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
1587 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
1588 * config/i386/i386-builtin.def: Add
1589 __builtin_ia32_vfnmaddpd256_mask,
1590 __builtin_ia32_vfnmaddpd256_maskz,
1591 __builtin_ia32_vfnmaddpd128_mask,
1592 __builtin_ia32_vfnmaddpd128_maskz,
1593 __builtin_ia32_vfnmaddps256_mask,
1594 __builtin_ia32_vfnmaddps256_maskz,
1595 __builtin_ia32_vfnmaddps128_mask,
1596 __builtin_ia32_vfnmaddps128_maskz,
1597 __builtin_ia32_vfnmaddpd512_mask,
1598 __builtin_ia32_vfnmaddpd512_maskz,
1599 __builtin_ia32_vfnmaddps512_mask,
1600 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
1601 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
1602 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
1603 __builtin_ia32_vfnmaddpd256.
1604 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
1605 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
1606 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
1608 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
1610 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
1612 (fmai_vmfnmadd_<mode><round_name>): Likewise.
1614 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1617 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
1618 __builtin_ia32_vfmsubpd512_mask.
1619 (_mm512_mask_fmsub_round_pd): Likewise.
1620 (_mm512_fmsub_pd): Likewise.
1621 (_mm512_mask_fmsub_pd): Likewise.
1622 (_mm512_maskz_fmsub_round_pd): Use
1623 __builtin_ia32_vfmsubpd512_maskz.
1624 (_mm512_maskz_fmsub_pd): Likewise.
1625 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
1626 (_mm512_mask_fmsub_round_ps): Likewise.
1627 (_mm512_fmsub_ps): Likewise.
1628 (_mm512_mask_fmsub_ps): Likewise.
1629 (_mm512_maskz_fmsub_round_ps): Use
1630 __builtin_ia32_vfmsubps512_maskz.
1631 (_mm512_maskz_fmsub_ps): Likewise.
1632 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
1633 __builtin_ia32_vfmsubpd256_mask.
1634 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
1635 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1636 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
1637 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1638 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1639 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
1640 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
1641 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
1642 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
1643 __builtin_ia32_vfmsubpd.
1644 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
1645 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
1646 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
1647 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
1648 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
1649 * config/i386/i386-builtin.def: Add
1650 __builtin_ia32_vfmsubpd256_mask,
1651 __builtin_ia32_vfmsubpd256_maskz,
1652 __builtin_ia32_vfmsubpd128_mask,
1653 __builtin_ia32_vfmsubpd128_maskz,
1654 __builtin_ia32_vfmsubps256_mask,
1655 __builtin_ia32_vfmsubps256_maskz,
1656 __builtin_ia32_vfmsubps128_mask,
1657 __builtin_ia32_vfmsubps128_maskz,
1658 __builtin_ia32_vfmsubpd512_mask,
1659 __builtin_ia32_vfmsubpd512_maskz,
1660 __builtin_ia32_vfmsubps512_mask,
1661 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
1662 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
1663 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
1664 __builtin_ia32_vfmsubpd256.
1665 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
1666 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
1667 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
1669 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
1671 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
1673 (fmai_vmfmsub_<mode><round_name>): Likewise.
1675 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1677 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
1678 Remove plus. Renamed to ...
1679 (*sub<mode>3<mask_name>_bcst): This.
1680 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
1681 (*add<mode>3<mask_name>_bcst): This.
1683 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1686 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
1688 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1691 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
1692 (_mm_or_epi32): Likewise.
1693 (_mm256_xor_epi32): Likewise.
1694 (_mm_xor_epi32): Likewise.
1695 (_mm256_or_epi64): Likewise.
1696 (_mm_or_epi64): Likewise.
1697 (_mm256_xor_epi64): Likewise.
1698 (_mm_xor_epi64): Likewise.
1700 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
1703 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
1705 2018-10-20 Jakub Jelinek <jakub@redhat.com>
1708 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
1710 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
1712 * doc/ux.texi: Move @section directly after @node.
1714 2018-10-19 Jakub Jelinek <jakub@redhat.com>
1718 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
1719 depend closely nested inside of loop with ordered clause with
1722 2018-10-19 David Malcolm <dmalcolm@redhat.com>
1724 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
1725 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
1726 * doc/ux.texi: New file.
1728 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1730 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
1731 be the first CR field allocated.
1733 2018-10-19 Richard Biener <rguenther@suse.de>
1736 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
1737 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
1739 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1742 * config/i386/sse.md
1743 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
1744 (*add<mode>3<mask_name>_bcst_2): Likewise.
1746 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1748 * config/i386/sse.md
1749 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1750 Replace nonimmediate_operand with register_operand.
1751 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1753 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1756 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
1758 PR rtl-optimization/87596
1759 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
1760 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
1761 for instructions in FROM..TO range.
1763 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
1765 * cfgexpand.c (expand_one_var): Use specific wording in error message
1766 for non-local frame variables.
1767 * stor-layout.c (layout_decl): Do not issue a warning for them.
1769 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
1771 * haifa-sched.c (priority): Add force_recompute parameter.
1772 (apply_replacement): Call priority () with force_recompute = true.
1773 (restore_pattern): Likewise.
1775 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1777 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
1778 HOST_BITS_PER_WIDE_INT.
1779 (test_vector_ops_duplicate): Likewise.
1781 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1784 * config/i386/sse.md (VF_AVX512): New.
1785 (avx512bcst): Likewise.
1786 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1788 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1790 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1793 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
1795 * doc/invoke.texi (-dumpversion): Improve grammar.
1796 (-dumpfullversion): Make more consistent with -dumpversion.
1798 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
1800 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
1801 Set JUMP_LABEL to the jump insn.
1802 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
1803 Predict emitted jump and add label to jump insn.
1805 2018-10-18 David Malcolm <dmalcolm@redhat.com>
1807 PR tree-optimization/87562
1808 * input.c (get_substring_ranges_for_loc): Use
1809 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
1810 getting the linemap for the endpoint. Verify that it's either
1811 in the same linemap as the start point's spelling location, or
1812 at least in the same file.
1814 2018-10-18 Richard Biener <rguenther@suse.de>
1816 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
1817 feed width-specific load/store costs through ix86_vec_cost.
1818 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
1819 (k8_cost): Likewise.
1820 (bdver_cost): Likewise.
1821 (znver1_cost): Likewise.
1822 (btver1_cost): Likewise.
1823 (btver2_cost): Likewise.
1825 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1827 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
1828 to simplify subreg of vec_merge.
1830 2018-10-18 Richard Biener <rguenther@suse.de>
1832 * config/i386/i386.c: Fix costing of vector FMA.
1834 2018-10-18 Richard Biener <rguenther@suse.de>
1836 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
1838 (ix86_builtin_vectorization_cost): For vec_construct properly
1839 cost insertion into SSE regs.
1840 (...): Adjust calls to ix86_vec_cost.
1842 2018-10-18 Richard Biener <rguenther@suse.de>
1846 2018-02-07 Richard Biener <rguenther@suse.de>
1848 PR tree-optimization/84204
1849 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
1852 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1855 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
1857 (test_vector_ops_duplicate): Add test for a scalar subreg of a
1858 VEC_MERGE of a VEC_DUPLICATE.
1860 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1862 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
1863 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
1864 * doc/standards.texi (C Language): Document C2X.
1865 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
1866 (rl78_option_override): Handle "GNU C2X" language name.
1868 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1870 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
1871 Document C17 as published in 2018.
1873 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
1876 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
1877 bail out if both sides do not have the same storage order.
1879 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1881 * bitmap.c (bitmap_head::dump): New.
1882 * bitmap.h (bitmap_head): Add dump().
1883 * gimple-ssa-evrp-analyze.c
1884 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
1885 (evrp_range_analyzer::set_ssa_range_info): Same.
1886 (evrp_range_analyzer::record_ranges_from_phis): Same.
1887 (evrp_range_analyzer::record_ranges_from_stmt): Same.
1888 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
1889 * gimple-ssa-sprintf.c (get_int_range): Same.
1890 (format_integer): Same.
1891 (sprintf_dom_walker::handle_gimple_call): Same.
1892 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
1893 (ipcp_vr_lattice::top_p): Same.
1894 (ipcp_vr_lattice::bottom_p): Same.
1895 (ipcp_vr_lattice::set_to_bottom): Same.
1896 (ipa_vr_operation_and_type_effects): Same.
1897 (propagate_vr_across_jump_function): Same.
1898 (ipcp_store_vr_results): Same.
1899 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
1900 (ipa_print_node_jump_functions_for_edge): Same.
1901 (ipa_get_value_range): Same.
1902 (ipa_compute_jump_functions_for_edge): Same.
1903 (ipa_write_jump_function): Same.
1904 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
1905 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1907 * vr-values.c (set_value_range_to_nonnegative): Same.
1908 (set_value_range_to_truthvalue): Same.
1909 (vr_values::get_value_range): Same.
1910 (vr_values::set_defs_to_varying): Same.
1911 (vr_values::update_value_range): Same.
1912 (symbolic_range_based_on_p): Same.
1913 (vr_values::op_with_boolean_value_range_p): Same.
1914 (vr_values::extract_range_for_var_from_comparison_expr): Same.
1915 (vr_values::extract_range_from_ssa_name): Same.
1916 (vr_values::extract_range_from_binary_expr): Same.
1917 (vr_values::extract_range_from_unary_expr): Same.
1918 (vr_values::extract_range_from_cond_expr): Same.
1919 (vr_values::extract_range_from_comparison): Same.
1920 (vr_values::check_for_binary_op_overflow): Same.
1921 (vr_values::extract_range_basic): Same.
1922 (vr_values::extract_range_from_assignment): Same.
1923 (compare_ranges): Same.
1924 (compare_range_with_value): Same.
1925 (vr_values::adjust_range_with_scev): Same.
1926 (vrp_valueize): Same.
1927 (vrp_valueize_1): Same.
1928 (vr_values::get_vr_for_comparison): Same.
1929 (vr_values::compare_name_with_value): Same.
1930 (vr_values::compare_names): Same.
1931 (vr_values::vrp_evaluate_conditional): Same.
1932 (find_case_label_ranges): Same.
1933 (vr_values::vrp_visit_switch_stmt): Same.
1934 (vr_values::extract_range_from_phi_node): Same.
1935 (vr_values::simplify_div_or_mod_using_ranges): Same.
1936 (vr_values::simplify_bit_ops_using_ranges): Same.
1937 (test_for_singularity): Same.
1938 (range_fits_type_p): Same.
1939 (vr_values::simplify_cond_using_ranges_1): Same.
1940 (vr_values::simplify_switch_using_ranges): Same.
1941 (vr_values::simplify_float_conversion_using_ranges): Same.
1942 (vr_values::two_valued_val_range_p): Same.
1943 (vr_values::add_equivalence): Move to value_range::equiv_add.
1944 * vr-values.h (vr_values::add_equivalence): Remove.
1945 (VR_INITIALIZER): Remove.
1946 * tree-vrp.c (value_range::set): New.
1947 (value_range::equiv_add): New.
1948 (value_range::value_range): New.
1949 (value_range::deep_copy): New.
1950 (value_range::check): New.
1951 (value_range::equal_p): New.
1952 (value_range::ignore_equivs_equal_p): New.
1953 (value_range::operator==): New.
1954 (value_range::operator!=): New.
1955 (value_range::symbolic_p): New.
1956 (value_range::numeric_p): New.
1957 (value_range::set_undefined): New.
1958 (value_range::set_varying): New.
1959 (value_range::may_contain_p): New.
1960 (value_range::equiv_clear): New.
1961 (value_range::singleton_p): New.
1962 (value_range::intersect): New.
1963 (value_range::dump): New.
1964 (value_range::set_and_canonicalize): New.
1965 (set_value_range): Adjust for value_range API.
1966 (set_value_range_to_undefined): Same.
1967 (set_value_range_to_varying): Same.
1968 (set_and_canonicalize_value_range): Same.
1969 (set_value_range_to_nonnull): Same.
1970 (set_value_range_to_null): Same.
1971 (range_is_null): Same.
1972 (range_is_nonnull): Same.
1973 (range_int_cst_p): Same.
1974 (range_int_cst_singleton_p): Same.
1975 (symbolic_range_p): Same.
1976 (range_includes_zero_p): Same.
1977 (value_range_constant_singleton): Same.
1978 (vrp_set_zero_nonzero_bits): Same.
1979 (ranges_from_anti_range): Same.
1980 (extract_range_into_wide_ints): Same.
1981 (extract_range_from_multiplicative_op): Same.
1982 (set_value_range_with_overflow): Same.
1983 (extract_range_from_binary_expr_1): Same.
1984 (extract_range_from_unary_expr): Same.
1985 (dump_value_range): Same.
1986 (debug_value_range): Same.
1987 (vrp_prop::check_array_ref): Same.
1988 (vrp_prop::check_mem_ref): Same.
1989 (vrp_prop::vrp_initialize): Same.
1990 (vrp_prop::visit_stmt): Same.
1991 (intersect_ranges): Same.
1992 (vrp_prop::visit_phi): Same.
1993 (vrp_prop::vrp_finalize): Same.
1994 (determine_value_range_1): Same.
1995 (determine_value_range): Same.
1996 (vrp_intersect_ranges_1): Rename to...
1997 (vrp_intersect_1): this.
1998 (vrp_intersect_ranges): Rename to...
1999 (value_range::intersect_helper): ...this.
2000 (vrp_meet_1): Rename to...
2001 (value_range::union_helper): ...this.
2002 (vrp_meet): Rename to...
2003 (value_range::union_): ...this.
2004 (copy_value_range): Remove.
2005 * tree-vrp.h (struct value_range): Rewrite into a proper class.
2006 (value_range::vrtype): New.
2007 (value_range::type): New.
2008 (value_range::equiv): New.
2009 (value_range::min): New.
2010 (value_range::max): New.
2011 (value_range::varying_p): New.
2012 (value_range::undefined_p): New.
2013 (value_range::null_p): New.
2014 (value_range::equiv_add): New.
2015 (copy_value_range): Remove.
2017 2018-10-17 David Malcolm <dmalcolm@redhat.com>
2019 * Makefile.in (SELFTEST_TARGETS): New.
2020 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
2021 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
2022 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
2024 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
2025 (selftest-c++-gdb, selftest-c++-valgrind): Move to
2027 * configure: Regenerate.
2028 * configure.ac (selftest_languages): New.
2030 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
2032 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
2033 overflow wraps argument.
2034 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
2035 wide_int_range_multiplicative_op.
2036 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
2037 overflow wraps argument.
2038 (wide_int_range_multiplicative_op): Same.
2039 (wide_int_range_lshift): Same.
2040 (wide_int_range_div): Same.
2041 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
2042 (wide_int_range_lshift): Same.
2043 (wide_int_range_div): Same.
2045 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
2047 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
2048 use sign as argument.
2049 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
2050 wide_int_range_shift_undefined_p.
2052 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2054 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
2056 (@despeculate_copy<ALLI_TI:mode>): ... This.
2057 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
2060 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2062 * config.gcc: Obsolete *-*-solaris2.10*.
2063 * doc/install.texi (Specific, *-*-solaris2*): Document it.
2065 2018-10-12 Jeff Law <law@redhat.com>
2067 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
2068 reg + sym +- const_int addressing modes.
2070 2018-10-15 David Malcolm <dmalcolm@redhat.com>
2072 * common.opt (fdiagnostics-minimum-margin-width=): New option.
2073 * diagnostic-show-locus.c (layout::layout): Apply the minimum
2075 (layout::start_annotation_line): Only print up to 3 of the
2076 margin character, to avoid touching the left-hand side.
2077 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
2078 minimum margin width, as set by test_diagnostic_context's ctor.
2079 (selftest::test_fixit_insert_containing_newline): Likewise.
2080 (selftest::test_fixit_insert_containing_newline_2): Likewise.
2081 (selftest::test_line_numbers_multiline_range): Clear
2082 dc.min_margin_width.
2083 * diagnostic.c (diagnostic_initialize): Initialize
2085 * diagnostic.h (struct diagnostic_context): Add field
2087 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
2088 * opts.c (common_handle_option): Handle
2089 OPT_fdiagnostics_minimum_margin_width_.
2090 * selftest-diagnostic.c
2091 (selftest::test_diagnostic_context::test_diagnostic_context):
2092 Initialize min_margin_width to 6.
2093 * toplev.c (general_init): Initialize global_dc->min_margin_width.
2095 2018-10-15 David Malcolm <dmalcolm@redhat.com>
2097 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
2098 Fix usage of "error_at_rich_loc" in the comment.
2100 2018-10-15 Renlin Li <renlin.li@arm.com>
2103 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
2104 if-conversioned loop when it contains ifn with types not
2105 supported by backend.
2106 * internal-fn.c (expand_direct_optab_fn): Add an assert.
2107 (direct_internal_fn_supported_p): New helper function.
2108 * internal-fn.h (direct_internal_fn_supported_p): Declare.
2110 2018-10-15 Jakub Jelinek <jakub@redhat.com>
2113 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
2114 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
2115 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
2117 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
2119 PR tree-optimization/87022
2120 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
2121 bits in dist vector rather than the first one.
2123 2018-10-15 Richard Biener <rguenther@suse.de>
2126 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
2127 (visit_loadstore): When a used restrict tag escaped verify that
2128 the points-to solution of "other" pointers do not include
2130 (compute_dependence_clique): If a used restrict tag escaped
2131 communicated that down to visit_loadstore.
2133 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
2135 * config/s390/s390.c (s390_expand_vec_init): Force vector element
2136 into reg if it isn't a general operand.
2138 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
2141 * config/i386/sse.md (*vec_dupv2di): Add register source to
2144 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
2147 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
2148 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
2149 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
2150 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
2152 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
2154 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
2155 (notice_args_size): Set it in the current trace if no insn that can
2156 throw internally has been seen yet.
2157 (connect_traces): When connecting args_size between traces, allow the
2158 incoming values not to match if there is an insn setting it before the
2159 first insn that can throw internally; in that case, force the creation
2160 of a CFI note on this latter insn.
2162 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
2164 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
2165 the base class, because using the injected-class-name was not clearly
2166 specified until DR 176.
2168 2018-10-12 Paul Koning <ni1d@arrl.net>
2170 * config/pdp11/pdp11.md (doloop_end): New expander.
2171 (doloop_end_insn): renamed from "doloop_end".
2172 (addqi3): New pattern.
2173 (subqi3): New pattern.
2174 * config/pdp11/predicates.md (incdec_operand): New predicate.
2176 2018-10-12 Yury Gribov <tetra2005@gmail.com>
2179 * real.c (format_helper::can_represent_integral_type_p): New function
2180 * real.h (format_helper::can_represent_integral_type_p): Ditto.
2181 * match.pd: New pattern.
2183 2018-10-12 Alexandre Oliva <oliva@adacore.com>
2185 * configure.ac: Introduce --enable-large-address-aware
2186 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
2187 * doc/install.texi: Document it.
2188 * configure, config.in: Rebuilt.
2189 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
2190 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
2191 (LINK_SPEC): Insert it.
2192 * config/i386/mingw-w64.h: Likewise.
2194 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
2196 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
2198 PR rtl-optimization/87600
2199 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
2201 2018-10-12 Paul Koning <ni1d@arrl.net>
2203 * doc/md.texi (doloop_end): Document that the pattern code may
2204 need to check operand mode.
2206 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
2208 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
2209 to zero-extend between int and floating-point registers.
2210 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
2211 ldp into floating-point registers. Add type and arch attributes.
2212 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
2213 Use f_loads for type attribute.
2215 2018-10-11 Martin Sebor <msebor@redhat.com>
2217 * doc/extend.texi (attribute packed): Correct typos.
2219 2018-10-11 Martin Sebor <msebor@redhat.com>
2221 * doc/extend.texi (attribute flatten): Mention interaction with
2224 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
2227 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
2228 Set new_decl virtual flag to zero.
2230 2018-10-11 Martin Sebor <msebor@redhat.com>
2233 * doc/extend.texi (attribute format_arg): Discuss using multiple
2234 attributes on a single function.
2236 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
2238 PR tree-optimization/86829
2239 * match.pd (sin (atan (x))): New simplification rules.
2240 (cos (atan (x))): Likewise.
2241 * real.c (build_sinatan_real): New function.
2242 * real.h (build_sinatan_real): Prototype.
2244 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2246 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
2248 (fold_mergeeo_helper): New helper function.
2249 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
2250 intrinsics. Correct some whitespace indentation issues.
2252 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
2255 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
2256 Use HOST_WIDE_INT_1U for shift.
2258 2018-10-11 Doug Rupp <rupp@adacore.com>
2259 Olivier Hainque <hainque@adacore.com>
2261 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
2262 Pass --relax to the linker for RTPs.
2263 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
2265 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
2266 Jan Hubicka <jh@suse.cz>
2267 Martin Jambor <mjambor@suse.cz>
2269 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
2270 the same elements are repeated rather than printing all of them.
2271 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
2272 "repeated" elements.
2273 * read-rtl-function.c (test_loading_repeat): New function.
2274 (read_rtl_function_c_tests): Call test_loading_repeat.
2275 * rtl-tests.c (test_dumping_repeat): New function.
2276 (rtl_tests_c_tests): Call test_dumping_repeat.
2278 2018-10-11 Richard Biener <rguenther@suse.de>
2280 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
2281 bdver?_cost): Unify to ...
2282 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
2283 * config/i386/i386.c (processor_cost_table): Adjust.
2285 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
2288 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
2289 the thunk when expanding to GIMPLE.
2291 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2293 * varasm.c (mergeable_string_section): Don't try to move zero-length
2294 strings to the merge section.
2296 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
2299 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
2301 2018-10-10 Jakub Jelinek <jakub@redhat.com>
2304 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
2305 to special_args set.
2307 2018-10-10 Richard Biener <rguenther@suse.de>
2309 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
2310 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
2311 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
2312 and recursing and pattern terminating the recursion on SSE
2313 vector width using ix86_expand_reduc.
2314 (reduc_sminmax_scal_<mode>): Split into part reducing to half
2315 width and recursing and SSE2 vector variant doing the final
2316 reduction with ix86_expand_reduc.
2317 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
2318 with terminating the recursion at AVX level, splitting that
2321 2018-10-09 David Malcolm <dmalcolm@redhat.com>
2323 * genmatch.c (error_cb): Rename to...
2324 (diagnostic_cb): ...this, converting int params to enums.
2325 (fatal_at): Update for renaming.
2326 (warning_at): Likewise.
2328 * input.c (selftest::ebcdic_execution_charset::apply):
2329 Update for renaming of...
2330 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
2332 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
2333 converting level and reason to enums.
2334 (class selftest::lexer_error_sink): Rename to...
2335 (class selftest::lexer_test_options): ...this, renaming field
2336 "m_errors" to "m_diagnostics".
2337 (selftest::lexer_test_options::apply): Update for renaming of...
2338 (selftest::lexer_test_options::on_error): ...this, renaming to...
2339 (selftest::lexer_test_options::on_diagnostic): ...this
2340 converting level and reason to enums.
2341 (selftest::test_lexer_string_locations_raw_string_unterminated):
2342 Update for renamings.
2343 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
2346 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
2348 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
2349 * config/rs6000/pmmintrin.h: New file.
2351 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
2353 PR tree-optimization/86659
2354 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
2356 2018-10-09 Richard Biener <rguenther@suse.de>
2358 PR tree-optimization/63155
2359 * tree-ssa-structalias.c: Include tree-ssa.h.
2360 (get_constraint_for_ssa_var): For undefs return nothing_id.
2361 (find_func_aliases): Cleanup PHI handling.
2363 2018-10-09 Richard Biener <rguenther@suse.de>
2365 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
2368 2018-10-09 Martin Liska <mliska@suse.cz>
2370 * asan.c (asan_emit_stack_protection): If a stack variable
2371 is located in a same file as current function, then emit
2372 line info into variable definition string.
2374 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2376 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
2379 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2381 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
2384 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2386 PR tree-optimization/86659
2387 * gimple-match.h (struct gimple_match_op): Add reverse field.
2388 (gimple_match_op::set_op): New overloaded method.
2389 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
2390 the REF_REVERSE_STORAGE_ORDER flag on the value.
2391 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
2392 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
2394 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
2397 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
2398 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
2400 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
2403 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
2404 Defined with __builtin_ia32_vfmaddsubpd512_mask.
2406 2018-10-08 Richard Biener <rguenther@suse.de>
2408 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
2409 cost the same as AVX128 ones.
2411 2018-10-08 Paul Koning <ni1d@arrl.net>
2413 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
2414 (expand_block_move): New function.
2415 * config/pdp11/pdp11.c (output_block_move): Remove.
2416 (expand_block_move): New function.
2417 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
2418 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
2419 (*movmemhi1): Remove.
2421 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
2423 * config/s390/2827.md: Increase latencies for some FP instructions.
2425 2018-10-08 Richard Biener <rguenther@suse.de>
2427 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
2429 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
2430 * tree-vectorizer.h (dump_stmt_cost): Adjust.
2431 (add_stmt_cost): Dump return value of the hook.
2433 2018-10-08 Richard Biener <rguenther@suse.de>
2435 PR tree-optimization/63155
2436 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
2437 (ssa_propagation_engine::ssa_propagate): Remove redundant
2438 bitmap bit clearing.
2440 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
2442 PR rtl-optimization/86939
2443 PR rtl-optimization/87479
2444 * ira.h (non_conflicting_reg_copy_p): New prototype.
2445 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
2446 (make_hard_regno_dead): Don't add conflicts for register
2447 ignore_reg_for_conflicts.
2448 (make_object_dead): Likewise.
2449 (non_conflicting_reg_copy_p): New function.
2450 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
2451 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
2452 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
2453 (make_hard_regno_dead): Don't add conflicts for register
2454 ignore_reg_for_conflicts. Remove special conflict handling of
2455 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
2456 check_pic_pseudo_p and update callers.
2457 (mark_pseudo_dead): Don't add conflicts for register
2458 ignore_reg_for_conflicts.
2459 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
2461 2018-10-05 Andrew Waterman <andrew@sifive.com>
2462 Jim Wilson <jimw@sifive.com>
2464 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
2465 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
2466 new pattern using HONOR_SNANS that emits one extra instruction.
2468 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2470 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
2471 patterns): Merge SI and DI patterns to a GPR pattern.
2472 (unnamed define_insn and define_split for record form of that): Merge
2473 to a single define_insn_and_split pattern.
2475 2018-10-05 David Malcolm <dmalcolm@redhat.com>
2478 * input.c (expand_location_to_spelling_point): Add param "aspect"
2479 and use rather than hardcoding LOCATION_ASPECT_CARET.
2480 (get_substring_ranges_for_loc): Handle the case of a single token
2481 within a macro expansion.
2482 * input.h (expand_location_to_spelling_point): Add "aspect" param,
2483 defaulting to LOCATION_ASPECT_CARET.
2485 2018-10-05 Paul Koning <ni1d@arrl.net>
2487 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
2488 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
2489 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2490 (pdp11_guard_type): New function.
2492 2018-10-05 Paul Koning <ni1d@arrl.net>
2494 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
2495 * config/pdp11/pdp11.opt (mfloat32): Remove.
2497 * doc/invoke.texi (pdp11 -mfloat32): Remove:
2498 (pdp11 -mfloat64): Remove.
2500 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2502 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
2503 (*cmp<mode>_cc_i387): Ditto.
2504 (*cmpu<mode>_cc_i387): Ditto.
2505 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2506 * config/i386/i386.c (ix86_expand_fp_compare): Remove
2508 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
2509 Emit x86_sahf_1 pattern.
2510 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
2511 (ix86_expand_carry_flag_compare): Ditto.
2513 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2515 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
2516 to reg_or_0_operand. Add "C" constraint.
2517 (*cmpxf_cc_i387): Ditto.
2518 (*cmp<mode>_i387): Change operand 2 predicate
2519 to nonimm_or_0_operand. Add "C" constraint.
2520 (*cmp<mode>_cc_i387): Ditto.
2521 (*cmp<mode>_0_i387): Remove insn pattern.
2522 (*cmp<mode>_0_cc_i387): Ditto.
2524 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2526 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
2527 * config/i386/predicates.md (nonimm_or_0_operand): Rename
2528 from vector_move_operand. Update all uses.
2530 2018-10-05 Martin Sebor <msebor@redhat.com>
2532 PR tree-optimization/87490
2533 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
2536 2018-10-05 Richard Biener <rguenther@suse.de>
2538 PR tree-optimization/63155
2539 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
2540 vertical space in dumpfiles.
2541 * tree-ssa-propagate.h
2542 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
2543 * tree-ssa-propagate.c (cfg_blocks_back): New global.
2544 (ssa_edge_worklist_back): Likewise.
2545 (curr_order): Likewise.
2546 (cfg_blocks_get): Remove abstraction.
2547 (cfg_blocks_add): Likewise.
2548 (cfg_blocks_empty_p): Likewise.
2549 (add_ssa_edge): Add to current or next worklist based on
2551 (add_control_edge): Likewise.
2552 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
2554 (ssa_propagation_engine::ssa_propagate): ... here. Unify
2555 iteration from CFG and SSA edge worklist so we process
2556 everything in RPO order, prioritizing forward progress
2558 (ssa_prop_init): Allocate new worklists, do not dump
2560 (ssa_prop_fini): Free new worklists.
2562 2018-10-05 Richard Biener <rguenther@suse.de>
2564 * tree-core.h (tree_block::abstract_flag): Remove.
2565 (tree_block::block_num): Make full 32bits.
2566 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
2567 * tree.h (BLOCK_ABSTRACT): Remove.
2568 * dwarf2out.c (gen_lexical_block_die): Remove dead code
2569 resulting from BLOCK_ABSTRACT being always false.
2570 (gen_inlined_subroutine_die): Likewise.
2571 (gen_block_die): Likewise.
2572 * tree.c (block_ultimate_origin): Likewise.
2573 * tree-pretty-print.c (dump_block_node): Remove code dealing
2574 with BLOCK_ABSTRACT.
2575 * tree-ssa-live.c (dump_scope_block): Likewise.
2576 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
2577 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
2579 2018-10-05 Richard Biener <rguenther@suse.de>
2581 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
2582 is asked for initialize mode to the component mode of the
2585 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
2588 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
2589 assembler for -mavx.
2590 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
2592 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2595 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
2597 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
2598 to be DImode. When using mffscrn, force the operand to a register.
2600 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
2602 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
2603 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
2604 X87MODEF mode iterator.
2605 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
2606 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
2607 X87MODEF mode iterator.
2609 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
2611 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
2612 -Wno-prio-ctor-dtor.
2614 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2616 * Makefile.in (OBJS): Add opt-problem.o.
2617 * dump-context.h: Include "selftest.h.
2618 (selftest::temp_dump_context): New forward decl.
2619 (class dump_context): Make friend of class
2620 selftest::temp_dump_context.
2621 (dump_context::dump_loc_immediate): New decl.
2622 (class dump_pretty_printer): Move here from dumpfile.c.
2623 (class temp_dump_context): Move to namespace selftest.
2624 (temp_dump_context::temp_dump_context): Add param
2625 "forcibly_enable_dumping".
2626 (selftest::verify_dumped_text):
2627 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
2628 (selftest::verify_item):
2629 (ASSERT_IS_TEXT): Move here from dumpfile.c.
2630 (ASSERT_IS_TREE): Likewise.
2631 (ASSERT_IS_GIMPLE): Likewise.
2632 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
2634 (dump_context::dump_loc_immediate): ...this new function.
2635 (class dump_pretty_printer): Move to dump-context.h.
2636 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
2637 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
2638 (temp_dump_context::temp_dump_context): Move to "selftest"
2639 namespace. Add param "forcibly_enable_dumping", and use it to
2640 conditionalize the use of m_pp;
2641 (selftest::verify_dumped_text): Make non-static.
2642 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
2643 (selftest::verify_item): Make non-static.
2644 (ASSERT_IS_TEXT): Move to dump-context.h.
2645 (ASSERT_IS_TREE): Likewise.
2646 (ASSERT_IS_GIMPLE): Likewise.
2647 (selftest::test_capture_of_dump_calls): Pass "true" for new
2648 param of temp_dump_context.
2649 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
2650 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
2652 * opt-problem.cc: New file.
2653 * opt-problem.h: New file.
2654 * optinfo-emit-json.cc
2655 (selftest::test_building_json_from_dump_calls): Pass "true" for
2656 new param of temp_dump_context.
2657 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
2658 (optinfo::emit_for_opt_problem): New function.
2659 (optinfo::emit): Clarity which emit_item is used.
2660 * optinfo.h (optinfo::get_dump_location): New accessor.
2661 (optinfo::emit_for_opt_problem): New decl.
2662 (optinfo::emit): Make const.
2663 * selftest-run-tests.c (selftest::run_tests): Call
2664 selftest::opt_problem_cc_tests.
2665 * selftest.h (selftest::opt_problem_cc_tests): New decl.
2666 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
2667 bool to opt_result, converting fprintf messages to
2668 opt_result::failure_at calls. Add "stmt" param for use by the
2670 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
2671 (runtime_alias_check_p): Convert return type from bool to
2672 opt_result, converting dump_printf calls to
2673 opt_result::failure_at, using the statement DDR_A for their
2675 (find_data_references_in_stmt): Convert return type from bool to
2676 opt_result, converting "return false" to opt_result::failure_at
2678 * tree-data-ref.h: Include "opt-problem.h".
2679 (dr_analyze_innermost): Convert return type from bool to opt_result,
2680 and add a const gimple * param.
2681 (find_data_references_in_stmt): Convert return type from bool to
2683 (runtime_alias_check_p): Likewise.
2684 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
2685 dr_analyze_innermost.
2686 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
2687 Convert return type from bool to opt_result, adding a message for
2688 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
2689 (vect_analyze_data_ref_dependence): Convert return type from bool
2690 to opt_result. Change sense of return type from "false"
2691 effectively meaning "no problems" to "false" meaning a problem,
2692 so that "return false" becomes "return opt_result::success".
2693 Convert "return true" calls to opt_result::failure_at, using
2694 the location of statement A rather than vect_location.
2695 (vect_analyze_data_ref_dependences): Convert return type from bool
2697 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
2698 calls to opt_result::failure_at, using the stmt location rather
2700 (vect_verify_datarefs_alignment): Convert return type from bool
2702 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
2703 into multiple more-tightly-scoped copies.
2704 (vect_analyze_data_refs_alignment): Convert return type from bool
2706 (vect_analyze_data_ref_accesses): Likewise, converting a
2707 "return false" to a "return opt_result::failure_at", adding a
2709 (vect_prune_runtime_alias_test_list): Convert return type from
2710 bool to opt_result, converting dump_printf_loc to
2711 opt_result::failure_at. Add a %G to show the pertinent statement,
2712 and use the stmt's location rather than vect_location.
2713 (vect_find_stmt_data_reference): Convert return type from
2714 bool to opt_result, converting dump_printf_loc to
2715 opt_result::failure_at, using stmt's location.
2716 (vect_analyze_data_refs): Convert return type from bool to
2717 opt_result. Convert "return false" to "return
2718 opt_result::failure_at", adding messages as needed.
2719 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
2720 type from bool to opt_result.
2721 (vect_determine_vf_for_stmt): Likewise.
2722 (vect_determine_vectorization_factor): Likewise, converting
2723 dump_printf_loc to opt_result::failure_at, using location of phi
2724 rather than vect_location.
2725 (vect_analyze_loop_form_1): Convert return type from bool to
2726 opt_result, converting dump_printf_loc calls, retaining the use of
2728 (vect_analyze_loop_form): Convert return type from loop_vec_info
2729 to opt_loop_vec_info.
2730 (vect_analyze_loop_operations): Convert return type from bool to
2731 opt_result, converting dump_printf_loc calls, using the location
2732 of phi/stmt rather than vect_location where available. Convert
2733 various "return false" to "return opt_result::failure_at" with
2734 "unsupported phi" messages.
2735 (vect_get_datarefs_in_loop): Convert return type from bool to
2736 opt_result. Add a message for the
2737 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
2738 (vect_analyze_loop_2): Convert return type from bool to
2739 opt_result. Ensure "ok" is set to a opt_result::failure_at before
2740 each "goto again;", adding new messages where needed.
2741 Add "unsupported grouped {store|load}" messages.
2742 (vect_analyze_loop): Convert return type from loop_vec_info to
2744 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
2746 * tree-vect-stmts.c (process_use): Likewise, converting
2747 dump_printf_loc call and using stmt location, rather than
2749 (vect_mark_stmts_to_be_vectorized): Likeise.
2750 (vect_analyze_stmt): Likewise, adding a %G.
2751 (vect_get_vector_types_for_stmt): Convert return type from bool to
2752 opt_result, converting dump_printf_loc calls and using stmt
2753 location, rather than vect_location.
2754 (vect_get_mask_type_for_stmt): Convert return type from tree to
2755 opt_tree, converting dump_printf_loc calls and using stmt location.
2756 * tree-vectorizer.c: Include "opt-problem.h.
2757 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
2758 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
2759 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
2760 enabled, use it to report at the top level "couldn't vectorize
2761 loop" followed by the problem.
2762 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
2763 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
2765 (vect_analyze_stmt): Likewise.
2766 (vect_get_vector_types_for_stmt): Likewise.
2767 (tree vect_get_mask_type_for_stmt): Likewise.
2768 (vect_analyze_data_ref_dependences): Likewise.
2769 (vect_enhance_data_refs_alignment): Likewise.
2770 (vect_analyze_data_refs_alignment): Likewise.
2771 (vect_verify_datarefs_alignment): Likewise.
2772 (vect_analyze_data_ref_accesses): Likewise.
2773 (vect_prune_runtime_alias_test_list): Likewise.
2774 (vect_find_stmt_data_reference): Likewise.
2775 (vect_analyze_data_refs): Likewise.
2776 (vect_analyze_loop): Convert return type from loop_vec_info to
2778 (vect_analyze_loop_form): Likewise.
2779 (vect_analyze_slp): Convert return type from bool to opt_result.
2781 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2783 * doc/invoke.texi (-fopt-info): Document new "internals"
2785 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
2786 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
2788 (optinfo_verbosity_options): Add "internals".
2789 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
2790 (dump_context::apply_dump_filter_p): New member function.
2791 (dump_context::dump_loc): Use apply_dump_filter_p rather than
2792 explicitly masking the dump_kind.
2793 (dump_context::begin_scope): Increment the scope depth first. Use
2794 apply_dump_filter_p rather than explicitly masking the dump_kind.
2795 (dump_context::emit_item): Use apply_dump_filter_p rather than
2796 explicitly masking the dump_kind.
2797 (dump_dec): Likewise.
2798 (dump_hex): Likewise.
2799 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
2800 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
2801 (opt_info_switch_p): Update handling of default
2802 MSG_OPTIMIZED_LOCATIONS to cope with default of
2803 MSG_PRIORITY_USER_FACING.
2804 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
2805 masking the dump_kind.
2806 (selftest::test_capture_of_dump_calls): Update test_dump_context
2807 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
2808 than MSG_ALL. Generalize scope test to be run at all four
2809 combinations of with/without MSG_PRIORITY_USER_FACING and
2810 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
2811 for each of the two values.
2812 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
2813 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
2814 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
2815 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
2816 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
2817 with MSG_PRIORITY_*.
2818 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
2819 dump messages as MSG_PRIORITY_USER_FACING.
2820 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
2821 about the interaction with MSG_PRIORITY_*.
2823 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2825 * varasm.c (output_constant): Add new parameter merge_strings.
2826 Make strings properly zero terminated in merge string sections.
2827 (mergeable_string_section): Don't fail if the last char is non-zero.
2828 (assemble_variable_contents): Handle merge string sections.
2829 (assemble_variable): Likewise.
2830 (assemble_constant_contents): Likewise.
2831 (output_constant_def_contents): Likewise.
2832 (output_constructor_array_range,
2833 output_constructor_regular_field): Adjust call to output_constant.
2834 (output_object_block): Adjust call to assemble_constant_contents
2835 and assemble_variable_contents.
2837 2018-10-04 Martin Liska <mliska@suse.cz>
2840 * cgraphunit.c (process_function_and_variable_attributes):
2841 Warn about a function with alias attribute and a body.
2843 2018-10-04 Martin Liska <mliska@suse.cz>
2846 * multiple_target.c (redirect_to_specific_clone): New function.
2847 (ipa_target_clone): Use it.
2848 * tree-inline.c: Fix comment.
2850 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2852 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
2854 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
2855 (gcc::dump_manager::register_pass): New member function, adapted
2856 from loop body in gcc::pass_manager::register_pass, adding a
2857 call to update_dfi_for_opt_info.
2858 (gcc::dump_manager::opt_info_enable_passes): Store the
2859 -fopt-info options into the new fields. Move the loop
2861 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
2863 * dumpfile.h (struct opt_pass): New forward decl.
2864 (gcc::dump_manager::register_pass): New decl.
2865 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
2866 (class gcc::dump_manager): Add fields "m_optgroup_flags",
2867 "m_optinfo_flags", and "m_optinfo_filename".
2868 * passes.c (gcc::pass_manager::register_pass): Move all of the
2869 dump-handling code to gcc::dump_manager::register_pass.
2871 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
2873 PR rtl-optimization/87466
2874 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
2875 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
2876 * doc/tm.texi: Regenerate.
2877 * ira-lives.c (process_bb_node_lives): Use the new target hook.
2878 * lra-lives.c (process_bb_lives): Likewise.
2879 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
2882 2018-10-04 Tamar Christina <tamar.christina@arm.com>
2884 * params.c (add_params): Fix initialization.
2886 2018-10-04 Martin Liska <mliska@suse.cz>
2888 PR gcov-profile/84107
2889 * tree-profile.c (init_ic_make_global_vars):
2890 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
2891 Come up with new ic_tuple* variables. Emit
2892 __gcov_indirect_call{,_topn} variables.
2893 (gimple_gen_ic_profiler): Access the variable
2895 (gimple_gen_ic_func_profiler): Access
2896 __gcov_indirect_call.callee field.
2897 (gimple_init_gcov_profiler): Use ptr_type_node.
2898 * value-prof.c (gimple_ic): Use ptr_type_node.
2900 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2902 PR tree-optimization/85787
2903 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
2904 into this function and add support for detecting multiple phis.
2905 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
2907 2018-10-04 Martin Liska <mliska@suse.cz>
2910 * ipa-inline.c (inline_to_all_callers_1):
2911 Call ultimate_alias_target for node being inlined.
2913 2018-10-03 Jeff Law <law@redhat.com>
2915 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
2917 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
2918 * tree.h (get_typenode_from_name): Prototype.
2920 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2922 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2923 Change operand 2 predicate to nonimmediate_operand.
2924 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2926 2018-10-03 Martin Sebor <msebor@redhat.com>
2927 Jeff Law <law@redhat.com>
2929 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
2931 (get_string_length): Detect unterminated arrays.
2932 (format_string): Same.
2933 (format_directive): Warn about unterminated arrays.
2934 (handle_gimple_call): Mark statements with no_warning as needed.
2936 2018-10-03 Jim Wilson <jimw@sifive.com>
2938 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
2939 also define __riscv_abi_rve. Delete trailing white space.
2941 2018-10-03 Paul Koning <ni1d@arrl.net>
2943 Enable LRA register allocator for PDP11.
2944 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
2947 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
2948 * config/pdp11/pdp11.opt (-mlra): New option.
2949 * doc/invoke.texi (PDP-11 Options): Document -mlra.
2951 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2953 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
2954 (*<absneg:code>extend<mode>xf2): Ditto.
2956 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2958 PR tree-optimization/87415
2959 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
2962 2018-10-02 Jeff Law <law@redhat.com>
2964 * gimple-fold.c (get_range_strlen): Only set *nonstr when
2965 an unterminated string is discovered. Bubble up range
2966 even for unterminated strings.
2967 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
2968 indicates the string was not terminated via NONSTR.
2970 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2972 * tree-vrp.c (extract_range_from_unary_expr): Special case all
2973 pointer conversions.
2974 Do not do anything special for anti-ranges.
2976 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
2978 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
2979 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
2981 2018-10-03 Martin Liska <mliska@suse.cz>
2983 PR gcov-profile/86109
2984 * coverage.c (coverage_begin_function): Do not
2985 mark lambdas as artificial.
2986 * tree-core.h (struct GTY): Remove tm_clone_flag
2987 and introduce new lambda_function.
2988 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
2990 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
2993 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
2994 P8_VECTOR and VSX are enabled.
2996 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2998 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
2999 0x3907 as CPU model number.
3001 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
3003 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
3004 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
3005 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
3006 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
3007 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
3008 * config/s390/s390.md: Likewise. Rename also the cpu attribute
3009 value from arch12 to z14.
3011 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
3013 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
3015 (isinf<mode>2): Ditto.
3017 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
3019 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
3020 before emitting fxam. Perform calculations in XFmode.
3022 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
3024 * match.pd (((X /[ex] A) +- B) * A): New transformation.
3026 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
3029 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
3030 * tree.c (signed_or_unsigned_type_for): Handle complex.
3032 2018-10-02 Jeff Law <law@redhat.com>
3034 * gimple-fold.c (get_range_strlen): Remove dead code.
3036 2018-10-02 Martin Sebor <msebor@redhat.com>
3037 Jeff Law <law@redhat.com>
3039 * builtins.c (unterminated_array): Add new arguments.
3040 If argument is not terminated, bubble up size and exact
3042 (expand_builtin_strnlen): Detect, avoid expanding
3043 and diagnose unterminated arrays.
3044 (c_strlen): Fill in offset of start of unterminated strings.
3045 * builtins.h (unterminated_array): Update prototype.
3047 2018-10-02 Richard Biener <rguenther@suse.de>
3049 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
3050 of haddv4df, first reduce to SSE width and exploit the fact
3051 that we only need element zero with the reduction result.
3052 (reduc_plus_scal_v2df): Likewise.
3054 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
3056 * dojump.h (do_jump): Delete.
3057 (do_jump_1): Likewise.
3058 (split_comparison): Move around.
3059 * dojump.c (do_jump): Make static.
3060 (do_jump_1): Likewise.
3061 (jumpifnot): Move around.
3062 (jumpifnot_1): Likewise.
3064 (jumpif_1): Likewise.
3065 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
3067 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
3069 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
3070 insns in the delay slot and add_insn_after for the jump insn.
3072 2018-10-02 Richard Biener <rguenther@suse.de>
3074 * tree-inline.c (expand_call_inline): Use the location of
3075 the callee declaration for the inline-entry marker.
3076 * final.c (notice_source_line): Remove special-casing of
3077 NOTE_INSN_INLINE_ENTRY.
3079 2018-10-01 Carl Love <cel@us.ibm.com>
3082 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
3083 (__builtin_mtfsb0): New.
3084 (__builtin_mtfsb1): New.
3085 ( __builtin_set_fpscr_rn): New.
3086 (__builtin_set_fpscr_drn): New.
3087 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
3088 (rs6000_expand_set_fpscr_rn_builtin): Add.
3089 (rs6000_expand_set_fpscr_drn_builtin): Add.
3090 (rs6000_expand_builtin): Add case statement entries for
3091 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
3092 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
3093 RS6000_BUILTIN_MFFSL.
3094 (rs6000_init_builtins): Add ftype initialization and def_builtin
3095 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
3096 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
3097 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
3098 rs6000_mffscdrn): Add define_insn.
3099 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
3100 * doc/extend.texi: Add documentation for the builtins.
3102 2018-10-01 Richard Biener <rguenther@suse.de>
3104 PR tree-optimization/87465
3105 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
3106 causing branch miscounts.
3108 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3110 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
3111 aarch64_option_default_param): New.
3112 (params.h): Include.
3113 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
3114 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
3115 stack-clash protection validation code.
3117 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3119 * params.c (validate_param): New.
3120 (add_params): Use it.
3121 (set_param_value): Refactor param validation into validate_param.
3122 (diagnostic.h): Include.
3123 * diagnostic.h (diagnostic_ready_p): New.
3125 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3127 * params.c (set_param_value):
3128 Add index of parameter being validated.
3129 * common/common-target.def (option_validate_param): New.
3130 * common/common-targhooks.h (default_option_validate_param): New.
3131 * common/common-targhooks.c (default_option_validate_param): New.
3132 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
3133 * doc/tm.texi: Regenerate.
3135 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3138 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3139 Add validation for stack-clash parameters and set defaults.
3141 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3144 * configure.ac: Add stack-clash-protection-guard-size.
3145 * doc/install.texi: Document it.
3146 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
3147 * params.def: Update comment for guard-size.
3148 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
3149 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
3150 * configure: Regenerate.
3152 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3155 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
3156 STACK_DYNAMIC_OFFSET): New.
3157 * config/aarch64/aarch64.c (aarch64_layout_frame):
3158 Update outgoing args size.
3159 (aarch64_stack_clash_protection_alloca_probe_range,
3160 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
3162 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3165 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
3167 * target.def (stack_clash_protection_alloca_probe_range): New.
3168 (stack_clash_protection_final_dynamic_probe): Remove.
3169 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
3170 (default_stack_clash_protection_final_dynamic_probe): Remove.
3171 * targhooks.c: Likewise.
3172 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
3173 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
3174 * doc/tm.texi: Regenerate.
3176 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3179 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
3180 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
3181 aarch64_clamp_to_uimm12_shift): New.
3182 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
3183 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
3185 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3188 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
3190 2018-10-01 Jeff Law <law@redhat.com>
3191 Richard Sandiford <richard.sandiford@linaro.org>
3192 Tamar Christina <tamar.christina@arm.com>
3195 * config/aarch64/aarch64.md
3196 (probe_stack_range): Add k (SP) constraint.
3197 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
3198 STACK_CLASH_MAX_UNROLL_PAGES): New.
3199 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
3200 stack probes for stack clash.
3201 (aarch64_allocate_and_probe_stack_space): New.
3202 (aarch64_expand_prologue): Use it.
3203 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
3204 (aarch64_sub_sp): Add emit_move_imm optional param.
3206 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
3208 PR tree-optimization/87261
3209 * match.pd: Remove trailing whitespace.
3210 Add (x & y) | ~(x | y) -> ~(x ^ y),
3211 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
3213 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
3215 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
3217 (ashlsi3_insn): Update instruction constraints.
3218 (ashrsi3_insn): Likewise.
3219 (rotrsi3): Likewise.
3220 (add_shift): Likewise.
3221 * config/arc/constraints.md (Csz): New 32 bit constraint. It
3222 avoids placing in the limm field small constants which, otherwise,
3223 could end into a small instruction.
3225 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
3227 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
3228 destination register is not odd-even.
3229 (umaddsidi4_split): Likewise.
3231 2018-10-01 Richard Biener <rguenther@suse.de>
3233 * tree-inline.c (expand_call_inline): Store origin of fn
3234 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
3235 * tree.c (block_ultimate_origin): Simplify and do some
3238 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
3240 * config/i386/mmx.md (EMMS): New int iterator.
3241 (emms): New int attribute.
3242 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
3243 EMMS int iterator. Explicitly declare clobbers.
3244 (mmx_emms): Remove expander.
3246 * config/i386/predicates.md (emms_operation): Remove predicate.
3247 (vzeroall_pattern): New predicate.
3248 (vzeroupper_pattern): Rename from vzeroupper_operation.
3249 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
3250 vzeroupper_pattern and vzeroall_pattern predicates.
3252 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
3254 PR rtl-optimization/86939
3255 * ira-lives.c (make_hard_regno_born): Rename from this...
3256 (make_hard_regno_live): ... to this. Remove update to conflict
3257 information. Update function comment.
3258 (make_hard_regno_dead): Add conflict information update. Update
3260 (make_object_born): Rename from this...
3261 (make_object_live): ... to this. Remove update to conflict information.
3262 Update function comment.
3263 (make_object_dead): Add conflict information update. Update function
3265 (mark_pseudo_regno_live): Call make_object_live.
3266 (mark_pseudo_regno_subword_live): Likewise.
3267 (mark_hard_reg_dead): Update function comment.
3268 (mark_hard_reg_live): Call make_hard_regno_live.
3269 (process_bb_node_lives): Likewise.
3270 * lra-lives.c (make_hard_regno_born): Rename from this...
3271 (make_hard_regno_live): ... to this. Remove update to conflict
3272 information. Remove now uneeded check_pic_pseudo_p argument.
3273 Update function comment.
3274 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
3275 to conflict information. Update function comment.
3276 (mark_pseudo_live): Remove update to conflict information. Update
3278 (mark_pseudo_dead): Add conflict information update.
3279 (mark_regno_live): Call make_hard_regno_live.
3280 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
3281 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
3283 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
3286 * config/i386/i386.c (construct_container): Use TImode for
3287 BLKmode values in 2 integer registers.
3289 2018-09-29 Jeff Law <law@redhat.com>
3291 * builtins.c (unterminated_array): Pass in c_strlen_data * to
3292 c_strlen rather than just a tree *.
3293 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
3294 Update recursive calls appropriately. If caller did not provide a
3295 suitable data pointer, create a local one. When a non-terminated
3296 string is discovered, bubble up information about the string via the
3297 c_strlen_data object.
3298 * builtins.h (c_strlen): Update prototype.
3299 (c_strlen_data): New structure.
3300 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
3301 For a type 2 call, if c_strlen indicates a non-terminated string
3302 use the length of the non-terminated string.
3303 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
3305 2018-09-29 Jakub Jelinek <jakub@redhat.com>
3308 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
3309 __m512d type for __A argument rather than __m512.
3311 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
3313 * match.pd (simple_comparison): Don't optimize if either operand is
3314 a function pointer when target needs function pointer canonicalization.
3316 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3318 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
3319 power5 .. power9 to remove indirection.
3320 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
3321 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
3322 ASM_CPU_476_SPEC): Delete.
3323 (ASM_CPU_SPEC): Adjust.
3324 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
3325 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
3327 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3329 * config.in: Delete HAVE_AS_DCI.
3330 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
3331 * config/rs6000/rs6000.h: Ditto.
3332 * configure.ac: Delete HAVE_AS_DCI.
3333 * configure: Regenerate.
3335 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3337 * config.in (HAVE_AS_LWSYNC): Delete.
3338 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
3339 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
3341 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
3342 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
3344 * configure.ac: Delete HAVE_AS_LWSYNC.
3345 * configure: Regenerate.
3347 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
3348 Pierre-Marie de Rodat <derodat@adacore.com>
3350 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
3351 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
3352 (cgraph_node::create_thunk): Add indirect_offset parameter.
3353 (thunk_adjust): Likewise.
3354 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
3355 and initialize the corresponding field with it.
3356 (cgraph_node::dump): Dump indirect_offset field.
3357 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
3358 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
3359 (thunk_adjust): Add indirect_offset parameter and deal with it.
3360 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
3361 pass it to thunk_adjust. Do not call the target hook if it's non-zero
3362 or if the thunk is external or local. Fix formatting. Do not chain
3363 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
3364 if any, in the GIMPLE representation.
3365 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
3366 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
3367 (input_node): Read indirect_offset field.
3368 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
3369 call to thunk_adjust.
3370 * tree-nested.c (struct nesting_info): Add thunk_p field.
3371 (create_nesting_tree): Set it.
3372 (convert_all_function_calls): Copy static chain from targets to thunks.
3373 (finalize_nesting_tree_1): Return early for thunks.
3374 (unnest_nesting_tree_1): Do not finalize thunks.
3375 (gimplify_all_functions): Do not gimplify thunks.
3377 2018-09-28 David Malcolm <dmalcolm@redhat.com>
3379 * opt-suggestions.c (option_proposer::build_option_suggestions):
3380 Release "option_values".
3382 2018-09-28 David Malcolm <dmalcolm@redhat.com>
3384 * coverage.c (get_coverage_counts): Convert problem-reporting dump
3385 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
3386 * dumpfile.c (kind_as_string): New function.
3387 (dump_loc): Rather than a hardcoded prefix of "note: ", use
3388 kind_as_string to vary the prefix based on dump_kind.
3389 (selftest::test_capture_of_dump_calls): Update for above.
3391 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
3393 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
3394 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
3396 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
3398 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
3399 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
3400 INVALID_REGNUM instead of FPSR_REG.
3401 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
3402 * config/i386/i386.md: Update comment of FP compares.
3403 (fldenv): Do not clobber FPSR_REG.
3405 2018-09-28 Richard Biener <rguenther@suse.de>
3407 * tree.h (BLOCK_ORIGIN): New.
3408 * omp-expand.c (grid_expand_target_grid_body): Assign
3409 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
3410 * tree-inline.c (remap_block): Likewise.
3411 * auto-profile.c (get_function_decl_from_block): Simplify
3412 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
3413 * langhooks.c (lhd_print_error_function): Likewise.
3414 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
3416 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
3417 * tree.c (block_nonartificial_location): Likewise.
3418 (block_ultimate_origin): Likewise.
3419 * tree-pretty-print.c (percent_K_format): Likewise. Remove
3420 no longer needed LTO case.
3422 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
3423 Jan Hubicka <jh@suse.cz>
3424 Martin Jambor <mjambor@suse.cz>
3426 * simplify-rtx.c (simplify_merge_mask): New function.
3427 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
3428 same masks are used in op1 or op2.
3429 (test_vec_merge): New function.
3430 (test_vector_ops): Call test_vec_merge.
3432 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
3434 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
3435 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
3436 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
3437 (cypress_costs): Set it.
3438 (supersparc_costs): Likewise.
3439 (hypersparc_costs): Likewise.
3440 (leon_cost): Likewise.
3441 (leon3_costs): Likewise.
3442 (sparclet_costs): Likewise.
3443 (ultrasparc_costs): Likewise.
3444 (ultrasparc_costs): Likewise.
3445 (niagara_costs): Likewise.
3446 (niagara2_costs): Likewise.
3447 (niagara3_costs): Likewise.
3448 (niagara4_costs): Likewise.
3449 (niagara7_costs): Likewise.
3450 (m8_costs): Likewise.
3451 (TARGET_CAN_FOLLOW_JUMP): Define.
3452 (pass_work_around_errata::gate): Minor tweak.
3453 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
3454 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
3455 Automaitcally clear MASK_FSMULD mask for V7 processors.
3456 (sparc_can_follow_jump): New static function.
3457 (output_ubranch): Deal with CROSSING_JUMP_P.
3458 (sparc_use_sched_lookahead): Rewrite using switch statement.
3459 (sparc_issue_rate): Reorder.
3460 (sparc_branch_cost): New function.
3462 2018-09-27 Martin Sebor <msebor@redhat.com>
3464 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
3465 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
3466 (int_fits_type_p): Same.
3468 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
3470 * config/i386/i386.md (FPCR_REG): Remove.
3471 (UNSPEC_FLDCW): Remove.
3472 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
3473 (x86_fldcw_1): Remove insn pattern.
3474 (fnstenv): Do not clobber FPCR_REG.
3476 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
3477 (CALL_USED_REGISTERS): Ditto.
3478 (REG_ALLOC_ORDER): Ditto.
3479 (REG_CLASS_CONTENTS): Ditto.
3480 (HI_REGISTER_NAMES): Ditto.
3481 (ADDITIONAL_REGISTER_NAMES): Use defines instead
3482 of numerical constants.
3483 * config/i386/i386.c (regclass_map): Remove fpsr register.
3484 (dbx_register_map): Ditto.
3485 (dbx64_register_map): Ditto.
3486 (svr4_dbx_register_map): Ditto.
3487 (print_reg): Do not handle FPCR_REG.
3489 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
3492 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
3493 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
3494 Delete, always treat as true.
3495 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
3496 Ditto. Simplify remaining code.
3497 * config/powerpcspe/powerpcspe.h: Ditto.
3498 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
3499 Simplify remaining code.
3500 (rs6000_expand_builtin): Ditto.
3501 * config/rs6000/rs6000.h: Ditto.
3502 * configure.ac: Ditto.
3503 * configure: Regenerate.
3505 2018-09-27 Martin Liska <mliska@suse.cz>
3507 * coverage.c (get_coverage_counts): Revert the formatting
3508 of missing profile opt info.
3510 2018-09-27 Richard Biener <rguenther@suse.de>
3514 * dwarf2out.c (set_block_origin_self): Do not mark outermost
3515 block as we do not output that.
3516 (gen_inlined_subroutine_die): Elide the originally outermost
3517 block, matching what we do for concrete instances.
3518 (decls_for_scope): Add parameter specifying whether to recurse
3521 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
3522 Tom de Vries <tom@codesourcery.com>
3526 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
3527 STORE_FLAG_VALUE == 1.
3529 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
3531 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
3532 constant definitions.
3533 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
3534 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
3535 ("speculation_barrier"): New expander definition.
3537 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
3539 PR gcov-profile/86957
3540 * common.opt: New warning option -Wmissing-profile.
3541 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
3542 * doc/invoke.texi: Document -Wmissing-profile.
3544 2018-09-26 Jim Wilson <jimw@sifive.com>
3546 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
3547 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
3549 2018-09-26 Martin Sebor <msebor@redhat.com>
3551 * tree.c (zerop): Change return type to bool.
3552 (integer_zerop, integer_onep, integer_each_onep): Same.
3553 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3554 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3555 (real_onep, real_minus_onep, chain_index): Same.
3556 (print_type_hash_statistics, type_list_equal): Same.
3557 * tree.h (zerop): Same.
3558 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
3559 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3560 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3561 (real_onep, real_minus_onep, chain_index): Same.
3562 (print_type_hash_statistics, type_list_equal): Same.
3564 2018-09-26 Jim Wilson <jimw@sifive.com>
3566 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
3568 2018-09-26 Jakub Jelinek <jakub@redhat.com>
3571 * config/i386/i386.c: Include debug.h and dwarf2out.h.
3572 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
3575 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
3577 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
3579 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3581 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
3582 and FP_SECOND_SSE_REGS.
3583 (REG_CLASS_NAMES): Ditto.
3584 (REG_CLASS_CONTENTS): Ditto.
3585 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
3586 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
3587 (ix86_preferred_output_reload_class): Ditto.
3588 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
3589 clobber constraint to "=&f".
3590 (fix_truncdi_i387): Ditto.
3591 (lrintxfdi2): Ditto.
3592 (fistdi2_<rounding>): Ditto.
3593 (fpremxf4_i387): Change "=u" constraint to "=f".
3594 (fprem1xf4_i387): Ditto.
3596 (fptanxf4_i387): Ditto.
3597 (fxtractxf3_i387): Ditto.
3598 (fscalexf4_i387): Ditto.
3599 (atan2xf3): Change "u" constraint to "f".
3600 (fyl2xxf3_i387): Ditto.
3601 (fyl2xp1xf3_i387): Ditto.
3603 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3606 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
3607 for removed I387_MASK_PM entity.
3610 2018-09-26 Jeff Law <law@redhat.com>
3612 2018-09-26 Alexey Neyman <stilor@att.net>
3614 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3615 headers are no longer pulled in by <isl/val.h>.
3617 2018-09-26 Richard Biener <rguenther@suse.de>
3620 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
3621 or concrete instance DIE to the tree. Create abstract origin
3622 attributes also for concrete instances.
3624 2018-09-26 Alexey Neyman <stilor@att.net>
3626 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3627 headers are no longer pulled in by <isl/val.h>.
3629 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
3631 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
3632 Use new helper functions.
3633 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
3634 Use new helper functions.
3635 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
3636 aarch_mm_needs_release): New declarations.
3637 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
3638 aarch_mm_needs_release): New.
3640 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
3642 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
3643 (arm32_output_mi_thunk): Deal with long calls.
3645 2018-09-26 Richard Biener <rguenther@suse.de>
3649 * tree-inline.c (expand_call_inline): When the location
3650 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
3651 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
3652 the inserted BLOCK to make inlined_function_outer_scope_p
3654 * dwarf2out.c (add_call_src_coords_attributes): Do not add
3655 coords for reserved locations.
3657 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3659 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
3660 (*call_indirect_nonlocal_sysv<mode>): Ditto.
3661 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
3662 (*sibcall_nonlocal_sysv<mode>): Ditto.
3663 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
3664 (<bd>_<mode>): Ditto.
3665 (<bd>tf_<mode>): Ditto.
3667 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3669 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
3670 control string as a list of templates instead of as C code.
3671 (*altivec_movti): Ditto.
3672 * config/rs6000/darwin.md (movdf_low_di): Ditto.
3674 2018-09-25 Jim Wilson <jimw@sifive.com>
3676 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
3677 when target symbol is weak.
3679 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
3682 * builtins.c (unterminated_array): Simplify.
3683 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
3684 where pointer arithmetic is safe.
3686 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3689 * config/rs6000/altivec.md (altivec_vspltb): Use
3690 const_0_to_15_operand instead of u5bit_cint_operand.
3691 (*altivec_vspltb_internal): Ditto.
3692 (altivec_vspltb_direct): Ditto.
3693 (altivec_vsplth): Use const_0_to_7_operand instead of
3695 (*altivec_vsplth_internal): Ditto.
3696 (altivec_vsplth_direct): Ditto.
3697 (altivec_vspltw): Use const_0_to_3_operand instead of
3699 (*altivec_vspltw_internal): Ditto.
3700 (altivec_vspltw_direct): Ditto.
3701 (altivec_vspltsf): Ditto.
3702 (*altivec_vspltsf_internal): Ditto.
3703 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
3704 various splats with the proper size immediate. Reorder the various
3705 cases by ascending size of immediate, and put all such together.
3707 2018-09-25 Richard Biener <rguenther@suse.de>
3710 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
3712 (maybe_create_die_with_external_ref): Do not create
3713 DW_TAG_imported_unit here.
3714 (add_abstract_origin_attribute): Handle external BLOCK refs.
3715 (dwarf2out_abstract_function): Simplify LTO case.
3716 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
3717 rather than using maybe_create_die_with_external_ref.
3719 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
3722 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
3723 (frndintxf2_mask_pm_i387): Ditto.
3724 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
3725 Enable for !flag_trapping_math.
3726 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
3727 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
3728 Change operand 1 predicate to nonimmediate_operand.
3729 (attr "i387_cw"): Remove mask_pm.
3730 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
3731 (enum ix86_entity): Remove I387_MASK_PM.
3732 * config/i386/i386.c (ix86_i387_mode_needed): Do not
3733 handle I387_MASK_PM.
3734 (ix86_mode_needed): Ditto.
3735 (ix86_mode_after): Ditto.
3736 (ix86_mode_entry): Ditto.
3737 (ix86_mode_exit): Ditto.
3738 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
3740 2018-09-25 Jakub Jelinek <jakub@redhat.com>
3742 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
3743 to_update_switch_stmts to vNULL instead of calling create on them
3746 2018-09-25 Richard Biener <rguenther@suse.de>
3748 PR tree-optimization/87402
3749 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
3750 (visit_phi): Re-instantiate handling of supposed to be VARYING
3751 but non-VARYING backedge value.
3753 2018-09-25 Richard Biener <rguenther@suse.de>
3756 * dwarf2out.c (struct sym_off_pair): New.
3757 (external_die_map): New global.
3758 (lookup_decl_die): When in LTO create DIEs lazily from the
3760 (lookup_block_die): New function, create DIEs lazily in LTO.
3761 (equate_block_to_die): New function.
3762 (dwarf2out_die_ref_for_decl): During WPA get the association
3763 from the external DIE map.
3764 (dwarf2out_register_external_die): Record mapping into the
3766 (maybe_create_die_with_external_ref): New function split out from
3767 DIE generation part of old dwarf2out_register_external_die.
3768 (add_abstract_origin_attribute): Do not return the DIE. When
3769 in LTO reference externals directly.
3770 (dwarf2out_abstract_function): When in LTO ignore calls for
3771 decls with external DIEs (already present abstract instances).
3772 (gen_call_site_die): Adjust.
3773 (add_high_low_attributes): Likewise.
3774 (gen_lexical_block_die): Likewise.
3775 (gen_inlined_subroutine_die): Likewie.
3776 (gen_block_die): Likewise.
3777 (dwarf2out_inline_entry): Likewise.
3778 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
3781 2018-09-25 Martin Liska <mliska@suse.cz>
3783 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
3784 integers and not by a float value.
3786 2018-09-25 Martin Liska <mliska@suse.cz>
3789 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
3791 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
3792 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
3793 * godump.c (go_format_type): Remove extra parenthesis.
3795 2018-09-25 Martin Liska <mliska@suse.cz>
3797 * alias.c (set_dest_equal_p): Remove unused function.
3798 * config/i386/i386.c (def_builtin_pure2): Likewise.
3799 * diagnostic-show-locus.c (class layout): Remove
3801 (layout::layout): Likewise here.
3802 * dump-context.h (class temp_dump_context): Likewise.
3803 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
3804 (add_AT_loclistsptr): Likewise.
3805 (add_AT_offset): Likewise.
3806 (get_AT_hi_pc): Likewise.
3807 (is_comdat_die): Likewise.
3808 (type_is_enum): Likewise.
3809 (ceiling): Likewise.
3810 (add_AT_vms_delta): Likewise.
3811 (is_class_die): Likewise.
3812 * edit-context.c (class line_event): Remove unused field.
3813 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
3815 * ipa-cp.c (ipa_get_vr_lat): Likewise.
3816 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
3817 (ok_for_base_p_nonstrict): Likewise.
3818 * tree-chrec.c (is_not_constant_evolution): Likewise.
3819 (chrec_fold_poly_cst): Likewise.
3820 * tree-if-conv.c (has_pred_critical_p): Likewise.
3821 * tree-ssa-coalesce.c (print_exprs): Likewise.
3822 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
3823 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
3824 * tree-vrp.c (value_ranges_intersect_p): Likewise.
3825 (value_range_nonnegative_p): Likewise.
3827 2018-09-25 Martin Liska <mliska@suse.cz>
3829 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
3830 Do not handle "GNU Pascal".
3831 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
3833 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
3834 from documentation. Likewise.
3835 * dbxout.c (dbxout_range_type): Likewise.
3836 * doc/cpp.texi: Likewise.
3837 * doc/extend.texi: Likewise.
3838 * doc/frontends.texi: Likewise.
3839 * doc/invoke.texi: Remove Pascal entry.
3840 * tree.def (CLEANUP_POINT_EXPR): Likewise.
3841 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
3843 2018-09-25 Martin Liska <mliska@suse.cz>
3846 * doc/invoke.texi: Document all parameters and remove default
3849 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
3852 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
3853 contains HOST_WIDE_INTs when computing its size.
3855 2018-09-24 Jim Wilson <jimw@sifive.com>
3858 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
3860 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
3862 2018-09-24 Andrew Pinski <apinski@marvell.com>
3864 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
3865 access prev before checking it for NULLness in the
3866 AARCH64_FUSE_CMP_BRANCH case.
3868 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
3871 * config/i386/i386.c (rest_of_insert_endbranch): Set
3872 endbr_queued_at_entrance to true and don't insert ENDBR if
3873 x86_function_profiler will be called.
3874 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
3876 * config/i386/i386.h (machine_function): Add
3877 endbr_queued_at_entrance.
3879 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3881 * genattrtab.c (mk_attr_alt): Use alternative_mask.
3882 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
3884 (check_attr_test): Use alternative_mask.
3885 (get_attr_value): Likewise.
3886 (compute_alternative_mask): Use alternative_mask and XWINT.
3887 (make_alternative_compare): Use alternative_mask.
3888 (attr_alt_subset_p): Use XWINT.
3889 (attr_alt_subset_of_compl_p): Likewise.
3890 (attr_alt_intersection): Use alternative_mask and XWINT.
3891 (attr_alt_union): Likewise.
3892 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
3893 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
3894 (simplify_test_exp): Use alternative_mask and XWINT.
3895 (write_test_expr): Use alternative_mask and XWINT, adjust bit
3896 number calculation to support 64 bits. Generate code that
3897 checks 64-bit masks.
3898 (main): Use alternative_mask.
3899 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
3901 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3904 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
3905 RETURN+USE when returning via %r14.
3907 2018-09-24 Martin Liska <mliska@suse.cz>
3909 * gcov.c (output_lines): Print colorization legend
3910 for both flag_use_colors and flag_use_hotness_colors.
3913 2018-09-24 Martin Liska <mliska@suse.cz>
3915 * coverage.c (get_coverage_counts): Use warning_at
3916 with current_function_decl location. Use %qD in warning
3919 2018-09-24 Martin Liska <mliska@suse.cz>
3921 * memory-block.h (memory_block_pool::release): Annotate with
3922 valgrind that the memory is not accessible.
3924 2018-09-24 Martin Liska <mliska@suse.cz>
3927 * asan.c: Make asan_handled_variables extern.
3929 * cfgexpand.c (expand_stack_vars): Make sure
3930 a representative is unpoison if another
3931 variable in the partition is handled by
3932 use-after-scope sanitization.
3934 2018-09-24 Richard Biener <rguenther@suse.de>
3936 PR tree-optimization/63155
3937 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
3938 the worklist when the edge of the respective argument isn't
3941 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3943 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
3944 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
3945 (MASK_CLASS_P): Update for rename.
3946 (MAYBE_MASK_CLASS_P): Ditto.
3947 (REG_CLASS_NAMES): Update.
3948 (REG_CLASS_CONTENT): Update.
3949 * config/i386/i386.c (regclass_map): Update for MASK_REG
3950 and ALL_MASK_REGS rename.
3951 * config/i386/constraints.md (Yk): Update for rename.
3954 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3956 * config/i386/i386.h (enum reg_class): Remove
3957 EVEX_SSE_REGS and MOD4_SSE_REGS.
3958 (REG_CLASS_NAMES): Update.
3959 (REG_CLASS_CONTENT): Update.
3960 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
3961 registers as ALL_SSE_REGS.
3962 (ix86_additional_allocno_class_p): Remove.
3963 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
3964 (ix86_register_priority): Lower priority of EVEX SSE registers.
3965 Use IN_RANGE macro where appropriate.
3966 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
3967 AVX-5124VNNIW checks.
3968 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
3969 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
3970 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
3971 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
3972 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
3973 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
3974 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
3975 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
3976 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
3977 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
3978 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
3979 * config/i386/constraints.md (Yh): Remove.
3981 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3983 * config/i386/i386.c (regclass_map): Declare integer REX registers
3986 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
3988 * doc/service.texi (Service): Switch the fsf.org link to https.
3990 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
3993 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3994 Define to speculation_safe_value_not_needed.
3996 2018-09-21 Florian Weimer <fweimer@redhat.com>
3999 * doc/extend.texi (Common Function Attributes): Mention that
4000 noreturn suppresses tail call optimization.
4002 2018-09-21 Jeff Law <law@redhat.com>
4004 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
4005 vr_values::cleanup_edges_and_switches.
4006 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
4008 (identify_jump_threads): Remove EDGE_IGNORE handling.
4009 (execute_vrp): Move handling of to_remove_edges and
4010 to_update_switch_stmts into vr_values class member functions.
4011 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
4012 (to_update_switch_stmts): Likewise.
4013 * vr-values.c: Include cfghooks.h.
4014 (vr_values::vr_values): Initialize to_remove_edges and
4015 to_update_switch_stmts.
4016 (vr_values::~vr_values): Verify to_remove_edges and
4017 to_update_switch_stmts are empty.
4018 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
4019 (vr_values::cleanup_edges_and_switches): New member function.
4020 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
4021 function. Add new data members.
4023 2018-09-21 David Malcolm <dmalcolm@redhat.com>
4025 PR tree-optimization/87309
4026 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
4027 calls with pflags and alt_flags.
4028 (selftest::test_capture_of_dump_calls): Add test of interaction of
4029 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
4031 2018-09-21 Olivier Hainque <hainque@adacore.com>
4033 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
4035 2018-09-21 Olivier Hainque <hainque@adacore.com>
4037 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
4039 2018-09-21 Olivier Hainque <hainque@adacore.com>
4041 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
4042 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
4043 (VXWORKS_LIBS_RTP): Minor reordering.
4045 2018-09-21 Olivier Hainque <hainque@adacore.com>
4047 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
4048 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
4050 2018-09-21 Olivier Hainque <hainque@adacore.com>
4052 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
4053 (PTRDIFF_TYPE): Likewise.
4055 2018-09-21 Olivier Hainque <hainque@adacore.com>
4057 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
4058 triplet, similar to support for VxWorks7.
4059 * config/vxworks-dummy.h: Provide a default definition
4060 of TARGET_VXWORKS64 to 0.
4062 2018-09-21 Olivier Hainque <hainque@adacore.com>
4064 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
4065 * config/vxworks-dummy.h: here.
4067 2018-09-21 Olivier Hainque <hainque@adacore.com>
4069 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
4071 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
4072 Bo Zhou <zbo.zhou@hisilicon.com>
4074 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
4075 * config/aarch64/aarch64-tune.md: Regenerated.
4076 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
4077 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
4078 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
4080 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
4081 Julian Brown <julian@codesourcery.com>
4083 * builtins.c (get_builtin_sync_mem): Handle address spaces.
4085 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
4087 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
4088 if the call takes a static chain.
4090 2018-09-21 Martin Liska <mliska@suse.cz>
4092 * auto-profile.c (autofdo_source_profile::read): Do not
4094 (read_profile): Do not add working sets.
4095 (read_autofdo_file): Remove sum_all.
4096 (afdo_callsite_hot_enough_for_early_inline): Remove const
4098 * coverage.c (struct counts_entry): Remove gcov_summary.
4099 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
4100 do not support GCOV_TAG_PROGRAM_SUMMARY.
4101 (get_coverage_counts): Remove summary and expected
4103 * coverage.h (get_coverage_counts): Likewise.
4104 * doc/gcov-dump.texi: Remove -w option.
4105 * gcov-dump.c (dump_working_sets): Remove.
4106 (main): Do not support '-w' option.
4107 (print_usage): Likewise.
4108 (tag_summary): Likewise.
4109 * gcov-io.c (gcov_write_summary): Do not dump
4111 (gcov_read_summary): Likewise.
4112 (gcov_histo_index): Remove.
4113 (gcov_histogram_merge): Likewise.
4114 (compute_working_sets): Likewise.
4115 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
4117 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
4118 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
4119 (GCOV_HISTOGRAM_SIZE): Remove.
4120 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
4121 (struct gcov_summary): Simplify rapidly just
4122 to runs and sum_max fields.
4123 (gcov_histo_index): Remove.
4124 (NUM_GCOV_WORKING_SETS): Likewise.
4125 (compute_working_sets): Likewise.
4126 * gcov-tool.c (print_overlap_usage_message): Remove
4127 trailing empty line.
4128 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
4129 (output_lines): Remove program related line.
4130 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
4131 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
4133 (input_profile_summary): Do not read it.
4134 (merge_profile_summaries): And do not merge it.
4135 (input_symtab): Do not call removed function.
4136 * modulo-sched.c (sms_schedule): Do not print sum_max.
4137 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
4138 removed when histogram method was invented.
4139 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
4141 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
4143 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
4144 and dump selected value.
4145 * profile.c (add_working_set): Remove.
4146 (get_working_sets): Likewise.
4147 (find_working_set): Likewise.
4148 (get_exec_counts): Do not work with working sets.
4149 (read_profile_edge_counts): Do not inform as sum_max is removed.
4150 (compute_branch_probabilities): Likewise.
4151 (compute_value_histograms): Remove argument for call of
4152 get_coverage_counts.
4153 * profile.h: Do not make gcov_summary const.
4155 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
4157 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
4159 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
4161 PR tree-optimization/86990
4162 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
4163 Check that the entire merged store group is made of constants only for
4166 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
4168 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
4169 (VTABLE_VERIFICATION_SPEC): Likewise.
4170 (SANITIZER_EARLY_SPEC): Likewise.
4171 (SANITIZER_SPEC): Likewise.
4172 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
4173 * doc/invoke.texi (Link Options): Document -r.
4175 2018-09-20 Richard Biener <rguenther@suse.de>
4178 * gimplify.c (gimplify_expr): Retain alignment of
4179 addressable lvalue in dereference.
4181 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
4184 * configure.ac: Check for .loc is_stmt support.
4185 * configure, config.in: Rebuilt.
4186 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
4189 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4191 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
4192 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
4195 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4197 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
4199 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
4202 2018-09-20 Marek Polacek <polacek@redhat.com>
4204 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
4206 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
4208 PR tree-optimization/87288
4209 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
4210 into account when determining PEELING_FOR_NITERS.
4212 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
4214 PR tree-optimization/86877
4215 * tree-vect-loop.c (vect_analyze_loop_2): Call
4216 vect_verify_datarefs_alignment.
4218 2018-09-19 Marek Polacek <polacek@redhat.com>
4220 * doc/invoke.texi: Document -Wclass-conversion.
4222 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
4224 * config/pa/pa.c (pa_adjust_priority): Delete.
4225 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
4227 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
4228 (atomic_storehi): Likewise.
4229 (atomic_storesi): Likewise.
4230 (atomic_loaddi): Restore compare and swap exchange loop code.
4232 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4234 PR rtl-optimization/86902
4235 * combine.c (try_combine): When changing the CC mode used, don't change
4236 an unrelated mode in other_insn to that new CC mode.
4238 2018-09-19 David Malcolm <dmalcolm@redhat.com>
4240 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
4241 with %T in place of calls to dump_generic_expr.
4242 (prune_runtime_alias_test_list): Likewise.
4243 (create_runtime_alias_checks): Likewise.
4244 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
4245 (vect_analyze_data_ref_dependence): Likewise.
4246 (vect_slp_analyze_data_ref_dependence): Likewise.
4247 (vect_record_base_alignment): Likewise. Use %G in place of call
4248 to dump_gimple_stmt.
4249 (vect_compute_data_ref_alignment): Likewise.
4250 (verify_data_ref_alignment): Likewise.
4251 (vect_find_same_alignment_drs): Likewise.
4252 (vect_analyze_group_access_1): Likewise.
4253 (vect_analyze_data_ref_accesses): Likewise.
4254 (dependence_distance_ge_vf): Likewise.
4255 (dump_lower_bound): Likewise.
4256 (vect_prune_runtime_alias_test_list): Likewise.
4257 (vect_find_stmt_data_reference): Likewise.
4258 (vect_analyze_data_refs): Likewise.
4259 (vect_create_addr_base_for_vector_ref): Likewise.
4260 (vect_create_data_ref_ptr): Likewise.
4261 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
4262 (vect_can_advance_ivs_p): Likewise.
4263 (vect_update_ivs_after_vectorizer): Likewise.
4264 (vect_gen_prolog_loop_niters): Likewise.
4265 (vect_prepare_for_masked_peels): Likewise.
4266 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
4267 (vect_determine_vectorization_factor): Likewise.
4268 (vect_is_simple_iv_evolution): Likewise.
4269 (vect_analyze_scalar_cycles_1): Likewise.
4270 (vect_analyze_loop_operations): Likewise.
4271 (report_vect_op): Likewise.
4272 (vect_is_slp_reduction): Likewise.
4273 (check_reduction_path): Likewise.
4274 (vect_is_simple_reduction): Likewise.
4275 (vect_create_epilog_for_reduction): Likewise.
4276 (vect_finalize_reduction:): Likewise.
4277 (vectorizable_induction): Likewise.
4278 (vect_transform_loop_stmt): Likewise.
4279 (vect_transform_loop): Likewise.
4280 (optimize_mask_stores): Likewise.
4281 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
4282 (vect_split_statement): Likewise.
4283 (vect_recog_over_widening_pattern): Likewise.
4284 (vect_recog_average_pattern): Likewise.
4285 (vect_determine_min_output_precision_1): Likewise.
4286 (vect_determine_precisions_from_range): Likewise.
4287 (vect_determine_precisions_from_users): Likewise.
4288 (vect_mark_pattern_stmts): Likewise.
4289 (vect_pattern_recog_1): Likewise.
4290 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
4291 (vect_record_max_nunits): Likewise.
4292 (vect_build_slp_tree_1): Likewise.
4293 (vect_build_slp_tree_2): Likewise.
4294 (vect_print_slp_tree): Likewise.
4295 (vect_analyze_slp_instance): Likewise.
4296 (vect_detect_hybrid_slp_stmts): Likewise.
4297 (vect_detect_hybrid_slp_1): Likewise.
4298 (vect_slp_analyze_operations): Likewise.
4299 (vect_slp_analyze_bb_1): Likewise.
4300 (vect_transform_slp_perm_load): Likewise.
4301 (vect_schedule_slp_instance): Likewise.
4302 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
4303 (vect_mark_stmts_to_be_vectorized): Likewise.
4304 (vect_init_vector_1): Likewise.
4305 (vect_get_vec_def_for_operand): Likewise.
4306 (vect_finish_stmt_generation_1): Likewise.
4307 (vect_check_load_store_mask): Likewise.
4308 (vectorizable_call): Likewise.
4309 (vectorizable_conversion): Likewise.
4310 (vectorizable_operation): Likewise.
4311 (vectorizable_load): Likewise.
4312 (vect_analyze_stmt): Likewise.
4313 (vect_is_simple_use): Likewise.
4314 (vect_get_vector_types_for_stmt): Likewise.
4315 (vect_get_mask_type_for_stmt): Likewise.
4316 * tree-vectorizer.c (increase_alignment): Likewise.
4318 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
4320 * doc/rtl.texi: Adjust vec_select description.
4321 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
4322 non-constant selectors.
4324 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
4326 * config/aarch64/aarch64-protos.h
4327 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
4328 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
4329 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
4330 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
4331 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
4332 (AARCH64_FL_PROFILE): Move index so flags are ordered.
4333 (AARCH64_ISA_RCPC8_4): New flag.
4334 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
4335 to aarch64_offset_9bit_signed_unscaled_p.
4336 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
4338 * config/aarch64/constraints.md (Ust): New constraint.
4339 * config/aarch64/predicates.md.
4340 (aarch64_9bit_offset_memory_operand): New predicate.
4341 (aarch64_rcpc_memory_operand): New predicate.
4343 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
4345 PR rtl-optimization/87361
4346 * rtlanal.c (nonzero_bits1): Revert accidental change.
4348 2018-09-19 Richard Biener <rguenther@suse.de>
4350 PR tree-optimization/87349
4351 PR tree-optimization/87342
4352 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
4354 2018-09-18 Marek Polacek <polacek@redhat.com>
4356 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
4357 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
4359 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
4361 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
4363 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
4365 PR rtl-optimization/86882
4366 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
4368 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
4370 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
4371 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
4373 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
4376 * doc/invoke.texi (Link Options): Fix formatting and grammar.
4378 2018-09-18 Richard Biener <rguenther@suse.de>
4381 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
4382 (compute_samebase_partition_bases): Likewise.
4383 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
4384 (gimple_can_coalesce_p): Simplify.
4386 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
4388 Handle a library implementation of ffs calling __builtin_ffs.
4389 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
4390 (mmix_init_libfuncs): New function: make __builtin_ffs expand
4393 2018-09-17 David Malcolm <dmalcolm@redhat.com>
4395 * diagnostic-show-locus.c (class layout_range): Add field
4397 (layout_range::layout_range): Add "original_idx" param and use it
4398 to initialize new field.
4399 (make_range): Use 0 for original_idx.
4400 (layout::layout): Pass in index to calls to
4401 maybe_add_location_range.
4402 (layout::maybe_add_location_range): Add param "original_idx" and
4403 pass it on to layout_range.
4404 (layout::print_any_labels): Pass on range->m_original_idx to
4406 (gcc_rich_location::add_location_if_nearby): Use 0 for
4408 * gcc-rich-location.h (text_range_label::get_text): Update for new
4410 (range_label_for_type_mismatch::get_text): Likewise.
4412 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
4414 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
4416 2018-09-17 David Malcolm <dmalcolm@redhat.com>
4418 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
4419 format_string_diagnostic_t.
4420 (fmtwarn_n): Likewise.
4421 * substring-locations.c
4422 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
4423 (format_warning_n_va): Convert to...
4424 (format_string_diagnostic_t::emit_warning_n_va): ...this.
4425 (format_warning_va): Convert to...
4426 (format_string_diagnostic_t::emit_warning_va): ...this.
4427 (format_warning_at_substring): Convert to...
4428 (format_string_diagnostic_t::emit_warning): ...this.
4429 (format_warning_at_substring_n): Convert to...
4430 (format_string_diagnostic_t::emit_warning_n): ...this.
4431 * substring-locations.h (class format_string_diagnostic_t): New
4433 (format_warning_va): Convert to
4434 format_string_diagnostic_t::emit_warning_va.
4435 (format_warning_n_va): Convert to
4436 format_string_diagnostic_t::emit_warning_n_va.
4437 (format_warning_at_substring): Convert to
4438 format_string_diagnostic_t::emit_warning.
4439 (format_warning_at_substring_n): Convert to
4440 format_string_diagnostic_t::emit_warning_n.
4442 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
4443 Bernd Schmidt <bernds_cb1@t-online.de>
4445 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
4448 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
4450 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
4451 operand 0 predicate to nonimmediate operand.
4452 (rint<mode>2_frndint): Remove insn pattern.
4453 (rint<mode>2): Change operand 1 predicate to general_operand.
4454 Extend operand 1 to XFmode and generate rintxf2 insn.
4455 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
4456 Do not use X87MODEF mode macro.
4457 (frndintxf2_<rounding>_i387): Rename from
4458 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
4459 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
4460 to XFmode and generate significandxf3 insn.
4462 2018-09-17 Richard Biener <rguenther@suse.de>
4464 PR tree-optimization/87328
4465 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
4466 visiting unexecutable backedges when not iterating.
4467 (do_rpo_vn): Mark all edges not executable even when not
4470 2018-09-17 Martin Jambor <mjambor@suse.cz>
4473 * doc/invoke.texi (Warning Options): Likewise.
4475 2018-09-17 Richard Biener <rguenther@suse.de>
4477 PR tree-optimization/87301
4478 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
4479 clean EH info from leftover copy assignments.
4481 2018-09-17 Martin Liska <mliska@suse.cz>
4483 PR gcov-profile/85871
4484 * gcov.c (output_intermediate_file): Fix out of bounds
4487 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
4489 * config/arc/arc.c: Object attributes for core4 not reflected
4491 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
4494 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
4496 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
4498 2018-09-17 Martin Liska <mliska@suse.cz>
4500 * doc/gcov.texi: Document new option --use-hotness-colors.
4501 * gcov.c (struct source_info): Declare new field.
4502 (source_info::source_info): Set default for maximum_count.
4503 (print_usage): Add new -q option.
4504 (process_args): Process it.
4505 (accumulate_line_info): Save src->maximum_count.
4506 (output_line_beginning): Make color line number if
4507 flag_use_hotness_colors is set.
4508 (output_line_details): Pass default argument value.
4509 (output_lines): Pass src->maximum_count.
4511 2018-09-17 Martin Liska <mliska@suse.cz>
4513 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
4514 Use processor_names table.
4515 * config/i386/i386.c (ix86_default_align): Use
4516 processor_cost_table for alignment values.
4517 (ix86_option_override_internal): Use processor_names.
4518 (ix86_function_specific_print): Likewise.
4519 * config/i386/i386.h (struct processor_costs):
4520 Add alignment values.
4521 (struct ptt): Remove and replace with const char *.
4522 * config/i386/x86-tune-costs.h (struct processor_costs):
4523 Declare default alignments for all costs.
4525 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
4527 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
4528 symbolics or VR_VARYING ranges for ABS_EXPR.
4529 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
4530 when range will wrap.
4532 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
4535 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
4536 before and after a JUMP_TABLE_DATA.
4538 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
4541 * dojump.c (do_compare_and_jump): Canonicalize function pointers
4542 when one operand is a function pointer. Use POINTER_TYPE_P and
4543 FUNC_OR_METHOD_TYPE_P.
4544 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
4545 * fold-const.c (build_range_check): Likewise.
4546 * match.pd (simple_comparison): Likewise.
4548 2018-09-14 David Malcolm <dmalcolm@redhat.com>
4551 * spellcheck.c (get_edit_distance_cutoff): New function.
4552 (selftest::test_edit_distance_unit_test_oneway): Rename to...
4553 (selftest::test_get_edit_distance_one_way): ...this.
4554 (selftest::test_get_edit_distance_unit): Rename to...
4555 (selftest::test_get_edit_distance_both_ways): ...this.
4556 (selftest::test_edit_distances): Move tests to this new function,
4557 and test some more pairs of strings. Update for above renaming.
4558 (selftest::get_old_cutoff): New function.
4559 (selftest::test_get_edit_distance_cutoff): New function.
4560 (selftest::assert_suggested_for): New function.
4561 (ASSERT_SUGGESTED_FOR): New macro.
4562 (selftest::assert_not_suggested_for): New function.
4563 (ASSERT_NOT_SUGGESTED_FOR): New macro.
4564 (selftest::test_suggestions): New function.
4565 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
4566 tests to selftest::test_edit_distances and call it. Add calls to
4567 selftest::test_get_edit_distance_cutoff and
4568 selftest::test_suggestions.
4569 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
4570 (best_match::consider): Replace hard-coded cutoff calculation with
4572 (best_match::get_cutoff): New declaration.
4573 (best_match::get_best_meaningful_candidate): Likewise.
4575 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4577 * builtins.c (fold_builtin_strlen): Remove TODO comment.
4579 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4582 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4584 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4585 terminated string literal.
4587 2018-09-14 Martin Sebor <msebor@redhat.com>
4589 * builtins.c (unterminated_array): Handle ARRAY_REF.
4590 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
4591 * builtins.h (unterminated_array): Declare extern.
4592 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
4594 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
4597 2018-09-14 Martin Sebor <msebor@redhat.com>
4598 Jeff Law <law@redhat.com>
4600 * builtins.c (unterminated_array): New.
4601 (expand_builtin_strcpy): Adjust.
4602 (expand_builtin_strcpy_args): Detect unterminated arrays.
4603 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
4604 unterminated arrays.
4605 * gimple-fold.h (get_maxval_strlen): Add argument.
4606 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
4608 * gimple-fold.c (get_range_strlen): Add argument.
4609 (get_maxval_strlen): Adjust.
4610 * gimple-fold.h (get_range_strlen): Add argument.
4612 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
4614 * config/i386/movdirintrin.h: Fix copyright year.
4616 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
4618 * reg-stack.c: Include regs.h.
4619 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
4620 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
4621 FIRST_STACK_REG, not DFmode.
4622 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
4623 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
4624 (change stack): Default register mode to the reg_raw_mode of
4625 FIRST_STACK_REG, not DFmode.
4626 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
4627 (*swapxf): Rename from swapxf.
4629 2018-09-14 Carl Love <cel@us.ibm.com>
4631 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
4632 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
4634 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
4637 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
4640 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
4643 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
4645 2018-09-14 Jason Merrill <jason@redhat.com>
4647 Fix --enable-gather-detailed-mem-stats.
4648 * hash-table.c (hash_table_usage): Change from variable to function.
4649 * hash-table.h: Adjust.
4650 * Makefile.in: Add missing dependencies on hash-table.h.
4652 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4654 PR tree-optimization/87259
4656 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
4657 execute_cse_reciprocals_1 has tried transforming.
4659 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
4661 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
4662 VR_VARYING for PLUS/MINUS_EXPR.
4664 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
4666 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
4669 2018-09-14 Richard Biener <rguenther@suse.de>
4672 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
4673 bits for the merged partition.
4675 2018-09-13 Martin Sebor <msebor@redhat.com>
4676 Bernd Edlinger <bernd.edlinger@hotmail.de>
4678 * builtins.h (c_srlen): Add argument.
4679 * builtins.c (warn_string_no_nul): New function.
4680 (c_strlen): Add argument and use it. Update recursive calls.
4681 Pass DECL argument to string_constant to get info on non
4682 terminated strings. Update *NONSTR as needed.
4683 (fold_builtin_strlen): Add argument to calls to c_strlen.
4684 Warn for unterminated arrays.
4685 (warn_string_no_null): Add prototype.
4686 * expr.c (string_constant): Update arguments. Update recursive
4687 calls appropriately. Detect missing NUL terminator and outermost
4688 declaration its missing in.
4689 Improve checks for arrays with nonzero lower bound or elements
4690 that are not a single byte. Simplify offset computation.
4691 Simplify checks for non-NUL terminated strings.
4692 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
4693 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
4695 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
4697 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
4699 * fold-const.c (c_getstr): Fix function comment. Remove unused third
4700 argument. Fix range checks.
4701 * fold-const.h (c_getstr): Adjust protoype.
4702 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
4703 string is constant but contains no NUL byte.
4705 * expr.c (string_constant): Adjust function comment.
4706 Remove bogus check for zero termination.
4708 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
4710 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
4711 (get_constant_size): Don't make STRING_CSTs larger than they are.
4712 (check_string_literal): New check function for STRING_CSTs.
4713 (output_constant): Use it.
4715 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
4718 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
4720 2018-09-13 Richard Biener <rguenther@suse.de>
4722 PR tree-optimization/87263
4723 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
4724 (struct unwind_state): Add max_rpo field.
4725 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
4726 Compute max_rpo, the max RPO number a block can be backwards reached
4727 from. Re-write non-iterating mode to a RPO ordered worklist approach,
4728 separating it from the iterating mode.
4730 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4732 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
4733 (rfs_decision): New scheduling decision.
4735 2018-09-13 Richard Biener <rguenther@suse.de>
4738 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
4739 (vn_nary_op_insert_pieces_predicated): Do not write useless
4740 valid_dominated_by_p entry outside of the allocated storage.
4742 2018-09-13 Omar Sandoval <osandov@osandov.com>
4743 Tom de Vries <tdevries@suse.de>
4746 * dwarf2out.c (is_c): New function.
4747 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
4749 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
4752 * config/aarch64/aarch64.md (*aarch64_bfxil):
4754 * config/aarch64/constraints.md (Ulc): Define.
4755 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
4757 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
4760 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4762 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
4763 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
4764 aarch64_layout_frame call.
4765 (aarch64_expand_epilogue): Likewise.
4766 (aarch64_initial_elimination_offset): Likewise.
4767 (aarch64_get_separate_components): Likewise.
4768 (aarch64_use_return_insn_p): Likewise.
4769 (aarch64_layout_frame): Remove unneeded check.
4771 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4773 * configure.ac: Only append
4774 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
4775 gcc_config_arguments if it was never reconfigured or last reconfigure
4776 was with different arguments.
4777 * configure: Regenerated.
4779 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4780 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4783 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
4784 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
4786 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4788 PR tree-optimization/87287
4789 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
4790 X % C == 0 to X % (unsigned) C == 0 optimization to ...
4791 * match.pd (X % C == 0): ... here. New optimization.
4793 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4796 * expr.h (maybe_optimize_mod_cmp): Declare.
4797 * expr.c (mod_inv): New function.
4798 (maybe_optimize_mod_cmp): New function.
4799 (do_store_flag): Use it.
4800 * cfgexpand.c (expand_gimple_cond): Likewise.
4802 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
4803 Julian Brown <julian@codesourcery.com>
4806 * gimplify.c (gimplify_scan_omp_clauses): Set
4807 target_firstprivatize_array_bases in OpenACC parallel and kernels
4808 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
4809 OpenACC data regions.
4811 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
4813 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
4814 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
4816 2018-09-12 Richard Biener <rguenther@suse.de>
4818 PR tree-optimization/87280
4819 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
4820 edge but unreachable target.
4821 (do_rpo_vn): For conservatively handling a PHI only mark
4822 the backedge executable but not the block reachable.
4824 2018-09-12 Richard Biener <rguenther@suse.de>
4826 PR tree-optimization/87266
4827 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
4830 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
4832 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
4834 ("trunc<BFP:mode><DFP_ALL:mode>2")
4835 ("trunc<DFP_ALL:mode><BFP:mode>2")
4836 ("extend<BFP:mode><DFP_ALL:mode>2")
4837 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
4838 according to the target operand type.
4840 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4841 Andreas Krebbel <krebbel@linux.ibm.com>
4843 PR tree-optimization/86844
4844 * gimple-ssa-store-merging.c
4845 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
4846 there are any overlapping stores in between them, make sure they are
4847 also coalesced or we give up completely.
4849 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4852 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
4853 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
4855 2018-09-12 Tom de Vries <tdevries@suse.de>
4857 * common.opt (gdescribe-dies): Add option.
4858 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
4859 attribute for artifical and nameless decls.
4860 (dwarf2out_register_external_die): Add description attribute to
4861 external reference die.
4862 (add_desc_attribute): New functions.
4863 (gen_subprogram_die): Add description attribute to
4864 DW_TAG_call_site_parameter.
4865 * tree-pretty-print.c (print_generic_expr_to_str): New function.
4866 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
4867 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
4869 (@item -gdescribe-dies): Add.
4871 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4873 * tree-vrp.c (vrp_shift_undefined_p): Remove.
4874 (extract_range_from_binary_expr_1: Call
4875 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
4876 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
4879 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4881 * gimple-ssa-warn-alloca.c
4882 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
4883 field for ALLOCA_BOUND_*_LARGE.
4885 2018-09-11 Nathan Sidwell <nathan@acm.org>
4887 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
4889 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4891 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
4892 for clobbers. Remove obsolete comment.
4894 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4896 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
4897 mpxchk, mpxld and mpxst types.
4898 (define_attr length_immediate): Remove all processing of mpx types.
4899 (define_attr prefix_0f): Ditto.
4900 (define_attr memory): Ditto.
4902 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4904 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
4905 (log<mode>2): Change operand 1 predicate to general_operand.
4906 Extend operand 1 to XFmode and generate logxf3 insn.
4907 (log10<mode>2): Change operand 1 predicate to general_operand.
4908 Extend operand 1 to XFmode and generate log10xf3 insn.
4909 (log2<mode>2): Change operand 1 predicate to general_operand.
4910 Extend operand 1 to XFmode and generate log2xf3 insn.
4911 (fyl2xp1_extend<mode>xf3_i387): Remove.
4912 (log1p<mode>2): Change operand 1 predicate to general_operand.
4913 Extend operand 1 to XFmode and generate log1pxf3 insn.
4914 (fxtract_extend<mode>xf3_i387): Remove.
4915 (logb<mode>2): Change operand 1 predicate to general_operand.
4916 Extend operand 1 to XFmode and generate logbxf3 insn.
4917 (ilogb<mode>2): Change operand 1 predicate to general_operand.
4918 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
4919 (significand<mode>2): Change operand 1 predicate to general_operand.
4920 Extend operand 1 to XFmode and generate significandxf3 insn.
4922 2018-09-11 Nathan Sidwell <nathan@acm.org>
4924 * gcc.c (perror_with_name, pfatal_with_name): Delete.
4925 (load_specs): Use fatal_error.
4926 (DELETE_IF_ORDINARY, process_command): Use error.
4927 (execute, run_attempt): Use fatal_error.
4929 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
4931 * diagnostic-core.h (sorry_at): New prototype.
4932 * diagnostic.c (sorry_at): New function.
4934 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
4936 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
4937 by zero as VR_UNDEFINED.
4939 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4941 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
4942 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
4943 (<sincos>mode2): New expander.
4944 (sincos_extend<mode>xf3_i387): Remove insn pattern.
4945 (sincos -> sin, cos splitters): Remove splitter patterns.
4946 (sincos<mode>3): Change operand 2 predicate to general_operand.
4947 Extend operand 2 to XFmode and generate sincosxf3 insn.
4948 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
4949 Change operand 3 predicate to const1_operand.
4950 (fptan_extend<mode>xf4_i387): Remove insn pattern.
4951 (tanxf2): Update operands in the call to fptanxf4_i387.
4952 (tan<mode>2): Change operand 1 predicate to general_operand.
4953 Extend operand 1 to XFmode and generate tanxf3 insn.
4954 (atan2xf3): Rename from *fpatanxf3_i387.
4955 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
4956 (atan2xf3): Remove expander.
4957 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
4958 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
4959 (atan<mode>2): Change operand 1 predicate to general_operand.
4960 Extend operand 1 to XFmode and generate atanxf3 insn.
4962 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4964 * config/i386/i386.md (x87/SSE constant load splitter): Use
4965 memory_operand instead of nonimmediate_operand for input operand
4968 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
4970 * config/i386/i386.md (float partial SSE register stall splitter): Move
4971 splitter near its instruction pattern.
4972 (float_extend partial SSE register stall splitter): Ditto.
4973 (float_truncate partial SSE register stall splitter): Ditto.
4975 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
4978 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4979 to speculation_safe_value_not_needed.
4982 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
4983 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
4984 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
4985 leaf_function_p, instead use has_hard_reg_initial_val.
4987 2018-09-09 Nathan Sidwell <nathan@acm.org>
4989 * gcc.h (pfatal_with_name): Don't declare here.
4990 * gcc.c (pfatal_with_name): Make static.
4992 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
4994 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
4997 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
4999 PR rtl-optimization/85458
5000 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
5001 priority hook to reduce the priority of EXPR.
5003 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
5005 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
5006 DImode for x87 on 32bit targets. Conditionally disable x87 modes
5007 with X87_ENABLE_FLOAT. Remove preparation code.
5008 (*float<SWI48:mode><MODEF:mode>2): Rename from
5009 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
5010 math using "enabled" attribute.
5011 (*floatdi<MODEF:mode>2_i387): Rename from
5012 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
5013 enable for 32bit targets only.
5014 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
5016 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
5017 as operand 1 predicate. Rewrite as define_insn_and_split.
5018 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
5020 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
5022 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
5023 to fallthru to FLOAT case.
5025 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
5028 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
5029 around folding of vec_sl to handle out of range shift values.
5031 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
5033 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
5034 Update callers to gen_fix_trunc<mode>_i387_fisttp
5035 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
5036 nonimmediate_operand.
5037 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
5038 and corresponding splitters.
5039 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
5040 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
5041 (fix_truncdi_i387_with_temp): Remove insn pattern
5042 and corresponding splitters.
5043 (fix_trunc<mode>_i387): Change operand 0 predicate to
5044 nonimmediate_operand.
5045 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
5046 and corresponding splitters.
5047 (*fistdi2_1): Remove.
5049 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
5050 (lrintxfdi2): Remove expander. Reimplement as define_insn.
5051 (*fist<mode>2_1): Remove.
5052 (fist<mode>2): Ditto.
5053 (fist<mode>2_with_temp): Remove insn pattern and corresponding
5055 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
5056 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
5057 (fistdi2_<rounding>): Change operand 0 predicate to
5058 nonimmediate_operand.
5059 (fistdi2_<rounding>_with_temp): Remove insn pattern
5060 and corresponding splitters.
5061 (fist<mode>2_<rounding>): Change operand 0 predicate to
5062 nonimmediate_operand.
5063 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
5064 and corresponding splitters.
5066 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
5068 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5070 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
5073 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
5075 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5076 early gimple folding of vec_splat().
5077 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
5078 * gimple-fold.h: Add an extern define for tree_vec_extract().
5080 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
5082 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
5083 wrappers around TREE_TYPE comparisons.
5085 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
5088 * config/s390/predicates.md: Add nonsym_memory_operand.
5089 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
5090 contains a SYMBOL_REF, load it into an intermediate pseudo.
5091 (s390_emit_compare_and_swap): Legitimize operand.
5092 * config/s390/s390.md: Use the new nonsym_memory_operand
5093 with UNSPECV_CAS patterns.
5095 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
5098 * config/s390/s390-passes.def: New file.
5099 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
5101 (make_pass_s390_early_mach): Add declaration.
5102 * config/s390/s390.c (make_pass_s390_early_mach):
5103 (s390_option_override): Remove dynamic registration.
5104 * config/s390/t-s390: Add s390-passes.def.
5106 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
5108 * config/s390/s390.c (s390_decompose_constant_pool_ref):
5109 Remove UNSPEC_LTREL_BASE check.
5110 (annotate_constant_pool_refs): Likewise.
5111 (find_constant_pool_ref): Likewise.
5112 (find_ltrel_base): Removed.
5113 (replace_ltrel_base): Removed.
5114 (s390_mainpool_finish): Remove replace_ltrel_base call.
5115 (s390_chunkify_start): Remove pending LTREL_BASE logic.
5116 (s390_chunkify_finish): Remove replace_ltrel_base call.
5117 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
5119 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
5122 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
5123 to speculation_safe_value_not_needed.
5125 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
5126 Bernd Schmidt <bernds_cb1@t-online.de>
5128 * config/nvptx/nvptx-opts.h: New file.
5129 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
5130 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
5132 (TARGET_SM35): New macro.
5133 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
5135 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
5137 (misa=): New option.
5138 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
5140 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
5142 * config/i386/i386.md (truncdfsf2): Remove expander.
5143 (truncdfsf2_with_temp): Ditto.
5144 (truncxf<mode>2): Ditto.
5145 (*truncdfsf_fast_mixed): Remove insn pattern.
5146 (*truncdfsf_fast_i387): Ditto.
5147 (*truncdfsf_mixed): Ditto.
5148 (*truncdfsf_i387): Ditto.
5149 (*truncdfsf2_i387_1): Ditto.
5150 (*truncxfsf2_mixed): Ditto.
5151 (*truncxfdf2_mixed): Ditto.
5152 (*truncxf<mode>2_i387_noop): Ditto. Update callers
5153 to call gen_truncxf<mode>2 instead.
5154 (*truncxf<mode>2_i387): Remove.
5155 (reg->reg splitters): Remove splitter pattern.
5156 (reg->mem splitters): Ditto.
5158 (truncdfsf2): New insn pattern.
5159 (truncxf<mode>2): Ditto.
5161 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5163 * tree-ssa-math-opts.c (is_mult_by): New function.
5164 (is_square_of): Use the above.
5165 (optimize_recip_sqrt): New function.
5166 (pass_cse_reciprocals::execute): Use the above.
5168 2018-09-05 Richard Biener <rguenther@suse.de>
5171 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
5172 to zero-init the emplaced vec.
5174 2018-09-05 Martin Liska <mliska@suse.cz>
5176 PR tree-optimization/87205
5177 * tree-switch-conversion.c (pass_lower_switch::execute):
5178 Group cases for switch statements.
5180 2018-09-05 Richard Biener <rguenther@suse.de>
5182 PR tree-optimization/87217
5183 * tree-ssa-sccvn.c (vuse_valueize): New.
5184 (vn_reference_lookup_pieces): Use it.
5185 (vn_reference_lookup): Likewise.
5187 2018-09-05 Nathan Sidwell <nathan@acm.org>
5190 * stor-layout.c (place_field): Scan forwards to check last
5191 bitfield when ms_bitfield_placement is in effect.
5193 2018-09-05 Richard Biener <rguenther@suse.de>
5196 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
5199 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
5200 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
5202 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
5203 * config.gcc (extra_objs): Build it.
5204 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
5206 * config/aarch64/aarch64-passes.def
5207 (pass_tag_collision_avoidance): New pass.
5208 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
5209 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
5210 (aarch64_classify_address): Remove static qualifier.
5211 (aarch64_address_info, aarch64_address_type): Move to...
5212 * config/aarch64/aarch64-protos.h: ... here.
5213 (make_pass_tag_collision_avoidance): New function.
5214 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
5217 2018-09-05 Martin Liska <mliska@suse.cz>
5219 * doc/gcov.texi: Update documentation of humar
5221 * gcov.c (format_count): Print one decimal place, it provides
5222 more fine number of situations like '1G' vs. '1.4G'.
5224 2018-09-05 Martin Liska <mliska@suse.cz>
5227 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
5228 * optc-gen.awk: Allow 'Var' for Deprecated options in order
5229 to generate a MASK value.
5231 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
5234 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
5235 if frame pointer isn't used.
5236 (compute_frame_pointer_to_fb_displacement): Likewise.
5238 2018-09-04 Jakub Jelinek <jakub@redhat.com>
5241 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
5242 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
5243 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
5244 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
5245 and OPTION_MASK_ISA_XSAVEC_UNSET.
5247 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
5249 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
5250 XOR operations in NAND case.
5252 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
5254 * wide-int-range.cc (wide_int_range_convert): New.
5255 * wide-int-range.h (wide_int_range_convert): New.
5256 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
5257 code into wide_int_range_convert.
5258 (extract_range_into_wide_ints): Do not munge anti range constants
5259 into the entire domain. Just return the range back.
5261 2018-09-04 Martin Liska <mliska@suse.cz>
5263 * genmatch.c (output_line_directive): Add new argument
5265 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
5267 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
5269 * doc/invoke.texi (Option Summary): Add whitespace.
5271 * doc/invoke.texi (Option Summary): Add -Waligned-new.
5273 2018-09-04 Richard Biener <rguenther@suse.de>
5275 PR tree-optimization/87211
5276 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
5277 backedge value we're supposed to treat as VARYING also number
5278 the PHI to VARYING in case it got a different value-number already.
5280 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
5282 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
5283 (extract_range_from_binary_expr_1): Do not call
5284 vrp_can_optimize_bit_op.
5285 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
5287 (wide_int_range_get_mask_and_bounds): New.
5288 (wide_int_range_optimize_bit_op): New.
5289 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
5290 (wide_int_range_bit_and): Same.
5291 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
5292 (wide_int_range_optimize_bit_op): New.
5293 (wide_int_range_get_mask_and_bounds): New.
5295 2018-09-04 Richard Biener <rguenther@suse.de>
5297 PR tree-optimization/87176
5298 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
5299 variable. When value-numbering a virtual PHI node make sure
5300 to not value-number to the backedge value.
5302 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
5304 * doc/extend.texi (Long Long, Hex Floats): Document support for
5305 long long and hex floats in more recent versions of ISO C++.
5307 2018-09-03 Richard Biener <rguenther@suse.de>
5309 PR tree-optimization/87177
5310 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
5313 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5315 * bb-reorder.c (edge_order): Convert to C-qsort-style
5316 tri-state comparator.
5317 (reorder_basic_blocks_simple): Change std::stable_sort to
5320 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5322 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
5323 tri-state comparator.
5324 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
5326 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5328 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
5329 (mergesort): ... here as maximum count for using netsort.
5330 (gcc_qsort): Set nlim to 3 if stable sort is requested.
5331 (gcc_stablesort): New.
5332 * system.h (gcc_stablesort): Declare.
5334 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5336 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
5337 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
5338 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
5340 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
5342 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
5343 lxsdx and stxsdx alternatives.
5344 (*mov<mode>_hardfloat64): Ditto.
5345 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
5347 2018-09-03 Richard Biener <rguenther@suse.de>
5349 PR tree-optimization/87200
5350 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
5353 2018-09-03 Martin Liska <mliska@suse.cz>
5355 PR tree-optimization/87201
5356 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
5357 Fix parenthesis in an expression.
5359 2018-09-03 Richard Biener <rguenther@suse.de>
5361 PR tree-optimization/87197
5362 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
5363 visited. CSE the VN_INFO hashtable lookup.
5365 PR tree-optimization/87169
5366 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
5367 iterating make sure there's no extra backedges from irreducible
5368 regions feeding the header. Mark the destination block
5371 2018-09-03 Martin Liska <mliska@suse.cz>
5374 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
5375 * common/common-targhooks.c (default_get_valid_option_values):
5377 * common/common-targhooks.h (default_get_valid_option_values):
5379 * common/config/i386/i386-common.c: Move processor_target_table
5381 (ix86_get_valid_option_values): New function.
5382 (TARGET_GET_VALID_OPTION_VALUES): New macro.
5383 * config/i386/i386.c (struct ptt): Move to i386-common.c.
5384 (PTA_*): Move all defined masks into i386-common.c.
5385 (ix86_function_specific_restore): Use new processor_cost_table.
5386 * config/i386/i386.h (struct ptt): Moved from i386.c.
5387 (struct pta): Likewise.
5388 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
5389 * doc/tm.texi.in: Likewise.
5390 * opt-suggestions.c (option_proposer::suggest_option):
5391 Pass prefix to build_option_suggestions.
5392 (option_proposer::get_completions): Likewise.
5393 (option_proposer::build_option_suggestions): Use the new target
5395 * opts.c (struct option_help_tuple): New struct.
5396 (print_filtered_help): Use the new target hook.
5398 2018-09-03 Martin Liska <mliska@suse.cz>
5401 * predict.c (set_even_probabilities): Add likely_edges
5402 argument and handle cases where we have precisely one
5404 (combine_predictions_for_bb): Catch also likely_edges.
5405 (tree_predict_by_opcode): Handle gswitch statements.
5406 * tree-cfg.h (find_case_label_for_value): New declaration.
5407 (find_taken_edge_switch_expr): Likewise.
5408 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
5409 Find pivot in decision tree based on probabily, not by number of
5412 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
5414 * doc/standards.texi (Standards): Update Objective-C reference.
5416 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5418 * doc/install.texi (Prerequisites): Update link for MPC.
5420 2018-09-01 Michael Matz <matz@suse.de>
5422 PR tree-optimization/87074
5423 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
5424 PHIs for outer-loop uses.
5426 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5428 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
5429 * doc/invoke.texi (C Dialect Options): Ditto.
5431 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5433 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
5435 2018-08-31 Richard Biener <rguenther@suse.de>
5437 PR tree-optimization/87168
5438 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
5439 (rpo_elim::eliminate_avail): When OP was not visited it must
5442 2018-08-31 David Malcolm <dmalcolm@redhat.com>
5444 * tree-vrp.c (copy_value_range): Convert param "from" from
5445 "value_range *" to "const value_range *".
5446 (range_is_null): Likewise for param "vr".
5447 (range_int_cst_p): Likewise.
5448 (range_int_cst_singleton_p): Likewise.
5449 (symbolic_range_p): Likewise.
5450 (value_ranges_intersect_p): Likewise for both params.
5451 (value_range_nonnegative_p): Likewise for param "vr".
5452 (value_range_constant_singleton): Likewise.
5453 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
5454 (extract_range_into_wide_ints): Likewise for param "vr".
5455 (extract_range_from_multiplicative_op): Likewise for params "vr0"
5457 (vrp_can_optimize_bit_op): Likewise.
5458 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
5460 (extract_range_from_unary_expr): Likewise.
5461 (debug_value_range): Likewise for param "vr".
5462 (value_range::dump): Add "const" qualifier.
5463 (vrp_prop::check_array_ref): Convert local "vr" from
5464 "value_range *" to "const value_range *".
5465 (vrp_prop::check_mem_ref): Likewise.
5466 (vrp_prop::visit_stmt): Likewise for local "old_vr".
5467 (vrp_intersect_ranges_1): Likewise for param "vr_1".
5468 (vrp_intersect_ranges): Likewise.
5469 (simplify_stmt_for_jump_threading): Likewise for local "vr".
5470 (vrp_prop::vrp_finalize): Likewise.
5471 * tree-vrp.h (value_range::dump): Add "const" qualifier.
5472 (vrp_intersect_ranges): Add "const" qualifier to params as above.
5473 (extract_range_from_unary_expr): Likewise.
5474 (value_range_constant_singleton): Likewise.
5475 (symbolic_range_p): Likewise.
5476 (copy_value_range): Likewise.
5477 (extract_range_from_binary_expr_1): Likewise.
5478 (range_int_cst_p): Likewise.
5479 (vrp_set_zero_nonzero_bits): Likewise.
5480 (range_int_cst_singleton_p): Likewise.
5482 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
5484 * config/aarch64/arm_neon.h (vabsd_s64): New.
5485 (vnegd_s64): Likewise.
5487 2018-08-31 Martin Jambor <mjambor@suse.cz>
5489 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
5491 2018-08-31 Martin Liska <mliska@suse.cz>
5493 * ipa-icf.c (sem_item::add_type): Use
5494 sem_item::m_type_hash_cache.
5495 * ipa-icf.h: Move the cache from sem_item_optimizer
5498 2018-08-31 Nathan Sidwell <nathan@acm.org>
5500 * doc/extend.texi (Backwards Compatibility): Remove implicit
5501 extern C leeway of () being (...).
5503 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5505 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
5507 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
5511 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
5513 2018-08-31 Jakub Jelinek <jakub@redhat.com>
5516 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
5517 gen_int_mode. Formatting fixes.
5519 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
5521 * target.def (custom_function_descriptors): Improve documentation.
5522 * doc/tm.texi.in (Trampolines): Expand discussion of function
5523 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
5524 beginning of the section.
5525 * doc/tm.texi: Regenerated.
5527 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
5529 * cfg.h (class auto_edge_flag): Spell out the template-id of the
5530 base class in the initializer list. This is a workaround for
5531 building with older GCC.
5532 (class auto_bb_flag): Likewise.
5534 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
5536 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
5537 (altivec_vcmpequ<VI_char>_p): Remove star.
5538 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
5540 (expand_strncmp_vec_sequence): New function.
5541 (emit_final_str_compare_vec): New function.
5542 (expand_strn_compare): Add support for vector strncmp.
5543 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
5544 length specification to bytes.
5545 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
5546 (vcmpnezb_p): New pattern.
5547 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
5548 for option -mstring-compare-inline-limit.
5550 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
5552 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
5553 (PTA_SKYLAKE): Add PTA_AES.
5554 (PTA_GOLDMONT): Likewise.
5556 2018-08-29 Jan Hubicka <jh@suse.cz>
5559 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
5560 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
5562 2018-08-29 Jan Hubicka <jh@suse.cz>
5564 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
5566 (hash_tree): Do not visit TYPE_STUB_DECL.
5567 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
5568 stream TYPE_STUB_DECL.
5569 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
5570 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
5571 after free_lang_data.
5572 (type_in_anonymous_namespace_p): Likewise.
5574 2018-08-29 Jan Hubicka <jh@suse.cz>
5576 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5577 comment that it has to be even number.
5578 (class sreal): Change m_sig type to int32_t.
5579 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5580 int64_t for temporary calculations.
5581 (sreal_verify_basics): Drop one bit from minimum and maximum.
5583 2018-08-30 Richard Biener <rguenther@suse.de>
5585 PR tree-optimization/87147
5586 * tree-ssa-sccvn.c (SSA_VISITED): New function.
5587 (visit_phi): When the degenerate result is from the backedge and
5588 we didn't visit its definition yet drop to VARYING.
5589 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
5591 2018-08-29 Jan Hubicka <jh@suse.cz>
5593 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
5595 (hash_tree): Likewise.
5597 2018-08-29 Jan Hubicka <jh@suse.cz>
5599 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
5600 and TYPE_NEXT_REF_TO.
5602 2018-08-29 Jan Hubicka <jh@suse.cz>
5604 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5605 comment that it has to be even number.
5606 (class sreal): Change m_sig type to int32_t.
5607 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5608 int64_t for temporary calculations.
5609 (sreal_verify_basics): Drop one bit from minimum and maximum.
5611 2018-08-30 Tamar Christina <tamar.christina@arm.com>
5613 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
5615 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
5618 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
5619 if to_add is negative.
5621 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
5624 * builtins.c (c_strlen): Improve range checks.
5626 2018-08-29 Martin Sebor <msebor@redhat.com>
5627 Jeff Law <law@redhat.com>
5629 PR tree-optimization/86714
5630 PR tree-optimization/86711
5631 * builtins.c (c_strlen): Add arguments to call to string_constant.
5632 * expr.c (string_constant): Add argument. Detect missing nul
5633 terminator and outermost declaration it's missing in.
5634 * expr.h (string_constant): Add argument.
5635 * fold-const.c (read_from_constant_string): Add arguments to call to
5637 (c_getstr): Likewise.
5638 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5640 * tree-ssa-strlen.c (get_stridx): Likewise.
5642 2018-08-29 Jan Hubicka <jh@suse.cz>
5644 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
5645 Do not stream DECL_VINDEX.
5646 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
5647 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
5648 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
5650 2018-08-29 Richard Biener <rguenther@suse.de>
5652 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
5653 virtual operands that are not default defs to honor region
5655 (rpo_vn_valueize): Remove ineffective code here.
5657 2018-08-29 Richard Biener <rguenther@suse.de>
5659 PR tree-optimization/87132
5660 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
5661 when skipping defs reachable over backedges.
5663 2018-08-29 Richard Biener <rguenther@suse.de>
5665 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
5666 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
5667 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
5668 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
5669 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
5670 (vn_reference_lookup_3): Remove use of const_parms.
5671 (free_rpo_vn): Do not free const_parms.
5672 (do_rpo_vn): Do not call init_const_parms.
5673 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
5674 SSA_NAME_POINTS_TO_READONLY_MEMORY.
5675 (call_may_clobber_ref_p_1): Likewise.
5677 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
5680 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
5682 (-ftree-scev-cprop): Document.
5684 2018-08-29 Jan Hubicka <jh@suse.cz>
5686 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
5688 (sreal constructor): Update.
5689 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
5690 sreal:operator/): Update.
5692 2018-08-29 Martin Liska <mliska@suse.cz>
5694 * tree-switch-conversion.c (switch_conversion::expand):
5695 Strenghten assumption about gswitch statements.
5697 2018-08-29 Richard Biener <rguenther@suse.de>
5699 PR tree-optimization/87117
5700 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
5701 re-value-number released SSA VDEFs.
5703 2018-08-29 Richard Biener <rguenther@suse.de>
5705 PR tree-optimization/87126
5706 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
5708 2018-08-28 Jim Wilson <jimw@sifive.com>
5710 * config/riscv/pic.md: Rewrite.
5711 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
5713 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
5714 (SOFTF, default_load, softload, softstore): New.
5716 2018-08-28 Jeff Law <law@redhat.com>
5718 * fold-const.c (fold_binary_loc): Remove recently added assert.
5720 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
5722 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
5723 to OP parmeter of generated function.
5725 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
5727 PR tree-optimization/87009
5728 * match.pd: Add boolean optimizations.
5730 2018-08-28 Martin Sebor <msebor@redhat.com>
5733 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
5734 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
5735 (pass_walloca::gate): Use it.
5736 (alloca_call_type): Same.
5737 (pass_walloca::execute): Same.
5738 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
5740 2018-08-28 David Malcolm <dmalcolm@redhat.com>
5742 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
5743 GCC_VERSION for usage of "__gcc_dump_printf__" format from
5746 2018-08-28 Richard Biener <rguenther@suse.de>
5748 PR tree-optimization/87124
5749 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
5750 constants before looking up avail.
5752 2018-08-28 Jakub Jelinek <jakub@redhat.com>
5755 * calls.c (maybe_warn_nonstring_arg): Punt early if
5756 warn_stringop_overflow is zero. Don't call get_range_strlen
5757 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
5758 Swap comparison operands to have constants on rhs. Only use
5759 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
5760 increment lenrng[0].
5762 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
5764 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
5765 use of tree_to_shwi. Remove duplicated test for the size being
5766 a whole number of bytes.
5768 2018-08-28 Richard Biener <rguenther@suse.de>
5770 PR tree-optimization/87117
5771 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
5772 Handle removed stmt without LHS (GIMPLE_NOP).
5774 2018-08-28 Richard Biener <rguenther@suse.de>
5776 PR tree-optimization/87117
5777 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
5778 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
5780 2018-08-28 Richard Biener <rguenther@suse.de>
5782 PR tree-optimization/87117
5783 * tree-ssa-pre.c (compute_avail): Do not make expressions
5784 with predicated values available.
5785 (get_expr_value_id): Assert we do not run into predicated value
5788 2018-08-28 Richard Biener <rguenther@suse.de>
5790 PR tree-optimization/87117
5791 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
5792 get virtual operands.
5793 (get_expr_operands): Handle STRING_CST like other decls.
5795 2018-08-28 Martin Liska <mliska@suse.cz>
5797 * tree.h: Update documentation of fndecl_built_in_p
5801 2018-08-27 Jeff Law <law@redhat.com>
5802 PR tree-optimization/87110
5803 * tree-ssa-dse.c (compute_trims): Handle non-constant
5806 2018-08-27 Martin Sebor <msebor@redhat.com>
5808 PR tree-optimization/86914
5809 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
5811 2018-08-27 Martin Sebor <msebor@redhat.com>
5813 PR tree-optimization/87112
5814 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
5815 the type of the bound argument.
5817 2018-08-27 Jeff Law <law@redhat.com>
5819 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
5820 type does not have a TYPE_SIZE_UNIT.
5822 2018-08-27 Steve Ellcey <sellcey@cavium.com>
5824 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
5825 with include of backend.h.
5827 2018-08-27 Richard Biener <rguenther@suse.de>
5829 PR tree-optimization/86927
5830 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
5831 use const cond reduction code.
5833 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
5835 PR tree-optimization/85758
5836 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
5838 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5841 * diagnostic-show-locus.c (class layout_range): Update for
5842 conversion of show_caret_p to a tri-state.
5843 (layout_range::layout_range): Likewise.
5844 (make_range): Likewise.
5845 (layout::maybe_add_location_range): Likewise.
5846 (layout::should_print_annotation_line_p): Don't show annotation
5847 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
5848 (layout::get_state_at_point): Update for conversion of
5849 show_caret_p to a tri-state. Bail out early for
5850 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
5851 underlining or source colorization.
5852 (gcc_rich_location::add_location_if_nearby): Update for conversion
5853 of show_caret_p to a tri-state.
5854 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
5855 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
5857 (selftest::test_one_liner_labels): Likewise.
5858 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
5859 conversion of show_caret_p to a tri-state.
5860 * pretty-print.c (text_info::set_location): Likewise.
5861 * pretty-print.h (text_info::set_location): Likewise.
5862 * substring-locations.c (format_warning_n_va): Likewise.
5863 * tree-diagnostic.c (default_tree_printer): Likewise.
5864 * tree-pretty-print.c (newline_and_indent): Likewise.
5866 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5869 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
5870 line above for line-insertion fix-it hints.
5871 (selftest::test_fixit_insert_containing_newline): Update the
5872 expected results, and add a test with line-numbering enabled.
5874 2018-08-27 Martin Liska <mliska@suse.cz>
5877 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
5878 params with DECL_HAS_VALUE_EXPR_P.
5880 2018-08-27 Martin Liska <mliska@suse.cz>
5882 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
5883 selected expansion strategy.
5885 2018-08-27 Martin Liska <mliska@suse.cz>
5887 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
5888 * builtins.c (is_builtin_fn): Likewise.
5889 * attribs.c (diag_attr_exclusions): Use new function
5890 fndecl_built_in_p and remove check for FUNCTION_DECL if
5892 (builtin_mathfn_code): Likewise.
5893 (fold_builtin_expect): Likewise.
5894 (fold_call_expr): Likewise.
5895 (fold_builtin_call_array): Likewise.
5896 (fold_call_stmt): Likewise.
5897 (set_builtin_user_assembler_name): Likewise.
5898 (is_simple_builtin): Likewise.
5899 * calls.c (gimple_alloca_call_p): Likewise.
5900 (maybe_warn_nonstring_arg): Likewise.
5901 * cfgexpand.c (expand_call_stmt): Likewise.
5902 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
5903 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
5904 (cgraph_node::verify_node): Likewise.
5905 * cgraphclones.c (build_function_decl_skip_args): Likewise.
5906 (cgraph_node::create_clone): Likewise.
5907 * config/arm/arm.c (arm_insert_attributes): Likewise.
5908 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
5909 * dse.c (scan_insn): Likewise.
5910 * expr.c (expand_expr_real_1): Likewise.
5911 * fold-const.c (operand_equal_p): Likewise.
5912 (fold_binary_loc): Likewise.
5913 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5914 * gimple-low.c (lower_stmt): Likewise.
5915 * gimple-pretty-print.c (dump_gimple_call): Likewise.
5916 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
5918 * gimple.c (gimple_build_call_from_tree): Likewise.
5919 (gimple_call_builtin_p): Likewise.
5920 (gimple_call_combined_fn): Likewise.
5921 * gimplify.c (gimplify_call_expr): Likewise.
5922 (gimple_boolify): Likewise.
5923 (gimplify_modify_expr): Likewise.
5924 (gimplify_addr_expr): Likewise.
5925 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
5926 * ipa-cp.c (determine_versionability): Likewise.
5927 * ipa-fnsummary.c (compute_fn_summary): Likewise.
5928 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
5929 * ipa-split.c (visit_bb): Likewise.
5930 (split_function): Likewise.
5931 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
5932 * lto-cgraph.c (input_node): Likewise.
5933 * lto-streamer-out.c (write_symbol): Likewise.
5934 * omp-low.c (setjmp_or_longjmp_p): Likewise.
5935 (lower_omp_1): Likewise.
5936 * predict.c (strip_predict_hints): Likewise.
5937 * print-tree.c (print_node): Likewise.
5938 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
5939 * trans-mem.c (is_tm_irrevocable): Likewise.
5940 (is_tm_load): Likewise.
5941 (is_tm_simple_load): Likewise.
5942 (is_tm_store): Likewise.
5943 (is_tm_simple_store): Likewise.
5944 (is_tm_abort): Likewise.
5945 (tm_region_init_1): Likewise.
5946 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
5947 * tree-cfg.c (verify_gimple_call): Likewise.
5948 (move_stmt_r): Likewise.
5949 (stmt_can_terminate_bb_p): Likewise.
5950 * tree-eh.c (lower_eh_constructs_2): Likewise.
5951 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
5952 * tree-inline.c (remap_gimple_stmt): Likewise.
5953 (copy_bb): Likewise.
5954 (estimate_num_insns): Likewise.
5955 (fold_marked_statements): Likewise.
5956 * tree-sra.c (scan_function): Likewise.
5957 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
5958 (optimize_stack_restore): Likewise.
5959 (pass_fold_builtins::execute): Likewise.
5960 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5961 (mark_all_reaching_defs_necessary_1): Likewise.
5962 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
5963 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5964 (pass_forwprop::execute): Likewise.
5965 * tree-ssa-loop-im.c (stmt_cost): Likewise.
5966 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5967 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
5968 * tree-ssa-strlen.c (get_string_length): Likewise.
5969 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
5970 (find_func_aliases_for_call): Likewise.
5971 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
5972 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
5973 * tree-tailcall.c (find_tail_calls): Likewise.
5974 * tree.c (need_assembler_name_p): Likewise.
5975 (free_lang_data_in_decl): Likewise.
5976 (get_call_combined_fn): Likewise.
5977 * ubsan.c (is_ubsan_builtin_p): Likewise.
5978 * varasm.c (incorporeal_function_p): Likewise.
5979 * tree.h (DECL_BUILT_IN): Remove and replace with
5981 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
5982 (fndecl_built_in_p): New.
5984 2018-08-27 Martin Liska <mliska@suse.cz>
5986 PR tree-optimization/86847
5987 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
5988 Dump also subtree probability.
5989 (switch_decision_tree::do_jump_if_equal): New function.
5990 (switch_decision_tree::emit_case_nodes): Handle special
5991 situations in balanced tree that can be emitted much simpler.
5992 Fix calculation of probabilities that happen in tree expansion.
5993 * tree-switch-conversion.h (struct cluster): Add
5995 (struct simple_cluster): Likewise.
5996 (struct case_tree_node): Add new function has_child.
5997 (do_jump_if_equal): New.
5999 2018-08-27 Martin Liska <mliska@suse.cz>
6001 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
6002 Add new argument to bit_test_cluster constructor.
6003 (bit_test_cluster::emit): Set bits really number of values
6005 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
6006 probability argument.
6007 * tree-switch-conversion.h (struct bit_test_cluster):
6008 Add m_handles_entire_switch.
6010 2018-08-27 Martin Liska <mliska@suse.cz>
6012 PR tree-optimization/86702
6013 * tree-switch-conversion.c (jump_table_cluster::emit):
6014 Make probabilities even for values in jump table
6015 according to number of cases handled.
6016 (switch_decision_tree::compute_cases_per_edge): Pass
6017 argument to reset_out_edges_aux function.
6018 (switch_decision_tree::analyze_switch_statement): Likewise.
6019 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
6022 2018-08-27 Martin Liska <mliska@suse.cz>
6024 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
6025 cfun argument explicitly.
6026 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
6027 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
6028 function gimple_switch_default_bb.
6029 (convert_switch_statements):
6031 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
6032 * stmt.c (label_to_block_fn): Use label_to_block and pass
6033 cfun argument explicitly and use gimple_switch_label_bb.
6034 (expand_case): Likewise.
6035 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
6036 cfun argument explicitly. Likewise.
6037 (make_edges_bb): Likewise.
6038 (make_cond_expr_edges): Likewise.
6039 (get_cases_for_edge): Likewise.
6040 (make_gimple_switch_edges): Likewise.
6041 (label_to_block_fn): Likewise.
6042 (label_to_block): Likewise.
6043 (make_goto_expr_edges): Likewise.
6044 (make_gimple_asm_edges): Likewise.
6045 (main_block_label): Likewise.
6046 (group_case_labels_stmt): Likewise.
6047 (find_taken_edge_computed_goto): Likewise.
6048 (find_taken_edge_switch_expr): Likewise.
6049 (gimple_verify_flow_info): Likewise.
6050 (gimple_redirect_edge_and_branch): Likewise.
6051 (gimple_switch_label_bb): New function.
6052 (gimple_switch_default_bb): Likewise.
6053 (gimple_switch_edge): Likewise.
6054 (gimple_switch_default_edge): Likewise.
6055 * tree-cfg.h (label_to_block_fn): Remove and replace ...
6056 (label_to_block): ... with this.
6057 (gimple_switch_label_bb): New.
6058 (gimple_switch_default_bb): Likewise.
6059 (gimple_switch_edge): Likewise.
6060 (gimple_switch_default_edge): Likewise.
6061 * tree-cfgcleanup.c (convert_single_case_switch): Use
6062 new gimple functions and pass new argument to label_to_block.
6063 (cleanup_control_flow_bb):
6064 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
6065 cfun argument explicitly.
6066 (make_eh_edges): Likewise.
6067 (redirect_eh_dispatch_edge): Likewise.
6068 (lower_resx): Likewise.
6069 (lower_eh_dispatch): Likewise.
6070 (maybe_remove_unreachable_handlers): Likewise.
6071 (unsplit_eh): Likewise.
6072 (cleanup_empty_eh): Likewise.
6073 (verify_eh_edges): Likewise.
6074 (verify_eh_dispatch_edge): Likewise.
6075 * tree-ssa-dom.c (record_edge_info): Likewise.
6076 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
6077 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
6078 (thread_through_normal_block): Likewise.
6079 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
6080 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
6081 * tree-switch-conversion.c (switch_conversion::collect): Use new
6083 (switch_conversion::check_final_bb): Likewise.
6084 (switch_conversion::gather_default_values): Pass new argument
6086 (switch_conversion::build_constructors): Likewise.
6087 (switch_decision_tree::compute_cases_per_edge): Use new
6088 gimple_switch_edge function.
6089 (switch_decision_tree::analyze_switch_statement): Pass new argument
6091 (switch_decision_tree::try_switch_expansion): Use
6092 gimple_switch_default_edge.
6093 * tree-vrp.c (find_switch_asserts): Pass new argument
6095 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
6096 (vr_values::simplify_switch_using_ranges): Likewise.
6098 2018-08-27 Richard Biener <rguenther@suse.de>
6100 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
6101 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
6103 * tree-ssa-sccvn.h (struct vn_pval): New structure.
6104 (struct vn_nary_op_s): Add unwind_to member. Add
6105 predicated_values flag and put result into a union together
6106 with a linked list of vn_pval.
6107 (struct vn_ssa_aux): Add name member to make maintaining
6108 a map of SSA name to vn_ssa_aux possible. Remove no longer
6109 needed info, dfsnum, low, visited, on_sccstack, use_processed
6110 and range_info_anti_range_p members.
6111 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
6112 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
6114 (vn_valueize): New global.
6115 (vn_context_bb): Likewise.
6116 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
6117 VN_INFO_PTR_INFO): Remove.
6118 * tree-ssa-sccvn.c: ... (rewrite)
6119 (pass_fre::execute): For -O2+ initialize loops and run
6120 RPO VN in optimistic mode (iterating). For -O1 and -Og
6121 run RPO VN in non-optimistic mode.
6122 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
6123 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
6124 * doc/invoke.texi (sccvn-max-scc-size): Remove.
6125 (rpo-vn-max-loop-depth): Document.
6126 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
6127 when valuezing the VUSE signals we walked out of the region.
6128 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
6129 (phi_translate): Set VN context block to use for availability
6131 (compute_avail): Likewise.
6132 (pre_valueize): New function.
6133 (pass_pre::execute): Adjust to the RPO VN API.
6135 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
6136 (propagate_constants_for_unrolling): Remove.
6137 (tree_unroll_loops_completely): Perform value-numbering
6138 on the unrolled bodies loop parent.
6140 2018-08-27 Richard Biener <rguenther@suse.de>
6142 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
6143 for partial antic compute.
6145 2018-08-27 Jakub Jelinek <jakub@redhat.com>
6147 PR rtl-optimization/87065
6148 * combine.c (simplify_if_then_else): Formatting fix.
6149 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
6151 (known_cond): Don't return const_true_rtx for vector modes. Use
6152 CONST0_RTX instead of const0_rtx. Formatting fixes.
6154 2018-08-27 Martin Liska <mliska@suse.cz>
6156 PR gcov-profile/87069
6157 * gcov.c (process_file): Record files already processed
6158 and warn about a file being processed multiple times.
6160 2018-08-27 Martin Liska <mliska@suse.cz>
6163 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6164 Set default values for x_aarch64_*_string strings.
6165 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
6166 prefix. For -mabi do not print '=ABI' in help and use
6167 <option_value> format for -msve-vector-bits and -moverride
6170 2018-08-26 Jeff Law <law@redhat.com>
6172 * config/mips/frame-header-opt.c: Include "backend.h" rather than
6175 2018-08-26 Marek Polacek <polacek@redhat.com>
6177 PR c++/87029, Implement -Wredundant-move.
6178 * doc/invoke.texi: Document -Wredundant-move.
6180 2018-08-25 Martin Sebor <msebor@redhat.com>
6182 PR tree-optimization/87059
6183 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
6184 to the same type as the other.
6185 * fold-const.c (fold_binary_loc): Assert expectation.
6187 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
6189 * config/darwin.c (machopic_legitimize_pic_address): Clean up
6190 extraneous parentheses, dead code section and formatting.
6192 2018-08-24 David Malcolm <dmalcolm@redhat.com>
6195 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
6196 wide enough for jumps in the line-numbering to be visible.
6197 (layout::print_gap_in_line_numbering): New member function.
6198 (layout::calculate_line_spans): When using line numbering, merge
6199 line spans that are only 1 line apart.
6200 (diagnostic_show_locus): When printing line numbers, show gaps in
6201 line numbering directly, rather than printing headers.
6202 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
6203 line-numbering with multiple line spans.
6204 (selftest::test_fixit_insert_containing_newline_2): Add test of
6205 line-numbering, in which the spans are close enough to be merged.
6207 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
6209 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
6210 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
6211 * tree-vrp.c (range_is_nonnull): Remove.
6212 (range_includes_zero_p): Accept value_range instead of min/max.
6213 (extract_range_from_binary_expr_1): Do not early bail on
6215 Use range_includes_zero_p instead of range_is_nonnull.
6216 (extract_range_from_unary_expr): Use range_includes_zero_p instead
6217 of range_is_nonnull.
6218 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
6219 special case VR_ANTI_RANGE.
6220 (vrp_finalize): Same.
6221 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
6223 (range_is_nonnull): Remove.
6224 * vr-values.c (vrp_stmt_computes_nonzero): Use
6225 range_includes_zero_p instead of range_is_nonnull.
6226 (extract_range_basic): Pass value_range to range_includes_zero_p
6227 instead of range_is_nonnull.
6229 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
6231 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
6232 * emit-rtl.h (rtl_data): Remove return_bnd.
6233 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
6234 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
6235 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
6236 (POINTER_BOUNDS_MODE): Remove definition.
6237 (make_pointer_bounds_mode): Remove.
6238 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
6239 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
6240 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
6241 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
6242 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
6243 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
6244 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
6246 * config/i386/i386-modes.def (BND32, BND64): Remove.
6247 * config/i386/i386.c (dbx_register_map): Remove bound registers.
6248 (dbx64_register_map): Ditto.
6249 (svr4_dbx_register_map): Ditto.
6250 (indirect_thunk_bnd_needed): Remove.
6251 (indirect_thunks_bnd_used): Ditto.
6252 (indirect_return_bnd_needed): Ditto.
6253 (indirect_return_via_cx_bnd): Ditto.
6254 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
6255 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
6256 (output_indirect_thunk): Ditto. Remove need_prefix argument.
6257 (output_indirect_thunk_function): Remove handling of
6258 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
6259 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
6260 (ix86_save_reg): Remove handling of crtl->return_bnd.
6261 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
6262 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
6263 and UNSPEC_BNDLX_ADDR.
6264 (ix86_output_indirect_branch_via_reg): Remove handling of
6265 indirect_thunk_prefix_bnd.
6266 (ix86_output_indirect_branch_via_push): Ditto.
6267 (ix86_output_function_return): Ditto.
6268 (ix86_output_indirect_function_return): Ditto.
6269 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
6270 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
6271 (CALL_USED_REGISTERS): Ditto.
6272 (REG_ALLOC_ORDER): Update for removal of bound registers.
6273 (HI_REGISTER_NAMES): Ditto.
6274 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
6275 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
6276 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
6277 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
6278 (FIRST_PSEUDO_REG): Update.
6279 (BND): Remove mode iterator.
6280 * config/i386/predicates.md (bnd_mem_operator): Remove.
6282 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
6284 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
6287 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
6289 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
6290 the case in which the permute needs only a single element and
6291 repeats for every vector of the result. Extend that case to
6292 handle variable-length vectors.
6293 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
6295 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
6298 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
6299 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
6301 2018-08-24 Richard Biener <rguenther@suse.de>
6303 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
6304 bb_flags_allocated members.
6305 (auto_flag): New RAII class for allocating flags.
6306 (auto_edge_flag): New RAII class for allocating edge flags.
6307 (auto_bb_flag): New RAII class for allocating bb flags.
6308 * cfgloop.c (verify_loop_structure): Allocate temporary edge
6310 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
6311 in favor of temporarily allocated BB flag.
6312 * hsa-brig.c: Re-order includes.
6313 * hsa-dump.c: Likewise.
6314 * hsa-regalloc.c: Likewise.
6315 * print-rtl.c: Likewise.
6316 * profile-count.c: Likewise.
6318 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
6321 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
6324 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
6327 * wide-int-range.cc (wide_int_range_div): Do not ignore result
6328 from wide_int_range_multiplicative_op.
6330 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6332 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
6335 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
6337 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
6340 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
6342 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
6343 full GENERIC expression used for replacement.
6345 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
6347 * tree-vrp.c (abs_extent_range): Remove.
6348 (extract_range_into_wide_ints): Pass wide ints by reference.
6349 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
6350 Pass wide ints by reference in all calls to
6351 extract_range_into_wide_ints.
6352 * wide-int-range.cc (wide_int_range_div): New.
6353 * wide-int-range.h (wide_int_range_div): New.
6354 (wide_int_range_includes_zero_p): New.
6355 (wide_int_range_zero_p): New.
6357 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
6359 * config/aarch64/aarch64.md (arches): New enum.
6360 (arch): New enum attr.
6361 (arch_enabled): New attr.
6362 (enabled): Now uses arch_enabled only.
6363 (simd, sve, fp16): Removed attribute.
6364 (fp): Attr now defined in terms of 'arch'.
6365 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
6366 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
6367 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
6368 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
6369 attributes into 'arch'.
6370 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
6371 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
6372 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
6373 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
6374 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
6375 'simd' attribute into 'arch'.
6376 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
6377 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
6378 Convert use of 'fp' attribute to 'arch'.
6379 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
6380 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
6382 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
6383 (different modes) Merge 'fp' and 'simd' into 'arch'.
6384 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
6387 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
6389 PR rtl-optimization/87026
6390 * expmed.c (canonicalize_comparison): If we can no longer create
6391 pseudoregisters, don't.
6393 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
6396 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
6398 * config/arm/arm.c (speculation_barrier_libfunc): New static
6400 (arm_init_libfuncs): Initialize it.
6401 (arm_emit_speculation_barrier): New function.
6402 * config/arm/arm.md (speculation_barrier): Call
6403 arm_emit_speculation_barrier for architectures that do not have
6405 (speculation_barrier_insn): Only match on Armv7 or later.
6407 2018-08-23 Richard Biener <rguenther@suse.de>
6410 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
6413 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
6415 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
6416 of single-vector TBLs.
6417 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
6420 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
6423 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
6424 aarch64_evpc_tbl guard.
6426 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
6428 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
6431 2018-08-22 Martin Sebor <msebor@redhat.com>
6434 * tree-pretty-print.c (pretty_print_string): Add argument.
6435 (dump_generic_node): Call to pretty_print_string with string size.
6437 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
6439 PR rtl-optimization/86771
6440 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
6441 of two SETs into those two SETs, one to be placed at i2, if that SETs
6442 destination is modified between i2 and i3.
6444 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6446 PR tree-optimization/86725
6447 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
6449 (vect_analyze_scalar_cycles_1): Check it.
6451 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6453 PR tree-optimization/86725
6454 * tree-vect-loop.c (vect_is_simple_reduction): When treating
6455 an outer loop phi as a double reduction, make sure that the
6456 single user of the phi result is an inner loop phi.
6458 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6460 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
6461 grouped stores with gaps to a strided group.
6463 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6465 * tree-vect-stmts.c (get_group_load_store_type)
6466 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
6467 first statement in a group.
6469 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6471 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
6472 the sequence used in gcc/gcc.c.
6474 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6477 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
6480 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6482 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
6483 Darwin10-specific unwinder-shim.
6484 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
6485 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
6486 New to cater for Darwin10 Rosetta.
6488 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6490 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
6491 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
6493 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6498 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
6499 (dwarf2out_switch_text_section): Generate a local label for the
6500 second function sub-section and apply it as the second FDE start
6502 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
6503 second sub-section start.
6505 2018-08-22 Richard Biener <rguenther@suse.de>
6507 PR tree-optimization/86988
6508 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
6510 2018-08-22 Richard Biener <rguenther@suse.de>
6512 PR tree-optimization/86945
6513 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
6515 2018-08-22 Alexandre Oliva <oliva@adacore.com>
6517 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
6518 a comment about how uses of r2 for .sdata2 come about.
6520 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
6522 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
6524 2018-08-21 Marek Polacek <polacek@redhat.com>
6526 PR c++/86981, Implement -Wpessimizing-move.
6527 * doc/invoke.texi: Document -Wpessimizing-move.
6529 2018-08-21 Jan Hubicka <jh@suse.cz>
6531 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
6532 * tree.h (is_redundant_typedef): Remove.
6533 * dwarf2out.c (is_redundant_typedef): Turn into static function.
6535 2018-08-21 Jan Hubicka <jh@suse.cz>
6537 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
6540 2018-08-21 Tamar Christina <tamar.christina@arm.com>
6542 * expmed.c (extract_low_bits): Reject invalid subregs early.
6544 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6547 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
6550 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6552 * config/vxworks.h: Guard vxworks_asm_out_constructor and
6553 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
6554 * config/vxworks.c: Likewise.
6556 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6558 * config/vxworks.c: Set targetm.have_ctors_dtors
6559 if HAVE_INITFINI_ARRAY_SUPPORT.
6560 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
6561 if HAVE_INITFINI_ARRAY_SUPPORT.
6563 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6565 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
6566 default search path for VxWorks < 7.
6568 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6570 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
6571 (get_format_string): Refer to c_getstr.
6573 2018-08-21 Tom de Vries <tdevries@suse.de>
6575 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6576 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6577 (debuginfo_early_stop): Declare.
6578 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6579 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6580 (debuginfo_early_stop): New function.
6581 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
6582 and debuginfo_early_stop.
6583 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
6584 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
6585 (general_init): Call debuginfo_early_init.
6586 (finalize): Call debuginfo_fini.
6587 (do_compile): Call debuginfo_init.
6588 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
6590 (@item -fdump-debug, @item -fdump-earlydebug): Add.
6592 2018-08-21 Tom de Vries <tdevries@suse.de>
6594 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
6595 flag_dump_noaddr and flag_dump_unnumbered.
6597 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
6599 * wide-int-range.cc (wide_int_range_abs): New.
6600 (wide_int_range_order_set): Rename from wide_int_range_min_max.
6601 * wide-int-range.h (wide_int_range_abs): New.
6602 (wide_int_range_min_max): New.
6603 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
6604 case to call wide_int_range_abs.
6605 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
6606 (extract_range_from_abs_expr): Delete.
6608 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
6611 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
6612 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
6615 2018-08-20 Nathan Sidwell <nathan@acm.org>
6616 Jeff Law <law@redhat.com>
6618 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
6619 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
6621 2018-08-20 David Malcolm <dmalcolm@redhat.com>
6624 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
6625 (decl_attributes): Likewise.
6626 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
6628 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
6629 * diagnostic-core.h (class auto_diagnostic_group): New class.
6630 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
6631 (diagnostic_report_diagnostic): Handle the first diagnostics within
6633 (emit_diagnostic): Add auto_diagnostic_group instance.
6635 (inform_n): Likewise.
6636 (warning): Likewise.
6637 (warning_at): Likewise.
6638 (warning_n): Likewise.
6639 (pedwarn): Likewise.
6640 (permerror): Likewise.
6642 (error_n): Likewise.
6643 (error_at): Likewise.
6645 (fatal_error): Likewise.
6646 (internal_error): Likewise.
6647 (internal_error_no_backtrace): Likewise.
6648 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
6649 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
6650 * diagnostic.h (struct diagnostic_context): Add fields
6651 "diagnostic_group_nesting_depth",
6652 "diagnostic_group_emission_count", "begin_group_cb",
6654 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
6655 Add auto_diagnostic_group instance(s).
6656 (find_explicit_erroneous_behavior): Likewise.
6657 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
6658 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
6659 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
6660 (gimplify_va_arg_expr): Likewise.
6661 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
6662 (HSA_SORRY_AT): Likewise.
6663 * ipa-devirt.c (compare_virtual_tables): Likewise.
6664 (warn_odr): Likewise.
6665 * multiple_target.c (expand_target_clones): Likewise.
6666 * opts-common.c (cmdline_handle_error): Likewise.
6667 * reginfo.c (globalize_reg): Likewise.
6668 * substring-locations.c (format_warning_n_va): Likewise.
6669 * tree-inline.c (expand_call_inline): Likewise.
6670 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
6671 * tree-ssa-loop-niter.c
6672 (do_warn_aggressive_loop_optimizations): Likewise.
6673 * tree-ssa-uninit.c (warn_uninit): Likewise.
6674 * tree.c (warn_deprecated_use): Likewise.
6676 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
6679 * config/i386/i386.md (eh_return): Always update EH return
6680 address in word_mode.
6682 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
6684 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
6685 TARGET_SPLIT_COMPLEX_ARG is defined.
6687 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6689 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
6691 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6694 * expr.c (expand_assignment): Assert that bitpos is positive.
6695 (store_field): Likewise
6696 (expand_expr_real_1): Make sure that bitpos is positive.
6697 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
6700 2018-08-20 Nathan Sidwell <nathan@acm.org>
6702 * Makefile.in (CPP_ID_DATA_H): Delete.
6703 (CPP_INTERNAL_H): Don't add it.
6704 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
6705 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
6707 2018-08-20 Richard Biener <rguenther@suse.de>
6709 PR tree-optimization/78655
6710 * tree-vrp.c (extract_range_from_binary_expr_1): Make
6711 pointer + offset nonnull if either operand is nonnull work.
6713 2018-08-20 Tom de Vries <tdevries@suse.de>
6715 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
6716 unless the referenced die describes the added property using
6717 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
6718 Otherwise, add a DW_AT_location to the referenced die.
6720 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
6723 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
6724 register_operand when calling ix86_set_reg_reg_cost.
6725 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
6726 Set *total to 0 for operands that satisfy x86_64_immediate_operand
6727 predicate and to 1 otherwise.
6729 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
6731 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
6732 emit a diagnostic that it is not supported and reset the option.
6733 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
6734 supported and consume the option. (ASM_FINAL_SPEC): New.
6736 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6738 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
6741 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
6743 C-SKY port: Documentation
6745 * doc/extend.texi (C-SKY Function Attributes): New section.
6746 * doc/invoke.texi (Option Summary): Add C-SKY options.
6747 (C-SKY Options): New section.
6748 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
6750 2018-08-17 Jojo <jijie_rong@c-sky.com>
6751 Huibin Wang <huibin_wang@c-sky.com>
6752 Sandra Loosemore <sandra@codesourcery.com>
6753 Chung-Lin Tang <cltang@codesourcery.com>
6755 C-SKY port: Backend implementation
6757 * config/csky/*: New.
6758 * common/config/csky/*: New.
6760 2018-08-17 Jojo <jijie_rong@c-sky.com>
6761 Huibin Wang <huibin_wang@c-sky.com>
6762 Sandra Loosemore <sandra@codesourcery.com>
6763 Chung-Lin Tang <cltang@codesourcery.com>
6764 Andrew Jenner <andrew@codesourcery.com>
6766 C-SKY port: Configury
6768 * config.gcc (csky-*-*): New.
6769 * configure.ac: Add csky to targets for dwarf2 debug_line support.
6770 * configure: Regenerated.
6772 2018-08-17 David Malcolm <dmalcolm@redhat.com>
6774 * dump-context.h: Include "dumpfile.h".
6775 (dump_context::dump_printf_va): Convert final param from va_list
6776 to va_list *. Convert from ATTRIBUTE_PRINTF to
6777 ATTRIBUTE_GCC_DUMP_PRINTF.
6778 (dump_context::dump_printf_loc_va): Likewise.
6779 * dumpfile.c: Include "stringpool.h".
6780 (make_item_for_dump_printf_va): Delete.
6781 (make_item_for_dump_printf): Delete.
6782 (class dump_pretty_printer): New class.
6783 (dump_pretty_printer::dump_pretty_printer): New ctor.
6784 (dump_pretty_printer::emit_items): New member function.
6785 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
6787 (dump_pretty_printer::emit_item): New member function.
6788 (dump_pretty_printer::stash_item): New member function.
6789 (dump_pretty_printer::format_decoder_cb): New member function.
6790 (dump_pretty_printer::decode_format): New member function.
6791 (dump_context::dump_printf_va): Reimplement in terms of
6792 dump_pretty_printer.
6793 (dump_context::dump_printf_loc_va): Convert final param from va_list
6795 (dump_context::begin_scope): Reimplement call to
6796 make_item_for_dump_printf.
6797 (dump_printf): Update for change to dump_printf_va.
6798 (dump_printf_loc): Likewise.
6799 (selftest::test_capture_of_dump_calls): Convert "stmt" from
6800 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
6801 with %T, %E, and %G.
6802 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
6803 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
6804 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
6805 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
6806 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
6807 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
6808 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
6809 within a dump_printf_loc call to "%wu".
6810 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
6811 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
6812 missing space after "=".
6813 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
6814 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
6815 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
6816 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
6817 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
6818 duplicate "vectorized" from message.
6820 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
6822 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
6823 polyNxK_t element's TYPE_STRING_FLAG.
6825 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6827 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
6828 (they were unnamed before). Fix comments.
6830 2018-08-17 Nathan Sidwell <nathan@acm.org>
6832 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
6834 2018-08-17 Richard Biener <rguenther@suse.de>
6836 PR tree-optimization/86841
6837 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
6839 2018-08-17 Martin Liska <mliska@suse.cz>
6841 * common.opt: Remove Warn, Init and Report for options with
6842 Ignore/Deprecated flag. Warning is done automatically for
6844 * config/i386/i386.opt: Likewise.
6845 * config/ia64/ia64.opt: Likewise.
6846 * config/rs6000/rs6000.opt: Likewise.
6847 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
6848 Remove usage of flag_check_pointer_bounds.
6849 * lto-wrapper.c (merge_and_complain): Do not handle
6850 OPT_fcheck_pointer_bounds.
6851 (append_compiler_options): Likewise.
6852 * opt-functions.awk: Do not handle Deprecated.
6853 * optc-gen.awk: Check that Var, Report and Init are not
6854 used for an option with Ignore/Deprecated flag.
6855 * opts-common.c (decode_cmdline_option): Do not report
6857 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
6859 * opts.h (struct cl_option): Remove cl_deprecated flag.
6860 (CL_ERR_DEPRECATED): Remove error enum value.
6862 2018-08-17 Richard Biener <rguenther@suse.de>
6865 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
6866 across a va-arg-pack using call adjust its return value accordingly.
6868 2018-08-16 Martin Sebor <msebor@redhat.com>
6870 PR tree-optimization/86853
6871 * gimple-ssa-sprintf.c (struct format_result): Rename member.
6872 (struct fmtresult): Add member and initialize it in ctors.
6873 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
6874 (format_string): Handle %S the same as %ls. Set MAYFAIL.
6875 (format_directive): Set POSUNDER4K when MAYFAIL is set.
6876 (parse_directive): Handle %C same as %c.
6877 (sprintf_dom_walker::compute_format_length): Adjust.
6878 (is_call_safe): Adjust.
6880 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6882 * builtins.c (c_strlen): Add new parameter eltsize. Use it
6883 for determining how to count the elements.
6884 * builtins.h (c_strlen): Adjust prototype.
6885 * expr.c (string_constant): Add new parameter mem_size.
6886 Set *mem_size appropriately.
6887 * expr.h (string_constant): Adjust protoype.
6888 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
6889 * gimple-fold.h (get_range_strlen): Adjust prototype.
6890 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
6891 (format_string): Call get_string_length with eltsize.
6893 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6895 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
6896 to emit the span, rather than setting it as the prefix.
6898 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6900 * diagnostic-show-locus.c (layout::start_annotation_line): Add
6901 "margin_char" parameter, defaulting to space. Use it in place
6902 of pp_space for the initial part of the margin.
6903 (layout::print_leading_fixits): Use '+' when filling the margin
6904 of line-insertion fix-it hints.
6906 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6908 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
6911 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6913 * config/rs6000/altivec.md: Don't set length attribute to the default
6915 * config/rs6000/darwin.md: Ditto.
6916 * config/rs6000/dfp.md: Ditto.
6917 * config/rs6000/htm.md: Ditto.
6918 * config/rs6000/rs6000.md: Ditto.
6919 * config/rs6000/sync.md: Ditto.
6920 * config/rs6000/vsx.md: Ditto.
6922 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6924 * config/rs6000/altivec.md: Don't set length attribute to the default
6925 value, for branch instructions.
6926 * config/rs6000/darwin.md: Ditto.
6927 * config/rs6000/rs6000.md: Ditto.
6929 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6931 * config/rs6000/rs6000.md (length): Always define as const_int 4.
6932 (unnamed conditional branch define_insn): Set length to 4 or 8
6933 depending on offset.
6934 (<bd>_<mode>): Similar, for alternative 0.
6935 (<bd>tf_<mode>): Ditto.
6937 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6939 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
6941 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
6943 * doc/rtl.texi: Replace old RTX class names with new names.
6946 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
6948 * expmed.h (canonicalize_comparison): New declaration.
6949 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
6950 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
6951 * optabs.c (prepare_cmp_insn): Likewise.
6952 * rtl.h (unsigned_condition_p): New function which checks if a
6953 comparison operator is unsigned.
6955 2018-08-16 Nathan Sidwell <nathan@acm.org>
6957 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
6958 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
6960 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6963 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
6964 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
6965 (mov<mov>): ..this and enable unconditionally.
6967 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6969 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
6971 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
6973 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
6974 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
6975 "Common" with "Target".
6977 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6979 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
6980 * doc/invoke.texi (mmitigate-rop): Remove.
6981 * config/i386/i386.c: Do not include "regrename.h".
6982 (ix86_rop_should_change_byte_p, reg_encoded_number)
6983 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
6985 (ix86_reorg): Remove call to ix86_mitigate_rop.
6986 * config/i386/i386.md (attr "modrm_class"): Remove.
6987 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
6988 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
6989 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
6991 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6993 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
6994 allow folding of mergeh() and mergel() for the float and double types.
6995 (fold_mergehl_helper): Rework to handle building a permute tree
6998 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
7000 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
7003 2018-08-15 David Malcolm <dmalcolm@redhat.com>
7005 * common.opt (fdiagnostics-show-labels): New option.
7006 * diagnostic-show-locus.c (class layout_range): Add field
7008 (class layout): Add field "m_show_labels_p".
7009 (layout_range::layout_range): Add param "label" and use it to
7011 (make_range): Pass in NULL for new "label" param of layout_range's
7013 (layout::layout): Initialize m_show_labels_p.
7014 (layout::maybe_add_location_range): Pass in loc_range->m_label
7015 when constructing layout_range instances.
7016 (struct line_label): New struct.
7017 (layout::print_any_labels): New member function.
7018 (layout::print_line): Call it if label-printing is enabled.
7019 (selftest::test_one_liner_labels): New test.
7020 (selftest::test_diagnostic_show_locus_one_liner): Call it.
7021 * diagnostic.c (diagnostic_initialize): Initialize
7022 context->show_labels_p.
7023 * diagnostic.h (struct diagnostic_context): Add field
7025 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7026 -fno-diagnostics-show-labels.
7027 * dwarf2out.c (gen_producer_string): Add
7028 OPT_fdiagnostics_show_labels to the ignored options.
7029 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
7031 (gcc_rich_location::maybe_add_expr): Likewise.
7032 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
7033 label" param, defaulting to NULL.
7034 (gcc_rich_location::add_expr): Add "label" param.
7035 (gcc_rich_location::maybe_add_expr): Likewise.
7036 (class text_range_label): New class.
7037 (class range_label_for_type_mismatch): New class.
7038 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
7039 of format_warning_va.
7040 (fmtwarn_n): Likewise for new params of format_warning_n_va.
7041 * lto-wrapper.c (merge_and_complain): Add
7042 OPT_fdiagnostics_show_labels to the "pick one setting" options.
7043 (append_compiler_options): Likewise to the dropped options.
7044 (append_diag_options): Likewise to the passed-on options.
7045 * opts.c (common_handle_option): Handle the new option.
7046 * selftest-diagnostic.c
7047 (test_diagnostic_context::test_diagnostic_context): Enable
7049 * substring-locations.c: Include "gcc-rich-location.h".
7050 (format_warning_n_va): Add "fmt_label" and "param_label" params
7051 and use them as appropriate.
7052 (format_warning_va): Add "fmt_label" and "param_label" params,
7053 passing them on to format_warning_n_va.
7054 (format_warning_at_substring): Likewise.
7055 (format_warning_at_substring_n): Likewise.
7056 * substring-locations.h (format_warning_va): Add "fmt_label" and
7057 "param_label" params.
7058 (format_warning_n_va): Likewise.
7059 (format_warning_at_substring): Likewise.
7060 (format_warning_at_substring_n): Likewise.
7061 * toplev.c (general_init): Initialize global_dc->show_labels_p.
7063 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
7066 * builtins.c (expand_builtin_memcmp): Do not expand the call
7067 when overflow is detected.
7069 2018-08-15 Martin Sebor <msebor@redhat.com>
7071 PR tree-optimization/71625
7072 * config/aarch64/aarch64-builtins.c
7073 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
7075 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
7077 * config/s390/s390.c (s390_reorg): Remove loop.
7079 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
7082 (darwin_function_switched_text_sections): Delete.
7083 * gcc/config/darwin.h
7084 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
7086 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
7089 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
7090 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
7091 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
7093 2018-08-15 Martin Liska <mliska@suse.cz>
7095 PR tree-optimization/86925
7096 * predict.c (expr_expected_value_1): When taking
7097 later predictor, assign also probability.
7098 Use fold_build2_initializer_loc in order to fold
7099 the expression in -frounding-math.
7101 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
7103 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
7105 (expand_vec_perm_1): Try the new method.
7107 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
7110 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
7111 Check whether lra_live_max_point is 0 before dividing.
7113 2018-08-14 Martin Sebor <msebor@redhat.com>
7115 PR tree-optimization/86650
7116 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
7117 (vrp_prop::check_mem_ref): Same.
7119 2018-08-13 Liu Hao <lh_mouse@126.com>
7121 * pretty-print.c (eat_esc_sequence): Swap the foreground and
7122 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
7123 and clear it thereafter, as it only works for DBCS.
7125 2018-08-13 Liu Hao <lh_mouse@126.com>
7127 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
7128 handle returned by _get_osf_handle().
7130 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7132 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
7133 for folding vec_perm.
7135 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7137 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
7138 Add support for gimple-folding of vec_pack() and vec_unpack()
7141 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7143 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
7144 vec_xst variants to the list.
7145 (rs6000_gimple_fold_builtin): Add support for folding unaligned
7146 vector loads and stores.
7148 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
7150 * config.gcc (rs6000-ibm-aix4.x): Delete.
7151 (rs6000-ibm-aix5.1): Delete.
7152 (rs6000-ibm-aix5.2): Delete.
7153 (rs6000-ibm-aix5.3): Delete.
7154 * config/rs6000/aix43.h: Delete.
7155 * config/rs6000/aix51.h: Delete.
7156 * config/rs6000/aix52.h: Delete.
7157 * config/rs6000/t-aix43: Delete.
7159 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
7161 * config/s390/s390.c (s390_decompose_constant_pool_ref):
7163 (s390_decompose_address): Factor out constant pool ref
7166 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7168 * config/nds32/nds32-predicates.c
7169 (nds32_can_use_bclr_p): Change return type as bool.
7170 (nds32_can_use_bset_p): Ditto.
7171 (nds32_can_use_btgl_p): Ditto.
7172 (nds32_can_use_bitci_p): Ditto.
7173 * config/nds32/nds32-protos.h
7174 (nds32_can_use_bclr_p): Change declaration.
7175 (nds32_can_use_bset_p): Ditto.
7176 (nds32_can_use_btgl_p): Ditto.
7177 (nds32_can_use_bitci_p): Ditto.
7179 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7181 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
7182 Support -msched-prolog-epilog option.
7183 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
7185 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7187 * common/config/nds32/nds32-common.c
7188 (nds32_option_optimization_table): Enalbe -malways-align.
7190 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7192 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
7194 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
7195 OPT_misr_secure_ case.
7196 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
7197 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
7198 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
7200 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
7201 (nds32_isr_info): New field security_level.
7202 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
7203 * config/nds32/nds32.md (return_internal): Consider critical attribute.
7204 * config/nds32/nds32.opt (misr-secure): New option.
7205 * config/nds32/nds32_init.inc: New file.
7206 * config/nds32/nds32_isr.h: New file.
7208 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
7210 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
7211 Update comment for atomic instructions.
7212 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
7213 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
7215 (atomic_loaddi): Revise fence expansion to only emit fence prior to
7216 load for __ATOMIC_SEQ_CST model.
7217 (atomic_loaddi_1): Remove float register target.
7218 (atomic_storedi): Handle CONST_INT values.
7219 (atomic_storedi_1): Remove float register source. Add special case
7221 (memory_barrier): New expander and insn.
7223 2018-08-11 Jakub Jelinek <jakub@redhat.com>
7225 PR tree-optimization/86835
7226 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
7227 new_stmt after def_gsi, make sure to insert new_square_stmt after
7228 that stmt, not 2 stmts before it.
7230 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
7233 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
7236 2018-08-10 Martin Liska <mliska@suse.cz>
7239 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
7241 * builtins.c (expand_builtin_expect_with_probability):
7243 (expand_builtin_expect_with_probability): New function.
7244 (build_builtin_expect_predicate): Add new argumnet probability
7245 for BUILT_IN_EXPECT_WITH_PROBABILITY.
7246 (fold_builtin_expect):
7249 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
7250 * builtins.h (fold_builtin_expect): Set new argument.
7251 * doc/extend.texi: Document __builtin_expect_with_probability.
7252 * doc/invoke.texi: Likewise.
7253 * gimple-fold.c (gimple_fold_call): Pass new argument.
7254 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
7255 also BUILT_IN_EXPECT_WITH_PROBABILITY.
7256 * predict.c (get_predictor_value): New function.
7257 (expr_expected_value): Add new argument probability. Assume
7258 that predictor and probability are always non-null.
7259 (expr_expected_value_1): Likewise. For __builtin_expect and
7260 __builtin_expect_with_probability set probability. Handle
7261 combination in binary expressions.
7262 (tree_predict_by_opcode): Simplify code by simply calling
7263 get_predictor_value.
7264 (pass_strip_predict_hints::execute): Add handling of
7265 BUILT_IN_EXPECT_WITH_PROBABILITY.
7266 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
7268 * tree.h (DECL_BUILT_IN_P): New function.
7270 2018-08-10 Martin Liska <mliska@suse.cz>
7272 PR tree-optimization/85799
7273 * passes.def: Add argument for pass_strip_predict_hints.
7274 * predict.c (class pass_strip_predict_hints): Add new argument
7276 (strip_predictor_early): New function.
7277 (pass_strip_predict_hints::execute): Call the function to
7279 (strip_predict_hints): New function.
7280 * predict.def: Fix comment.
7282 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
7284 * Makefile.in: Clarify which tm.texi to copy over to assert the
7285 right to grant a GFDL license for all.
7287 2018-08-09 Jeff Law <law@redhat.com>
7289 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
7292 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
7294 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
7297 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7299 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
7300 reductions for variable-length vectors.
7302 2018-08-09 David Malcolm <dmalcolm@redhat.com>
7305 * common.opt (fdiagnostics-show-line-numbers): New option.
7306 * diagnostic-show-locus.c (class layout): Add fields
7307 "m_show_line_numbers_p" and "m_linenum_width";
7308 (num_digits): New function.
7309 (test_num_digits): New function.
7310 (layout::layout): Initialize new fields. Update m_x_offset
7311 logic to handle any left margin.
7312 (layout::print_source_line): Print line number when requested.
7313 (layout::start_annotation_line): New member function.
7314 (layout::print_annotation_line): Call it.
7315 (layout::print_leading_fixits): Likewise.
7316 (layout::print_trailing_fixits): Likewise. Update calls to
7317 move_to_column for new parameter.
7318 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
7319 it to potentially call start_annotation_line.
7320 (layout::show_ruler): Call start_annotation_line.
7321 (selftest::test_line_numbers_multiline_range): New selftest.
7322 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
7323 and selftest::test_line_numbers_multiline_range.
7324 * diagnostic.c (diagnostic_initialize): Initialize
7325 show_line_numbers_p.
7326 * diagnostic.h (struct diagnostic_context): Add field
7327 "show_line_numbers_p".
7328 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7329 -fno-diagnostics-show-line-numbers.
7330 * dwarf2out.c (gen_producer_string): Add
7331 OPT_fdiagnostics_show_line_numbers to the ignored options.
7332 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
7333 one setting" options.
7334 (append_compiler_options): Likewise to the dropped options.
7335 (append_diag_options): Likewise to the passed-on options.
7336 * opts.c (common_handle_option): Handle the new option.
7337 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
7339 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
7341 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
7342 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
7343 third argument of type "const signed char" to descriptions of
7344 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
7345 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
7346 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
7347 __builtin_bcdsub_ov functions.
7349 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7351 PR tree-optimization/86858
7352 * tree-vect-loop.c (vect_is_simple_reduction): Restore
7353 flow_bb_inside_loop_p calls.
7355 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7357 PR tree-optimization/86871
7358 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
7359 instead of gimple_assign_lhs.
7361 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
7364 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
7365 register constraint to operand 0.
7366 (add<mode>3_carryinC): Likewise.
7367 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
7369 2018-08-09 Martin Liska <mliska@suse.cz>
7372 * common.opt: Remove extra line.
7374 2018-08-09 Martin Liska <mliska@suse.cz>
7376 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
7377 at the end of a line, make first letter capital and end up
7378 a sentence with a dot.
7379 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
7380 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
7381 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
7382 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
7383 (PARAM_MAX_ISL_OPERATIONS): Likewise.
7384 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
7385 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
7386 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
7387 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
7388 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
7389 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
7390 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
7391 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
7392 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
7393 (PARAM_TREE_REASSOC_WIDTH): Likewise.
7394 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
7395 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
7396 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
7398 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
7401 * config/s390/s390.c (s390_option_override_internal): Reduce the
7402 stack-clash-protection-probe-interval param if it would be too big
7405 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
7408 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
7409 * config/m68k/m68k.c (handle_move_double): Don't call
7410 m68k_final_prescan_insn.
7411 (m68k_adjust_decorated_operand): Renamed from
7412 m68k_final_prescan_insn, remove first and third operand and
7414 (print_operand): Call it.
7415 (print_operand_address): Call it.
7417 2018-08-08 Nathan Sidwell <nathan@acm.org>
7419 * diagnostic.c (diagnostic_report_current_module): Use
7420 linemap_included_from & linemap_included_from_linemap.
7422 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
7424 * config/aarch64/aarch64-cores.def: Add phecda core.
7425 * config/aarch64/aarch64-tune.md: Regenerate.
7426 * doc/invoke.texi: Add phecda core.
7428 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
7431 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
7433 * config/s390/s390.md ("movti"): Add more alternatives for
7434 constant to GPR copies.
7436 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
7438 * config/s390/s390.c: Fix whitespace damage throughout the file.
7439 * config/s390/s390.h: Likewise.
7440 * config/s390/tpf.h: Likewise.
7442 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
7444 * config/s390/s390.c (s390_loadrelative_operand_p):
7445 Remove TARGET_CPU_ZARCH usages.
7446 (s390_rtx_costs): Likewise.
7447 (s390_legitimate_constant_p): Likewise.
7448 (s390_cannot_force_const_mem): Likewise.
7449 (legitimate_reload_constant_p): Likewise.
7450 (s390_preferred_reload_class): Likewise.
7451 (legitimize_pic_address): Likewise.
7452 (legitimize_tls_address): Likewise.
7453 (s390_split_branches): Removed.
7454 (s390_add_execute): Removed.
7455 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
7456 (s390_mainpool_start): Likewise.
7457 (s390_mainpool_finish): Likewise.
7458 (s390_mainpool_cancel): Removed.
7459 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
7460 (s390_chunkify_cancel): Likewise.
7461 (s390_return_addr_rtx): Likewise.
7462 (s390_register_info): Remove split_branches_pending_p uages.
7463 (s390_optimize_register_info): Likewise.
7464 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
7465 split_branches_pending_p usages.
7466 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
7467 (s390_load_got): Likewise.
7468 (s390_expand_split_stack_prologue): Likewise.
7469 (output_asm_nops): Likewise.
7470 (s390_function_profiler): Likewise.
7471 (s390_emit_call): Likewise.
7472 (s390_conditional_register_usage): Likewise.
7473 (s390_optimize_prologue): Likewise.
7474 (s390_reorg): Remove TARGET_CPU_ZARCH and
7475 split_branches_pending_p usages.
7476 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
7478 (s390_output_indirect_thunk_function): Likewise.
7479 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
7480 (TARGET_CPU_ZARCH_P): Removed.
7481 (struct machine_function): Remove split_branches_pending_p.
7482 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
7484 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
7486 * common/config/s390/s390-common.c (processor_flags_table):
7488 * config.gcc: Remove with_arch/with_tune support.
7489 * config/s390/2064.md: Remove cpu attribute comparisons.
7490 * config/s390/driver-native.c (s390_host_detect_local_cpu):
7492 * config/s390/linux.h (ASM_SPEC):
7493 Remove -march support.
7494 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
7495 Use a table to get an arch level.
7496 * config/s390/s390-opts.h (enum processor_type):
7498 * config/s390/s390.c
7499 (processor_table): Remove entries, add arch_level values.
7500 (s390_issue_rate): Remove cases.
7501 (s390_option_override): Adjust
7502 s390_option_override_internal() call.
7503 (s390_option_override_internal): Remove deprecation warning.
7504 (s390_valid_target_attribute_tree): Adjust
7505 s390_option_override_internal() call.
7506 * config/s390/s390.h (struct s390_processor):
7507 Share with s390-c.c, add arch_level field.
7508 * config/s390/s390.md:
7509 Remove occurrences in cpu attribute.
7510 * config/s390/s390.opt: Remove -march/-mtune support.
7511 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
7512 * doc/invoke.texi: Remove deprecation warning.
7514 2018-08-08 Luis Machado <luis.machado@linaro.org>
7516 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
7518 (qdf24xx_tunings): Set vector cost structure to
7519 qdf24xx_vector_cost.
7521 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
7522 <register_sextend>: Set to 3.
7524 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
7527 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
7528 * config/aarch64/aarch64-simd.md
7529 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
7530 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
7531 (aarch64_frecpx<mode>): ...this new pattern.
7532 * config/aarch64/aarch64-simd-builtins.def: Remove comment
7533 about aarch64_frecp<FRECP:frecp_suffix><mode>.
7535 2018-08-07 Martin Liska <mliska@suse.cz>
7538 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
7539 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
7540 * predict.def (PRED_MALLOC_NONNULL): New predictor.
7541 * doc/extend.texi: Document that malloc attribute adds
7544 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
7547 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7548 Define to speculation_safe_value_not_needed.
7550 2018-08-06 Jeff Law <law@redhat.com>
7552 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
7553 the vr_values instance to cprop_into_stmt.
7554 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
7555 (cprop_operand): Also query EVRP to determine if OP is a constant.
7557 2018-08-06 Nathan Sidwell <nathan@acm.org>
7559 * diagnostic.c (diagnostic_report_current_module): Reroll
7560 included-at loop. Translate text.
7562 2018-08-06 David Malcolm <dmalcolm@redhat.com>
7564 * function-tests.c (selftest::test_expansion_to_rtl): Call
7565 free_after_compilation.
7567 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7569 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
7571 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
7573 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
7574 loops with memory block operations from getting unrolled.
7576 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
7579 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7580 Define to speculation_safe_value_not_needed.
7582 2018-08-06 Jeff Law <law@redhat.com>
7584 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
7587 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7590 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
7591 with all enabled __intN types.
7593 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
7595 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7597 * alias.c (record_set): Check for clobber high.
7598 * cfgexpand.c (expand_gimple_stmt): Likewise.
7599 * combine-stack-adj.c (single_set_for_csa): Likewise.
7600 * combine.c (find_single_use_1): Likewise.
7601 (set_nonzero_bits_and_sign_copies): Likewise.
7602 (get_combine_src_dest): Likewise.
7603 (is_parallel_of_n_reg_sets): Likewise.
7604 (try_combine): Likewise.
7605 (record_dead_and_set_regs_1): Likewise.
7606 (reg_dead_at_p_1): Likewise.
7607 (reg_dead_at_p): Likewise.
7608 * dce.c (deletable_insn_p): Likewise.
7609 (mark_nonreg_stores_1): Likewise.
7610 (mark_nonreg_stores_2): Likewise.
7611 * df-scan.c (df_find_hard_reg_defs): Likewise.
7612 (df_uses_record): Likewise.
7613 (df_get_call_refs): Likewise.
7614 * dwarf2out.c (mem_loc_descriptor): Likewise.
7615 * haifa-sched.c (haifa_classify_rtx): Likewise.
7616 * ira-build.c (create_insn_allocnos): Likewise.
7617 * ira-costs.c (scan_one_insn): Likewise.
7618 * ira.c (equiv_init_movable_p): Likewise.
7619 (rtx_moveable_p): Likewise.
7620 (interesting_dest_for_shprep): Likewise.
7621 * jump.c (mark_jump_label_1): Likewise.
7622 * postreload-gcse.c (record_opr_changes): Likewise.
7623 * postreload.c (reload_cse_simplify): Likewise.
7624 (struct reg_use): Add source expr.
7625 (reload_combine): Check for clobber high.
7626 (reload_combine_note_use): Likewise.
7627 (reload_cse_move2add): Likewise.
7628 (move2add_note_store): Likewise.
7629 * print-rtl.c (print_pattern): Likewise.
7630 * recog.c (decode_asm_operands): Likewise.
7631 (store_data_bypass_p): Likewise.
7632 (if_test_bypass_p): Likewise.
7633 * regcprop.c (kill_clobbered_value): Likewise.
7634 (kill_set_value): Likewise.
7635 * reginfo.c (reg_scan_mark_refs): Likewise.
7636 * reload1.c (maybe_fix_stack_asms): Likewise.
7637 (eliminate_regs_1): Likewise.
7638 (elimination_effects): Likewise.
7639 (mark_not_eliminable): Likewise.
7640 (scan_paradoxical_subregs): Likewise.
7641 (forget_old_reloads_1): Likewise.
7642 * reorg.c (find_end_label): Likewise.
7643 (try_merge_delay_insns): Likewise.
7644 (redundant_insn): Likewise.
7645 (own_thread_p): Likewise.
7646 (fill_simple_delay_slots): Likewise.
7647 (fill_slots_from_thread): Likewise.
7648 (dbr_schedule): Likewise.
7649 * resource.c (update_live_status): Likewise.
7650 (mark_referenced_resources): Likewise.
7651 (mark_set_resources): Likewise.
7652 * rtl.c (copy_rtx): Likewise.
7653 * rtlanal.c (reg_referenced_p): Likewise.
7654 (single_set_2): Likewise.
7655 (noop_move_p): Likewise.
7656 (note_stores): Likewise.
7657 * sched-deps.c (sched_analyze_reg): Likewise.
7658 (sched_analyze_insn): Likewise.
7660 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7662 * cse.c (invalidate_reg): New function extracted from...
7663 (invalidate): ...here.
7664 (canonicalize_insn): Check for clobber high.
7665 (invalidate_from_clobbers): invalidate clobber highs.
7666 (invalidate_from_sets_and_clobbers): Likewise.
7667 (count_reg_usage): Check for clobber high.
7668 (insn_live_p): Likewise.
7669 * cselib.c (cselib_expand_value_rtx_1):Likewise.
7670 (cselib_invalidate_regno): Check for clobber in setter.
7671 (cselib_invalidate_rtx): Pass through setter.
7672 (cselib_invalidate_rtx_note_stores):
7673 (cselib_process_insn): Check for clobber high.
7674 * cselib.h (cselib_invalidate_rtx): Add operand.
7676 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7678 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
7679 (mark_not_eliminable): Likewise.
7680 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
7681 * lra-lives.c (process_bb_lives): Check for clobber high.
7682 * lra.c (new_insn_reg): Remember clobber highs.
7683 (collect_non_operand_hard_regs): Check for clobber high.
7684 (lra_set_insn_recog_data): Likewise.
7685 (add_regs_to_insn_regno_info): Likewise.
7686 (lra_update_insn_regno_info): Likewise.
7688 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7690 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
7691 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
7693 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7695 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
7696 (copy_insn_1): Likewise.
7697 (gen_hard_reg_clobber_high): New gen function.
7698 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
7699 * genemit.c (gen_exp): Likewise.
7700 (gen_emit_seq): Pass through info.
7701 (gen_insn): Check for CLOBBER_HIGH.
7702 (gen_expand): Pass through info.
7703 (gen_split): Likewise.
7704 (output_add_clobbers): Likewise.
7705 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
7706 (remove_clobbers): Likewise.
7707 * rtl.h (gen_hard_reg_clobber_high): New declaration.
7709 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7711 * doc/rtl.texi (clobber_high): Add.
7712 (parallel): Add in clobber high
7713 * rtl.c (rtl_check_failed_code3): Add function.
7714 * rtl.def (CLOBBER_HIGH): Add expression.
7715 * rtl.h (RTL_CHECKC3): Add macro.
7716 (rtl_check_failed_code3): Add declaration.
7717 (XC3EXP): Add macro.
7719 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
7722 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
7723 cfun->machine->max_used_stack_alignment if needed.
7725 2018-08-04 Martin Sebor <msebor@redhat.com>
7727 PR tree-optimization/86571
7728 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
7731 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
7734 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7737 2018-08-03 Jeff Law <law@redhat.com>
7740 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7741 Define to speculation_safe_value_not_needed.
7743 2018-08-03 David Malcolm <dmalcolm@redhat.com>
7745 * doc/gcov.texi (-x): Remove duplicate "to".
7746 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
7747 (-Wif-not-aligned): Remove duplicate "is".
7748 (-flto): Remove duplicate "the".
7749 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
7750 duplicate "v5.00.b".
7751 (MSP430 Options): Remove duplicate "and" from the description
7752 of "-mgprel-sec=regexp".
7753 (x86 Options): Remove duplicate copies of "vmldLog102" and
7754 vmlsLog104 from description of "-mveclibabi=type".
7756 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
7758 * internal-fn.h (first_commutative_argument): Declare.
7759 * internal-fn.c (first_commutative_argument): New function.
7760 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
7761 restrictions for pattern statements. Use first_commutative_argument
7762 to look for commutative operands in calls to internal functions.
7764 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
7766 * Makefile.in (wide-int-range.o): New.
7767 * tree-vrp.c: Move all the wide_int_* functions to...
7768 * wide-int-range.cc: ...here.
7769 * tree-vrp.h: Move all the wide_int_* prototypes to...
7770 * wide-int-range.h: ...here.
7772 2018-08-03 Tom de Vries <tdevries@suse.de>
7774 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
7776 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
7777 * except.c (output_function_exception_table): Do early exit if
7778 targetm_common.except_unwind_info (&global_options) == UI_NONE.
7780 2018-08-03 Martin Liska <mliska@suse.cz>
7782 * predict.c (dump_prediction): Change to 2 digits
7785 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
7787 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
7789 (falkor_am_1_gtov_gtov): ... a new insn reservation.
7791 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
7793 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
7794 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
7795 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
7796 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
7797 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
7799 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7801 * diagnostic-show-locus.c (diagnostic_show_locus): Use
7802 pp_take_prefix when saving the existing prefix.
7803 * diagnostic.c (diagnostic_append_note): Likewise.
7804 * langhooks.c (lhd_print_error_function): Likewise.
7805 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
7806 param's type. Free the existing prefix.
7807 (pp_take_prefix): New function.
7808 (pretty_printer::pretty_printer): Drop the prefix parameter.
7809 Rename the length parameter to match the comment.
7810 (pretty_printer::~pretty_printer): Free the prefix.
7811 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
7813 (struct pretty_printer): Drop the "const" from "prefix" field's
7814 type and clarify memory management.
7815 (pp_set_prefix): Drop the "const" from the 2nd param.
7816 (pp_take_prefix): New decl.
7818 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
7820 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
7821 for word_mode_ok here instead of passing as argument.
7822 (expand_block_compare): Change select_block_compare_mode() call.
7823 (expand_strncmp_gpr_sequence): New function.
7824 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
7826 2018-08-02 Jeff Law <law@redhat.com>
7829 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7830 Define to speculation_safe_value_not_needed.
7833 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7834 Define to speculation_safe_value_not_needed.
7836 2018-08-02 Tom de Vries <tdevries@suse.de>
7839 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
7840 function. Return UI_TARGET unconditionally.
7841 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
7842 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
7844 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7846 * genemit.c (print_overload_test): Fix typo.
7848 2018-08-02 Richard Biener <rguenther@suse.de>
7850 PR tree-optimization/86816
7851 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
7852 which checks for value availability before querying it.
7853 (gvn_uses_equal): Use it.
7854 (same_succ_hash): Likewise.
7855 (gimple_equal_p): Likewise.
7857 2018-08-02 Nick Clifton <nickc@redhat.com>
7860 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7861 Define to speculation_safe_value_not_needed.
7864 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7865 Define to speculation_safe_value_not_needed.
7868 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7869 Define to speculation_safe_value_not_needed.
7872 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7873 Define to speculation_safe_value_not_needed.
7876 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7877 Define to speculation_safe_value_not_needed.
7880 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7881 Define to speculation_safe_value_not_needed.
7884 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7885 Define to speculation_safe_value_not_needed.
7888 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7889 Define to speculation_safe_value_not_needed.
7892 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7893 speculation_safe_value_not_needed.
7896 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
7897 to speculation_safe_value_not_needed.
7899 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7901 * doc/md.texi: Expand the documentation of instruction names
7902 to mention port-local uses. Document '@' in pattern names.
7903 * read-md.h (overloaded_instance, overloaded_name): New structs.
7905 (md_reader::handle_overloaded_name): New member function.
7906 (md_reader::get_overloads): Likewise.
7907 (md_reader::m_first_overload): New member variable.
7908 (md_reader::m_next_overload_ptr): Likewise.
7909 (md_reader::m_overloads_htab): Likewise.
7910 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
7911 m_next_overload_ptr and m_overloads_htab.
7912 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
7913 (get_mode_token, get_code_token, get_int_token): New functions.
7914 (map_attr_string): Add an optional argument that passes back
7915 the associated iterator.
7916 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
7917 (md_reader::handle_overloaded_name, add_overload_instance): New
7919 (apply_iterators): Handle '@' names. Report an error if '@'
7920 is used without iterators.
7921 (initialize_iterators): Initialize the new iterator_group fields.
7922 * genopinit.c (handle_overloaded_code_for)
7923 (handle_overloaded_gen): New functions.
7924 (main): Use them to print declarations of maybe_code_for_* and
7925 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
7926 * genemit.c (print_overload_arguments, print_overload_test)
7927 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
7928 (main): Use it to print definitions of maybe_code_for_* and
7929 maybe_gen_* functions.
7930 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
7931 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
7932 instead of explicit mode checks.
7933 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
7934 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
7935 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
7936 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
7937 (aarch64_expand_compare_and_swap): Likewise
7938 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
7939 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
7940 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
7941 (aarch64_constant_pool_reload_icode): Delete.
7942 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
7943 instead of aarch64_constant_pool_reload_icode. Use
7944 code_for_aarch64_reload_mov instead of explicit mode checks.
7945 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
7946 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
7947 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
7948 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
7949 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
7950 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
7951 (aarch64_atomic_load_op_code): Delete.
7952 (aarch64_emit_atomic_load_op): Likewise.
7953 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
7954 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
7955 instead of aarch64_emit_atomic_load_op.
7956 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
7957 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
7958 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
7959 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
7960 character before the pattern name.
7961 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7962 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
7963 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
7964 (aarch64_frecps<mode>): Likewise.
7965 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
7966 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
7967 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
7968 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
7969 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
7971 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7973 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
7974 Allow HFmode constants if TARGET_FP_F16INST.
7976 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
7979 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
7980 No longer check last store for clobber of address register.
7982 2018-08-02 Martin Liska <mliska@suse.cz>
7984 PR gcov-profile/86817
7985 * gcov.c (process_all_functions): New function.
7987 (process_file): Move functions processing to
7988 process_all_functions.
7990 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7992 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
7993 "const" to the "gimple *" and "rtx_insn *" parameters.
7994 * dumpfile.h (dump_user_location_t::dump_user_location_t):
7996 (dump_location_t::dump_location_t): Likewise.
7998 2018-08-01 Martin Sebor <msebor@redhat.com>
8000 PR tree-optimization/86650
8001 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
8002 rather than a "gcall *". Directly pass the data of interest
8003 to percent_K_format, rather than building a temporary CALL_EXPR
8005 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
8006 (gimple_fold_builtin_strncat): Adjust.
8007 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
8008 gcall* argument with gimple*.
8009 * gimple-ssa-warn-restrict.c (check_call): Same.
8010 (wrestrict_dom_walker::before_dom_children): Same.
8011 (builtin_access::builtin_access): Same.
8012 (check_bounds_or_overlap): Same
8013 (maybe_diag_overlap): Same.
8014 (maybe_diag_offset_bounds): Same.
8015 * tree-diagnostic.c (default_tree_printer): Move usage of
8016 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
8018 * tree-pretty-print.c (percent_K_format): Add argument.
8019 * tree-pretty-print.h: Add argument.
8020 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
8021 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
8022 (maybe_diag_stxncpy_trunc): Same.
8023 (handle_builtin_stxncpy): Same.
8024 (handle_builtin_strcat): Same.
8026 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8028 * match.pd: Optimise pointer range checks.
8030 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8032 PR tree-optimization/86758
8033 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
8034 to remove pattern statements.
8036 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8038 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
8039 result of dfs_enumerate_from when constructing stmt_vec_infos,
8040 instead of additionally calling get_loop_body.
8042 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8044 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
8046 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
8047 When creating an iv, assert that the step is not known to be zero.
8048 (vect_setup_realignment): Update call accordingly.
8049 * tree-vect-stmts.c (vectorizable_store): Likewise.
8050 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
8052 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8054 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
8055 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
8056 (vectorizable_reduction): Likewise.
8057 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8058 (vect_detect_hybrid_slp_stmts): Likewise.
8059 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8061 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
8063 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
8064 (wide_int_set_zero_nonzero_bits): ...this.
8065 (zero_nonzero_bits_from_vr): Rename to...
8066 (vrp_set_zero_nonzero_bits): ...this.
8067 (extract_range_from_multiplicative_op_1): Abstract wide int
8069 (wide_int_range_multiplicative_op): ...here.
8070 (extract_range_from_binary_expr_1): Extract wide int binary
8071 operations into their own functions.
8072 (wide_int_range_lshift): New.
8073 (wide_int_range_can_optimize_bit_op): New.
8074 (wide_int_range_shift_undefined_p): New.
8075 (wide_int_range_bit_xor): New.
8076 (wide_int_range_bit_ior): New.
8077 (wide_int_range_bit_and): New.
8078 (wide_int_range_trunc_mod): New.
8079 (extract_range_into_wide_ints): New.
8080 (vrp_shift_undefined_p): New.
8081 (extract_range_from_multiplicative_op): New.
8082 (vrp_can_optimize_bit_op): New.
8083 * tree-vrp.h (value_range::dump): New.
8084 (wide_int_range_multiplicative_op): New.
8085 (wide_int_range_lshift):New.
8086 (wide_int_range_shift_undefined_p): New.
8087 (wide_int_range_bit_xor): New.
8088 (wide_int_range_bit_ior): New.
8089 (wide_int_range_bit_and): New.
8090 (wide_int_range_trunc_mod): New.
8091 (zero_nonzero_bits_from_bounds): Rename to...
8092 (wide_int_set_zero_nonzero_bits): ...this.
8093 (zero_nonzero_bits_from_vr): Rename to...
8094 (vrp_set_zero_nonzero_bits): ...this.
8095 (range_easy_mask_min_max): Rename to...
8096 (wide_int_range_can_optimize_bit_op): this.
8097 * vr-values.c (simplify_bit_ops_using_ranges): Rename
8098 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
8100 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8102 * tree-vectorizer.h (vect_orig_stmt): New function.
8103 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
8104 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
8105 (vect_create_epilog_for_reduction): Likewise.
8106 (vectorizable_live_operation): Likewise.
8107 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8108 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
8109 * tree-vect-stmts.c (vectorizable_call): Likewise.
8110 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
8112 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8114 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
8116 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
8117 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
8118 (vect_transform_loop): Likewise.
8119 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8121 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8123 * tree-vectorizer.h (vect_schedule_slp): Return void.
8124 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8125 (vect_schedule_slp): Likewise.
8127 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8129 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
8131 (vect_transform_loop): Update calls accordingly. Schedule SLP
8132 instances before the main loop, if any exist.
8134 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8136 PR tree-optimization/86749
8137 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
8138 If the lhs is used in a COND_EXPR, check that it is being used
8139 as the "then" or "else" value.
8141 2018-08-01 Tom de Vries <tdevries@suse.de>
8144 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
8145 speculation_safe_value_not_needed.
8147 2018-08-01 Richard Biener <rguenther@suse.de>
8149 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
8152 2018-08-01 Martin Liska <mliska@suse.cz>
8154 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
8155 format how successful transformation is dumped.
8156 (gimple_mod_pow2_value_transform): Likewise.
8157 (gimple_mod_subtract_transform): Likewise.
8158 (gimple_stringops_transform): Likewise.
8160 2018-08-01 Martin Liska <mliska@suse.cz>
8163 * value-prof.c (interesting_stringop_to_profile_p):
8164 Simplify the code and add BUILT_IN_MEMMOVE.
8165 (gimple_stringops_transform): Likewise.
8167 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
8169 * config/aarch64/aarch64-simd.md
8170 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
8171 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
8172 use GPI iterator instead of SI mode.
8174 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
8176 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
8177 rs6000_speculation_barrier.
8178 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
8179 new barrier pattern name.
8181 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
8183 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
8184 (speculation_barrier): New insn.
8186 2018-08-01 Richard Biener <rguenther@suse.de>
8189 * graphite.h: Include isl/id.h and isl/space.h to allow build
8192 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
8195 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
8196 mode for COFF targets.
8197 * defaults.h (TARGET_COFF): Define.
8198 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
8199 TARGET_COFF): Define.
8200 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
8201 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
8202 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
8204 2018-07-31 Alexandre Oliva <oliva@adacore.com>
8206 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
8207 * gimple-streamer-out.c (output_bb): Save it.
8208 * lto-streamer-in.c (input_struct_function_base): Restore
8209 instance discriminator if available. Create map on demand.
8210 * lto-streamer-out.c (output_struct_function_base): Save it if
8212 * final.c (decl_to_instance_map): Document LTO strategy.
8214 2018-07-31 Alexandre Oliva <oliva@adacore.com>
8215 Olivier Hainque <hainque@adacore.com>
8217 * debug.h (decl_to_instance_map_t): New type.
8218 (decl_to_instance_map): Declare.
8219 (maybe_create_decl_to_instance_map): New inline function.
8220 * final.c (bb_discriminator, last_bb_discriminator): New statics,
8221 to track basic block discriminators.
8222 (final_start_function_1): Initialize them.
8223 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
8225 (decl_to_instance_map): New variable.
8226 (map_decl_to_instance, maybe_set_discriminator): New functions.
8227 (notice_source_line): Set discriminator.
8229 2018-07-31 Ian Lance Taylor <iant@golang.org>
8231 * targhooks.c (default_have_speculation_safe_value): Add
8234 2018-07-31 David Malcolm <dmalcolm@redhat.com>
8236 * dump-context.h: Include "pretty-print.h".
8237 (dump_context::refresh_dumps_are_enabled): New decl.
8238 (dump_context::emit_item): New decl.
8239 (class dump_context): Add fields "m_test_pp" and
8241 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
8242 (temp_dump_context::get_dumped_text): New decl.
8243 (class temp_dump_context): Add field "m_pp".
8244 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
8245 (dump_context::refresh_dumps_are_enabled): ...and add a test for
8247 (set_dump_file): Update for above change.
8248 (set_alt_dump_file): Likewise.
8249 (dump_loc): New overload, taking a pretty_printer *.
8250 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
8251 to any test pretty-printer.
8252 (make_item_for_dump_gimple_stmt): New function, adapted from
8253 optinfo::add_gimple_stmt.
8254 (dump_context::dump_gimple_stmt): Call it, and use the result,
8255 eliminating the direct usage of dump_file and alt_dump_file in
8256 favor of indirectly using them via emit_item.
8257 (make_item_for_dump_gimple_expr): New function, adapted from
8258 optinfo::add_gimple_expr.
8259 (dump_context::dump_gimple_expr): Call it, and use the result,
8260 eliminating the direct usage of dump_file and alt_dump_file in
8261 favor of indirectly using them via emit_item.
8262 (make_item_for_dump_generic_expr): New function, adapted from
8264 (dump_context::dump_generic_expr): Call it, and use the result,
8265 eliminating the direct usage of dump_file and alt_dump_file in
8266 favor of indirectly using them via emit_item.
8267 (make_item_for_dump_printf_va): New function, adapted from
8268 optinfo::add_printf_va.
8269 (make_item_for_dump_printf): New function.
8270 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
8271 and use the result, eliminating the direct usage of dump_file and
8272 alt_dump_file in favor of indirectly using them via emit_item.
8273 (make_item_for_dump_dec): New function.
8274 (dump_context::dump_dec): Call it, and use the result,
8275 eliminating the direct usage of dump_file and alt_dump_file in
8276 favor of indirectly using them via emit_item.
8277 (make_item_for_dump_symtab_node): New function, adapted from
8278 optinfo::add_symtab_node.
8279 (dump_context::dump_symtab_node): Call it, and use the result,
8280 eliminating the direct usage of dump_file and alt_dump_file in
8281 favor of indirectly using them via emit_item.
8282 (dump_context::begin_scope): Reimplement, avoiding direct usage
8283 of dump_file and alt_dump_file in favor of indirectly using them
8285 (dump_context::emit_item): New member function.
8286 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
8287 Set up test pretty-printer on the underlying context. Call
8288 refresh_dumps_are_enabled.
8289 (temp_dump_context::~temp_dump_context): Call
8290 refresh_dumps_are_enabled.
8291 (temp_dump_context::get_dumped_text): New member function.
8292 (selftest::verify_dumped_text): New function.
8293 (ASSERT_DUMPED_TEXT_EQ): New macro.
8294 (selftest::test_capture_of_dump_calls): Run all tests twice, with
8295 and then without optinfo enabled. Add uses of
8296 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
8297 * dumpfile.h: Update comment for the dump_* API.
8298 * optinfo-emit-json.cc
8299 (selftest::test_building_json_from_dump_calls): Update for new
8300 param for temp_dump_context ctor.
8301 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
8302 and "m_owned" field.
8303 (optinfo_item::~optinfo_item): Likewise.
8304 (optinfo::add_item): New member function.
8305 (optinfo::emit): Update comment.
8306 (optinfo::add_string): Delete.
8307 (optinfo::add_printf): Delete.
8308 (optinfo::add_printf_va): Delete.
8309 (optinfo::add_gimple_stmt): Delete.
8310 (optinfo::add_gimple_expr): Delete.
8311 (optinfo::add_tree): Delete.
8312 (optinfo::add_symtab_node): Delete.
8313 (optinfo::add_dec): Delete.
8314 * optinfo.h (class dump_context): New forward decl.
8315 (optinfo::add_item): New decl.
8316 (optinfo::add_string): Delete.
8317 (optinfo::add_printf): Delete.
8318 (optinfo::add_printf_va): Delete.
8319 (optinfo::add_gimple_stmt): Delete.
8320 (optinfo::add_gimple_expr): Delete.
8321 (optinfo::add_tree): Delete.
8322 (optinfo::add_symtab_node): Delete.
8323 (optinfo::add_dec): Delete.
8324 (optinfo::add_poly_int): Delete.
8325 (optinfo_item::optinfo_item): Remove "owned" param.
8326 (class optinfo_item): Remove field "m_owned".
8328 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
8331 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
8332 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
8333 requested variable alignment.
8334 (expand_one_ssa_partition): Likewise.
8335 (expand_one_var): Likewise.
8337 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8339 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
8340 to speculation_safe_value_not_needed.
8342 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8344 * targhooks.h (speculation_safe_value_not_needed): New prototype.
8345 * targhooks.c (speculation_safe_value_not_needed): New function.
8346 * target.def (have_speculation_safe_value): Update documentation.
8347 * doc/tm.texi: Regenerated.
8349 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8351 * config/aarch64/iterators.md (ALLI_TI): New iterator.
8352 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
8354 (despeculate_copy<ALLI:mode>_insn): New insn.
8355 (despeculate_copyti_insn): New insn.
8356 (despeculate_simple<ALLI:mode>): New insn
8357 (despeculate_simpleti): New insn.
8358 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
8360 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
8361 aarch64_speculation_safe_value.
8362 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
8364 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8366 * config/aarch64/aarch64-speculation.cc: New file.
8367 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
8368 before pass_reorder_blocks.
8369 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
8371 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
8372 X14 and X15 when tracking speculation.
8373 * config/aarch64/aarch64.md (register name constants): Add
8374 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
8375 (unspec): Add UNSPEC_SPECULATION_TRACKER.
8376 (speculation_barrier): New insn attribute.
8377 (cmp<mode>): Allow SP in comparisons.
8378 (speculation_tracker): New insn.
8379 (speculation_barrier): Add speculation_barrier attribute.
8380 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
8381 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
8382 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
8384 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8386 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
8387 aarch64_track_speculation is true.
8388 (tb<optab><mode>1): Likewise.
8389 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
8390 generate CB[N]Z when tracking speculation.
8391 (aarch64_split_compare_and_swap): Likewise.
8392 (aarch64_split_atomic_op): Likewise.
8394 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8396 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
8398 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8400 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
8401 (speculation_barrier): New insn.
8403 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8405 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
8406 * config/arm/arm.md (speculation_barrier): New expand.
8407 (speculation_barrier_insn): New pattern.
8409 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8411 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
8412 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
8413 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
8414 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
8416 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
8417 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
8418 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
8419 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
8420 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
8421 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
8422 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
8423 * builtins.c (expand_speculation_safe_value): New function.
8424 (expand_builtin): Call it.
8425 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
8426 * doc/extend.texi: Document __builtin_speculation_safe_value.
8427 * doc/md.texi: Document "speculation_barrier" pattern.
8428 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
8429 TARGET_HAVE_SPECULATION_SAFE_VALUE.
8430 * doc/tm.texi: Regenerated.
8431 * target.def (have_speculation_safe_value, speculation_safe_value): New
8433 * targhooks.c (default_have_speculation_safe_value): New function.
8434 (default_speculation_safe_value): New function.
8435 * targhooks.h (default_have_speculation_safe_value): Add prototype.
8436 (default_speculation_safe_value): Add prototype.
8438 2018-07-31 David Malcolm <dmalcolm@redhat.com>
8440 * dump-context.h (dump_context::dump_loc): New decl.
8441 * dumpfile.c (dump_context::dump_loc): New member function.
8442 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
8443 and dump_gimple_stmt.
8444 (dump_context::dump_gimple_expr_loc): Likewise, using
8446 (dump_context::dump_generic_expr_loc): Likewise, using
8448 (dump_context::dump_printf_loc_va): Likewise, using
8450 (dump_context::begin_scope): Explicitly using the global function
8451 "dump_loc", rather than the member function.
8453 2018-07-31 Martin Sebor <msebor@redhat.com>
8455 PR tree-optimization/86741
8456 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
8458 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
8460 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
8461 depend on whether prefetch instructions will be emitted or not.
8462 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
8463 will be emitted or not.
8464 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
8465 (TARGET_SETMEM_PFD): New macros.
8467 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8469 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
8470 (NULL_STMT_VEC_INFO): Delete.
8471 (stmt_vec_info::operator*): Likewise.
8472 (stmt_vec_info::operator gimple *): Likewise.
8473 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
8474 of NULL_STMT_VEC_INFO.
8475 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8476 (vect_reassociating_reduction_p): Likewise.
8477 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
8478 (vectorizable_store): Likewise.
8479 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
8480 (vec_info::free_stmt_vec_infos): Likewise.
8482 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8484 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
8485 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
8487 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8489 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
8490 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
8491 (vec_info::free_stmt_vec_info): New private member functions.
8492 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
8493 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
8494 * tree-parloops.c (gather_scalar_reductions): Remove calls to
8495 set_stmt_vec_info_vec and free_stmt_vec_infos.
8496 * tree-vect-loop.c (_loop_vec_info): Remove call to
8497 set_stmt_vec_info_vec.
8498 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
8499 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
8500 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
8501 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
8502 (vec_info::free_stmt_vec_info): ...these new functions. Remove
8503 assignments in {vec_info::,}new_stmt_vec_info that are redundant
8504 with the clearing in the xcalloc.
8505 (stmt_vec_info_vec): Delete.
8506 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
8507 (vectorize_loops): Likewise.
8508 (vec_info::~vec_info): Remove argument from call to
8509 free_stmt_vec_infos.
8510 (vec_info::add_stmt): Remove vinfo argument from call to
8513 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8515 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
8516 rather than a gimple stmt.
8517 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
8518 information for pattern statements when passed the original
8519 statement; instead wait to be passed the pattern statement itself.
8520 Don't call set_vinfo_for_stmt here.
8521 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
8522 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
8523 stmt_vec_infos here.
8524 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
8525 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
8526 stmt_vec_infos entry.
8528 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8530 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
8531 * tree-vectorizer.c (vec_info::replace_stmt): New function.
8532 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
8533 * tree-vect-stmts.c (vectorizable_call): Likewise.
8534 (vectorizable_simd_clone_call): Likewise.
8536 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8538 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
8539 * tree-vectorizer.c (vec_info::remove_stmt): New function.
8540 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
8541 * tree-vect-loop.c (vect_transform_loop): Likewise.
8542 * tree-vect-slp.c (vect_schedule_slp): Likewise.
8543 * tree-vect-stmts.c (vect_remove_stores): Likewise.
8545 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8547 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
8548 (vect_dr_stmt): Delete.
8549 * tree-vectorizer.c (vec_info::lookup_dr): New function.
8550 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
8552 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
8553 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
8554 (vect_verify_datarefs_alignment, vect_peeling_supportable)
8555 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8556 (vect_analyze_data_refs): Likewise.
8557 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
8559 (vect_find_same_alignment_drs): Likewise.
8560 (vect_slp_analyze_node_dependences): Update calls accordingly.
8561 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8562 instead of DR_VECT_AUX.
8563 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
8564 of a vector data references. Use vec_info::lookup_dr instead of
8566 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
8567 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8568 instead of DR_VECT_AUX.
8570 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8572 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
8574 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8576 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8577 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8578 (vect_gen_prolog_loop_niters): Likewise.
8580 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8582 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
8583 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
8584 (vect_known_alignment_in_bytes, vect_dr_behavior)
8585 (vect_get_scalar_dr_size): Take references as dr_vec_infos
8586 instead of data_references. Update calls to other routines for
8587 which the same change has been made.
8588 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
8589 dr_vec_infos instead of stmt_vec_infos.
8590 (vect_analyze_data_ref_dependence): Update call accordingly.
8591 (vect_slp_analyze_data_ref_dependence)
8592 (vect_record_base_alignments): Use DR_VECT_AUX.
8593 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
8594 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
8595 (vector_alignment_reachable_p, vect_get_data_access_cost)
8596 (vect_peeling_supportable, vect_analyze_group_access_1)
8597 (vect_analyze_group_access, vect_analyze_data_ref_access)
8598 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
8599 (vect_compile_time_alias, vect_small_gap_p)
8600 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
8601 (vect_supportable_dr_alignment): Take references as dr_vec_infos
8602 instead of data_references. Update calls to other routines for
8603 which the same change has been made.
8604 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
8605 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8606 (vect_slp_analyze_and_verify_node_alignment)
8607 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8608 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8609 (vect_setup_realignment): Use dr_vec_infos. Update calls after
8611 (_vect_peel_info::dr): Replace with...
8612 (_vect_peel_info::dr_info): ...this new field.
8613 (vect_peeling_hash_get_most_frequent)
8614 (vect_peeling_hash_choose_best_peeling): Update accordingly.
8615 (vect_peeling_hash_get_lowest_cost):
8616 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
8617 routines for which the same change has been made.
8618 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
8620 * tree-vect-loop-manip.c (get_misalign_in_elems)
8621 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
8623 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8624 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
8625 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
8626 (get_group_load_store_type, get_negative_load_store_type)
8627 (vect_get_data_ptr_increment, vectorizable_store)
8628 (vectorizable_load): Likewise.
8629 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
8630 Update calls to other routines for which the same change has been made.
8632 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8634 * tree-vectorizer.h (vec_info::move_dr): New member function.
8635 (dataref_aux): Rename to...
8636 (dr_vec_info): ...this and add "dr" and "stmt" fields.
8637 (_stmt_vec_info::dr_aux): Update accordingly.
8638 (_stmt_vec_info::data_ref_info): Delete.
8639 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
8640 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
8641 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
8642 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
8644 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
8645 (STMT_VINFO_DR_INFO): New macro.
8646 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
8647 (set_dr_misalignment): Update after rename of dataref_aux.
8648 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
8649 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
8650 initialization of STMT_VINFO_DATA_REF.
8651 * tree-vectorizer.c (vec_info::move_dr): New function.
8652 * tree-vect-patterns.c (vect_recog_bool_pattern)
8653 (vect_recog_mask_conversion_pattern)
8654 (vect_recog_gather_scatter_pattern): Use it.
8655 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
8656 the "dr" and "stmt" fields of dr_vec_info instead of
8657 STMT_VINFO_DATA_REF.
8659 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8661 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
8662 (is_pattern_stmt_p): Use it.
8663 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
8664 on pattern statements.
8666 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8668 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
8669 original stmt as a stmt_vec_info rather than a gimple stmt.
8670 (vect_pattern_recog_1): Take the statement directly as a
8671 stmt_vec_info, rather than via a gimple_stmt_iterator.
8672 Update call to vect_mark_pattern_stmts.
8673 (vect_pattern_recog): Update calls accordingly.
8675 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8677 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
8678 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
8679 a vect_def_type for the first argument.
8680 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
8681 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
8682 operand if it isn't defined by a vectorized statement.
8683 (vect_build_gather_load_calls): Remove the mask_dt argument and
8684 update calls to vect_get_vec_def_for_stmt_copy.
8685 (vectorizable_bswap): Likewise the dt argument.
8686 (vectorizable_call): Update calls to vectorizable_bswap and
8687 vect_get_vec_def_for_stmt_copy.
8688 (vectorizable_simd_clone_call, vectorizable_assignment)
8689 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
8690 (vectorizable_comparison): Update calls to
8691 vect_get_vec_def_for_stmt_copy.
8692 (vectorizable_store): Likewise. Remove now-unnecessary calls to
8694 (vect_get_loop_based_defs): Remove dt argument and update call
8695 to vect_get_vec_def_for_stmt_copy.
8696 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
8697 and vect_get_vec_def_for_stmt_copy.
8698 (vectorizable_load): Update calls to vect_build_gather_load_calls
8699 and vect_get_vec_def_for_stmt_copy.
8700 * tree-vect-loop.c (vect_create_epilog_for_reduction)
8701 (vectorizable_reduction, vectorizable_live_operation): Update calls
8702 to vect_get_vec_def_for_stmt_copy.
8704 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8706 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
8707 and gimple stmt arguments with a stmt_vec_info.
8708 (vect_record_base_alignments): Update calls accordingly.
8709 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
8710 and gimple stmt arguments with a stmt_vec_info.
8711 (vect_build_slp_tree_1): Remove vinfo argument and update call
8712 to vect_record_max_nunits.
8713 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
8714 and vect_record_max_nunits.
8716 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8718 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
8719 file and take a stmt_vec_info instead of a gimple stmt.
8720 (supportable_widening_operation, vect_finish_replace_stmt)
8721 (vect_finish_stmt_generation, vect_get_store_rhs)
8722 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
8723 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
8724 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
8725 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
8726 (vect_create_data_ref_ptr, bump_vector_ptr)
8727 (vect_permute_store_chain, vect_setup_realignment)
8728 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
8729 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
8730 (vectorizable_reduction, vectorizable_induction)
8731 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
8732 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
8733 than gimple stmts as arguments.
8734 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
8735 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
8736 (can_group_stmts_p, vect_check_gather_scatter)
8737 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8738 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
8739 (vect_permute_load_chain, vect_shift_permute_load_chain)
8740 (vect_transform_grouped_load)
8741 (vect_record_grouped_load_vectors): Likewise.
8742 * tree-vect-loop.c (vect_fixup_reduc_chain)
8743 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
8744 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
8745 (vectorizable_reduction, vectorizable_induction)
8746 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
8747 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
8748 (vect_get_load_store_mask): Likewise.
8749 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
8750 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
8751 * tree-vect-stmts.c (vect_mark_relevant)
8752 (is_simple_and_all_uses_invariant)
8753 (exist_non_indexing_operands_for_use_p, process_use)
8754 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
8755 (vect_get_vec_def_for_operand, vect_get_vec_defs)
8756 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
8757 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
8758 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
8759 (get_negative_load_store_type, get_load_store_type)
8760 (vect_check_load_store_mask, vect_check_store_rhs)
8761 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
8762 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8763 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8764 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8765 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
8766 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
8767 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
8768 (supportable_widening_operation): Likewise.
8770 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8772 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
8773 a stmt_vec_info instead of a gcall.
8774 (vect_check_gather_scatter): Update call accordingly.
8775 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
8777 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
8778 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
8779 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
8780 instead of a gimple stmt.
8781 (vect_transform_loop): Update calls accordingly.
8782 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
8783 stmt_vec_infos instead of gimple stmts.
8784 (vect_analyze_slp_instance): Update use accordingly.
8785 * tree-vect-stmts.c (read_vector_array, write_vector_array)
8786 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
8787 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
8788 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
8789 (vect_gen_widened_results_half, vect_get_loop_based_defs)
8790 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
8791 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
8792 down to subroutines.
8794 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8796 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
8797 of the worklist from a vector of gimple stmts to a vector of
8799 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8800 (vect_mark_stmts_to_be_vectorized): Likewise
8802 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8804 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
8805 statement before passing it to vect_analyze_stmt.
8806 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
8807 the chain of phi vector definitions. Track the exit phi via its
8809 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
8810 STMT_VINFO_REDUC_DEF.
8811 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
8812 stmt_vec_infos to handle the statement chains.
8813 (vect_get_slp_defs): Record the first statement in the node
8814 using a stmt_vec_info.
8815 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
8816 statements here and pass their stmt_vec_info down to subroutines.
8817 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
8818 down to vect_finish_stmt_generation.
8819 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
8820 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
8821 stmt_vec_infos to subroutines.
8822 (vect_remove_stores): Use stmt_vec_infos to handle the statement
8825 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8827 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
8828 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
8829 (vect_permute_store_chain, vect_setup_realignment)
8830 (vect_permute_load_chain, vect_shift_permute_load_chain)
8831 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
8832 stmts internally, and when passing values to other vectorizer routines.
8833 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8834 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
8835 (vect_analyze_loop_operations, get_initial_def_for_reduction)
8836 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
8837 (vectorizable_reduction, vectorizable_induction)
8838 (vectorizable_live_operation, vect_transform_loop_stmt)
8839 (vect_transform_loop): Likewise.
8840 * tree-vect-patterns.c (vect_reassociating_reduction_p)
8841 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
8842 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
8843 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8844 (vect_slp_analyze_node_operations_1): Likewise.
8845 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8846 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
8847 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
8848 (vect_finish_stmt_generation_1, get_group_load_store_type)
8849 (get_load_store_type, vect_build_gather_load_calls)
8850 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8851 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8852 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8853 (vectorizable_store, vectorizable_load, vectorizable_condition)
8854 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
8855 (supportable_widening_operation): Likewise.
8856 (vect_get_vector_types_for_stmt): Likewise.
8857 * tree-vectorizer.h (vect_dr_behavior): Likewise.
8859 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8861 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8862 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
8863 (vect_permute_store_chain, vect_permute_load_chain)
8864 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
8865 repeated stmt_vec_info lookups.
8866 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8867 (vect_update_ivs_after_vectorizer): Likewise.
8868 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
8869 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8870 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
8871 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8872 (vect_bb_slp_scalar_cost): Likewise.
8873 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
8875 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8877 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
8878 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
8879 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
8880 to get gassigns and gcalls, rather than operating on generc gimple
8882 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
8883 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
8884 (vectorizable_load, vect_analyze_stmt): Likewise.
8885 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
8887 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8889 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
8890 return stmt_vec_infos rather than gimple stmts. Do not accept
8892 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
8894 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8895 Update use of get_later_stmt.
8896 (vect_get_constant_vectors): Update call accordingly.
8897 (vect_schedule_slp_instance): Likewise
8898 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
8899 (vect_slp_analyze_instance_dependence): Likewise.
8900 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
8902 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8904 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
8905 (stmt_info_for_cost::stmt_info): ...this new field.
8906 (add_stmt_costs): Update accordingly.
8907 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8908 (vect_get_known_peeling_cost): Likewise.
8909 (vect_estimate_min_profitable_iters): Likewise.
8910 * tree-vect-stmts.c (record_stmt_cost): Likewise.
8912 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8914 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
8915 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8916 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8918 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
8920 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8922 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
8923 a gimple stmt to a stmt_vec_info.
8924 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
8926 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8928 * tree-vectorizer.h (vec_info::grouped_stores): Change from
8929 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8930 (_loop_vec_info::reduction_chains): Likewise.
8931 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
8933 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8935 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8937 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
8938 a gimple stmt to a stmt_vec_info.
8939 (_stmt_vec_info::next_element): Likewise.
8940 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
8941 (vect_slp_analyze_and_verify_node_alignment)
8942 (vect_analyze_group_access_1, vect_analyze_group_access)
8943 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
8944 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
8945 (vect_supportable_dr_alignment): Update accordingly.
8946 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
8947 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
8948 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
8949 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
8950 * tree-vect-slp.c (vect_build_slp_tree_1)
8951 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
8952 (vect_split_slp_store_group, vect_analyze_slp_instance)
8953 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
8954 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
8955 (get_group_load_store_type, get_load_store_type)
8956 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
8957 (vect_transform_stmt, vect_remove_stores): Likewise.
8959 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8961 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
8963 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8964 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
8965 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
8966 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
8967 (vector_alignment_reachable_p, vect_get_data_access_cost)
8968 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
8969 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
8970 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8971 (vect_analyze_group_access_1, vect_analyze_group_access)
8972 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
8973 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
8974 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
8975 result of vect_dr_stmt and use the stmt_vec_info instead of
8976 the associated gimple stmt.
8977 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8978 (vect_gen_prolog_loop_niters): Likewise.
8979 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8981 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8983 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
8984 to a vec<stmt_vec_info>.
8985 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
8986 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
8988 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
8989 to a vec<stmt_vec_info>.
8990 (bst_traits::value_type, bst_traits::value_type): Likewise.
8991 (bst_traits::hash): Update accordingly.
8992 (vect_get_and_check_slp_defs): Change the stmts parameter from
8993 a vec<gimple *> to a vec<stmt_vec_info>.
8994 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
8995 (vect_build_slp_tree): Likewise.
8996 (vect_build_slp_tree_2): Likewise. Update uses of
8997 SLP_TREE_SCALAR_STMTS.
8998 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
8999 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
9000 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
9001 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
9002 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
9003 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
9004 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
9005 (vect_get_constant_vectors, vect_get_slp_defs)
9006 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
9007 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
9008 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
9009 instead of gimple stmts.
9010 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
9011 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
9012 (vect_slp_analyze_instance_dependence): Update uses of
9013 SLP_TREE_SCALAR_STMTS.
9014 (vect_slp_analyze_and_verify_node_alignment): Likewise.
9015 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
9016 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
9017 (get_initial_defs_for_reduction): Likewise.
9018 (vect_create_epilog_for_reduction): Likewise.
9019 (vectorize_fold_left_reduction): Likewise.
9020 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
9021 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
9022 (can_vectorize_live_stmts): Likewise.
9024 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9026 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
9027 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
9028 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
9030 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
9032 (gather_scalar_reductions): Update after above interface changes.
9033 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
9034 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
9036 (vect_force_simple_reduction): Likewise.
9037 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
9038 LOOP_VINFO_REDUCTIONS.
9039 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
9041 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9043 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
9044 a gimple stmt to a stmt_vec_info.
9045 * tree-vect-loop.c (vect_active_double_reduction_p)
9046 (vect_force_simple_reduction, vectorizable_reduction): Update
9049 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9051 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
9052 vec<gimple *> to a vec<stmt_vec_info>.
9053 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
9054 the reduction_phis argument from a vec<gimple *> to a
9056 (vectorizable_reduction): Likewise the phis local variable that
9057 is passed to vect_create_epilog_for_reduction. Update for new type
9058 of SLP_TREE_VEC_STMTS.
9059 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
9060 (vectorizable_live_operation): Likewise.
9061 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
9062 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
9064 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9066 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
9067 a gimple stmt to a stmt_vec_info.
9068 (vectorizable_condition, vectorizable_live_operation)
9069 (vectorizable_reduction, vectorizable_induction): Pass back the
9070 vectorized statement as a stmt_vec_info.
9071 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
9072 use of STMT_VINFO_VEC_STMT.
9073 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
9074 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
9075 as stmt_vec_infos rather than gimple stmts.
9076 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
9078 (vectorizable_live_operation): Likewise.
9079 (vectorizable_reduction, vectorizable_induction): Likewise,
9080 updating use of STMT_VINFO_VEC_STMT.
9081 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
9082 of STMT_VINFO_VEC_STMT.
9083 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
9084 (vectorizable_simd_clone_call, vectorizable_conversion)
9085 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
9086 (vectorizable_store, vectorizable_load, vectorizable_condition)
9087 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
9088 from a gimple stmt to a stmt_vec_info.
9089 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
9090 pointer to a stmt_vec_info to the vectorizable_* routines.
9092 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9094 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
9095 a gimple stmt to a stmt_vec_info.
9096 (is_pattern_stmt_p): Update accordingly.
9097 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
9098 (vect_record_grouped_load_vectors): Likewise.
9099 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
9100 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
9101 (vect_model_reduction_cost): Likewise.
9102 (vect_create_epilog_for_reduction): Likewise.
9103 (vectorizable_reduction, vectorizable_induction): Likewise.
9104 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9105 Return the stmt_vec_info for the pattern statement.
9106 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
9107 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
9108 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
9109 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
9110 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
9111 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
9112 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
9113 (free_stmt_vec_info, vect_is_simple_use): Likewise.
9115 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9117 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
9118 (vect_finish_stmt_generation): Likewise.
9119 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
9120 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
9121 (vect_build_gather_load_calls): Use the return value of the above
9122 functions instead of a separate call to vinfo_for_stmt. Use narrow
9123 scopes for the input gimple stmt and wider scopes for the associated
9124 stmt_vec_info. Use vec_info::lookup_def when setting these
9125 stmt_vec_infos from an SSA_NAME definition.
9126 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
9127 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
9128 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
9129 (vectorizable_store, vectorizable_load, vectorizable_condition)
9130 (vectorizable_comparison): Likewise.
9131 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
9132 (vectorizable_reduction): Likewise.
9134 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9136 * tree-vectorizer.h (vect_is_simple_use): Add an optional
9137 stmt_vec_info * parameter before the optional gimple **.
9138 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
9139 (process_use, vect_get_vec_def_for_operand_1): Update callers.
9140 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
9141 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9142 (vectorizable_live_operation): Likewise.
9143 * tree-vect-patterns.c (type_conversion_p): Likewise.
9144 (vect_look_through_possible_promotion): Likewise.
9145 (vect_recog_rotate_pattern): Likewise.
9146 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
9148 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9150 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
9151 a typedef to a wrapper class.
9152 (NULL_STMT_VEC_INFO): New macro.
9153 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
9154 (stmt_vec_info::operator*): New function.
9155 (stmt_vec_info::operator gimple *): Likewise.
9156 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
9157 (add_stmt_costs): Likewise.
9158 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
9159 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
9160 (vect_get_known_peeling_cost): Likewise.
9161 (vect_estimate_min_profitable_iters): Likewise.
9162 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9163 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
9164 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
9165 (vectorizable_store, free_stmt_vec_infos): Likewise.
9166 (new_stmt_vec_info): Change return type of xcalloc to
9169 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9171 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
9172 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
9173 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
9174 a single_imm_use-based sequence.
9175 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
9177 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9179 * tree-vectorizer.h (vec_info::lookup_def): Declare.
9180 * tree-vectorizer.c (vec_info::lookup_def): New function.
9181 * tree-vect-patterns.c (vect_get_internal_def): Use it.
9182 (vect_widened_op_tree): Likewise.
9183 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
9184 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
9185 (vectorizable_reduction): Likewise.
9186 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
9188 (vect_is_slp_reduction): Update calls accordingly. Use
9189 vec_info::lookup_def.
9190 (vect_is_simple_reduction): Likewise
9191 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
9193 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9195 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
9196 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
9197 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
9199 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
9200 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
9201 (vect_update_vf_for_slp, vect_analyze_loop_operations)
9202 (vect_is_slp_reduction, vectorizable_induction)
9203 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
9204 * tree-vect-patterns.c (vect_init_pattern_stmt):
9205 (vect_determine_min_output_precision_1, vect_determine_precisions)
9206 (vect_pattern_recog): Likewise.
9207 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
9208 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
9209 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
9210 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
9211 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
9212 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
9213 info field from a loop to a loop_vec_info.
9215 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9217 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
9218 (vec_info::add_stmt): Declare.
9219 * tree-vectorizer.c (vec_info::add_stmt): New function.
9220 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
9221 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
9222 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
9223 (vectorizable_induction): Likewise.
9224 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
9225 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
9226 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
9227 (vectorizable_load): Likewise.
9228 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9229 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
9230 (vect_recog_gather_scatter_pattern): Likewise.
9231 (append_pattern_def_seq): Likewise. Remove a check that is
9232 performed by add_stmt itself.
9234 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9236 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
9237 which make_ssa_name was called with new_stmt before new_stmt
9240 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9242 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
9244 (vect_is_slp_reduction): ...here...
9245 (vect_is_simple_reduction): ...and here. Remove repetition of tests
9246 that are already known to be false.
9248 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9250 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
9251 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
9252 STMT_VINFO_NUM_SLP_USES when it's true.
9253 (vect_free_slp_instance): Add a final_p parameter and pass it to
9255 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
9256 (vect_analyze_slp_instance): Likewise.
9257 (vect_slp_analyze_operations): Likewise.
9258 (vect_slp_analyze_bb_1): Likewise.
9259 * tree-vectorizer.c (vec_info): Likewise.
9260 * tree-vect-loop.c (vect_transform_loop): Likewise.
9262 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9264 * tree-vect-loop.c (vectorizable_reduction): Assert that the
9265 function is not called for second and subsequent members of
9268 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9270 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
9271 cases for nested loops from here to ...
9272 (vect_create_epilog_for_reduction): ...here. Only call
9273 vect_is_simple_use for inner-loop reductions.
9275 2018-07-31 Martin Liska <mliska@suse.cz>
9277 PR gcov-profile/85338
9278 PR gcov-profile/85350
9279 PR gcov-profile/85372
9280 * profile.c (struct location_triplet): New.
9281 (struct location_triplet_hash): Likewise.
9282 (output_location): Do not output a BB that
9283 is already recorded for a line.
9284 (branch_prob): Use streamed_locations.
9286 2018-07-31 Martin Liska <mliska@suse.cz>
9288 PR gcov-profile/85370
9289 * coverage.c (coverage_begin_function): Do not mark target
9290 clones as artificial functions.
9292 2018-07-31 Martin Liska <mliska@suse.cz>
9294 PR gcov-profile/83813
9295 PR gcov-profile/84758
9296 PR gcov-profile/85217
9297 PR gcov-profile/85332
9298 * profile.c (branch_prob): Do not record GOTO expressions
9299 for GIMPLE statements which locations are already streamed.
9301 2018-07-31 Olivier Hainque <hainque@adacore.com>
9303 * gcc.c (handle_spec_function): Accept a soft_matched_part
9304 argument, as do_spec_1. Pass it down to ...
9305 (eval_spec_function): Accept a soft_matched_part argument,
9306 and pass it down to ...
9307 (do_spec_2): Accept a soft_matched_part argument, and pass
9308 it down to do_spec_1.
9309 (do_spec_1): Pass soft_matched_part to handle_spec_function.
9310 (handle_braces): Update call to handle_spec_function.
9311 (driver::set_up_specs): Update calls to do_spec_2.
9312 (compare_debug_dump_opt_spec_function): Likewise.
9313 (compare_debug_self_opt_spec_function): Likewise.
9315 2018-07-31 Olivier Hainque <hainque@adacore.com>
9317 * common.opt (nolibc): New option.
9318 * doc/invoke.texi (Link Options): Document it.
9319 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
9320 * config/alpha/linux.h: Likewise.
9321 * config/arc/elf.h: Likewise.
9322 * config/arm/uclinux-elf.h: Likewise.
9323 * config/arm/unknown-elf.h: Likewise.
9324 * config/avr/avrlibc.h: Likewise.
9325 * config/bfin/bfin.h: Likewise.
9326 * config/bfin/linux.h: Likewise.
9327 * config/bfin/uclinux.h: Likewise.
9328 * config/darwin.h: Likewise.
9329 * config/darwin10.h: Likewise.
9330 * config/darwin12.h: Likewise.
9331 * config/gnu-user.h: Likewise.
9332 * config/lm32/uclinux-elf.h: Likewise.
9333 * config/pa/pa-hpux11.h: Likewise.
9334 * config/pa/pa64-hpux.h: Likewise.
9335 * config/sparc/sparc.h: Likewise.
9337 2018-07-31 Olivier Hainque <hainque@adacore.com>
9339 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
9340 undefined variables.
9342 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
9345 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
9348 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
9350 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
9351 terminated string literal.
9353 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
9355 PR rtl-optimization/85160
9356 * combine.c (is_just_move): New function.
9357 (try_combine): Allow combining two instructions into two if neither of
9358 the original instructions was a move.
9360 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
9363 * doc/extend.texi (Global Register Variables): Discourage use of type
9365 (Local Register Variables): Likewise.
9367 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
9369 PR tree-optimization/86506
9370 * hwint.h (ceil_log2): Resync with hwint.c implementation.
9372 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
9375 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
9376 hard_regno, make sure no insn between `from` and `to` clobbers it.
9378 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
9379 Tom de Vries <tdevries@suse.de>
9381 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
9382 (PTX_DEFAULT_RUNTIME_DIM): ... this.
9383 (nvptx_goacc_validate_dims): Set default worker and gang dims to
9384 PTX_DEFAULT_RUNTIME_DIM.
9385 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
9387 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
9389 * config/pa/pa.c (pa_output_addr_vec): Align address table.
9390 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
9391 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
9393 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
9395 * config/rs6000/constraints.md (wG constraint): Delete, no longer
9397 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
9398 predicate to reflect toc fusion has been deleted.
9399 (toc_fusion_mem_raw): Delete, no longer used.
9400 (toc_fusion_mem_wrapped): Likewise.
9401 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
9403 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
9404 Delete, no longer used.
9405 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
9406 meant to be used for toc fusion.
9407 (rs6000_debug_print_mode): Delete toc fusion debugging.
9408 (rs6000_debug_reg_global): Likewise.
9409 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
9410 fusion and secondary reload support that were never used.
9411 (rs6000_option_override_internal): Delete TOC fusion, that was only
9412 partially defined, and it did not work unless you also used the
9414 (rs6000_legitimate_address_p): Delete TOC fusion support.
9415 (rs6000_opt_masks): Likewise.
9416 (fusion_wrap_memory_address): Delete function, no longer used.
9417 (fusion_split_address); Delete TOC fusion support.
9418 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
9419 longer used with toc fusion being deleted.
9420 (TARGET_TOC_FUSION_FP): Likewise.
9421 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
9423 (toc fusion spliter): Delete TOC fusion support.
9424 (toc_fusionload_<mode>): Likewise.
9425 (toc_fusionload_di): Likewise.
9426 (fusion_gpr_load_<mode>): Delete generator function, this insn no
9427 longer needs to be named. Rename predicate to delete TOC fusion.
9428 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
9429 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
9430 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
9431 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
9432 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
9434 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9436 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
9437 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
9438 __int128 in built-in function prototypes.
9439 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
9440 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
9442 2018-07-27 Martin Sebor <msebor@redhat.com>
9444 PR tree-optimization/86696
9445 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
9446 types, including enums.
9447 (handle_char_store): Be prepared for the above function to fail.
9449 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
9451 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
9452 when optimization level is lower than 2 or optimize for size.
9454 2018-07-26 Martin Sebor <msebor@redhat.com>
9456 PR tree-optimization/86043
9457 PR tree-optimization/86042
9458 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
9459 (get_string_cst_length): Rename...
9460 (get_min_string_length): ...to this. Add argument.
9461 (handle_char_store): Extend to handle multi-character stores by
9463 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
9464 * tree.h (initializer_zerop): Add argument.
9466 2018-07-26 Jakub Jelinek <jakub@redhat.com>
9469 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
9470 declare target to variables if they have always,{to,from,tofrom} map
9473 2018-07-26 Martin Liska <mliska@suse.cz>
9476 * lto-wrapper.c: Add linker_output as prefix
9477 for ltrans_output_file.
9479 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
9481 PR rtl-optimization/85805
9482 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
9483 value for hard registers if that was written in the same mode.
9485 2018-07-26 Martin Liska <mliska@suse.cz>
9487 PR gcov-profile/86536
9488 * gcov.c (format_gcov): Use printf format %.*f directly
9489 and do not handle special values.
9491 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9493 * common/config/arc/arc-common.c (arc_option_optimization_table):
9494 Update default optimizations for size.
9496 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9498 * config/arc/arc.md (movsf_insn): Add short instruction selection.
9499 * config/arc/constraints.md (CfZ): New constraint.
9500 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
9501 (subsf3_fpu): Likewise.
9502 (cmpsf_fpu): Likewise.
9503 (cmpsf_fpu_uneq): Likewise.
9505 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9507 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
9509 (arc_is_uncached_mem_p): uncached applies to both the variable and
9512 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9514 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
9517 2018-07-25 David Malcolm <dmalcolm@redhat.com>
9519 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
9520 field "m_scopes" from vec to auto_vec.
9522 2018-07-25 Martin Liska <mliska@suse.cz>
9524 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
9527 2018-07-25 Richard Biener <rguenther@suse.de>
9530 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
9531 special wrt context_die late.
9532 (gen_subprogram_die): Re-use DIEs in local scope.
9534 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
9536 PR tree-optimization/86644
9537 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
9539 2018-07-25 Martin Liska <mliska@suse.cz>
9542 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
9543 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
9545 2018-07-25 Martin Liska <mliska@suse.cz>
9548 * params.def: Explain ASan abbreviation and provide
9549 a documentation link.
9551 2018-07-24 Martin Sebor <msebor@redhat.com>
9553 PR tree-optimization/86622
9554 PR tree-optimization/86532
9555 * builtins.h (string_length): Declare.
9556 * builtins.c (c_strlen): Correct handling of non-constant offsets.
9557 (check_access): Be prepared for non-constant length ranges.
9558 (string_length): Make extern.
9559 * expr.c (string_constant): Only handle the minor non-constant
9560 array index. Use string_constant to compute the length of
9561 a generic string constant.
9563 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
9565 PR tree-optimization/86618
9566 * tree-vect-stmts.c (vectorizable_call): Don't take the address
9567 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
9569 2018-07-24 David Malcolm <dmalcolm@redhat.com>
9571 PR tree-optimization/86636
9572 * json.cc (json::object::set): Fix comment. Add assertions.
9573 (json::array::append): Move here from json.h. Add comment and an
9575 (json::string::string): Likewise.
9576 * json.h (json::array::append): Move to json.cc.
9577 (json::string::string): Likewise.
9578 * optinfo-emit-json.cc
9579 (optrecord_json_writer::impl_location_to_json): Assert that we
9580 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
9581 wrapper around it. Expand the location once, rather than three
9583 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
9584 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
9586 (optrecord_json_writer::optinfo_to_json): Likewise, in four
9587 places. Fix some overlong lines.
9589 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
9591 * config/aarch64/aarch64-simd.md
9592 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
9593 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
9594 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
9595 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
9596 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
9597 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
9598 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
9599 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
9600 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
9602 2018-07-24 Jakub Jelinek <jakub@redhat.com>
9605 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
9606 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
9607 and abs_d == d, do the power of two handling if profitable.
9609 2018-07-24 Richard Biener <rguenther@suse.de>
9611 * match.pd: Add BIT_FIELD_REF canonicalizations.
9613 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9616 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
9618 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9620 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
9621 terminated STRING_CST object.
9623 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9625 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
9627 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9629 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
9630 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
9631 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
9632 the elements into a register.
9633 (rs6000_split_v4si_init_di_reg): Delete.
9634 (rs6000_split_v4si_init): Delete.
9635 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
9636 (vsx_init_v4si): Rewrite as a define_expand.
9638 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9640 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
9641 zero_extend argument from memory): New.
9643 2018-07-22 Martin Sebor <msebor@redhat.com>
9646 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
9647 diagnosing calls with unknown arguments unless -Walloca-larger-than
9648 is restricted to less than PTRDIFF_MAX bytes.
9650 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
9652 * doc/gcov.texi (Invoking Gcov): Editorial changes.
9654 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9656 * pretty-print.c (text_info::set_location): Remove redundant
9657 "line_table" parameter from call to rich_location::set_range.
9659 2018-07-20 Martin Sebor <msebor@redhat.com>
9662 * builtins.c (expand_builtin_alloca): Adjust.
9663 * calls.c (alloc_max_size): Simplify.
9664 * cgraphunit.c (cgraph_node::expand): Adjust.
9665 * common.opt (larger_than_size, warn_frame_larger_than): Remove
9667 (frame_larger_than_size): Same.
9668 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
9669 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
9671 * doc/invoke.texi (GCC Command Options): Document option arguments.
9672 Explain byte-size arguments and suffixes.
9673 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
9674 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
9675 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
9676 * doc/options.texi (UInteger): Expand.
9677 (Host_Wide_Int, ByteSize): Document new properties.
9678 * final.c (final_start_function_1): Include sizes in an error message.
9679 * function.c (frame_offset_overflow): Same.
9680 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
9681 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
9682 Diagnose unbounded alloca calls only for limits of less than
9684 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
9685 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
9687 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
9688 only for limits of less than PTRDIFF_MAX.
9689 * langhooks-def.h (lhd_handle_option): Change function argument
9691 * langhooks.c (lhd_handle_option): Same.
9692 * langhooks.h (handle_option): Same.
9693 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
9695 (var_type, var_type_struct): Same.
9696 (var_set): Handle ByteSize flag.
9697 * optc-gen.awk: Add comments to output to ease debugging. Make
9698 use of HOST_WIDE_INT where appropriate.
9699 * opts-gen-save.awk: Use %lx to format unsigned long.
9700 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
9701 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
9702 arguments. Parse bytes-size suffixes.
9703 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
9704 (enum_value_to_arg): Same.
9705 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
9706 (handle_option): Adjust.
9707 (generate_option): Change function argument to HOST_WIDE_INT.
9708 (cmdline_handle_error): Adjust.
9709 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
9710 (set_option): Change function argument to HOST_WIDE_INT.
9711 (option_enabled): Handle cl_host_wide_int.
9712 (get_option_state): Handle CLVC_SIZE.
9713 (control_warning_option): Same.
9714 * opts.c (common_handle_option): Change function argument to
9715 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
9716 OPT_Wvla_larger_than_.
9717 * opts.h (enum cl_var_type): Add an enumerator.
9718 * stor-layout.c (layout_decl): Print a more meaningful warning.
9719 * toplev.c (output_stack_usage): Adjust.
9721 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
9723 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
9724 call to inline_expand_builtin_string_cmp.
9725 (expand_builtin_strcmp): Likewise.
9726 (expand_builtin_strncmp): Likewise.
9727 (inline_string_cmp): Delete the last parameter, change char_type_node
9728 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
9730 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
9731 the inlining expansion on target where the type of the call has same or
9732 narrower precision than unsigned char.
9734 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9736 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
9737 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
9738 * common.opt (fsave-optimization-record): New option.
9739 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
9740 * doc/invoke.texi (-fsave-optimization-record): New option.
9741 * dumpfile.c: Include "optinfo-emit-json.h".
9742 (struct kv_pair): Move to coretypes.h.
9743 (optgroup_options): Make non-static.
9744 (dump_context::end_scope): Call
9745 optimization_records_maybe_pop_dump_scope.
9746 * dumpfile.h (optgroup_options): New decl.
9747 * json.cc: New file.
9749 * optinfo-emit-json.cc: New file.
9750 * optinfo-emit-json.h: New file.
9751 * optinfo.cc: Include "optinfo-emit-json.h".
9752 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
9753 (optinfo_enabled_p): Check optimization_records_enabled_p.
9754 (optinfo_wants_inlining_info_p): Likewise.
9755 * optinfo.h: Update comment.
9756 * profile-count.c (profile_quality_as_string): New function.
9757 * profile-count.h (profile_quality_as_string): New decl.
9758 (profile_count::quality): New accessor.
9759 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
9760 and optinfo_emit_json_cc_tests.
9761 * selftest.h (selftest::json_cc_tests): New decl.
9762 (selftest::optinfo_emit_json_cc_tests): New decl.
9763 * toplev.c: Include "optinfo-emit-json.h".
9764 (compile_file): Call optimization_records_finish.
9765 (do_compile): Call optimization_records_start.
9766 * tree-ssa-live.c: Include optinfo.h.
9767 (remove_unused_scope_block_p): Retain inlining information if
9768 optinfo_wants_inlining_info_p returns true.
9770 2018-07-20 Richard Biener <rguenther@suse.de>
9773 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
9774 to cover -flto-partition=none.
9776 2018-07-20 Martin Liska <mliska@suse.cz>
9778 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
9779 (get_decl_source_range): Remove unused function.
9781 2018-07-20 Richard Biener <rguenther@suse.de>
9783 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
9784 (struct vn_phi_s): Likewise.
9785 (struct vn_reference_s): Likewise.
9786 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
9787 for searching the slot of an entry known to be in the hash itself.
9788 (vn_phi_hasher::equal): Likewise.
9789 (vn_reference_hasher::equal): Likewise.
9790 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
9792 (optimistic_info, current_info): Remove, keeping only valid_info.
9793 (vn_reference_lookup_1): Remove fallback lookup.
9794 (vn_reference_lookup_2): Likewise.
9795 (vn_nary_op_lookup_1): Likewise.
9796 (vn_phi_lookup): Likewise.
9797 (vn_nary_build_or_lookup_1): Make sure to not chain the built
9799 (vn_reference_insert): Adjust, chain the inserted hash element
9800 at last_inserted_ref.
9801 (vn_reference_insert_pieces): Likewise.
9802 (visit_reference_op_call): Likewise.
9803 (vn_nary_op_insert_into): Chain the inserted hash element at
9805 (vn_nary_op_insert_pieces): Adjust.
9806 (vn_nary_op_insert): Likewise.
9807 (vn_nary_op_insert_stmt): Likewise.
9808 (vn_phi_insert): Adjust, chain the inserted hash element at
9810 (process_scc): Remove clearing and copying the optimistic
9811 table. Instead remove elements inserted during an optimistic
9812 iteration from the single table we maintain.
9813 (init_scc_vn): Adjust.
9814 (free_scc_vn): Likewise.
9815 (sccvn_dom_walker::record_cond): Likewise.
9816 (sccvn_dom_walker::after_dom_children): Likewise.
9818 2018-07-19 Martin Sebor <msebor@redhat.com>
9820 PR tree-optimization/84047
9821 PR tree-optimization/83776
9822 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
9823 (check_array_bounds): Call it.
9825 2018-07-19 Martin Sebor <msebor@redhat.com>
9827 * align.h (align_flags): Use member initialization.
9829 2018-07-19 David Malcolm <dmalcolm@redhat.com>
9831 * Makefile.in (OBJS): Add optinfo.o.
9832 * coretypes.h (class symtab_node): New forward decl.
9833 (struct cgraph_node): New forward decl.
9834 (class varpool_node): New forward decl.
9835 * dump-context.h: New file.
9836 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
9838 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
9839 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
9840 (set_alt_dump_file): Likewise.
9841 (dump_context::~dump_context): New dtor.
9842 (dump_gimple_stmt): Move implementation to...
9843 (dump_context::dump_gimple_stmt): ...this new member function.
9844 Add the stmt to any pending optinfo, creating one if need be.
9845 (dump_gimple_stmt_loc): Move implementation to...
9846 (dump_context::dump_gimple_stmt_loc): ...this new member function.
9847 Start a new optinfo and add the stmt to it.
9848 (dump_gimple_expr): Move implementation to...
9849 (dump_context::dump_gimple_expr): ...this new member function.
9850 Add the stmt to any pending optinfo, creating one if need be.
9851 (dump_gimple_expr_loc): Move implementation to...
9852 (dump_context::dump_gimple_expr_loc): ...this new member function.
9853 Start a new optinfo and add the stmt to it.
9854 (dump_generic_expr): Move implementation to...
9855 (dump_context::dump_generic_expr): ...this new member function.
9856 Add the tree to any pending optinfo, creating one if need be.
9857 (dump_generic_expr_loc): Move implementation to...
9858 (dump_context::dump_generic_expr_loc): ...this new member
9859 function. Add the tree to any pending optinfo, creating one if
9861 (dump_printf): Move implementation to...
9862 (dump_context::dump_printf_va): ...this new member function. Add
9863 the text to any pending optinfo, creating one if need be.
9864 (dump_printf_loc): Move implementation to...
9865 (dump_context::dump_printf_loc_va): ...this new member function.
9866 Start a new optinfo and add the stmt to it.
9867 (dump_dec): Move implementation to...
9868 (dump_context::dump_dec): ...this new member function. Add the
9869 value to any pending optinfo, creating one if need be.
9870 (dump_context::dump_symtab_node): New member function.
9871 (dump_context::get_scope_depth): New member function.
9872 (dump_context::begin_scope): New member function.
9873 (dump_context::end_scope): New member function.
9874 (dump_context::ensure_pending_optinfo): New member function.
9875 (dump_context::begin_next_optinfo): New member function.
9876 (dump_context::end_any_optinfo): New member function.
9877 (dump_context::s_current): New global.
9878 (dump_context::s_default): New global.
9879 (dump_scope_depth): Delete global.
9880 (dumpfile_ensure_any_optinfo_are_flushed): New function.
9881 (dump_symtab_node): New function.
9882 (get_dump_scope_depth): Reimplement in terms of dump_context.
9883 (dump_begin_scope): Likewise.
9884 (dump_end_scope): Likewise.
9885 (selftest::temp_dump_context::temp_dump_context): New ctor.
9886 (selftest::temp_dump_context::~temp_dump_context): New dtor.
9887 (selftest::verify_item): New function.
9888 (ASSERT_IS_TEXT): New macro.
9889 (ASSERT_IS_TREE): New macro.
9890 (ASSERT_IS_GIMPLE): New macro.
9891 (selftest::test_capture_of_dump_calls): New test.
9892 (selftest::dumpfile_c_tests): Call it.
9893 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
9894 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
9895 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
9896 descriptive comment.
9897 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
9898 (dump_node, dump_bb): Move these unrelated decls.
9899 (class dump_manager): Add leading comment.
9900 * optinfo.cc: New file.
9901 * optinfo.h: New file.
9903 2018-07-19 Michael Collison <michael.collison@arm.com>
9904 Richard Henderson <rth@redhat.com>
9906 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
9907 (subti): Handle op1 zero.
9908 (subvti4, usub4ti4): New.
9909 (*sub<GPI>3_compare1_imm): New.
9910 (sub<GPI>3_carryinCV): New.
9911 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
9912 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
9914 2018-07-19 Michael Collison <michael.collison@arm.com>
9915 Richard Henderson <rth@redhat.com>
9917 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
9918 (addti3): Create simpler code if low part is already known to be 0.
9919 (addvti4, uaddvti4): New.
9920 (*add<GPI>3_compareC_cconly_imm): New.
9921 (*add<GPI>3_compareC_cconly): New.
9922 (*add<GPI>3_compareC_imm): New.
9923 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
9924 handle constants within this pattern..
9925 (*add<GPI>3_compareV_cconly_imm): New.
9926 (*add<GPI>3_compareV_cconly): New.
9927 (*add<GPI>3_compareV_imm): New.
9928 (add<GPI>3_compareV): New.
9929 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
9930 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
9931 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
9932 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
9933 with 'comparison' operator.
9934 (*add<GPI>3_compareV_cconly_imm): Ditto.
9935 (*add<GPI>3_compareV_cconly): Ditto.
9936 (*add<GPI>3_compareV_imm): Ditto.
9937 (add<GPI>3_compareV): Ditto.
9938 (add<mode>3_carryinC): Ditto.
9939 (*add<mode>3_carryinC_zero): Ditto.
9940 (*add<mode>3_carryinC): Ditto.
9941 (add<mode>3_carryinV): Ditto.
9942 (*add<mode>3_carryinV_zero): Ditto.
9943 (*add<mode>3_carryinV): Ditto.
9945 2018-07-19 Michael Collison <michael.collison@arm.com>
9946 Richard Henderson <rth@redhat.com>
9948 * config/aarch64/aarch64-modes.def (CC_V): New.
9949 * config/aarch64/aarch64-protos.h
9950 (aarch64_addti_scratch_regs): Declare
9951 (aarch64_subvti_scratch_regs): Declare.
9952 (aarch64_expand_subvti): Declare.
9953 (aarch64_gen_unlikely_cbranch): Declare
9954 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
9955 for signed overflow using CC_Vmode.
9956 (aarch64_get_condition_code_1): Handle CC_Vmode.
9957 (aarch64_gen_unlikely_cbranch): New function.
9958 (aarch64_addti_scratch_regs): New function.
9959 (aarch64_subvti_scratch_regs): New function.
9960 (aarch64_expand_subvti): New function.
9962 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9964 * config/aarch64/aarch64-option-extensions.def: New entry for profile
9966 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
9967 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
9970 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9973 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
9974 address check not strict.
9976 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9978 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
9980 (store_pair_lanes<mode>): Likewise.
9981 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
9982 enum value 'ADDR_QUERY_LDP_STP_N'.
9983 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
9984 (aarch64_print_address_internal): Add declaration.
9985 (aarch64_print_ldpstp_address): Remove.
9986 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
9987 (aarch64_print_operand): Change printing of 'y'.
9988 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
9989 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
9990 'true' rather than '1'.
9991 * config/aarch64/constraints.md (Uml): Likewise.
9992 (Uml): Rename to Umn.
9995 2018-07-19 Richard Biener <rguenther@suse.de>
9997 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
9999 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
10000 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
10001 (vn_reference_hasher): Likewise.
10002 (struct vn_tables_s): Remove obstack and alloc-pool members.
10003 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
10004 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
10005 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
10006 (vn_reference_insert_pieces): Likewise.
10007 (alloc_vn_nary_op_noinit): Adjust.
10008 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
10009 (vn_phi_eq): Adjust.
10010 (shared_lookup_phiargs): Remove.
10011 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
10012 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
10013 (visit_reference_op_call): Likewise.
10014 (copy_nary, copy_phi, copy_reference): Remove.
10015 (process_scc): Rewind the obstack when iterating. Do not
10016 copy the elements to valid_info but just move them from one
10017 hashtable to the other.
10018 (allocate_vn_table): Adjust.
10019 (free_vn_table): Likewise.
10020 (init_scc_vn): Likewise.
10021 (free_scc_vn): Likewise.
10023 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
10026 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
10027 indirect_return as function type attribute.
10028 (ix86_attribute_table): Change indirect_return to function
10030 * doc/extend.texi: Update indirect_return attribute.
10032 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
10034 * wide-int.h (widest2_int): New.
10035 * gimple-fold.c (arith_overflowed_p): Use it.
10036 * tree.h (widest2_int_cst): New.
10037 * tree-vrp.c (wide_int_binop_overflow): Rename from
10038 vrp_int_const_binop.
10039 Rewrite to work on trees.
10040 (extract_range_from_multiplicative_op_1): Abstract code to...
10041 (wide_int_range_min_max): ...here.
10042 (wide_int_range_cross_product): ...and here.
10043 (extract_range_from_binary_expr_1): Abstract overflow code to...
10044 (wide_int_range_mult_wrapping): ...here.
10045 * tree-vrp.h (wide_int_range_cross_product): New.
10046 (wide_int_range_mult_wrapping): New.
10048 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
10049 Julia Koval <julia.koval@intel.com>
10051 * config/i386/x86-tune-costs.h (skylake_memcpy,
10052 skylake_memset): Replace rep_prefix with unrolling for size 512.
10054 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
10056 PR middle-end/86544
10057 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
10058 comparision with EQ_EXPR in last stmt.
10060 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
10062 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
10063 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
10064 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
10065 previously known as "PowerPC AltiVec Built-in Functions". Move
10066 some material to new subsubsections "PowerPC AltiVec Built-in
10067 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
10069 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
10070 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
10071 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
10072 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
10074 2018-07-18 Richard Biener <rguenther@suse.de>
10076 PR tree-optimization/86557
10077 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
10080 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
10082 * config/s390/s390.c (s390_function_profiler): Generate CFI.
10084 2018-07-17 Jeff Law <law@redhat.com>
10086 * config/arm/arm.c (get_label_padding): Update for recent
10087 changes to label_to_alignment.
10089 PR tree-optimization/86010
10090 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
10092 * config/mips/mips.c (vr4130_align_insns): Update for recent
10093 changes to label_to_alignment.
10095 * config/frv/frv.c (frv_label_align): Update for recent changes
10096 to label_to_alignment.
10098 * config/nios2/nios2.c (nios2_label_align): Update for recent
10099 changes which dropped ALIGN_LABELS_LOG.
10101 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
10103 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
10104 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
10106 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10108 * config/arc/arc.c (arc_label_align): Use align_labels instead of
10109 deprecated align_labels_log.
10111 2018-07-17 Richard Biener <rguenther@suse.de>
10114 * dwarf2out.c (init_sections_and_labels): Always generate
10115 a debug_line_str_section for early LTO debug.
10116 (dwarf2out_finish): Reset debug_line_str_hash output early.
10117 Bump counter for extra dwarf5 .debug_loc labels to not conflict
10119 (dwarf2out_early_finish): Output debug_line_str.
10121 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
10123 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
10124 index register on z196 or later.
10126 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
10128 * config/s390/s390.c (s390_default_align): Set default function
10130 (s390_override_options_after_change): Call s390_default align.
10131 (s390_option_override_internal): Call s390_default_align.
10132 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
10134 2018-07-17 Jakub Jelinek <jakub@redhat.com>
10136 PR middle-end/86542
10137 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
10138 to _looptemp_ clauses, other than the first two.
10140 2018-07-17 Martin Liska <mliska@suse.cz>
10142 * opts.c: Do not enable OPT_falign_* for -Os.
10144 2018-07-17 Martin Liska <mliska@suse.cz>
10146 * align.h (MAX_CODE_ALIGN): New.
10147 (MAX_CODE_ALIGN_VALUE): New.
10148 * common/config/i386/i386-common.c (ix86_handle_option):
10149 (MAX_CODE_ALIGN): Moved to align.h.
10150 * final.c (MAX_CODE_ALIGN): Likewise.
10151 * opts.c (parse_and_check_align_values):
10152 (MAX_CODE_ALIGN): Likewise.
10153 (MAX_CODE_ALIGN_VALUE): Likewise.
10155 2018-07-17 Martin Liska <mliska@suse.cz>
10157 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
10158 in order to fulfil coding style.
10159 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
10160 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10161 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10162 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
10163 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
10164 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
10165 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
10166 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
10167 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10169 2018-07-17 Martin Liska <mliska@suse.cz>
10171 * align.h: New file.
10172 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
10174 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
10175 align_flags of label_to_alignment.
10176 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
10178 * config/m68k/m68k.c: Do not use removed align_labels_value and
10180 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
10181 (LOOP_ALIGN): Likewise.
10182 (LABEL_ALIGN): Likewise.
10183 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
10184 Remove not used macro.
10185 (rs6000_loop_align): Change return type to align_flags.
10186 (rs6000_loop_align_max_skip): Remove.
10187 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
10188 Change return type to align_flags.
10189 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
10190 Remove not used macro.
10191 (rs6000_loop_align): Change return type to align_flags.
10192 (rs6000_loop_align_max_skip): Remove.
10193 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
10194 * config/rx/rx-protos.h (rx_align_for_label): Make it
10196 * config/rx/rx.c (rx_align_for_label): Change return type
10198 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
10200 into align_flags class.
10201 (LABEL_ALIGN): Likewise.
10202 (LOOP_ALIGN): Likewise.
10203 * config/s390/s390.c (s390_label_align): Use align_flags
10205 (s390_asm_output_function_label): Likewise.
10206 * config/sh/sh.c (sh_override_options_after_change):
10207 Use align_flags class directly without macros.
10208 (find_barrier): Likewise.
10209 (barrier_align): Likewise.
10210 (sh_loop_align): Likewise.
10211 * config/spu/spu.c (spu_option_override):
10212 Use align_flags_tuple::get_value instead of removed macros.
10213 (spu_sched_init): Likewise.
10214 * config/spu/spu.h (GTY): Likewise.
10215 * config/visium/visium.c (visium_option_override):
10216 Set "8" as default secondary alignment.
10217 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
10218 in order to guarantee secondary alignment of 8.
10219 * coretypes.h: Include align.h header file.
10220 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
10221 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
10222 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
10223 * doc/tm.texi.in: Likewise.
10224 * final.c (struct label_alignment): Remove not used structure.
10225 (LABEL_ALIGN): Change type to align_flags.
10226 (LOOP_ALIGN): Likewise.
10227 (JUMP_ALIGN): Likewise.
10228 (default_loop_align_max_skip): Remove.
10229 (default_label_align_max_skip): Likewise.
10230 (default_jump_align_max_skip): Likewise.
10231 (default_label_align_after_barrier_max_skip):
10232 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
10233 (LABEL_TO_MAX_SKIP): Remove.
10234 (label_to_alignment): Return align_flags type instead of integer.
10235 (label_to_max_skip): Remove.
10236 (align_fuzz): Use align_flags type.
10237 (compute_alignments): Use align_flags type and use align_flags::max
10238 to combine multiple alignments.
10239 (grow_label_align): Grow vec instead of C array.
10240 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
10241 (shorten_branches): Use align_flags type and use align_flags::max
10242 to combine multiple alignments.
10243 (final_scan_insn_1): Remove usage of secondary alignment that comes
10244 from label alignment, but instead use proper secondary alignment
10245 which is computed in grow_label_align.
10246 * flags.h (struct align_flags_tuple): Move to align.h.
10247 (struct align_flags): Likewise.
10248 (state_align_loops): Rename to align_loops.
10249 (state_align_jumps): Rename to align_jumps.
10250 (state_align_labels): Rename to align_labels.
10251 (state_align_functions): Rename to align_functions.
10252 (align_loops_log): Remove.
10253 (align_jumps_log): Remove.
10254 (align_labels_log): Remove.
10255 (align_functions_log): Remove.
10256 (align_loops_max_skip): Remove.
10257 (align_jumps_max_skip): Remove.
10258 (align_labels_max_skip): Remove.
10259 (align_functions_max_skip): Remove.
10260 (align_loops_value): Remove.
10261 (align_jumps_value): Remove.
10262 (align_labels_value): Remove.
10263 (align_functions_value): Remove.
10264 * output.h (label_to_alignment): Change return type to align_flags.
10265 (label_to_max_skip): Remove.
10266 * target.def: Remove loop_align_max_skip, label_align_max_skip,
10267 jump_align_max_skip macros.
10268 * targhooks.h (default_loop_align_max_skip): Remove.
10269 (default_label_align_max_skip): Likewise.
10270 (default_jump_align_max_skip): Likewise.
10271 (default_label_align_after_barrier_max_skip): Remove.
10272 * toplev.c (read_log_maxskip): Use ::normalize function.
10273 (parse_N_M): Remove not used argument and also call ::normalize.
10274 (parse_alignment_opts): Do not pass unused arguments.
10275 * varasm.c (assemble_start_function): Use directly align_functions
10276 instead of removed macros.
10277 * system.h: Do not poison removed macros.
10279 2018-07-17 Jakub Jelinek <jakub@redhat.com>
10281 PR middle-end/86539
10282 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
10283 and cond temporaries don't have reference type if iterator has
10284 pointer type. For init use &for_pre_body instead of pre_p if
10285 for_pre_body is non-empty.
10287 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10289 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
10290 double-double modes to SFmode directly directly.
10291 (trunc<mode>sf2_fprs): Delete.
10293 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10295 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
10296 for conversions between IFmode and the decimal floating point modes.
10297 (init_float128_ieee): Use the correct names for conversions between
10298 KFmode and the decimal floating point modes.
10300 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10302 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
10303 for the conversions between TDmode and IFmode.
10304 (init_float128_ieee): Use more correct names for the conversions
10305 between TDmode and KFmode.
10307 2018-07-16 Jakub Jelinek <jakub@redhat.com>
10309 PR tree-optimization/86526
10310 * builtins.c (expand_builtin_memcmp): Formatting fixes.
10311 (inline_expand_builtin_string_cmp): Likewise.
10312 (inline_string_cmp): Likewise. Use c_readstr instead of
10313 builtin_memcpy_read_str. Add unit_mode temporary.
10315 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
10317 PR middle-end/86528
10318 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
10319 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
10321 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
10323 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
10324 Alphabetize prototypes of built-in functions, separating out
10325 built-in functions that are listed in this section but should be
10326 described elsewhere.
10328 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
10331 * expmed.c (emit_store_flag): Do not emit setcc followed by a
10332 conditional move when trapping comparison was split to a
10333 non-trapping one (and vice versa).
10335 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10337 * config/s390/s390.c (s390_function_profiler): Generate nops
10338 instead of profiler call sequences.
10339 * config/s390/s390.opt: Add the new option.
10341 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10343 * config/s390/s390.c (s390_function_profiler): Generate
10344 __mcount_loc section.
10345 * config/s390/s390.opt: Add the new option.
10347 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10349 * common.opt: Add the new warning.
10350 * config/s390/s390.c (s390_function_profiler): Emit "brasl
10351 %r0,__fentry__" when -mfentry is specified.
10352 (s390_option_override_internal): Disallow -mfentry for 31-bit
10354 * config/s390/s390.opt: Add the new option.
10356 2018-07-16 Richard Biener <rguenther@suse.de>
10359 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
10360 for function-local FUNCTION_DECL and RESULT_DECL immediately.
10362 2018-07-16 Martin Liska <mliska@suse.cz>
10365 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
10368 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
10370 * config/arc/arcHS.md: Update ARCHS scheduling rules.
10372 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
10374 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
10376 * config/arc/arc-cpus.def (hs4x): New cpu.
10378 * config/arc/arc-tables.opt: Regenerate.
10379 * config/arc/arc.c (arc_sched_issue_rate): New function.
10380 (TARGET_SCHED_ISSUE_RATE): Define.
10381 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
10382 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
10384 (attr tune): Add ARCHS4x tune values.
10385 (attr tune_dspmpy): Define.
10386 (*tst): Correct instruction type.
10387 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
10388 * config/arc/arcHS4x.md: New file.
10389 * config/arc/fpu.md: Update instruction type attributes.
10390 * config/arc/t-multilib: Regenerate.
10392 2018-07-16 Tom de Vries <tdevries@suse.de>
10395 * var-tracking.c (vt_initialize): Fix pre_dec handling.
10397 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10399 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
10402 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
10404 PR tree-optimization/86514
10405 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
10406 conversion to a boolean type from a type with greater precision.
10408 2018-07-16 Tom de Vries <tdevries@suse.de>
10410 * var-tracking.c (vt_initialize): Print adjusted insn slim if
10411 dump_flags request TDF_SLIM.
10413 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
10415 * fold-const.c (int_const_binop_1): Abstract...
10416 (wide_int_binop): ...wide int code here.
10417 (poly_int_binop): ...poly int code here.
10418 Abstract the rest of int_const_binop_1 into int_const_binop.
10419 * fold-const.h (wide_int_binop): New.
10420 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
10421 Remove useless PLUS/MINUS_EXPR case.
10422 (zero_nonzero_bits_from_vr): Move wide int code...
10423 (zero_nonzero_bits_from_bounds): ...here.
10424 (extract_range_from_binary_expr_1): Move mask optimization code...
10425 (range_easy_mask_min_max): ...here.
10426 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
10427 (range_easy_mask_min_max): New.
10429 2018-07-15 Jeff Law <law@redhat.com>
10432 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
10435 2018-07-14 Jim Wilson <jimw@sifive.com>
10437 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
10439 2018-07-14 Paul Koning <ni1d@arrl.net>
10441 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
10443 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
10445 * lto-streamer-out.c (copy_function_or_variable): Dump info about
10448 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
10449 Steve Munroe <munroesj52@gmail.com>
10451 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
10452 (_mm_andnot_si128): Likewise.
10453 (_mm_or_si128): Likewise.
10454 (_mm_xor_si128): Likewise.
10456 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
10458 PR middle-end/78809
10459 * builtins.c (expand_builtin_memcmp): Inline the calls first
10460 when result_eq is false.
10461 (expand_builtin_strcmp): Inline the calls first.
10462 (expand_builtin_strncmp): Likewise.
10463 (inline_string_cmp): New routine. Expand a string compare
10464 call by using a sequence of char comparison.
10465 (inline_expand_builtin_string_cmp): New routine. Inline expansion
10466 a call to str(n)cmp/memcmp.
10467 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
10469 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
10471 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
10473 * config/arm/driver-arm.c: Include arm-native.h.
10474 (host_detect_local_cpu): Use auto-generated data tables.
10475 (vendors, arm_cpu_table): Delete. Move part information to ...
10476 * config/arm/arm-cpus.in: ... here.
10477 * config/arm/parsecpu.awk (gen_native): New function.
10478 (vendor, part): New CPU fields.
10479 (END): Add support for building the native CPU detection tables.
10480 * config/arm/t-arm (arm-native.h): Add build rule.
10481 (driver-arm.o): Add dependency on arm-native.h.
10483 2018-07-13 Richard Biener <rguenther@suse.de>
10485 PR middle-end/85974
10486 * match.pd (addr1 - addr2): Allow either of the operand to
10489 2018-07-13 Tom de Vries <tdevries@suse.de>
10491 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
10494 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
10496 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
10497 arrays instead of numbered variables.
10499 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
10501 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
10502 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
10504 2018-07-13 Richard Biener <rguenther@suse.de>
10507 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
10508 instead of get_context_die.
10510 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10511 Richard Biener <rguenther@suse.de>
10513 PR middle-end/86489
10514 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
10515 that the loop latch destination where phi is defined.
10517 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
10519 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
10520 (riscv_expand_epilogue): Add assertion to check interrupt mode.
10521 (riscv_set_current_function): Extract getting interrupt type to new
10523 (riscv_get_interrupt_type): New function.
10524 (riscv_merge_decl_attributes): New function, checking interrupt type is
10526 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
10528 2018-07-12 Paul Koning <ni1d@arrl.net>
10530 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
10533 2018-07-12 Paul Koning <ni1d@arrl.net>
10535 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
10536 zero reference, add doloop_end instead.
10537 * doc/md.texi (decrement_and_branch_until_zero): Remove.
10538 (Looping patterns): Remove decrement_and_branch_until_zero. Add
10539 detail for doloop_end.
10541 2018-07-12 Martin Sebor <msebor@redhat.com>
10544 * attribs.c (decl_attributes): Reject conflicting attributes before
10545 calling attribute handlers.
10547 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10549 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
10551 (gcc::dump_manager::get_dump_file_name): likewise.
10552 (dump_begin): Likewise.
10553 * dumpfile.h (dump_begin): Update prototype.
10554 (gcc::dump_manager::get_dump_file_name,
10555 gcc::dump_manager::get_dump_file_name): Update prototype.
10557 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10559 * internal-fn.h (vectorizable_internal_fn_p): New function.
10560 * tree-vect-slp.c (compatible_calls_p): Likewise.
10561 (vect_build_slp_tree_1): Remove nops argument. Handle calls
10562 to internal functions.
10563 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
10565 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10567 * fold-const.h (inverse_conditions_p): Declare.
10568 * fold-const.c (inverse_conditions_p): New function.
10569 * match.pd: Use inverse_conditions_p. Add folds of view_converts
10570 that test the inverse condition of a conditional internal function.
10571 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
10572 * internal-fn.c (internal_fn_mask_index): Handle conditional
10573 internal functions.
10574 (vectorized_internal_fn_supported_p): New function.
10575 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
10576 (any_pred_load_store): Replace with...
10577 (need_to_predicate): ...this new variable.
10578 (redundant_ssa_names): New variable.
10579 (ifcvt_can_use_mask_load_store): Move initial checks to...
10580 (ifcvt_can_predicate): ...this new function. Handle tree codes
10581 for which a conditional internal function exists.
10582 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
10583 instead of ifcvt_can_use_mask_load_store. Update after variable
10585 (predicate_load_or_store): New function, split out from
10586 predicate_mem_writes.
10587 (check_redundant_cond_expr): New function.
10588 (value_available_p): Likewise.
10589 (predicate_rhs_code): Likewise.
10590 (predicate_mem_writes): Rename to...
10591 (predicate_statements): ...this. Use predicate_load_or_store
10592 and predicate_rhs_code.
10593 (combine_blocks, tree_if_conversion): Update after above name changes.
10594 (ifcvt_local_dce): Handle redundant_ssa_names.
10595 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
10596 general conditional functions.
10597 * tree-vect-stmts.c (vectorizable_call): Likewise.
10599 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10600 Alan Hayward <alan.hayward@arm.com>
10601 David Sherwood <david.sherwood@arm.com>
10603 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
10604 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
10605 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
10606 plus and minus and convert them into IFN_COND_FMA-based sequences.
10607 (convert_mult_to_fma): Handle conditional plus and minus.
10609 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10611 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
10612 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
10613 (cond_fnms_optab): New optabs.
10614 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
10615 internal functions.
10616 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
10617 * internal-fn.h (get_conditional_internal_fn): Declare.
10618 (get_unconditional_internal_fn): Likewise.
10619 * internal-fn.c (cond_ternary_direct): New macro.
10620 (expand_cond_ternary_optab_fn): Likewise.
10621 (direct_cond_ternary_optab_supported_p): Likewise.
10622 (FOR_EACH_COND_FN_PAIR): Likewise.
10623 (get_conditional_internal_fn): New function.
10624 (get_unconditional_internal_fn): Likewise.
10625 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
10626 (gimple_match_op::gimple_match_op): Add a new overload for 5
10628 (gimple_match_op::set_op): Likewise.
10629 (gimple_resimplify5): Declare.
10630 * genmatch.c (decision_tree::gen): Generate simplifications for
10632 * gimple-match-head.c (gimple_simplify): Define an overload for
10633 5 operands. Handle calls with 5 arguments in the top-level overload.
10634 (convert_conditional_op): Handle conversions from unconditional
10635 internal functions to conditional ones.
10636 (gimple_resimplify5): New function.
10637 (build_call_internal): Pass a fifth operand.
10638 (maybe_push_res_to_seq): Likewise.
10639 (try_conditional_simplification): Try converting conditional
10640 internal functions to unconditional internal functions.
10641 Handle 3-operand unconditional forms.
10642 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
10643 Define ternary equivalents of the current rules for binary conditional
10644 internal functions.
10645 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
10646 ternary operations.
10647 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
10648 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
10649 (optab): Handle them.
10650 (SVE_COND_FP_TERNARY): New int iterator.
10651 (sve_fmla_op, sve_fmad_op): New int attributes.
10652 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
10653 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
10654 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
10656 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10658 * target.def (preferred_else_value): New target hook.
10659 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
10660 * doc/tm.texi: Regenerate.
10661 * targhooks.h (default_preferred_else_value): Declare.
10662 * targhooks.c (default_preferred_else_value): New function.
10663 * internal-fn.h (conditional_internal_fn_code): Declare.
10664 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
10665 (get_conditional_internal_fn): Use it.
10666 (conditional_internal_fn_code): New function.
10667 * gimple-match.h (gimple_match_cond): New struct.
10668 (gimple_match_op): Add a cond member function.
10669 (gimple_match_op::gimple_match_op): Update all forms to take a
10671 * genmatch.c (expr::gen_transform): Use the same condition as res_op
10672 for the suboperation, but don't specify a particular else_value.
10673 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
10674 (visit_nary_op, visit_reference_op_load): Pass
10675 gimple_match_cond::UNCOND to the gimple_match_op constructor.
10676 * gimple-match-head.c: Include tree-eh.h
10677 (convert_conditional_op): New function.
10678 (maybe_resimplify_conditional_op): Likewise.
10679 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
10680 (gimple_resimplify2): Likewise.
10681 (gimple_resimplify3): Likewise.
10682 (gimple_resimplify4): Likewise.
10683 (maybe_push_res_to_seq): Return null for conditional operations.
10684 (try_conditional_simplification): New function.
10685 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
10687 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
10689 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
10691 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
10693 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10695 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
10697 (hash_tree): Do not hash DECL_FCONTEXT
10698 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
10699 Do not stream DECL_FCONTEXT.
10700 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
10701 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
10703 2018-07-12 Richard Biener <rguenther@suse.de>
10706 * dwarf2out.c (gen_block_die): Only output blocks when they have
10707 at least one !DECL_IGNORED_P variable.
10709 2018-07-12 Richard Biener <rguenther@suse.de>
10712 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
10713 Remove -mieee-fp handling.
10715 2018-07-12 Richard Biener <rguenther@suse.de>
10717 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
10718 left-over from last patch.
10720 2018-07-12 Jakub Jelinek <jakub@redhat.com>
10722 PR tree-optimization/86492
10723 * gimple-ssa-store-merging.c
10724 (imm_store_chain_info::coalesce_immediate_stores): Call
10725 check_no_overlap even for the merge_overlapping case. Formatting fix.
10727 2018-07-12 Richard Biener <rguenther@suse.de>
10729 PR middle-end/86479
10730 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
10731 move possibly trapping operations into the conditional.
10733 2018-07-12 Richard Biener <rguenther@suse.de>
10735 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
10736 (vn_lookup_simplify_result): Remove recursion limit applied
10738 (vn_nary_build_or_lookup_1): Adjust.
10739 (try_to_simplify): Likewise.
10740 * gimple-match-head.c (gimple_resimplify1): Instead apply one
10742 (gimple_resimplify2): Likewise.
10743 (gimple_resimplify3): Likewise.
10744 (gimple_resimplify4): Likewise.
10746 2018-07-11 Jakub Jelinek <jakub@redhat.com>
10748 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
10749 Use __mmask64 type instead of __mmask8 for __M argument.
10750 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
10751 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
10753 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
10754 __mmask16 for __M argument.
10755 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
10756 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
10757 to __mmask16 instead of __mmask8.
10758 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
10759 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
10760 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
10761 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
10762 instead of __mmask16 for __U argument.
10763 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
10764 __mmask16 instead of __mmask8 for __U argument.
10765 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
10767 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
10769 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
10771 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
10773 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
10775 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
10777 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
10779 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
10781 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
10783 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
10785 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
10786 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
10787 return type as well as __M argument type and all casts from __mmask8
10789 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
10790 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
10791 return type as well as __M argument type and all casts from __mmask8
10793 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
10794 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
10795 return type as well as __M argument type and all casts from __mmask8
10797 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
10798 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
10799 return type as well as __M argument type and all casts from __mmask8
10801 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
10802 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
10805 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
10807 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
10808 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
10811 2018-07-11 Paul Koning <ni1d@arrl.net>
10813 * doc/md.texi (define_subst): Document how multiple occurrences of
10814 the same argument in the replacement pattern are handled.
10816 2018-07-11 Paul Koning <ni1d@arrl.net>
10818 * doc/extend.texi (Common Variable Attributes): Move "mode" into
10819 alphabetical order.
10820 (Common Type Attributes): Add "mode" attribute.
10822 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
10824 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
10825 stream DECL_ORIGINAL_TYPE.
10826 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
10827 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
10828 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
10829 Do not walk original type.
10830 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
10832 (write_ts_decl_non_common_tree_pointers): Do not stream
10834 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
10835 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
10837 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10839 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
10840 thread twice from the same starting edge.
10842 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10844 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
10845 * gimple.c (gimple_call_nonnull_result_p): ...here...
10846 (gimple_call_nonnull_arg): ...and here.
10847 * gimple.h (gimple_call_nonnull_result_p): New.
10848 (gimple_call_nonnull_arg): New.
10850 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
10852 * config/arm/arm-cpus.in: Move information from fpu field of each
10853 cpu definition to the isa field.
10854 * config/arm/parsecpu.awk (fpu): Delete match rule.
10855 (gen_comm_data): Don't add bits from the CPU's FPU entry.
10857 2018-07-11 Richard Biener <rguenther@suse.de>
10860 * dwarf2out.c (init_sections_and_labels): Use
10861 output_asm_line_debug_info consistently.
10862 (dwarf2out_early_finish): Likewise.
10863 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
10866 2018-07-11 Richard Biener <rguenther@suse.de>
10868 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
10869 Rework father_bb setting in a way to avoid propagating constants
10870 multiple times on a loop body.
10872 2018-07-10 Mark Wielaard <mark@klomp.org>
10875 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
10877 2018-07-10 Richard Biener <rguenther@suse.de>
10879 * hash-map.h (hash_map::iterator::operator*): Return
10880 references to key and value.
10882 2018-07-10 Jakub Jelinek <jakub@redhat.com>
10885 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
10886 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
10887 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
10888 (gimplify_omp_for): For composite loops, move outer
10889 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
10890 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
10891 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
10892 TREE_LIST for both the original class iterator and the "last" helper
10893 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
10894 loop, remember has_decl_expr from outer composite loops for the
10895 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
10897 2018-07-09 Martin Sebor <msebor@redhat.com>
10899 PR middle-end/77357
10900 PR middle-end/86428
10901 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
10902 accessing implicitly initialized array elements.
10903 * expr.c (string_constant): Handle string initializers of
10904 character arrays within aggregates.
10905 * gimple-fold.c (fold_array_ctor_reference): Add argument.
10906 Store element offset. As a special case, handle zero size.
10907 (fold_nonarray_ctor_reference): Same.
10908 (fold_ctor_reference): Add argument. Store subobject offset.
10909 * gimple-fold.h (fold_ctor_reference): Add argument.
10911 2018-07-09 Paul Koning <ni1d@arrl.net>
10913 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
10914 (pdp11_insn_cost): New function.
10915 (pdp11_md_asm_adjust): New function.
10916 (TARGET_INVALID_WITHIN_DOLOOP): Define.
10917 (pdp11_rtx_costs): Update to match machine better.
10918 (output_addr_const_pdp11): Correct format mismatch warnings.
10919 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
10920 * config/pdp11/pdp11.md: General change to add base_cost and/or
10921 length attributes for use by new pdp11_insn_cost function.
10922 (MIN_BRANCH): Correct definition.
10924 (doloop_end): Use standard pattern name for looping pattern.
10925 (doloop_end_nocc): New.
10926 (movsf): Add another constraint alternative.
10927 (zero_extendqihi2): Add constraint alternatives for not in place
10929 (zero_extendhisi2): Remove.
10930 (shift patterns): Add CC handling variants.
10934 (define_peephole2): New peephole to recognize mov that sets CC for
10937 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10939 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
10940 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
10943 2018-07-09 Jakub Jelinek <jakub@redhat.com>
10946 * real.c (real_nextafter): Return true if result is denormal.
10948 2018-07-09 Martin Liska <mliska@suse.cz>
10950 * common.opt: Add back wrongly removed attribute.
10952 2018-07-09 Richard Biener <rguenther@suse.de>
10955 * dwarf2out.c (gen_block_die): For an early generated DIE
10956 always output high/low PC attributes.
10958 2018-07-09 Tom de Vries <tdevries@suse.de>
10960 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
10961 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
10962 onto VAR_DECL with abstract origin.
10964 2018-07-07 Jim Wilson <jimw@sifive.com>
10966 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
10968 2018-07-07 Tom de Vries <tdevries@suse.de>
10970 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
10972 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
10974 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
10976 (combine_bound): Use wide-int overflow calculation instead of
10978 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
10980 * fold-const.c (int_const_binop_2): Same.
10981 (extract_muldiv_1): Same.
10982 (fold_div_compare): Same.
10983 (fold_abs_const): Same.
10985 * poly-int.h (add): Same.
10989 * predict.c (predict_iv_comparison): Same.
10990 * profile-count.c (slow_safe_scale_64bit): Same.
10991 * simplify-rtx.c (simplify_const_binary_operation): Same.
10992 * tree-chrec.c (tree_fold_binomial): Same.
10993 * tree-data-ref.c (split_constant_offset_1): Same.
10994 * tree-if-conv.c (idx_within_array_bound): Same.
10995 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
10996 * tree-ssa-phiopt.c (minmax_replacement): Same.
10997 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
10998 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
10999 * vr-values.c (vr_values::adjust_range_with_scev): Same.
11000 * wide-int.cc (wi::add_large): Same.
11001 (wi::mul_internal): Same.
11002 (wi::sub_large): Same.
11003 (wi::divmod_internal): Same.
11004 * wide-int.h: Change overflow type to overflow_type for neg, add,
11005 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
11006 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
11007 mul_internal, divmod_internal.
11008 (overflow_type): New enum.
11009 (accumulate_overflow): New.
11011 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
11013 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
11014 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
11016 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
11018 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
11019 argument is checked for zero before entering loop, avoid checking again.
11021 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
11023 * gimplify.h (generic_expr_could_trap_p): Set as global function.
11024 * gimplify.h (generic_expr_could_trap_p): Likwise.
11025 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
11027 2018-07-06 Jakub Jelinek <jakub@redhat.com>
11029 PR tree-optimization/86401
11030 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
11031 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
11032 (fold_bit_and_mask): ... here. New helper function for match.pd.
11033 * fold-const.h (fold_bit_and_mask): Declare.
11034 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
11036 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
11039 * target.def (translate_mode_attribute): New hook.
11040 * targhooks.h (default_translate_mode_attribute): Declare.
11041 * targhooks.c (default_translate_mode_attribute): New function.
11042 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
11043 * doc/tm.texi: Regenerate.
11044 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
11045 (rs6000_translate_mode_attribute): New function.
11047 2018-07-06 Paul Koning <ni1d@arrl.net>
11049 * doc/md.texi (define_split): Document DONE and FAIL.
11050 (define_peephole2): Ditto.
11052 2018-07-05 Jeff Law <law@redhat.com>
11054 PR tree-optimization/86010
11055 * tree-ssa-dse.c (compute_trims): More aggressively trim at
11056 both the head and tail of mem* and str* calls.
11058 2018-07-05 Jim Wilson <jimw@sifive.com>
11060 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
11062 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
11064 * config/aarch64/aarch64-simd.md: correct flags text for
11065 MIN_EXPR replacement.
11067 2018-07-05 James Clarke <jrtc27@jrtc27.com>
11069 * configure: Regenerated.
11071 2018-07-05 Carl Love <cel@us.ibm.com>
11073 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
11074 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
11075 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
11076 VSX_BUILTIN_DOUBLEL_V4SF.
11078 2018-07-05 Martin Sebor <msebor@redhat.com>
11081 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
11082 than its domain to compute its the upper bound of a char array.
11084 2018-07-05 Nathan Sidwell <nathan@acm.org>
11086 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
11087 * doc/cpp.texi: Update comment.
11088 * doc/tm.texi: Rebuilt.
11089 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
11090 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
11091 * doc/extend.texi (Backwards Compatibility): Clarify it is system
11092 headers affected by extern "C".
11093 * system.h: Poison NO_IMPLICIT_EXTERN_C.
11094 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
11095 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
11096 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
11097 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
11098 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
11099 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
11100 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
11101 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
11102 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
11103 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
11104 config/sparc/sp64-elf.h, config/spu/spu.h,
11105 config/stormy16/stormy16.h, config/v850/v850.h,
11106 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
11107 define NO_IMPLICIT_EXTERN_C.
11108 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
11110 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11113 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
11114 instead of GET_MODE_SIZE when comparing Units.
11116 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11119 * rtlanal.c (set_noop_p): Constrain on mode change,
11120 include hard-reg-set.h
11122 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11124 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
11126 2018-07-05 Jakub Jelinek <jakub@redhat.com>
11129 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
11132 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
11134 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
11136 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
11139 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
11141 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
11143 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
11145 PR middle-end/86380
11146 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
11148 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
11150 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
11151 neg_*_op* variables.
11153 2018-07-04 Martin Liska <mliska@suse.cz>
11155 * tree-switch-conversion.c: Define
11156 max_ratio_for_speed and max_ratio_for_size constants.
11158 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
11159 Martin Liska <mliska@suse.cz>
11161 PR middle-end/66240
11164 * common.opt: Rename align options with 'str_' prefix.
11165 * common/config/i386/i386-common.c (set_malign_value): New
11167 (ix86_handle_option): Use it to set -falign-* options/
11168 * config/aarch64/aarch64-protos.h (struct tune_params): Change
11169 type from int to string.
11170 * config/aarch64/aarch64.c: Update default values from int
11172 * config/alpha/alpha.c (alpha_override_options_after_change):
11174 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
11175 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11176 max skip conditionally.
11177 * config/i386/freebsd.h (SUBALIGN_LOG): New.
11178 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11179 max skip conditionally.
11180 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11181 max skip conditionally.
11182 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
11183 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11184 max skip conditionally.
11185 * config/i386/i386.c (struct ptt): Change type from int to
11187 (ix86_default_align): Set default values.
11188 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
11189 max skip conditionally.
11190 * config/i386/iamcu.h (SUBALIGN_LOG): New.
11191 (ASM_OUTPUT_MAX_SKIP_ALIGN):
11192 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
11193 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11194 max skip conditionally.
11195 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
11196 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
11197 * config/i386/x86-64.h (SUBALIGN_LOG): New.
11198 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11199 max skip conditionally.
11200 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
11201 * config/ia64/ia64.c (ia64_option_override): Set default values
11202 for alignment options.
11203 * config/m68k/m68k.c: Handle new str_align_* options.
11204 * config/mips/mips.c (mips_set_compression_mode): Change
11206 (mips_option_override): Set default values for options.
11207 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
11209 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11211 * config/rx/rx.c (rx_option_override): Likewise.
11212 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
11213 (LABEL_ALIGN): Use align_labels_log.
11214 (LOOP_ALIGN): Use align_loops_align.
11215 * config/s390/s390.c (s390_asm_output_function_label): Use new
11217 * config/sh/sh.c (sh_override_options_after_change):
11218 Change type of constants.
11219 * config/spu/spu.c (spu_sched_init): Likewise.
11220 * config/sparc/sparc.c (sparc_option_override): Set default
11221 values for options.
11222 * config/visium/visium.c (visium_option_override): Likewise.
11223 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
11224 emit p2align format with last argument if it's not needed.
11225 * doc/invoke.texi: Document extended format of -falign-*.
11226 * final.c: Use align_labels alignment.
11227 * flags.h (struct target_flag_state): Change type to use
11229 (struct align_flags_tuple): New.
11230 (struct align_flags): Likewise.
11231 (align_loops_log): Redefine macro to use new types.
11232 (align_loops_max_skip): Redefine macro to use new types.
11233 (align_jumps_log): Redefine macro to use new types.
11234 (align_jumps_max_skip): Redefine macro to use new types.
11235 (align_labels_log): Redefine macro to use new types.
11236 (align_labels_max_skip): Redefine macro to use new types.
11237 (align_functions_log): Redefine macro to use new types.
11238 (align_loops): Redefine macro to use new types.
11239 (align_jumps): Redefine macro to use new types.
11240 (align_labels): Redefine macro to use new types.
11241 (align_functions): Redefine macro to use new types.
11242 (align_functions_max_skip): Redefine macro to use new types.
11243 (align_loops_value): New macro.
11244 (align_jumps_value): New macro.
11245 (align_labels_value): New macro.
11246 (align_functions_value): New macro.
11247 * function.c (invoke_set_current_function_hook): Propagate
11248 alignment values from flags to global variables default in
11250 * ipa-icf.c (sem_function::equals_wpa): Use
11251 cl_optimization_option_eq instead of memcmp.
11252 * lto-streamer.h (cl_optimization_stream_out): Support streaming
11254 (cl_optimization_stream_in): Likewise.
11255 * optc-save-gen.awk: Support strings in cl_optimization.
11256 * opth-gen.awk: Likewise.
11257 * opts.c (finish_options): Remove error checking of invalid
11259 (MAX_CODE_ALIGN): Remove.
11260 (MAX_CODE_ALIGN_VALUE): Likewise.
11261 (parse_and_check_align_values): New function.
11262 (check_alignment_argument): Likewise.
11263 (common_handle_option): Use check_alignment_argument.
11264 * opts.h (parse_and_check_align_values): Declare.
11265 * toplev.c (init_alignments): Remove.
11266 (read_log_maxskip): New.
11267 (parse_N_M): Likewise.
11268 (parse_alignment_opts): Likewise.
11269 (backend_init_target): Remove usage of init_alignments.
11270 * toplev.h (parse_alignment_opts): Declare.
11271 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
11273 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
11274 * tree.c (cl_option_hasher::equal): New.
11275 * varasm.c: Use new global macros.
11277 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
11279 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
11280 Use a simpler align directive also if MAXSKIP = ALIGN-1.
11281 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11282 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11283 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11284 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
11285 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
11286 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
11287 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
11288 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11289 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11290 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11291 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11293 2018-07-04 Martin Liska <mliska@suse.cz>
11294 Jonathan Wakely <jwakely@redhat.com>
11296 * coverage.c: Use correct type.
11297 * doc/invoke.texi: Language correction.
11299 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
11302 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
11303 ENDBRANCH for non-tail call which may return via indirect branch.
11304 * doc/extend.texi: Document indirect_return attribute.
11306 2018-07-03 Martin Sebor <msebor@redhat.com>
11308 PR tree-optimization/86274
11309 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
11311 (format_floating): Correct handling of infinities and NaNs.
11313 2018-07-03 Martin Sebor <msebor@redhat.com>
11315 * print-tree.c (print_real_cst): New function.
11316 (print_node_brief): Call it.
11317 (print_node): Ditto.
11319 2018-07-03 Jeff Law <law@redhat.com>
11321 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
11322 into a single pattern.
11324 * config/h8300/h8300.md (ors code_iterator): New.
11325 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
11326 a single pattern and single splitter.
11327 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
11328 (iorqi3_1, xorqi3_1): Likewise.
11329 (iorqi3, xorqi3 expanders): Similarly.
11331 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
11332 (movmd_internal) into a single pattern using the P mode iterator.
11333 (movmd splitters): Similarly.
11334 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
11335 (movsd splitters): Similarly.
11337 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
11338 ADDB, ADDW and ADDL into a single ADD attribute which selects the
11339 right table based on the size of the operand.
11340 * config/h8300/h8300.md (length_table): Corresponding changes. All
11341 references to "addb", "addw" and "addl" changed to "add".
11342 (btst patterns): Merge two variants into a single pattern.
11343 (tstqi, tsthi): Likewise.
11344 (addhi3_incdec, addsi3_incdec): Likewise.
11345 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
11346 (mulhi3, mulsi3): Likewise.
11347 (udivhi3, udivsi3): Likewise.
11348 (divhi3, divsi3): Likewise.
11349 (andorqi3, andorhi3, andorsi3): Likewise.
11351 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
11354 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
11355 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
11357 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11359 PR tree-optimization/85694
11360 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
11361 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
11363 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
11364 (<u>avg<mode>3_ceil): New patterns.
11366 2018-07-03 David Malcolm <dmalcolm@redhat.com>
11368 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
11369 scan-tree-dump directive.
11370 * gcc.dg/vect/slp-perm-2.c: Likewise.
11371 * gcc.dg/vect/slp-perm-3.c: Likewise.
11372 * gcc.dg/vect/slp-perm-5.c: Likewise.
11373 * gcc.dg/vect/slp-perm-6.c: Likewise.
11374 * gcc.dg/vect/slp-perm-7.c: Likewise.
11375 * gcc.dg/vect/slp-perm-8.c: Likewise.
11377 2018-07-03 Marek Polacek <polacek@redhat.com>
11379 PR middle-end/86202
11380 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
11382 2018-07-03 Richard Biener <rguenther@suse.de>
11385 * tree-ssa-structalias.c (find_func_clobbers): Properly
11386 handle indirect calls.
11388 2018-07-03 Jeff Law <law@redhat.com>
11390 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
11391 (shifts): New code iterator.
11392 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
11393 expander. Fix HImode handling on H8/SX.
11394 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
11395 (subqi3, subhi3, subsi3 expanders): Likewise.
11396 (andqi3, andhi3, andsi3 expanders): Likewise.
11397 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
11398 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
11399 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
11400 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
11401 (zero_extendqihi2, zero_extendqisi2): Likewise.
11402 (extendqihi2, extendqisi2): Likewise.
11403 (rotlqi3, rotlhi3, rotlsi3): Likewise.
11404 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
11405 (rotlqi3_1, rotlhi3_1): Likewise.
11406 (logicalhi3_sn, logicalsi3_sn): Likewise.
11407 (logicalhi3, logicalsi3): Likewise.
11409 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11411 * tree-vect-patterns.c (vect_recog_rotate_pattern)
11412 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11413 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
11414 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
11415 type to append_pattern_def_seq instead of creating a stmt_vec_info
11417 (build_mask_conversion): Likewise. Remove vinfo argument.
11418 (vect_add_conversion_to_patterm): Likewise, renaming to...
11419 (vect_add_conversion_to_pattern): ...this.
11420 (vect_recog_mask_conversion_pattern): Update call to
11421 build_mask_conversion. Pass the vector type to
11422 append_pattern_def_seq here too.
11423 (vect_recog_gather_scatter_pattern): Update call to
11424 vect_add_conversion_to_pattern.
11426 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11428 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
11429 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11430 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
11431 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
11432 STMT_VINFO_PATTERN_DEF_SEQ to null here.
11433 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
11434 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
11435 append_pattern_def_seq instead of new_pattern_def_seq.
11436 (vect_recog_divmod_pattern): Do both of the above.
11437 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
11440 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11442 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
11443 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
11444 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
11445 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
11446 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
11447 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11448 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
11449 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
11450 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
11451 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
11452 parameter with a single stmt_vec_info.
11453 (vect_recog_func_ptr): Likewise.
11454 (vect_recog_gather_scatter_pattern): Likewise, folding in...
11455 (vect_try_gather_scatter_pattern): ...this.
11456 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
11457 the stmt_vec_info of the statement to be matched. Don't clear
11458 STMT_VINFO_RELATED_STMT.
11459 (vect_pattern_recog): Update call accordingly.
11461 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11463 PR tree-optimization/85694
11464 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
11465 (uavgM3_ceil): Document new optabs.
11466 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
11467 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
11469 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
11470 (savg_ceil_optab): New optabs.
11471 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
11472 (vect_vect_recog_func_ptrs): Add it.
11473 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
11474 constant directly from the associated lhs.
11476 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11478 * tree-vect-patterns.c (vect_split_statement): New function.
11479 (vect_convert_input): Use it to try to split an existing cast.
11481 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11483 * poly-int.h (print_hex): New function.
11484 * dumpfile.h (dump_dec, dump_hex): Declare.
11485 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
11486 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
11487 min_input_precision, operation_precision and operation_sign.
11488 * tree-vect-patterns.c (vect_get_range_info): New function.
11489 (vect_same_loop_or_bb_p, vect_single_imm_use)
11490 (vect_operation_fits_smaller_type): Delete.
11491 (vect_look_through_possible_promotion): Add an optional
11492 single_use_p parameter.
11493 (vect_recog_over_widening_pattern): Rewrite to use new
11494 stmt_vec_info infomration. Handle one operation at a time.
11495 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
11496 (vect_truncatable_operation_p, vect_set_operation_type)
11497 (vect_set_min_input_precision): New functions.
11498 (vect_determine_min_output_precision_1): Likewise.
11499 (vect_determine_min_output_precision): Likewise.
11500 (vect_determine_precisions_from_range): Likewise.
11501 (vect_determine_precisions_from_users): Likewise.
11502 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
11503 (vect_vect_recog_func_ptrs): Put over_widening first.
11505 (vect_pattern_recog): Call vect_determine_precisions.
11507 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11509 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
11510 statements that have been replaced by further pattern statements.
11511 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
11513 2018-07-03 Richard Biener <rguenther@suse.de>
11515 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
11516 always set *dt. Dump vectype in vectype overload.
11517 * dumpfile.h (dump_gimple_expr): New function.
11518 (dump_gimple_expr_loc): Likewise.
11519 * dumpfile.c (dump_gimple_expr): New function.
11520 (dump_gimple_expr_loc): Likewise.
11522 2018-07-02 Jeff Law <law@redhat.com>
11524 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
11525 the H8/300, H8/300H and H8/S variants into a single pattern.
11526 (movhi_h8300, movqi_h8300hs): Similarly.
11527 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
11528 (QHI mode iterator): New.
11530 * config/h8300/h8300.md: Remove trailing whitespace.
11532 2018-07-02 Jim Wilson <jimw@sifive.com>
11534 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
11535 instead of emit_insn for interrupt returns.
11536 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
11537 (riscv_sret, riscv_uret): Likewise.
11539 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11541 * pretty-print.c (selftest::test_pp_format): Move save and restore
11542 of quotes to class auto_fix_quotes, and add an instance.
11543 * selftest.c: Include "intl.h".
11544 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
11545 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
11546 * selftest.h (selftest::auto_fix_quotes): New class.
11548 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
11550 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
11551 (aarch64_sve_prepare_conditional_op): Remove.
11552 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
11553 Allow aarch64_simd_reg_or_zero as select operand; remove
11554 the aarch64_sve_prepare_conditional_op call.
11555 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
11556 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11557 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
11558 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
11559 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
11560 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
11561 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
11562 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
11563 and a splitters to match all of the *_any patterns.
11564 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
11566 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
11567 (SVE_COND_FP_BINARY_REV): Remove.
11568 (sve_int_op_rev, sve_fp_op_rev): New.
11569 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
11570 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
11571 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
11572 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
11573 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
11574 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
11575 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
11576 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
11577 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
11579 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
11580 Remove match_dup 1 from the inner unspec.
11581 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11583 * config/aarch64/aarch64.md (movprfx): New attr.
11584 (length): Default movprfx to 8.
11585 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
11586 (*madd<SVE_I>, *msub<SVE_I): Likewise.
11587 (*<su>mul<SVE_I>3_highpart): Likewise.
11588 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
11589 (*v<ASHIFT><SVE_I>3): Likewise.
11590 (*<su><MAXMIN><SVE_I>3): Likewise.
11591 (*<su><MAXMIN><SVE_F>3): Likewise.
11592 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
11593 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
11594 (*div<SVE_F>4): Likewise.
11596 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
11598 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
11601 2018-07-02 Richard Biener <rguenther@suse.de>
11603 PR tree-optimization/86363
11604 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
11605 memset argument refers to a non-variable address.
11607 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11609 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
11610 {PLUS,MINUS}_EXPR code to...
11611 (adjust_symbolic_bound): ...here,
11612 (combine_bound): ...here,
11613 (set_value_range_with_overflow): ...and here.
11615 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11617 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
11619 (extract_range_from_abs_expr): ...here.
11621 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
11623 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
11624 -fno-omit-frame-pointer when not optimizing.
11626 2018-07-02 Martin Liska <mliska@suse.cz>
11629 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
11630 (propagate_nothrow): Likewise.
11632 2018-07-02 Martin Liska <mliska@suse.cz>
11635 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
11637 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11639 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
11640 function in r262149, changing "loc" param from source_location to
11641 const dump_location_t &.
11642 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
11643 declaration, as above.
11645 2018-07-01 Paul Koning <ni1d@arrl.net>
11647 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11648 -munit-asm, -mgnu-asm, -mdec-asm.
11649 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
11650 (pdp11_output_labelref): New.
11651 (pdp11_output_def): New.
11652 (pdp11_output_addr_vec_elt): New.
11653 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
11654 %# and %@ format codes.
11655 (pdp11_option_override): New.
11656 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
11657 (pdp11_output_ident): New.
11658 (pdp11_asm_named_section): New.
11659 (pdp11_asm_init_sections): New.
11660 (pdp11_file_start): New.
11661 (pdp11_file_end): New.
11662 (output_ascii): Use .ascii/.asciz for -mdec-asm.
11663 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
11664 %o, like %c but octal.
11665 (pdp11_option_override): New.
11666 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
11668 (DATA_SECTION_ASM_OP): Ditto.
11669 (READONLY_DATA_SECTION_ASM_OP): New.
11670 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
11671 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
11672 (ASM_OUTPUT_LABELREF): Ditto.
11673 (ASM_OUTPUT_DEF): Ditto.
11674 (ASM_OUTPUT_EXTERNAL): New.
11675 (ASM_OUTPUT_SOURCE_FILENAME): New.
11676 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
11677 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
11678 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
11679 %# and %@ format codes.
11680 * config/pdp11/pdp11.opt (mgnu-asm): New.
11681 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
11682 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
11683 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
11685 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
11687 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
11688 dereferencing path[] beyond its length.
11690 (debug_all_paths): New.
11691 (rewire_first_differing_edge): New.
11692 (adjust_paths_after_duplication): New.
11693 (duplicate_thread_path): Call adjust_paths_after_duplication.
11695 (thread_through_all_blocks): Add new argument to
11696 duplicate_thread_path.
11698 2018-06-30 Jim Wilson <jimw@sifive.com>
11700 * config/riscv/predicates.md (p2m1_shift_operand): New.
11701 (high_mask_shift_operand): New.
11702 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
11703 pattern using p2m1_shift_operand.
11704 (lshsi3_zero_extend_3+2): New combiner pattern using
11705 high_mask_shift_operand.
11707 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11709 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
11711 (vect_recog_rotate_pattern): ...here.
11712 (vect_convert_input): Try to insert casts of invariants in the
11714 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
11715 preheader to be empty.
11717 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11719 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
11720 vector type. If given, install it in the new statement's
11721 STMT_VINFO_VECTYPE.
11722 (vect_element_precision): New function.
11723 (vect_unpromoted_value): New struct.
11724 (vect_unpromoted_value::vect_unpromoted_value): New function.
11725 (vect_unpromoted_value::set_op): Likewise.
11726 (vect_look_through_possible_promotion): Likewise.
11727 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
11728 (vect_widened_op_tree, vect_convert_input): Likewise.
11729 (vect_convert_inputs, vect_convert_output): Likewise.
11730 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
11731 to handle the optional cast of the multiplication result and
11732 vect_widened_op_tree to detect the widened multiplication itself.
11733 Do not require the input and output of promotion casts to have
11734 the same sign, but base the signedness of the operation on the
11735 input rather than the result. If the pattern includes two
11736 promotions, check that those promotions have the same sign.
11737 Do not restrict the MULT_EXPR handling to a double-width result;
11738 handle quadruple-width results and wider. Use vect_convert_inputs
11739 to convert the inputs to the common type.
11740 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
11741 to handle the optional cast of the ABS result. Also allow a sign
11742 change or a sign extension between the ABS and MINUS.
11743 Use vect_widened_op_tree to detect the widened subtraction and use
11744 vect_convert_inputs to convert the inputs to the common type.
11745 (vect_handle_widen_op_by_const): Delete.
11746 (vect_recog_widen_op_pattern): New function.
11747 (vect_recog_widen_mult_pattern): Use it.
11748 (vect_recog_widen_shift_pattern): Likewise.
11749 (vect_recog_widen_sum_pattern): Use
11750 vect_look_through_possible_promotion to handle the promoted
11753 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11755 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
11756 the containing gimple_seq *.
11757 * gimple-iterator.h (gsi_for_stmt): Declare it.
11758 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
11759 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
11760 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11761 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11762 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
11764 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
11766 (vect_mark_pattern_stmts): ...here. Handle cases in which the
11767 statement being replaced is part of an existing pattern
11768 definition sequence, inserting the new pattern statements before
11770 (vect_pattern_recog_1): Don't return a bool. If the statement
11771 is already part of a pattern, instead apply pattern matching
11772 to the pattern definition statements. Don't clear the
11773 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
11774 (vect_pattern_recog): Don't break after the first match;
11775 continue processing the pattern definition statements instead.
11776 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
11778 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11780 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
11781 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11782 (vect_recog_widen_sum_pattern): Use it.
11784 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11786 * tree-vect-loop.c (vectorizable_reduction): Assert that the
11787 phi is not a pattern statement and has not been replaced by
11788 a pattern statement.
11789 * tree-vect-patterns.c (type_conversion_p): Don't check
11790 STMT_VINFO_IN_PATTERN_P.
11791 (vect_recog_vector_vector_shift_pattern): Likewise.
11792 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
11793 the pattern statement rather than the original statement; check
11794 directly for a WIDEN_MULT_EXPR here.
11795 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
11796 vect_is_simple_use to return the pattern statement rather
11797 than the original statement; use is_pattern_stmt_p to check
11798 for such a pattern statement.
11799 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
11800 to return the pattern statement rather than the original statement;
11801 don't do the same transformation here.
11802 (vect_is_simple_use): If the defining statement has been replaced
11803 by a pattern statement, return the pattern statement instead.
11804 Remove the corresponding (local) transformation from the vectype
11807 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11809 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
11810 end and default to null.
11811 * tree-vect-loop.c (vect_create_epilog_for_reduction)
11812 (vectorizable_reduction): Update calls accordingly, dropping the
11813 gimple ** argument if the passed-back statement isn't needed.
11814 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
11815 (vect_recog_rotate_pattern): Likewise.
11816 (vect_recog_mask_conversion_pattern): Likewise.
11817 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11818 (vect_mask_constant_operand_p): Likewise.
11819 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
11820 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
11821 (get_group_load_store_type, get_load_store_type): Likewise.
11822 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
11823 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
11824 (vectorizable_conversion, vectorizable_assignment): Likewise.
11825 (vectorizable_shift, vectorizable_operation): Likewise.
11826 (vectorizable_store, vect_is_simple_cond): Likewise.
11827 (vectorizable_condition, vectorizable_comparison): Likewise.
11828 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
11829 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
11830 and move it to the end. Cope with null def_stmt_outs.
11832 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
11834 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
11836 2018-06-29 Jeff Law <law@redhat.com>
11838 * config/v850/v850.c (v850_legitimate_address_p): Handle large
11839 displacements for TARGET_V850E2V3 and newer.
11840 (TARGET_LRA_P): Remove. Defaults to LRA now.
11841 * config/v850/v850.md (sign23byte_load): Remove.
11842 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
11843 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
11845 2018-06-29 Martin Liska <mliska@suse.cz>
11848 * coverage.c (coverage_init): Mangle full path name.
11849 * doc/invoke.texi: Document the change.
11850 * gcov-io.c (mangle_path): New.
11851 * gcov-io.h (mangle_path): Likewise.
11852 * gcov.c (mangle_name): Use mangle_path for path mangling.
11854 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11856 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
11857 if starting source register is not even.
11859 2018-06-29 Martin Liska <mliska@suse.cz>
11861 PR tree-optimization/86263
11862 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
11863 Make edge redirection.
11865 2018-06-29 David Malcolm <dmalcolm@redhat.com>
11867 * dumpfile.c (dump_loc): Add indentation based on scope depth.
11868 (dump_scope_depth): New variable.
11869 (get_dump_scope_depth): New function.
11870 (dump_begin_scope): New function.
11871 (dump_end_scope): New function.
11872 * dumpfile.h (get_dump_scope_depth): New declaration.
11873 (dump_begin_scope): New declaration.
11874 (dump_end_scope): New declaration.
11875 (class auto_dump_scope): New class.
11876 (AUTO_DUMP_SCOPE): New macro.
11877 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
11880 2018-06-29 Richard Biener <rguenther@suse.de>
11882 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
11883 compute_all_dependences succeeds.
11884 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
11885 exceed --param loop-max-datarefs-for-datadeps.
11887 2018-06-29 Jakub Jelinek <jakub@redhat.com>
11889 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
11891 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
11894 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
11895 alternative 0 in preferred_for_speed attribute.
11897 2018-06-28 Paul Koning <ni1d@arrl.net>
11899 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
11900 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
11901 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
11902 * config/pdp11/pdp11.md: Correct "length" attribute calculation
11903 for shift insn patterns.
11905 2018-06-28 David Malcolm <dmalcolm@redhat.com>
11907 * cgraph.c (cgraph_node::get_body): Replace assignments to
11908 "dump_file" with calls to set_dump_file.
11909 * dumpfile.c (alt_dump_file): Make static, and group with...
11910 (alt_flags): ...this definition.
11911 (dumps_are_enabled): New variable.
11912 (refresh_dumps_are_enabled): New function.
11913 (set_dump_file): New function.
11914 (set_alt_dump_file): New function.
11915 (gcc::dump_manager::dump_start): Replace assignments to
11916 "dump_file" and "alt_dump_file" with calls to set_dump_file and
11918 (gcc::dump_manager::dump_finish): Likewise.
11919 * dumpfile.h (alt_dump_file): Delete decl.
11920 (dumps_are_enabled): New variable decl.
11921 (set_dump_file): New function decl.
11922 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
11924 * tree-nested.c (lower_nested_functions): Replace assignments to
11925 "dump_file" with calls to set_dump_file.
11927 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11929 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
11930 goto_locus of each outgoing edge of each basic block.
11932 2018-06-28 Richard Biener <rguenther@suse.de>
11934 * dwarf2out.c (decl_scope_table): Remove.
11935 (push_decl_scope): Likewise.
11936 (pop_decl_scope): Likewise.
11937 (gen_type_die_for_member): Do not call push/pop_decl_scope.
11938 (gen_struct_or_union_type_die): Likewise.
11939 (gen_tagged_type_die): Likewise.
11940 (dwarf2out_init): Do not initialize decl_scope_table.
11941 (dwarf2out_c_finalize): Do not free it.
11943 2018-06-28 Richard Biener <rguenther@suse.de>
11945 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
11946 deciding whether to not re-use a DIE.
11948 2018-06-28 Richard Biener <rguenther@suse.de>
11950 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
11951 DW_AT_abstract_origin attribute.
11953 2018-06-28 Martin Liska <mliska@suse.cz>
11955 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
11956 Use newly introduced constants.
11957 * tree-switch-conversion.h (struct jump_table_cluster):
11958 Define max_ratio_for_size and max_ratio_for_speed.
11960 2018-06-28 Martin Liska <mliska@suse.cz>
11962 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11963 Add new checking assert to catch invalid state.
11964 (jump_table_cluster::can_be_handled): Handle single case
11966 (jump_table_cluster::is_beneficial): Bail out for such case.
11967 (bit_test_cluster::find_bit_tests):
11968 Add new checking assert to catch invalid state.
11969 (bit_test_cluster::can_be_handled): Handle single case
11971 (bit_test_cluster::is_beneficial): Bail out for such case.
11972 (switch_decision_tree::analyze_switch_statement):
11975 2018-06-28 Martin Liska <mliska@suse.cz>
11977 * common.opt: Introduce -completion option.
11978 * gcc.c (driver_handle_option): Handle it.
11979 (driver::main): Print completions if completion
11981 * opt-suggestions.c (option_proposer::get_completions):
11983 (option_proposer::suggest_completion): Likewise.
11984 (option_proposer::find_param_completions): Likewise.
11985 (verify_autocompletions): Likewise.
11986 (test_completion_valid_options): Likewise.
11987 (test_completion_valid_params): Likewise.
11988 (in_completion_p): Likewise.
11989 (empty_completion_p): Likewise.
11990 (test_completion_partial_match): Likewise.
11991 (test_completion_garbage): Likewise.
11992 (opt_proposer_c_tests): Likewise.
11993 * opt-suggestions.h: Declare new functions.
11994 * opts.c (common_handle_option): Handle OPT__completion_.
11995 * selftest-run-tests.c (selftest::run_tests): Add
11996 opt_proposer_c_tests.
11997 * selftest.c (assert_str_startswith): New.
11998 * selftest.h (assert_str_startswith): Likewise.
11999 (opt_proposer_c_tests): New.
12000 (ASSERT_STR_STARTSWITH): Likewise.
12002 2018-06-28 Martin Liska <mliska@suse.cz>
12004 * Makefile.in: Add opt-suggestions.o.
12005 * gcc-main.c: Include opt-suggestions.h.
12006 * gcc.c (driver::driver): Likewise.
12007 (driver::~driver): Remove m_option_suggestions.
12008 (driver::build_option_suggestions): Moved to option_proposer.
12009 (driver::suggest_option): Likewise.
12010 (driver::handle_unrecognized_options): Use option_proposer.
12011 * gcc.h (class driver): Add new memver m_option_proposer.
12012 * opt-suggestions.c: New file.
12013 * opt-suggestions.h: New file.
12015 2018-06-28 Martin Liska <mliska@suse.cz>
12017 * vec.h (class auto_string_vec): New (moved from auto_argvec).
12018 (auto_string_vec::~auto_string_vec): Likewise.
12020 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
12022 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
12023 prevent_decl_creation_for_types fields up and add reset_location field.
12024 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
12025 statement if id->reset_location is true.
12026 (copy_edges_for_bb): Do not set goto_locus on the new edges if
12027 id->reset_location is true.
12028 (copy_phis_for_bb): Force input_location on the arguments if
12029 id->reset_location is true.
12030 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
12031 is set on the function to be inlined.
12033 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
12035 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
12037 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
12039 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
12040 registers for Pmode.
12041 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
12042 hard registers for the clobbered pseudo.
12044 2018-06-27 Paul Koning <ni1d@arrl.net>
12046 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
12047 mutually exclusive options.
12048 * config/pdp11/constraints.md (h): New constraint.
12049 (O): Update definition to match shift code generation.
12050 (D): New constraint.
12051 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
12053 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
12055 (output_jump): Change arguments.
12056 (pdp11_fixed_cc_regs): New function.
12057 (pdp11_cc_mode): Ditto.
12058 (pdp11_expand_shift): Ditto.
12059 (pdp11_assemble_shift): Ditto.
12060 (pdp11_small_shift): Ditto.
12061 (pdp11_branch_cost): Remove.
12062 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
12064 (pdp11_register_move_cost): Update for CC registers.
12065 (pdp11_rtx_costs): Add case for LSHIFTRT.
12066 (pdp11_output_jump): Add CCNZ mode conditional branches.
12067 (notice_update_cc_on_set): Remove.
12068 (pdp11_cc_mode): New function.
12069 (simple_memory_operand): Correct pre/post decrement case.
12070 (no_side_effect_operand): New function.
12071 (pdp11_regno_reg_class): Add CC_REGS class.
12072 (pdp11_fixed_cc_regs): New function.
12073 (pdp11_small_shift): New function.
12074 (pdp11_expand_shift): New function to expand shift insns.
12075 (pdp11_assemble_shift): New function to output shifts.
12076 (pdp11_branch_cost): Remove.
12077 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
12078 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
12079 (WCHAR_TYPE): Ditto.
12080 (PTRDIFF_TYPE): Ditto.
12081 (ADJUST_INSN_LENGTH): New macro.
12082 (FIXED_REGISTERS): Add CC registers.
12083 (CALL_USED_REGISTERS): Ditto.
12084 (reg_class): Ditto.
12085 (REG_CLASS_NAMES): Ditto.
12086 (REG_CLASS_CONTENTS): Ditto.
12087 (SELECT_CC_MODE): Use new function.
12088 (TARGET_FLAGS_REGNUM): New macro.
12089 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
12090 (cc0_reg_rtx): Remove.
12091 (CC_STATUS_MDEP): Remove.
12092 (CC_STATUS_MDEFP_INIT): Remove.
12093 (CC_IN_FPU): Remove.
12094 (NOTICE_UPDATE_CC): Remove.
12095 (REGISTER_NAMES): Add CC registers.
12096 (BRANCH_COST): Change to constant 1.
12097 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
12099 * config/pdp11/pdp11.opt (mbcopy): Remove.
12100 (mbcopy-builtin): Remove.
12101 (mbranch-cheap): Remove.
12102 (mbranch-expensive): Remove.
12103 * config/pdp11/predicates.md (expand_shift_operand): Update to
12104 match shift code generation.
12105 (ccnz_operator): New predicate.
12106 * doc/invoke.texi (PDP-11 Options): Remove deleted options
12107 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
12108 Remove non-existent option -mabshi, -mno-abshi. Document mutually
12110 * doc/md.texi (PDP-11): Document new D and h constraints. Update
12111 description of O constraint.
12113 2018-06-27 Jeff Law <law@redhat.com>
12114 Austin Law <austinklaw@gmail.com>
12116 * config/v850/v850.md (addsi3_set_flags): New pattern.
12117 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
12118 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
12119 (zero_extendhisi2_v850_set_flags): Likewise.
12120 (zero_extendqisi2_v850_set_flags): Likewise.
12121 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
12122 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
12123 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
12125 * config/v850/v850-protos.h (notice_update_cc): Remove.
12126 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
12127 (v850_print_operand): Handle 'D' and "d".
12128 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
12129 Add handling of arithmetic/logical operations compared against zero.
12130 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
12131 Do not look at v850_compare_op, instead get mode from last argument.
12132 (v850_gen_compare): Remove
12133 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
12134 after reload for prologue insns.
12135 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
12137 (construct_save_jarl): Likewise.
12138 (TARGET_FLAGS_REGNUM): Define.
12139 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
12140 (NOTICE_UPDATE_CC): Remove.
12141 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
12142 than cc0. Conditionalize on reload_completed.
12143 (cmpsi_insn, setfcc_insn): Likewise.
12144 (tst1 splitter): Turn into define_and_split which sets the flags
12146 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
12147 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
12148 (cstoresf4, cstoredf4): Clobber the flags.
12149 (cmpsi, cmpsf, cmpdf): Remove expanders.
12150 (setf_insn): Remove pattern.
12151 (addsi3): Turn into define_and_split which clobbers the flags after
12152 reload and a suitable pattern (addsi3_clobber_flags) for use after
12154 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
12155 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
12156 (ashrsi3, ashrsi3_v850e2): Likewise.
12157 (bins): Clobber the flags.
12158 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
12159 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
12160 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
12161 (call_value_internal_short, call_value_internal_long): Likewise.
12162 (callt_save_interrupt, callt_return_interrupt): Likewise.
12163 (save_interrupt, return_interrupt): Likewise.
12164 (callt_save_all_interrupt, save_all_interrupt): Likewise.
12165 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
12166 (restore_all_interrupt, _restore_all_interrupt): Likewise.
12167 (All FP comparisons): Only allow after reload has completed.
12169 (divh, divhu): Tweak output template.
12170 (branch_z_normal, branch_z_invert): Remove
12171 (branch_nz_normal, branch_nz_invert): Likewise.
12172 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
12174 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
12175 * config/v850/v850.c (notice_update_cc): Remove.
12176 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
12177 (CC_NO_CARRY): Likewise.
12178 (NOTICE_UPDATE_CC): Define to nothing.
12179 * config/v850/v850.md: Remove block comment on cc0 handling
12180 Remove "cc" attribute from all patterns. Remove cc_status handling
12181 from all patterns. Minor formatting fixes.
12183 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12185 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
12186 (cortex-a76.cortex-a55): Likewise.
12187 * config/aarch64/aarch64-tune.md: Regenerate.
12188 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
12189 cortex-a76.cortex-a55.
12191 2018-06-27 Jeff Law <law@redhat.com>
12193 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
12194 (MULTILIB_DIRNAMES): Similarly.
12196 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
12198 * gimple.h (gimple_return_retbnd): Delete.
12199 (gimple_return_set_retbnd): Likewise.
12200 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
12201 gimple_return_set_retbnd.
12202 * gimple-pretty-print.c (dump_gimple_return): Remove call to
12203 gimple_return_retbnd and adjust.
12204 * tree-inline.h (struct copy_body_data): Remove retbnd field.
12205 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
12206 Explicitly return NULL in a couple more cases. Move assertion
12207 on debug statements and remove unreachable code.
12208 (reset_debug_binding): Do not test id->retbnd.
12209 (expand_call_inline): Do not set it.
12211 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
12213 * configure.ac: Add --disable-gcov option.
12214 * configure: Regenerate.
12215 * Makefile.in: Honour @enable_gcov@.
12216 * doc/install.texi: Document --disable-gcov.
12218 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12220 * config/arm/arm-cpus.in (cortex-a76): New entry.
12221 (cortex-a76.cortex-a55): Likewise.
12222 * config/arm/arm-tables.opt: Regenerate.
12223 * config/arm/arm-tune.md: Likewise.
12224 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
12225 * doc/invoke.texi (ARM Options): Document cortex-a76 and
12226 cortex-a76.cortex-a55.
12228 2018-06-27 Tamar Christina <tamar.christina@arm.com>
12231 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
12233 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
12235 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
12237 (EPILOGUE_USES): Likewise.
12239 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
12241 * tree-inline.c (remap_location): New function extracted from...
12242 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
12243 (copy_phis_for_bb): ...here. Call remap_location.
12244 (copy_cfg_body): Adjust call to copy_edges_for_bb.
12246 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
12248 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
12249 unaligned vsx for 16B memset.
12251 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
12254 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
12255 ieee128_float_type_node to long_double_type_node unless
12256 TARGET_LONG_DOUBLE_128 is set.
12258 2018-06-26 David Malcolm <dmalcolm@redhat.com>
12260 * cfgloop.c (get_loop_location): Convert return type from
12261 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
12262 by implicit construction from rtx_insn *, and using
12263 dump_user_location_t::from_function_decl for the fallback case.
12264 * cfgloop.h (get_loop_location): Convert return type from
12265 location_t to dump_user_location_t.
12266 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
12267 dump_printf_loc to pass in a dump_location_t rather than a
12268 location_t, via the gimple stmt.
12269 * coverage.c (get_coverage_counts): Update calls to
12270 dump_printf_loc to pass in dump_location_t rather than a
12272 * doc/optinfo.texi (Dump types): Convert example of
12273 dump_printf_loc from taking "locus" to taking "insn". Update
12274 description of the "_loc" calls to cover dump_location_t.
12275 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
12277 (dump_user_location_t::dump_user_location_t): New constructors,
12278 from gimple *stmt and rtx_insn *.
12279 (dump_user_location_t::from_function_decl): New function.
12280 (dump_loc): Make static.
12281 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
12282 const dump_location_t &.
12283 (dump_generic_expr_loc): Delete.
12284 (dump_printf_loc): Convert param "loc" from location_t to
12285 const dump_location_t &.
12286 (selftest::test_impl_location): New function.
12287 (selftest::dumpfile_c_tests): New function.
12288 * dumpfile.h: Include "profile-count.h".
12289 (class dump_user_location_t): New class.
12290 (struct dump_impl_location_t): New struct.
12291 (class dump_location_t): New class.
12292 (dump_printf_loc): Convert 2nd param from source_location to
12293 const dump_location_t &.
12294 (dump_generic_expr_loc): Delete.
12295 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
12296 const dump_location_t &.
12297 * gimple-fold.c (fold_gimple_assign): Update call to
12298 dump_printf_loc to pass in a dump_location_t rather than a
12299 location_t, via the gimple stmt.
12300 (gimple_fold_call): Likewise.
12301 * gimple-loop-interchange.cc
12302 (loop_cand::analyze_iloop_reduction_var): Update for change to
12303 check_reduction_path.
12304 (tree_loop_interchange::interchange): Update for change to
12305 find_loop_location.
12306 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
12307 change in return-type of find_loop_location.
12308 (graphite_regenerate_ast_isl): Likewise.
12309 * graphite-optimize-isl.c (optimize_isl): Likewise.
12310 * graphite.c (graphite_transform_loops): Likewise.
12311 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
12312 pass in a dump_location_t rather than a location_t, via the
12314 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12315 * ipa.c (walk_polymorphic_call_targets): Likewise.
12316 * loop-unroll.c (report_unroll): Convert "locus" param from
12317 location_t to dump_location_t.
12318 (decide_unrolling): Update for change to get_loop_location's
12320 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
12321 location_t to dump_user_location_t.
12322 (grid_find_single_omp_among_assignments_1): Updates calls to
12323 dump_printf_loc to pass in a dump_location_t rather than a
12324 location_t, via the gimple stmt.
12325 (grid_parallel_clauses_gridifiable): Convert "tloc" from
12326 location_t to dump_location_t. Updates calls to dump_printf_loc
12327 to pass in a dump_location_t rather than a location_t, via the
12329 (grid_inner_loop_gridifiable_p): Likewise.
12330 (grid_dist_follows_simple_pattern): Likewise.
12331 (grid_gfor_follows_tiling_pattern): Likewise.
12332 (grid_target_follows_gridifiable_pattern): Likewise.
12333 (grid_attempt_target_gridification): Convert initialization
12334 of local "grid" from memset to zero-initialization; FIXME: does
12335 this require C++11? Update call to dump_printf_loc to pass in a
12336 optinfo_location rather than a location_t, via the gimple stmt.
12337 * profile.c (read_profile_edge_counts): Updates call to
12338 dump_printf_loc to pass in a dump_location_t rather than a
12340 (compute_branch_probabilities): Likewise.
12341 * selftest-run-tests.c (selftest::run_tests): Call
12343 * selftest.h (dumpfile_c_tests): New decl.
12344 * tree-loop-distribution.c (pass_loop_distribution::execute):
12345 Update for change in return type of find_loop_location.
12346 * tree-parloops.c (parallelize_loops): Likewise.
12347 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
12348 "locus" from location_t to dump_user_location_t.
12349 (canonicalize_loop_induction_variables): Likewise.
12350 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
12351 for change in return type of find_loop_location.
12352 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
12353 to dump_printf_loc to pass in a dump_location_t rather than a
12354 location_t, via the stmt.
12355 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
12357 * tree-vect-loop-manip.c (find_loop_location): Convert return
12358 type from source_location to dump_user_location_t.
12359 (vect_do_peeling): Update for above change.
12360 (vect_loop_versioning): Update for change in type of
12362 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
12363 from location_t to dump_user_location_t.
12364 (vect_estimate_min_profitable_iters): Update for change in type
12366 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
12367 location_t to dump_location_t.
12368 (vect_slp_bb): Update for change in type of vect_location.
12369 * tree-vectorizer.c (vect_location): Convert from source_location
12370 to dump_user_location_t.
12371 (try_vectorize_loop_1): Update for change in vect_location's type.
12372 (vectorize_loops): Likewise.
12373 (increase_alignment): Likewise.
12374 * tree-vectorizer.h (vect_location): Convert from source_location
12375 to dump_user_location_t.
12376 (find_loop_location): Convert return type from source_location to
12377 dump_user_location_t.
12378 (check_reduction_path): Convert 1st param from location_t to
12379 dump_user_location_t.
12380 * value-prof.c (check_counter): Update call to dump_printf_loc to
12381 pass in a dump_user_location_t rather than a location_t; update
12382 call to error_at for change in type of "locus".
12383 (check_ic_target): Update call to dump_printf_loc to
12384 pass in a dump_user_location_t rather than a location_t, via the
12387 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
12389 * config/s390/s390.h (enum processor_flags): Do not use
12390 default tune parameter when -march was specified.
12392 2018-06-26 Jakub Jelinek <jakub@redhat.com>
12395 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
12396 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
12399 2018-06-26 Richard Biener <rguenther@suse.de>
12401 PR tree-optimization/86287
12403 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
12404 (vect_analyze_loop): Initialize n_stmts.
12406 2018-06-26 Richard Biener <rguenther@suse.de>
12408 PR middle-end/86271
12409 * fold-const.c (fold_convertible_p): Pointer extension
12412 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
12415 * dwarf2out.c (loc_list_has_views): Adjust comments.
12416 (dw_loc_list): Split single cross-partition range with
12419 2018-06-25 Jeff Law <law@redhat.com>
12421 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
12422 on -mbig-switch by default.
12424 * config/v850/predicates.md (const_float_1_operand): Fix match_code
12426 (const_float_0_operand): Remove unused predicate.
12427 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
12428 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
12429 (recipsf2): New expander. Original pattern now called
12431 (recipdf2, recipdf2_insn): Similarly.
12432 (rsqrtsf2, rsqrtsf2_insn): Similarly
12433 (rsqrtdf2, rsqrtdf2_insn): Similarly
12435 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
12437 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
12438 Simplify logic for FreeBSD (twice).
12440 2018-06-25 Martin Sebor <msebor@redhat.com>
12442 PR tree-optimization/86204
12443 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
12444 a strnlen result if it's less than the length of the string.
12446 2018-06-25 Martin Sebor <msebor@redhat.com>
12448 PR tree-optimization/85700
12449 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
12450 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
12451 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
12453 2018-06-25 Martin Sebor <msebor@redhat.com>
12455 * doc/extend.texi (Zero-length arrays): Update and clarify.
12457 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
12459 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
12460 added IEEE/IBM long double multilib support on PowerPC little
12461 endian Linux systems.
12462 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
12463 (MULTILIB_DEFAULTS): Likewise.
12464 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12466 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
12467 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
12468 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
12470 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
12472 PR middle-end/86311
12473 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
12474 (REORDER_45): Likewise.
12476 2018-06-25 Jeff Law <law@redhat.com>
12478 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
12479 dividend to 32 bits. Adjust length.
12480 (udivmodhi4): Cleanup output template. Fix length.
12482 2018-06-25 Carl Love <cel@us.ibm.com>
12484 * config/rs6000/vsx.md: Change word selector to prefered location.
12486 2018-06-25 Richard Biener <rguenther@suse.de>
12488 PR tree-optimization/86304
12489 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
12490 epilogue-if-converted loops as well.
12492 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
12494 * lto-section-out.c (lto_begin_section): Do not print section
12495 name for noaddr and unnumbered dumps.
12497 2018-06-25 Richard Biener <rguenther@suse.de>
12499 * tree-vectorizer.h (struct vec_info_shared): New structure
12500 with parts split out from struct vec_info and loop_nest from
12501 struct _loop_vec_info.
12502 (struct vec_info): Adjust accordingly.
12503 (struct _loop_vec_info): Likewise.
12504 (LOOP_VINFO_LOOP_NEST): Adjust.
12505 (LOOP_VINFO_DATAREFS): Likewise.
12506 (LOOP_VINFO_DDRS): Likewise.
12507 (struct _bb_vec_info): Likewise.
12508 (BB_VINFO_DATAREFS): Likewise.
12509 (BB_VINFO_DDRS): Likewise.
12510 (struct _stmt_vec_info): Add dr_aux member.
12511 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
12512 (DR_MISALIGNMENT_UNINITIALIZED): New.
12513 (set_dr_misalignment): Adjust.
12514 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
12515 (vect_analyze_loop): Adjust prototype.
12516 (vect_analyze_loop_form): Likewise.
12517 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12518 Compute dependences lazily.
12519 (vect_record_base_alignments): Use shared datarefs/ddrs.
12520 (vect_verify_datarefs_alignment): Likewise.
12521 (vect_analyze_data_refs_alignment): Likewise.
12522 (vect_analyze_data_ref_accesses): Likewise.
12523 (vect_analyze_data_refs): Likewise.
12524 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
12525 constructor parameter for shared part.
12526 (vect_analyze_loop_form): Pass in shared part and adjust.
12527 (vect_analyze_loop_2): Pass in storage for the number of
12528 stmts. Move loop nest finding to the caller. Compute
12530 (vect_analyze_loop): Pass in shared part.
12531 (vect_transform_loop): Verify shared datarefs are unchanged.
12532 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
12533 constructor parameter for shared part.
12534 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
12535 (vect_slp_bb): Verify shared datarefs are unchanged before
12537 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
12539 (new_stmt_vec_info): Initialize DR_AUX misalignment to
12540 DR_MISALIGNMENT_UNINITIALIZED.
12541 * tree-vectorizer.c (vec_info::vec_info): Add constructor
12542 parameter for shared part.
12543 (vec_info::~vec_info): Adjust.
12544 (vec_info_shared::vec_info_shared): New.
12545 (vec_info_shared::~vec_info_shared): Likewise.
12546 (vec_info_shared::save_datarefs): Likewise.
12547 (vec_info_shared::check_datarefs): Likewise.
12548 (try_vectorize_loop_1): Construct shared part live for analyses
12549 of a single loop for multiple vector sizes.
12550 * tree-parloops.c (gather_scalar_reductions): Adjust.
12552 2018-06-25 Richard Biener <rguenther@suse.de>
12554 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
12555 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
12556 (vect_analyze_data_refs): Remove similar code from here and
12557 simplify accordingly.
12559 2018-06-25 Richard Biener <rguenther@suse.de>
12561 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
12562 for reverse storage order accesses rather than asserting
12563 they cannot happen here.
12565 2018-06-25 Tom de Vries <tdevries@suse.de>
12568 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
12569 Use data16 instead of .byte for insn prefix.
12571 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
12574 * parser.c (make_char_string_pack): Pass this literal chars
12575 through cpp_interpret_string.
12576 (cp_parser_userdef_numeric_literal): Check the result of
12577 make_char_string_pack.
12579 2018-06-24 Maya Rashish <coypu@sdf.org>
12581 * ginclude/stddef.h: Simplify conditions around avoiding
12582 re-definition of __size_t.
12584 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12586 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
12587 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
12589 2018-06-22 Maya Rashish <coypu@sdf.org>
12591 * doc/invoke.texi (mno-fancy-math-387): Update for changes
12592 made to OpenBSD and NetBSD through the years.
12594 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
12596 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12597 behavior of vec_pack (vector double, vector double) to match
12598 behavior of vec_float2 (vector double, vector double).
12600 2018-06-22 Olivier Hainque <hainque@adacore.com>
12602 * gimplify.c (gimplify_function_tree): Prevent creation
12603 of a trampoline for the address of the current function
12604 passed to entry/exit instrumentation hooks.
12606 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
12609 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
12612 2018-06-22 Martin Liska <mliska@suse.cz>
12614 PR tree-optimization/86263
12615 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12616 Bail out if is_enabled is false.
12617 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
12619 (jump_table_cluster::is_enabled): New function.
12621 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12623 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
12624 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
12625 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
12626 (lto_input_ts_binfo_tree_pointers): Likewise.
12627 * tree-streamer-out.c (streamer_write_tree_bitfields,
12628 write_ts_binfo_tree_pointers): Likewise.
12629 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
12631 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12633 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
12635 2018-06-22 Martin Liska <mliska@suse.cz>
12637 * symbol-summary.h (get): Make it pure and inline move
12638 functionality from ::get function.
12639 (get): Remove and inline into ::get and ::get_create.
12640 (get_create): Move code from ::get function.
12642 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12645 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
12646 -x assembler-with-cpp.
12648 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12650 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
12651 _FILE_OFFSET_BITS=64 for C++.
12653 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
12655 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
12656 conversion insn that shows up when pr85657-3.c is compiled using
12657 IEEE 128-bit long double.
12658 (neg<mode>2_internal): Use the correct mode to check whether the
12659 mode is IBM extended.
12660 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
12661 multiply and divide external functions from being created more
12664 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12666 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
12668 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
12669 the edge can be forwarded.
12670 (cfg_layout_merge_blocks): Likewise.
12672 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12674 * except.c (finish_eh_generation): Commit edge insertions only after
12675 the EH edges have been redirected from post-landing to landing pads.
12677 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12679 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
12680 create_tmp_var_for to create the FRAME decl.
12681 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
12683 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12685 * tree-inline.c (copy_edges_for_bb): Minor tweak.
12686 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
12687 debug statement when resetting its value.
12688 (expand_call_inline): Copy the locus of the call onto the assignment
12689 of the return value, if any. Use local variable in more cases.
12691 2018-06-21 Martin Liska <mliska@suse.cz>
12693 * ipa-pure-const.c (propagate_nothrow): Use
12694 funct_state_summaries->get.
12695 (dump_malloc_lattice): Likewise.
12696 (propagate_malloc): Likewise.
12698 2018-06-21 Richard Biener <rguenther@suse.de>
12700 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
12701 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
12702 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
12704 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
12705 BLOCK_ABSTRACT_ORIGIN unconditionally.
12707 2018-06-21 David Malcolm <dmalcolm@redhat.com>
12709 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
12711 * ipa-reference.c (ipa_reference_c_finalize): Delete
12712 ipa_ref_opt_sum_summaries and set it to NULL.
12714 2018-06-21 Tom de Vries <tdevries@suse.de>
12716 PR tree-optimization/85859
12717 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
12718 test with comment from bb_no_side_effects_p.
12720 2018-06-21 Richard Biener <rguenther@suse.de>
12722 PR tree-optimization/86232
12723 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
12724 max for constant niter.
12726 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12728 * config/aarch64/aarch64-simd.md
12729 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
12731 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12733 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
12734 Make opernads of the unspec commutative.
12736 2018-06-21 Richard Biener <rguenther@suse.de>
12738 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
12739 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12740 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
12741 (vect_analyze_data_ref_dependence): Re-order checks to deal with
12743 (vect_record_base_alignments): Do not record base alignment
12744 for gathers or scatters.
12745 (vect_compute_data_ref_alignment): Drop return value that is always
12746 true. Bail out early for gathers or scatters.
12747 (vect_enhance_data_refs_alignment): Bail out early for gathers
12749 (vect_find_same_alignment_drs): Likewise.
12750 (vect_analyze_data_refs_alignment): Remove dead code.
12751 (vect_slp_analyze_and_verify_node_alignment): Likewise.
12752 (vect_analyze_data_refs): For possible gathers or scatters do
12753 not create an alternate DR, just check their possible validity
12754 and mark them. Adjust DECL_NONALIASED handling to not rely
12755 on DR_BASE_ADDRESS.
12756 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
12757 update inits of gathers or scatters.
12758 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
12759 Also copy gather/scatter flag to pattern vinfo.
12761 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12763 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12764 behavior of vec_packsu (vector unsigned long long, vector unsigned
12765 long long) to match behavior of vec_packs with same signature.
12767 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
12768 Thomas Schwinge <thomas@codesourcery.com>
12769 Cesar Philippidis <cesar@codesourcery.com>
12771 * gimplify.c (gimplify_scan_omp_clauses): Add support for
12772 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12773 (gimplify_adjust_omp_clauses): Likewise.
12774 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
12775 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
12776 (gimplify_omp_target_update): Update handling of acc update and
12778 * omp-low.c (install_var_field): Remove unused parameter
12779 base_pointers_restrict.
12780 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
12781 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
12783 (omp_target_base_pointers_restrict_p): Delete.
12784 (scan_omp_target): Update call to scan_sharing_clauses.
12785 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
12787 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
12788 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12789 (convert_local_omp_clauses): Likewise.
12790 * tree-pretty-print.c (dump_omp_clause): Likewise.
12791 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
12793 (omp_clause_code_name): Likewise.
12795 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12798 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
12801 PR tree-optimization/86231
12802 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
12803 anti-range don't overwrite *vr0min before using it to compute *vr0max.
12805 2018-06-20 Tom de Vries <tdevries@suse.de>
12807 PR tree-optimization/86097
12808 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
12809 iv type if signedness of iv type is not the same as that of *nit.
12811 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12813 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
12814 EDGE_EH edges, verify they are all EDGE_EH.
12816 2018-06-20 Maya Rashish <coypu@sdf.org>
12818 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
12820 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12822 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
12823 * config/aarch64/aarch64.c (xgene1_tunings): Add
12824 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
12825 (aarch64_mode_valid_for_sched_fusion_p):
12826 Allow 16-byte modes.
12827 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
12828 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
12830 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
12832 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
12833 * config/aarch64/iterators.md (VQ2): New mode iterator.
12835 2018-06-20 Martin Liska <mliska@suse.cz>
12837 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
12838 Change default ratio from 10 to 8.
12840 2018-06-20 Martin Liska <mliska@suse.cz>
12842 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12844 (bit_test_cluster::find_bit_tests): Likewise.
12845 (switch_decision_tree::analyze_switch_statement): Find clusters.
12846 * tree-switch-conversion.h (struct jump_table_cluster): Document
12849 2018-06-20 Martin Liska <mliska@suse.cz>
12851 * tree-switch-conversion.c (switch_conversion::collect):
12852 Record m_uniq property.
12853 (switch_conversion::expand): Bail out for special conditions.
12854 (group_cluster::~group_cluster): New.
12855 (group_cluster::group_cluster): Likewise.
12856 (group_cluster::dump): Likewise.
12857 (jump_table_cluster::emit): New.
12858 (switch_decision_tree::fix_phi_operands_for_edges): New.
12859 (struct case_node): Remove struct.
12860 (jump_table_cluster::can_be_handled): New.
12861 (case_values_threshold): Moved to header.
12862 (reset_out_edges_aux): Likewise.
12863 (jump_table_cluster::is_beneficial): New.
12864 (bit_test_cluster::can_be_handled): Likewise.
12865 (add_case_node): Remove.
12866 (bit_test_cluster::is_beneficial): New.
12867 (case_bit_test::cmp): New.
12868 (bit_test_cluster::emit): New.
12869 (expand_switch_as_decision_tree_p): Remove.
12870 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
12871 (fix_phi_operands_for_edge): Likewise.
12872 (switch_decision_tree::analyze_switch_statement): New.
12873 (compute_cases_per_edge): Move ...
12874 (switch_decision_tree::compute_cases_per_edge): ... here.
12875 (try_switch_expansion): Likewise.
12876 (switch_decision_tree::try_switch_expansion): Likewise.
12877 (record_phi_operand_mapping): Likewise.
12878 (switch_decision_tree::record_phi_operand_mapping): Likewise.
12879 (emit_case_decision_tree): Likewise.
12880 (switch_decision_tree::emit): Likewise.
12881 (balance_case_nodes): Likewise.
12882 (switch_decision_tree::balance_case_nodes): Likewise.
12883 (dump_case_nodes): Likewise.
12884 (switch_decision_tree::dump_case_nodes): Likewise.
12885 (emit_jump): Likewise.
12886 (switch_decision_tree::emit_jump): Likewise.
12887 (emit_cmp_and_jump_insns): Likewise.
12888 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
12889 (emit_case_nodes): Likewise.
12890 (switch_decision_tree::emit_case_nodes): Likewise.
12891 (conditional_probability): Remove.
12892 * tree-switch-conversion.h (enum cluster_type): New.
12894 (struct cluster): Likewise.
12895 (cluster::cluster): Likewise.
12896 (struct simple_cluster): Likewise.
12897 (simple_cluster::simple_cluster): Likewise.
12898 (struct group_cluster): Likewise.
12899 (struct jump_table_cluster): Likewise.
12900 (struct bit_test_cluster): Likewise.
12901 (struct min_cluster_item): Likewise.
12902 (struct case_tree_node): Likewise.
12903 (case_tree_node::case_tree_node): Likewise.
12904 (jump_table_cluster::case_values_threshold): Likewise.
12905 (struct case_bit_test): Likewise.
12906 (struct switch_decision_tree): Likewise.
12907 (struct switch_conversion): Likewise.
12908 (switch_decision_tree::reset_out_edges_aux): Likewise.
12910 2018-06-20 Martin Liska <mliska@suse.cz>
12912 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
12913 (hoist_edge_and_branch_if_true): Likewise.
12914 (expand_switch_using_bit_tests_p): Likewise.
12915 (struct case_bit_test): Likewise.
12916 (case_bit_test_cmp): Likewise.
12917 (emit_case_bit_tests): Likewise.
12918 (switch_conversion::switch_conversion): New class.
12919 (struct switch_conv_info): Remove old struct.
12920 (collect_switch_conv_info): More to ...
12921 (switch_conversion::collect): ... this.
12922 (check_range): Likewise.
12923 (switch_conversion::check_range): Likewise.
12924 (check_all_empty_except_final): Likewise.
12925 (switch_conversion::check_all_empty_except_final): Likewise.
12926 (check_final_bb): Likewise.
12927 (switch_conversion::check_final_bb): Likewise.
12928 (create_temp_arrays): Likewise.
12929 (switch_conversion::create_temp_arrays): Likewise.
12930 (free_temp_arrays): Likewise.
12931 (gather_default_values): Likewise.
12932 (switch_conversion::gather_default_values): Likewise.
12933 (build_constructors): Likewise.
12934 (switch_conversion::build_constructors): Likewise.
12935 (constructor_contains_same_values_p): Likewise.
12936 (switch_conversion::contains_same_values_p): Likewise.
12937 (array_value_type): Likewise.
12938 (switch_conversion::array_value_type): Likewise.
12939 (build_one_array): Likewise.
12940 (switch_conversion::build_one_array): Likewise.
12941 (build_arrays): Likewise.
12942 (switch_conversion::build_arrays): Likewise.
12943 (gen_def_assigns): Likewise.
12944 (switch_conversion::gen_def_assigns): Likewise.
12945 (prune_bbs): Likewise.
12946 (switch_conversion::prune_bbs): Likewise.
12947 (fix_phi_nodes): Likewise.
12948 (switch_conversion::fix_phi_nodes): Likewise.
12949 (gen_inbound_check): Likewise.
12950 (switch_conversion::gen_inbound_check): Likewise.
12951 (process_switch): Use the newly created class.
12952 (switch_conversion::expand): New.
12953 (switch_conversion::~switch_conversion): New.
12954 * tree-switch-conversion.h: New file.
12956 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12958 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
12959 tree-vect-patterns.c.
12960 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
12961 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
12962 (vect_recog_sad_pattern): Likewise.
12963 (vect_recog_widen_sum_pattern): Likewise.
12964 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
12965 (vect_recog_widen_shift_pattern): Remove the type_in argument.
12966 (vect_recog_rotate_pattern): Likewise.
12967 (vect_recog_mult_pattern): Likewise.
12968 (vect_recog_vector_vector_shift_pattern): Likewise.
12969 (vect_recog_divmod_pattern): Likewise.
12970 (vect_recog_mixed_size_cond_pattern): Likewise.
12971 (vect_recog_bool_pattern): Likewise.
12972 (vect_recog_mask_conversion_pattern): Likewise.
12973 (vect_try_gather_scatter_pattern): Likewise.
12974 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
12975 (vect_recog_over_widening_pattern): Likewise.
12976 (vect_recog_gather_scatter_pattern): Likewise.
12977 (vect_recog_func_ptr): Move from tree-vectorizer.h
12978 (vect_vect_recog_func_ptrs): Move further down the file.
12979 (vect_recog_func): Likewise. Remove the third argument.
12980 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
12981 (vect_pattern_recog_1): Expect the pattern function to do any
12982 necessary target tests. Also expect it to provide a vector type.
12983 Remove the type_in handling.
12985 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12987 * tree-vect-patterns.c (vect_pattern_detected): New function.
12988 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
12989 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
12990 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
12991 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
12992 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
12993 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
12994 (vect_recog_mask_conversion_pattern)
12995 (vect_try_gather_scatter_pattern): Likewise.
12997 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12999 * tree-vect-patterns.c (vect_get_internal_def): New function.
13000 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
13001 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
13002 (search_type_for_mask_1): Use it.
13004 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13006 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
13007 redundant WIDEN_SUM_EXPR handling.
13008 (vect_recog_sad_pattern): Likewise.
13010 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13012 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
13013 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
13014 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
13015 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
13016 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
13018 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13020 * tree-vect-stmts.c (vectorizable_call): Make sure that we
13021 use the stmt_vec_info of the original bb statement for the
13022 new zero assignment, even if the call is part of a pattern.
13024 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13026 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
13027 that the sequence is attached to the original statement rather
13028 than the pattern statement.
13029 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
13030 PATTERN_DEF_SEQ from the original statement rather than
13031 the main pattern statement.
13032 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
13033 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
13034 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
13036 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13038 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
13039 definition statements before the early exit for statements that aren't
13041 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
13043 (vect_transform_loop): ...here. Process pattern definition
13044 statements without first checking whether the main pattern
13045 statement is live or relevant.
13047 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
13049 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
13050 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
13052 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
13054 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
13055 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
13056 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
13057 (expand_block_compare): Change select_block_compare_mode call.
13058 (expand_strncmp_align_check): Use new functions, fix comment.
13059 (emit_final_str_compare_gpr): New function.
13060 (expand_strn_compare): Refactor and clean up code.
13061 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
13063 2018-06-19 Tony Reix <tony.reix@atos.com>
13064 Damien Bergamini <damien.bergamini@atos.com>
13065 David Edelsohn <dje.gcc@gmail.com>
13067 * collect2.c (static_obj): New variable.
13068 (static_libs): New variable.
13069 (is_in_list): Uncomment declaration.
13070 (main): Track AIX libraries linked statically.
13071 (is_in_list): Uncomment definition.
13072 (scan_prog_file): Don't add AIX shared libraries initializer
13073 to constructor list if linking statically.
13075 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
13077 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
13079 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
13081 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
13083 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
13086 2018-06-19 Martin Liska <mliska@suse.cz>
13088 * config/i386/i386.c (ix86_can_inline_p): Do not use
13089 ipa_fn_summaries::get_create.
13090 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
13092 (devirtualization_time_bonus): Likewise.
13093 (ipcp_propagate_stage): Likewise.
13094 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
13095 (edge_set_predicate): Likewise.
13096 (evaluate_conditions_for_known_args): Likewise.
13097 (evaluate_properties_for_edge): Likewise.
13098 (ipa_call_summary::reset): Tranform to ...
13099 (ipa_call_summary::~ipa_call_summary): ... this.
13100 (ipa_fn_summary::reset): Transform to ...
13101 (ipa_fn_summary::~ipa_fn_summary): ... this.
13102 (ipa_fn_summary_t::remove): Rename to ...
13103 (ipa_fn_summary_t::remove_callees): ... this.
13104 (ipa_fn_summary_t::duplicate): Use placement new
13105 instead of memory copy.
13106 (ipa_call_summary_t::duplicate): Likewise.
13107 (ipa_call_summary_t::remove): Remove.
13108 (dump_ipa_call_summary): Change get_create to get.
13109 (ipa_dump_fn_summary): Dump only when summary exists.
13110 (analyze_function_body): Use symbol_summary::get instead
13112 (compute_fn_summary): Likewise.
13113 (estimate_edge_devirt_benefit): Likewise.
13114 (estimate_edge_size_and_time): Likewise.
13115 (inline_update_callee_summaries): Likewise.
13116 (remap_edge_change_prob): Likewise.
13117 (remap_edge_summaries): Likewise.
13118 (ipa_merge_fn_summary_after_inlining): Likewise.
13119 (write_ipa_call_summary): Likewise.
13120 (ipa_fn_summary_write): Likewise.
13121 (ipa_free_fn_summary): Likewise.
13122 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
13123 (struct ipa_call_summary): Likewise.
13124 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
13126 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
13127 (estimate_size_after_inlining): Likewise.
13128 (estimate_growth): Likewise.
13129 (growth_likely_positive): Likewise.
13130 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
13131 (inline_call): Likewise.
13132 * ipa-inline.c (caller_growth_limits): Likewise.
13133 (can_inline_edge_p): Likewise.
13134 (can_inline_edge_by_limits_p): Likewise.
13135 (compute_uninlined_call_time): Likewise.
13136 (compute_inlined_call_time): Likewise.
13137 (want_inline_small_function_p): Likewise.
13138 (edge_badness): Likewise.
13139 (update_caller_keys): Likewise.
13140 (update_callee_keys): Likewise.
13141 (inline_small_functions): Likewise.
13142 (inline_to_all_callers_1): Likewise.
13143 (dump_overall_stats): Likewise.
13144 (early_inline_small_functions): Likewise.
13145 (early_inliner): Likewise.
13146 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13147 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13148 * ipa-pure-const.c (malloc_candidate_p): Likewise.
13149 * ipa-split.c (execute_split_functions): Likewise.
13150 * symbol-summary.h: Likewise.
13151 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
13153 2018-06-19 Richard Biener <rguenther@suse.de>
13155 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
13156 (vectorize_loops): ... here. Fix dbgcnt handling.
13157 (try_vectorize_loop): Wrap try_vectorize_loop_1.
13159 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
13162 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
13163 ieee128 argument takes up only one (vector) register, not two (floating
13166 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
13168 * gimplify.c (gimplify_init_constructor): Really never clear for an
13169 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
13171 2018-06-19 Richard Biener <rguenther@suse.de>
13173 PR tree-optimization/86179
13174 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
13175 after failed recognition.
13177 2018-06-18 Martin Sebor <msebor@redhat.com>
13179 PR middle-end/85602
13180 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
13181 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
13182 Handle integer subtraction.
13183 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
13184 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
13186 2018-06-18 David Malcolm <dmalcolm@redhat.com>
13188 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
13189 param from rtx to rtx_insn *.
13190 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
13192 (frv_ifcvt_modify_insn): Likwise.
13193 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
13194 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
13195 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
13196 as_a <rtx_insn *> cast to local "unprotected_region" once
13197 it's been established that it's not NULL or pc_rtx.
13198 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
13199 param "sethi" from rtx to rtx_insn *.
13200 (nds32_group_float_insns): Likewise for param "insn".
13201 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
13203 (vax_output_int_subtract): Likewise.
13204 * config/vax/vax.c (vax_output_int_add): Likewise for param
13206 (vax_output_int_subtract): Likewise.
13207 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
13208 (emit_pattern_after): Likewise for param "after".
13209 (emit_insn_after): Likewise.
13210 (emit_jump_insn_after): Likewise.
13211 (emit_call_insn_after): Likewise.
13212 (emit_debug_insn_after): Likewise.
13213 (emit_pattern_before): Likewise for param "before".
13214 (emit_insn_before): Likewise.
13215 (emit_jump_insn_before): Likewise.
13216 * final.c (get_insn_template): Likewise for param "insn", removing
13218 * output.h (get_insn_template): Likewise for 2nd param.
13219 * rtl.h (emit_insn_before): Likewise.
13220 (emit_jump_insn_before): Likewise.
13221 (emit_debug_insn_before_noloc): Likewise.
13222 (emit_insn_after): Likewise.
13223 (emit_jump_insn_after): Likewise.
13224 (emit_call_insn_after): Likewise.
13225 (emit_debug_insn_after): Likewise.
13226 (set_insn_deleted): Likewise for param.
13228 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
13231 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
13232 floating point modes, so that IFmode is numerically greater than
13233 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
13234 to declare the ordering. This prevents IFmode from being
13235 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
13236 machine. Include rs6000-modes.h to share the fractional values
13237 between genmodes* and the rest of the compiler.
13238 (IFmode): Likewise.
13239 (KFmode): Likewise.
13240 (TFmode): Likewise.
13241 * config/rs6000/rs6000-modes.h: New file.
13242 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
13243 meaning of rs6000_long_double_size so that 126..128 selects an
13244 appropriate 128-bit floating point type.
13245 (rs6000_option_override_internal): Likewise.
13246 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
13247 (TARGET_LONG_DOUBLE_128): Change the meaning of
13248 rs6000_long_double_size so that 126..128 selects an appropriate
13249 128-bit floating point type.
13250 (LONG_DOUBLE_TYPE_SIZE): Update comment.
13251 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
13252 source and destination to match the standard usage.
13253 (truncifkf2): Likewise.
13254 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
13255 ISA 2.07 to use an explicit clobber, instead of passing in a
13257 (copysign<mode>3_soft): Likewise.
13259 2018-06-18 David Malcolm <dmalcolm@redhat.com>
13261 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13262 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
13263 (vect_slp_analyze_instance_dependence): Likewise.
13264 (vect_enhance_data_refs_alignment): Likewise.
13265 (vect_analyze_data_refs_alignment): Likewise.
13266 (vect_slp_analyze_and_verify_instance_alignment
13267 (vect_analyze_data_ref_accesses): Likewise.
13268 (vect_prune_runtime_alias_test_list): Likewise.
13269 (vect_analyze_data_refs): Likewise.
13270 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
13271 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
13272 (vect_analyze_scalar_cycles_1): Likewise.
13273 (vect_get_loop_niters): Likewise.
13274 (vect_analyze_loop_form_1): Likewise.
13275 (vect_update_vf_for_slp): Likewise.
13276 (vect_analyze_loop_operations): Likewise.
13277 (vect_analyze_loop): Likewise.
13278 (vectorizable_induction): Likewise.
13279 (vect_transform_loop): Likewise.
13280 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
13281 * tree-vect-slp.c (vect_analyze_slp): Likewise.
13282 (vect_make_slp_decision): Likewise.
13283 (vect_detect_hybrid_slp): Likewise.
13284 (vect_slp_analyze_operations): Likewise.
13285 (vect_slp_bb): Likewise.
13286 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
13287 (vectorizable_bswap): Likewise.
13288 (vectorizable_call): Likewise.
13289 (vectorizable_simd_clone_call): Likewise.
13290 (vectorizable_conversion): Likewise.
13291 (vectorizable_assignment): Likewise.
13292 (vectorizable_shift): Likewise.
13293 (vectorizable_operation): Likewise.
13294 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
13296 2018-06-18 Martin Sebor <msebor@redhat.com>
13298 PR tree-optimization/81384
13299 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
13300 * builtins.c (expand_builtin_strnlen): New function.
13301 (expand_builtin): Call it.
13302 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
13303 * builtins.def (BUILT_IN_STRNLEN): New.
13304 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
13305 Warn for bounds in excess of maximum object size.
13306 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
13307 single-value ranges. Handle strnlen.
13308 (handle_builtin_strlen): Handle strnlen.
13309 (strlen_check_and_optimize_stmt): Same.
13310 * doc/extend.texi (Other Builtins): Document strnlen.
13312 2018-06-18 Maya Rashish <coypu@sdf.org>
13314 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
13315 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
13316 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
13318 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
13320 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
13322 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13324 * tree.c (escaped_string::escape): Replace cast to char * by
13325 const_cast<char *> (unescaped).
13327 2018-06-18 Nick Clifton <nickc@redhat.com>
13330 * tree.c (escaped_string): New class. Converts an unescaped
13331 string into its escaped equivalent.
13332 (warn_deprecated_use): Use the new class to convert the
13333 deprecation message, if present.
13334 (test_escaped_strings): New self test.
13335 (test_c_tests): Add test_escaped_strings.
13336 * doc/extend.texi (deprecated): Add a note that the
13337 deprecation message is affected by the -fmessage-length
13338 option, and that control characters will be escaped.
13339 (#pragma GCC error): Document this pragma.
13340 (#pragma GCC warning): Likewise.
13341 * doc/invoke.texi (-fmessage-length): Document this option's
13342 effect on the #warning and #error preprocessor directives and
13343 the deprecated attribute.
13345 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
13347 * tree.c (decl_value_expr_lookup): Revert latest change.
13348 (decl_value_expr_insert): Likewise.
13350 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
13352 * gimplify.c (nonlocal_vlas): Delete.
13353 (nonlocal_vla_vars): Likewise.
13354 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
13356 (gimplify_body): Do not create and destroy nonlocal_vlas.
13357 * tree-nested.c: Include diagnostic.h.
13358 (use_pointer_in_frame): Tweak.
13359 (lookup_field_for_decl): Add assertion and declare the transformation.
13360 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
13361 internal error when the reference is in a wrong context. Do not
13362 create a debug decl by default.
13363 (note_nonlocal_block_vlas): Delete.
13364 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
13365 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
13366 create a debug decl by default.
13367 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
13368 call is in a wrong context.
13369 (fixup_vla_decls): New function.
13370 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
13371 debug variables were created.
13372 * tree.c (decl_value_expr_lookup): Add checking assertion.
13373 (decl_value_expr_insert): Likewise.
13375 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
13377 PR middle-end/82479
13378 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
13379 * tree-scalar-evolution.c (interpret_expr): Likewise.
13380 (expression_expensive_p): Likewise.
13381 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
13382 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
13383 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
13384 (ssa_defined_by_minus_one_stmt_p): New.
13386 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
13388 PR middle-end/64946
13389 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
13390 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
13391 * dojump.c (do_jump): Likewise.
13392 * expr.c (expand_expr_real_2): Check operand type's sign.
13393 * fold-const.c (const_unop): Handle ABSU_EXPR.
13394 (fold_abs_const): Likewise.
13395 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
13396 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
13397 (strip_sign_op_1): Likesise.
13398 * match.pd: Add new pattern to generate ABSU_EXPR.
13399 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
13400 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
13401 * tree-eh.c (operation_could_trap_helper_p): Likewise.
13402 * tree-inline.c (estimate_operator_cost): Likewise.
13403 * tree-pretty-print.c (dump_generic_node): Likewise.
13404 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
13405 * tree.def (ABSU_EXPR): New.
13407 2018-06-16 Jakub Jelinek <jakub@redhat.com>
13409 PR middle-end/86095
13410 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
13411 documented as preserved for backward compatibility only.
13412 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
13414 PR rtl-optimization/86108
13415 * bb-reorder.c (create_forwarder_block): Renamed to ...
13416 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
13417 jump from new landing pad to the second part.
13418 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
13421 2018-06-15 Jakub Jelinek <jakub@redhat.com>
13423 PR middle-end/85878
13424 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
13425 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
13426 Only call store_expr for halves if the mode is the same.
13428 PR middle-end/86123
13429 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
13430 Fix up comment formatting.
13432 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13434 * typed-splay-tree.h (typed_splay_tree::remove): New function.
13435 (typed_splay_tree::closure,
13436 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
13437 (typed_splay_tree::typed_splay_tree,
13438 typed_splay_tree::operator =): Declared private.
13439 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
13440 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
13441 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
13442 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
13443 typed_splay_tree::splay_tree_splay,
13444 typed_splay_tree::splay_tree_foreach_helper,
13445 typed_splay_tree::splay_tree_insert,
13446 typed_splay_tree::splay_tree_remove,
13447 typed_splay_tree::splay_tree_lookup,
13448 typed_splay_tree::splay_tree_predecessor,
13449 typed_splay_tree::splay_tree_successor,
13450 typed_splay_tree::splay_tree_min,
13451 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
13452 (typed_splay_tree::root, typed_splay_tree::comp,
13453 typed_splay_tree::delete_key,
13454 typed_splay_tree::delete_value): New data members.
13455 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
13456 typed_splay_tree::remove.
13458 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
13460 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
13461 -mginv and -mno-ginv to the assembler.
13462 * config/mips/mips.opt (-mcrc): New option.
13463 (-mginv): Likewise.
13464 * doc/invoke.text (-mcrc): Document.
13465 (-mginv): Likewise.
13467 2018-06-15 Nick Clifton <nickc@redhat.com>
13470 * tree.c (escaped_string): New class. Converts an unescaped
13471 string into its escaped equivalent.
13472 (warn_deprecated_use): Use the new class to convert the
13473 deprecation message, if present.
13474 (test_escaped_strings): New self test.
13475 (test_c_tests): Add test_escaped_strings.
13476 * doc/extend.texi (deprecated): Add a note that the
13477 deprecation message is affected by the -fmessage-length
13478 option, and that control characters will be escaped.
13479 (#pragma GCC error): Document this pragma.
13480 (#pragma GCC warning): Likewise.
13481 * doc/invoke.texi (-fmessage-length): Document this option's
13482 effect on the #warning and #error preprocessor directives and
13483 the deprecated attribute.
13485 2018-06-15 Richard Biener <rguenther@suse.de>
13487 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
13488 here, also noting vector size used.
13489 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
13490 size used in MSG_OPTIMIZED_LOCATIONS dump.
13491 (pass_slp_vectorize::execute): Adjust.
13493 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
13496 * config/arc/arc.c (arc_return_address_register): Fix
13499 2018-06-15 Richard Biener <rguenther@suse.de>
13501 PR middle-end/86159
13502 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
13503 leave useless conversion stripping to force_gimple_operand_gsi.
13504 (gimplify_build2): Likewise.
13505 (gimplify_build1): Likewise.
13507 2018-06-15 Richard Biener <rguenther@suse.de>
13509 PR middle-end/86076
13510 * tree-cfg.c (move_stmt_op): unshare invariant addresses
13511 before adjusting their block.
13513 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
13515 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
13516 multilibs for *-*-rtems*.
13517 * config/riscv/t-rtems: New file.
13519 2018-06-14 Jakub Jelinek <jakub@redhat.com>
13521 PR middle-end/86122
13522 * match.pd ((A +- CST1) +- CST2): Punt if last resort
13523 unsigned_type_for returns NULL.
13526 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
13527 subregs of multi-word pseudos unless the float mode has word size.
13529 2018-06-14 Richard Biener <rguenther@suse.de>
13531 PR middle-end/86139
13532 * tree-vect-generic.c (build_word_mode_vector_type): Remove
13533 duplicate and harmful type_hash_canon.
13534 * tree.c (type_hash_canon): Assert we didn't find ourselves.
13536 2018-06-14 Richard Biener <rguenther@suse.de>
13539 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
13542 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
13544 * config/rtems.h (STDINT_LONG32): Define.
13546 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
13547 Prachi Godbole <prachi.godbole@imgtec.com>
13549 * config/mips/mips-cpus.def: Define P6600.
13550 * config/mips/mips-tables.opt: Regenerate.
13551 * config/mips/mips.c (mips_ucbranch_type): New enum.
13552 (mips_rtx_cost_data): Add support for P6600.
13553 (mips_issue_rate): Likewise.
13554 (mips_multipass_dfa_lookahead): Likewise.
13555 (mips_avoid_hazard): Likewise.
13556 (mips_reorg_process_insns): Likewise.
13557 (mips_classify_branch_p6600): New function.
13558 * config/mips/mips.h (TUNE_P6600): New define.
13559 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
13560 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
13561 * config/mips/mips.md: Include p6600.md.
13562 (processor): Add p6600.
13563 * config/mips/p6600.md: New file.
13564 * doc/invoke.texi: Add p6600 to supported architectures.
13566 2018-06-13 Martin Sebor <msebor@redhat.com>
13568 PR tree-optimization/86114
13569 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
13571 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
13573 2018-06-13 Richard Biener <rguenther@suse.de>
13575 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
13576 Properly set vector type of the intermediate stmt.
13577 * tree-vect-stmts.c (vectorizable_operation): The destination
13578 var always has vectype_out type.
13580 2018-06-13 Jeff Law <law@redhat.com>
13582 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
13583 integer 0 for argument to print_rtl_with_bb.
13584 (rl78_reorg): Likewise.
13586 2018-06-13 David Malcolm <dmalcolm@redhat.com>
13588 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
13589 from rtx to rtx_insn *.
13590 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
13592 (add_sched_insns_for_speculation): Likewise for local "target",
13593 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13594 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
13595 from rtx_insn ** to rtx_code_label **.
13596 (reorg_emit_nops): Likewise.
13597 (c6x_reorg): Likewise for local "call_labels".
13598 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
13600 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
13601 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
13602 the loops over LABEL_REFS.
13603 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
13605 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13606 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
13607 (split_branches): Strengthen local "olabel" from rtx to
13608 rtx_insn *, adding a safe_as_a cast.
13609 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
13611 (add_insn_after): Likewise for first two params.
13612 (add_insn_before): Likewise.
13613 (remove_insn): Likewise for param.
13614 (emit_pattern_before_noloc): Likewise for second and third params.
13615 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
13616 (emit_call_insn_before_noloc): Likewise.
13617 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
13619 (emit_barrier_before): Likewise.
13620 (emit_label_before): Strengthen "label" param from "rtx" to
13621 "rtx_code_label *". Strengthen "before" param from "rtx" to
13623 (emit_insn_after_1): Strengthen "after" param from "rtx" to
13625 (emit_pattern_after_noloc): Likewise.
13626 (emit_insn_after_noloc): Likewise.
13627 (emit_jump_insn_after_noloc): Likewise.
13628 (emit_call_insn_after_noloc): Likewise.
13629 (emit_debug_insn_after_noloc): Likewise.
13630 (emit_barrier_after): Likewise.
13631 (emit_label_after): Likewise for both params.
13632 (emit_pattern_after_setloc): Likewise for "after" param. Convert
13633 "loc" param from "int" to "location_t".
13634 (emit_insn_after_setloc): Likewise.
13635 (emit_jump_insn_after_setloc): Likewise.
13636 (emit_call_insn_after_setloc): Likewise.
13637 (emit_debug_insn_after_setloc): Likewise.
13638 (emit_pattern_before_setloc): Likewise for "before" param. Convert
13639 "loc" param from "int" to "location_t".
13640 (emit_pattern_before): Convert NULL_RTX to NULL.
13641 (emit_insn_before_setloc): Convert "loc" param from "int" to
13643 (emit_jump_insn_before_setloc): Likewise.
13644 (emit_call_insn_before_setloc): Likewise.
13645 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
13646 rtx_insn *. Convert "loc" param from "int" to "location_t".
13647 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
13648 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
13649 Convert 3rd param from "int" to "location_t".
13650 (emit_barrier_before, emit_barrier_after, next_real_insn):
13651 Strengthen param from rtx to rtx_insn *.
13652 (emit_label_before): Strengthen 1st param from "rtx" to
13653 "rtx_code_label *". Strengthen 2nd param from "rtx" to
13655 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
13656 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
13657 Strengthen 2nd param from "rtx" to "rtx_insn *".
13658 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
13659 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
13660 Likewise. Convert 3rd param from "int" to "location_t".
13661 (emit_label_after): Strengthen 1st param from "rtx" to
13662 "rtx_code_label *".
13663 (next_real_insn, remove_insn): Strengthen param from "rtx" to
13665 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
13666 from "rtx" to "rtx_insn *".
13668 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
13670 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
13671 bodies streamed in with -Q.
13672 * dumpfile.c (dump_files): Add lto-stream-out dump file.
13673 * dumpfile.h (tree_dump_index): Add lto_stream_out.
13674 * gimple-streamer-out.c: Include gimple-pretty-print.h
13675 (output_bb): Dump stmts streamed.
13676 * lto-section-out.c: Include print-tree.h
13677 (lto_begin_section): Dump sections created.
13678 (lto_output_decl_index): Dump decl encoded.
13679 * lto-streamer-out.c: Include print-tree.h
13680 (create_output_block): Dump output block created.
13681 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
13682 (output_function): Dump function output.
13683 (output_constructor): Dump constructor streamed.
13684 (write_global_stream): Output indexes encoded.
13685 (produce_asm_for_decls): Dump streams encoded.
13686 * lto-streamer.c (streamer_dump_file): New global var.
13687 * lto-streamer.h (streamer_dump_file): Declare.
13688 * passes.c (ipa_write_summaries): Initialize streamer dump.
13689 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
13692 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
13695 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
13696 offsets for register save directives. Emit a second batch of save
13697 directives, if need be, when the function accesses prior frames.
13699 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13701 * config/arc/fpu.md (fmasf4): Force operand to register.
13702 (fnmasf4): Likewise.
13704 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13706 * config/arc/arc-protos.h (arc_pad_return): Remove.
13707 * config/arc/arc.c (machine_function): Remove force_short_suffix
13709 (arc_print_operand): Adjust printing of '&'.
13710 (arc_verify_short): Remove conditional printing of short suffix.
13711 (arc_final_prescan_insn): Remove reference to size_reason.
13712 (pad_return): New function.
13713 (arc_reorg): Call pad_return.
13714 (arc_pad_return): Remove.
13715 (arc_init_machine_status): Remove reference to force_short_suffix.
13716 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
13717 (attr length): When attribute iscompact is true force to 2
13718 regardless; in the case of maybe check if we want to force the
13719 instruction to have 4 bytes length.
13720 (nopv): Change it to generate 4 byte long nop as well.
13721 (blockage): New pattern.
13722 (simple_return): Remove call to arc_pad_return.
13723 (p_return_i): Likewise.
13725 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13727 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
13729 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13731 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
13734 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13736 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
13737 for ARC700 and ARCv2.
13739 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
13742 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
13743 operands[2] instead of operands[1].
13746 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13748 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
13749 case, check whether the outer register overlaps an unallocatable
13750 register, not just whether it fits the required class.
13752 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13754 * poly-int.h (can_div_trunc_p): Add new overload in which all values
13756 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
13757 (memrefs_conflict_p): Likewise.
13758 (init_alias_analysis): Likewise.
13759 * cfgexpand.c (expand_debug_expr): Likewise.
13760 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
13761 * cse.c (fold_rtx): Likewise.
13762 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
13763 * expr.c (emit_block_move_hints): Likewise.
13764 (clear_storage_hints, push_block, emit_push_insn): Likewise.
13765 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
13766 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
13767 (emit_group_store): Likewise.
13768 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
13769 to read the PRE/POST_MODIFY increment.
13770 * calls.c (store_one_arg): Use strip_offset.
13771 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
13773 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
13775 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
13776 (simplify_binary_operation_1): Extend CONST_INT handling to
13778 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
13779 than a HOST_WIDE_INT.
13780 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
13782 (adjust_mems, add_stores): Update accodingly.
13783 (vt_canonicalize_addr): Track polynomial offsets.
13784 (emit_note_insn_var_location): Likewise.
13785 (vt_add_function_parameter): Likewise.
13786 (vt_initialize): Likewise.
13788 2018-06-12 Jeff Law <law@redhat.com>
13790 * config.gcc (alpha*-*-freebsd*): Remove.
13791 * config/alpha/freebsd.h: Remove.
13793 2018-06-12 David Malcolm <dmalcolm@redhat.com>
13796 * spellcheck-tree.c (levenshtein_distance): Rename to...
13797 (get_edit_distance): ...this, and update for underlying renaming.
13798 * spellcheck-tree.h (levenshtein_distance): Rename to...
13799 (get_edit_distance): ...this.
13800 * spellcheck.c (levenshtein_distance): Rename to...
13801 (get_edit_distance): ...this. Convert from Levenshtein distance
13802 to Damerau-Levenshtein distance by supporting transpositions of
13803 adjacent characters. Rename "v1" to "v_next" and "v0" to
13805 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
13806 (selftest::test_edit_distance_unit_test_oneway): ...this, and
13807 update for underlying renaming.
13808 (selftest::levenshtein_distance_unit_test): Rename to...
13809 (selftest::test_get_edit_distance_unit): ...this, and update for
13810 underlying renaming.
13811 (selftest::test_find_closest_string): Add example from PR 69968
13812 where transposition helps
13813 (selftest::test_metric_conditions): Update for renaming.
13814 (selftest::test_metric_conditions): Likewise.
13815 (selftest::spellcheck_c_tests): Likewise.
13816 * spellcheck.h (levenshtein_distance): Rename both overloads to...
13817 (get_edit_distance): ...this.
13818 (best_match::consider): Update for renaming.
13820 2018-06-12 Martin Sebor <msebor@redhat.com>
13822 PR tree-optimization/85259
13823 * builtins.c (compute_objsize): Handle constant offsets.
13824 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
13825 true iff a warning has been issued.
13826 * gimple.h (gimple_nonartificial_location): New function.
13827 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
13828 gimple_nonartificial_location and handle -Wno-system-headers.
13829 (handle_builtin_stxncpy): Same.
13831 2018-06-12 Martin Sebor <msebor@redhat.com>
13834 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
13836 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13838 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13839 BUILTIN_VEC_XST entries for pointer to double and long long.
13841 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
13844 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13846 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13849 2018-06-12 Martin Liska <mliska@suse.cz>
13851 * doc/options.texi: Document IntegerRange.
13853 2018-06-12 Martin Liska <mliska@suse.cz>
13855 * config/i386/i386.opt: Make MPX-related options as Deprecated.
13856 * opt-functions.awk: Handle Deprecated flag.
13857 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
13859 (read_cmdline_option): Report warning for a deprecated option.
13860 * opts.h (struct cl_option): Add new field cl_deprecated.
13861 (CL_ERR_DEPRECATED): New.
13863 2018-06-12 Martin Liska <mliska@suse.cz>
13865 * doc/options.texi: Document Deprecated option flag.
13867 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13869 * config/arc/arc-arch.h (arc_extras): New enum.
13870 (arc_cpu_t):Add field extra.
13871 (arc_cpu_types): Consider the extras.
13872 * config/arc/arc-cpus.def: Add extras info.
13873 * config/arc/arc-opts.h (processor_type): Consider extra field.
13874 * config/arc/arc.c (arc_override_options): Handle extra field.
13876 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13878 * config/arc/arc-arch.h: Update ARC_OPTX macro.
13879 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
13881 * config/arc/arc.c (arc_init): Update pic warning.
13882 (irq_range): Update irq range parsing warnings.
13883 (arc_override_options): Update various warning messages.
13884 (arc_handle_aux_attribute): Likewise.
13886 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
13888 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
13890 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13892 * doc/sourcebuild.texi: Document usage of line number 0 in verify
13893 compiler messages directives.
13895 2018-06-12 Matthew Fortune <mfortune@gmail.com>
13897 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
13898 * config/mips/mips-tables.opt: Regenerate.
13899 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
13901 * doc/invoke.texi: Document -march=i6500.
13903 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
13905 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
13906 (i6400_gpmul): Add cpu_unit.
13907 (i6400_gpdiv): Likewise.
13908 (i6400_msa_add_d): Update reservations.
13909 (i6400_msa_int_add) Likewise.
13910 (i6400_msa_short_logic3) Likewise.
13911 (i6400_msa_short_logic2) Likewise.
13912 (i6400_msa_short_logic) Likewise.
13913 (i6400_msa_move) Likewise.
13914 (i6400_msa_cmp) Likewise.
13915 (i6400_msa_short_float2) Likewise.
13916 (i6400_msa_div_d) Likewise.
13917 (i6400_msa_long_logic1) Likewise.
13918 (i6400_msa_long_logic2) Likewise.
13919 (i6400_msa_mult) Likewise.
13920 (i6400_msa_long_float2) Likewise.
13921 (i6400_msa_long_float4) Likewise.
13922 (i6400_msa_long_float5) Likewise.
13923 (i6400_msa_long_float8) Likewise.
13924 (i6400_fpu_fadd): Include frint type.
13925 (i6400_fpu_store): New define_insn_reservation.
13926 (i6400_fpu_load): Likewise.
13927 (i6400_fpu_move): Likewise.
13928 (i6400_fpu_fcmp): Likewise.
13929 (i6400_fpu_fmadd): Likewise.
13930 (i6400_int_mult): Include imul3nc type and update reservation.
13931 (i6400_int_div): Include idiv3 type and update reservation.
13932 (i6400_int_load): Update to check type not move_type.
13933 (i6400_int_store): Likewise.
13934 (i6400_int_prefetch): Set zero latency.
13936 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
13938 * gcc.c: Document new %@{...} sequence.
13939 (LINK_COMMAND_SPEC): Use it for the -L switches.
13940 (cpp_unique_options): Use it for the -I switches.
13941 (at_file_argbuf): New global variable.
13942 (in_at_file): Likewise.
13943 (alloc_args): Create at_file_argbuf.
13944 (clear_args): Truncate at_file_argbuf.
13945 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
13946 (open_at_file): New function.
13947 (close_at_file): Likewise.
13948 (create_at_file): Delete.
13949 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
13952 (validate_switches_from_spec): Deal with %@{...} sequence.
13953 (validate_switches): Likewise.
13954 (driver::finalize): Call clear_args.
13956 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
13958 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
13960 2018-06-11 Martin Sebor <msebor@redhat.com>
13962 * doc/invoke.texi (-Wall): List -Wc++17-compat.
13963 (Wno-class-memaccess): Add @opindex.
13964 (Wno-templates, Wno-multiple-inheritance): Same.
13965 (Wno-virtual-inheritance, Wno-namespaces): Same.
13966 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
13967 (Wno-format-overflow, Wno-format-truncation): Same.
13968 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
13969 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
13970 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
13971 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
13972 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
13973 (Wno-misspelled-isr): Same.
13975 2018-06-11 Martin Sebor <msebor@redhat.com>
13977 * PR tree-optimization/86083
13978 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
13980 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
13982 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
13984 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
13987 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
13988 on the correct operand.
13989 (*movdi_internal64): Ditto.
13991 2018-06-11 Martin Liska <mliska@suse.cz>
13993 PR tree-optimization/86089
13994 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
13996 2018-06-11 Julia Koval <julia.koval@intel.com>
13998 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
13999 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
14000 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
14002 2018-06-11 Olivier Hainque <hainque@adacore.com>
14004 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
14005 for Ada with strict dwarf2.
14007 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
14010 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
14013 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
14015 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
14017 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
14019 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
14022 2018-06-08 Martin Liska <mliska@suse.cz>
14024 * tree-cfg.h (debug_function): Fix argument type to match
14027 2018-06-08 Martin Liska <mliska@suse.cz>
14029 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
14030 Remove usage of MPX-related (and removed) fields.
14031 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
14033 2018-06-08 David Malcolm <dmalcolm@redhat.com>
14035 * cfg.c (debug): Use TDF_NONE rather than 0.
14036 * cfghooks.c (debug): Likewise.
14037 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
14038 (struct dump_option_value_info): Convert to...
14039 (struct kv_pair): ...this template type.
14040 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
14042 (optinfo_verbosity_options): Likewise.
14043 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
14045 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
14046 than int for "optgroup_flags" param.
14047 (dump_generic_expr_loc): Use dump_flags_t rather than int for
14049 (dump_dec): Likewise.
14050 (dump_finish): Use TDF_NONE rather than 0.
14051 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
14052 rather than int for "optgroup_flags" param. Use TDF_NONE rather
14053 than 0. Update for change to option_ptr.
14054 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
14055 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
14056 0. Update for changes to optinfo_verbosity_options and
14058 (opt_info_switch_p): Convert optgroup_flags from int to
14060 (dump_basic_block): Use dump_flags_t rather than int
14061 for "dump_kind" param.
14062 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
14063 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
14064 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
14065 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
14066 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
14067 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
14068 TDF_NONE): Convert from macros to...
14069 (enum dump_flag): ...this new enum.
14070 (dump_flags_t): Update to use enum.
14071 (operator|, operator&, operator~, operator|=, operator&=):
14072 Implement for dump_flags_t.
14073 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
14074 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
14075 Convert from macros to...
14076 (enum optgroup_flag): ...this new enum.
14077 (optgroup_flags_t): New typedef.
14078 (operator|, operator|=): Implement for optgroup_flags_t.
14079 (struct dump_file_info): Convert field "alt_flags" to
14080 dump_flags_t. Convert field "optgroup_flags" to
14082 (dump_basic_block): Use dump_flags_t rather than int for param.
14083 (dump_generic_expr_loc): Likewise.
14084 (dump_dec): Likewise.
14085 (dump_register): Convert param "optgroup_flags" to
14087 (opt_info_enable_passes): Likewise.
14088 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
14090 * gimple-pretty-print.c (debug): Likewise.
14091 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
14092 (merged_store_group::apply_stores): Likewise.
14093 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
14094 * gimple.c (verify_gimple_pp): Likewise.
14095 * graphite-poly.c (print_pbb_body): Likewise.
14096 * passes.c (pass_manager::register_one_dump_file): Convert
14097 local "optgroup_flags" to optgroup_flags_t.
14098 * print-tree.c (print_node): Use TDF_NONE rather than 0.
14100 (debug_body): Likewise.
14101 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
14102 to optgroup_flags_t.
14103 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
14105 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
14106 (convert_mult_to_fma): Likewise.
14107 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
14108 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
14109 * tree-vect-data-refs.c (dump_lower_bound): Convert param
14110 "dump_kind" to dump_flags_t.
14112 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
14114 * config/rs6000/rs6000.c (min, max): Delete.
14116 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
14118 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
14119 -mabi=spe and -mabi=no-spe.
14121 2018-06-08 Martin Liska <mliska@suse.cz>
14123 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
14124 where we expect an existing summary.
14126 2018-06-08 Martin Liska <mliska@suse.cz>
14128 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
14129 * ipa-inline.h (estimate_edge_growth): Likewise.
14131 2018-06-08 Martin Liska <mliska@suse.cz>
14133 * cgraph.c (function_version_hasher::hash): Use
14134 cgraph_node::get_uid ().
14135 (function_version_hasher::equal):
14136 * cgraph.h (cgraph_node::get_uid): New method.
14137 * ipa-inline.c (update_caller_keys): Use
14138 cgraph_node::get_uid ().
14139 (update_callee_keys): Likewise.
14140 * ipa-utils.c (searchc): Likewise.
14141 (ipa_reduced_postorder): Likewise.
14142 * lto-cgraph.c (input_node): Likewise.
14143 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
14144 * symbol-summary.h (symtab_insertion): Likewise.
14145 (symtab_removal): Likewise.
14146 (symtab_duplication): Likewise.
14147 * tree-pretty-print.c (dump_function_header): Likewise.
14148 * tree-sra.c (convert_callers_for_node): Likewise.
14150 2018-06-08 Martin Liska <mliska@suse.cz>
14152 * cgraph.c (symbol_table::create_edge): Always assign a new
14154 (symbol_table::free_edge): Do not recycle numbers.
14155 * cgraph.h (cgraph_edge::get): New method.
14156 * symbol-summary.h (symtab_removal): Use it.
14157 (symtab_duplication): Likewise.
14158 (call_summary::hashable_uid): Remove.
14160 2018-06-08 Martin Liska <mliska@suse.cz>
14162 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
14163 (initialize_growth_caches): Remove.
14164 (free_growth_caches): Likewise.
14165 (do_estimate_edge_time): Use edge_growth_cache.
14166 (do_estimate_edge_size): Likewise.
14167 (do_estimate_edge_hints): Likewise.
14168 * ipa-inline.c (reset_edge_caches): Likewise.
14169 (recursive_inlining): Likewise.
14170 (inline_small_functions): Likewise.
14171 * ipa-inline.h (initialize_growth_caches): Remove.
14172 (estimate_edge_size): Likewise.
14173 (estimate_edge_time): Likewise.
14174 (estimate_edge_hints): Likewise.
14175 (reset_edge_growth_cache): Likewise.
14176 * symbol-summary.h (call_summary::remove): New method.
14178 2018-06-08 Martin Liska <mliska@suse.cz>
14180 * ipa-cp.c (class edge_clone_summary): New summary.
14181 (grow_edge_clone_vectors): Remove.
14182 (ipcp_edge_duplication_hook): Remove.
14183 (class edge_clone_summary_t): New call_summary class.
14184 (ipcp_edge_removal_hook): Remove.
14185 (edge_clone_summary_t::duplicate): New function.
14186 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
14187 (create_specialized_node): Likewise.
14188 (ipcp_driver): Initialize edge_clone_summaries and do not
14191 2018-06-08 Martin Liska <mliska@suse.cz>
14193 * symbol-summary.h (get): New function.
14194 (call_summary::m_initialize_when_cloning): New class member.
14196 2018-06-08 Martin Liska <mliska@suse.cz>
14198 * cgraph.c (cgraph_node::remove): Do not recycle uid.
14199 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
14200 (symbol_table::allocate_cgraph_symbol): Do not set uid.
14201 * passes.c (uid_hash_t): Record removed_nodes by their uids.
14202 (remove_cgraph_node_from_order): Use the removed_nodes set.
14203 (do_per_function_toporder): Likwise.
14204 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
14205 instead of summary_uid.
14206 (symtab_removal): Likewise.
14207 (symtab_duplication): Likewise.
14209 2018-06-08 Martin Liska <mliska@suse.cz>
14211 * ipa-cp.c (ipcp_store_bits_results): Use
14212 ipcp_transformation_sum.
14213 (ipcp_store_vr_results): Likewise.
14214 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
14216 (ipcp_transformation_initialize): ... this.
14217 (ipa_set_node_agg_value_chain):
14218 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
14219 (write_ipcp_transformation_info): Likewise.
14220 (read_ipcp_transformation_info): Likewise.
14221 (ipcp_update_bits): Likewise.
14222 (ipcp_update_vr): Likewise.
14223 (ipcp_transform_function): Likewise.
14224 * ipa-prop.h: Rename ipcp_transformation_summary to
14225 ipcp_transformation.
14226 (class ipcp_transformation_t): New function summary.
14227 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
14228 (ipa_get_agg_replacements_for_node): Likewise.
14230 2018-06-08 Martin Liska <mliska@suse.cz>
14232 * ipa-pure-const.c (struct funct_state_d): Do it class instead
14234 (class funct_state_summary_t): New function_summary class.
14235 (has_function_state): Remove.
14236 (get_function_state): Likewise.
14237 (set_function_state): Likewise.
14238 (add_new_function): Likewise.
14239 (funct_state_summary_t::insert): New function.
14240 (duplicate_node_data): Remove.
14241 (remove_node_data): Remove.
14242 (funct_state_summary_t::duplicate): New function.
14243 (register_hooks): Create new funct_state_summaries.
14244 (pure_const_generate_summary): Use it.
14245 (pure_const_write_summary): Likewise.
14246 (pure_const_read_summary): Likewise.
14247 (propagate_pure_const): Likewise.
14248 (propagate_nothrow): Likewise.
14249 (dump_malloc_lattice): Likewise.
14250 (propagate_malloc): Likewise.
14251 (execute): Do not register hooks, just remove summary
14253 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
14256 2018-06-08 Martin Liska <mliska@suse.cz>
14258 * ipa-reference.c (remove_node_data): Remove.
14259 (duplicate_node_data): Likewise.
14260 (class ipa_ref_var_info_summary_t): New class.
14261 (class ipa_ref_opt_summary_t): Likewise.
14262 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
14263 (get_reference_optimization_summary): Use
14264 ipa_ref_opt_sum_summaries.
14265 (set_reference_vars_info): Remove.
14266 (set_reference_optimization_summary): Likewise.
14267 (ipa_init): Create summaries.
14268 (init_function_info): Use function summary.
14269 (ipa_ref_opt_summary_t::duplicate): New function.
14270 (ipa_ref_opt_summary_t::remove): New function.
14271 (get_read_write_all_from_node): Fix GNU coding style.
14272 (propagate): Use function summary.
14273 (write_node_summary_p): Fix GNU coding style.
14274 (stream_out_bitmap): Likewise.
14275 (ipa_reference_read_optimization_summary): Use function summary.
14276 (ipa_reference_c_finalize): Do not release hooks.
14278 2018-06-08 Martin Liska <mliska@suse.cz>
14280 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
14281 (analyze_function_body): Extract multiple calls of get_create.
14282 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
14283 * ipa-inline.c (recursive_inlining): Use ::get method.
14284 * ipa-inline.h (estimate_edge_growth): Likewise.
14286 2018-06-08 Martin Liska <mliska@suse.cz>
14288 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
14290 (hsa_function_summary::hsa_function_summary): Use the new enum
14292 (hsa_gpu_implementation_p): Use hsa_summaries::get.
14293 * hsa-gen.c (hsa_get_host_function): Likewise.
14294 (get_brig_function_name): Likewise.
14295 * ipa-hsa.c (process_hsa_functions): Likewise.
14296 (ipa_hsa_write_summary): Likewise.
14297 * symbol-summary.h (symtab_duplication): Use ::get function/
14298 (get): New function.
14300 2018-06-08 Martin Liska <mliska@suse.cz>
14302 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
14304 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
14305 (hsa_register_kernel): Likewise.
14306 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
14307 * hsa-gen.c (hsa_get_host_function): Likewise.
14308 (get_brig_function_name): Likewise.
14309 (generate_hsa): Likewise.
14310 (pass_gen_hsail::execute): Likewise.
14311 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
14312 (devirtualization_time_bonus): Likewise.
14313 (ipcp_propagate_stage): Likewise.
14314 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
14315 (edge_set_predicate): Likewise.
14316 (evaluate_conditions_for_known_args): Likewise.
14317 (evaluate_properties_for_edge): Likewise.
14318 (ipa_fn_summary::reset): Likewise.
14319 (ipa_fn_summary_t::duplicate): Likewise.
14320 (dump_ipa_call_summary): Likewise.
14321 (ipa_dump_fn_summary): Likewise.
14322 (analyze_function_body): Likewise.
14323 (compute_fn_summary): Likewise.
14324 (estimate_edge_devirt_benefit): Likewise.
14325 (estimate_edge_size_and_time): Likewise.
14326 (estimate_calls_size_and_time): Likewise.
14327 (estimate_node_size_and_time): Likewise.
14328 (inline_update_callee_summaries): Likewise.
14329 (remap_edge_change_prob): Likewise.
14330 (remap_edge_summaries): Likewise.
14331 (ipa_merge_fn_summary_after_inlining): Likewise.
14332 (ipa_update_overall_fn_summary): Likewise.
14333 (read_ipa_call_summary): Likewise.
14334 (inline_read_section): Likewise.
14335 (write_ipa_call_summary): Likewise.
14336 (ipa_fn_summary_write): Likewise.
14337 (ipa_free_fn_summary): Likewise.
14338 * ipa-hsa.c (process_hsa_functions): Likewise.
14339 (ipa_hsa_write_summary): Likewise.
14340 (ipa_hsa_read_section): Likewise.
14341 * ipa-icf.c (sem_function::merge): Likewise.
14342 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
14343 (do_estimate_edge_time): Likewise.
14344 (estimate_size_after_inlining): Likewise.
14345 (estimate_growth): Likewise.
14346 (growth_likely_positive): Likewise.
14347 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
14348 (inline_call): Likewise.
14349 * ipa-inline.c (caller_growth_limits): Likewise.
14350 (can_inline_edge_p): Likewise.
14351 (can_inline_edge_by_limits_p): Likewise.
14352 (compute_uninlined_call_time): Likewise.
14353 (compute_inlined_call_time): Likewise.
14354 (want_inline_small_function_p): Likewise.
14355 (edge_badness): Likewise.
14356 (update_caller_keys): Likewise.
14357 (update_callee_keys): Likewise.
14358 (recursive_inlining): Likewise.
14359 (inline_small_functions): Likewise.
14360 (inline_to_all_callers_1): Likewise.
14361 (dump_overall_stats): Likewise.
14362 (early_inline_small_functions): Likewise.
14363 (early_inliner): Likewise.
14364 * ipa-inline.h (estimate_edge_growth): Likewise.
14365 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14366 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14367 * ipa-prop.h (IPA_NODE_REF): Likewise.
14368 (IPA_EDGE_REF): Likewise.
14369 * ipa-pure-const.c (malloc_candidate_p): Likewise.
14370 (propagate_malloc): Likewise.
14371 * ipa-split.c (execute_split_functions): Likewise.
14372 * symbol-summary.h: Rename get to get_create.
14374 (get_create): Likewise.
14375 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
14377 2018-06-08 Martin Liska <mliska@suse.cz>
14379 * symbol-summary.h (release): Move definition out of class
14381 (symtab_removal): Likewise.
14382 (symtab_duplication): Likewise.
14384 2018-06-08 Martin Liska <mliska@suse.cz>
14386 * symbol-summary.h (function_summary): Move constructor
14387 implementation out of class declaration.
14388 (release): Likewise.
14389 (symtab_insertion): Likewise.
14390 (symtab_removal): Likewise.
14391 (symtab_duplication): Likewise.
14394 2018-06-08 Martin Liska <mliska@suse.cz>
14396 * Makefile.in: Remove support for MPX (macros, related functions,
14397 fields in cgraph_node, ...).
14398 * builtin-types.def (BT_BND): Likewise.
14399 (BT_FN_BND_CONST_PTR): Likewise.
14400 (BT_FN_CONST_PTR_BND): Likewise.
14401 (BT_FN_VOID_PTR_BND): Likewise.
14402 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
14403 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
14404 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
14405 (expand_builtin_mempcpy_with_bounds): Likewise.
14406 (expand_builtin_memset_with_bounds): Likewise.
14407 (expand_builtin_memset_args): Likewise.
14408 (std_expand_builtin_va_start): Likewise.
14409 (expand_builtin): Likewise.
14410 (expand_builtin_with_bounds): Likewise.
14411 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
14412 (DEF_LIB_BUILTIN_CHKP): Likewise.
14413 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
14414 (DEF_CHKP_BUILTIN): Likewise.
14415 (BUILT_IN_MEMCPY): Likewise.
14416 (BUILT_IN_MEMMOVE): Likewise.
14417 (BUILT_IN_MEMPCPY): Likewise.
14418 (BUILT_IN_MEMSET): Likewise.
14419 (BUILT_IN_STPCPY): Likewise.
14420 (BUILT_IN_STRCAT): Likewise.
14421 (BUILT_IN_STRCHR): Likewise.
14422 (BUILT_IN_STRCPY): Likewise.
14423 (BUILT_IN_STRLEN): Likewise.
14424 (BUILT_IN_MEMCPY_CHK): Likewise.
14425 (BUILT_IN_MEMMOVE_CHK): Likewise.
14426 (BUILT_IN_MEMPCPY_CHK): Likewise.
14427 (BUILT_IN_MEMSET_CHK): Likewise.
14428 (BUILT_IN_STPCPY_CHK): Likewise.
14429 (BUILT_IN_STRCAT_CHK): Likewise.
14430 (BUILT_IN_STRCPY_CHK): Likewise.
14431 * calls.c (store_bounds): Likewise.
14432 (emit_call_1): Likewise.
14433 (special_function_p): Likewise.
14434 (maybe_warn_nonstring_arg): Likewise.
14435 (initialize_argument_information): Likewise.
14436 (finalize_must_preallocate): Likewise.
14437 (compute_argument_addresses): Likewise.
14438 (expand_call): Likewise.
14439 * cfgexpand.c (expand_call_stmt): Likewise.
14440 (expand_return): Likewise.
14441 (expand_gimple_stmt_1): Likewise.
14442 (pass_expand::execute): Likewise.
14443 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14444 (cgraph_node::remove): Likewise.
14445 (cgraph_node::dump): Likewise.
14446 (cgraph_node::verify_node): Likewise.
14447 * cgraph.h (chkp_function_instrumented_p): Likewise.
14448 (symtab_node::get_alias_target): Likewise.
14449 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
14450 (cgraph_local_p): Likewise.
14451 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
14452 (cgraph_edge::rebuild_references): Likewise.
14453 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
14454 (walk_polymorphic_call_targets): Likewise.
14455 (cgraph_node::expand_thunk): Likewise.
14456 (symbol_table::output_weakrefs): Likewise.
14457 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
14458 (ix86_handle_option): Likewise.
14459 * config/i386/constraints.md: Likewise.
14460 * config/i386/i386-builtin-types.def (BND): Likewise.
14464 * config/i386/i386-builtin.def (BDESC_END): Likewise.
14465 (BDESC_FIRST): Likewise.
14467 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
14468 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
14469 * config/i386/i386.c (enum reg_class): Likewise.
14470 (ix86_target_string): Likewise.
14471 (ix86_option_override_internal): Likewise.
14472 (ix86_conditional_register_usage): Likewise.
14473 (ix86_valid_target_attribute_inner_p): Likewise.
14474 (ix86_set_indirect_branch_type): Likewise.
14475 (ix86_set_current_function): Likewise.
14476 (ix86_function_arg_regno_p): Likewise.
14477 (init_cumulative_args): Likewise.
14478 (ix86_function_arg_advance): Likewise.
14479 (ix86_function_arg): Likewise.
14480 (ix86_pass_by_reference): Likewise.
14481 (ix86_function_value_regno_p): Likewise.
14482 (ix86_function_value_1): Likewise.
14483 (ix86_function_value_bounds): Likewise.
14484 (ix86_return_in_memory): Likewise.
14485 (ix86_setup_incoming_vararg_bounds): Likewise.
14486 (ix86_va_start): Likewise.
14487 (indirect_thunk_need_prefix): Likewise.
14488 (print_reg): Likewise.
14489 (ix86_print_operand): Likewise.
14490 (ix86_expand_call): Likewise.
14491 (ix86_output_function_return): Likewise.
14492 (reg_encoded_number): Likewise.
14493 (BDESC_VERIFYS): Likewise.
14494 (ix86_init_mpx_builtins): Likewise.
14495 (ix86_init_builtins): Likewise.
14496 (ix86_emit_cmove): Likewise.
14497 (ix86_emit_move_max): Likewise.
14498 (ix86_expand_builtin): Likewise.
14499 (ix86_builtin_mpx_function): Likewise.
14500 (ix86_get_arg_address_for_bt): Likewise.
14501 (ix86_load_bounds): Likewise.
14502 (ix86_store_bounds): Likewise.
14503 (ix86_load_returned_bounds): Likewise.
14504 (ix86_store_returned_bounds): Likewise.
14505 (ix86_class_likely_spilled_p): Likewise.
14506 (ix86_hard_regno_mode_ok): Likewise.
14507 (x86_order_regs_for_local_alloc): Likewise.
14508 (ix86_mitigate_rop): Likewise.
14509 (ix86_bnd_prefixed_insn_p): Likewise.
14510 (ix86_mpx_bound_mode): Likewise.
14511 (ix86_make_bounds_constant): Likewise.
14512 (ix86_initialize_bounds): Likewise.
14513 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
14514 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
14515 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
14516 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
14517 (TARGET_CHKP_BOUND_MODE): Likewise.
14518 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
14519 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
14520 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
14521 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
14522 * config/i386/i386.h (TARGET_MPX): Likewise.
14523 (TARGET_MPX_P): Likewise.
14524 (VALID_BND_REG_MODE): Likewise.
14525 (FIRST_BND_REG): Likewise.
14526 (LAST_BND_REG): Likewise.
14527 (enum reg_class): Likewise.
14528 (BND_REG_P): Likewise.
14529 (BND_REGNO_P): Likewise.
14530 (BNDmode): Likewise.
14531 (ADJUST_INSN_LENGTH): Likewise.
14532 * config/i386/i386.md: Likewise.
14533 * config/i386/i386.opt: Likewise.
14534 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
14535 (defined): Likewise.
14536 (LINK_MPX): Likewise.
14537 (MPX_SPEC): Likewise.
14538 (LIBMPX_SPEC): Likewise.
14539 (LIBMPXWRAPPERS_SPEC): Likewise.
14540 (CHKP_SPEC): Likewise.
14541 * config/i386/predicates.md: Likewise.
14542 * dbxout.c (dbxout_type): Likewise.
14543 * doc/extend.texi: Likewise.
14544 * doc/invoke.texi: Likewise.
14545 * doc/md.texi: Likewise.
14546 * doc/tm.texi: Likewise.
14547 * doc/tm.texi.in: Likewise.
14548 * dwarf2out.c (is_base_type): Likewise.
14549 (gen_formal_types_die): Likewise.
14550 (gen_subprogram_die): Likewise.
14551 (gen_type_die_with_usage): Likewise.
14552 (gen_decl_die): Likewise.
14553 (dwarf2out_late_global_decl): Likewise.
14554 * expr.c (expand_assignment): Likewise.
14555 (emit_storent_insn): Likewise.
14556 (store_expr_with_bounds): Likewise.
14557 (store_expr): Likewise.
14558 (expand_expr_real_1): Likewise.
14559 * expr.h (store_expr_with_bounds): Likewise.
14560 * function.c (use_register_for_decl): Likewise.
14561 (struct bounds_parm_data): Likewise.
14562 (assign_parms_augmented_arg_list): Likewise.
14563 (assign_parm_find_entry_rtl): Likewise.
14564 (assign_parm_is_stack_parm): Likewise.
14565 (assign_parm_load_bounds): Likewise.
14566 (assign_bounds): Likewise.
14567 (assign_parms): Likewise.
14568 (expand_function_start): Likewise.
14569 * gcc.c (CHKP_SPEC): Likewise.
14570 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14571 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
14572 (wrestrict_dom_walker::check_call): Likewise.
14573 * gimple.c (gimple_build_call_from_tree): Likewise.
14574 * gimple.h (enum gf_mask): Likewise.
14575 (gimple_call_with_bounds_p): Likewise.
14576 (gimple_call_set_with_bounds): Likewise.
14577 * gimplify.c (gimplify_init_constructor): Likewise.
14578 * ipa-cp.c (initialize_node_lattices): Likewise.
14579 (propagate_constants_across_call): Likewise.
14580 (find_more_scalar_values_for_callers_subset): Likewise.
14581 * ipa-hsa.c (process_hsa_functions): Likewise.
14582 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
14583 * ipa-icf.c (sem_function::merge): Likewise.
14584 * ipa-inline.c (early_inliner): Likewise.
14585 * ipa-pure-const.c (warn_function_noreturn): Likewise.
14586 (warn_function_cold): Likewise.
14587 (propagate_pure_const): Likewise.
14588 * ipa-ref.h (enum GTY): Likewise.
14589 * ipa-split.c (find_retbnd): Likewise.
14590 (consider_split): Likewise.
14591 (split_function): Likewise.
14592 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
14593 * ipa.c (walk_polymorphic_call_targets): Likewise.
14594 (symbol_table::remove_unreachable_nodes): Likewise.
14595 (process_references): Likewise.
14596 (cgraph_build_static_cdtor_1): Likewise.
14597 * lto-cgraph.c (lto_output_node): Likewise.
14598 (output_refs): Likewise.
14599 (compute_ltrans_boundary): Likewise.
14600 (input_overwrite_node): Likewise.
14601 (input_node): Likewise.
14602 (input_cgraph_1): Likewise.
14603 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
14604 * passes.c (pass_manager::execute_early_local_passes): Likewise.
14605 (class pass_chkp_instrumentation_passes): Likewise.
14606 (make_pass_chkp_instrumentation_passes): Likewise.
14607 * passes.def: Likewise.
14608 * rtl.h (struct GTY): Likewise.
14609 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
14610 * stor-layout.c (layout_type): Likewise.
14611 * symtab.c: Likewise.
14612 * target.def: Likewise.
14613 * targhooks.c (default_chkp_bound_type): Likewise.
14614 (default_chkp_bound_mode): Likewise.
14615 (default_builtin_chkp_function): Likewise.
14616 (default_chkp_function_value_bounds): Likewise.
14617 (default_chkp_make_bounds_constant): Likewise.
14618 (default_chkp_initialize_bounds): Likewise.
14619 * targhooks.h (default_chkp_bound_type): Likewise.
14620 (default_chkp_bound_mode): Likewise.
14621 (default_builtin_chkp_function): Likewise.
14622 (default_chkp_function_value_bounds): Likewise.
14623 (default_chkp_make_bounds_constant): Likewise.
14624 (default_chkp_initialize_bounds): Likewise.
14625 * toplev.c (compile_file): Likewise.
14626 (process_options): Likewise.
14627 * tree-core.h (DEF_BUILTIN): Likewise.
14628 (DEF_BUILTIN_CHKP): Likewise.
14629 * tree-inline.c (declare_return_variable): Likewise.
14630 (remap_gimple_stmt): Likewise.
14631 (copy_bb): Likewise.
14632 (initialize_inlined_parameters): Likewise.
14633 (expand_call_inline): Likewise.
14634 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
14635 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
14636 (make_pass_ipa_chkp_produce_thunks): Likewise.
14637 (make_pass_chkp): Likewise.
14638 (make_pass_chkp_opt): Likewise.
14639 (make_pass_chkp_instrumentation_passes): Likewise.
14640 * tree-pretty-print.c (dump_generic_node): Likewise.
14641 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
14642 * tree-ssa-dce.c (propagate_necessity): Likewise.
14643 (eliminate_unnecessary_stmts): Likewise.
14644 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
14645 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
14646 * tree-ssa-sccvn.h: Likewise.
14647 * tree-ssa-strlen.c (get_string_length): Likewise.
14648 (valid_builtin_call): Likewise.
14649 (adjust_last_stmt): Likewise.
14650 (handle_builtin_strchr): Likewise.
14651 (handle_builtin_strcpy): Likewise.
14652 (handle_builtin_stxncpy): Likewise.
14653 (handle_builtin_memcpy): Likewise.
14654 (handle_builtin_strcat): Likewise.
14655 (strlen_check_and_optimize_stmt): Likewise.
14656 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
14657 * tree-streamer-in.c: Likewise.
14658 * tree-streamer.c (record_common_node): Likewise.
14659 * tree.c (tree_code_size): Likewise.
14660 (wide_int_to_tree_1): Likewise.
14661 (type_contains_placeholder_1): Likewise.
14662 (build_common_tree_nodes): Likewise.
14663 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
14664 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
14665 (POINTER_BOUNDS_P): Likewise.
14666 (BOUNDED_TYPE_P): Likewise.
14667 (BOUNDED_P): Likewise.
14668 (CALL_WITH_BOUNDS_P): Likewise.
14669 (pointer_bounds_type_node): Likewise.
14670 * value-prof.c (gimple_ic): Likewise.
14671 * var-tracking.c (vt_add_function_parameters): Likewise.
14672 * varasm.c (make_decl_rtl): Likewise.
14673 (assemble_start_function): Likewise.
14674 (output_constant): Likewise.
14675 (maybe_assemble_visibility): Likewise.
14676 * varpool.c (ctor_for_folding): Likewise.
14677 * chkp-builtins.def: Remove.
14678 * ipa-chkp.c: Remove.
14679 * ipa-chkp.h: Remove.
14680 * rtl-chkp.c: Remove.
14681 * rtl-chkp.h: Remove.
14682 * tree-chkp-opt.c: Remove.
14683 * tree-chkp.c: Remove.
14684 * tree-chkp.h: Remove.
14686 2018-06-07 Carl Love <cel@us.ibm.com>
14688 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14689 vextract_fp_from_shortl): Add BE support.
14691 2018-06-07 Paul Koning <ni1d@arrl.net>
14693 * compare-elim.c (try_merge_compare): Don't merge compare if
14694 address contains a side effect.
14695 (try_eliminate_compare): Likewise.
14697 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
14699 * config.gcc: Support "tremont".
14700 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
14701 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14703 * config/i386/i386.c (m_TREMONT): Define.
14704 (processor_target_table): Add "tremont".
14705 (PTA_TREMONT): Define.
14706 (ix86_lea_outperforms): Add TARGET_TREMONT.
14707 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
14708 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
14709 and M_INTEL_GOLDMONT_PLUS.
14710 (fold_builtin_cpu): Add "tremont".
14711 (ix86_add_stmt_cost): Add TARGET_TREMONT.
14712 (ix86_option_override_internal): Add "tremont".
14713 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
14714 (processor_type): Add PROCESSOR_TREMONT.
14715 * config/i386/x86-tune.def: Add m_TREMONT.
14716 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
14718 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14720 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
14721 symbol defined for msp430i* devices to be lower case.
14723 2018-06-07 Richard Biener <rguenther@suse.de>
14725 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
14726 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
14727 Properly wrap signed arithmetic if overflow wraps.
14729 2018-06-07 Jakub Jelinek <jakub@redhat.com>
14731 PR tree-optimization/69615
14732 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
14733 of a cast from a same precision integral SSA_NAME in a bb dominated
14734 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
14735 cast to utype if rhs2 has already a compatible type.
14737 2018-06-07 Richard Biener <rguenther@suse.de>
14739 PR tree-optimization/85935
14740 * graphite-scop-detection.c (find_params_in_bb): Analyze
14741 condition operands with respect to the correct loop. Assert
14742 the analysis doesn't fail.
14744 2018-06-04 Carl Love <cel@us.ibm.com>
14746 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
14747 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
14748 as it is slightly cheaper.
14749 (first_match_or_eos_index_<mode>):
14750 Calculate index using natural element order.
14751 (first_match_index_<mode>):
14752 Calculate index using natural element order.
14753 (first_match_or_eos_index_<mode>):
14754 Calculate index using natural order.
14755 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
14756 for BE and LE modes.
14757 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
14758 P9V_BUILTIN_VCLZLSBB_V16QI.
14759 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
14762 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14764 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
14765 indentation and line wrap for many prototypes. Add missing
14766 @smallexample directives around block of prototypes for vec_xl and
14769 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
14771 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
14772 track if we pass or return IEEE 128-bit floating point.
14773 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
14774 C++ mangling that is compatible with GCC 8.1.
14775 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
14776 (init_cumulative_args): Note if we pass or return IEEE 128-bit
14777 floating point types.
14778 (rs6000_function_arg_advance_1): Likewise.
14779 (rs6000_mangle_type): Optionally generate mangled names that match
14780 what GCC 8.1 generated for IEEE 128-bit floating point types.
14781 (rs6000_globalize_decl_name): If we have an external function that
14782 passes or returns IEEE 128-bit types, generate a weak reference
14783 from the mangled name used in GCC 8.1 to the current mangled
14785 (rs6000_init_builtins): Make __ibm128 use the long double type if
14786 long double is IBM extended double. Make __float128 use the long
14787 double type if long double is IEEE 128-bit.
14790 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
14791 macro for __ibm128 built-in functions.
14792 (PACK_IF): Add __ibm128 pack/unpack functions.
14793 (UNPACK_IF): Likewise.
14794 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
14795 enable long double built-in functions if long double is IEEE
14796 128-bit floating point.
14797 (rs6000_invalid_builtin): Update long double built-in function
14799 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
14800 functions, adjust the built-in function to use the long double
14801 built-in function if __ibm128 and long double are the same type.
14802 * doc/extend.texi (PowerPC builtins): Update documention for
14803 __builtin_{,un}pack_longdouble. Add documentation for
14804 __builtin_{,un}pack_ibm128.
14806 2018-06-06 Jim Wilson <jimw@sifive.com>
14808 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
14809 (struct machine_function): New field interrupt_mode.
14810 (riscv_handle_type_attribute): New function. Add forward declaration.
14811 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
14812 (riscv_expand_epilogue): Check interrupt_mode field.
14813 (riscv_set_current_function): Check interrupt attribute args and
14814 set interrupt_mode field.
14815 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
14816 (riscv_sret, riscv_uret): New.
14817 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
14818 new arguments to interrupt attribute.
14820 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
14823 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
14824 Don't handle -mcpu=power8 if -mpower9-vector is also used.
14826 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14828 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
14829 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
14830 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
14831 several redundant entries.
14833 2018-06-06 David Malcolm <dmalcolm@redhat.com>
14835 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
14836 type from "rtx" to "rtx_insn *".
14837 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
14838 for local "call_insn", removing cast.
14839 (ix86_expand_call): Likewise, introducing a "call_insn" local.
14841 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
14843 PR tree-optimization/86066
14844 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
14845 for BIT_INSERT_EXPR stores.
14847 2018-06-06 Richard Biener <rguenther@suse.de>
14849 PR tree-optimization/86062
14850 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
14851 component refs ontop
14852 of to be offsetted base.
14854 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14856 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
14857 to be static and remove check on interrupt attribute name.
14859 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
14861 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14862 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
14864 2018-06-05 Steve Ellcey <sellcey@cavium.com>
14867 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
14869 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
14870 Remove second argument, change how error is called.
14871 (aarch64_layout_arg): Remove second argument from
14872 aarch64_err_no_fpadvsimd call.
14873 (aarch64_init_cumulative_args): Ditto.
14874 (aarch64_gimplify_va_arg_expr): Ditto.
14875 * config/aarch64/aarch64.md (mov<mode>): Ditto.
14877 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
14879 * config/i386/i386.md (simple_return_indirect_internal): New expander.
14880 (*simple_return_indirect_internal<mode>): Rename from
14881 simple_return_indirect_internal. Use W mode iterator.
14882 (rstorssp): New expander.
14883 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
14884 (clrssbsy): New expander.
14885 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
14887 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14889 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
14891 (cmse_check_pointed_object): Likewise.
14893 2018-06-05 Martin Liska <mliska@suse.cz>
14895 PR gcov-profile/47618
14896 * doc/invoke.texi: Document how -fprofile-dir format
14899 2018-06-05 Richard Biener <rguenther@suse.de>
14901 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
14902 removal pretend DOM info isn't available so we do not update
14903 it and only remove edges, not dominated blocks. Actually free
14904 DOM info in case we removed something. Remove unreachable blocks.
14905 (mfb_keep_latches): Work with either DOM info or marked backedges.
14906 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
14907 first. Mark backedges if DOM info isn't available.
14908 (Re-)compute DOM info after cleanup_control_flow_pre.
14910 2018-06-05 Richard Biener <rguenther@suse.de>
14912 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
14913 (locus_discrim_hasher::hash): Adjust.
14914 (locus_discrim_hasher::equal): Likewise.
14915 (next_discriminator_for_locus): Work on line directly.
14916 (same_line_p): Pass in expanded locus1 as well.
14917 (assign_discriminators): Avoid redundant location expansions.
14919 2018-06-05 Richard Biener <rguenther@suse.de>
14921 PR tree-optimization/86046
14922 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
14923 if required after clearing TREE_ADDRESSABLE.
14925 2018-06-05 Richard Biener <rguenther@suse.de>
14927 PR tree-optimization/86047
14928 * tree-ssa-loop.c (for_each_index): Glob handling of all
14929 decls and constants and really handle all of them.
14931 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14934 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
14935 qualifier_void_pointer and qualifier_const_void_pointer.
14936 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
14937 (arm_init_builtins): Handle the above.
14938 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
14939 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
14940 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
14943 2018-06-05 Martin Liska <mliska@suse.cz>
14945 * auto-profile.c (read_autofdo_file): Do not use
14946 gcov_ctr_summary struct.
14947 (afdo_callsite_hot_enough_for_early_inline): Likewise.
14948 * coverage.c (struct counts_entry): Likewise.
14949 (read_counts_file): Read just single summary entry.
14950 (get_coverage_counts): Use gcov_summary struct.
14951 * coverage.h (get_coverage_counts): Likewise.
14952 * gcov-dump.c (dump_working_sets): Likewise.
14953 (tag_summary): Dump just single summary.
14954 * gcov-io.c (gcov_write_summary): Write just histogram
14956 (gcov_read_summary): Read just single summary.
14957 (compute_working_sets): Use gcov_summary struct.
14958 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
14959 of GCOV_COUNTERS_SUMMABLE.
14960 (GCOV_COUNTERS_SUMMABLE): Remove.
14961 (GCOV_FIRST_VALUE_COUNTER): Replace with
14962 GCOV_COUNTER_V_INTERVAL.
14963 (struct gcov_ctr_summary): Remove.
14964 (struct gcov_summary): Directly use fields of former
14966 (compute_working_sets): Use gcov_summary struct.
14967 * gcov.c (read_count_file): Do not use ctrs fields.
14968 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
14970 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
14972 * profile.c: Likewise.
14973 * profile.h: Likewise.
14975 2018-06-05 Martin Liska <mliska@suse.cz>
14977 PR gcov-profile/84846
14978 * gcov.c (output_lines): Print working directory only
14979 in intermediate format.
14981 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
14983 * config/s390/s390-builtin-types.def: Add void function type.
14984 * config/s390/s390-builtins.def: Use the function type for the
14987 2018-06-04 Jim Wilson <jimw@sifive.com>
14989 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
14991 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
14992 and maybe_eh_return. Change regno to unsigned int. Use new args to
14993 handle EH_RETURN_DATA_REGNO registers properly.
14994 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
14995 (riscv_expand_epilogue): Update comment. Change argument name and
14996 type. Update code to use new name and type. Pass new args to
14997 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
14999 * config/riscv/riscv.md (NORMAL_RETURN): New.
15000 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
15001 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
15002 (eh_return): Call gen_eh_return_internal and emit barrier.
15003 (eh_return_internal): Call riscv_expand_epilogue.
15005 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
15007 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
15008 bit_insertion field and declare can_be_merged_into method.
15009 (merged_store_group::can_be_merged_into): New method.
15010 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
15011 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
15012 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
15014 2018-06-04 Richard Biener <rguenther@suse.de>
15016 PR tree-optimization/85955
15017 * builtins.c (fold_builtin_sincos): Convert pointers to
15018 destination to appropriate type before dereferencing.
15020 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
15022 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
15024 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
15026 * expr.c (expand_expr_real_1): Force the operand into memory if
15027 its TYPE_MODE is BLKmode and if there is no integer mode for
15028 the number of bits being extracted.
15030 2018-06-04 Jakub Jelinek <jakub@redhat.com>
15034 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
15035 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
15037 2018-06-04 Richard Biener <rguenther@suse.de>
15039 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
15040 (cleanup_tree_cfg_noloop): ... single caller. Do
15041 start_recording_case_labels later.
15043 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
15045 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
15046 to _IMMINTRIN_H_INCLUDED.
15047 * config/i386/pconfigintrin.h: Ditto.
15048 * config/i386/waitpkgintrin.h: Ditto.
15049 * config/i386/immintrin.h: Add includes for sgxintrin.h,
15050 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
15051 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
15052 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
15053 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
15054 waitpkgintrin.h and cldemoteintrin.h.
15056 2018-06-04 Richard Biener <rguenther@suse.de>
15058 PR tree-optimization/86038
15059 * tracer.c (find_best_successor): Check probability for
15060 being initialized, bail out if not.
15062 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
15065 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
15066 of bits to ignore when comparing architectures.
15068 2018-06-04 Jakub Jelinek <jakub@redhat.com>
15070 PR tree-optimization/69615
15071 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
15072 maximum or minimum of the type, try to merge it also as if
15073 range1 is + [-, x - 1] or + [x + 1, -].
15076 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
15078 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
15080 PR tree-optimization/86034
15081 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
15082 the unsigned bitfield type in a bit insertion sequence if it does not
15083 have a larger precision than the bitfield size.
15084 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
15086 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
15088 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
15090 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
15092 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
15093 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
15094 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
15095 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
15097 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
15099 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
15100 Disable -fdelete-null-pointer-checks for ELF toolchain.
15102 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
15103 Kito Cheng <kito.cheng@gmail.com>
15105 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
15106 (nds32le-*-*, nds32be-*-*): Integrate checking process.
15107 (nds32*-*-*): Add glibc and uclibc conditions.
15108 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
15109 (TARGET_EXCEPT_UNWIND_INFO): Define.
15110 * config/nds32/elf.h: New file.
15111 * config/nds32/linux.h: New file.
15112 * config/nds32/nds32-elf.opt: New file.
15113 * config/nds32/nds32-linux.opt: New file.
15114 * config/nds32/nds32-fp-as-gp.c
15115 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
15116 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
15118 (nds32_asm_file_end): Ditto.
15119 (nds32_print_operand): Ditto.
15120 (nds32_insert_attributes): Ditto.
15121 (nds32_init_libfuncs): New function.
15122 (TARGET_HAVE_TLS): Define.
15123 (TARGET_INIT_LIBFUNCS): Define.
15124 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
15126 (TARGET_ELF): Apply different mcmodel setting.
15127 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
15128 been migrated into elf.h and linux.h files.
15129 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
15130 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
15131 (mcmodel): The content has been migrated into nds32-elf.opt and
15132 nds32-linux.opt files.
15133 * config/nds32/t-elf: New file.
15134 * config/nds32/t-linux: New file.
15136 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
15137 Shiva Chen <shiva0217@gmail.com>
15139 * config/nds32/constants.md (unspec_volatile_element): Add
15140 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
15141 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
15143 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
15144 (make_pass_nds32_fp_as_gp): Declare.
15145 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
15147 (nds32_asm_function_end_prologue): Remove unused asm output.
15148 (nds32_asm_function_begin_epilogue): Remove unused asm output.
15149 (nds32_asm_file_start): Output necessary fp_as_gp information.
15150 (nds32_option_override): Adjust register usage.
15151 (nds32_expand_prologue): Consider fp_as_gp situation.
15152 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
15153 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
15156 (simple_return): Ditto.
15157 (omit_fp_begin): Output special directive for fp_as_gp.
15158 (omit_fp_end): Output special directive for fp_as_gp.
15159 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
15160 mforbid-fp-as-gp): New options.
15162 2018-06-01 Mark Wielaard <mark@klomp.org>
15164 * dwarf2out.c (dwarf2out_finish): Remove generation of
15165 DW_AT_loclists_base.
15167 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
15169 * gimple-ssa-store-merging.c: Include gimple-fold.h.
15170 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
15171 (struct merged_store_group): Add bit_insertion field.
15172 (dump_char_array): Use standard hexadecimal format.
15173 (merged_store_group::merged_store_group): Set bit_insertion to false.
15174 (merged_store_group::apply_stores): Use optimal buffer size. Deal
15175 with BIT_INSERT_EXPR stores. Move up code updating the mask and
15176 also print the mask in the dump file.
15177 (pass_store_merging::gate): Minor tweak.
15178 (imm_store_chain_info::coalesce_immediate): Fix wrong association
15179 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
15180 stores with INTEGER_CST stores.
15181 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
15182 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
15183 and use it throughout. Generate bit insertion sequences if need be.
15184 (pass_store_merging::process_store): Remove redundant condition.
15185 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
15187 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
15189 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
15190 the 128-bit floating point types. Fix function comment.
15192 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15194 * config/aarch64/aarch64-simd.md
15195 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
15197 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
15200 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
15202 PR tree-optimization/85989
15203 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
15205 (backprop::intersect_uses): Check it when deciding whether this
15206 is a backedge reference.
15207 (backprop::process_block): Add each phi to m_visited_phis
15208 after visiting it, then clear it at the end.
15210 2018-06-01 Richard Biener <rguenther@suse.de>
15212 * tree-vectorizer.h (vect_dr_stmt): New function.
15213 (vect_get_load_cost): Adjust.
15214 (vect_get_store_cost): Likewise.
15215 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
15216 Use vect_dr_stmt instead of DR_SMTT.
15217 (vect_record_base_alignments): Likewise.
15218 (vect_calculate_target_alignment): Likewise.
15219 (vect_compute_data_ref_alignment): Likewise and make static.
15220 (vect_update_misalignment_for_peel): Likewise.
15221 (vect_verify_datarefs_alignment): Likewise.
15222 (vector_alignment_reachable_p): Likewise.
15223 (vect_get_data_access_cost): Likewise. Pass down
15224 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
15225 (vect_get_peeling_costs_all_drs): Likewise.
15226 (vect_peeling_hash_get_lowest_cost): Likewise.
15227 (vect_enhance_data_refs_alignment): Likewise.
15228 (vect_find_same_alignment_drs): Likewise.
15229 (vect_analyze_data_refs_alignment): Likewise.
15230 (vect_analyze_group_access_1): Likewise.
15231 (vect_analyze_group_access): Likewise.
15232 (vect_analyze_data_ref_access): Likewise.
15233 (vect_analyze_data_ref_accesses): Likewise.
15234 (vect_vfa_segment_size): Likewise.
15235 (vect_small_gap_p): Likewise.
15236 (vectorizable_with_step_bound_p): Likewise.
15237 (vect_prune_runtime_alias_test_list): Likewise.
15238 (vect_analyze_data_refs): Likewise.
15239 (vect_supportable_dr_alignment): Likewise.
15240 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
15241 (vect_gen_prolog_loop_niters): Likewise.
15242 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
15243 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
15245 (vect_recog_mask_conversion_pattern): Likewise.
15246 (vect_try_gather_scatter_pattern): Likewise.
15247 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
15248 to vect_get_store_cost.
15249 (vect_get_store_cost): Get stmt_info instead of DR.
15250 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
15251 (vect_get_load_cost): Get stmt_info instead of DR.
15253 2018-06-01 Richard Biener <rguenther@suse.de>
15255 PR middle-end/86017
15256 * gimple-fold.c (var_decl_component_p): Also allow offsetted
15257 vars wrapped in MEM_REFs.
15259 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
15261 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15262 Fix subreg tests so that we only return a choice between
15263 GENERAL_REGS and FP_REGS if the original classes included both.
15265 2018-06-01 Richard Biener <rguenther@suse.de>
15268 * tree-ssa-structalias.c (get_function_part_constraint):
15269 Handle NULL fi->decl.
15270 (find_func_aliases_for_call): Properly handle indirect
15271 fi from direct call.
15272 (find_func_clobbers): Likewise.
15273 (ipa_pta_execute): Likewise.
15274 (create_variable_info_for): For functions that are ifunc_resolver
15275 resolve to a varinfo that contains the result of the resolver call.
15276 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
15279 2018-05-31 Michael Collison <michael.collison@arm.com>
15281 * config/aarch64/aarch64.md:
15282 (*fix_to_zero_extenddfdi2): New pattern.
15283 * gcc.target/aarch64/fix_extend1.c: New testcase.
15285 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
15287 PR middle-end/78809
15288 PR middle-end/83026
15289 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
15290 and BUILT_IN_STRNCMP_EQ.
15291 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
15292 BUILT_IN_STRNCMP_EQ.
15293 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
15294 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
15295 (gimple_fold_builtin): Likewise.
15296 * tree-ssa-strlen.c (compute_string_length): New function.
15297 (determine_min_obsize): New function.
15298 (handle_builtin_string_cmp): New function to handle calls to
15299 string compare functions.
15300 (strlen_optimize_stmt): Add handling to builtin string compare
15302 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15303 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
15304 * tree.c (build_common_builtin_nodes): Add new defines of
15305 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
15307 2018-05-31 Jakub Jelinek <jakub@redhat.com>
15310 * bb-reorder.c (pass_partition_blocks::gate): Return false for
15311 functions with naked attribute.
15313 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
15315 * config/i386/sse.md (avx_vec_concat<mode>):
15316 Substitute concat_tg_mode mode attribute with xtg_mode.
15317 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
15318 (concat_tg_mode): Remove mode attribute.
15320 2018-05-31 Martin Sebor <msebor@redhat.com>
15323 * calls.c (alloc_max_size): Correct a logic error/typo.
15324 Treat excessive arguments as infinite. Warn for invalid arguments.
15325 * doc/invoke.texi (-Walloc-size-larger-than): Update.
15327 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
15330 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
15331 and movx for Haswell.
15333 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
15334 Cesar Philippidis <cesar@codesourcery.com>
15336 PR middle-end/85879
15337 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
15338 when emitting error on private/firstprivate reductions.
15339 * omp-low.c (lower_omp_target): Avoid reference-type processing
15340 on pointers for firstprivate clause.
15342 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
15344 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
15347 * config/aarch64/aarch64-simd.md
15348 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
15349 (aarch64_ld1_x3_<mode>): Likewise
15350 (aarch64_st1x2<VALLDIF:mode>): Likewise
15351 (aarch64_st1_x2_<mode>): Likewise
15352 (aarch64_st1x3<VALLDIF:mode>): Likewise
15353 (aarch64_st1_x3_<mode>): Likewise
15354 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
15355 (vld1_s8_x3): Likewise.
15356 (vld1_u16_x3): Likewise.
15357 (vld1_s16_x3): Likewise.
15358 (vld1_u32_x3): Likewise.
15359 (vld1_s32_x3): Likewise.
15360 (vld1_u64_x3): Likewise.
15361 (vld1_s64_x3): Likewise.
15362 (vld1_f16_x3): Likewise.
15363 (vld1_f32_x3): Likewise.
15364 (vld1_f64_x3): Likewise.
15365 (vld1_p8_x3): Likewise.
15366 (vld1_p16_x3): Likewise.
15367 (vld1_p64_x3): Likewise.
15368 (vld1q_u8_x3): Likewise.
15369 (vld1q_s8_x3): Likewise.
15370 (vld1q_u16_x3): Likewise.
15371 (vld1q_s16_x3): Likewise.
15372 (vld1q_u32_x3): Likewise.
15373 (vld1q_s32_x3): Likewise.
15374 (vld1q_u64_x3): Likewise.
15375 (vld1q_s64_x3): Likewise.
15376 (vld1q_f16_x3): Likewise.
15377 (vld1q_f32_x3): Likewise.
15378 (vld1q_f64_x3): Likewise.
15379 (vld1q_p8_x3): Likewise.
15380 (vld1q_p16_x3): Likewise.
15381 (vld1q_p64_x3): Likewise.
15382 (vst1_s64_x2): Likewise.
15383 (vst1_u64_x2): Likewise.
15384 (vst1_f64_x2): Likewise.
15385 (vst1_s8_x2): Likewise.
15386 (vst1_p8_x2): Likewise.
15387 (vst1_s16_x2): Likewise.
15388 (vst1_p16_x2): Likewise.
15389 (vst1_s32_x2): Likewise.
15390 (vst1_u8_x2): Likewise.
15391 (vst1_u16_x2): Likewise.
15392 (vst1_u32_x2): Likewise.
15393 (vst1_f16_x2): Likewise.
15394 (vst1_f32_x2): Likewise.
15395 (vst1_p64_x2): Likewise.
15396 (vst1q_s8_x2): Likewise.
15397 (vst1q_p8_x2): Likewise.
15398 (vst1q_s16_x2): Likewise.
15399 (vst1q_p16_x2): Likewise.
15400 (vst1q_s32_x2): Likewise.
15401 (vst1q_s64_x2): Likewise.
15402 (vst1q_u8_x2): Likewise.
15403 (vst1q_u16_x2): Likewise.
15404 (vst1q_u32_x2): Likewise.
15405 (vst1q_u64_x2): Likewise.
15406 (vst1q_f16_x2): Likewise.
15407 (vst1q_f32_x2): Likewise.
15408 (vst1q_f64_x2): Likewise.
15409 (vst1q_p64_x2): Likewise.
15410 (vst1_s64_x3): Likewise.
15411 (vst1_u64_x3): Likewise.
15412 (vst1_f64_x3): Likewise.
15413 (vst1_s8_x3): Likewise.
15414 (vst1_p8_x3): Likewise.
15415 (vst1_s16_x3): Likewise.
15416 (vst1_p16_x3): Likewise.
15417 (vst1_s32_x3): Likewise.
15418 (vst1_u8_x3): Likewise.
15419 (vst1_u16_x3): Likewise.
15420 (vst1_u32_x3): Likewise.
15421 (vst1_f16_x3): Likewise.
15422 (vst1_f32_x3): Likewise.
15423 (vst1_p64_x3): Likewise.
15424 (vst1q_s8_x3): Likewise.
15425 (vst1q_p8_x3): Likewise.
15426 (vst1q_s16_x3): Likewise.
15427 (vst1q_p16_x3): Likewise.
15428 (vst1q_s32_x3): Likewise.
15429 (vst1q_s64_x3): Likewise.
15430 (vst1q_u8_x3): Likewise.
15431 (vst1q_u16_x3): Likewise.
15432 (vst1q_u32_x3): Likewise.
15433 (vst1q_u64_x3): Likewise.
15434 (vst1q_f16_x3): Likewise.
15435 (vst1q_f32_x3): Likewise.
15436 (vst1q_f64_x3): Likewise.
15437 (vst1q_p64_x3): Likewise.
15439 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15441 * config/msp430/msp430.c (msp430_output_labelref): Prepend
15442 user_label_prefix to name.
15444 * tree-core.h: Update comment about the format of NAME string
15445 passed to handler in attribute_spec.
15447 * config/msp430/msp430.md: Remove erroneous subreg expression from
15448 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
15449 zero_extend{q,h}isi2.
15451 2018-05-30 Borislav Petkov <bp@suse.de>
15453 * doc/extend.texi: Document some architecture specific
15454 constraints and sort entries.
15456 2018-05-30 Martin Sebor <msebor@redhat.com>
15458 PR middle-end/85369
15459 * builtins.c (expand_builtin_stpcpy_1): New function.
15460 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
15461 only if the former succeeds.
15463 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
15465 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
15468 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15470 * doc/invoke.texi (-flinker-output): Document
15472 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15474 * passes.c (ipa_write_summaries): Only modify statements if body
15476 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
15477 incrementally linking.
15478 (ipa_passes): Likewise.
15479 * lto-cgraph.c (lto_output_node): When incrementally linking do not
15480 pass down resolution info.
15481 * common.opt (flag_incremental_link): Update info.
15482 * gcc.c (plugin specs): Turn flinker-output=* to
15483 -plugin-opt=-linker-output-known
15484 * toplev.c (compile_file): Also cut compilation when doing incremental
15486 * flag-types. (enum lto_partition_model): Add
15487 LTO_LINKER_OUTPUT_NOLTOREL.
15488 (invoke.texi): Add -flinker-output docs.
15489 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
15490 link same way as WPA; do not stream in dead initializers.
15492 * dwarf2out.c (dwarf2out_die_ref_for_decl,
15493 darf2out_register_external_decl): Support incremental link.
15495 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15497 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
15499 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15501 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
15502 it down to simple_object_copy_lto_debug_sections.
15503 (run_gcc): Determine incremental LTO link time and configure
15504 lto1 into non-wpa mode, disable renaming of debug sections.
15506 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
15508 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
15509 descriptions of various incorrectly documented functions.
15511 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
15514 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15515 address check not strict.
15517 2018-05-30 Richard Biener <rguenther@suse.de>
15519 PR tree-optimization/85964
15520 * tracer.c (better_p): Drop initialized count check, we only
15521 call the function with initialized counts now.
15522 (find_best_successor): Do find a best edge if one
15523 has uninitialized count.
15524 (find_best_predecessor): Likewise. Do BB frequency check only
15525 if count is initialized.
15527 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
15529 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
15530 (aarch64_ldrstr_offset_compare): New.
15531 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
15532 load/store orderings.
15533 (aarch64_gen_adjusted_ldpstp): Likewise.
15535 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15537 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15538 Check for subset of GENERAL_REGS and FP_REGS.
15539 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
15542 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
15544 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
15545 and wi::to_poly_offset. Add the current offset and then check
15546 whether the sum fits, rather than using an unchecked addition of
15547 a checked term. Check for a shwi rather than a uhwi.
15548 * expr.c (get_bit_range): Use tree_to_poly_uint64.
15549 (store_constructor): Use poly_int_tree_p.
15550 (expand_expr_real_1): Likewise.
15551 * function.c (assign_temp): Likewise.
15552 * fold-const.c (const_binop): Use poly_int_tree_p and
15553 wi::to_poly_offset.
15554 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
15556 * ipa-icf-gimple.c (func_checker::compare_operand): Use
15557 to_poly_offset for MEM offsets.
15558 * ipa-icf.c (sem_variable::equals): Likewise.
15559 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
15560 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
15561 wi::to_poly_offset for BIT_FIELD_REF offsets.
15562 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
15563 wi::to_poly_offset.
15564 * var-tracking.c (emit_note_insn_var_location): Use
15565 tree_to_poly_uint64.
15567 2018-05-29 Jim Wilson <jimw@sifive.com>
15569 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
15571 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
15574 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
15575 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
15577 (sse4_1_round<mode>2): Use nonimmediate_operand
15578 for operand 1 predicate.
15580 2018-05-29 Martin Sebor <msebor@redhat.com>
15581 Richard Biener <rguenther@suse.de>
15584 * calls.c (get_size_range): Call determine_value_range instead
15585 of get_value_range..
15586 * tree-vrp.h (determine_value_range): Declared new function.
15587 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
15589 2018-05-29 Richard Biener <rguenther@suse.de>
15591 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
15592 sure to use non-pattern stmts for get_earlier_stmt arguments.
15593 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
15594 called on pattern stmts.
15595 (get_later_stmt): Likewise.
15597 2018-05-29 Martin Liska <mliska@suse.cz>
15599 PR gcov-profile/85759
15600 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
15603 2018-05-29 Jakub Jelinek <jakub@redhat.com>
15605 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
15607 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
15611 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15612 VEC_PACK_FLOAT_EXPR): New tree codes.
15613 * tree-pretty-print.c (op_code_prio): Handle
15614 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15615 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15616 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15617 * tree-inline.c (estimate_operator_cost): Likewise.
15618 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
15619 * fold-const.c (const_binop): Likewise.
15620 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
15621 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15622 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
15623 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
15624 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15625 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15626 * expr.c (expand_expr_real_2): Likewise.
15627 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
15628 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
15629 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
15631 * optabs.c (expand_widen_pattern_expr): For
15632 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
15633 sign from result type rather than operand's type.
15634 (expand_binop_directly): For vec_packu_float_optab and
15635 vec_packs_float_optab allow result type to be different from operand's
15637 * optabs-tree.c (optab_for_tree_code): Handle
15638 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15639 VEC_PACK_FLOAT_EXPR. Formatting fixes.
15640 * tree-vect-generic.c (expand_vector_operations_1): Handle
15641 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15642 VEC_PACK_FLOAT_EXPR.
15643 * tree-vect-stmts.c (supportable_widening_operation): Handle
15645 (supportable_narrowing_operation): Handle FLOAT_EXPR.
15646 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
15647 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
15648 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
15649 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
15651 (vec_pack<floatprefix>_float_<mode>): New expander.
15652 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
15654 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
15655 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
15656 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
15657 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
15658 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
15660 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
15661 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
15662 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15663 VEC_PACK_FLOAT_EXPR): Document.
15665 2018-05-29 Richard Biener <rguenther@suse.de>
15667 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
15669 (stmt_vec_info_vec): Make pointer.
15670 (init_stmt_vec_info_vec): Remove.
15671 (free_stmt_vec_info_vec): Likewise.
15672 (set_stmt_vec_info_vec): New function.
15673 (free_stmt_vec_infos): Likewise.
15674 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
15675 (set_vinfo_for_stmt): Likewise.
15676 (get_earlier_stmt): Likewise.
15677 (get_later_stmt): Likewise.
15678 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
15679 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
15680 (vec_info::~vec_info): Free stmt_vec_infos.
15681 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
15682 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
15683 (pass_slp_vectorize::execute): Likewise.
15684 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
15685 (free_stmt_vec_info_vec): Likewise.
15686 (set_stmt_vec_info_vec): New function.
15687 (free_stmt_vec_infos): Likewise.
15688 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
15689 the global stmt_vec_info_vec.
15690 * tree-parloops.c (gather_scalar_reductions): Use
15691 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
15694 2018-05-29 Richard Biener <rguenther@suse.de>
15696 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
15698 2018-05-29 Martin Liska <mliska@suse.cz>
15699 David Malcolm <dmalcolm@redhat.com>
15701 * vec.c (test_reverse): New.
15702 (vec_c_tests): Add new test.
15703 * vec.h (vl_ptr>::reverse): New function.
15705 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
15707 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
15709 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
15712 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15714 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
15716 2018-05-28 Richard Biener <rguenther@suse.de>
15718 PR tree-optimization/85933
15719 * tree-vect-data-refs.c (vect_record_base_alignments): Only
15720 look at stmts marked as vectorizable.
15722 2018-05-28 Richard Biener <rguenther@suse.de>
15724 PR tree-optimization/85934
15725 * tree-vect-generic.c (expand_vector_operations_1): Hoist
15726 vector boolean check before scalar optimization.
15728 2018-05-28 Jakub Jelinek <jakub@redhat.com>
15730 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
15733 2018-05-28 Mark Wielaard <mark@klomp.org>
15735 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
15736 if it is an expression containing a minus sign.
15738 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
15740 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
15742 2018-05-27 Paul Koning <ni1d@arrl.net>
15744 * config/pdp11/pdp11.md (truncsihi2): Remove.
15746 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
15747 Chung-Ju Wu <jasonwucj@gmail.com>
15749 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
15751 (unaligned_store_dw): Ditto.
15752 * config/nds32/nds32-memory-manipulation.c
15753 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
15754 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
15755 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
15756 (emit_setmem_word_loop): Rename to ...
15757 (emit_setmem_doubleword_loop): ... this.
15758 (nds32_gen_dup_4_byte_to_word_value): New function.
15759 (nds32_gen_dup_8_byte_to_double_word_value): New function.
15760 (nds32_expand_setmem_loop): Refine implementation.
15761 (nds32_expand_setmem_loop_v3m): Ditto.
15762 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
15765 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15767 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
15769 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15771 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
15772 (nds32_init_machine_status): Initialize machine->attr_naked_p and
15773 machine->attr_no_prologue_p.
15774 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
15775 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
15776 (nds32_expand_epilogue): Consider attr_naked_p.
15777 (nds32_expand_epilogue_v3pop): Likewise.
15778 (nds32_can_use_return_insn): Likewise.
15779 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
15780 attr_no_prologue_p fields.
15781 * config/nds32/nds32.opt (mret-in-naked-func): New option.
15783 2018-05-27 Jakub Jelinek <jakub@redhat.com>
15786 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
15788 * config/i386/sse.md
15789 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
15791 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
15793 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
15795 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
15797 (*<floatsuffix>floatv2div2sf2): Rename to ...
15798 (*float<floatunssuffix>v2div2sf2): ... this.
15799 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
15800 (float<floatunssuffix>v2div2sf2_mask): ... this.
15801 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
15802 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
15803 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
15805 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
15807 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15809 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15811 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15813 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15815 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
15816 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
15817 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
15818 gen_ufix_truncv8dfv8si2.
15819 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
15820 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
15821 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
15822 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
15823 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
15824 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
15825 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
15826 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
15828 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15832 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
15834 2018-05-25 Jim Wilson <jimw@sifive.com>
15836 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
15837 * config/riscv/riscv.c (struct machine_function): Add
15838 interrupt_handler_p and attribute_checked_p fields.
15839 (riscv_attribute_table): Add interrupt.
15840 (riscv_interrupt_type_p): New.
15841 (riscv_save_reg_p): Save extra regs for interrupt handler.
15842 (riscv_use_save_libcall): Return false for interrupt handler.
15843 (riscv_first_stack_step): Add forward declaration.
15844 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
15845 for interrupt handler with large frame. Use it for saved reg list.
15846 (riscv_expand_prologue): Move flag_stack_usage_info support to
15847 eliminate duplication.
15848 (riscv_expand_epilogue): Generate mret for interrupt handler.
15849 (riscv_epilogue_uses): New.
15850 (riscv_can_use_return_insn): Return false for interrupt handler.
15851 (riscv_function_ok_for_sibcall): Likewise.
15852 (riscv_set_current_function): Add interrupt handler support.
15853 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
15854 * config/riscv/riscv.md (UNSPECV_MRET): New.
15856 (riscv_frflags, riscv_fsflags): Use tab after opcode.
15858 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
15860 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
15862 PR tree-optimization/85712
15863 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
15864 this candidate has already been replaced in-situ by a copy.
15866 2018-05-25 Jason Merrill <jason@redhat.com>
15868 PR c++/80485 - inline function non-zero address.
15869 * symtab.c (nonzero_address): Check DECL_COMDAT.
15871 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
15874 * config/alpha/alpha.md (ashlsi3): New insn pattern.
15875 (*ashlsi_se): Rename from *ashldi_se. Define as sign
15876 extension of SImode operation. Use const123_operand predicate.
15877 (*saddsi_1): Remove.
15878 (*saddl_se_1): Ditto.
15879 (*ssubsi_1): Ditto.
15880 (*ssubl_se_1): Ditto.
15881 * config/alpha/predicates.md (const123_operand): New predicate.
15882 * config/alpha/constraints.md (P): Use IN_RANGE.
15884 2018-05-25 Richard Biener <rguenther@suse.de>
15886 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
15888 (ref_maybe_used_by_stmt_p): Likewise.
15889 (stmt_may_clobber_ref_p): Likewise.
15890 (stmt_may_clobber_ref_p_1): Likewise.
15891 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
15893 (ref_maybe_used_by_stmt_p): Likewise.
15894 (stmt_may_clobber_ref_p): Likewise.
15895 (stmt_may_clobber_ref_p_1): Likewise.
15896 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
15897 the alias oracle to disambiguate DRs with stmts DR analysis
15899 (vect_analyze_data_refs): Do not give up on not analyzable
15900 DRs for BB vectorization. Remove code truncating the dataref
15903 2018-05-25 Jakub Jelinek <jakub@redhat.com>
15906 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
15907 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
15908 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
15910 2018-05-25 Richard Biener <rguenther@suse.de>
15912 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
15913 function, combining stmt data ref gathering and fatal analysis
15915 (vect_analyze_data_refs): Remove now redudnant code and simplify.
15916 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
15917 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
15918 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
15919 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
15921 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15923 PR tree-optimization/85720
15924 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
15925 SCC if all partitions are builtins.
15926 (version_loop_by_alias_check): New parameter. Generate cancelable
15927 runtime alias check if all partitions are builtins.
15928 (distribute_loop): Update call to above function.
15930 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15932 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
15933 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
15934 (parm_default_def_partition_arg): Ditto.
15935 (set_parm_default_def_partition): Ditto.
15936 (get_parm_default_def_partitions): Ditto and make it static.
15937 (get_undefined_value_partitions): Ditto and make it static.
15938 (remove_ssa_form): Refactor call to init_var_map here.
15939 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
15940 computation for loop region.
15941 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
15942 (register_default_def): Delete.
15943 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
15944 (parm_default_def_partition_arg): Ditto.
15945 (set_parm_default_def_partition): Ditto.
15946 (get_parm_default_def_partitions): Ditto and make it static.
15947 (get_undefined_value_partitions): Ditto and make it static.
15948 (coalesce_with_default, coalesce_with_default): Update comment.
15949 (create_coalesce_list_for_region): New func factored out from
15950 create_outofssa_var_map.
15951 (populate_coalesce_list_for_outofssa): New func factored out from
15952 create_outofssa_var_map and coalesce_ssa_name.
15953 (create_outofssa_var_map): Delete.
15954 (coalesce_ssa_name): Refactor to support live range computation.
15955 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
15956 (get_parm_default_def_partitions): Delete.
15957 (get_undefined_value_partitions): Ditto.
15958 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
15959 computation for loop region.
15960 (new_tree_live_info, loe_visit_block): Ditto.
15961 (live_worklist, set_var_live_on_entry): Ditto.
15962 (calculate_live_on_exit, verify_live_on_entry): Ditto.
15963 * tree-ssa-live.h (struct _var_map): New fields.
15964 (init_var_map): Change decl.
15965 (region_contains_p): New.
15967 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15969 * tree-ssa-live.h (live_merge_and_clear): Delete.
15971 2018-05-25 Richard Biener <rguenther@suse.de>
15974 * tree-dump.c (dequeue_and_dump): Remove access to removed
15975 operand 2 of a SWITCH_EXPR.
15977 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15979 * doc/sourcebuild.texi (vect_double_cond_arith): Include
15980 multiplication and division.
15981 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
15982 (cond_udiv@var{m}, cond_umod@var{m}): Document.
15983 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
15984 (cond_udiv_optab, cond_umod_optab): New optabs.
15985 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
15986 (IFN_COND_RDIV): New internal functions.
15987 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
15988 TRUNC_MOD_EXPR and RDIV_EXPR.
15989 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
15990 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
15992 (SVE_INT_BINARY): Include mult.
15993 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
15994 (optab, sve_int_op): Handle mult.
15995 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
15997 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
15998 for SVE_INT_BINARY_SD.
16000 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16002 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
16003 (optab, sve_int_op): Handle div and udiv.
16004 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
16005 for SVE_INT_BINARY_SD.
16006 (*<optab><mode>3): New insn for the same.
16008 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16010 * tree-vect-patterns.c: Include predict.h.
16011 (vect_recog_divmod_pattern): Restrict check for division support
16012 to when optimizing for size.
16014 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16016 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
16017 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
16018 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
16019 (gimple_match_op::set_op): Likewise.
16020 (gimple_resimplify4): Declare.
16021 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
16022 (expr::gen_transform): Likewise.
16023 (decision_tree::gen): Generate a simplification routine for 4 operands.
16024 * gimple-match-head.c (gimple_simplify): Add an overload for
16025 4 operands. In the top-level function, handle up to 4 call
16026 arguments and call gimple_resimplify4.
16027 (gimple_resimplify4): New function.
16028 (build_call_internal): Pass a fourth operand.
16029 (maybe_push_to_seq): Likewise.
16030 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
16031 Fold VEC_COND_EXPRs of an operation and a default value into
16032 an IFN_COND_* function if possible.
16033 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
16035 (SVE_COND_FP_BINARY): Include them.
16036 (optab, sve_fp_op): Handle them.
16037 (SVE_INT_BINARY_REV): New code iterator.
16038 (SVE_COND_FP_BINARY_REV): New int iterator.
16039 (commutative): New int attribute.
16040 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
16042 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
16044 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
16045 (*cond_<optab><mode>): New patterns for reversed operands.
16047 2018-05-25 Richard Biener <rguenther@suse.de>
16049 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
16050 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
16051 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
16052 (STMT_VINFO_GROUPED_ACCESS): Adjust.
16053 * tree-vect-data-refs.c (everywhere): Adjust users.
16054 * tree-vect-loop.c (everywhere): Likewise.
16055 * tree-vect-slp.c (everywhere): Likewise.
16056 * tree-vect-stmts.c (everywhere): Likewise.
16057 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
16059 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16061 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
16063 (gcc_cv_as_section_exclude): ... this.
16064 Try Solaris as #exclude syntax.
16065 * configure: Regenerate.
16066 * config.in: Regenerate.
16067 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
16069 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
16070 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
16072 * varasm.c (default_elf_asm_named_section): Don't check if
16073 HAVE_GAS_SECTION_EXCLUDE is defined.
16075 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16077 * doc/md.texi: Update the documentation of the cond_* optabs
16078 to mention the new final operand. Fix GET_MODE_NUNITS call.
16079 Describe the scalar case too.
16080 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
16081 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
16083 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
16084 (get_conditional_internal_fn): Update comment.
16085 * tree-vect-loop.c (vectorizable_reduction): Pass the original
16086 accumulator value as a final argument to conditional functions.
16087 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
16088 a define_expand and add an "else" operand. Assert for now that
16089 the else operand is equal to operand 2. Use SVE_INT_BINARY and
16090 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
16091 (*cond_<optab><mode>): New patterns.
16092 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
16093 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
16094 (UNSPEC_COND_EOR): Delete.
16095 (optab): Remove associated mappings.
16096 (SVE_INT_BINARY): New code iterator.
16097 (sve_int_op): Remove int attribute and add "minus" to the code
16099 (SVE_COND_INT_OP): Delete.
16100 (SVE_COND_FP_OP): Rename to...
16101 (SVE_COND_FP_BINARY): ...this.
16103 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16105 * optabs.c (can_reuse_operands_p): New function.
16106 (maybe_legitimize_operands): Try to reuse the results for
16109 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
16111 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
16112 Add {q} suffix to insn mnemonic.
16114 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16116 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
16117 (msp430_warn_func_return): New.
16119 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
16121 * fold-const.c (tree_nonzero_bits): New function.
16122 * fold-const.h (tree_nonzero_bits): Likewise.
16123 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
16124 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
16126 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
16130 * varasm.c (assemble_alias): Check ifunc_resolver only on
16133 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
16136 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
16137 when memory input operand is handled.
16139 2018-05-24 Luis Machado <luis.machado@linaro.org>
16141 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
16143 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
16145 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
16147 * match.pd: Delay FMA folds until after vectorization.
16149 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
16152 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
16153 address check not strict.
16155 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
16157 * gimple-match.h (gimple_match_op): New class.
16158 (mprts_hook): Replace parameters with a gimple_match_op *.
16159 (maybe_build_generic_op): Likewise.
16160 (gimple_simplified_result_is_gimple_val): Replace parameters with
16161 a const gimple_match_op *.
16162 (gimple_simplify): Replace code_helper * and tree * parameters with
16163 a gimple_match_op * parameter.
16164 (gimple_resimplify1): Replace code_helper *, tree and tree *
16165 parameters with a gimple_match_op * parameter.
16166 (gimple_resimplify2): Likewise.
16167 (gimple_resimplify3): Likewise.
16168 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
16169 parameters with a gimple_match_op * parameter.
16170 * gimple-match-head.c (gimple_simplify): Change prototypes of
16171 auto-generated functions to take a gimple_match_op * instead of
16172 separate code_helper * and tree * parameters. Make the same
16173 change in the top-level overload and update calls to the
16174 gimple_resimplify routines. Update calls to the auto-generated
16175 functions and to maybe_push_res_to_seq in the publicly-facing
16176 operation-specific gimple_simplify overloads.
16177 (gimple_match_op::MAX_NUM_OPS): Define.
16178 (gimple_resimplify1): Replace rcode and ops with a single res_op
16179 parameter. Update call to gimple_simplify.
16180 (gimple_resimplify2): Likewise.
16181 (gimple_resimplify3): Likewise.
16182 (mprts_hook): Replace parameters with a gimple_match_op *.
16183 (maybe_build_generic_op): Likewise.
16184 (build_call_internal): Replace type, nargs and ops with
16185 a gimple_match_op *.
16186 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
16187 with a single gimple_match_op *. Update calls to mprts_hook,
16188 build_call_internal and gimple_simplified_result_is_gimple_val.
16189 Factor out code that is common to the tree_code and combined_fn cases.
16190 * genmatch.c (expr::gen_transform): Replace tem_code and
16191 tem_ops with a gimple_match_op called tem_op. Update calls
16192 to the gimple_resimplify functions and maybe_push_res_to_seq.
16193 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
16194 res_ops. Update call to the gimple_resimplify functions.
16195 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
16196 (decision_tree::gen): Make the functions take a gimple_match_op *
16197 called res_op instead of separate res_code and res_ops parameters.
16198 Update call accordingly.
16199 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
16200 and ops with a single res_op parameter. Update calls to
16201 maybe_build_generic_op and maybe_push_res_to_seq.
16202 (fold_stmt_1): Update calls to gimple_simplify and
16203 replace_stmt_with_simplification.
16204 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
16205 and gimple_simplified_result_is_gimple_val.
16206 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
16208 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
16209 with a gimple_match_op *.
16210 (vn_nary_build_or_lookup): Likewise. Update call to
16211 vn_nary_build_or_lookup_1.
16212 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
16213 gimple_match_op *. Update calls to the gimple_resimplify routines
16214 and to gimple_simplified_result_is_gimple_val.
16215 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
16216 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
16217 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
16218 (visit_nary_op): Likewise.
16219 (visit_reference_op_load): Likewise.
16221 2018-05-23 Luis Machado <luis.machado@linaro.org>
16223 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
16224 modifier for printing the step amount.
16226 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16229 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
16232 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
16234 * doc/sourcebuild.texi (Endianness): New subsubsection.
16236 2018-05-23 Luis Machado <luis.machado@linaro.org>
16238 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16239 <prefetch_dynamic_strides>: New const bool field.
16240 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16241 prefetch_dynamic_strides.
16242 (exynosm1_prefetch_tune): Likewise.
16243 (thunderxt88_prefetch_tune): Likewise.
16244 (thunderx_prefetch_tune): Likewise.
16245 (thunderx2t99_prefetch_tune): Likewise.
16246 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
16248 (aarch64_override_options_internal): Update to set
16249 PARAM_PREFETCH_DYNAMIC_STRIDES.
16250 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
16251 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
16252 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
16253 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
16254 prefetch-dynamic-strides setting.
16256 2018-05-23 Luis Machado <luis.machado@linaro.org>
16258 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16259 <minimum_stride>: New const int field.
16260 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16261 minimum_stride field defaulting to -1.
16262 (exynosm1_prefetch_tune): Likewise.
16263 (thunderxt88_prefetch_tune): Likewise.
16264 (thunderx_prefetch_tune): Likewise.
16265 (thunderx2t99_prefetch_tune): Likewise.
16266 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
16267 <default_opt_level>: Set to 3.
16268 (aarch64_override_options_internal): Update to set
16269 PARAM_PREFETCH_MINIMUM_STRIDE.
16270 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
16271 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
16272 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
16273 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
16274 stride is constant and is below the minimum stride threshold.
16276 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16278 * config/arm/arm-cpus.in (mode26): Delete.
16279 (armv4): Delete mode26 reference.
16280 * config/arm/arm.c (arm_configure_build_target): Delete use of
16283 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
16285 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
16287 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
16288 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
16290 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
16291 (floatunsdidf2): Ditto.
16293 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
16295 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
16296 (fixuns_trunc<mode>si2_avx512f): Ditto.
16297 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
16298 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
16299 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
16301 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
16303 PR rtl-optimization/79985
16304 * df-scan.c (df_insn_refs_collect): Remove special case for
16305 global registers and asm statements.
16307 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
16309 * extend.texi (Global Register Variables): Rewrite the bullet list.
16310 Note that the register is available for allocation. Note that access
16311 via inline asm must use constraints. Add note about async-signal
16312 handlers. Remove paragraph about automagic register selection.
16314 2018-05-23 Richard Biener <rguenther@suse.de>
16316 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
16317 of fixed offset from memset VN.
16319 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
16321 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
16322 first_interp field.
16323 (alloc_cand_and_find_basis): Initialize first_interp field.
16324 (slsr_process_mul): Modify first_interp field.
16325 (slsr_process_add): Likewise.
16326 (slsr_process_cast): Modify first_interp field for each new
16328 (slsr_process_copy): Likewise.
16329 (dump_candidate): Dump first_interp field.
16330 (replace_mult_candidate): Process all interpretations, not just
16332 (replace_rhs_if_not_dup): Likewise.
16333 (replace_one_candidate): Likewise.
16335 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
16337 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
16339 (aarch64_needs_frame_chain): New function.
16340 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
16342 2018-05-23 Sudakshina Das <sudi.das@arm.com>
16345 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
16346 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
16347 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
16348 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
16349 as true for strict-align.
16350 (aarch64_can_inline_p): Perform checks even when callee has no
16351 attributes to check for strict alignment.
16352 * doc/extend.texi (AArch64 Function Attributes): Document
16354 * doc/invoke.texi: (AArch64 Options): Likewise.
16356 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
16358 PR tree-optimization/85853
16359 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
16360 the handling of the root of the node to...
16361 (vect_slp_analyze_node_operations_1): ...this new function,
16362 and run the whole thing with the child nodes' def types
16363 set according to their SLP node's def type.
16365 2018-05-23 Richard Biener <rguenther@suse.de>
16367 PR middle-end/85874
16368 * tree-data-ref.c (create_runtime_alias_checks): Defer
16369 and ignore overflow warnings.
16371 2018-05-23 Yury Gribov <tetra2005@gmail.com>
16373 PR tree-optimization/85822
16374 * tree-vrp.c (is_masked_range_test): Fix handling of negative
16377 2018-05-23 Richard Biener <rguenther@suse.de>
16379 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
16380 memset constants via native_interpret_expr.
16382 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
16385 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
16387 (cgraph_node::create_alias): Likewise.
16388 (cgraph_node::get_availability): Check ifunc_resolver instead
16389 of looking up ifunc attribute.
16390 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
16391 * varasm.c (do_assemble_alias): Likewise.
16392 (assemble_alias): Likewise.
16393 (default_binds_local_p_3): Likewise.
16394 * cgraph.h (cgraph_node): Add ifunc_resolver.
16395 (cgraph_node::only_called_directly_or_aliased_p): Return false
16396 for IFUNC resolver.
16397 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
16399 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
16400 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
16401 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
16402 instead of looking up ifunc attribute.
16404 2018-05-22 Luis Machado <luis.machado@linaro.org>
16406 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
16408 2018-05-22 Martin Sebor <msebor@redhat.com>
16410 PR middle-end/85359
16411 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
16412 only when expasion succeeds.
16413 (expand_builtin_strcmp): Same.
16414 (expand_builtin_strncmp): Same.
16416 2018-05-22 Martin Sebor <msebor@redhat.com>
16418 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
16420 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
16421 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16423 * config/aarch64/aarch64-ldpstp.md: Replace uses of
16424 aarch64_mem_pair_operand with memory_operand and delete operand swapping
16426 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
16427 Add check for legitimate_address.
16428 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
16429 (aarch64_swap_ldrstr_operands): New.
16430 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
16433 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
16434 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16436 * config/aarch64/aarch64.md: New patterns to generate stp
16438 (store_pair_sw, store_pair_dw): New patterns to generate stp for
16439 single words and double words.
16440 (load_pair_sw, load_pair_dw): Likewise.
16441 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
16443 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
16445 * config/aarch64/aarch64-ldpstp.md: Modify peephole
16446 for different mode ldpstp and add peephole for merged zero stores.
16447 Likewise for loads.
16448 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
16450 (aarch64_gen_store_pair): Rename calls to match new patterns.
16451 (aarch64_gen_load_pair): Rename calls to match new patterns.
16452 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
16453 (load_pair<DREG:mode><DREG2:mode>): ... This.
16454 (store_pair<mode>): Rename to...
16455 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
16456 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
16457 New mode iterators.
16458 (V_INT_EQUIV): Handle SImode.
16459 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
16462 2018-05-22 Martin Sebor <msebor@redhat.com>
16465 * calls.c (maybe_warn_nonstring_arg): Use string length to set
16466 or ajust the presumed bound on an operation to avoid unnecessary
16469 2018-05-22 Martin Sebor <msebor@redhat.com>
16471 PR tree-optimization/85826
16472 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
16473 assuming that a DECL necesarily has a constant size.
16475 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
16477 PR middle-end/85862
16478 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
16480 2018-05-22 Richard Biener <rguenther@suse.de>
16482 PR tree-optimization/85834
16483 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
16484 non-constant and non-zero memset arguments.
16486 2018-05-22 Martin Liska <mliska@suse.cz>
16489 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
16491 2018-05-22 Richard Biener <rguenther@suse.de>
16493 PR tree-optimization/85863
16494 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
16495 comparisons when vectype is specified.
16496 (vectorizable_condition): Do not specify vectype for
16497 vect_is_simple_cond when SLP vectorizing.
16499 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
16502 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
16503 define __ibm128 as long double.
16504 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
16505 as a distinct type when IEEE 128-bit support is enabled.
16506 (init_float128_ieee): Fix up conversions between IFmode and IEEE
16507 128-bit types to use the correct functions.
16508 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
16509 convert between 128-bit floating point types that have different
16510 modes but the same representation, instead of using gen_lowpart to
16512 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
16514 (IFKF_reg): New attributes to give the register constraints for
16516 (extend<mode>tf2_internal): New insns to mark an explicit
16517 conversion between 128-bit floating point types that have a
16518 different mode but share the same representation.
16520 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
16522 PR tree-optimization/85814
16523 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
16524 a null return from get_strinfo when unsharing the next
16525 strinfo in the chain.
16527 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16530 * varasm.c (weak_finish): Clean up weak_decls.
16532 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16534 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
16535 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
16536 UNSPEC_UADALP values.
16537 * config/aarch64/iterators.md (ABAL): New int iterator.
16540 (sur): Add mappings for the above.
16541 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
16543 (aarch64_<sur>abal<mode>_4): Likewise.
16544 (aarch64_<sur>adalp<mode>_3): Likewise.
16545 (<sur>sadv16qi): New define_expand.
16547 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
16549 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
16550 (*movdf_internal): Ditto.
16551 (*rcpsf2_sse): Ditto.
16552 (*rsqrtsf2_sse): Ditto.
16553 (*sqrt<mode>2_sse): Ditto.
16555 2018-05-21 Tamar Christina <tamar.christina@arm.com>
16557 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
16559 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
16560 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
16561 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
16562 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
16564 * config/aarch64/arm_neon.h: Likewise.
16565 * config/aarch64/iterators.md (VQ_I): New.
16567 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
16569 * config.gcc: Add arc/t-multilib-linux to tmake_file for
16571 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
16574 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16576 * config/nds32/constraints.md (S): New constraint.
16577 * config/nds32/nds32.md (call_internal): Use constraint S.
16578 (call_value_internal): Likewise.
16579 (sibcall_internal): Likewise.
16580 (sibcall_value_internal): Likewise.
16582 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16583 Chung-Ju Wu <jasonwucj@gmail.com>
16585 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
16586 into consideration.
16588 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16589 Chung-Ju Wu <jasonwucj@gmail.com>
16591 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
16592 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
16593 (nds32_rtx_costs_impl): Simplify.
16594 (nds32_address_cost_impl): Simplify.
16595 (nds32_init_rtx_costs): New function.
16596 (nds32_rtx_costs_speed_prefer): Likewise.
16597 (nds32_rtx_costs_size_prefer): Likewise.
16598 (nds32_address_cost_speed_prefer): Likewise.
16599 (nds32_address_cost_speed_fwprop): Likewise.
16600 (nds32_address_cost_size_prefer): Likewise.
16601 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
16602 * config/nds32/nds32.c (nds32_option_override): Use
16603 nds32_init_rtx_costs function.
16605 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16607 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
16608 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
16609 (TARGET_PIPELINE_N8): Likewise.
16610 (TARGET_PIPELINE_N10): Likewise.
16611 (TARGET_PIPELINE_N13): Likewise.
16612 (TARGET_PIPELINE_GRAYWOLF): Likewise.
16614 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
16616 * config/nds32/nds32-fpu.md: Update copyright year.
16618 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16620 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
16622 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16624 * config/nds32/nds32.c
16625 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
16626 * config/nds32/nds32.opt (minline-asm-r15): New option.
16628 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16630 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
16632 * config/nds32/nds32.md (abssi2): New pattern.
16634 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
16636 * config/i386/i386.md (rex64namesuffix): New mode attribute.
16637 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
16638 Merge insn pattern from sse_cvtsi2ss<round_name> and
16639 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
16640 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
16641 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
16642 using SWI48 mode iterator.
16643 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
16644 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
16645 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
16646 pattern from sse_cvttss2si<round_saeonly_name>
16647 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
16648 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
16649 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
16650 using SWI48 mode iterator.
16651 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
16652 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
16653 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
16654 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
16655 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
16656 using SWI48 mode iterator.
16657 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
16658 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
16659 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
16660 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
16661 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
16662 SWI48 mode iterator.
16663 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
16664 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
16665 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
16666 pattern from sse_cvttsd2si<round_saeonly_name>
16667 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
16669 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16671 * config/nds32/nds32-md-auxiliary.c
16672 (nds32_valid_smw_lwm_base_p): Refine.
16673 (nds32_output_smw_single_word): Refine.
16674 (nds32_output_smw_double_word): New.
16675 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
16677 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16679 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
16680 (nds32_output_stack_pop): Refine.
16681 (nds32_expand_unaligned_load): Refine.
16682 (nds32_expand_unaligned_store): Refine.
16684 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16685 Chung-Ju Wu <jasonwucj@gmail.com>
16687 * config/nds32/constants.md: Add TP_REGNUM constant.
16688 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
16689 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
16691 * config/nds32/nds32-doubleword.md: Consider flag_pic.
16692 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
16693 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
16694 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
16695 and PIC code generation.
16696 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
16698 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
16700 * config/nds32/nds32.md: Support TLS and PIC.
16701 * config/nds32/nds32.c: Support TLS and PIC.
16702 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
16703 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
16706 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16708 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
16709 mode with E_ prefix.
16711 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16712 Chung-Ju Wu <jasonwucj@gmail.com>
16714 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
16715 * config/nds32/nds32-md-auxiliary.c
16716 (symbolic_reference_mentioned_p): New.
16717 (nds32_legitimize_ict_address): New.
16718 (nds32_expand_ict_move): New.
16719 (nds32_indirect_call_referenced_p): New.
16720 (nds32_symbol_binds_local_p): Delete.
16721 (nds32_long_call_p): Modify.
16722 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
16723 * config/nds32/nds32-protos.h
16724 (symbolic_reference_mentioned_p): Declare.
16725 (nds32_legitimize_ict_address): Declare.
16726 (nds32_expand_ict_move): Declare.
16727 (nds32_indirect_call_referenced_p): Declare.
16728 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
16729 (nds32_relax_group): Use nds32_ict_const_p as condition.
16730 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
16731 (nds32_asm_file_start): Output ict_model directive in asm code.
16732 (nds32_legitimate_address_p): Consider indirect call.
16733 (nds32_print_operand): Consider indirect call.
16734 (nds32_print_operand_address): Consider indirect call.
16735 (nds32_insert_attributes): Handle "indirect_call" attribute.
16736 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16737 (TARGET_LEGITIMATE_CONSTANT_P): Define.
16738 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
16739 (TARGET_DELEGITIMIZE_ADDRESS): Define.
16740 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
16741 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
16742 (TARGET_ICT_MODEL_SMALL): Define.
16743 (TARGET_ICT_MODEL_LARGE): Define.
16744 * config/nds32/nds32.md (movsi): Consider ict model.
16745 (call, call_value): Consider ict model.
16746 (sibcall, sibcall_value): Consider ict model.
16747 * config/nds32/nds32.opt (mict-model): New option.
16748 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
16751 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
16752 Monk Chiang <sh.chiang04@gmail.com>
16753 Jim Wilson <jimw@sifive.com>
16755 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
16756 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
16757 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
16758 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
16759 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
16760 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
16761 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
16762 compute save_libcall_adjustment properly.
16763 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
16764 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
16765 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
16766 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
16767 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
16768 (ABI_SPEC): Handle mabi=ilp32e.
16769 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
16770 (RVE): Add RVE mask.
16771 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
16772 <-march>: Add rv32e as an example.
16774 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
16777 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
16778 (intra_create_variable_infos): Handle C++ constructors.
16780 2018-05-18 Martin Liska <mliska@suse.cz>
16782 * passes.def: Remove a redundant pass.
16784 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
16787 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
16789 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16791 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
16793 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
16795 (armv2, armv2a, armv3, armv3m): Delete architectures.
16796 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
16797 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
16798 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
16800 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
16801 (*mulsidi3adddi): Likewise.
16802 (mulsidi3): Likewise.
16803 (*mulsidi3_nov6): Likewise.
16804 (umulsidi3): Likewise.
16805 (umulsidi3_nov6): Likewise.
16806 (umaddsidi4): Likewise.
16807 (*umulsidi3adddi): Likewise.
16808 (smulsi3_highpart): Likewise.
16809 (*smulsi3_highpart_nov6): Likewise.
16810 (umulsi3_highpart): Likewise.
16811 (*umulsi3_highpart_nov6): Likewise.
16812 * config/arm/arm.h (arm_arch3m): Delete.
16813 * config/arm/arm.c (arm_arch3m): Delete.
16814 (arm_option_override_internal): Update armv3-related comment.
16815 (arm_configure_build_target): Delete use of isa_bit_mode32.
16816 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
16817 (arm_rtx_costs_internal): Delete check of arm_arch3m.
16818 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
16819 (mulsa3): Likewise.
16820 (mulusa3): Likewise.
16821 * config/arm/arm-protos.h (arm_arch3m): Delete.
16822 * config/arm/arm-tables.opt: Regenerate.
16823 * config/arm/arm-tune.md: Likewise.
16824 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
16825 deleted architectures.
16827 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16829 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
16830 (armv5t, armv5te): New features.
16831 (ARMv5, ARMv5e): Delete fgroups.
16832 (ARMv5t, ARMv5te): Adjust for above changes.
16833 (ARMv6m): Likewise.
16834 (armv5, armv5e): Delete arches.
16835 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
16837 (*call_reg_arm): Likewise.
16838 (*call_value_reg_armv5): Likewise.
16839 (*call_value_reg_arm): Likewise.
16840 (*call_symbol): Likewise.
16841 (*call_value_symbol): Likewise.
16842 (*sibcall_insn): Likewise.
16843 (*sibcall_value_insn): Likewise.
16844 (clzsi2): Likewise.
16845 (prefetch): Likewise.
16846 (define_split and define_peephole2 dependent on arm_arch5):
16848 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
16850 (TARGET_ARM_QBIT): Likewise.
16851 (TARGET_DSP_MULTIPLY): Likewise.
16852 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
16853 (arm_arch5, arm_arch5e): Delete.
16854 (arm_arch5t, arm_arch5te): Declare.
16855 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
16856 (arm_arch5t): Declare.
16857 (arm_option_reconfigure_globals): Update for the above.
16858 (arm_options_perform_arch_sanity_checks): Update comment, replace
16859 use of arm_arch5 with arm_arch5t.
16860 (use_return_insn): Likewise.
16861 (arm_emit_call_insn): Likewise.
16862 (output_return_instruction): Likewise.
16863 (arm_final_prescan_insn): Likewise.
16864 (arm_coproc_builtin_available): Likewise.
16865 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
16866 arm_arch5e with arm_arch5t and arm_arch5te.
16867 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
16868 (arm_arch5t, arm_arch5te): Declare.
16869 * config/arm/arm-tables.opt: Regenerate.
16870 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
16871 * config/arm/t-multilib: Likewise.
16872 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
16873 instead of arm_arch5.
16874 (*call_reg_thumb1): Likewise.
16875 (*call_value_reg_thumb1_v5): Likewise.
16876 (*call_value_reg_thumb1): Likewise.
16877 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
16879 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
16881 2018-05-18 Martin Liska <mliska@suse.cz>
16883 PR gcov-profile/84846
16884 * doc/gcov.texi: Document -t option of gcov tool.
16886 2018-05-18 Martin Liska <mliska@suse.cz>
16888 PR gcov-profile/84846
16889 * gcov.c (print_usage): Add new -t option.
16890 (process_args): Handle the option.
16891 (generate_results): Use stdout as output when requested by
16894 2018-05-18 Martin Liska <mliska@suse.cz>
16896 PR gcov-profile/84846
16897 * coverage.c (coverage_init): Write PWD to .gcno file.
16898 * doc/gcov.texi: Document how working directory is printed.
16899 * gcov-dump.c (dump_gcov_file): Print PWD.
16900 * gcov.c (output_intermediate_file): Likewise.
16901 (read_graph_file): Read PWD string.
16902 (output_lines): Print PWD.
16904 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16906 PR middle-end/85817
16907 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
16908 for retval and return false if all args to phi are zero.
16910 2018-05-18 Richard Biener <rguenther@suse.de>
16912 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
16914 (evrp_dom_walker::before_dom_children): Call it.
16916 2018-05-18 Richard Biener <rguenther@suse.de>
16918 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
16919 results when processing array refs with variable index.
16921 2018-05-18 Toon Moene <toon@moene.org>
16923 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
16924 directly after that of -floop-interchange. Indicate that both
16925 options are enabled by default when specifying -O3.
16927 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16929 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
16930 iterator. Delete separate integer-mode vec_set<mode> expander.
16931 (aarch64_simd_vec_setv2di): Delete.
16932 (vec_setv2di): Delete.
16933 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
16934 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
16935 the "w, r" alternative.
16937 2018-05-18 Martin Liska <mliska@suse.cz>
16939 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
16940 * tree-pass.h (make_pass_lower_switch_O0): New function.
16941 * tree-switch-conversion.c (node_has_low_bound): Remove.
16942 (node_has_high_bound): Likewise.
16943 (node_is_bounded): Likewise.
16944 (class pass_lower_switch): Make it a template type and create
16946 (pass_lower_switch::execute): Add template argument.
16947 (make_pass_lower_switch): New function.
16948 (make_pass_lower_switch_O0): New function.
16949 (do_jump_if_equal): Remove.
16950 (emit_case_nodes): Simplify to just handle all 3 cases and leave
16951 all the hard work to tree optimization passes.
16953 2018-05-18 Martin Liska <mliska@suse.cz>
16955 * dbgcnt.c (limit_low): Renamed from limit.
16956 (limit_high): New variable.
16957 (dbg_cnt_is_enabled): Check for upper limit.
16958 (dbg_cnt): Adjust dumping.
16959 (dbg_cnt_set_limit_by_index): Add new argument for high
16961 (dbg_cnt_set_limit_by_name): Likewise.
16962 (dbg_cnt_process_single_pair): Parse new format.
16963 (dbg_cnt_process_opt): Use strtok.
16964 (dbg_cnt_list_all_counters): Remove 'value' and add
16966 * doc/invoke.texi: Document changes.
16968 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
16970 * doc/sourcebuild.texi (scalar_all_fma): Document.
16971 * tree.def (FMA_EXPR): Delete.
16972 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
16973 * internal-fn.c (ternary_direct): New macro.
16974 (expand_ternary_optab_fn): Likewise.
16975 (direct_ternary_optab_supported_p): Likewise.
16976 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
16977 * builtins.c (fold_builtin_fma): Delete.
16978 (fold_builtin_3): Don't call it.
16979 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
16980 * expr.c (expand_expr_real_2): Likewise.
16981 * fold-const.c (operand_equal_p): Likewise.
16982 (fold_ternary_loc): Likewise.
16983 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
16984 * gimple.c (DEFTREECODE): Likewise.
16985 * gimplify.c (gimplify_expr): Likewise.
16986 * optabs-tree.c (optab_for_tree_code): Likewise.
16987 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16988 * tree-eh.c (operation_could_trap_p): Likewise.
16989 (stmt_could_throw_1_p): Likewise.
16990 * tree-inline.c (estimate_operator_cost): Likewise.
16991 * tree-pretty-print.c (dump_generic_node): Likewise.
16992 (op_code_prio): Likewise.
16993 * tree-ssa-loop-im.c (stmt_cost): Likewise.
16994 * tree-ssa-operands.c (get_expr_operands): Likewise.
16995 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
16996 * fold-const-call.h (fold_fma): Delete.
16997 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
16998 CFN_FNMA and CFN_FNMS.
16999 (fold_fma): Delete.
17000 * genmatch.c (combined_fn): New enum.
17001 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
17002 (commutative_op): New function.
17003 (commutate): Use it. Handle more than 2 operands.
17004 (dt_operand::gen_gimple_expr): Use commutative_op.
17005 (parser::parse_expr): Allow :c to be used with non-binary
17006 operators if the commutative operand is known.
17007 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
17008 CFN_FMS, CFN_FNMA and CFN_FNMS.
17009 (backprop::process_assign_use): Remove FMA_EXPR handling.
17010 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
17011 (gen_hsa_fma): New function.
17012 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
17013 IFN_FNMA and IFN_FNMS.
17014 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
17015 * gimple-fold.h (follow_all_ssa_edges): Declare.
17016 * gimple-fold.c (follow_all_ssa_edges): New function.
17017 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
17018 gimple_build interface and use follow_all_ssa_edges to fold the result.
17019 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
17020 instead of checking for optabs directly.
17021 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
17022 rather than FMA_EXPRs.
17023 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
17024 call to IFN_FMA instead of an FMA_EXPR.
17026 2018-05-17 Jim Wilson <jimw@sifive.com>
17028 * expr.c (do_tablejump): When converting index to Pmode, if we have a
17029 sign extended promoted subreg, and the range does not have the sign bit
17030 set, then do a sign extend.
17032 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
17033 test, check for sign extended subreg and/or constant operands, and
17034 do a sign extend in that case.
17036 2018-05-17 Steve Ellcey <sellcey@cavium.com>
17038 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
17039 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
17041 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
17042 Change logics_shift_reg to logics_shift_imm.
17043 (thunderx2t99_fp_loadpair_basic): Delete.
17044 (thunderx2t99_fp_storepair_basic): Delete.
17045 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
17046 (thunderx2t99_asimd_polynomial): Delete.
17047 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
17048 and neon_fp_mul_d_scalar_q.
17049 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
17050 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
17051 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
17052 (thunderx2t99_asimd_lut): Add missing tbl types.
17053 (thunderx2t99_asimd_ext): Delete.
17054 (thunderx2t99_asimd_load1_1_mult): Delete.
17055 (thunderx2t99_asimd_load1_2_mult): Delete.
17056 (thunderx2t99_asimd_load1_ldp): New.
17057 (thunderx2t99_asimd_load1): New.
17058 (thunderx2t99_asimd_load2): Add missing *load2* types.
17059 (thunderx2t99_asimd_load3): New.
17060 (thunderx2t99_asimd_load4): New.
17061 (thunderx2t99_asimd_store1_1_mult): Delete.
17062 (thunderx2t99_asimd_store1_2_mult): Delete.
17063 (thunderx2t99_asimd_store2_mult): Delete.
17064 (thunderx2t99_asimd_store2_onelane): Delete.
17065 (thunderx2t99_asimd_store_stp): New.
17066 (thunderx2t99_asimd_store1): New.
17067 (thunderx2t99_asimd_store2): New.
17068 (thunderx2t99_asimd_store3): New.
17069 (thunderx2t99_asimd_store4): New.
17071 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
17073 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
17074 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
17076 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
17077 Segher Boessenkool <segher@kernel.crashing.org>
17080 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
17083 2018-05-17 Richard Biener <rguenther@suse.de>
17085 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
17086 for pruning loop and prune defs feeding only already visited PHIs.
17088 2018-05-17 Richard Biener <rguenther@suse.de>
17090 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
17092 2018-05-17 Bin Cheng <bin.cheng@arm.com>
17093 Richard Biener <rguenther@suse.de>
17095 PR tree-optimization/85793
17096 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
17097 for VMAT_ELEMENTWISE.
17099 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
17101 * internal-fn.h (lookup_internal_fn): Declare
17102 * internal-fn.c (lookup_internal_fn): New function.
17103 * gimple.c (gimple_build_call_from_tree): Handle calls to
17104 internal functions.
17105 * gimple-pretty-print.c (dump_gimple_call): Print "." before
17106 internal function names.
17107 * tree-pretty-print.c (dump_generic_node): Likewise.
17108 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
17110 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
17112 * gimple-fold.h (gimple_build): Make the function forms take
17113 combined_fn rather than built_in_function.
17114 (gimple_simplify): Likewise.
17115 * gimple-match-head.c (gimple_simplify): Likewise.
17116 * gimple-fold.c (gimple_build): Likewise.
17117 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
17118 rather than gimple_build_call_internal.
17119 (get_initial_defs_for_reduction): Likewise.
17120 (vect_create_epilog_for_reduction): Likewise.
17121 (vectorizable_live_operation): Likewise.
17123 2018-05-17 Martin Liska <mliska@suse.cz>
17125 * gimple-ssa-sprintf.c (format_directive): Do not use
17126 space in between 'G_' and '('.
17128 2018-05-17 Jakub Jelinek <jakub@redhat.com>
17131 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
17132 even if the mask is not all ones.
17135 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
17137 (ix86_gimple_fold_builtin): Likewise.
17140 * config/i386/i386.c: Include tree-vector-builder.h.
17141 (ix86_vector_shift_count): New function.
17142 (ix86_fold_builtin): Fold shift builtins by scalar count.
17143 (ix86_gimple_fold_builtin): Likewise.
17145 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
17146 _mm512_setzero): New intrinsics.
17148 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
17149 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17151 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
17152 code generation for cases where splatting a value is not useful.
17153 * simplify-rtx.c (simplify_ternary_operation): Simplify
17154 vec_merge across a vec_duplicate and a paradoxical subreg forming
17155 a vector mode to a vec_concat.
17157 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
17159 * config.gcc: Support "goldmont-plus".
17160 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17162 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17163 PROCESSOR_GOLDMONT_PLUS.
17164 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
17165 (processor_target_table): Add "goldmont-plus".
17166 (PTA_GOLDMONT_PLUS): Define.
17167 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
17168 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
17169 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
17170 (fold_builtin_cpu): Add "goldmont-plus".
17171 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
17172 (ix86_option_override_internal): Add "goldmont-plus".
17173 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
17174 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
17175 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
17176 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
17178 2018-05-17 Richard Biener <rguenther@suse.de>
17180 PR tree-optimization/85757
17181 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
17182 remove defs that only feed that PHI from further processing.
17184 2018-05-16 Jim Wilson <jimw@sifive.com>
17186 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
17188 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
17190 2018-05-16 Mark Wielaard <mark@klomp.org>
17192 * dwarf2out.c (count_index_strings): New function.
17193 (output_indirect_strings): Call count_index_strings and generate
17194 header for dwarf_version >= 5.
17196 2018-05-16 Mark Wielaard <mark@klomp.org>
17198 * dwarf2out.c (dwarf_FORM): New function.
17199 (set_indirect_string): Use dwarf_FORM.
17200 (reset_indirect_string): Likewise.
17201 (size_of_die): Likewise.
17202 (value_format): Likewise.
17203 (output_die): Likewise.
17204 (add_skeleton_AT_string): Likewise.
17205 (output_macinfo_op): Likewise.
17206 (index_string): Likewise.
17207 (output_index_string_offset): Likewise.
17208 (output_index_string): Likewise.
17209 (count_index_strings): Likewise.
17211 2018-05-16 Carl Love <cel@us.ibm.com>
17213 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17214 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
17216 2018-05-16 Martin Jambor <mjambor@suse.cz>
17218 * ipa-prop.c (ipa_free_all_edge_args): Remove.
17219 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
17221 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
17223 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
17224 (fnma<mode>4): Likewise.
17225 (fms<mode>4): Likewise.
17226 (fnms<mode>4): Likewise.
17227 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
17228 (aarch64_fnma<mode>4): Likewise.
17229 (aarch64_fms<mode>4): Likewise.
17230 (aarch64_fnms<mode>4): Likewise.
17231 (aarch64_fnmadd<mode>4): Likewise.
17233 2018-05-16 Jason Merrill <jason@redhat.com>
17235 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
17237 2018-05-16 Richard Biener <rguenther@suse.de>
17239 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
17240 (dump_stmt_cost): Declare.
17241 (add_stmt_cost): Dump cost we add.
17242 (add_stmt_costs): New function.
17243 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
17244 No longer exported.
17245 (vect_analyze_stmt): Adjust prototype.
17246 (vectorizable_condition): Likewise.
17247 (vectorizable_live_operation): Likewise.
17248 (vectorizable_reduction): Likewise.
17249 (vectorizable_induction): Likewise.
17250 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
17251 cost vector to pass to vectorizable_ and record afterwards.
17252 (vect_model_reduction_cost): Take cost vector argument and adjust.
17253 (vect_model_induction_cost): Likewise.
17254 (vectorizable_reduction): Likewise.
17255 (vectorizable_induction): Likewise.
17256 (vectorizable_live_operation): Likewise.
17257 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
17258 SLP_TREE_NUMBER_OF_VEC_STMTS.
17259 (vect_analyze_slp_cost_1): Remove.
17260 (vect_analyze_slp_cost): Likewise.
17261 (vect_slp_analyze_node_operations): Take visited args and
17262 a target cost vector. Avoid processing already visited stmt sets.
17263 (vect_slp_analyze_operations): Use a local cost vector to gather
17264 costs and register those of non-discarded instances.
17265 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
17266 (vect_schedule_slp_instance): Remove copying of
17267 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
17269 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
17270 adding cost. Record cost entry location.
17271 (vect_prologue_cost_for_slp_op): Function to compute cost of
17272 a constant or invariant generated for SLP vect in the prologue,
17273 split out from vect_analyze_slp_cost_1.
17274 (vect_model_simple_cost): Make static. Adjust for SLP costing.
17275 (vect_model_promotion_demotion_cost): Likewise.
17276 (vect_model_store_cost): Likewise, make static.
17277 (vect_model_load_cost): Likewise.
17278 (vectorizable_bswap): Add cost vector arg and adjust.
17279 (vectorizable_call): Likewise.
17280 (vectorizable_simd_clone_call): Likewise.
17281 (vectorizable_conversion): Likewise.
17282 (vectorizable_assignment): Likewise.
17283 (vectorizable_shift): Likewise.
17284 (vectorizable_operation): Likewise.
17285 (vectorizable_store): Likewise.
17286 (vectorizable_load): Likewise.
17287 (vectorizable_condition): Likewise.
17288 (vectorizable_comparison): Likewise.
17289 (can_vectorize_live_stmts): Likewise.
17290 (vect_analyze_stmt): Likewise.
17291 (vect_transform_stmt): Adjust calls to vectorizable_*.
17292 * tree-vectorizer.c: Include gimple-pretty-print.h.
17293 (dump_stmt_cost): New function.
17295 2018-05-16 Richard Biener <rguenther@suse.de>
17297 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
17298 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
17299 * tree-ssa-dse.c: Include tree-ssa-loop.h.
17300 (check_name): New callback.
17301 (dse_classify_store): Track cycles via a visited bitmap of PHI
17302 defs and simplify handling of in-loop and across loop dead stores
17303 and properly fail for loop-variant refs. Handle byte-tracking with
17304 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
17307 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
17309 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
17310 (vect_get_mask_type_for_stmt): Likewise.
17311 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
17313 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
17314 to determine the statement's vector type and the vector type that
17315 should be used for calculating nunits. Deal with cases in which
17316 the type has to be deferred.
17317 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
17318 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
17319 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
17320 (vect_determine_vf_for_stmt): New functions, split out from...
17321 (vect_determine_vectorization_factor): ...here.
17322 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
17323 (vect_get_mask_type_for_stmt): New functions, split out from
17324 vect_determine_vectorization_factor.
17326 2018-05-16 Richard Biener <rguenther@suse.de>
17328 * tree-cfg.c (verify_gimple_assign_ternary): Properly
17329 verify the [VEC_]COND_EXPR embedded comparison.
17331 2018-05-15 Martin Sebor <msebor@redhat.com>
17333 PR tree-optimization/85753
17334 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
17335 RECORD_TYPE in addition to ARRAY_TYPE.
17337 2018-05-15 Martin Sebor <msebor@redhat.com>
17339 PR middle-end/85643
17340 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
17342 2018-05-15 Richard Biener <rguenther@suse.de>
17344 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
17345 add by_clobber_p one. Change algorithm to collect all defs
17346 representing uses we need to walk and try reducing them to
17347 a single one before failing.
17348 (dse_dom_walker::dse_optimize_stmt): Adjust.
17350 2018-05-13 Mark Wielaard <mark@klomp.org>
17352 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
17353 (size_of_loc_descr): Likewise.
17354 (output_loc_operands): Likewise.
17355 (output_loc_operands_raw): Likewise.
17356 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
17357 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
17358 (hash_loc_operands): Likewise.
17359 (compare_loc_operands): Likewise.
17361 2018-05-14 Mark Wielaard <mark@klomp.org>
17363 * dwarf2out.c (count_index_addrs): New function.
17364 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
17366 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17368 PR tree-optimization/83648
17369 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
17370 return value as malloc candidate.
17372 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17375 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
17376 param as true in call to suggest_attribute.
17378 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
17380 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
17381 -mreadonly-in-sdata.
17383 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17385 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
17387 (aarch64_crypto_aesd_fused): Likewise.
17389 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
17391 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
17392 (movsi_aarch64): Likewise.
17393 (load_pairsi): Likewise.
17394 (load_pairdi): Likewise.
17395 (store_pairsi): Likewise.
17396 (store_pairdi): Likewise.
17397 (load_pairsf): Likewise.
17398 (load_pairdf): Likewise.
17399 (store_pairsf): Likewise.
17400 (store_pairdf): Likewise.
17401 (zero_extend): Likewise.
17402 (trunc): Swap alternatives.
17403 (fcvt_target): Add '?' to prefer w over r.
17405 2018-05-14 Jakub Jelinek <jakub@redhat.com>
17408 * config/i386/i386.md: Disallow non-commutative arithmetics in
17409 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
17410 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
17411 in the peephole2 before it.
17413 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
17415 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
17416 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
17417 (ix86_handle_option): Handle -mcldemote.
17418 * config.gcc: New header.
17419 * config/i386/cldemoteintrin.h: New file.
17420 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
17421 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17423 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17424 OPTION_MASK_ISA_CLDEMOTE.
17425 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
17426 (ix86_valid_target_attribute_inner_p): Ditto.
17427 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
17428 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
17429 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
17430 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
17431 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
17433 * config/i386/i386.opt: Add -mcldemote.
17434 * config/i386/x86intrin.h: New header.
17435 * doc/invoke.texi: Add -mcldemote.
17437 2018-05-14 Richard Biener <rguenther@suse.de>
17439 * doc/match-and-simplify.texi: Adjust :s documentation.
17441 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
17443 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
17444 intended memcpy size.
17445 (REORDER_45): Likewise.
17447 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
17449 * sort.cc: New file.
17450 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
17451 * vec.c (qsort_chk): Use gcc_qsort.
17452 * Makefile.in (OBJS-libcommon): Add sort.o.
17453 (build/sort.o): New target. Use it...
17454 (BUILD_RTL): ... here, and...
17455 (build/gencfn-macros): ... here, and...
17456 (build/genmatch): ... here.
17458 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17459 Chung-Ju Wu <jasonwucj@gmail.com>
17461 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
17462 * config/nds32/nds32-graywolf.md: New file.
17463 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
17464 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
17466 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
17467 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
17468 * config/nds32/nds32.md (pipeline_model): Add graywolf.
17469 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
17470 * config/nds32/pipelines.md: Include n15 settings.
17472 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17473 Chung-Ju Wu <jasonwucj@gmail.com>
17475 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
17476 * config/nds32/nds32-n13.md: New file.
17477 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
17478 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
17480 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
17481 * config/nds32/nds32.md (pipeline_model): Add n13.
17482 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
17483 * config/nds32/pipelines.md: Include n13 settings.
17485 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17486 Chung-Ju Wu <jasonwucj@gmail.com>
17488 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
17489 * config/nds32/nds32-n10.md: New file.
17490 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
17491 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
17493 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
17494 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
17495 * config/nds32/nds32.md (pipeline_model): Add n10.
17496 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
17497 * config/nds32/pipelines.md: Include n10 settings.
17499 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
17500 Kito Cheng <kito.cheng@gmail.com>
17501 Chung-Ju Wu <jasonwucj@gmail.com>
17503 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
17504 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
17505 Add enum values for DSP extension instructions.
17506 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
17508 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
17509 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
17510 New code iterators.
17511 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
17512 * config/nds32/nds32-dspext.md: New file for DSP implementation.
17513 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
17514 * config/nds32/nds32-intrinsic.md: Likewise.
17515 * config/nds32/nds32_intrinsic.h: Likewise.
17516 * config/nds32/nds32-md-auxiliary.c: Likewise.
17517 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
17518 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
17519 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
17520 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
17521 * config/nds32/nds32-protos.h: New declarations for DSP extension.
17522 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
17523 TYPE_DMAC in switch statement.
17524 * config/nds32/nds32.c: New checking and implementation for DSP
17525 extension instructions.
17526 * config/nds32/nds32.h: Likewise.
17527 * config/nds32/nds32.md: Likewise.
17528 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
17529 * config/nds32/predicates.md: Implement new predicates for DSP
17532 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17534 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
17535 Reformat alternatives and attributes so it is easier to identify
17536 which constraints/attributes go with which instruction.
17537 (mov<mode>_hardfloat32, FMOVE64): Likewise.
17538 (mov<mode>_softfloat32, FMOVE64): Likewise.
17539 (mov<mode>_hardfloat64, FMOVE64): Likewise.
17540 (mov<mode>_softfloat64, FMOVE64): Likewise.
17542 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17544 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17546 (Basic PowerPC Built-in Functions): The new name of the
17547 subsection previously known as "PowerPC Built-in Functions".
17548 (Basic PowerPC Built-in Functions Available on all Configurations):
17550 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
17551 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
17552 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
17553 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
17555 2018-05-11 Martin Jambor <mjambor@suse.cz>
17558 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
17561 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17564 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
17566 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
17568 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
17569 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
17570 (ix86_handle_option): Handle -mwaitpkg.
17571 * config.gcc: New header.
17572 * config/i386/cpuid.h (bit_WAITPKG): New bit.
17573 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
17574 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
17576 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17577 OPTION_MASK_ISA_WAITPKG.
17578 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
17579 (ix86_option_override_internal): Add PTA_WAITPKG.
17580 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
17581 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
17582 IX86_BUILTIN_TPAUSE.
17583 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
17584 __builtin_ia32_umwait and __builtin_ia32_tpause.
17585 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
17586 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
17587 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
17588 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
17589 UNSPECV_TPAUSE): New.
17590 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
17591 * config/i386/i386.opt: Add -mwaitpkg.
17592 * config/i386/waitpkgintrin.h: New file.
17593 * config/i386/x86intrin.h: New header.
17594 * doc/invoke.texi: Add -mwaitpkg.
17596 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17599 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
17601 (cortex-m0): Use armv6s-m isa.
17602 (cortex-m0plus): Likewise.
17603 (cortex-m1): Likewise.
17604 (cortex-m0.small-multiply): Likewise.
17605 (cortex-m0plus.small-multiply): Likewise.
17606 (cortex-m1.small-multiply): Likewise.
17608 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
17609 Jakub Jelinek <jakub@redhat.com>
17611 PR tree-optimization/85692
17612 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
17613 source permute as well.
17615 2018-05-11 Martin Liska <mliska@suse.cz>
17618 * doc/extend.texi: Document LLVM style format for no_sanitize
17621 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
17623 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
17624 mode_supports_vsx_dform_quad to mode_supports_dq_form.
17625 (mode_supports_vsx_dform_quad): Likewise.
17626 (mode_supports_vmx_dform): Move these functions to be next to the
17627 other mode_supports functions.
17628 (mode_supports_dq_form): Likewise.
17629 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
17630 mode_supports_dq_form.
17631 (reg_offset_addressing_ok_p): Likewise.
17632 (offsettable_ok_by_alignment): Likewise.
17633 (rs6000_legitimate_offset_address_p): Likewise.
17634 (legitimate_lo_sum_address_p): Likewise.
17635 (rs6000_legitimize_address): Likewise.
17636 (rs6000_legitimize_reload_address): Likewise.
17637 (rs6000_secondary_reload_inner): Likewise.
17638 (rs6000_preferred_reload_class): Likewise.
17639 (rs6000_output_move_128bit): Likewise.
17641 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17643 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
17644 Generate SImode target register for null target.
17645 <case IX86_BUILTIN_XGETBV>: Ditto.
17646 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
17647 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
17649 2018-05-10 Carl Love <cel@us.ibm.com>
17651 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17652 dcbtt and dcbtstt if operands[2] is 0.
17654 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17657 * config/i386/sse.md (usadv64qi): New expander.
17659 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
17661 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
17662 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
17663 -maltivec=be support.
17664 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
17665 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
17666 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17667 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
17668 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
17669 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
17670 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
17671 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
17672 altivec_vsumsws): Adjust.
17673 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
17674 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
17675 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
17677 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
17678 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
17679 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
17680 (altivec_lve<VI_char>x): Delete expand.
17681 (*altivec_lve<VI_char>x_internal): Rename to...
17682 (altivec_lve<VI_char>x): ... this.
17683 (altivec_lvxl_<mode>): Delete expand.
17684 (*altivec_lvxl_<mode>_internal): Rename to ...
17685 (altivec_lvxl_<mode>): ... this.
17686 (altivec_stvxl_<mode>): Delete expand.
17687 (*altivec_stvxl_<mode>_internal): Rename to ...
17688 (altivec_stvxl_<mode>): ... this.
17689 (altivec_stve<VI_char>x): Delete expand.
17690 (*altivec_stve<VI_char>x_internal): Rename to ...
17691 (altivec_stve<VI_char>x): ... this.
17692 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
17693 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
17694 reduc_plus_scal_<mode>): Adjust.
17695 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
17697 (rs6000_cpu_cpp_builtins): Adjust.
17698 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
17699 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
17700 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
17701 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
17702 -maltivec=be support.
17703 (rs6000_split_vec_extract_var): Adjust.
17704 (rs6000_split_v4si_init): Adjust.
17705 (swap_selector_for_mode): Delete.
17706 (altivec_expand_lvx_be, altivec_expand_stvx_be,
17707 altivec_expand_stvex_be): Delete.
17708 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
17709 -maltivec=be support.
17710 (rs6000_gimple_fold_builtin): Ditto.
17711 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
17713 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
17714 (TARGET_DIRECT_MOVE_64BIT): Adjust.
17715 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
17716 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
17717 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
17718 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
17719 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
17720 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
17721 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
17722 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
17723 anonymous split): Adjust.
17724 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
17725 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
17727 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
17729 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
17730 when --with-gxx-include-dir is also specified.
17731 * configure: Regenerate.
17733 2018-05-09 Jim Wilson <jimw@sifive.com>
17736 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
17737 * config/riscv/t-withmultilib: New.
17738 * config/riscv/withmultilib.h: New.
17739 * doc/install.texi: Document RISC-V --with-multilib-list support.
17741 2018-05-09 Richard Biener <rguenther@suse.de>
17743 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
17745 (vect_bb_vectorization_profitable_p): Adjust. Compute
17746 actual scalar cost using the cost vector and the add_stmt_cost
17749 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17751 PR rtl-optimization/85645
17752 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
17753 in the REG_CFA_REGISTER note for LR, don't leave it empty.
17755 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17757 PR rtl-optimization/85645
17758 * shrink-wrap.c (spread_components): Return a boolean saying if
17759 anything was changed.
17760 (try_shrink_wrapping_separate): Iterate spread_components until
17761 nothing changes anymore.
17763 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17765 PR rtl-optimization/85645
17766 * regrename.c (build_def_use): Also kill the chains that include the
17767 destination of a REG_CFA_REGISTER note.
17769 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17771 PR rtl-optimization/85645
17772 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
17773 insn that has a REG_CFA_REGISTER note.
17775 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
17777 * cfgexpand.c (expand_clobber): New function.
17778 (expand_gimple_stmt_1): Use it.
17779 * tree-vect-stmts.c (vect_clobber_variable): New function,
17781 (vectorizable_simd_clone_call): ...here.
17782 (vectorizable_store): Emit a clobber either side of an
17783 IFN_STORE_LANES sequence.
17784 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
17786 2018-05-09 Tom de Vries <tom@codesourcery.com>
17789 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
17790 (define_insn "trap_if_false"): Add exit after trap.
17792 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
17794 PR rtl-optimization/85638
17795 * bb-reorder.c: Include common/common-target.h.
17796 (create_forwarder_block): New function extracted from...
17797 (fix_up_crossing_landing_pad): ...here. Rename into...
17798 (dw2_fix_up_crossing_landing_pad): ...this.
17799 (sjlj_fix_up_crossing_landing_pad): New function.
17800 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
17801 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
17802 from both partitions and exit the loop after one iteration.
17804 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17807 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17809 (Basic PowerPC Built-in Functions): The new name of the
17810 subsection previously known as "PowerPC Built-in Functions".
17811 (Basic PowerPC Built-in Functions Available on all Configurations):
17813 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17815 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17817 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17819 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17822 2018-05-08 Jim Wilson <jimw@sifive.com>
17824 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
17825 (LD_EMUL_SUFFIX): New.
17826 (LINK_SPEC): Use it.
17828 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17830 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17832 (Basic PowerPC Built-in Functions): The new name of the
17833 subsection previously known as "PowerPC Built-in Functions".
17834 (Basic PowerPC Built-in Functions Available on all Configurations):
17836 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17838 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17840 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17842 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17845 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17848 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
17849 after cmpelim optimization.
17851 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
17853 * config.gcc: Support "goldmont".
17854 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
17855 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17856 PROCESSOR_GOLDMONT.
17857 * config/i386/i386.c (m_GOLDMONT): Define.
17858 (processor_target_table): Add "goldmont".
17859 (PTA_GOLDMONT): Define.
17860 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
17861 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
17862 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
17863 (fold_builtin_cpu): Add "goldmont".
17864 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
17865 (ix86_option_override_internal): Add "goldmont".
17866 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
17867 (processor_type): Add PROCESSOR_GOLDMONT.
17868 * config/i386/i386.md: Add CPU "glm".
17869 * config/i386/glm.md: New file.
17870 * config/i386/x86-tune.def: Add m_GOLDMONT.
17871 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
17873 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17876 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
17878 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
17879 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
17880 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
17883 * config/i386/i386.c (ix86_fold_builtin): Handle
17884 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
17887 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
17888 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
17890 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
17893 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
17894 (check_arch): Likewise.
17895 (check_fpu): Return the result rather than printing it.
17896 (end arch): Fix operator precedence.
17897 (end cpu): Likewise.
17898 (END): Print the result from check_fpu.
17900 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17901 Alan Hayward <alan.hayward@arm.com>
17902 David Sherwood <david.sherwood@arm.com>
17904 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
17905 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
17906 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
17907 (*fcmuo<mode>_and): New patterns.
17909 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17911 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
17912 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
17913 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
17914 (cmp_op, sve_imm_con): New code attributes.
17915 (SVE_COND_INT_CMP, imm_con): Delete.
17916 (cmp_op): Remove above unspecs from int attribute.
17917 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
17919 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
17920 comparison-specific unspecs.
17921 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
17922 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
17923 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
17924 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
17925 (*vec_fcm<cmp_op><mode>): Rename to...
17926 (*fcm<cmp_op><mode>): ...this and adjust likewise.
17927 (*vec_fcmuo<mode>): Rename to...
17928 (*fcmuo<mode>): ...this and adjust likewise.
17929 (*pred_fcm<cmp_op><mode>): New pattern.
17930 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
17931 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
17933 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
17935 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
17936 (aarch64_emit_sve_predicated_cond): New function.
17937 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
17938 (aarch64_emit_unspec_cond_or): Replace with...
17939 (aarch64_emit_sve_or_conds): ...this new function. Use
17940 aarch64_emit_sve_ptrue_op for the individual comparisons and
17941 aarch64_emit_binop to OR them together.
17942 (aarch64_emit_inverted_unspec_cond): Replace with...
17943 (aarch64_emit_sve_inverted_cond): ...this new function. Use
17944 aarch64_emit_sve_ptrue_op for the comparison and
17945 aarch64_emit_unop to invert the result.
17946 (aarch64_expand_sve_vec_cmp_float): Update after the above
17947 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
17949 2018-05-07 Nathan Sidwell <nathan@acm.org>
17951 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
17952 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
17953 (Backwards Compatibility): Likewise.
17955 2018-05-07 Luis Machado <luis.machado@linaro.org>
17959 2018-05-07 Luis Machado <luis.machado@linaro.org>
17961 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17962 <prefetch_dynamic_strides>: New const bool field.
17963 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17964 prefetch_dynamic_strides.
17965 (exynosm1_prefetch_tune): Likewise.
17966 (thunderxt88_prefetch_tune): Likewise.
17967 (thunderx_prefetch_tune): Likewise.
17968 (thunderx2t99_prefetch_tune): Likewise.
17969 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17971 (aarch64_override_options_internal): Update to set
17972 PARAM_PREFETCH_DYNAMIC_STRIDES.
17973 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17974 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17975 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17976 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17977 prefetch-dynamic-strides setting.
17979 2018-05-07 Luis Machado <luis.machado@linaro.org>
17981 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17982 <minimum_stride>: New const int field.
17983 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17984 minimum_stride field.
17985 (exynosm1_prefetch_tune): Likewise.
17986 (thunderxt88_prefetch_tune): Likewise.
17987 (thunderx_prefetch_tune): Likewise.
17988 (thunderx2t99_prefetch_tune): Likewise.
17989 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17990 (aarch64_override_options_internal): Update to set
17991 PARAM_PREFETCH_MINIMUM_STRIDE.
17992 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17993 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17994 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17995 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17996 stride is constant and is below the minimum stride threshold.
17998 2018-05-07 Luis Machado <luis.machado@linaro.org>
18000 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
18003 2018-05-07 Luis Machado <luis.machado@linaro.org>
18005 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18006 <prefetch_dynamic_strides>: New const bool field.
18007 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18008 prefetch_dynamic_strides.
18009 (exynosm1_prefetch_tune): Likewise.
18010 (thunderxt88_prefetch_tune): Likewise.
18011 (thunderx_prefetch_tune): Likewise.
18012 (thunderx2t99_prefetch_tune): Likewise.
18013 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
18015 (aarch64_override_options_internal): Update to set
18016 PARAM_PREFETCH_DYNAMIC_STRIDES.
18017 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
18018 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
18019 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
18020 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
18021 prefetch-dynamic-strides setting.
18023 2018-05-07 Luis Machado <luis.machado@linaro.org>
18025 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18026 <minimum_stride>: New const int field.
18027 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18028 minimum_stride field.
18029 (exynosm1_prefetch_tune): Likewise.
18030 (thunderxt88_prefetch_tune): Likewise.
18031 (thunderx_prefetch_tune): Likewise.
18032 (thunderx2t99_prefetch_tune): Likewise.
18033 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
18034 (aarch64_override_options_internal): Update to set
18035 PARAM_PREFETCH_MINIMUM_STRIDE.
18036 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
18037 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
18038 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
18039 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
18040 stride is constant and is below the minimum stride threshold.
18042 2018-05-06 Jakub Jelinek <jakub@redhat.com>
18045 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
18046 the type is addressable. Don't force op into register if it has
18049 2018-05-05 Roland McGrath <mcgrathr@google.com>
18052 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
18053 any section for which we don't know a specific type it should have,
18054 regardless of name. Previously this was done only for the exact
18055 names ".init_array", ".fini_array", and ".preinit_array".
18056 (default_elf_asm_named_section): Add comment about
18057 relationship with default_section_type_flags and SECTION_NOTYPE.
18058 (get_section): Don't consider it a type conflict if one side has
18059 SECTION_NOTYPE and the other doesn't, as long as neither has the
18060 SECTION_BSS et al used in the default_section_type_flags logic.
18062 2018-05-05 Tom de Vries <tom@codesourcery.com>
18065 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
18066 (workaround_barsyncs): New function.
18067 (nvptx_reorg): Use workaround_barsyncs.
18068 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
18069 (define_expand "nvptx_membar_cta"): New define_expand.
18070 (define_insn "*nvptx_membar_cta"): New insn.
18072 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18074 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
18075 To improve optimization opportunities.
18076 * builtin-types.def: The new needed builtin types for the above.
18078 2018-05-04 Richard Biener <rguenther@suse.de>
18080 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
18081 * gimple-ssa-store-merging.c
18082 (imm_store_chain_info::output_merged_store): Remove redundant create,
18083 release split_store vector contents on failure.
18084 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
18085 scalar stmt vector on cache hit.
18087 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
18089 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18091 * config.gcc (powerpc-xilinx-eabi*): Remove.
18092 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
18094 (fusion_addis_mem_combo_load): Ditto.
18095 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
18097 (rs6000_cpu_cpp_builtins): Ditto.
18098 * config/rs6000/rs6000-linux.c
18099 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
18100 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
18101 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
18103 (rs6000_setup_reg_addr_masks): Ditto.
18104 (rs6000_init_hard_regno_mode_ok): Ditto.
18105 (rs6000_option_override_internal): Ditto.
18106 (legitimate_lo_sum_address_p): Ditto.
18107 (rs6000_legitimize_address): Ditto.
18108 (rs6000_legitimize_reload_address): Ditto.
18109 (rs6000_legitimate_address_p): Ditto.
18110 (abi_v4_pass_in_fpr): Ditto.
18111 (setup_incoming_varargs): Ditto.
18112 (rs6000_gimplify_va_arg): Ditto.
18113 (rs6000_split_multireg_move): Ditto.
18114 (rs6000_savres_strategy): Ditto.
18115 (rs6000_emit_prologue_components): Ditto.
18116 (rs6000_emit_epilogue_components): Ditto.
18117 (rs6000_emit_prologue): Ditto.
18118 (rs6000_emit_epilogue): Ditto.
18119 (rs6000_elf_file_end): Ditto.
18120 (rs6000_function_value): Ditto.
18121 (rs6000_libcall_value): Ditto.
18122 * config/rs6000/rs6000.h: Ditto.
18123 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
18124 (TARGET_MINMAX): ... this. New.
18125 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
18126 * config/rs6000/rs6000.md: Remove Xilinx FP support.
18127 (*movsi_internal1_single): Delete.
18128 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
18129 mfpu=, mxilinx-fpu): Delete.
18130 * config/rs6000/singlefp.h: Delete.
18131 * config/rs6000/sysv4.h: Remove Xilinx FP support.
18132 * config/rs6000/t-rs6000: Ditto.
18133 * config/rs6000/t-xilinx: Delete.
18134 * config/rs6000/titan.md: Adjust for fp_type removal.
18135 * config/rs6000/vsx.md: Remove Xilinx FP support.
18136 (VStype_simple): Delete.
18137 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
18138 * config/rs6000/xfpu.h: Delete.
18139 * config/rs6000/xfpu.md: Delete.
18140 * config/rs6000/xilinx.h: Delete.
18141 * config/rs6000/xilinx.opt: Delete.
18142 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
18143 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
18145 2018-05-04 Tom de Vries <tom@codesourcery.com>
18148 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
18151 2018-05-04 Richard Biener <rguenther@suse.de>
18153 PR middle-end/85627
18154 * tree-complex.c (update_complex_assignment): We are always in SSA form.
18155 (expand_complex_div_wide): Likewise.
18156 (expand_complex_operations_1): Likewise.
18157 (expand_complex_libcall): Preserve EH info of the original stmt.
18158 (tree_lower_complex): Handle removed blocks.
18159 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
18160 on complex multiplication and division libcall builtins.
18162 2018-05-04 Richard Biener <rguenther@suse.de>
18164 PR middle-end/85574
18165 * fold-const.c (negate_expr_p): Restrict negation of operand
18166 zero of a division to when we know that can happen without
18168 (fold_negate_expr_1): Likewise.
18170 2018-05-04 Jakub Jelinek <jakub@redhat.com>
18173 * real.h (real_nextafter): Declare.
18174 * real.c (real_nextafter): New function.
18175 * fold-const-call.c (fold_const_nextafter): New function.
18176 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
18177 CASE_CFN_NEXTTOWARD.
18178 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
18179 even when arg1_mode is different from arg0_mode.
18181 2018-05-03 Nathan Sidwell <nathan@acm.org>
18183 * doc/extend.texi (Deprecated Features): Remove
18184 -ffriend-injection.
18185 (Backwards Compatibility): Likewise.
18186 * doc/invoke.texi (C++ Language Options): Likewise.
18187 (C++ Dialect Options): Likewise.
18189 2018-05-03 Jakub Jelinek <jakub@redhat.com>
18192 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
18193 _mm512_mask_mullox_epi64): New intrinsics.
18195 2018-05-03 Tom de Vries <tom@codesourcery.com>
18198 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18199 dump files): Add offload-tree.
18201 2018-05-03 Richard Biener <rguenther@suse.de>
18203 PR tree-optimization/85615
18204 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
18205 to loops not nested in BBs loop father to avoid creating multi-entry
18208 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18210 PR tree-optimization/70291
18211 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
18212 arguments. Change return type to tree. Emit libcall as a new
18213 statement rather than replacing existing one when inplace_p is true.
18214 (expand_complex_multiplication_components): New function.
18215 (expand_complex_multiplication): Expand floating-point complex
18216 multiplication using the above.
18217 (expand_complex_division): Rename inner_type parameter to type.
18218 Update expand_complex_libcall call-site.
18219 (expand_complex_operations_1): Update expand_complex_multiplication
18220 and expand_complex_division call-sites.
18222 2018-05-02 Jakub Jelinek <jakub@redhat.com>
18225 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
18226 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
18227 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
18228 the highest significant bit of the shift count mask is clear. In
18229 check whether and[sq]i3 is needed verify that all significant bits
18230 of the shift count other than the highest are set.
18232 2018-05-02 Tom de Vries <tom@codesourcery.com>
18235 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
18236 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
18237 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
18238 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
18239 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
18240 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
18241 __builtin_goacc_parlevel_size.
18243 2018-05-02 Richard Biener <rguenther@suse.de>
18245 PR tree-optimization/85597
18246 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
18247 do not use split vect_get_vec_defs call but call vect_get_slp_defs
18250 2018-05-02 Tom de Vries <tom@codesourcery.com>
18253 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18254 dump files): Add ltrans-tree.
18256 2018-05-02 Tom de Vries <tom@codesourcery.com>
18259 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18260 dump files): Add wpa-ipa.
18262 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
18264 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
18265 powerpc*-*-linux*paired* target.
18266 * config/rs6000/750cl.h: Delete.
18267 * config/rs6000/paired.h: Delete.
18268 * config/rs6000/paired.md: Delete.
18269 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
18271 * config/rs6000/rs6000-builtin.def: Remove paired float support.
18272 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18273 comment. Remove paired float support.
18274 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
18275 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
18277 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
18278 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
18280 * config/rs6000/rs6000.c: Remove paired float support.
18281 (paired_expand_vector_init, paired_expand_vector_move,
18282 paired_emit_vector_compare, paired_emit_vector_cond_expr,
18283 (paired_expand_lv_builtin, paired_expand_stv_builtin,
18284 paired_expand_builtin, paired_expand_predicate_builtin,
18285 paired_init_builtins): Delete.
18286 * config/rs6000/rs6000.h: Remove paired float support.
18287 * config/rs6000/rs6000.md: Remove paired float support.
18288 (move_from_CR_ov_bit): Delete.
18289 * config/rs6000/rs6000.opt (mpaired): Delete.
18290 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
18291 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
18293 2018-05-02 Richard Biener <rguenther@suse.de>
18295 PR middle-end/85567
18296 * gimplify.c (gimplify_save_expr): When in SSA form allow
18297 SAVE_EXPRs to compute to SSA vars.
18299 2018-05-02 Jakub Jelinek <jakub@redhat.com>
18302 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
18303 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
18304 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
18305 clobber operands[2], instead use a new pseudo. Formatting fixes.
18307 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
18309 PR tree-optimization/85586
18310 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
18311 exit early for statements in the same group if the accesses are
18314 2018-05-02 Tom de Vries <tom@codesourcery.com>
18317 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
18320 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
18322 PR tree-optimization/85143
18323 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
18325 2018-05-01 Tom de Vries <tom@codesourcery.com>
18328 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
18329 not found" error message.
18331 2018-05-01 Tom de Vries <tom@codesourcery.com>
18334 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
18335 * vec.c (test_ordered_remove_if): New function.
18336 (vec_c_tests): Call test_ordered_remove_if.
18337 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
18338 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
18339 * tree-vect-patterns.c (vect_pattern_recog_1): Use
18340 VEC_ORDERED_REMOVE_IF.
18342 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18344 PR tree-optimization/82665
18345 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
18346 pointer subtraction where arguments come from a memchr call.
18348 2018-05-01 Jakub Jelinek <jakub@redhat.com>
18350 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
18351 --push-state --as-needed and --pop-state instead of --as-needed and
18352 --no-as-needed if ld supports it.
18353 * configure: Regenerated.
18356 * doc/install.texi2html: Replace _002d with - and _002a with * in
18357 generated html files using sed.
18359 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18362 * gcc-rich-location.c (blank_line_before_p): New function.
18363 (use_new_line): New function.
18364 (gcc_rich_location::add_fixit_insert_formatted): New function.
18365 * gcc-rich-location.h
18366 (gcc_rich_location::add_fixit_insert_formatted): New function.
18368 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18370 * selftest.c (assert_streq): Rename "expected" and "actual" to
18371 "val1" and "val2". Extend NULL-handling to cover both inputs
18372 symmetrically, while still requiring both to be non-NULL for a pass.
18373 * selftest.h (assert_streq): Rename "expected" and "actual" to
18375 (ASSERT_EQ): Likewise.
18376 (ASSERT_EQ_AT): Likewise.
18377 (ASSERT_KNOWN_EQ): Likewise.
18378 (ASSERT_KNOWN_EQ_AT): Likewise.
18379 (ASSERT_NE): Likewise.
18380 (ASSERT_MAYBE_NE): Likewise.
18381 (ASSERT_MAYBE_NE_AT): Likewise.
18382 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
18383 the assertion to pass.
18384 (ASSERT_STREQ_AT): Likewise.
18386 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
18388 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
18389 interaction with -pie.
18391 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18393 * selftest.h: Fix alphabetization of per-source-file selftest
18396 2018-04-30 Jason Merrill <jason@redhat.com>
18398 PR c++/61982 - dead stores to destroyed objects.
18399 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
18402 2018-04-30 Jason Merrill <jason@redhat.com>
18404 * tree.c (build_clobber): New.
18405 * tree.h: Declare it.
18406 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
18408 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18410 * diagnostic-show-locus.c (layout::layout): Update for
18411 location_get_source_line returning a char_span.
18412 (struct char_span): Move to input.h.
18413 (struct correction): Update for fields in char_span becoming
18415 (struct source_line): Update for location_get_source_line
18416 returning a char_span.
18417 (layout::print_line): Likewise.
18418 * edit-context.c (edited_file::print_content): Likewise.
18419 (edited_file::print_diff_hunk): Likewise.
18420 (edited_file::print_run_of_changed_lines): Likewise.
18421 (edited_file::get_num_lines): Likewise.
18422 (edited_line::edited_line): Likewise.
18423 * final.c (asm_show_source): Likewise.
18424 * input.c (location_get_source_line): Convert return type
18425 from const char * to char_span, losing the final "line_len"
18427 (dump_location_info): Update for the above.
18428 (get_substring_ranges_for_loc): Likewise. Use a char_span
18429 when handling the literal within the line.
18430 (test_reading_source_line): Update for location_get_source_line
18431 returning a char_span.
18432 * input.h (class char_span): Move here from
18433 diagnostic-show-locus.c, converting from a struct to a class.
18434 Make data members private.
18435 (char_span::operator bool): New.
18436 (char_span::length): New.
18437 (char_span::get_buffer): New.
18438 (char_span::operator[]): New.
18439 (char_span::subspan): Make const.
18440 (char_span::xstrdup): New.
18441 (location_get_source_line): Convert return type from const char *
18442 to char_span, losing the final "line_size" param.
18444 2018-04-30 Jan Hubicka <jh@suse.cz>
18446 * lto-wrapper.c (ltrans_priorities): New static var.
18447 (cmp_priority): New.
18448 (run_gcc): Read priorities and if doing parallel build order
18449 the Makefile by them.
18451 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18453 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
18455 2018-04-30 Richard Biener <rguenther@suse.de>
18457 * tree-cfg.c (verify_address): Remove base argument, add
18458 flag whether to check TREE_ADDRESSABLE and do that.
18459 (verify_expr): Remove.
18460 (verify_types_in_gimple_reference): Add pieces from verify_expr.
18461 (verify_gimple_assign_single): Likewise.
18462 (verify_gimple_switch): Likewise.
18463 (verify_expr_location_1): Dereference tp once. Add (disabled)
18464 piece from verify_expr.
18465 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
18467 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18469 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
18471 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18473 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
18474 (small_data_pattern): Likewise.
18475 (arc_rewrite_small_data): Likewise.
18476 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
18477 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
18478 (get_symbol_alignment): New function.
18479 (legitimate_small_data_address_p): Likewise.
18480 (legitimate_scaled_address): Update, call
18481 legitimate_small_data_address_p.
18482 (output_sdata): New static variable.
18483 (arc_print_operand): Update how we handle small data operands.
18484 (arc_print_operand_address): Likewise.
18485 (arc_legitimate_address_p): Update, use
18486 legitimate_small_data_address_p.
18487 (arc_rewrite_small_data_p): Remove.
18488 (arc_rewrite_small_data_1): Likewise.
18489 (arc_rewrite_small_data): Likewise.
18490 (small_data_pattern): Likewise.
18491 (compact_sda_memory_operand): Update to use
18492 legitimate_small_data_address_p and get_symbol_alignment.
18493 (prepare_move_operands): Don't rewite sdata pattern.
18494 (prepare_extend_operands): Remove.
18495 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
18497 (zero_extendqisi2): Likewise.
18498 (zero_extendhisi2): Likewise.
18499 (extendqihi2): Likewise.
18500 (extendqisi2): Likewise.
18501 (extendhisi2): Likewise.
18502 (addsi3): Likewise.
18503 (subsi3): Likewise.
18504 (andsi3): Likewise.
18505 * config/arc/constraints.md (Usd): Change it to memory constraint.
18507 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18509 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
18510 as source of std instructions.
18511 * config/arc/arc.md (movsi_insn): Update pattern predicate to
18512 allow 6-bit constants as source for store instructions.
18513 (movdi_insn): Update instruction pattern to allow 6-bit constants
18514 as source for store instructions.
18516 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
18518 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
18520 2018-04-30 Nathan Sidwell <nathan@acm.org>
18521 Sandra Loosemore <sandra@codesourcery.com>
18523 * dumpfile.c (dump_open): Allow '-' for stdout.
18524 * doc/invoke.texi (Developer Options): Document dump filename
18525 determination early. Document stdin/stdout selection.
18527 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
18529 Microblaze Target: PIC data text relative
18531 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
18532 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
18534 * config/microblaze/microblaze.h (microblaze_constant_address_p):
18535 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
18536 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
18537 New addressing mode for data-text relative position indepenedent code.
18538 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
18539 'ADDRESS_SYMBOLIC_TXT_REL'.
18540 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
18541 (microblaze_legitimate_pic_operand): Exclude function calls from
18542 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
18543 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
18545 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18546 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18547 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
18548 for 'address + offset'.
18549 (microblaze_expand_prologue): Add new function prologue call for
18551 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
18552 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
18553 table in case of TARGET_PIC_DATA_TEXT_REL.
18554 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
18555 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
18556 Add new macros 'UNSPEC_TEXT',
18557 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
18558 + exclude function calls from 'UNSPEC_PLT' in case of data text
18560 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18561 new target hook for generating address diff vector tables in case of
18563 * doc/tm.texi : Regenerate.
18564 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
18565 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
18566 of addr diff vector generation.
18567 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18568 target hook definition.
18569 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
18570 Add default function for generate_pic_addr_diff_vec -> flag_pic.
18571 * doc/invoke.texi (Add new pic option): Add new microblaze pic
18572 option for data text relative.
18574 2018-04-30 Richard Biener <rguenther@suse.de>
18576 * tree-chrec.h (evolution_function_is_constant_p): Remove
18578 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
18580 2018-04-30 Richard Biener <rguenther@suse.de>
18583 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
18585 2018-04-30 Richard Biener <rguenther@suse.de>
18587 PR tree-optimization/28364
18588 PR tree-optimization/85275
18589 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
18590 copying first exit test.
18592 2018-04-28 Mark Wielaard <mark@klomp.org>
18594 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
18595 dwarf_version >= 5.
18596 (dwarf_AT): Handle DW_AT_addr_base.
18597 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
18599 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
18602 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
18603 (*ashl<dwi>3_doubleword_mask_1): Ditto.
18604 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
18605 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
18607 2018-04-28 Richard Biener <rguenther@suse.de>
18609 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
18610 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
18611 to reflect use. Only add interesting stmts.
18613 2018-04-27 Martin Jambor <mjambor@suse.cz>
18616 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
18617 the jump function allows for passing through aggregate values.
18619 2018-04-27 David Malcolm <dmalcolm@redhat.com>
18621 * input.h (in_system_header_at): Convert from macro to inline
18623 (from_macro_expansion_at): Likewise.
18624 (from_macro_definition_at): Likewise.
18626 2018-04-27 Jeff Law <law@redhat.com>
18628 * config.gcc: Mark tile* targets as deprecated/obsolete.
18630 2018-04-27 Richard Biener <rguenther@suse.de>
18632 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
18635 2018-04-27 Richard Biener <rguenther@suse.de>
18637 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
18639 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
18641 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
18642 with Yd constraint. Set "preferred_for_speed" attribute from
18643 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
18644 with Yd constraint.
18645 (*movdi_internal): Ditto.
18646 (movti_interunit splitters): Remove
18647 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
18648 (movdi_interunit splitters): Ditto.
18649 * config/i386/constraints.md (Ye): Remove.
18650 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
18652 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18655 * config/aarch64/constraints.md (Usg): Limit to 31.
18656 (Usj): Limit to 63.
18658 2018-04-27 Jakub Jelinek <jakub@redhat.com>
18660 PR tree-optimization/85529
18661 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
18662 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
18663 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
18664 zero extension or masking of the MSB bit.
18665 (optimize_range_tests): Add FIRST_BB argument, pass it through
18666 to optimize_range_tests_var_bound.
18667 (maybe_optimize_range_tests, reassociate_bb): Adjust
18668 optimize_range_tests callers.
18670 2018-04-26 Richard Biener <rguenther@suse.de>
18671 Jakub Jelinek <jakub@redhat.com>
18673 * cgraph.h (symbol_table): Just declare debug method here.
18674 * symtab.c (symbol_table::debug): Define.
18676 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
18678 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
18680 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
18682 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
18683 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
18684 (*movdi_internal): Substitute Yi and Yj constraint with x
18685 and Ym and Yn constraint with y constraint. Update "isa"
18686 attribute and set "preferred_for_speed" attribute from
18687 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
18688 (*movsi_internal): Ditto.
18689 (*movdf_internal): Ditto.
18690 (*movsf_internal): Ditto.
18691 (*zero_extendsidi2): Ditto.
18692 * config/i386/sse.md (vec_set<mode>_0): Ditto.
18693 (sse2_loadld): Ditto.
18694 (*vec_extract<ssevecmodelower>_0): Ditto.
18695 (*vec_extractv4si_0_zext_sse4): Ditto.
18696 (vec_concatv2di): Ditto.
18697 (*vec_dup<mode>): Ditto.
18698 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
18699 * config/i386/constraints.md (Yi): Remove.
18704 2018-04-26 Nathan Sidwell <nathan@acm.org>
18706 * dumpfile.c (dump_open): New.
18707 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
18708 (dump_finish): Detect stdio/stderr by value not name.
18710 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
18712 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
18714 2018-04-26 Tom de Vries <tom@codesourcery.com>
18717 * config/nvptx/nvptx.c (verify_neutering_jumps)
18718 (verify_neutering_labels): New function
18719 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
18721 2018-04-26 Tom de Vries <tom@codesourcery.com>
18724 * config/nvptx/nvptx.c (needs_neutering_p): New function.
18725 (nvptx_single): Use needs_neutering_p to skip over insns that do not
18728 2018-04-26 Richard Biener <rguenther@suse.de>
18729 Tom de Vries <tom@codesourcery.com>
18732 * lto-streamer-out.c (output_function): Fixup loops if required to match
18733 discovery done in the reader.
18735 2018-04-26 Richard Biener <rguenther@suse.de>
18737 PR tree-optimization/85116
18738 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
18739 have a loop exit from the single latch predecessor. Remove
18740 case of header with just condition.
18741 (ch_base::copy_headers): Exclude infinite loops from any
18743 (pass_ch::execute): Record exits.
18745 2018-04-26 Richard Biener <rguenther@suse.de>
18747 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
18748 prologue cost vector and pass it to vect_get_load_cost.
18749 (vect_get_peeling_costs_all_drs): Likewise.
18750 (vect_peeling_hash_get_lowest_cost): Likewise.
18751 (vect_enhance_data_refs_alignment): Likewise.
18753 2018-04-26 Richard Biener <rguenther@suse.de>
18755 PR middle-end/85450
18756 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
18757 checking of integer<->pointer conversions.
18758 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
18759 sign-/zero-extending pointer types.
18760 (expand_omp_for_static_chunk): Likewise.
18762 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
18763 Jean Lee <xiaoyur347@gmail.com>
18765 * config/mips/mips.c (mips_asan_shadow_offset): New function.
18766 (TARGET_ASAN_SHADOW_OFFSET): Define.
18767 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
18768 true for -fsanitize=address.
18770 2018-04-25 Mark Wielaard <mark@klomp.org>
18772 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
18775 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18777 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
18778 than "alu", remove explicit "memory" and "imm_disp" attributes.
18779 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
18781 PR middle-end/85414
18782 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
18783 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
18784 gen_lowpart_no_emit.
18786 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
18789 * config/i386/i386.c (ix86_expand_builtin): Change memory
18790 operand to XI, extend p0 to Pmode.
18791 * config/i386/i386.md: Change unspec volatile and operand
18792 1 mode to XI, change operand 0 mode to P.
18794 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18796 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
18797 GET_MODE_MASK before any checking.
18798 (nds32_can_use_bset_p): Likewise.
18799 (nds32_can_use_btgl_p): Likewise.
18801 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18803 * config/nds32/nds32-doubleword.md: New define_split pattern for
18804 illegal register number.
18806 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18808 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
18810 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18812 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
18814 2018-04-25 Richard Biener <rguenther@suse.de>
18816 * lto-streamer.h (LTO_major_version): Bump to 8.
18818 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18820 * BASE-VER: Set to 9.0.0.
18822 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
18824 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
18825 in __abskf2 and __powikf2.
18827 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18830 * config/aarch64/constraints.md (Usg, Usj): New constraints.
18831 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
18832 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
18833 Use the above on operand 2. Reindent.
18834 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
18836 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18839 * common/config/i386/i386-common.c (ix86_handle_option): Don't
18841 * config/i386/i386.opt (mcet): Removed.
18842 * doc/install.texi: Remove -mcet documentation.
18843 * doc/invoke.texi: Likewise.
18845 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18848 * doc/install.texi: Remove -mcet from bootstrap-cet.
18850 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18853 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
18854 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
18858 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
18859 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
18860 containing a CONST_VECTOR.
18862 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
18864 * doc/install.texi: Update newlib dependency for nvptx.
18866 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18869 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
18870 instead of INTVAL when shifting x left.
18872 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
18874 PR tree-optimization/85478
18875 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
18876 vect_grouped_store_supported for single element vectors.
18878 2018-04-24 Richard Biener <rguenther@suse.de>
18881 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
18882 load cost increase to the case of non-constant step.
18884 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18887 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
18888 destination if any_malformed_asm.
18890 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
18892 PR middle-end/85496
18893 * expr.c (store_field): In the bitfield case, if the value comes from
18894 a function call and is returned in registers by means of a PARALLEL,
18895 do not change the mode of the temporary unless BLKmode and VOIDmode.
18897 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
18899 PR rtl-optimization/85423
18900 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
18901 dependencies to debug insns when the previous insn is non-debug.
18903 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18905 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
18906 enums into a single definition.
18907 (fls): Fix predicates and printing.
18910 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18912 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
18913 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
18914 and short u6 immediate.
18915 (check_if_valid_sleep_operand): Remove.
18916 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
18918 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18920 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
18921 flag_always_save_lp condition.
18922 * config/nds32/nds32.opt (malways-save-lp): New option.
18924 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18926 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
18927 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
18928 * config/nds32/nds32.h
18929 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
18930 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
18932 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18934 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
18935 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
18937 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18938 Chung-Ju Wu <jasonwucj@gmail.com>
18940 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
18942 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
18943 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
18945 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18947 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
18949 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18951 * config/nds32/nds32-protos.h (nds32_data_alignment,
18952 nds32_local_alignment): Declare.
18953 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
18954 nds32_local_alignment): New functions.
18955 (TARGET_CONSTANT_ALIGNMENT): Define.
18956 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
18958 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18960 * config/nds32/nds32.c
18961 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
18962 (TARGET_MODES_TIEABLE_P): Likewise.
18964 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18966 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
18967 level Ofast and Og.
18969 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
18970 Chung-Ju Wu <jasonwucj@gmail.com>
18972 * config/nds32/constants.md (unspec_volatile_element): Add enum values
18973 for unaligned access.
18974 * config/nds32/nds32-intrinsic.c: Implementation of expanding
18976 * config/nds32/nds32-intrinsic.md: Likewise.
18977 * config/nds32/nds32_intrinsic.h: Likewise.
18978 * config/nds32/nds32.h (nds32_builtins): Likewise.
18979 * config/nds32/nds32.opt (munaligned-access): New option.
18980 * config/nds32/nds32.c (nds32_asm_file_start): Display
18981 flag_unaligned_access status.
18983 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18985 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
18986 -mno-relax is present.
18987 * config/riscv/linux.h (LINK_SPEC): Ditto.
18989 2018-04-20 Martin Sebor <msebor@redhat.com>
18992 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
18994 (gimple_fold_builtin_stxcpy_chk): Same.
18995 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
18997 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
19000 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
19001 __powikf2 when long double is IEEE 128-bit.
19003 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
19005 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
19006 step to make sure stack always aligned.
19008 2018-04-20 Carl Love <cel@us.ibm.com>
19011 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
19012 size check for arg0.
19014 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
19015 Tom de Vries <tom@codesourcery.com>
19018 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
19019 Emit insns for calls too.
19020 (nvptx_find_par): Always look for worker-level predecessor insn.
19021 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
19023 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
19024 (nvptx_process_pars): Propagate frames for calls.
19026 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
19029 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
19031 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
19032 (ix86_handle_option): Don't handle OPT_mibt.
19033 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
19035 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
19037 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
19038 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
19039 (ix86_target_macros): Define __CET__ with flag_cf_protection
19040 for -fcf-protection.
19041 * config/i386/i386.c (isa2_opts): Remove -mibt.
19042 * config/i386/i386.h (TARGET_IBT): Removed.
19043 (TARGET_IBT_P): Likewise.
19044 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
19045 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
19046 * config/i386/i386.opt (mcet): Update help message.
19047 (mshstk): Likewise.
19049 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
19050 -mcet as an alias for -mshstk.
19052 2018-04-20 Richard Biener <rguenther@suse.de>
19054 PR middle-end/85475
19055 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
19056 complexity by forcing a single use of the multiply operand.
19058 2018-04-20 Martin Jambor <mjambor@suse.cz>
19061 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
19062 recursion dependency to only apply to non-clones.
19064 2018-04-20 Martin Jambor <mjambor@suse.cz>
19067 * ipa-cp.c (create_specialized_node): Check that clones of
19068 self-recursive edges exist during IPA-CP.
19070 2018-04-19 Toon Moene <toon@moene.org>
19072 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
19075 2018-04-19 Jakub Jelinek <jakub@redhat.com>
19077 PR tree-optimization/85467
19078 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
19079 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
19080 VECTOR_CST element to type.
19082 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
19085 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
19086 * config/i386/i386.md (builtin_setjmp_setup): Removed.
19087 (builtin_longjmp): Likewise.
19088 (save_stack_nonlocal): New pattern.
19089 (restore_stack_nonlocal): Likewise.
19091 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
19094 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
19095 Replace ASM_OUTPUT_LABEL with fprintf.
19097 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
19100 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
19101 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
19102 * config/i386/i386-c.c (ix86_target_macros_internal): Also
19103 define __IBT__ and __SHSTK__ for -fcf-protection.
19104 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
19106 (ix86_trampoline_init): Likewise.
19107 (x86_output_mi_thunk): Likewise.
19108 (ix86_notrack_prefixed_insn_p): Likewise.
19109 (ix86_option_override_internal): Don't disallow -fcf-protection.
19110 * config/i386/i386.md (rdssp<mode>): Also enable for
19112 (incssp<mode>): Likewise.
19113 (nop_endbr): Likewise.
19114 * config/i386/i386.opt (mcet): Change help message to built-in
19117 (mshstk): Likewise.
19118 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
19119 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
19120 enable CET built-in functions.
19122 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
19124 * common/config/i386/i386-common.c
19125 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
19126 OPTION_MASK_ISA_MOVDIRI_UNSET,
19127 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
19128 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
19129 * config.gcc (movdirintrin.h): New header.
19130 * config/i386/cpuid.h (bit_MOVDIRI,
19131 bit_MOVDIR64B): New bits.
19132 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
19134 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
19135 (VOID, PVOID, PCVOID)): New function types.
19136 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
19137 __builtin_ia32_directstoreu_u64,
19138 __builtin_ia32_movdir64b): New builtins.
19139 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
19140 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
19142 (ix86_valid_target_attribute_inner_p): Ditto.
19143 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
19144 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
19145 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
19146 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
19147 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
19148 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
19149 (movdiri<mode>, movdir64b_<mode>): New.
19150 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
19151 * config/i386/immintrin.h: Include movdirintrin.h.
19152 * config/i386/movdirintrin.h: New file.
19153 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
19155 2018-04-19 Richard Biener <rguenther@suse.de>
19157 PR middle-end/85455
19158 * cfg.c (clear_bb_flags): When loop state says we have
19159 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
19161 2018-04-19 Richard Biener <rguenther@suse.de>
19163 PR tree-optimization/84737
19164 * tree-vect-data-refs.c (vect_copy_ref_info): New function
19165 copying restrict info.
19166 (vect_setup_realignment): Use it.
19167 * tree-vectorizer.h (vect_copy_ref_info): Declare.
19168 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
19169 the first DR to all generated stores.
19170 (vectorizable_load): Likewise for loads.
19172 2018-04-19 Jakub Jelinek <jakub@redhat.com>
19174 PR tree-optimization/85446
19175 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
19176 the integral and pointer types to have the same precision.
19178 * doc/install.texi: Document --disable-cet being the default and
19181 2018-04-18 Martin Liska <mliska@suse.cz>
19183 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
19186 2018-04-18 Martin Liska <mliska@suse.cz>
19189 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
19192 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
19193 arguments if they are comparable.
19195 2018-04-18 Martin Liska <mliska@suse.cz>
19198 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
19201 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
19204 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
19207 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
19208 ENDBR after calling __morestack.
19210 2018-04-18 David Malcolm <dmalcolm@redhat.com>
19213 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
19214 by using gcc_base_ver to generate a gcc_driver_version, and use
19215 it when generating GCC_DRIVER_NAME.
19216 * configure: Regenerate.
19218 2018-04-18 Jakub Jelinek <jakub@redhat.com>
19221 * config.gcc: Obsolete powerpc*-*-*spe*.
19223 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19226 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
19227 (stabstr_D): Change type of unum from unsigned int to
19228 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
19231 2018-04-17 Jim Wilson <jimw@sifive.com>
19234 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
19235 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
19236 Set arg_pointer_offset after using pretend_args_size.
19238 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19240 PR rtl-optimization/85431
19241 * dse.c (record_store): Ignore zero width stores.
19244 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
19245 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
19246 __builtin_stack_restore rather than after it.
19247 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
19248 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
19249 argument instead of virtual_dynamic_stack_rtx.
19251 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
19253 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
19255 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19256 Add note to error message to explain internal mapping of overloaded
19257 built-in function name to non-overloaded built-in function name.
19258 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
19261 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
19264 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
19265 where the inputs overlap with the output.
19267 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19269 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
19270 (=v, v) alternative and explicit "memory" attribute.
19271 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
19272 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19274 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
19275 "sselog1" type instead of "sselog".
19276 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
19277 "sselog". Remove explicit "memory" attribute.
19278 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
19279 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19281 (vec_extract_hi_v32hi): Merge all alternatives into one, use
19282 "sselog1" type instead of "sselog". Remove explicit "memory"
19284 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
19285 use "sselog1" type instead of "sselog". Remove explicit "memory"
19287 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
19288 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19290 (vec_extract_hi_v64qi): Merge all alternatives into one, use
19291 "sselog1" type instead of "sselog". Remove explicit "memory"
19293 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
19294 use "sselog1" type instead of "sselog". Remove explicit "memory"
19298 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
19300 PR middle-end/85414
19301 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
19304 2018-04-17 Martin Jambor <mjambor@suse.cz>
19307 * ipa-cp.c (create_specialized_node): Call
19308 expand_all_artificial_thunks if necessary.
19310 2018-04-17 Martin Liska <mliska@suse.cz>
19313 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
19314 in message, remote space in between '_G' and '('.
19316 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19319 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
19320 avx512f_vmcmp<mode>3<round_saeonly_name>,
19321 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
19322 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
19323 avx512f_rndscale<mode><round_saeonly_name>,
19324 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
19325 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
19326 Use %<iptr>2 instead of %2 for -masm=intel.
19327 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
19328 avx512f_vcvttss2usi<round_saeonly_name>,
19329 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
19331 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
19332 avx512f_vcvttsd2usi<round_saeonly_name>,
19333 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
19334 Use %q1 instead of %1 for -masm=intel.
19335 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
19336 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
19337 of %3 for -masm=intel.
19338 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
19340 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
19342 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
19344 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
19345 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
19347 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
19349 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
19350 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
19351 %g1 and one with %0 and %1.
19352 (avx512er_vmrcp28<mode><round_saeonly_name>,
19353 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
19354 %1 for -masm=intel.
19355 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
19356 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
19357 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
19358 of %0 and %{%4%} for -masm=intel.
19359 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
19360 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
19361 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
19362 order of %0 and %{%5%}%{z%} for -masm=intel.
19364 2018-04-17 Jan Hubicka <jh@suse.cz>
19367 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
19369 2018-04-17 Martin Liska <mliska@suse.cz>
19372 * multiple_target.c (create_dispatcher_calls): Set apostrophes
19373 for target_clone error message. Make default implementation
19374 clone to be a local declaration.
19375 (separate_attrs): Add new argument and check for an empty
19377 (expand_target_clones): Handle it.
19378 (ipa_target_clone): Make redirection just for target_clones
19381 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
19382 Tom de Vries <tom@codesourcery.com>
19384 PR middle-end/84955
19385 * omp-expand.c (expand_oacc_for): Add dummy false branch for
19386 tiled basic blocks without omp continue statements.
19388 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
19391 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
19392 vec_extract expression as having side effects to make sure it gets
19395 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
19398 * config/i386/i386.c (get_builtin_code_for_version): Check
19401 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
19404 * config.gcc: Support "skylake".
19405 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19407 * config/i386/i386.c (m_SKYLAKE): Define.
19408 (processor_target_table): Add "skylake".
19409 (ix86_option_override_internal): Add "skylake".
19410 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
19411 PROCESSOR_CANNONLAKE.
19412 (get_builtin_code_for_version): Fix priority for
19413 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
19414 PROCESSOR_SKYLAKE-AVX512.
19415 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
19416 (processor_type): Add PROCESSOR_SKYLAKE.
19418 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
19419 Jason Merrill <jason@redhat.com>
19422 * convert.c (convert_to_integer_1): Use direct recursion for
19423 enumeral types and types with a precision less than the number
19424 of bits in their mode.
19426 2018-04-16 Julia Koval <julia.koval@intel.com>
19429 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
19430 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
19432 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
19435 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
19436 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
19437 and -mno-direct-move.
19439 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
19442 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
19443 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
19444 Ensure negative shifts result in {0}.
19446 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
19448 PR rtl-optimization/79916
19449 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
19450 regs (if any) to define how to gnerate SD moves when LRA is in
19453 2018-04-13 Jakub Jelinek <jakub@redhat.com>
19455 PR rtl-optimization/85393
19456 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
19457 * except.c (expand_dw2_landing_pad_for_region): Make static.
19458 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
19459 a label and unconditional jump to old_bb, rather than
19460 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
19463 PR rtl-optimization/85376
19464 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
19465 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
19466 instead of a specific value.
19468 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
19469 Bin Cheng <bin.cheng@arm.com>
19471 PR tree-optimization/82965
19472 PR tree-optimization/83991
19473 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
19474 by_profile_only parameter.
19475 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
19476 information if the loop was predicted to iterate too many times.
19477 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
19479 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
19482 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
19485 2018-04-13 Martin Liska <mliska@suse.cz>
19486 Jakub Jelinek <jakub@redhat.com>
19488 PR middle-end/81657
19489 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
19490 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
19491 * builtins.c (expand_builtin_memory_copy_args): Use
19492 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
19493 handle dest_addr == pc_rtx.
19495 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19498 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
19499 asked to not generate direct moves.
19500 (fix_trunc<mode>si2_stfiwx): Similar.
19501 (fix_trunc<mode>si2_internal): Similar.
19503 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19506 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
19507 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
19508 lookup if dest in some wider mode is known to be const0_rtx and
19509 if so, record permanent equivalence for it to be ZERO_EXTEND of
19510 the narrower mode destination.
19512 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19514 * lto-streamer-out.c (output_function): Revert 259346.
19515 * omp-expand.c (expand_oacc_for): Likewise.
19517 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
19519 PR rtl-optimization/85354
19520 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
19521 * sel-sched.c (sel_global_init): ... here.
19523 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
19526 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
19527 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
19528 mode for PE-COFF targets.
19529 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
19530 (i386_pe_asm_lto_end): Likewise.
19531 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
19532 (TARGET_ASM_LTO_END): Likewise.
19533 * config/i386/winnt.c (saved_debug_info_level): New static variable.
19534 (i386_pe_asm_lto_start): New function.
19535 (i386_pe_asm_lto_end): Likewise.
19537 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19538 Richard Biener <rguenther@suse.de>
19540 PR middle-end/84955
19541 * lto-streamer-out.c (output_function): Fix CFG loop state before
19543 * omp-expand.c (expand_oacc_for): Handle calls to internal
19544 functions like regular functions.
19546 2018-04-12 Richard Biener <rguenther@suse.de>
19549 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
19550 for the early LTO debug to properly generate references to it
19551 during DIE emission. Do not re-use that for the skeleton for
19553 (dwarf2out_early_finish): Likewise.
19555 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19558 * config/i386/sse.md
19559 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
19560 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
19561 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
19562 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
19563 and output is a reg, avoid creating invalid lowpart subreg, but
19564 instead split into a 512-bit move. Don't split if not AVX512VL,
19565 input is xmm16+ reg and output is a mem.
19566 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
19567 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
19568 xmm16+ reg and output is a mem.
19570 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19572 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
19573 also for flag_dwarf2_cfi_asm.
19575 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19577 PR rtl-optimization/85342
19578 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
19579 a bool scalar var inside of the loop instead. Don't try to update
19580 recog_data.operand after failed apply_change_group.
19582 2018-04-12 Tom de Vries <tom@codesourcery.com>
19585 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
19586 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
19587 array with flexible array member as array without given dimension.
19588 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
19589 argument for undefined param to true.
19591 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
19594 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19595 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
19596 from PowerPC section.
19597 * config/rs6000/sysv4.opt (mcall-): Improve help text.
19598 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
19599 help text that is too long.
19600 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
19601 help text that is too long.
19602 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
19603 help text that is too long.
19605 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
19607 * config/alpha/alpha.md (stack_probe_internal): Rename
19608 from "probe_stack". Update all callers.
19610 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19612 PR rtl-optimization/84566
19613 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
19614 sched_macro_fuse_insns.
19616 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19619 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
19620 (compute_block_dependences): ... from here.
19622 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19624 PR tree-optimization/85331
19625 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
19626 from int to HOST_WIDE_INT.
19628 2018-04-11 Martin Jambor <mjambor@suse.cz>
19631 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
19632 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
19633 not the same as the source val.
19634 (cgraph_edge_brings_value_p): New parameter.
19635 (gather_edges_for_value): Pass destination value to
19636 cgraph_edge_brings_value_p.
19637 (perhaps_add_new_callers): Likewise.
19638 (get_info_about_necessary_edges): Likewise and exclude values brought
19639 only by self-recursive edges.
19640 (create_specialized_node): Redirect only clones of self-calling edges.
19641 (+self_recursive_pass_through_p): New function.
19642 (find_more_scalar_values_for_callers_subset): Use it.
19643 (find_aggregate_values_for_callers_subset): Likewise.
19644 (known_aggs_to_agg_replacement_list): Removed.
19645 (decide_whether_version_node): Re-calculate known constants for all
19646 remaining context clones.
19648 2018-04-11 Richard Biener <rguenther@suse.de>
19651 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
19652 from early DWARF output.
19653 (dwarf2out_early_finish): Output line info unconditionally into
19654 early DWARF and add reference to it.
19656 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19659 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
19660 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
19661 other than V2DFmode using iptr mode attribute.
19662 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
19664 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19666 PR rtl-optimization/84659
19667 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
19669 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19672 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
19674 (output_loc_list): Pass address of a dummy size variable even in the
19675 locview handling loop.
19676 (index_location_lists): Add comment on why skip_loc_list_entry can't
19679 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
19682 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
19685 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
19688 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19689 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
19690 and -mstring-compare-inline-limit.
19692 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19695 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
19696 for stack clash protection in a register whenever we need it to be in
19699 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19701 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
19702 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
19704 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19707 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
19709 (mlong-double-): Ditto.
19710 * config/rs6000/sysv4.opt (msdata=): Ditto.
19711 (mtls-size=): Ditto.
19713 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19715 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19716 erroneous entries for
19717 "vector int vec_ldl (int, long int *)", and
19718 "vector unsigned int vec_ldl (int, unsigned long int *)".
19719 Add comments and entries for
19720 "vector bool char vec_ldl (int, bool char *)",
19721 "vector bool short vec_ldl (int, bool short *)",
19722 "vector bool int vec_ldl (int, bool int *)",
19723 "vector bool long long vec_ldl (int, bool long long *)",
19724 "vector pixel vec_ldl (int, pixel *)",
19725 "vector long long vec_ldl (int, long long *)",
19726 "vector unsigned long long vec_ldl (int, unsigned long long *)".
19727 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
19728 type tree bool_long_long_type_node and correct definition of
19729 bool_V2DI_type_node to make reference to this new type tree.
19730 (rs6000_mangle_type): Replace erroneous reference to
19731 bool_long_type_node with bool_long_long_type_node.
19732 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
19733 comments to emphasize sign distinctions for char and int types and
19734 replace RS6000_BTI_bool_long constant with
19735 RS6000_BTI_bool_long_long constant. Also add comment to restrict
19736 use of RS6000_BTI_pixel.
19737 (bool_long_type_node): Remove this macro definition.
19738 (bool_long_long_type_node): New macro definition
19740 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19742 PR rtl-optimization/85300
19743 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
19744 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
19745 simplify_unary_operation fails.
19747 2018-04-10 Martin Liska <mliska@suse.cz>
19749 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
19750 cgraph_edge and ipa_ref.
19752 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19756 * config/i386/sse.md
19757 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
19758 computation of the VEC_MERGE selector from mask.
19759 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
19760 Fix decoding of the VEC_MERGE selector into mask.
19762 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19764 PR tree-optimization/85286
19765 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
19767 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19769 * final.c (final_1): Set insn_last_address as well as
19770 insn_current_address.
19772 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19775 * explow.c (emit_stack_probe): Call validize_mem on memory location
19776 before passing it to gen_probe_stack. Create address operand and
19777 legitimize it for the probe_stack_address case.
19779 2018-04-09 Jan Hubicka <jh@suse.cz>
19782 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
19783 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
19784 * tree.c (free_lang_data_in_type): Fix handling of binfos;
19786 (free_lang_data): Rebuild type inheritance graph.
19788 2018-04-09 Martin Sebor <msebor@redhat.com>
19790 * invoke.texi (-finline-small-functions): Mention other optimization
19792 (-findirect-inlining, -fpartial-inlining): Same.
19793 (-finline-functions-called-once): Same.
19794 (-freorder-blocks-and-partition): Same.
19796 2018-04-09 Jan Hubicka <jh@suse.cz>
19799 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
19800 jumps; choose last target that matches the criteria (i.e.
19801 no partition changes for non-crossing jumps).
19802 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
19803 support for redirecting crossing jumps to non-crossing.
19805 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
19807 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
19808 also for naked functions.
19810 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
19812 * config/arc/arc.md (add_shift): New pattern.
19813 (add_shift2): Likewise.
19814 (sub_shift): Likewise.
19815 (sub_shift_cmp0_noout): Likewise.
19816 (compare_si_ashiftsi): Likewise.
19817 (xbfu_cmp0_noout): New combine pattern.
19818 (xbfu_cmp0"): Likewise.
19819 (movsi_set_cc_insn): Place the predicable variant first.
19820 (commutative_binary_cmp0_noout): Remove clobber.
19821 (commutative_binary_cmp0): New pattern.
19822 (noncommutative_binary_cmp0): Likewise.
19823 (noncommutative_binary_cmp0_noout): Likewise.
19824 (noncommutative_binary_comparison_result_used): Removed.
19825 (rsub_cmp0): New pattern.
19826 (rsub_cmp0_noout): Likewise.
19827 (extzvsi): Changed, keep only meaningful variants.
19828 (SQH, SEZ): New iterators.
19829 (SQH_postfix): New mode attribute.
19830 (SEZ_prefix): New code attribute.
19831 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
19832 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
19833 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
19834 of numerical value.
19835 (noncommutative_operator): Check the availability of barrel
19838 2018-04-09 Richard Biener <rguenther@suse.de>
19840 PR tree-optimization/85284
19841 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
19842 Only use the niter constraining form of simple_iv when the exit
19843 is always executed.
19845 2018-04-09 Tom de Vries <tom@codesourcery.com>
19848 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
19849 (define_expand "*memory_barrier"): New define_expand.
19850 (define_insn "memory_barrier"): New insn.
19852 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19854 PR rtl-optimization/80463
19855 PR rtl-optimization/83972
19856 PR rtl-optimization/83480
19858 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
19859 correct producer for the insn.
19860 (tidy_control_flow): Fixup seqnos in case of debug insns.
19862 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19864 PR rtl-optimization/83913
19866 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
19867 different sched-times when merging exprs.
19869 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19871 PR rtl-optimization/83962
19873 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
19874 tidy_fallthru_edge and tidy_control_flow.
19876 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19878 PR rtl-optimization/83530
19880 * sel-sched.c (force_next_insn): New global variable.
19881 (remove_insn_for_debug): When force_next_insn is true, also leave only
19882 next insn in the ready list.
19883 (sel_sched_region): When the region wasn't scheduled, make another pass
19884 over it with force_next_insn set to 1.
19886 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19888 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
19890 * config/nds32/constants.md (unspec_volatile_element): Add enum values
19891 for interrupt control.
19892 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19893 functions for interrupt control.
19894 * config/nds32/nds32-intrinsic.md: Likewise.
19895 * config/nds32/nds32_intrinsic.h: Likewise.
19896 * config/nds32/nds32.h (nds32_builtins): Likewise.
19898 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19900 * config/nds32/nds32.c (nds32_init_machine_status,
19901 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
19902 strict_aligned_p field.
19903 (nds32_expand_to_rtl_hook): New function.
19904 (TARGET_EXPAND_TO_RTL_HOOK): Define.
19905 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
19907 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19908 Chung-Ju Wu <jasonwucj@gmail.com>
19910 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
19911 * config/nds32/nds32-n7.md: New file.
19912 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
19913 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
19915 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
19916 * config/nds32/nds32.md (pipeline_model): Add n7.
19917 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
19918 * config/nds32/pipelines.md: Include n7 settings.
19920 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19921 Chung-Ju Wu <jasonwucj@gmail.com>
19923 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
19924 * config/nds32/nds32-e8.md: New file.
19925 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
19926 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
19928 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
19929 * config/nds32/nds32.md (pipeline_model): Add e8.
19930 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
19931 * config/nds32/pipelines.md: Include e8 settings.
19933 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19934 Chung-Ju Wu <jasonwucj@gmail.com>
19936 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
19937 * config/nds32/nds32-n8.md: New file.
19938 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
19939 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
19941 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
19942 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
19943 * config/nds32/nds32.md (pipeline_model): Add n8.
19944 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
19945 * config/nds32/pipelines.md: Include n8 settings.
19947 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19948 Chung-Ju Wu <jasonwucj@gmail.com>
19950 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
19951 * config/nds32/nds32-n9-2r1w.md: New file.
19952 * config/nds32/nds32-n9-3r2w.md: New file.
19953 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
19954 nds32_register_ports): New or modify for cpu n9.
19955 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
19957 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
19958 * config/nds32/nds32-utils.c: New file.
19959 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
19960 TARGET_MUL_SLOW): Define.
19961 * config/nds32/nds32.md (pipeline_model): New attribute.
19962 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
19963 New options that support cpu n9.
19964 * config/nds32/pipelines.md: Include n9 settings.
19965 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
19967 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19969 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
19970 information if necessary.
19971 (output_cond_branch_compare_zero): Likewise.
19972 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
19973 (nds32_target_alignment): Refine for alignment.
19974 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
19975 (FUNCTION_BOUNDARY): Modify.
19976 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
19978 * config/nds32/nds32.opt (malways-align, malign-functions): New.
19980 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19982 * config/nds32/constants.md (unspec_volatile_element): Add values for
19983 TLB operation and data prefetch.
19984 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19985 functions for TLB operation and data prefetch.
19986 * config/nds32/nds32-intrinsic.md: Likewise.
19987 * config/nds32/nds32_intrinsic.h: Likewise.
19988 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
19989 (nds32_print_operand): Likewise.
19990 * config/nds32/nds32.h (nds32_builtins): Likewise.
19992 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19993 Andrew Pinski <pinsika@gcc.gnu.org>
19995 PR middle-end/82976
19996 * match.pd: Use constant_boolean_node of correct type instead of
19997 boolean_true_node or boolean_false_node for simplifying
19998 pointer comparisons to zero.
20000 2018-04-07 Jakub Jelinek <jakub@redhat.com>
20002 PR tree-optimization/80021
20003 * tree.c (verify_type_variant): Make error call in verify_variant_match
20004 translatable and remove final full stop.
20006 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20008 * config/nds32/constants.md (unspec_volatile_element): Add
20009 UNSPEC_VOLATILE_EH_RETURN.
20010 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
20011 nds32_output_stack_pop): Support dwarf exception handling process.
20012 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
20013 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
20014 exception handling process.
20015 (nds32_compute_stack_frame): Likewise.
20016 (nds32_return_addr_rtx): Likewise.
20017 (nds32_initial_elimination_offset): Likewise.
20018 (nds32_expand_prologue): Likewise.
20019 (nds32_expand_epilogue): Likewise.
20020 (nds32_dynamic_chain_address): New function.
20021 * config/nds32/nds32.h (machine_function): Add fields for dwarf
20022 exception handling.
20023 (DYNAMIC_CHAIN_ADDRESS): Define.
20024 (EH_RETURN_DATA_REGNO): Define.
20025 (EH_RETURN_STACKADJ_RTX): Define.
20026 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
20027 patterns for dwarf exception handling.
20029 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20031 * config/nds32/nds32.h: Clean up obsolete macros.
20033 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20035 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
20036 Add enum values for particular instructions.
20037 * config/nds32/nds32-intrinsic.c: Implementation of expanding
20038 particular intrinsic functions.
20039 * config/nds32/nds32-intrinsic.md: Likewise.
20040 * config/nds32/nds32_intrinsic.h: Likewise.
20041 * config/nds32/nds32.h (nds32_builtins): Likewise.
20042 * config/nds32/nds32.md (type): Add pbsad and pbsada.
20043 (btst, ave): New patterns for particular instructions.
20045 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20047 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
20048 Add enum values for atomic load/store and memory sync.
20049 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
20051 * config/nds32/nds32-intrinsic.md: Likewise.
20052 * config/nds32/nds32_intrinsic.h: Likewise.
20053 * config/nds32/nds32.h (nds32_builtins): Likewise.
20055 2018-04-07 Jakub Jelinek <jakub@redhat.com>
20057 PR tree-optimization/85257
20058 * fold-const.c (native_encode_vector): If not all elts could fit
20059 and off is -1, return 0 rather than offset.
20060 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
20061 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
20062 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
20063 adjust buffer in native_interpret_expr call.
20065 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20067 * config/nds32/constants.md (unspec_volatile_element): Add cache
20068 control enum values.
20069 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
20070 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
20071 * config/nds32/nds32.c (nds32_cctl_names): New.
20072 (nds32_print_operand): Handle cache control register names.
20073 * config/nds32/nds32.h (nds32_builtins): New enum values.
20074 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
20076 * config/nds32/nds32.md (type): Add mmu.
20077 * config/nds32/pipelines.md (simple_insn): Add mmu.
20079 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20081 * config/nds32/nds32.md (type): Remove call.
20082 * config/nds32/pipelines.md (simple_insn): Likewise.
20084 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20086 * config/nds32/constants.md (unspec_volatile_element): Add
20087 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
20088 UNSPEC_VOLATILE_FMFCFG.
20089 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
20090 description for fmfcfg and fmfcsr.
20091 (bdesc_1arg): Add fmtcsr.
20092 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
20093 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
20094 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
20095 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
20096 unspec_fmfcfg): New patterns.
20097 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
20098 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
20099 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
20100 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
20101 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
20102 __nds32__fmfcfg): Define.
20104 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20106 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
20107 intrinsic register names.
20108 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
20109 intrinsic register enum values and macros.
20111 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20113 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
20114 for load/store addressing form.
20115 (nds32_print_operand_address): Likewise.
20117 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
20120 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
20121 based on LABEL_REF. Remove useless assertion.
20122 (pic_address_needs_scratch): Fix formatting.
20123 (sparc_legitimize_pic_address): Minor tweaks.
20124 (sparc_delegitimize_address): Adjust assertion accordingly.
20125 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
20126 into symbolic_operand.
20127 (movsi_high_pic_label_ref): Likewise.
20128 (movsi_lo_sum_pic_label_ref): Likewise.
20129 (movdi_pic_label_ref): Likewise.
20130 (movdi_high_pic_label_ref): Likewise.
20131 (movdi_lo_sum_pic_label_ref): Likewise.
20133 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
20135 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
20136 custom LIB_SPEC setup.
20138 2018-04-06 Ruslan Bukin <br@bsdpad.com>
20139 Kito Cheng <kito.cheng@gmail.com>
20141 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
20142 * config/riscv/freebsd.h: New.
20144 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20146 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
20147 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
20150 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20151 Kito Cheng <kito.cheng@gmail.com>
20153 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
20154 nds32_output_call, nds32_symbol_binds_local_p): New functions.
20155 * config/nds32/nds32-protos.h (nds32_output_call,
20156 nds32_output_return): Declare.
20157 * config/nds32/nds32.md: Refine all the call and return patterns.
20159 2018-04-06 Jakub Jelinek <jakub@redhat.com>
20162 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
20163 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
20165 PR rtl-optimization/84872
20166 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
20167 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
20168 EDGE_CROSSING edge.
20170 2018-04-06 Tamar Christina <tamar.christina@arm.com>
20172 * expr.c (copy_blkmode_to_reg): Revert 254862.
20173 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
20175 2018-04-06 Richard Biener <rguenther@suse.de>
20177 PR middle-end/85244
20178 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
20179 after seeing a component reference with an adjacent field. Treat
20180 refs to arrays at struct end of external decls similar to
20181 refs to unconstrained commons.
20183 2018-04-06 Jakub Jelinek <jakub@redhat.com>
20186 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
20187 look through SAVE_EXPRs with non-side-effects argument. Adjust
20189 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
20192 2018-04-06 Richard Biener <rguenther@suse.de>
20194 PR middle-end/85180
20195 * alias.c (find_base_term): New wrapper around find_base_term
20196 unwinding CSELIB_VAL_PTR changes.
20197 (find_base_term): Do not restore CSELIB_VAL_PTR during the
20200 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20202 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
20204 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
20205 constant definitions.
20206 ("nop"): lr 0,0 -> nopr r0
20207 ("nop_lr0", "nop_lr1"): New insn definitions.
20209 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20211 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
20212 NDS32_V3PUSH_AVAILABLE_P macro.
20214 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
20215 Chung-Ju Wu <jasonwucj@gmail.com>
20217 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
20218 (nds32*-*-*): Add float and fpu_config into supported_defaults.
20219 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
20220 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
20221 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
20222 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
20223 * config/nds32/constraints.md: New constraints and checking for hard
20224 float configuration.
20225 * config/nds32/iterators.md: New mode iterator and attribute for hard
20226 float configuration.
20227 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
20229 * config/nds32/nds32-fpu.md: New file.
20230 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
20231 deal with hard float code generation.
20232 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
20234 (abi_type, float_reg_number): New enum type.
20235 * config/nds32/nds32-predicates.c: New predicates for hard float.
20236 * config/nds32/nds32-protos.h: Declare functions for hard float.
20237 * config/nds32/nds32.c: Implementation for hard float configuration.
20238 * config/nds32/nds32.h: Definitions for hard float configuration.
20239 * config/nds32/nds32.md: Include hard float machine description and
20240 modify patterns for hard float configuration.
20241 * config/nds32/nds32.opt: New options for hard float configuration.
20242 * config/nds32/predicates.md: New predicates for hard float
20245 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20247 * common/config/nds32/nds32-common.c
20248 (nds32_option_optimization_table): Enable -mreleax-hint by default.
20250 2018-04-05 Jakub Jelinek <jakub@redhat.com>
20252 PR middle-end/85195
20253 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
20254 CONSTRUCTOR_ELT (ctor, ...)->value.
20256 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
20259 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
20261 2018-04-05 Tom de Vries <tom@codesourcery.com>
20264 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
20267 2018-04-05 Shiva Chen <shiva0217@gmail.com>
20268 Kito Cheng <kito.cheng@gmail.com>
20270 * config/nds32/constraints.md (U33): Fine-tune checking condition.
20271 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
20272 * config/nds32/nds32.h (nds32_16bit_address_type): Add
20273 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
20275 2018-04-05 Shiva Chen <shiva0217@gmail.com>
20276 Kito Cheng <kito.cheng@gmail.com>
20278 * config/nds32/constraints.md (Ufe): New memory constraint.
20279 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
20280 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
20281 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
20283 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
20284 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
20286 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20288 * config/nds32/nds32.md: Use optimize_size in the condition for
20289 alu-shift instructions.
20291 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20293 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
20295 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20297 * config/nds32/nds32.md (negsi2): Refine pattern.
20299 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
20300 Chung-Ju Wu <jasonwucj@gmail.com>
20302 * config/nds32/iterators.md (shift_rotate): New code iterator.
20303 (shift): New code attribute.
20304 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
20305 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
20306 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
20307 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
20308 bit-wise operations.
20309 (andsi3, *andsi3): Ditto.
20310 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
20311 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
20312 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
20313 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
20314 nds32_ior_operand, nds32_xor_operand): New predicates.
20316 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20318 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
20319 (addsi3, subsi3): ... this.
20321 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20323 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
20325 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20327 * config/nds32/nds32.md: Adjust indention.
20329 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
20331 * config/nds32/nds32.md (feature): New attribute.
20333 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20335 * config/nds32/nds32.md (subtype): New attribute.
20337 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20340 * config/arm/arm-builtins.c (arm_expand_builtin): Change
20341 expansion to perform a bitwise AND of the argument followed by a
20342 boolean negation of the result.
20344 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
20346 PR rtl-optimization/84878
20347 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
20348 the basic block. Assert the use reference is not artificial and that
20349 it has an associated insn.
20351 2018-04-04 Michael Matz <matz@suse.de>
20353 * builtins.c (compute_objsize): Pass correct operand
20354 to array_at_struct_end_p.
20356 2018-04-04 Richard Biener <rguenther@suse.de>
20359 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
20360 from contexts for DINFO_LEVEL_TERSE and below.
20362 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20364 * config/nds32/nds32-doubleword.md (move_<mode>): Require
20365 resiter_operand condition.
20366 * config/nds32/nds32.md (*move<mode>): Ditto.
20368 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20369 Monk Chiang <sh.chiang04@gmail.com>
20371 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
20373 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20375 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
20377 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20378 Kito Cheng <kito.cheng@gmail.com>
20380 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
20381 nds32_cond_code_str, output_cond_branch,
20382 output_cond_branch_compare_zero, nds32_expand_cbranch,
20383 nds32_expand_cstore, nds32_expand_movcc,
20384 nds32_output_cbranchsi4_equality_zero,
20385 nds32_output_cbranchsi4_equality_reg,
20386 nds32_output_cbranchsi4_equality_reg_or_const_int,
20387 nds32_output_cbranchsi4_greater_less_zero: New functions.
20388 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
20389 nds32_expand_cstore, nds32_expand_movcc,
20390 nds32_output_cbranchsi4_equality_zero,
20391 nds32_output_cbranchsi4_equality_reg,
20392 nds32_output_cbranchsi4_equality_reg_or_const_int,
20393 nds32_output_cbranchsi4_greater_less_zero): Declare.
20394 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
20395 nds32_rimm11s_operand): New predicates.
20396 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
20397 * config/nds32/nds32.md: Rewrite all the branch and conditional move
20400 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20402 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
20403 * config/nds32/nds32.md: Ditto.
20404 * config/nds32/pipelines.md: Ditto.
20406 2018-04-04 Richard Biener <rguenther@suse.de>
20408 PR tree-optimization/85168
20409 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
20410 propagating abnormals.
20412 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20414 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
20416 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20417 Kito Cheng <kito.cheng@gmail.com>
20419 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
20420 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
20421 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
20422 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
20423 * config/nds32/nds32.md (sibcall_internal): New.
20424 (sibcall_register): Remove.
20425 (sibcall_immediate): Remove.
20426 (sibcall_value_internal): New.
20427 (sibcall_value_register): Remove.
20428 (sibcall_value_immediate): Remove.
20429 * config/nds32/predicates.md (nds32_general_register_operand): New.
20430 (nds32_call_address_operand): New.
20432 2018-04-03 Jakub Jelinek <jakub@redhat.com>
20434 PR rtl-optimization/85167
20435 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
20436 bb_defs if *split_p, instead preinitialize it to NULL.
20438 PR tree-optimization/85156
20439 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
20440 evaluating the argument multiple times.
20442 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
20444 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
20446 (_mm_cvtpd_ps): Likewise.
20447 (_mm_cvttpd_epi32): Likewise.
20448 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
20449 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
20450 vector, pixel, and bool following altivec.h include.
20452 2018-04-03 Martin Sebor <msebor@redhat.com>
20454 * doc/extend.texi (Common Function Attributes): Clarify.
20455 (const attribute): Likewise.
20456 (pure attribute): Likewise.
20458 2018-04-03 Jakub Jelinek <jakub@redhat.com>
20461 * config/i386/i386.c (ix86_expand_vector_set): Use
20462 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
20464 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
20466 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
20467 instructions when changing rounding bits to preserve precision bits
20468 in the x87 control word.
20470 2018-04-03 Martin Liska <mliska@suse.cz>
20472 PR tree-optimization/82491
20473 * rtl.h (strip_offset_and_add): Replace += suboffset with
20474 poly_uint64 () + suboffset.
20476 2018-03-29 Martin Liska <mliska@suse.cz>
20477 Martin Jambor <mjambor@suse.cz>
20480 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
20481 param_type is not an integral or pointer type.
20483 2018-04-03 Richard Biener <rguenther@suse.de>
20485 * sese.h (recompute_all_dominators): Remove.
20487 2018-04-02 Martin Sebor <msebor@redhat.com>
20489 * doc/invoke.texi (-Wrestrict): Fix typos.
20491 2018-04-02 Jim Wilson <jimw@sifive.com>
20493 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
20494 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
20495 (<optab>di3, <optab>si3_extend): Likewise.
20496 (<optab>si3_mask, <optab>si3_mask_1): New.
20497 (<optab>di3_mask, <optab>di3_mask_1): New.
20498 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
20499 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
20500 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
20502 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
20504 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
20507 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
20509 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
20510 (nds32_canonicalize_comparison): New function.
20512 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20513 Kito Cheng <kito.cheng@gmail.com>
20514 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20516 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
20517 * config/nds32/constants.md (unspec_volatile_element): Add
20518 UNSPEC_VOLATILE_RELAX_GROUP.
20519 * config/nds32/nds32-relax-opt.c: New file.
20520 * config/nds32/nds32-predicates.c
20521 (nds32_symbol_load_store_p): New function.
20522 * config/nds32/nds32-protos.h
20523 (nds32_symbol_load_store_p): Declare function.
20524 (make_pass_nds32_relax_opt): Declare new rtl pass function.
20525 * config/nds32/nds32.c
20526 (nds32_register_pass): New function to register pass.
20527 (nds32_register_passes): New function to register passes.
20528 * config/nds32/nds32.md (relax_group): New pattern.
20529 * config/nds32/nds32.opt (mrelax-hint): New option.
20530 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
20532 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
20534 * config/nds32/t-nds32: Modify files dependency.
20536 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20538 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
20539 (PROFILE_HOOK): Define its implementation.
20541 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20543 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
20544 type and 32-bit size.
20546 2018-04-01 Jakub Jelinek <jakub@redhat.com>
20548 PR middle-end/85090
20549 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
20550 (V_128_256): New mode iterator.
20551 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
20552 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
20553 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
20555 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
20556 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
20558 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20561 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
20562 NaN inputs correctly.
20564 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
20567 * config/rs6000/vsx.md (??r): New mode attribute.
20568 (*vsx_mov<mode>_64bit): Use it.
20569 (*vsx_mov<mode>_32bit): Likewise.
20571 2018-03-30 Martin Sebor <msebor@redhat.com>
20573 PR tree-optimization/84818
20574 * builtins.c (check_access): Use warning_n.
20576 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20579 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
20581 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
20584 2018-03-30 Julia Koval <julia.koval@intel.com>
20587 * x86-tune.def (movx, partial_reg_dependency): Enable for
20590 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
20592 PR inline-asm/84985
20593 * lra-constraints.c (process_alt_operands): Move setting
20594 this_alternative_matches below.
20596 2018-03-29 Martin Liska <mliska@suse.cz>
20599 * doc/invoke.texi: Document how LTO works with debug info.
20600 Describe auto-load support of binutils. Mention 'x86-64'
20601 as valid option value of -march option.
20603 2018-03-29 Jakub Jelinek <jakub@redhat.com>
20605 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
20608 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
20609 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
20610 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
20613 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20616 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
20617 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
20618 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
20619 for RS6000_BTM_POWERPC64.
20620 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
20621 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
20622 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
20625 (DIVDEU): Likewise.
20627 2018-03-28 Carl Love <cel@us.ibm.com>
20630 2017-09-27 Carl Love <cel@us.ibm.com>
20632 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
20633 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
20634 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
20637 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20639 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
20640 instead of __vector bool.
20641 (_mm_max_pu8): Likewise.
20642 (_mm_min_pi16): Likewise.
20644 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20647 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
20648 (DIVWEUO): Likewise.
20649 (DIVDEO): Likewise.
20650 (DIVDEUO): Likewise.
20651 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
20652 DIVWEUO and DIVDEUO.
20653 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
20654 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
20655 (div_extend): Likewise.
20656 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
20658 (__builtin_divweuo): Likewise.
20659 (__builtin_divdeo): Likewise.
20660 (__builtin_divdeuo): Likewise.
20662 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20665 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
20666 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
20668 PR tree-optimization/82004
20669 * gimple-match-head.c (optimize_pow_to_exp): New function.
20670 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
20671 Don't fold to exp if optimize_pow_to_exp is false.
20673 2018-03-28 Martin Liska <mliska@suse.cz>
20676 * calls.c (initialize_argument_information): Fix trailing space.
20677 * common.opt: Fix typo and provide better explanation for
20678 -fsanitize-coverage option.
20679 * config/i386/i386.opt: Fix typo.
20681 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20682 Martin Liska <mliska@suse.cz>
20685 * gimplify.c (asan_poison_variable): Don't do the check for
20686 gimplify_omp_ctxp here.
20687 (gimplify_decl_expr): Do it here.
20688 (gimplify_target_expr): Likewise.
20690 2018-03-28 Martin Liska <mliska@suse.cz>
20693 * config/i386/i386.c (ix86_function_arg_advance): Do not call
20694 chkp_type_bounds_count if MPX is not enabled.
20696 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
20698 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
20700 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20703 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
20704 function to create the function decl for complex long double
20705 multiply and divide for -mabi=ieeelongdouble.
20706 (init_float128_ieee): Call it.
20708 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
20711 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
20712 -fcf-protection=branch -mibt.
20713 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
20715 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20718 * config/arm/arm.c (arm_valid_symbolic_address): Handle
20719 arm_word_relocations.
20721 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
20724 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
20725 extern array declarations.
20727 2018-03-27 Richard Biener <rguenther@suse.de>
20729 PR middle-end/84067
20730 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
20731 explicit single_use checks.
20733 2018-03-27 Richard Biener <rguenther@suse.de>
20735 PR tree-optimization/85082
20736 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
20739 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20741 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
20742 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
20743 Turn on fasynchronous-unwind-tables and funwind-tables.
20745 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
20748 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
20749 (*bmi_blsr_<mode>_ccz): Ditto.
20751 2018-03-26 Tom de Vries <tom@codesourcery.com>
20753 PR tree-optimization/85063
20754 * omp-general.c (offloading_function_p): New function. Factor out
20756 * omp-offload.c (pass_omp_target_link::gate): ... here.
20757 * omp-general.h (offloading_function_p): Declare.
20758 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
20759 with attribute omp declare target for offloading functions.
20761 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
20763 PR tree-optimization/84005
20764 * tree-data-ref.h (get_base_for_alignment): Declare.
20765 * tree-data-ref.c (get_base_for_alignment_1): New function.
20766 (get_base_for_alignment): Likewise.
20767 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20768 get_base_for_alignment to find a suitable base object, instead
20769 of always using drb->base_address.
20771 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20773 PR inline-asm/85022
20774 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
20775 known size by default.
20777 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
20779 PR inline-asm/85030
20780 * lra-constraints.c (process_alt_operands): Don't match BLKmode
20781 and non BLKmode operands.
20783 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20786 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
20787 Clean up attributes.
20789 2018-03-23 Richard Biener <rguenther@suse.de>
20792 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
20793 we are going to emit early debug for LTO.
20795 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20797 PR inline-asm/85034
20798 * function.c (match_asm_constraints_1): Don't optimize if input
20799 doesn't satisfy general_operand predicate for output's mode.
20801 PR inline-asm/85022
20802 * alias.c (write_dependence_p): Don't require for x_canonicalized
20803 non-VOIDmode if x has VOIDmode.
20806 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
20807 just don't try to optimize it rather than assert it never happens.
20809 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20811 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
20812 macro expansions for definition of ST_INTERNAL_<mode> and
20813 LD_INTERNAL_<mode> builtins.
20814 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
20816 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
20818 (altivec_expand_st_builtin): Likewise.
20819 (altivec_expand_builtin): Remove calls to deleted functions.
20820 (rs6000_address_for_altivec): Delete this function.
20821 * config/rs6000/vector.md: Remove expands for
20822 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
20824 2018-03-22 Sudakshina Das <sudi.das@arm.com>
20827 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
20828 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
20829 re-computing once computed.
20830 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
20831 (arm_init_machine_status): Initialize
20832 machine->static_chain_stack_bytes.
20834 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20837 * doc/extend.texi: Add four new prototypes for vec_ld.
20838 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
20839 definitions for more logical presentation.
20840 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
20841 entries for V1TI variants of __builtin_altivec_ld builtin.
20842 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
20843 handling of V1TI variant of LVX icode pattern.
20844 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
20845 (rs6000_gimple_fold_builtin): Likewise.
20846 (altivec_init_builtins): Add code to define
20847 __builtin_altivec_lvx_v1ti function.
20849 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20851 PR inline-asm/84941
20852 * function.c (match_asm_constraints_1): Don't do the optimization
20853 if input isn't a REG, SUBREG, MEM or constant.
20855 2018-03-22 Tom de Vries <tom@codesourcery.com>
20857 PR tree-optimization/84956
20858 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
20859 bb_has_abnormal_pred.
20861 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20864 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
20865 DECL_INITIAL (decl) to decl at the end.
20866 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
20867 adjust the comment.
20869 2018-03-21 Joseph Myers <joseph@codesourcery.com>
20871 * doc/extend.texi (__builtin_tgmath): Document when complex
20872 integer types are treated as _Complex _Float64.
20874 2018-03-21 Tom de Vries <tom@codesourcery.com>
20876 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
20878 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20880 PR tree-optimization/84960
20881 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
20882 if it is ENTRY block, move them into single succ of ENTRY in that case.
20884 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
20886 PR tree-optimization/84811
20887 * poly-int.h (poly_span_traits): Remove the T3 parameter and
20888 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
20889 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
20890 (known_subrange_p): Update accordingly. Cast each value involved
20891 in the size comparison, rather than casting the result of the
20894 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20896 PR tree-optimization/84982
20897 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
20898 by flipping the least significant bit rather than all bits from
20899 bitpos to bitpos + bitsize - 1.
20901 2018-03-21 Nathan Sidwell <nathan@acm.org>
20903 * doc/extend.texi (Deprecated Features): Remove mention of
20904 long-deleted deprecations.
20906 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20909 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
20910 * configure: Regenerate.
20912 2018-03-21 Tom de Vries <tom@codesourcery.com>
20914 PR tree-optimization/83126
20915 * tree-parloops.c (num_phis): New function.
20916 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
20918 2018-03-21 Nathan Sidwell <nathan@acm.org>
20920 * doc/extend.texi (Deprecated Features): Update deprecated flags,
20921 mention anon-struct/union members and trailing attributes.
20923 2018-03-21 Bin Cheng <bin.cheng@arm.com>
20925 PR tree-optimization/84969
20926 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
20927 builtin memset partitions if they set different rhs values.
20929 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20931 PR rtl-optimization/84989
20932 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
20933 VEC_DUPLICATE with scalar result mode.
20935 2018-03-21 Martin Liska <mliska@suse.cz>
20938 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
20939 not intended return statement.
20941 2018-03-21 Martin Liska <mliska@suse.cz>
20944 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
20945 (chkp_find_bound_slots_1): Limit number of iterations.
20947 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
20950 * Minor grammar fixes for x86 options.
20952 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20955 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
20956 holding REG_CFA_RESTORE notes, instead turn them into a USE.
20958 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
20961 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
20962 (altivec_lvx_<mode>_1op): Likewise.
20963 (altivec_stvx_<mode>_2op): Likewise.
20964 (altivec_stvx_<mode>_1op): Likewise.
20965 (altivec_lvx_<VM2:mode>): New define_expand.
20966 (altivec_stvx_<VM2:mode>): Likewise.
20967 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
20968 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20969 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
20970 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20971 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
20972 (rs6000_gen_lvx): Likewise.
20973 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
20974 (altivec_expand_stv_builtin): Likewise.
20975 (altivec_expand_builtin): Likewise.
20976 * config/rs6000/vector.md: Likewise.
20978 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20981 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
20984 2018-03-20 Richard Biener <rguenther@suse.de>
20987 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
20988 sign-conversions as zero, fall back to standard scalar_stmt
20991 2018-03-20 Martin Liska <mliska@suse.cz>
20994 * predict.c (rebuild_frequencies): Handle case when we have
20995 PROFILE_ABSENT, but flag_guess_branch_prob is false.
20997 2018-03-20 Jakub Jelinek <jakub@redhat.com>
21000 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
21001 flag_section_anchors.
21002 * varasm.c (use_blocks_for_decl_p): Remove hack for
21003 dw2_force_const_mem.
21006 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
21008 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
21009 be created, use lowpart_subreg of operands[0] rather than operands[0]
21011 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
21012 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
21013 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
21014 and n constraint instead of aarch64_shift_imm_di and Usd.
21015 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
21016 (*aarch64_<optab>_reg_minus<mode>3): ... this.
21018 2018-03-20 Sudakshina Das <sudi.das@arm.com>
21021 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
21022 to favor GPR over NEON registers.
21023 (<shift>di3_neon): Likewise.
21025 2018-03-20 Tom de Vries <tom@codesourcery.com>
21028 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
21029 (nvptx_process_pars): Emit bar.sync asap and alap.
21031 2018-03-20 Tom de Vries <tom@codesourcery.com>
21034 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
21035 seen_label if seen_label is already set.
21037 2018-03-20 Jakub Jelinek <jakub@redhat.com>
21040 * config/i386/i386.c (fold_builtin_cpu): For features above 31
21041 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
21042 Use 1U instead of 1. Formatting fixes.
21045 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
21046 instead of TREE_TYPE (s1) for the return value.
21048 2018-03-19 Jakub Jelinek <jakub@redhat.com>
21050 PR tree-optimization/84946
21051 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
21052 bitsize + bitsize in poly_uint64 rather than poly_int64.
21055 * dwarf2asm.c: Include fold-const.c.
21056 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
21057 of decl rather than decl itself.
21059 PR rtl-optimization/84643
21060 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
21062 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
21065 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
21066 calling assemble_variable.
21068 2018-03-19 Sudakshina Das <sudi.das@arm.com>
21071 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
21072 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
21074 2018-03-19 Jim Wilson <jimw@sifive.com>
21077 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
21078 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
21079 (riscv_first_stack_step): Likewise.
21080 (riscv_option_override): Use STACK_BOUNDARY instead of
21081 MIN_STACK_BOUNDARY.
21082 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
21083 MIN_STACK_BOUNDARY.
21084 (BIGGEST_ALIGNMENT): Set to 128.
21085 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
21086 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
21089 2018-03-19 Richard Biener <rguenther@suse.de>
21091 PR tree-optimization/84933
21092 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
21093 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
21095 2018-03-19 Richard Biener <rguenther@suse.de>
21097 PR tree-optimization/84859
21098 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
21099 (cond_if_else_store_replacement): Perform sinking operation on
21100 single-store BBs regardless of MAX_STORES_TO_SINK setting.
21101 Generalize what a BB with a single eligible store is.
21103 2018-03-19 Richard Biener <rguenther@suse.de>
21105 PR tree-optimization/84929
21106 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
21107 chrec_is_positive against non-chrec arg.
21109 2018-03-19 Tamar Christina <tamar.christina@arm.com>
21112 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
21114 2018-03-18 Martin Liska <mliska@suse.cz>
21116 PR rtl-optimization/84635
21117 * regrename.c (build_def_use): Use matches_mode only when
21120 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
21122 PR tree-optimization/84913
21123 * tree-vect-loop.c (vectorizable_reduction): Don't try to
21124 vectorize chains of COND_EXPRs.
21126 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21128 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
21130 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21132 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
21134 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21136 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
21138 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
21139 Kito Cheng <kito.cheng@gmail.com>
21141 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
21142 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
21143 (nds32_adjust_reg_alloc_order): New function.
21144 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
21146 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
21148 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
21149 nds32_print_operand, nds32_print_operand_address): Use
21150 HOST_WIDE_INT_PRINT_DEC instead.
21152 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
21154 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
21156 2018-03-17 Jakub Jelinek <jakub@redhat.com>
21159 * config/i386/i386.c (initial_ix86_tune_features,
21160 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
21161 unsigned long long.
21162 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
21163 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
21164 rather than 1u << ix86_tune. Formatting fix.
21165 (ix86_option_override_internal): Change ix86_arch_mask from
21166 unsigned int to unsigned HOST_WIDE_INT, initialize to
21167 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
21168 (ix86_function_specific_restore): Likewise.
21170 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21173 * postreload.c (reload_combine_recognize_pattern): Perform
21174 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
21175 truncate_int_for_mode the result for the destination's mode.
21178 * hsa-gen.c (mem_type_for_type): Fix comment typo.
21179 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
21181 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21184 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
21187 * lra-assigns.c (lra_split_hard_reg_for): Don't use
21188 regno_allocno_class_array and sorted_pseudos.
21189 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
21190 insns where regno is used.
21192 2018-03-16 Martin Liska <mliska@suse.cz>
21195 * multiple_target.c (create_dispatcher_calls): Redirect
21196 reference in the symbol table.
21198 2018-03-16 Martin Liska <mliska@suse.cz>
21201 * multiple_target.c (create_dispatcher_calls): Redirect also
21204 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21208 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
21209 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
21210 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
21212 2018-03-16 Julia Koval <julia.koval@intel.com>
21214 * doc/invoke.texi (Skylake Server): Add CLWB.
21215 Cannonlake): Remove CLWB.
21217 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21219 PR tree-optimization/84841
21220 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
21222 (FLOAT_ONE_CONST_TYPE): Define.
21223 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
21224 (sort_by_operand_rank): Put entries with higher constant_type last
21225 rather than first to match comments.
21227 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
21229 * config/nios2/nios2.md (movsi_internal): Fix thinko in
21232 2018-03-15 Jakub Jelinek <jakub@redhat.com>
21235 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
21236 check and use address of target always.
21238 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
21241 * config/i386/i386.c (indirect_thunk_needed): Update comments.
21242 (indirect_thunk_bnd_needed): Likewise.
21243 (indirect_thunks_used): Likewise.
21244 (indirect_thunks_bnd_used): Likewise.
21245 (indirect_return_needed): New.
21246 (indirect_return_bnd_needed): Likewise.
21247 (output_indirect_thunk_function): Add a bool argument for
21249 (output_indirect_thunk_function): Don't generate alias for
21250 function return thunk.
21251 (ix86_code_end): Call output_indirect_thunk_function to generate
21252 function return thunks.
21253 (ix86_output_function_return): Set indirect_return_bnd_needed
21254 and indirect_return_needed instead of indirect_thunk_bnd_needed
21255 and indirect_thunk_needed.
21257 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
21259 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
21260 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
21261 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
21263 2018-03-15 David Malcolm <dmalcolm@redhat.com>
21264 Paul Hua <paul.hua.gm@gmail.com>
21267 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
21269 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
21271 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
21272 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
21273 resp. SFmode cases.
21275 2018-03-15 Tamar Christina <tamar.christina@arm.com>
21278 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
21279 instead of GET_MODE_SIZE when comparing Units.
21281 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
21284 * varasm.c (hash_section): Return an unchangeble hash value
21285 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
21286 Return !aarch64_can_use_per_function_literal_pools_p ().
21288 2018-03-15 Jakub Jelinek <jakub@redhat.com>
21291 * optabs.c (emit_conditional_move): Pass address of cmode's copy
21292 rather than address of cmode as last argument to prepare_cmp_insn.
21294 2018-03-15 Julia Koval <julia.koval@intel.com>
21296 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
21297 F_AVX512VNNI, F_AVX512BITALG): New.
21299 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
21302 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
21303 insn for floating-point loads and stores.
21305 2018-03-14 Carl Love <cel@us.ibm.com>
21307 * config/rs6000/rs6000-c.c: Add macro definitions for
21308 ALTIVEC_BUILTIN_VEC_PERMXOR.
21309 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
21310 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
21311 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
21312 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
21314 * config/doc/extend.texi: Add prototypes for vec_permxor.
21316 2018-03-14 David Malcolm <dmalcolm@redhat.com>
21319 * diagnostic-show-locus.c (class layout_point): Convert m_line
21320 from int to linenum_type.
21321 (line_span::comparator): Use linenum "compare" function when
21322 comparing line numbers.
21323 (test_line_span): New function.
21324 (layout_range::contains_point): Convert param "row" from int to
21326 (layout_range::intersects_line_p): Likewise.
21327 (layout::will_show_line_p): Likewise.
21328 (layout::print_source_line): Likewise.
21329 (layout::should_print_annotation_line_p): Likewise.
21330 (layout::print_annotation_line): Likewise.
21331 (layout::print_leading_fixits): Likewise.
21332 (layout::annotation_line_showed_range_p): Likewise.
21333 (struct line_corrections): Likewise for field m_row.
21334 (line_corrections::line_corrections): Likewise for param "row".
21335 (layout::print_trailing_fixits): Likewise.
21336 (layout::get_state_at_point): Likewise.
21337 (layout::get_x_bound_for_row): Likewise.
21338 (layout::print_line): Likewise.
21339 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
21340 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
21341 * input.c (selftest::test_linenum_comparisons): New function.
21342 (selftest::input_c_tests): Call it.
21343 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
21344 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
21345 * selftest.h (ASSERT_GT): New macro.
21346 (ASSERT_GT_AT): New macro.
21347 (ASSERT_LT): New macro.
21348 (ASSERT_LT_AT): New macro.
21350 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
21352 PR rtl-optimization/84780
21353 * combine.c (distribute_links): Don't make a link based on pc_rtx.
21355 2018-03-14 Martin Liska <mliska@suse.cz>
21357 * tree.c (record_node_allocation_statistics): Use
21358 get_stats_node_kind.
21359 (get_stats_node_kind): New function extracted from
21360 record_node_allocation_statistics.
21361 (free_node): Use get_stats_node_kind.
21363 2018-03-14 Richard Biener <rguenther@suse.de>
21365 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
21366 that the value-set of ANTIC_IN doesn't grow.
21369 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
21371 (BB_VISITED_WITH_VISITED_SUCCS): New define.
21372 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
21374 2018-03-14 Julia Koval <julia.koval@intel.com>
21376 * config.gcc (icelake-client, icelake-server): New.
21378 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
21379 (initial_ix86_arch_features): Ditto.
21380 (PTA_SKYLAKE): Add SGX.
21381 (PTA_ICELAKE): Remove.
21382 (PTA_ICELAKE_CLIENT): New.
21383 (PTA_ICELAKE_SERVER): New.
21384 (ix86_option_override_internal): Split up icelake on icelake client and
21386 (get_builtin_code_for_version): Ditto.
21387 (fold_builtin_cpu): Ditto.
21388 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
21389 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
21390 * config/i386/i386.h (processor_type): Ditto.
21391 * doc/invoke.texi: Ditto.
21393 2018-03-14 Jakub Jelinek <jakub@redhat.com>
21396 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
21397 INTEGER_CST offset, add it together with bitpos / 8 and
21398 sign extend based on POINTER_SIZE.
21402 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21405 * config/i386/constraints.md (Yc): New register constraint.
21406 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21407 Use Yc constraint for alternative 2 of operand 0. Remove
21408 preferred_for_speed attribute.
21410 2018-03-14 Richard Biener <rguenther@suse.de>
21412 PR tree-optimization/84830
21413 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
21414 with the old one to avoid oscillations.
21416 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
21419 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
21421 (assign_by_spills): Return a flag of reload assignment failure.
21422 Do not process the reload assignment failures. Do not spill other
21423 reload pseudos if they has the same reg class. Update n if
21425 (lra_assign): Add a return arg. Set up from the result of
21426 assign_by_spills call.
21427 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21428 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21429 usage_insns if it is not NULL.
21430 (spill_hard_reg_in_range): New function.
21431 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21432 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21433 function prototypes.
21434 (lra_assign): Change prototype.
21435 * lra.c (lra): Add code to deal with fails by splitting hard reg
21438 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
21440 * config/riscv/riscv.opt (mrelax): New option.
21441 * config/riscv/riscv.c (riscv_file_start): Emit ".option
21442 "norelax" when riscv_mrelax is disabled.
21443 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
21445 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21448 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
21449 reassociation for int modes.
21451 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21453 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
21454 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
21456 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
21457 * config/aarch64/aarch64-sve.md
21458 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
21459 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
21460 (*extend<mode><Vwide>2): Rename to...
21461 (aarch64_sve_extend<mode><Vwide>2): ...this.
21462 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
21463 renaming the old pattern to...
21464 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
21466 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
21467 define_expand, renaming the old pattern to...
21468 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
21469 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
21470 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
21471 account when deciding which SVE instruction the optab should use.
21472 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
21474 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21476 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
21477 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
21478 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
21479 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
21480 (tlsdesc_small_<mode>): Turn a define_expand and use
21481 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
21482 (tlsdesc_small_advsimd_<mode>): ...this.
21483 (tlsdesc_small_sve_<mode>): New pattern.
21485 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21487 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
21488 (UNSPEC_UMUL_HIGHPART): New constants.
21489 (MUL_HIGHPART): New int iteraor.
21490 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
21491 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
21493 (*<su>mul<mode>3_highpart): New define_insn.
21495 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
21498 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
21501 2018-03-13 Martin Liska <mliska@suse.cz>
21504 * (sem_item_optimizer::sem_item_optimizer): Initialize new
21506 (sem_item_optimizer::~sem_item_optimizer): Release it.
21507 (sem_item_optimizer::merge_classes): Register variable aliases.
21508 (sem_item_optimizer::fixup_pt_set): New function.
21509 (sem_item_optimizer::fixup_points_to_sets): Likewise.
21510 * ipa-icf.h: Declare new variables and functions.
21512 2018-03-13 Jakub Jelinek <jakub@redhat.com>
21514 PR middle-end/84834
21515 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
21516 integer_pow2p@2 and test integer_pow2p in condition.
21517 (A < 0 ? C : 0): Similarly for @1.
21519 PR middle-end/84831
21520 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
21521 characters starting at p contain '\0' character, don't look beyond
21525 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
21526 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
21529 * reg-stack.c (change_stack): Change update_end var from int to
21530 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
21531 also call set_block_for_insn on the newly added insns and rescan.
21534 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
21535 on the last operand.
21538 * tree.c (stabilize_reference_1): Return save_expr (e) for
21539 STATEMENT_LIST even if it doesn't have side-effects.
21541 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
21543 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
21545 2018-03-12 Renlin Li <renlin.li@arm.com>
21547 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
21548 aarch64_output_scalar_simd_mov_immediate.
21550 2018-03-12 Martin Sebor <msebor@redhat.com>
21552 PR tree-optimization/83456
21553 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
21554 for perfectly overlapping calls to memcpy.
21555 (gimple_fold_builtin_memory_chk): Same.
21556 (gimple_fold_builtin_strcpy): Handle no-warning.
21557 (gimple_fold_builtin_stxcpy_chk): Same.
21558 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
21560 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21562 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
21563 parameter. Use it for SFmode.
21564 (rs6000_function_arg_advance_1): Adjust.
21565 (rs6000_function_arg): Adjust.
21566 (rs6000_gimplify_va_arg): Pass false for that new parameter.
21568 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21570 PR rtl-optimization/84169
21571 PR rtl-optimization/84780
21572 * combine.c (can_combine_p): Check for a 2-insn combination whether
21573 the destination register is used between the two insns, too.
21575 2018-03-12 Richard Biener <rguenther@suse.de>
21577 PR tree-optimization/84803
21578 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
21579 for refs DR analysis didn't process.
21581 2018-03-12 Richard Biener <rguenther@suse.de>
21583 PR tree-optimization/84777
21584 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
21585 force-vectorize loops ignore whether we are optimizing for size.
21587 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
21589 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
21590 (TARGET_MD_ASM_ADJUST): Define.
21592 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
21593 Kito Cheng <kito.cheng@gmail.com>
21594 Chung-Ju Wu <jasonwucj@gmail.com>
21596 * config/nds32/nds32.c (nds32_compute_stack_frame,
21597 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
21598 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
21599 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
21600 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
21601 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
21602 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
21603 * config/nds32/nds32.md (prologue, epilogue): Use macro
21604 NDS32_V3PUSH_AVAILABLE_P to do checking.
21606 2018-03-11 Jakub Jelinek <jakub@redhat.com>
21609 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
21610 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
21611 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
21612 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
21613 addition of most attributes on !orig_type_die or the attribute not
21614 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
21616 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21617 Chung-Ju Wu <jasonwucj@gmail.com>
21619 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
21620 __NDS32_VH__ macro.
21621 * config/nds32/nds32.opt (mvh): New option.
21623 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21624 Chung-Ju Wu <jasonwucj@gmail.com>
21626 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
21628 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
21629 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
21632 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21633 Chung-Ju Wu <jasonwucj@gmail.com>
21635 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
21637 * config/nds32/nds32-multiple.md (strlensi): New pattern.
21638 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
21640 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
21641 Kito Cheng <kito.cheng@gmail.com>
21642 Chung-Ju Wu <jasonwucj@gmail.com>
21644 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
21645 UNSPEC_FFMISM and UNSPEC_FLMISM.
21646 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
21647 for ffb, ffmism and flmism.
21648 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
21649 (unspec_ffmism): Ditto.
21650 (unspec_flmism): Ditto.
21651 (nds32_expand_builtin_impl): Check if string extension is available.
21652 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
21653 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
21655 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
21658 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21661 * lra-assigns.c (assign_by_spills): Return a flag of reload
21662 assignment failure. Do not process the reload assignment
21663 failures. Do not spill other reload pseudos if they has the same
21665 (lra_assign): Add a return arg. Set up from the result of
21666 assign_by_spills call.
21667 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21668 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21669 usage_insns if it is not NULL.
21670 (spill_hard_reg_in_range): New function.
21671 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21672 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21673 function prototypes.
21674 (lra_assign): Change prototype.
21675 * lra.c (lra): Add code to deal with fails by splitting hard reg
21678 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
21681 * config/i386/i386.opt: Replace Enforcment with Enforcement.
21683 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
21686 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
21687 (dw_val_node): Add val_symbolic_view.
21688 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
21689 (symview_upper_bound): New.
21690 (new_line_info_table): Initialize symviews_since_reset.
21691 (dwarf2out_source_line): Count symviews_since_reset and set
21692 symview_upper_bound.
21693 (dw_val_equal_p): Handle symview.
21694 (add_AT_symview): New.
21695 (print_dw_val): Handle symview.
21696 (attr_checksum, attr_checksum_ordered): Likewise.
21697 (same_dw_val_p, size_of_die): Likewise.
21698 (value_format, output_die): Likewise.
21699 (add_high_low_attributes): Use add_AT_symview for entry_view.
21700 (dwarf2out_finish): Reset symview_upper_bound, clear
21703 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
21706 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
21707 Add strict argument and use it.
21708 (rs6000_split_multireg_move): Update for new strict argument.
21709 (mem_operand_gpr): Disallow all non-offsettable addresses.
21710 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
21712 2018-03-09 Jakub Jelinek <jakub@redhat.com>
21715 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
21716 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
21717 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
21720 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
21721 decl, use remap_type if we want to use the type.
21723 2018-03-09 Martin Sebor <msebor@redhat.com>
21725 PR tree-optimization/84526
21726 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21728 (builtin_access::generic_overlap): Be prepared to handle non-array
21731 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
21733 PR rtl-optimization/84682
21734 * lra-constraints.c (process_address_1): Check is_address flag
21735 for address constraints.
21736 (process_alt_operands): Likewise.
21737 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
21738 preprocess_constraints.
21739 * recog.h (preprocess_constraints): Add oploc parameter.
21741 * recog.c (preprocess_constraints): Test address_operand for
21742 CT_ADDRESS constraints.
21744 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21747 * lra-assigns.c (assign_by_spills): Return a flag of reload
21748 assignment failure. Do not process the reload assignment
21749 failures. Do not spill other reload pseudos if they has the same
21751 (lra_assign): Add a return arg. Set up from the result of
21752 assign_by_spills call.
21753 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21754 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21755 usage_insns if it is not NULL.
21756 (spill_hard_reg_in_range): New function.
21757 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21758 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21759 function prototypes.
21760 (lra_assign): Change prototype.
21761 * lra.c (lra): Add code to deal with fails by splitting hard reg
21764 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21767 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
21768 Accept complain bool parameter. Only emit errors if it is true.
21769 (arm_parse_cpu_option_name): Likewise.
21770 (arm_target_thumb_only): Adjust callers of the above.
21771 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
21772 prototype to take a default true bool parameter.
21773 (arm_parse_arch_option_name): Likewise.
21775 2018-03-09 David Malcolm <dmalcolm@redhat.com>
21776 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
21780 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
21781 * configure: Regenerate.
21782 * configure.ac ("linker --version-script option"): New.
21783 ("linker soname option"): New.
21785 2018-03-09 Richard Biener <rguenther@suse.de>
21787 PR tree-optimization/84775
21788 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
21789 immediate uses of predicate stmts and mark them modified.
21792 PR tree-optimization/84178
21793 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21795 (version_loop_for_if_conversion): Delay update_ssa call.
21796 (tree_if_conversion): Delay update_ssa until after predicate
21799 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
21802 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
21803 when the function accesses prior frames.
21805 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21808 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
21809 gen_llsym, otherwise call maybe_gen_llsym.
21811 PR inline-asm/84742
21812 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
21813 has ',' character inside of it.
21815 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21818 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
21819 as clobbering CC_REGNUM.
21821 2018-03-08 Richard Biener <rguenther@suse.de>
21823 PR middle-end/84552
21824 * tree-scalar-evolution.c: Include tree-into-ssa.h.
21825 (follow_copies_to_constant): Do not follow SSA names registered
21828 2018-03-08 Richard Biener <rguenther@suse.de>
21830 PR tree-optimization/84178
21831 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21833 (version_loop_for_if_conversion): Delay update_ssa call.
21834 (tree_if_conversion): Delay update_ssa until after predicate
21837 2018-03-08 David Malcolm <dmalcolm@redhat.com>
21839 PR tree-optimization/84178
21840 * tree-if-conv.c (release_bb_predicate): Remove the
21841 the assertion that the stmts have NULL use_ops.
21842 Discard the statements, asserting that they haven't
21843 yet been added to a BB.
21845 2018-03-08 Richard Biener <rguenther@suse.de>
21847 PR tree-optimization/84746
21848 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
21849 (phi_translate): Pass in destination ANTIC_OUT set.
21850 (phi_translate_1): Likewise. For a simplified result lookup
21851 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
21852 (phi_translate_set): Adjust.
21853 (do_pre_regular_insertion): Likewise.
21854 (do_pre_partial_partial_insertion): Likewise.
21856 2018-03-08 Martin Liska <mliska@suse.cz>
21858 PR gcov-profile/84735
21859 * doc/gcov.texi: Document usage of profile files.
21860 * gcov-io.h: Document changes in the format.
21862 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
21866 * dwarf2out.c (struct dw_line_info_table): Update comments for
21868 (FORCE_RESET_NEXT_VIEW): New.
21869 (FORCE_RESETTING_VIEW_P): New.
21870 (RESETTING_VIEW_P): Check for -1 too.
21871 (ZERO_VIEW_P): Likewise.
21872 (new_line_info_table): Force-reset next view.
21873 (dwarf2out_begin_function): Likewise.
21874 (dwarf2out_source_line): Simplify zero_view_p initialization.
21875 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
21876 view directly. Omit view when omitting .loc at line 0.
21878 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21880 PR tree-optimization/84740
21881 * tree-switch-conversion.c (process_switch): Call build_constructors
21882 only if info.phi_count is non-zero.
21884 PR tree-optimization/84739
21885 * tree-tailcall.c (find_tail_calls): Check call arguments against
21886 DECL_ARGUMENTS (current_function_decl) rather than
21887 DECL_ARGUMENTS (func) when checking for tail recursion.
21889 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21891 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
21892 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
21893 Volker Reichelt's entry and add entries for people that perform
21894 GCC fuzzy testing and report numerous bugs.
21896 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21899 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
21900 readonly data in sdata, if that is disabled.
21901 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
21902 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
21903 -mreadonly-in-sdata option.
21905 2018-03-07 Martin Sebor <msebor@redhat.com>
21907 PR tree-optimization/84468
21908 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
21909 basic block when looking for nul assignment.
21911 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
21914 * except.h (output_function_exception_table): Adjust prototype.
21915 * except.c (output_function_exception_table): Remove FNNAME parameter
21916 and add SECTION parameter. Ouput one part of the table at a time.
21917 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
21918 the first part of the exception table and emit unwind directives.
21919 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
21920 (i386_pe_seh_cold_init): Likewise.
21921 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
21922 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
21923 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
21924 (ix86_output_call_insn): Emit a nop in one more case for SEH.
21925 * config/i386/winnt.c: Include except.h.
21926 (struct seh_frame_state): Add reg_offset, after_prologue and
21927 in_cold_section fields.
21928 (i386_pe_seh_end_prologue): Set seh->after_prologue.
21929 (i386_pe_seh_cold_init): New function.
21930 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
21931 to seh->in_cold_section.
21932 (seh_emit_push): Record the offset of the push.
21933 (seh_emit_save): Record the offet of the save.
21934 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
21935 Test seh->after_prologue to disregard the epilogue.
21936 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
21937 (i386_pe_end_cold_function): New function.
21939 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21942 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
21943 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
21946 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
21948 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
21949 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
21951 PR middle-end/84723
21952 * multiple_target.c: Include tree-inline.h and intl.h.
21953 (expand_target_clones): Diagnose and fail if node->definition and
21954 !tree_versionable_function_p (node->decl).
21956 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
21958 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
21960 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
21961 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
21962 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
21964 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21967 * combine.c (try_combine): Use reg_or_subregno instead of handling
21968 just paradoxical SUBREGs and REGs.
21970 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
21972 * config/arc/arc.c (arc_finalize_pic): Remove function.
21973 (arc_must_save_register): We use single base PIC register, remove
21974 checks to save/restore the PIC register.
21975 (arc_expand_prologue): Likewise.
21976 * config/arc/arc-protos.h (arc_set_default_type_attributes):
21978 (arc_verify_short): Likewise.
21979 (arc_attr_type): Likewise.
21980 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
21981 (walk_stores): Likewise.
21982 (arc_address_cost): Make it static.
21983 (arc_verify_short): Likewise.
21984 (branch_dest): Likewise.
21985 (arc_attr_type): Likewise.
21986 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
21987 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
21988 (arc_final_prescan_insn): Remove inserting the nops due to
21989 hardware hazards. It is done in reorg step.
21990 (insn_length_variant_t): Remove.
21991 (insn_length_parameters_t): Likewise.
21992 (arc_insn_length_parameters): Likewise.
21993 (arc_get_insn_variants): Likewise.
21994 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
21996 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21998 PR inline-asm/84683
21999 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
22002 PR tree-optimization/84687
22003 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
22005 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
22007 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22009 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
22010 Rename to ppc_speculation_barrier.
22011 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
22012 __builtin_ppc_speculation_barrier.
22014 2018-03-05 Jakub Jelinek <jakub@redhat.com>
22017 * combine.c (combine_simplify_rtx): Don't try to simplify if
22018 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
22021 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
22023 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
22024 to 32 bytes when compiling for POWER9.
22026 2018-03-05 Jakub Jelinek <jakub@redhat.com>
22029 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
22030 regparm >= 3 with no arg reg available also for calls with
22031 flag_force_indirect_call. Pass decl to ix86_function_regparm.
22034 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
22036 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
22038 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
22041 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
22043 2018-03-05 Richard Biener <rguenther@suse.de>
22045 PR tree-optimization/84486
22046 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
22047 When inserting a __builtin_assume_aligned call set the LHS
22048 SSA name alignment info accordingly.
22050 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
22052 PR tree-optimization/84114
22053 * config/aarch64/aarch64.c (aarch64_reassociation_width)
22054 Avoid reassociation of FLOAT_MODE addition.
22056 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
22058 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
22059 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
22060 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
22061 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
22062 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
22063 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
22064 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
22066 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
22067 __builtin_ia32_wbinvd): New builtins.
22068 (SPECIAL_ARGS2): New.
22069 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
22070 (SPECIAL_ARGS2): New.
22071 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
22072 (ix86_valid_target_attribute_inner_p): Ditto.
22073 (ix86_init_mmx_sse_builtins): Add special_args2.
22074 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
22075 TARGET_WBNOINVD_P): New.
22076 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
22077 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
22078 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
22079 * config/i386/immintrin.h (_wbinvd): New intrinsic.
22080 * config/i386/pconfigintrin.h: New file.
22081 * config/i386/wbnoinvdintrin.h: Ditto.
22082 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
22084 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
22086 2018-03-05 Richard Biener <rguenther@suse.de>
22088 PR tree-optimization/84670
22089 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
22091 (BB_VISITED_WITH_VISITED_SUCCS): New define.
22092 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
22093 (compute_antic_aux): Only assert the number of values in ANTIC_IN
22094 doesn't grow if all successors (recursively) were visited at least
22097 2018-03-05 Richard Biener <rguenther@suse.de>
22099 PR tree-optimization/84650
22100 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
22101 if executed in the loop pipeline.
22103 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
22105 * doc/configfiles.texi (Configuration Files): Move info about
22106 conditionalizing $target-protos.h to...
22107 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
22108 differs from $target-protos.h.
22110 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
22111 Chung-Ju Wu <jasonwucj@gmail.com>
22113 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
22114 * config/nds32/nds32-multiple.md (setmemsi): Define.
22115 * config/nds32/nds32-memory-manipulation.c
22116 (nds32_gen_dup_4_byte_to_word_value): New.
22117 (emit_setmem_word_loop): New.
22118 (emit_setmem_byte_loop): New.
22119 (nds32_expand_setmem_loop): New.
22120 (nds32_expand_setmem_loop_v3m): New.
22121 (nds32_expand_setmem_unroll): New.
22122 (nds32_expand_setmem): New.
22124 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22125 Chung-Ju Wu <jasonwucj@gmail.com>
22127 * config/nds32/nds32-memory-manipulation.c
22128 (nds32_emit_load_store): New.
22129 (nds32_emit_post_inc_load_store): New.
22130 (nds32_emit_mem_move): New.
22131 (nds32_emit_mem_move_block): New.
22132 (nds32_expand_movmemsi_loop_unknown_size): New.
22133 (nds32_expand_movmemsi_loop_known_size): New.
22134 (nds32_expand_movmemsi_loop): New.
22135 (nds32_expand_movmemsi_unroll): New.
22136 (nds32_expand_movmemqi): Rename ...
22137 (nds32_expand_movmemsi): ... to this.
22138 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
22139 (movmemsi): ... to this.
22140 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
22141 (nds32_expand_movmemsi): ... to this.
22143 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22144 Monk Chiang <sh.chiang04@gmail.com>
22145 Chung-Ju Wu <jasonwucj@gmail.com>
22147 * config/nds32/nds32-protos.h
22148 (nds32_expand_load_multiple): New arguments.
22149 (nds32_expand_store_multiple): Ditto.
22150 (nds32_valid_multiple_load_store): Rename ...
22151 (nds32_valid_multiple_load_store_p): ... to this.
22152 * config/nds32/nds32-memory-manipulation.c
22153 (nds32_expand_load_multiple): Refine implementation.
22154 (nds32_expand_store_multiple): Ditto.
22155 * config/nds32/nds32-multiple.md
22156 (load_multiple): Update nds32_expand_load_multiple interface.
22157 (store_multiple): Update nds32_expand_store_multiple interface.
22158 * config/nds32/nds32-predicates.c
22159 (nds32_valid_multiple_load_store): Rename ...
22160 (nds32_valid_multiple_load_store_p): ... to this and refine
22162 * config/nds32/predicates.md
22163 (nds32_load_multiple_and_update_address_operation): New predicate.
22164 (nds32_store_multiple_and_update_address_operation): New predicate.
22166 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22167 Chung-Ju Wu <jasonwucj@gmail.com>
22169 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
22170 (combo): New attribute.
22171 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
22173 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
22175 * config/nds32/nds32.opt: Change -mcmodel= default value.
22177 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
22178 Monk Chiang <sh.chiang04@gmail.com>
22179 Chung-Ju Wu <jasonwucj@gmail.com>
22181 * config/nds32/constants.md (unspec_element): New enum.
22182 * config/nds32/constraints.md (Umw): New constraint.
22183 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
22184 * config/nds32/nds32-intrinsic.md: Likewise.
22185 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
22186 (nds32_valid_smw_lwm_base_p): New.
22187 (nds32_output_smw_single_word): New.
22188 (nds32_output_lmw_single_word): New.
22189 (nds32_expand_unaligned_load): New.
22190 (nds32_expand_unaligned_store): New.
22191 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
22192 (nds32_output_smw_single_word): Declare.
22193 (nds32_output_lmw_single_word): Declare.
22194 (nds32_expand_unaligned_load): Declare.
22195 (nds32_expand_unaligned_store): Declare.
22196 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
22197 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
22198 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
22199 NDS32_BUILTIN_UASTORE_DW.
22200 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
22203 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
22204 Kito Cheng <kito.cheng@gmail.com>
22205 Chung-Ju Wu <jasonwucj@gmail.com>
22207 * config/nds32/nds32-intrinsic.c
22208 (nds32_expand_builtin_null_ftype_reg): Delete.
22209 (nds32_expand_builtin_reg_ftype_imm): Ditto.
22210 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
22211 (nds32_read_argument): New.
22212 (nds32_legitimize_target): Ditto.
22213 (nds32_legitimize_argument): Ditto.
22214 (nds32_check_constant_argument): Ditto.
22215 (nds32_expand_unop_builtin): Ditto.
22216 (nds32_expand_unopimm_builtin): Ditto.
22217 (nds32_expand_binop_builtin): Ditto.
22218 (nds32_builtin_decl_impl): Ditto.
22219 (builtin_description): Ditto.
22220 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
22221 (nds32_init_builtins_impl): Ditto.
22222 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
22223 (nds32_builtin_decl): New.
22224 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
22225 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
22227 2018-03-02 Jeff Law <law@redhat.com>
22229 * reorg.c (stop_search_p): Handle DEBUG_INSN.
22230 (redundant_insn, fill_simple_delay_slots): Likewise.
22231 (fill_slots_from_thread): Likewise.
22232 * resource.c (mark_referenced_resources): Likewise.
22233 (mark_set_resources, find_dead_or_set_registers): Likewise.
22235 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22237 * substring-locations.h (format_warning_va): Formatting fix for
22238 ATTRIBUTE_GCC_DIAG.
22239 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
22241 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
22242 * substring-locations.c: Include intl.h.
22243 (format_warning_va): Turned into small wrapper around
22244 format_warning_n_va, renamed to ...
22245 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
22246 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
22248 (format_warning_at_substring_n): New function.
22249 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
22250 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
22251 format_warning_at_substring with just a shorter name instead of
22252 const function pointer.
22253 (fmtwarn_n): New function.
22254 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
22255 appropriate, get rid of all the fmtstr temporaries, move conditionals
22256 with G_() wrapped string literals directly into fmtwarn arguments,
22257 cast dir.len to (int), formatting fixes.
22259 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
22261 * doc/invoke.texi: Remove "Cilk Plus" references.
22263 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22264 Richard Biener <rguenther@suse.de>
22267 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
22268 for error or warning attributes if CALL_FROM_THUNK_P is set.
22271 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22274 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
22275 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
22278 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
22279 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
22281 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
22282 instead of -1U in last predictors element's probability member.
22284 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
22287 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
22288 arguments if they are comparable.
22290 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
22292 PR tree-optimization/84634
22293 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
22294 masks and masked_loop_p with a single loop_masks, making sure it's
22295 null for bb vectorization.
22297 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
22299 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
22300 (vect_analyze_data_ref_access): Use loop->safe_len rather than
22301 loop->force_vectorize to check whether there is no alias.
22303 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22306 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
22308 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
22310 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
22311 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
22312 instead of a loop around prev_real_insn.
22313 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
22316 PR inline-asm/84625
22317 * config/i386/i386.c (ix86_print_operand): Use conditional
22318 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
22321 2018-03-02 Richard Biener <rguenther@suse.de>
22323 PR tree-optimization/84427
22324 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
22325 (bitmap_set_subtract_values): Rewrite to handle multiple
22328 (prune_clobbered_mems): Likewise.
22329 (phi_translate): Take edge instead of pred/phiblock.
22330 (phi_translate_1): Likewise.
22331 (phi_translate_set): Likewise. Insert all translated
22332 exprs for a value into the set, keeping possibly multiple
22333 expressions per value.
22334 (compute_antic_aux): Adjust for phi_translate changes.
22335 When intersecting union the expressions and prune those
22336 not in the final value set, keeping possibly multiple
22337 expressions per value. Do not use value-insertion
22338 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
22339 all expressions. Add verification that the value-sets
22340 only shrink during iteration.
22341 (compute_partial_antic_aux): Adjust for the phi_translate changes.
22342 (do_pre_regular_insertion): Likewise.
22343 (do_pre_partial_partial_insertion): Likewise.
22345 2018-03-02 Richard Biener <rguenther@suse.de>
22348 * config/darwin.c (saved_debug_info_level): New static global.
22349 (darwin_asm_lto_start): Disable debug info generation for LTO out.
22350 (darwin_asm_lto_end): Restore debug info generation settings.
22352 2018-03-01 Martin Liska <mliska@suse.cz>
22355 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
22356 volatile arguments.
22358 2018-03-01 Richard Biener <rguenther@suse.de>
22361 * dwarf2out.c (gen_variable_die): Properly handle late VLA
22362 type annotation with LTO when debug was disabled at compile-time.
22364 2018-03-01 Matthew Fortune <mfortune@gmail.com>
22366 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
22368 (mips_final_postscan_insn): Likewise.
22370 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
22372 PR rtl-optimization/84528
22373 * alias.c (init_alias_target): Add commentary.
22374 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
22375 a unique base value if the frame pointer is not eliminated
22376 to the stack pointer.
22378 2018-03-01 Tom de Vries <tom@codesourcery.com>
22380 PR rtl-optimization/83327
22381 * lra-int.h (hard_regs_spilled_into): Declare.
22382 * lra.c (hard_regs_spilled_into): Define.
22383 (init_reg_info): Init hard_regs_spilled_into.
22384 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
22385 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
22386 (process_bb_lives): Handle hard_regs_spilled_into.
22387 (lra_create_live_ranges_1): Before doing liveness propagation, clear
22388 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
22390 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
22392 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
22393 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
22394 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
22395 * config/rs6000/aix72.h: New file.
22397 2018-02-28 Jakub Jelinek <jakub@redhat.com>
22399 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
22400 instead of warning_at with conditional singular and plural messages
22404 * stor-layout.c (update_alignment_for_field): For
22405 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
22406 && !DECL_PACKED (field), do the alignment update, just use
22407 only desired_align instead of MAX (type_align, desired_align)
22409 (place_field): Don't do known_align < desired_align handling
22410 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
22411 is non-NULL, instead do it after rli->prev_field handling and
22412 only if not within a bitfield word. For DECL_PACKED (field)
22413 use type_align of BITS_PER_UNIT.
22415 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
22417 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
22418 superfluous parentheses and trailing spaces.
22420 2018-02-28 Richard Biener <rguenther@suse.de>
22422 PR tree-optimization/84584
22423 * graphite-scop-detection.c (scop_detection::add_scop): Discard
22424 SCoPs with fake exit edge.
22426 2018-02-28 Martin Liska <mliska@suse.cz>
22429 * timevar.c (timer::print): Fix format to properly print 100%
22432 2018-02-28 Richard Biener <rguenther@suse.de>
22434 PR middle-end/84607
22435 * genmatch.c (capture_info::walk_match): Do not mark
22436 captured expressions without operands as expr_p given
22437 they act more like predicates and should be subject to
22438 "lost tail" side-effect preserving.
22440 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
22442 PR rtl-optimization/81611
22443 * auto-inc-dec.c (attempt_change): Move dead note from
22444 mem_insn if it's the next use of regno
22445 (find_address): Take address use of reg holding
22446 non-incremented value. Add parm to limit search to the named
22448 (merge_in_block): Attempt to use a mem insn that is the next
22449 use of the original regno.
22451 2018-02-27 Martin Sebor <msebor@redhat.com>
22454 * doc/invoke.texi (-Wmissing-attributes): New option.
22455 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
22457 2018-02-27 Martin Sebor <msebor@redhat.com>
22459 PR translation/84207
22460 * diagnostic-core.h (warning_n, error_n, inform_n): Change
22461 n argument to unsigned HOST_WIDE_INT.
22462 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
22463 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
22464 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
22465 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
22467 2018-02-27 Richard Biener <rguenther@suse.de>
22469 PR tree-optimization/84512
22470 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
22471 Do not use the estimate returned from record_stmt_cost for
22472 the scalar iteration cost but sum properly using add_stmt_cost.
22474 2018-02-27 Richard Biener <rguenther@suse.de>
22476 PR tree-optimization/84466
22477 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
22478 Adjust last change to less strictly validate use operands.
22480 2018-02-27 Martin Liska <mliska@suse.cz>
22482 PR gcov-profile/84548
22483 * gcov.c (process_file): Allow partial overlap and consider it
22484 also as group functions.
22485 (output_lines): Properly calculate range of lines for a group.
22487 2018-02-27 Martin Liska <mliska@suse.cz>
22489 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
22490 'ggc' suffixes. Change first column width.
22491 (timer::print): Fix formatting of the column.
22493 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
22495 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
22496 preserve inline entry blocks for the sake of debug inline
22497 entry point markers alone.
22498 (remove_unused_locals): Suggest in comments a better place to
22499 force the preservation of inline entry blocks that are
22500 otherwise unused, but do not preserve them.
22502 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22504 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
22506 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22509 * config/i386/constraints.md (Bs): Replace
22510 ix86_indirect_branch_register with
22511 TARGET_INDIRECT_BRANCH_REGISTER.
22513 * config/i386/i386.md (indirect_jump): Likewise.
22514 (tablejump): Likewise.
22515 (*sibcall_memory): Likewise.
22516 (*sibcall_value_memory): Likewise.
22517 Peepholes of indirect call and jump via memory: Likewise.
22518 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
22519 (*sibcall_value_GOT_32): Likewise.
22520 * config/i386/predicates.md (indirect_branch_operand): Likewise.
22521 (GOT_memory_operand): Likewise.
22522 (call_insn_operand): Likewise.
22523 (sibcall_insn_operand): Likewise.
22524 (GOT32_symbol_operand): Likewise.
22525 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
22527 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22529 PR rtl-optimization/83496
22530 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
22531 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
22532 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
22533 redundant insn, if any.
22534 (relax_delay_slots): Likewise.
22535 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
22537 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
22539 PR tree-optimization/83965
22540 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
22541 that grouped statements are part of a reduction chain. Return
22542 true if the statement is not marked as a reduction itself but
22543 is part of a group.
22544 (vect_recog_dot_prod_pattern): Don't check whether the statement
22545 is part of a group here.
22546 (vect_recog_sad_pattern): Likewise.
22547 (vect_recog_widen_sum_pattern): Likewise.
22549 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22552 * final.c (rest_of_clean_state): Also look for calls inside sequences.
22554 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22557 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
22559 (ix86_output_indirect_function_return): New prototype.
22560 (ix86_split_simple_return_pop_internal): Likewise.
22561 * config/i386/i386.c (indirect_return_via_cx): New.
22562 (indirect_return_via_cx_bnd): Likewise.
22563 (indirect_thunk_name): Handle return va CX_REG.
22564 (output_indirect_thunk_function): Create alias for
22565 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
22566 (ix86_output_indirect_jmp): Remove the bool argument.
22567 (ix86_output_indirect_function_return): New function.
22568 (ix86_split_simple_return_pop_internal): Likewise.
22569 * config/i386/i386.md (*indirect_jump): Don't pass false
22570 to ix86_output_indirect_jmp.
22571 (*tablejump_1): Likewise.
22572 (simple_return_pop_internal): Change it to define_insn_and_split.
22573 Call ix86_split_simple_return_pop_internal to split it for
22574 -mfunction-return=.
22575 (simple_return_indirect_internal): Call
22576 ix86_output_indirect_function_return instead of
22577 ix86_output_indirect_jmp.
22579 2018-02-26 Jakub Jelinek <jakub@redhat.com>
22582 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
22583 memset and value initialization afterwards.
22585 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
22587 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
22589 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22592 * common/config/aarch64/aarch64-common.c
22593 (aarch_option_optimization_table[]): Switch
22594 off fomit-frame-pointer
22596 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22597 Chung-Ju Wu <jasonwucj@gmail.com>
22599 * config/nds32/nds32-multiple.md (load_multiple): Disallow
22601 (store_multiple): Ditto.
22603 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22605 * config.gcc: Add --with-cpu support for nds32 target.
22606 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
22607 * config/nds32/nds32.opt: Add -mcpu= option.
22609 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
22611 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
22612 isel=yes): Warn for these deprecated options.
22614 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
22616 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
22617 ISA_2_5_MASKS_EMBEDDED.
22619 2018-02-23 Jakub Jelinek <jakub@redhat.com>
22621 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
22622 p->max as pointers rather than using iterative_hash_expr.
22624 2018-02-23 Carl Love <cel@us.ibm.com>
22626 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
22627 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
22629 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
22630 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
22631 P8V_BUILTIN_VEC_VUNSIGNED2.
22633 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
22636 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
22637 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
22639 * lra-constraints.c (curr_insn_transform): Set up
22640 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
22642 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
22643 * lra-eliminations.c (spill_pseudos): Ditto.
22644 (process_insn_for_elimination): Ditto.
22645 * lra-lives.c (reg_early_clobber_p): Use the new macros.
22646 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
22647 LRA_NON_CLOBBERED_ALT.
22649 2018-02-22 Martin Sebor <msebor@redhat.com>
22651 PR tree-optimization/84480
22652 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
22653 to maybe_diag_stxncpy_trunc. Call it.
22654 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
22655 from gimple_fold_builtin_strcpy. Print inlining stack.
22656 (handle_builtin_stxncpy): Print inlining stack.
22657 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
22659 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
22662 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
22663 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
22664 and -fcheck-pointer-bounds are used together.
22665 (indirect_thunk_prefix): New enum.
22666 (indirect_thunk_need_prefix): New function.
22667 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
22668 "_nt" instead of "_bnd" for NOTRACK prefix.
22669 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
22670 (output_indirect_thunk_function): Likewise.
22672 (ix86_code_end): Update output_indirect_thunk_function calls.
22673 (ix86_output_indirect_branch_via_reg): Replace
22674 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
22675 (ix86_output_indirect_branch_via_push): Likewise.
22676 (ix86_output_function_return): Likewise.
22677 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
22678 incompatible with -fcf-protection=branch and
22679 -fcheck-pointer-bounds.
22681 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22684 * config/aarch64/aarch64.c (aarch64_print_address_internal):
22685 Change gcc_assert call to output_operand_lossage.
22687 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22689 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
22691 2018-02-22 DJ Delorie <dj@redhat.com>
22692 Sebastian Perta <sebastian.perta@renesas.com>
22693 Oleg Endo <olegendo@gcc.gnu.org>
22695 * config/rx/rx.c (rx_rtx_costs): New function.
22696 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
22698 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22700 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
22702 2018-02-22 Martin Liska <mliska@suse.cz>
22705 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
22706 Add "native" as a possible value.
22708 2018-02-22 Martin Liska <mliska@suse.cz>
22711 * config/i386/i386.c (ix86_option_override_internal):
22712 Add "native" as a possible value for -march and -mtune.
22714 2018-02-22 Jakub Jelinek <jakub@redhat.com>
22717 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
22718 to all type variants.
22720 PR tree-optimization/84503
22721 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
22722 width as info->bitpos + info->bitsize - start.
22723 (merged_store_group::merge_overlapping): Simplify width computation.
22724 (check_no_overlap): New function.
22725 (imm_store_chain_info::try_coalesce_bswap): Compute expected
22726 start + width and last_order of the group, fail if check_no_overlap
22728 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
22729 to group if check_no_overlap fails.
22731 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22733 * config/rs6000/altivec.md: Delete contraint arguments to
22734 define_expand, define_split, and define_peephole2, and in
22735 define_insn_and_split if always unused.
22736 * config/rs6000/darwin.md: Ditto.
22737 * config/rs6000/dfp.md: Ditto.
22738 * config/rs6000/rs6000.md: Ditto.
22739 * config/rs6000/sync.md: Ditto.
22740 * config/rs6000/vector.md: Ditto.
22741 * config/rs6000/vsx.md: Ditto.
22743 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22745 * config/rs6000/altivec.md: Write output control strings as braced
22746 blocks instead of double-quoted strings.
22747 * config/rs6000/darwin.md: Ditto.
22748 * config/rs6000/rs6000.md: Ditto.
22749 * config/rs6000/vector.md: Ditto.
22750 * config/rs6000/vsx.md: Ditto.
22752 2018-02-21 Jason Merrill <jason@redhat.com>
22754 PR c++/84314 - ICE with templates and fastcall attribute.
22755 * attribs.c (build_type_attribute_qual_variant): Remove assert.
22757 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22759 * ipa-cp.c (determine_versionability): Fix comment typos.
22761 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22764 * ipa-cp.c (determine_versionability): Do not version functions caling
22767 2018-02-21 Martin Liska <mliska@suse.cz>
22770 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
22771 Add "native" as a possible value.
22772 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
22773 the macro when native cpu detection is available.
22775 2018-02-21 Martin Liska <mliska@suse.cz>
22778 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
22779 Add "native" as a possible value.
22780 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
22781 when native cpu detection is available.
22783 2018-02-21 Jakub Jelinek <jakub@redhat.com>
22784 Martin Sebor <msebor@redhat.com>
22786 PR tree-optimization/84478
22787 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
22789 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
22790 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
22791 support which is conservatively correct, for 2 only stay conservative
22792 for maxlen. Formatting and comment capitalization fixes. Add STRICT
22793 argument to the 2 argument get_range_strlen, adjust 6 arg
22794 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
22796 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
22797 (gimple_fold_builtin_strlen): Pass true as last argument to
22800 2018-02-20 Martin Sebor <msebor@redhat.com>
22802 PR middle-end/84095
22803 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
22804 (builtin_memref::set_base_and_offset): Same. Handle inner references.
22805 (builtin_memref::builtin_memref): Factor out parts into
22806 set_base_and_offset and call it.
22808 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22810 PR middle-end/84406
22811 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
22812 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
22813 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
22814 search at the associated MODE_INT.
22816 2018-02-20 Jeff Law <law@redhat.com>
22818 PR middle-end/82123
22819 PR tree-optimization/81592
22820 PR middle-end/79257
22821 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
22822 for range data rather than using global data.
22823 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
22824 range data rather than using global data.
22825 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
22826 pass it to children as needed.
22827 (struct directive::fmtresult): Similarly.
22828 (struct directive::set_width): Similarly.
22829 (struct directive::set_precision): Similarly.
22830 (format_integer, format_directive, parse_directive): Similarly.
22831 (format_none): Accept unnamed vr_values parameter.
22832 (format_percent, format_floating, format_character): Similarly.
22833 (format_string, format_plain): Similarly.
22834 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
22835 the EVRP range analyzer for range data rather than using global data.
22836 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
22837 gimple-ssa-evrp-analyze.h
22838 (class sprintf_dom_walker): Add after_dom_children member function.
22839 Add evrp_range_analyzer member.
22840 (sprintf_dom_walker::before_dom_children): Call into the EVRP
22841 range analyzer as needed.
22842 (sprintf_dom_walker::after_dom_children): New member function.
22843 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
22845 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
22846 (evrp_range_analyzer::pop_to_marker): Likewise.
22848 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22850 PR tree-optimization/84419
22851 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
22852 with the required type if its current type is compatible but
22855 2018-02-20 Jakub Jelinek <jakub@redhat.com>
22857 PR middle-end/82004
22858 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
22859 after vectorization.
22861 2018-02-20 Martin Liska <mliska@suse.cz>
22864 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
22865 possible values if we don't have a hint.
22867 2018-02-20 Martin Liska <mliska@suse.cz>
22871 * final.c (shorten_branches): Build align_tab array with one
22873 * opts.c (finish_options): Add alignment option limit check.
22874 (MAX_CODE_ALIGN): Likewise.
22875 (MAX_CODE_ALIGN_VALUE): Likewise.
22876 * doc/invoke.texi: Document maximum allowed option value for
22877 all -falign-* options.
22879 2018-02-19 Jakub Jelinek <jakub@redhat.com>
22882 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
22883 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
22884 * var-tracking.c (emit_note_insn_var_location): Remove all references
22885 to NOTE_INSN_CALL_ARG_LOCATION.
22886 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
22887 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
22888 Use copy_rtx_if_shared.
22889 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
22890 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
22891 (dwarf2out_var_location): Remove handling of
22892 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
22894 * final.c (final_scan_insn): Remove all references to
22895 NOTE_INSN_CALL_ARG_LOCATION.
22896 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
22897 before dumping final insns.
22898 * except.c (emit_note_eh_region_end): Remove all references to
22899 NOTE_INSN_CALL_ARG_LOCATION.
22900 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
22901 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
22902 * config/arc/arc.c (hwloop_optimize): Likewise.
22903 * config/arm/arm.c (create_fix_barrier): Likewise.
22904 * config/s390/s390.c (s390_chunkify_start): Likewise.
22905 * config/sh/sh.c (find_barrier): Likewise.
22906 * config/i386/i386.c (rest_of_insert_endbranch,
22907 ix86_seh_fixup_eh_fallthru): Likewise.
22908 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
22909 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
22910 * config/frv/frv.c (frv_function_prologue): Likewise.
22911 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
22913 (note_outside_basic_block_p): Remove all references to
22914 NOTE_INSN_CALL_ARG_LOCATION.
22915 * gengtype.c (adjust_field_rtx_def): Likewise.
22916 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
22918 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
22919 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
22922 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
22925 PR tree-optimization/84452
22926 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
22927 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
22930 2018-02-19 Martin Liska <mliska@suse.cz>
22933 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
22935 2018-02-19 Martin Liska <mliska@suse.cz>
22936 Richard Sandiford <richard.sandiford@linaro.org>
22938 PR tree-optimization/82491
22939 * gimple-fold.c (get_base_constructor): Make earlier bail out
22942 2018-02-19 Carl Love <cel@us.ibm.com>
22944 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
22945 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
22947 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
22948 P8V_BUILTIN_VEC_NEG.
22950 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
22952 * config/rl78/rl78.md (movdf): New define expand.
22954 2018-02-19 Martin Liska <mliska@suse.cz>
22957 * doc/invoke.texi: Fix typo.
22958 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
22960 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
22962 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
22963 handle rs6000_single_float and rs6000_double_float specially for
22966 2018-02-16 Jeff Law <law@redhat.com>
22968 * config/rx/rx.c (add_pop_cfi_notes): New function.;
22969 (pop_regs): Use it.
22971 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22974 * ipa-inline.c (inline_small_functions): Fix a typo.
22976 2018-02-16 Nathan Sidwell <nathan@acm.org>
22978 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
22980 2018-02-16 Carl Love <cel@us.ibm.com>
22982 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
22983 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
22984 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
22985 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
22986 expansion to P8V_BUILTIN_VEC_FLOAT2.
22988 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
22990 PR rtl-optimization/70023
22991 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
22992 src_regno into account.
22994 2018-02-16 Carl Love <cel@us.ibm.com>
22996 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
22997 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
22998 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
22999 * config/rs6000/rs6000.c: Remove case statements for
23000 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
23001 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
23002 and P9V_BUILTIN_VEC_VINSERT4B.
23003 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
23004 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
23005 * config/rs6000/vsx.md:
23006 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
23009 2018-02-16 Carl Love <cel@us.ibm.com>
23011 * config/rs6000/altivec.h: Add builtin names vec_extract4b
23013 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
23015 * config/rs6000/rs6000-c.c: Add the definitions for
23016 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
23017 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
23018 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
23019 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
23020 definition for insert4b and define insn *insert3b_internal.
23021 * doc/extend.texi: Add documentation for vec_extract4b.
23023 2018-02-16 Nathan Sidwell <nathan@acm.org>
23025 * doc/extend.texi (Backwards Compatibility): Mention friend
23026 injection. Note for-scope is deprecated.
23027 * doc/invoke.texi (-ffriend-injection): Deprecate.
23029 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
23031 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
23032 that moved to I2, also allow destinations that are a paradoxical
23033 subreg (instead of a normal reg).
23035 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
23038 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
23041 2018-02-16 Richard Biener <rguenther@suse.de>
23043 PR tree-optimization/84037
23044 PR tree-optimization/84016
23046 * config/i386/i386.c (ix86_builtin_vectorization_cost):
23047 Adjust vec_construct for the fact we need additional higher latency
23048 128bit inserts for AVX256 and AVX512 vector builds.
23049 (ix86_add_stmt_cost): Scale vector construction cost for
23052 2018-02-16 Richard Biener <rguenther@suse.de>
23054 PR tree-optimization/84417
23055 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
23056 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
23057 (non_rewritable_lvalue_p): Likewise, use poly-ints.
23059 2018-02-16 Martin Liska <mliska@suse.cz>
23062 * internal-fn.def (ASAN_CHECK): Set proper flags.
23063 (ASAN_MARK): Likewise.
23065 2018-02-16 Julia Koval <julia.koval@intel.com>
23067 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
23068 from PTA_CANNONLAKE.
23070 2018-02-16 Jakub Jelinek <jakub@redhat.com>
23073 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
23074 Use ++iter rather than iter++ for std::list iterators.
23075 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
23076 defer deleting them until all nodes in the forest are processed. Do
23077 free even leaf nodes. Change to_process into auto_vec.
23080 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
23081 * vec.h (vec_default_construct): Use memset instead of placement new
23082 if BROKEN_VALUE_INITIALIZATION is defined.
23083 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
23084 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
23087 PR rtl-optimization/83723
23088 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
23089 * lra.c (lra_substitute_pseudo): Likewise. If true, use
23090 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
23092 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
23094 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
23096 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
23098 PR rtl-optimization/81443
23099 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
23100 from inner REGs to paradoxical SUBREGs.
23102 2018-02-16 Richard Biener <rguenther@suse.de>
23104 PR tree-optimization/84399
23105 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
23106 For operands we can analyze at their definition make sure we can
23107 analyze them at each use as well.
23109 2018-02-16 Richard Biener <rguenther@suse.de>
23111 PR tree-optimization/84190
23112 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
23113 volatile accesses if the decl isn't volatile.
23115 2018-02-15 Jason Merrill <jason@redhat.com>
23117 PR c++/84314 - ICE with templates and fastcall attribute.
23118 * attribs.c (build_type_attribute_qual_variant): Don't clobber
23119 TYPE_CANONICAL on an existing type.
23121 2018-02-15 Jakub Jelinek <jakub@redhat.com>
23123 PR tree-optimization/84383
23124 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
23125 dstoff nor call operand_equal_p if dstbase is NULL.
23127 PR tree-optimization/84334
23128 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
23129 also a CONSTANT_CLASS_P, punt.
23131 2018-02-14 Jim Wilson <jimw@sifive.com>
23133 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
23134 first SMALL_OPERAND check. New local min_second_step. Move assert
23135 to where locals are set. Add TARGET_RVC support.
23136 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
23138 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
23140 * doc/invoke.texi: Correct -Wformat-overflow code sample.
23142 2018-02-14 Martin Sebor <msebor@redhat.com>
23144 PR tree-optimization/83698
23145 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
23146 arrays constrain the offset range to their bounds.
23147 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
23148 (builtin_access::overlap): Avoid setting the size of overlap if it's
23150 (maybe_diag_overlap): Also consider arrays when deciding what values
23151 of offsets to include in diagnostics.
23153 2018-02-14 Martin Sebor <msebor@redhat.com>
23156 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
23157 that correspond to the kind of a declaration.
23159 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
23162 * config/pa/pa.md: Load address of PIC label using the linkage table
23163 if the label is nonlocal.
23165 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23167 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
23168 warning message if user requests -maltivec=be.
23169 * doc/invoke.texi: Document deprecation of -maltivec=be.
23171 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
23174 * config/rs6000/rs6000-c.c: Update definitions for
23175 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
23176 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
23178 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23181 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
23182 add _get_ssp intrinsics. Remove argument from
23183 __builtin_ia32_rdssp[d|q].
23184 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
23185 * config/i386/i386-builtin.def: Remove argument from
23186 __builtin_ia32_rdssp[d|q].
23187 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
23188 ix86_expand_special_args_builtin for _rdssp[d|q].
23189 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
23190 Clear register before usage.
23191 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
23192 Add documentation for new _get_ssp and _inc_ssp intrinsics.
23194 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
23196 PR tree-optimization/84357
23197 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
23198 operand 1 of an ARRAY_REF too.
23200 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
23203 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
23204 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
23206 (set_of_reg): New struct.
23207 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
23208 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
23209 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
23211 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
23212 Split into bitclr, bitset, bitinvert patterns if appropriate.
23213 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
23214 use rx_fuse_in_memory_bitop.
23215 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
23216 to named insn, correct maximum insn length.
23218 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
23221 * machmode.def: Define a complex mode for PARTIAL_INT.
23222 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
23224 * doc/rtl.texi: Document CSPImode.
23225 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
23227 (msp430_hard_regno_nregs_with_padding): Likewise.
23229 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
23232 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
23234 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23236 PR rtl-optimization/84169
23237 * combine.c (try_combine): New variable split_i2i3. Set it to true if
23238 we generated a parallel as new i3 and we split that to new i2 and i3
23239 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
23240 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
23241 those to i2, not i1. Partially rewrite this scan code.
23243 2018-02-13 Jakub Jelinek <jakub@redhat.com>
23246 * stor-layout.c (place_field): For variable length fields, adjust
23247 offset_align afterwards not just based on the field's alignment,
23248 but also on the size.
23250 PR middle-end/84309
23251 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
23252 of exps and logs in the use_exp2 case.
23254 2018-02-13 Jeff Law <law@redhat.com>
23256 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
23257 entry for "vector".
23259 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
23262 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
23266 * common.opt (gas-loc-support, gas-locview-support): New.
23267 (ginline-points, ginternal-reset-location-views): New.
23268 * doc/invoke.texi: Document them. Use @itemx where intended.
23269 (gvariable-location-views): Adjust.
23270 * target.def (reset_location_view): New.
23271 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
23272 (TARGET_RESET_LOCATION_VIEW): New.
23273 * doc/tm.texi: Rebuilt.
23274 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
23275 (dwarf2out_default_as_locview_support): New.
23276 (output_asm_line_debug_info): Use option variables.
23277 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
23278 (output_loc_list): Likewise.
23279 (add_high_low_attributes): Check option variables.
23280 Don't output entry view attribute in strict mode.
23281 (gen_inlined_subroutine_die): Check option variables.
23282 (dwarf2out_inline_entry): Likewise.
23283 (init_sections_and_labels): Likewise.
23284 (dwarf2out_early_finish): Likewise.
23285 (maybe_reset_location_view): New, from...
23286 (dwarf2out_var_location): ... here. Call it.
23287 * debug.h (dwarf2out_default_as_loc_support): Declare.
23288 (dwarf2out_default_as_locview_support): Declare.
23289 * hooks.c (hook_int_rtx_insn_0): New.
23290 * hooks.h (hook_int_rtx_insn_0): Declare.
23291 * toplev.c (process_options): Take -gas-loc-support and
23292 -gas-locview-support from dwarf2out. Enable
23293 -gvariable-location-views by default only with locview
23294 assembler support. Enable -ginternal-reset-location-views by
23295 default only if the target defines the corresponding hook.
23296 Enable -ginline-points by default if location views are
23297 enabled; force it disabled if statement frontiers are
23299 * tree-inline.c (expand_call_inline): Check option variables.
23300 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
23302 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
23304 PR tree-optimization/84321
23305 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
23306 handling. Also check whether the anti-range contains any values
23307 that satisfy the mask; switch to a VR_RANGE if not.
23309 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
23312 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
23314 2018-02-13 Martin Jambor <mjambor@suse.cz>
23317 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
23318 of call statements, also set location of a load to a temporary.
23320 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
23322 * config/rl78/rl78.c (add_vector_labels): New function.
23323 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
23324 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
23325 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
23326 which checks that no arguments are passed.
23327 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
23328 * doc/extend.texi: Documentation for the new attribute.
23330 2018-02-13 Andreas Schwab <schwab@suse.de>
23332 * config/riscv/linux.h (CPP_SPEC): Define.
23334 2018-02-13 Jakub Jelinek <jakub@redhat.com>
23337 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
23338 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
23339 OPTION_MASK_ISA_AES as first argument to def_builtin_const
23340 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
23341 instead of OPTION_MASK_ISA_PCLMUL as first argument to
23342 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
23343 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
23344 temporarily for AES and PCLMUL builtins.
23346 PR tree-optimization/84339
23347 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
23348 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
23351 PR middle-end/84309
23352 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
23353 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
23354 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
23356 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
23358 * omp-simd-clone.h: New file.
23359 * omp-simd-clone.c: Include omp-simd-clone.h.
23360 (expand_simd_clones): No longer static.
23361 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
23362 cgraph.h and omp-simd-clone.h.
23363 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
23364 (vect_recog_widen_shift_pattern): Formatting fix.
23365 (vect_pattern_recog_1): Don't check optab for calls.
23368 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
23369 operands[2] into a REG before using gen_lowpart on it.
23371 2018-02-12 Jeff Law <law@redhat.com>
23374 * config/sh/sh.c (find_barrier): Consider a sibling call
23377 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
23378 successfully back substituting a reg.
23380 2018-02-12 Richard Biener <rguenther@suse.de>
23382 PR tree-optimization/84037
23383 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
23384 parameter, move visited init to caller.
23385 (vect_slp_analyze_operations): Separate cost from validity
23386 check, initialize visited once for all instances.
23387 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
23389 * tree-vect-stmts.c (vect_model_simple_cost): Make early
23391 (vect_model_promotion_demotion_cost): Likewise.
23392 (vectorizable_bswap): Guard cost modeling with !slp_node
23393 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
23395 (vectorizable_call): Likewise.
23396 (vectorizable_conversion): Likewise.
23397 (vectorizable_assignment): Likewise.
23398 (vectorizable_shift): Likewise.
23399 (vectorizable_operation): Likewise.
23400 (vectorizable_store): Likewise.
23401 (vectorizable_load): Likewise.
23402 (vectorizable_condition): Likewise.
23403 (vectorizable_comparison): Likewise.
23405 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
23408 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
23409 (ASAN_MARK): Fix fnspec to account for return value, change pointer
23410 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
23412 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23414 PR middle-end/83665
23415 * params.def (inline-min-speedup): Increase from 8 to 15.
23416 (max-inline-insns-auto): Decrease from 40 to 30.
23417 * ipa-split.c (consider_split): Add some buffer for function to
23418 be considered inlining candidate.
23419 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
23422 2018-02-12 Richard Biener <rguenther@suse.de>
23424 PR tree-optimization/84037
23425 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
23426 matched stmts if we cannot swap the non-matched ones.
23428 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
23430 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
23431 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
23432 _mm_maskz_scalef_round_ss): New intrinsics.
23433 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
23434 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
23435 __builtin_ia32_scalefss_round): Remove.
23436 (__builtin_ia32_scalefsd_mask_round,
23437 __builtin_ia32_scalefss_mask_round): New intrinsics.
23438 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
23439 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
23440 ((match_operand:VF_128 2 "<round_nimm_predicate>"
23441 "<round_constraint>")): Changed to ...
23442 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
23443 "<round_scalar_constraint>")): ... this.
23444 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
23445 %0, %1, %2<round_op3>}"): Changed to ...
23446 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
23447 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
23448 %2<round_scalar_mask_op3>}"): ... this.
23449 * config/i386/subst.md (round_scalar_nimm_predicate): New.
23451 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
23453 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
23454 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
23455 (_mm_maskz_sqrt_round_ss): New intrinsics.
23456 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
23457 (__builtin_ia32_sqrtsd_mask_round)
23458 (__builtin_ia32_sqrtss_mask_round): New builtins.
23459 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
23460 (__builtin_ia32_sqrtss_round): Remove.
23461 (__builtin_ia32_sqrtsd_mask_round)
23462 (__builtin_ia32_sqrtss_mask_round): New builtins.
23463 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
23464 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
23465 ((match_operand:VF_128 1 "vector_operand"
23466 "xBm,<round_constraint>")): Changed to ...
23467 ((match_operand:VF_128 1 "vector_operand"
23468 "xBm,<round_scalar_constraint>")): ... this.
23469 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
23470 %0, %2, %<iptr>1<round_op3>}): Changed to ...
23471 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
23472 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
23473 %<iptr>1<round_scalar_mask_op3>}): ... this.
23474 ((set_attr "prefix" "<round_prefix>")): Changed to ...
23475 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
23477 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
23480 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
23481 Cast vec_cmpeq result to correct type.
23482 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
23483 Cast vec_cmpgt result to correct type.
23485 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
23487 * final.c (final_scan_insn_1): Renamed from...
23488 (final_scan_insn): ... this. New wrapper, to recover
23489 seen from the outermost call in recursive ones.
23490 * config/sparc/sparc.c (output_return): Drop seen from call.
23491 (output_sibcall): Likewise.
23492 * config/visium/visium.c (output_branch): Likewise.
23494 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
23496 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
23499 2018-02-10 Alan Modra <amodra@gmail.com>
23502 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
23503 Specify LR as an input.
23505 2018-02-10 Jakub Jelinek <jakub@redhat.com>
23508 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
23509 remove_member_access_dummy_vars): New functions.
23510 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
23511 lower_omp_1, execute_lower_omp): Use them.
23513 PR rtl-optimization/84308
23514 * shrink-wrap.c (spread_components): Release todo vector.
23516 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
23518 PR rtl-optimization/57193
23519 * ira-color.c (struct allocno_color_data): Add member
23520 conflict_allocno_hard_prefs.
23521 (update_conflict_allocno_hard_prefs): New.
23522 (bucket_allocno_compare_func): Add a preference based on
23523 conflict_allocno_hard_prefs.
23524 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
23525 (color_allocnos): Remove a dead code. Initiate
23526 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
23528 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23531 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
23532 constraint from =wa to wa. Avoid a subreg on the output operand,
23533 instead use a pseudo and subreg it in a move.
23534 (p9_xxbrd_<mode>): Changed to ...
23535 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
23536 (p9_xxbrd_v2df): New expander.
23537 (p9_xxbrw_<mode>): Changed to ...
23538 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
23539 (p9_xxbrw_v4sf): New expander.
23541 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23543 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
23545 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
23548 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
23549 multiply in 32-bit mode.
23550 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
23551 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
23554 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23556 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
23557 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
23558 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
23559 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
23561 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
23564 * dwarf2out.c (is_trivial_indirect_ref): New function.
23565 (dwarf2out_late_global_decl): Do not generate a location
23566 attribute for variables that have a non-trivial DECL_VALUE_EXPR
23567 and that are not defined in the current unit.
23569 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23571 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
23572 instead of a libcall for UNORDERED.
23574 2018-02-09 Tamar Christina <tamar.christina@arm.com>
23577 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
23578 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
23580 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23583 * config/s390/s390.c (s390_set_current_function): Invoke
23584 s390_indirect_branch_settings also if fndecl didn't change.
23586 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23588 * config/rs6000/rs6000.md (blockage): Set length to zero.
23590 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23592 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
23594 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23597 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
23598 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
23602 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
23603 PARALLEL incoming that failed vt_get_decl_and_offset check.
23605 PR middle-end/84237
23606 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
23607 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
23609 (get_variable_section): For decls in named .bss* sections pass true as
23610 second argument to bss_initializer_p.
23612 2018-02-09 Marek Polacek <polacek@redhat.com>
23613 Jakub Jelinek <jakub@redhat.com>
23616 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
23617 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
23618 Sync some changes from cxx_fold_indirect_ref.
23620 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23622 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
23624 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
23625 (BLOCK_INLINE_ENTRY_LABEL): New.
23626 (dwarf2out_var_location): Disregard inline entry markers.
23627 (inline_entry_data): New struct.
23628 (inline_entry_data_hasher): New hashtable type.
23629 (inline_entry_data_hasher::hash): New.
23630 (inline_entry_data_hasher::equal): New.
23631 (inline_entry_data_table): New variable.
23632 (add_high_low_attributes): Add DW_AT_entry_pc and
23633 DW_AT_GNU_entry_view attributes if a pending entry is found
23634 in inline_entry_data_table. Add old entry_pc attribute only
23635 if debug nonbinding markers are disabled.
23636 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
23637 markers are enabled.
23638 (block_within_block_p, dwarf2out_inline_entry): New.
23639 (dwarf2out_finish): Check that no entries remained in
23640 inline_entry_data_table.
23641 * final.c (reemit_insn_block_notes): Handle inline entry notes.
23642 (final_scan_insn, notice_source_line): Likewise.
23643 (rest_of_clean_state): Skip inline entry markers.
23644 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
23646 * gimple.c (gimple_build_debug_inline_entry): New.
23647 * gimple.h (enum gimple_debug_subcode): Add
23648 GIMPLE_DEBUG_INLINE_ENTRY.
23649 (gimple_build_debug_inline_entry): Declare.
23650 (gimple_debug_inline_entry_p): New.
23651 (gimple_debug_nonbind_marker_p): Adjust.
23652 * insn-notes.def (INLINE_ENTRY): New.
23653 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
23654 inline entry marker notes.
23655 (print_insn): Likewise.
23656 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
23657 (INSN_DEBUG_MARKER_KIND): Likewise.
23658 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
23659 * tree-inline.c (expand_call_inline): Build and insert
23660 debug_inline_entry stmt.
23661 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
23662 inline entry blocks early, if nonbind markers are enabled.
23663 (dump_scope_block): Dump fragment info.
23664 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
23665 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
23666 (gimple_build_debug_inline_entry): New.
23667 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
23668 Enable/disable inline entry points too.
23669 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
23670 (DEBUG_INSN): Describe inline entry markers.
23672 * common.opt (gvariable-location-views): New.
23673 (gvariable-location-views=incompat5): New.
23674 * config.in: Rebuilt.
23675 * configure: Rebuilt.
23676 * configure.ac: Test assembler for view support.
23677 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
23678 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
23679 * dwarf2out.c (var_loc_view): New typedef.
23680 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
23681 (dwarf2out_locviews_in_attribute): New.
23682 (dwarf2out_locviews_in_loclist): New.
23683 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
23684 (enum dw_line_info_opcode): Add LI_adv_address.
23685 (struct dw_line_info_table): Add view.
23686 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
23687 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
23688 (zero_view_p): New variable.
23689 (ZERO_VIEW_P): New macro.
23690 (output_asm_line_debug_info): New.
23691 (struct var_loc_node): Add view.
23692 (add_AT_view_list, AT_loc_list): New.
23693 (add_var_loc_to_decl): Add view param. Test it against last.
23694 (new_loc_list): Add view params. Record them.
23695 (AT_loc_list_ptr): Handle loc and view lists.
23696 (view_list_to_loc_list_val_node): New.
23697 (print_dw_val): Handle dw_val_class_view_list.
23698 (size_of_die): Likewise.
23699 (value_format): Likewise.
23700 (loc_list_has_views): New.
23701 (gen_llsym): Set vl_symbol too.
23702 (maybe_gen_llsym, skip_loc_list_entry): New.
23703 (dwarf2out_maybe_output_loclist_view_pair): New.
23704 (output_loc_list): Output view list or entries too.
23705 (output_view_list_offset): New.
23706 (output_die): Handle dw_val_class_view_list.
23707 (output_dwarf_version): New.
23708 (output_compilation_unit_header): Use it.
23709 (output_skeleton_debug_sections): Likewise.
23710 (output_rnglists, output_line_info): Likewise.
23711 (output_pubnames, output_aranges): Update version comments.
23712 (output_one_line_info_table): Output view numbers in asm comments.
23713 (dw_loc_list): Determine current endview, pass it to new_loc_list.
23714 Call maybe_gen_llsym.
23715 (loc_list_from_tree_1): Adjust.
23716 (add_AT_location_description): Create view list attribute if
23717 needed, check it's absent otherwise.
23718 (convert_cfa_to_fb_loc_list): Adjust.
23719 (maybe_emit_file): Call output_asm_line_debug_info for test.
23720 (dwarf2out_var_location): Reset views as needed. Precompute
23721 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
23722 attribute. Set view.
23723 (new_line_info_table): Reset next view.
23724 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
23725 (dwarf2out_source_line): Likewise. Output view resets and labels to
23726 the assembler, or select appropriate line info opcodes.
23727 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
23728 (optimize_string_length): Catch it. Adjust.
23729 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
23730 dw_val_class_view_list, and remove it if no longer needed.
23731 (hash_loc_list): Hash view numbers.
23732 (loc_list_hasher::equal): Compare them.
23733 (optimize_location_lists): Check whether a view list symbol is
23734 needed, and whether the locview attribute is present, and
23735 whether they match. Remove the locview attribute if no longer
23737 (index_location_lists): Call skip_loc_list_entry for test.
23738 (dwarf2out_finish): Call output_asm_line_debug_info for test.
23739 Use output_dwarf_version.
23740 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
23741 (struct dw_val_node): Add val_view_list.
23742 * final.c (SEEN_NEXT_VIEW): New.
23743 (set_next_view_needed): New.
23744 (clear_next_view_needed): New.
23745 (maybe_output_next_view): New.
23746 (final_start_function): Rename to...
23747 (final_start_function_1): ... this. Take pointer to FIRST,
23748 add SEEN parameter. Emit param bindings in the initial view.
23749 (final_start_function): Reintroduce SEEN-less interface.
23750 (final): Rename to...
23751 (final_1): ... this. Take SEEN parameter. Output final pending
23752 next view at the end.
23753 (final): Reintroduce seen-less interface.
23754 (final_scan_insn): Output pending next view before switching
23755 sections or ending a block. Mark the next view as needed when
23756 outputting variable locations. Notify debug backend of section
23757 changes, and of location view changes.
23758 (rest_of_handle_final): Adjust.
23759 * toplev.c (process_options): Autodetect value for debug variable
23760 location views option. Warn on incompat5 without -gdwarf-5.
23761 * doc/invoke.texi (gvariable-location-views): New.
23762 (gvariable-location-views=incompat5): New.
23763 (gno-variable-location-views): New.
23765 2018-02-08 David Malcolm <dmalcolm@redhat.com>
23767 PR tree-optimization/84136
23768 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
23769 that the result of find_edge is non-NULL.
23771 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23774 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
23775 storing integer register in SImode. Fix cost of 256 and 512
23776 byte aligned SSE register store.
23778 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23780 * config/i386/i386.c (ix86_multiplication_cost): Fix
23781 multiplication cost for TARGET_AVX512DQ.
23783 2018-02-08 Marek Polacek <polacek@redhat.com>
23785 PR tree-optimization/84238
23786 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
23789 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23791 PR tree-optimization/84265
23792 * tree-vect-stmts.c (vectorizable_store): Don't treat
23793 VMAT_CONTIGUOUS accesses as grouped.
23794 (vectorizable_load): Likewise.
23796 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23798 PR tree-optimization/81635
23799 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
23800 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
23801 (test_round_for_mask): New functions.
23802 (wide_int_cc_tests): Call test_round_for_mask.
23803 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
23804 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
23805 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
23806 range returned by get_range_info.
23808 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23811 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
23812 * symtab.c: Include builtins.h
23813 (symtab_node::output_to_lto_symbol_table_p): Move here
23814 from lto-streamer-out.c:output_symbol_p.
23815 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
23816 (output_symbol_p): Move all logic to symtab.c
23817 (produce_symtab): Update.
23819 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23821 * config/s390/s390-opts.h (enum indirect_branch): Define.
23822 * config/s390/s390-protos.h (s390_return_addr_from_memory)
23823 (s390_indirect_branch_via_thunk)
23824 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
23825 (enum s390_indirect_branch_type): Define.
23826 * config/s390/s390.c (struct s390_frame_layout, struct
23827 machine_function): Remove.
23828 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
23829 (indirect_branch_table_label_no, indirect_branch_table_name):
23831 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
23832 (enum s390_indirect_branch_option): Define.
23833 (s390_return_addr_from_memory): New function.
23834 (s390_handle_string_attribute): New function.
23835 (s390_attribute_table): Add new attribute handler.
23836 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
23837 (s390_indirect_branch_via_thunk): New function.
23838 (s390_indirect_branch_via_inline_thunk): New function.
23839 (s390_function_ok_for_sibcall): When jumping via thunk disallow
23840 sibling call optimization for non z10 compiles.
23841 (s390_emit_call): Force indirect branch target to be a single
23842 register. Add r1 clobber for non-z10 compiles.
23843 (s390_emit_epilogue): Emit return jump via return_use expander.
23844 (s390_reorg): Handle JUMP_INSNs as execute targets.
23845 (s390_option_override_internal): Perform validity checks for the
23846 new command line options.
23847 (s390_indirect_branch_attrvalue): New function.
23848 (s390_indirect_branch_settings): New function.
23849 (s390_set_current_function): Invoke s390_indirect_branch_settings.
23850 (s390_output_indirect_thunk_function): New function.
23851 (s390_code_end): Implement target hook.
23852 (s390_case_values_threshold): Implement target hook.
23853 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
23855 * config/s390/s390.h (struct s390_frame_layout)
23856 (struct machine_function): Move here from s390.c.
23857 (TARGET_INDIRECT_BRANCH_NOBP_RET)
23858 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
23859 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
23860 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
23861 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
23862 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
23863 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
23864 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
23865 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
23866 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
23867 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
23868 (mnemonic attribute): Add values which aren't recognized
23870 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
23871 pattern for branch conversion. Fix mnemonic attribute.
23872 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
23873 indirect branch via thunk if requested.
23874 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
23875 ("*indirect_jump"): Disable for branch conversion using out of
23877 ("indirect_jump_via_thunk<mode>_z10")
23878 ("indirect_jump_via_thunk<mode>")
23879 ("indirect_jump_via_inlinethunk<mode>_z10")
23880 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
23881 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
23882 ("casesi_jump_via_inlinethunk<mode>_z10")
23883 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
23884 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
23885 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
23886 ("*indirect2_jump"): Disable for branch conversion.
23887 ("casesi_jump"): Turn into expander and expand patterns for branch
23889 ("return_use"): New expander.
23890 ("*return"): Emit return via thunk and rename it to ...
23891 ("*return<mode>"): ... this one.
23892 * config/s390/s390.opt: Add new options and and enum for the
23895 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23897 * lra-constraints.c (match_reload): Unconditionally use
23898 gen_lowpart_SUBREG, rather than selecting between that
23899 and equivalent gen_rtx_SUBREG code.
23901 2018-02-08 Richard Biener <rguenther@suse.de>
23903 PR tree-optimization/84233
23904 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
23905 changed flag instead of boguously re-using phi_inserted.
23907 2018-02-08 Martin Jambor <mjambor@suse.cz>
23909 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
23910 static local variables.
23912 2018-02-08 Richard Biener <rguenther@suse.de>
23914 PR tree-optimization/84278
23915 * tree-vect-stmts.c (vectorizable_store): When looking for
23916 smaller vector types to perform grouped strided loads/stores
23917 make sure the mode is supported by the target.
23918 (vectorizable_load): Likewise.
23920 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23922 * config/aarch64/aarch64.c (aarch64_components_for_bb):
23923 Increase LDP/STP opportunities by adding adjacent callee-saves.
23925 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23927 PR rtl-optimization/84068
23928 PR rtl-optimization/83459
23929 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
23931 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
23933 PR tree-optimization/84224
23934 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
23935 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
23936 non-zero arguments.
23938 2018-02-07 Iain Sandoe <iain@codesourcery.com>
23941 * config/rs6000/altivec.md (*restore_world): Remove LR use.
23942 * config/rs6000/predicates.md (restore_world_operation): Adjust op
23943 count, remove one USE.
23945 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23947 * doc/install.texi (Configuration): Document the
23948 --with-long-double-format={ibm,ieee} PowerPC configuration
23952 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
23953 Convert from define_expand to be define_insn_and_split. Rework
23954 float/double/_Float128 conversions to QI/HI/SImode to work with
23955 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
23956 conversions to QI/HImode types did a store and then a load to
23957 truncate the value. For conversions to VSX registers, don't split
23958 the insn, instead emit the code directly. Use the code iterator
23959 any_fix to combine signed and unsigned conversions.
23960 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
23961 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
23962 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23963 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
23964 (fix_<mode>di2_hw): Likewise.
23965 (fixuns_<mode>di2_hw): Likewise.
23966 (fix_<mode>si2_hw): Likewise.
23967 (fixuns_<mode>si2_hw): Likewise.
23968 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
23969 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23970 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
23971 fix<uns>_trunc<SFDF:mode>si2_p8.
23972 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
23973 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23974 (fix<uns>_<mode>_mem): Likewise.
23975 (fctiw<u>z_<mode>_mem): Likewise.
23976 (fix<uns>_<mode>_mem): Likewise.
23977 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
23978 the register allocator from doing a direct move to the GPRs to do
23979 a store, and instead use the ISA 3.0 store byte/half-word from
23980 vector register instruction. For IEEE 128-bit floating point,
23981 also optimize stores of 32-bit ints.
23982 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
23984 2018-02-07 Alan Hayward <alan.hayward@arm.com>
23986 * genextract.c (push_pathstr_operand): New function to support
23988 (walk_rtx): Call push_pathstr_operand.
23989 (print_path): Support [a-zA-Z].
23991 2018-02-07 Richard Biener <rguenther@suse.de>
23993 PR tree-optimization/84037
23994 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
23995 (cse_and_gimplify_to_preheader): Declare.
23996 (vect_get_place_in_interleaving_chain): Likewise.
23997 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
23999 (_loop_vec_info::~_loop_vec_info): Delete it.
24000 (cse_and_gimplify_to_preheader): New function.
24001 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
24002 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
24003 (vectorizable_load): Likewise. For grouped stores always base
24004 the IV on the first element.
24005 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
24006 condition before gimplifying.
24008 2018-02-07 Jakub Jelinek <jakub@redhat.com>
24010 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
24011 *DIV_EXPR and *MOD_EXPR.
24013 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
24016 * config/i386/i386.c (ix86_option_override_internal): Mask out
24017 the CF_SET bit when checking -fcf-protection.
24019 2018-02-07 Tom de Vries <tom@codesourcery.com>
24022 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
24025 2018-02-07 Richard Biener <rguenther@suse.de>
24027 PR tree-optimization/84204
24028 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
24031 PR tree-optimization/84205
24032 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
24033 special-case isl_ast_op_zdiv_r.
24035 PR tree-optimization/84223
24036 * graphite-scop-detection.c (gather_bbs::before_dom_children):
24037 Only add conditions from within the region.
24038 (gather_bbs::after_dom_children): Adjust.
24040 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
24043 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
24044 * config/avr/avr.md: Only post-reload split REG-REG moves if
24045 either register is GENERAL_REG_P.
24047 2018-02-07 Jakub Jelinek <jakub@redhat.com>
24049 PR tree-optimization/84235
24050 * tree-ssa-scopedtables.c
24051 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
24052 if the subtraction is performed in floating point type where NaNs are
24053 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
24054 build 1. Formatting fix.
24056 2018-02-06 Jakub Jelinek <jakub@redhat.com>
24059 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
24060 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
24061 and skip it regardless of bb boundaries. Use CALL_P macro,
24062 don't test INSN_P (insn) together with CALL_P or JUMP_P check
24063 unnecessarily, formatting fix.
24065 2018-02-06 Michael Collison <michael.collison@arm.com>
24067 * config/arm/thumb2.md:
24068 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
24069 (*thumb_mov_notscc): Ditto.
24071 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24074 * config/rs6000/rs6000.md (su code attribute): Use "u" for
24075 unsigned_fix, not "s".
24077 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24079 * configure.ac (gcc_fn_eh_frame_ro): New function.
24080 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
24081 correct .eh_frame permissions.
24082 * configure: Regenerate.
24084 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
24086 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
24087 irrelevant options.
24089 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24091 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24092 Display warning message for -mno-speculate-indirect-jumps.
24094 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
24096 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
24098 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
24100 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
24102 PR tree-optimization/84225
24103 * tree-eh.c (find_trapping_overflow): Only call
24104 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
24106 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
24109 * config/i386/i386.c: Reimplement the check of possible options
24110 -mibt/-mshstk conbination. Change error messages.
24111 * doc/invoke.texi: Fix a typo: remove extra '='.
24113 2018-02-06 Marek Polacek <polacek@redhat.com>
24115 PR tree-optimization/84228
24116 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
24118 2018-02-06 Tamar Christina <tamar.christina@arm.com>
24121 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
24122 emitted arch directives.
24123 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
24124 __ARM_FEATURE_COPROC before changing architectures.
24126 2018-02-06 Richard Biener <rguenther@suse.de>
24128 * config/i386/i386.c (print_reg): Fix typo.
24129 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
24131 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
24133 * configure: Regenerate.
24135 2018-02-05 Martin Sebor <msebor@redhat.com>
24137 PR tree-optimization/83369
24138 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
24141 2018-02-05 Martin Liska <mliska@suse.cz>
24143 * doc/invoke.texi: Cherry-pick upstream r323995.
24145 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
24147 * ira.c (ira_init_register_move_cost): Adjust comment.
24149 2018-02-05 Martin Liska <mliska@suse.cz>
24151 PR gcov-profile/84137
24152 * doc/gcov.texi: Fix typo in documentation.
24154 2018-02-05 Martin Liska <mliska@suse.cz>
24156 PR gcov-profile/83879
24157 * doc/gcov.texi: Document necessity of --dynamic-list-data when
24158 using dlopen functionality.
24160 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
24162 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
24163 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
24164 _mm_maskz_range_ss, _mm_mask_range_round_ss,
24165 _mm_maskz_range_round_ss): New intrinsics.
24166 (__builtin_ia32_rangesd128_round)
24167 (__builtin_ia32_rangess128_round): Remove.
24168 (__builtin_ia32_rangesd128_mask_round,
24169 __builtin_ia32_rangess128_mask_round): New builtins.
24170 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
24171 __builtin_ia32_rangess128_round): Remove.
24172 (__builtin_ia32_rangesd128_mask_round,
24173 __builtin_ia32_rangess128_mask_round): New builtins.
24174 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
24175 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
24176 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
24177 "<round_saeonly_constraint>")): Changed to ...
24178 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
24179 "<round_saeonly_scalar_constraint>")): ... this.
24180 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24181 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
24182 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
24183 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
24184 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
24186 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
24188 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
24190 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
24191 Remove all values except native, 8540 and 8548.
24193 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
24195 * config/i386/i386.c (ix86_output_function_return): Pass
24196 INVALID_REGNUM, instead of -1, as invalid register number to
24197 indirect_thunk_name and output_indirect_thunk.
24199 2018-02-02 Julia Koval <julia.koval@intel.com>
24201 * config.gcc: Add -march=icelake.
24202 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
24203 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
24204 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
24205 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
24206 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
24207 (processor_target_table): Add icelake.
24208 (ix86_option_override_internal): Handle new PTAs.
24209 (get_builtin_code_for_version): Handle icelake.
24210 (M_INTEL_COREI7_ICELAKE): New.
24211 (fold_builtin_cpu): Handle icelake.
24212 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
24213 * doc/invoke.texi: Add -march=icelake.
24215 2018-02-02 Julia Koval <julia.koval@intel.com>
24217 * config/i386/i386.c (ix86_option_override_internal): Change flags type
24218 to wide_int_bitmask.
24219 * wide-int-bitmask.h: New.
24221 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
24224 * config/i386/i386.md: Replace Pmode with word_mode in
24225 builtin_setjmp_setup and builtin_longjmp to support x32.
24227 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
24231 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
24233 (rs6000_supported_cpu_names): New static variable.
24234 (linux_cpu_translation_table): Likewise.
24235 (elf_platform) <cpu>: Define new static variable and use it.
24236 Translate kernel AT_PLATFORM name to canonical name if needed.
24237 Error if platform name is unknown.
24239 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
24242 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
24244 2018-02-01 Jeff Law <law@redhat.com>
24247 * config/i386/i386.c (release_scratch_register_on_entry): Add new
24248 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
24249 the scratch if RELEASE_VIA_POP is false.
24250 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
24251 If we have to save a temporary register, decrement SIZE appropriately.
24252 Pass new arguments to release_scratch_register_on_entry.
24253 (ix86_adjust_stack_and_probe): Likewise.
24254 (ix86_emit_probe_stack_range): Pass new arguments to
24255 release_scratch_register_on_entry.
24257 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
24259 PR rtl-optimization/84157
24260 * combine.c (change_zero_ext): Use REG_P predicate in
24261 front of HARD_REGISTER_P predicate.
24263 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
24265 * config/avr/avr.c (avr_option_override): Move disabling of
24266 -fdelete-null-pointer-checks to...
24267 * common/config/avr/avr-common.c (avr_option_optimization_table):
24270 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24272 PR tree-optimization/81635
24273 * tree-data-ref.c (split_constant_offset_1): For types that
24274 wrap on overflow, try to use range info to prove that wrapping
24277 2018-02-01 Renlin Li <renlin.li@arm.com>
24280 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
24281 TAILCALL_ADDR_REGS.
24282 (aarch64_register_move_cost): Likewise.
24283 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
24284 TAILCALL_ADDR_REGS.
24285 (REG_CLASS_NAMES): Likewise.
24286 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
24287 TAILCALL_ADDR_REGS. Remove IP registers.
24288 * config/aarch64/aarch64.md (Ucs): Update register constraint.
24290 2018-02-01 Richard Biener <rguenther@suse.de>
24292 * domwalk.h (dom_walker::dom_walker): Add additional constructor
24293 for specifying RPO order and allow NULL for that.
24294 * domwalk.c (dom_walker::dom_walker): Likewise.
24295 (dom_walker::walk): Handle NULL RPO order.
24296 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
24298 (rewrite_update_dom_walker): Likewise.
24299 (mark_def_dom_walker): Likewise.
24301 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24303 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
24304 (aarch64_maybe_expand_sve_subreg_move): Declare.
24305 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
24306 * config/aarch64/predicates.md (aarch64_any_register_operand): New
24308 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
24309 that are semantically a reverse operation.
24310 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
24311 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
24312 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
24314 (aarch64_can_change_mode_class): For big-endian, forbid changes
24315 between two SVE modes if they have different element sizes.
24317 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24319 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
24320 the TImode handling for big-endian targets.
24322 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24324 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
24325 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
24327 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
24328 Remove BSWAP handing for big-endian targets and use the form of
24329 LD1RQ appropariate for the mode.
24331 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24333 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
24334 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
24335 duplicated element.
24337 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24340 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
24341 check for operands that need to go through aarch64_sve_reload_be.
24343 2018-02-01 Jakub Jelinek <jakub@redhat.com>
24345 PR tree-optimization/81661
24346 PR tree-optimization/84117
24347 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
24348 * tree-eh.c: Include gimplify.h.
24349 (find_trapping_overflow, replace_trapping_overflow,
24350 rewrite_to_non_trapping_overflow): New functions.
24351 * tree-vect-loop.c: Include tree-eh.h.
24352 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
24353 * tree-data-ref.c: Include tree-eh.h.
24354 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
24356 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
24358 PR rtl-optimization/84123
24359 * combine.c (change_zero_ext): Check if hard register satisfies
24360 can_change_dest_mode before calling gen_lowpart_SUBREG.
24362 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
24365 * ira.c (ira_init_register_move_cost): Remove assert.
24367 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24369 PR rtl-optimization/84071
24370 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
24371 * doc/tm.texi: Regenerate.
24373 2018-01-31 Richard Biener <rguenther@suse.de>
24375 PR tree-optimization/84132
24376 * tree-data-ref.c (analyze_miv_subscript): Properly
24377 check whether evolution_function_is_affine_multivariate_p
24378 before calling gcd_of_steps_may_divide_p.
24380 2018-01-31 Julia Koval <julia.koval@intel.com>
24383 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24384 * config/i386/i386.md (rdpid_rex64) New.
24385 (rdpid): Make 32bit only.
24387 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
24390 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
24391 an IDENTIFIER_NODE for FUNCTION_TYPE's.
24393 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24396 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
24398 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
24400 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24402 PR rtl-optimization/84071
24403 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
24404 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
24406 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
24408 * config/arc/arc.c (arc_handle_aux_attribute): New function.
24409 (arc_attribute_table): Add 'aux' attribute.
24410 (arc_in_small_data_p): Consider aux like variables.
24411 (arc_is_aux_reg_p): New function.
24412 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
24413 (arc_get_aux_arg): New function.
24414 (prepare_move_operands): Handle aux-register access.
24415 (arc_handle_aux_attribute): New function.
24416 * doc/extend.texi (ARC Variable attributes): Add subsection.
24418 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
24420 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
24421 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
24422 (arc_attribute_table): Add 'uncached' attribute.
24423 (arc_print_operand): Print '.di' flag for uncached memory
24425 (arc_in_small_data_p): Do not consider for small data the uncached
24427 (arc_is_uncached_mem_p): New function.
24428 * config/arc/predicates.md (compact_store_memory_operand): Check
24429 for uncached memory accesses.
24430 (nonvol_nonimm_operand): Likewise.
24431 * doc/extend.texi (ARC Type Attribute): New subsection.
24433 2018-01-31 Jakub Jelinek <jakub@redhat.com>
24436 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
24437 falign-loops=): Add Optimization flag.
24439 2018-01-30 Jeff Law <law@redhat.com>
24442 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
24443 INT_REGISTERS_SAVED. Check it prior to calling
24444 get_scratch_register_on_entry.
24445 (ix86_adjust_stack_and_probe): Similarly.
24446 (ix86_emit_probe_stack_range): Similarly.
24447 (ix86_expand_prologue): Corresponding changes.
24449 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24452 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
24453 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
24455 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
24458 * lra-constraints.c (curr_insn_transform): Process AND in the
24461 2018-01-30 Jakub Jelinek <jakub@redhat.com>
24463 PR rtl-optimization/83986
24464 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
24465 dependence against last_pending_memory_flush in addition to
24466 pending_jump_insns.
24468 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
24470 PR tree-optimization/81611
24471 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
24474 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24477 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
24480 2018-01-30 Richard Biener <rguenther@suse.de>
24481 Jakub Jelinek <jakub@redhat.com>
24483 PR tree-optimization/84111
24484 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
24485 inner loops added during recursion, as they don't have up-to-date
24488 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
24491 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
24492 (can_inline_edge_by_limits_p): ... here.
24493 (can_early_inline_edge_p, check_callers,
24494 update_caller_keys, update_callee_keys, recursive_inlining,
24495 add_new_edges_to_heap, speculation_useful_p,
24496 inline_small_functions,
24497 inline_small_functions, flatten_function,
24498 inline_to_all_callers_1): Update.
24500 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
24502 * profile-count.c (profile_count::combine_with_ipa_count): Handle
24505 2018-01-30 Richard Biener <rguenther@suse.de>
24507 PR tree-optimization/83008
24508 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
24509 invariant and constant vector uses in stmts when they need
24510 more than one stmt.
24512 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24515 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
24516 * configure: Regenerate.
24518 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24520 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
24522 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
24523 Use gen_rtx_REG rather than gen_lowpart.
24525 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24527 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
24528 rather than 0 when creating partial subregs.
24530 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24532 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
24535 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
24538 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
24539 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
24540 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
24541 flags. This restores the settings used before the 2017-07-24.
24542 Turning off pre increment/decrement/modify allows IVOPTS to
24543 optimize DF/SF loops where the index is an int.
24545 2018-01-29 Richard Biener <rguenther@suse.de>
24546 Kelvin Nilsen <kelvin@gcc.gnu.org>
24549 * tree-vect-stmts.c (vectorizable_call): Don't call
24550 targetm.vectorize_builtin_md_vectorized_function if callee is
24553 2018-01-22 Carl Love <cel@us.ibm.com>
24555 * doc/extend.tex: Fix typo in second arg in
24556 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
24558 2018-01-29 Richard Biener <rguenther@suse.de>
24560 PR tree-optimization/84086
24561 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
24562 (flush_ssaname_freelist): When SSA names were released reset
24563 the SCEV hash table.
24565 2018-01-29 Richard Biener <rguenther@suse.de>
24567 PR tree-optimization/84057
24568 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
24569 removed paths when removing edges.
24571 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
24573 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
24574 -mfunction-return=@var{choice}.
24576 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
24578 PR diagnostic/84034
24579 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
24580 Handle CR like TAB.
24581 (layout::print_source_line): Likewise.
24582 (test_get_line_width_without_trailing_whitespace): Add test cases.
24584 2018-01-27 Jakub Jelinek <jakub@redhat.com>
24586 PR middle-end/84040
24587 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
24590 2018-01-26 Jim Wilson <jimw@sifive.com>
24592 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
24594 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
24597 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24599 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
24600 and CMP + SUB-immediate -> SUBS.
24602 2018-01-26 Martin Sebor <msebor@redhat.com>
24604 PR tree-optimization/83896
24605 * tree-ssa-strlen.c (get_string_len): Rename...
24606 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
24607 Avoid assuming length is constant.
24608 (handle_char_store): Use HOST_WIDE_INT for string length.
24610 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
24613 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
24614 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
24616 2018-01-26 Richard Biener <rguenther@suse.de>
24618 PR rtl-optimization/84003
24619 * dse.c (record_store): Only record redundant stores when
24620 the earlier store aliases at least all accesses the later one does.
24622 2018-01-26 Jakub Jelinek <jakub@redhat.com>
24624 PR rtl-optimization/83985
24625 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
24626 REG_CFA_RESTORE insns.
24627 (delete_unmarked_insns): Don't ignore separate shrink wrapping
24628 REG_CFA_RESTORE insns here.
24631 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
24632 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
24634 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24636 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
24637 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
24638 (arc_init): Likewise.
24639 (arc_override_options): Likewise.
24640 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
24642 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
24644 * config/arc/arc.h (TARGET_DBNZ): Define.
24645 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
24646 properly set the tune attribute.
24647 (dbnz): Use TARGET_DBNZ guard.
24648 * config/arc/arc.opt (mtune): Add core3 option.
24650 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24652 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
24653 recognize new pic like addresses.
24654 (arc_delegitimize_address): Clean up.
24656 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24658 * config/arc/arc-arches.def: Option mrf16 valid for all
24660 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
24661 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
24662 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
24663 * config/arc/arc-tables.opt: Regenerate.
24664 * config/arc/arc.c (arc_conditional_register_usage): Handle
24665 reduced register file case.
24666 (arc_file_start): Set must have build attributes.
24667 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
24668 mrf16 option value.
24669 * config/arc/arc.opt (mrf16): Add new option.
24670 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
24671 * config/arc/genmultilib.awk: Handle new mrf16 option.
24672 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
24673 * config/arc/t-multilib: Regenerate.
24674 * doc/invoke.texi (ARC Options): Document mrf16 option.
24676 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24678 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
24679 * config/arc/arc.c (arc_handle_secure_attribute): New function.
24680 (arc_attribute_table): Add 'secure_call' attribute.
24681 (arc_print_operand): Print secure call operand.
24682 (arc_function_ok_for_sibcall): Don't optimize tail calls when
24684 (arc_is_secure_call_p): New function. * config/arc/arc.md
24685 (call_i): Add support for sjli instruction.
24686 (call_value_i): Likewise.
24687 * config/arc/constraints.md (Csc): New constraint.
24689 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24690 John Eric Martin <John.Martin@emmicro-us.com>
24692 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
24693 * config/arc/arc.c (_arc_jli_section): New struct.
24694 (arc_jli_section): New type.
24695 (rc_jli_sections): New static variable.
24696 (arc_handle_jli_attribute): New function.
24697 (arc_attribute_table): Add jli_always and jli_fixed attribute.
24698 (arc_file_end): New function.
24699 (TARGET_ASM_FILE_END): Define.
24700 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
24701 (arc_add_jli_section): New function.
24702 (jli_call_scan): Likewise.
24703 (arc_reorg): Call jli_call_scan.
24704 (arc_output_addsi): Remove 'S' from printing asm operand.
24705 (arc_is_jli_call_p): New function.
24706 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
24708 (movhi_insn): Likewise.
24709 (movsi_insn): Likewise.
24710 (movsi_set_cc_insn): Likewise.
24711 (loadqi_update): Likewise.
24712 (load_zeroextendqisi_update): Likewise.
24713 (load_signextendqisi_update): Likewise.
24714 (loadhi_update): Likewise.
24715 (load_zeroextendhisi_update): Likewise.
24716 (load_signextendhisi_update): Likewise.
24717 (loadsi_update): Likewise.
24718 (loadsf_update): Likewise.
24719 (movsicc_insn): Likewise.
24720 (bset_insn): Likewise.
24721 (bxor_insn): Likewise.
24722 (bclr_insn): Likewise.
24723 (bmsk_insn): Likewise.
24724 (bicsi3_insn): Likewise.
24725 (cmpsi_cc_c_insn): Likewise.
24726 (movsi_ne): Likewise.
24727 (movsi_cond_exec): Likewise.
24728 (clrsbsi2): Likewise.
24729 (norm_f): Likewise.
24739 (call_i): Remove 'S' asm letter, add jli instruction.
24740 (call_value_i): Likewise.
24741 * config/arc/arc.op (mjli-always): New option.
24742 * config/arc/constraints.md (Cji): New constraint.
24743 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
24745 (subsf3_fpx): Likewise.
24746 (mulsf3_fpx): Likewise.
24747 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
24749 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
24750 function attrbutes.
24751 * doc/invoke.texi (ARC): Document mjli-always option.
24753 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
24755 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
24756 avoid addition with 0 and use incw and decw where possible.
24758 2018-01-26 Richard Biener <rguenther@suse.de>
24760 PR tree-optimization/81082
24761 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
24762 association if it requires casting to unsigned.
24763 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
24764 from fold_plusminus_mult_expr to catch important cases late when
24765 range info is available.
24767 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24769 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
24770 * configure.ac (hidden_linkonce): New test.
24771 * configure: Regenerate.
24772 * config.in: Regenerate.
24774 2018-01-26 Julia Koval <julia.koval@intel.com>
24776 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
24777 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
24778 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
24779 _mm_mask_bitshuffle_epi64_mask): Fix type.
24780 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
24781 USI_FTYPE_V4DI_V4DI_USI): Remove.
24782 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
24783 __builtin_ia32_vpshufbitqmb256_mask,
24784 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
24785 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
24786 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
24788 2018-01-26 Alan Modra <amodra@gmail.com>
24791 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
24792 UNSPEC_VBPERMQ. Sort other unspecs.
24794 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
24796 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
24798 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
24800 PR middle-end/83055
24801 * predict.c (drop_profile): Do not push/pop cfun; update also
24803 (handle_missing_profiles): Fix logic looking for zero profiles.
24805 2018-01-25 Jakub Jelinek <jakub@redhat.com>
24807 PR middle-end/83977
24808 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
24809 on functions with #pragma omp declare simd or functions with simd
24811 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
24812 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
24813 Remove trailing \n from warning_at calls.
24815 2018-01-25 Tom de Vries <tom@codesourcery.com>
24818 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24819 for neutered workers.
24821 2018-01-24 Joseph Myers <joseph@codesourcery.com>
24824 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
24825 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
24827 2018-01-24 Jeff Law <law@redhat.com>
24830 * i386.c (get_probe_interval): Move to earlier point.
24831 (ix86_compute_frame_layout): If -fstack-clash-protection and
24832 the frame is larger than the probe interval, then use pushes
24833 to save registers rather than reg->mem moves.
24834 (ix86_expand_prologue): Remove conditional for int_registers_saved
24837 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
24840 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
24841 min/max for never referenced object.
24843 2018-01-24 Jakub Jelinek <jakub@redhat.com>
24845 PR middle-end/83977
24846 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
24848 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
24849 attributes from DECL_ATTRIBUTES (decl) without affecting
24850 DECL_ATTRIBUTES (current_function_decl).
24851 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
24852 functions with non-NULL DECL_ABSTRACT_ORIGIN.
24854 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
24856 PR tree-optimization/83979
24857 * fold-const.c (fold_comparison): Use constant_boolean_node
24858 instead of boolean_{true,false}_node.
24860 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
24862 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
24865 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24867 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24868 Simplify the clause that sets the length attribute.
24869 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24870 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
24871 clause that sets the length attribute.
24872 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24874 2018-01-24 Tom de Vries <tom@codesourcery.com>
24877 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
24878 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
24879 Add strict parameter.
24880 (prevent_branch_around_nothing): Insert dummy insn between branch to
24881 label and label with no ptx insn inbetween.
24882 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
24884 2018-01-24 Tom de Vries <tom@codesourcery.com>
24887 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24888 for neutered threads in warp.
24889 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
24891 2018-01-24 Richard Biener <rguenther@suse.de>
24893 PR tree-optimization/83176
24894 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
24897 2018-01-24 Richard Biener <rguenther@suse.de>
24899 PR tree-optimization/82819
24900 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
24901 code generating pluses that are no-ops in the target precision.
24903 2018-01-24 Richard Biener <rguenther@suse.de>
24905 PR middle-end/84000
24906 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
24908 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24910 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
24911 to merge probabilities.
24912 * predict.c (probably_never_executed): Also mark as cold functions
24913 with global 0 profile and guessed local profile.
24914 * profile-count.c (profile_probability::combine_with_count): New
24916 * profile-count.h (profile_probability::operator*,
24917 profile_probability::operator*=, profile_probability::operator/,
24918 profile_probability::operator/=): Reduce precision to adjusted
24919 and set value to guessed on contradictory divisions.
24920 (profile_probability::combine_with_freq): Remove.
24921 (profile_probability::combine_wiht_count): Declare.
24922 (profile_count::force_nonzero):: Set to adjusted.
24923 (profile_count::probability_in):: Set quality to adjusted.
24924 * tree-ssa-tail-merge.c (replace_block_by): Use
24925 combine_with_count.
24927 2018-01-23 Andrew Waterman <andrew@sifive.com>
24928 Jim Wilson <jimw@sifive.com>
24930 * config/riscv/riscv.c (riscv_stack_boundary): New.
24931 (riscv_option_override): Set riscv_stack_boundary. Handle
24932 riscv_preferred_stack_boundary_arg.
24933 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
24934 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
24935 (STACK_BOUNDARY): Set to riscv_stack_boundary.
24936 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
24937 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
24938 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
24940 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
24943 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
24944 of struct ix86_frame.
24945 (ix86_expand_epilogue): Likewise. Add a local variable for
24946 the reg_save_offset field in struct ix86_frame.
24948 2018-01-23 Bin Cheng <bin.cheng@arm.com>
24950 PR tree-optimization/82604
24951 * tree-loop-distribution.c (enum partition_kind): New enum item
24952 PKIND_PARTIAL_MEMSET.
24953 (partition_builtin_p): Support above new enum item.
24954 (generate_code_for_partition): Ditto.
24955 (compute_access_range): Differentiate cases that equality can be
24956 proven at all loops, the innermost loops or no loops.
24957 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
24958 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
24959 (finalize_partitions, distribute_loop): Don't fuse partition of
24960 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
24961 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
24962 parloop is enabled.
24964 2018-01-23 Martin Liska <mliska@suse.cz>
24966 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
24967 order to ignore the predictor.
24968 (PRED_POLYMORPHIC_CALL): Likewise.
24969 (PRED_RECURSIVE_CALL): Likewise.
24971 2018-01-23 Martin Liska <mliska@suse.cz>
24973 * tree-profile.c (tree_profiling): Print function header to
24974 aware reader which function we are working on.
24975 * value-prof.c (gimple_find_values_to_profile): Do not print
24976 not interesting value histograms.
24978 2018-01-23 Martin Liska <mliska@suse.cz>
24980 * profile-count.h (enum profile_quality): Add
24981 profile_uninitialized as the first value. Do not number values
24982 as they are zero based.
24983 (profile_count::verify): Update sanity check.
24984 (profile_probability::verify): Likewise.
24986 2018-01-23 Nathan Sidwell <nathan@acm.org>
24988 * doc/invoke.texi (ffor-scope): Deprecate.
24990 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24992 PR tree-optimization/83510
24993 * domwalk.c (set_all_edges_as_executable): New function.
24994 (dom_walker::dom_walker): Convert bool param
24995 "skip_unreachable_blocks" to enum reachability. Move setup of
24996 edge flags to set_all_edges_as_executable and only do it when
24997 reachability is REACHABLE_BLOCKS.
24998 * domwalk.h (enum dom_walker::reachability): New enum.
24999 (dom_walker::dom_walker): Convert bool param
25000 "skip_unreachable_blocks" to enum reachability.
25001 (set_all_edges_as_executable): New decl.
25002 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
25003 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
25005 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
25006 but converting true to REACHABLE_BLOCKS.
25007 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
25009 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
25010 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
25011 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
25013 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
25014 if check_all_array_refs will be called.
25016 2018-01-23 David Malcolm <dmalcolm@redhat.com>
25018 * tree.c (selftest::test_location_wrappers): Add more test
25021 2018-01-23 David Malcolm <dmalcolm@redhat.com>
25023 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
25024 (selftest::test_bit_in_range): Likewise.
25026 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
25029 * doc/sourcebuild.texi (vect_float): Say that the selector
25030 only describes the situation when -funsafe-math-optimizations is on.
25031 (vect_float_strict): Document.
25033 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
25035 PR tree-optimization/83965
25036 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
25037 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
25038 instead of checking only for a reduction.
25039 (vect_recog_widen_sum_pattern): Likewise.
25041 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
25043 * predict.c (probably_never_executed): Only use precise profile info.
25044 (compute_function_frequency): Skip after inlining hack since we now
25045 have quality checking.
25047 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
25049 * profile-count.h (profile_probability::very_unlikely,
25050 profile_probability::unlikely, profile_probability::even): Set
25051 precision to guessed.
25053 2018-01-23 Richard Biener <rguenther@suse.de>
25055 PR tree-optimization/83963
25056 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25057 Properly terminate dominator walk when crossing the exit edge not
25058 when visiting its source block.
25060 2018-01-23 Jakub Jelinek <jakub@redhat.com>
25063 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
25064 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
25066 2018-01-22 Jakub Jelinek <jakub@redhat.com>
25068 PR tree-optimization/83957
25069 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
25070 semicolon after for body surrounded by braces.
25072 PR tree-optimization/83081
25073 * profile-count.h (profile_probability::split): New method.
25074 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
25075 Use profile_probability::split.
25076 (do_compare_rtx_and_jump): Fix adjustment of probabilities
25077 when splitting a single conditional jump into 2.
25079 2018-01-22 David Malcolm <dmalcolm@redhat.com>
25081 PR tree-optimization/69452
25082 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
25085 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25087 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
25088 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
25089 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
25091 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25093 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
25095 * config/rl78/rl78.md (movdi): New define_expand.
25096 * config/rl78/rl78.c (rl78_split_movdi): New function.
25098 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25101 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
25103 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
25104 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
25105 128-bit to produce an UNSPEC move to get the double word with the
25106 signbit and then a shift directly to do signbit.
25107 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
25108 implementation with a new version that just does either a direct
25109 move or a regular move. Move memory interface to separate insns.
25110 Move insns so they are next to the expander.
25111 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
25112 with signbit move. Split big and little endian case.
25113 (signbit<mode>2_dm_mem_le): Likewise.
25114 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
25115 (signbit<mode>2_dm2): Likewise.
25117 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25119 * config/rl78/rl78.md (anddi3): New define_expand.
25121 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25123 * config/rl78/rl78.md (umindi3): New define_expand.
25125 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25127 * config/rl78/rl78.md (smindi3): New define_expand.
25129 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25131 * config/rl78/rl78.md (smaxdi3): New define_expand.
25133 2018-01-22 Carl Love <cel@us.ibm.com>
25135 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
25136 LVX_V1TI): Add macro expansion.
25137 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
25138 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
25139 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
25140 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
25141 Change check to determine if the instruction is a byte reversing
25142 entry. Fix typo in comment.
25143 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
25144 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
25145 Add def_builtin calls for new builtins.
25146 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
25147 Add define_insn expansion.
25149 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25151 * config/rl78/rl78.md (umaxdi3): New define_expand.
25153 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25155 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
25156 for non-QImode registers.
25158 2018-01-22 Richard Biener <rguenther@suse.de>
25160 PR tree-optimization/83963
25161 * graphite-scop-detection.c (scop_detection::get_sese): Delay
25162 including the loop exit block.
25163 (scop_detection::merge_sese): Likewise.
25164 (scop_detection::add_scop): Do it here instead.
25166 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25168 * doc/sourcebuild.texi (arm_softfloat): Document.
25170 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
25173 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
25174 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
25175 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
25177 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25178 David Edelsohn <dje.gcc@gmail.com>
25181 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
25182 Change "crset eq" to "crset 2".
25183 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
25184 (*call_indirect_aix<mode>_nospec): Likewise.
25185 (*call_value_indirect_aix<mode>_nospec): Likewise.
25186 (*call_indirect_elfv2<mode>_nospec): Likewise.
25187 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
25188 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
25189 change assembly output from . to $.
25190 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
25191 (indirect_jump<mode>_nospec): Change assembly output from . to $.
25192 (*tablejump<mode>_internal1_nospec): Likewise.
25194 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
25197 * config/sh/sh_optimize_sett_clrt.cc:
25198 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
25200 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
25202 PR tree-optimization/83940
25203 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
25204 offset_dt to vect_constant_def rather than vect_unknown_def_type.
25205 (vect_check_load_store_mask): Add a mask_dt_out parameter and
25206 use it to pass back the definition type.
25207 (vect_check_store_rhs): Likewise rhs_dt_out.
25208 (vect_build_gather_load_calls): Add a mask_dt argument and use
25209 it instead of a call to vect_is_simple_use.
25210 (vectorizable_store): Update calls to vect_check_load_store_mask
25211 and vect_check_store_rhs. Use the dt returned by the latter instead
25212 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
25213 instead of calls to vect_is_simple_use. Pass the scalar rather
25214 than the vector operand to vect_is_simple_use when handling
25215 second and subsequent copies of an rhs value.
25216 (vectorizable_load): Update calls to vect_check_load_store_mask
25217 and vect_build_gather_load_calls. Use the cached mask_dt and
25218 gs_info.offset_dt instead of calls to vect_is_simple_use.
25220 2018-01-20 Jakub Jelinek <jakub@redhat.com>
25222 PR middle-end/83945
25223 * tree-emutls.c: Include gimplify.h.
25224 (lower_emutls_2): New function.
25225 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
25226 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
25227 it before further processing.
25230 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
25231 UINTVAL (trueop1) instead of INTVAL (op1).
25233 2018-01-19 Jakub Jelinek <jakub@redhat.com>
25237 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
25238 INCOMING_FRAME_SP_OFFSET if not defined.
25239 (scan_trace): Add ENTRY argument. If true and
25240 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
25241 emit a note to adjust the CFA offset.
25242 (create_cfi_notes): Adjust scan_trace callers.
25243 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
25244 INCOMING_FRAME_SP_OFFSET in the CIE.
25245 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
25246 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
25248 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
25249 * doc/tm.texi: Regenerated.
25251 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25253 PR rtl-optimization/83147
25254 * lra-constraints.c (remove_inheritance_pseudos): Use
25255 lra_substitute_pseudo_within_insn.
25257 2018-01-19 Tom de Vries <tom@codesourcery.com>
25258 Cesar Philippidis <cesar@codesourcery.com>
25261 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
25263 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
25266 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
25267 spaces for function labels.
25269 2018-01-19 Martin Liska <mliska@suse.cz>
25271 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
25272 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
25273 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
25274 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
25275 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
25276 (PRED_CONST_RETURN): Change from 69 to 65.
25277 (PRED_NULL_RETURN): Change from 91 to 71.
25278 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
25279 (PRED_LOOP_GUARD): Change from 66 to 73.
25281 2018-01-19 Martin Liska <mliska@suse.cz>
25283 * predict.c (predict_insn_def): Add new assert.
25284 (struct branch_predictor): Change type to signed integer.
25285 (test_prediction_value_range): Amend test to cover
25286 PROB_UNINITIALIZED.
25287 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
25288 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
25289 (PRED_LOOP_ITERATIONS_MAX): Likewise.
25290 (PRED_LOOP_IV_COMPARE): Likewise.
25291 * predict.h (PROB_UNINITIALIZED): Define new constant.
25293 2018-01-19 Martin Liska <mliska@suse.cz>
25295 * predict.c (dump_prediction): Add new format for
25296 analyze_brprob.py script which is enabled with -details
25298 * profile-count.h (precise_p): New function.
25300 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
25302 PR tree-optimization/83922
25303 * tree-vect-loop.c (vect_verify_full_masking): Return false if
25304 there are no statements that need masking.
25305 (vect_active_double_reduction_p): New function.
25306 (vect_analyze_loop_operations): Use it when handling phis that
25307 are not in the loop header.
25309 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
25311 PR tree-optimization/83914
25312 * tree-vect-loop.c (vectorizable_induction): Don't convert
25313 init_expr or apply the peeling adjustment for inductions
25314 that are nested within the vectorized loop.
25316 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25318 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
25321 2018-01-18 Jakub Jelinek <jakub@redhat.com>
25325 * function.h (gimplify_parameters): Add gimple_seq * argument.
25326 * function.c: Include gimple.h and options.h.
25327 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
25328 for the added local temporaries if needed.
25329 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
25330 if there are any parameter cleanups, wrap whole body into a
25331 try/finally with the cleanups.
25333 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25336 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25337 Use GET_MODE_CLASS for scalar floating point.
25339 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
25343 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
25344 Fix call of call_cgraph_insertion_hooks.
25346 2018-01-18 Martin Sebor <msebor@redhat.com>
25348 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
25350 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
25353 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
25356 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
25359 * common.opt: (-ffile-prefix-map): New option.
25360 * opts.c (common_handle_option): Defer it.
25361 * opts-global.c (handle_common_deferred_options): Handle it.
25362 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
25363 * file-prefix-map.h: New file.
25364 (remap_debug_filename, add_debug_prefix_map): ...here.
25365 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
25366 * final.c (debug_prefix_map, add_debug_prefix_map
25367 remap_debug_filename): Move to...
25368 * file-prefix-map.c: New file.
25369 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
25370 generalize, get rid of alloca(), use strrchr() instead of strchr().
25371 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
25372 Implement in terms of add_prefix_map().
25373 (remap_macro_filename, remap_debug_filename): Implement in term of
25375 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
25376 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
25377 * dbxout.c: Include file-prefix-map.h.
25378 * varasm.c: Likewise.
25379 * vmsdbgout.c: Likewise.
25380 * xcoffout.c: Likewise.
25381 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
25382 * doc/cppopts.texi (-fmacro-prefix-map): Document.
25383 * doc/invoke.texi (-ffile-prefix-map): Document.
25384 (-fdebug-prefix-map): Update description.
25386 2018-01-18 Martin Liska <mliska@suse.cz>
25388 * config/i386/i386.c (indirect_thunk_name): Document that also
25390 (output_indirect_thunk): Document why both instructions
25391 (pause and lfence) are generated.
25393 2018-01-18 Richard Biener <rguenther@suse.de>
25395 PR tree-optimization/83887
25396 * graphite-scop-detection.c
25397 (scop_detection::get_nearest_dom_with_single_entry): Remove.
25398 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
25399 (scop_detection::merge_sese): Re-implement with a flood-fill
25400 algorithm that properly finds a SESE region if it exists.
25402 2018-01-18 Jakub Jelinek <jakub@redhat.com>
25405 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
25406 pointer_diff optimizations use view_convert instead of convert.
25408 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25410 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
25411 Generate different code for -mno-speculate-indirect-jumps.
25412 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
25413 (*call_indirect_aix<mode>): Disable for
25414 -mno-speculate-indirect-jumps.
25415 (*call_indirect_aix<mode>_nospec): New define_insn.
25416 (*call_value_indirect_aix<mode>): Disable for
25417 -mno-speculate-indirect-jumps.
25418 (*call_value_indirect_aix<mode>_nospec): New define_insn.
25419 (*sibcall_nonlocal_sysv<mode>): Generate different code for
25420 -mno-speculate-indirect-jumps.
25421 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
25423 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
25425 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
25426 long double type, set the flags for noting the default long double
25427 type, even if we don't pass or return a long double type.
25429 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
25432 * ipa-inline.c (flatten_function): Do not overwrite final inlining
25435 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
25437 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
25438 support for merge[hl].
25439 (fold_mergehl_helper): New helper function.
25440 (tree-vector-builder.h): New #include for tree_vector_builder usage.
25441 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
25442 (altivec_vmrglw_direct): Add xxmrglw insn.
25444 2018-01-17 Andrew Waterman <andrew@sifive.com>
25446 * config/riscv/riscv.c (riscv_conditional_register_usage): If
25447 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
25449 2018-01-17 David Malcolm <dmalcolm@redhat.com>
25452 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
25453 call the lto_location_cache before reading the
25454 DECL_SOURCE_LOCATION of the types.
25456 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25457 Richard Sandiford <richard.sandiford@linaro.org>
25459 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
25460 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
25461 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
25462 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
25463 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
25465 * config/aarch64/constraints.md (aarch64_movti_operand):
25467 * config/aarch64/predicates.md (Uti): Add new constraint.
25469 2018-01-17 Carl Love <cel@us.ibm.com>
25471 * config/rs6000/vsx.md (define_expand xl_len_r,
25472 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
25473 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
25475 (define_expand, define_insn): Move the shift left from the
25476 define_insn to the define_expand for lxvl and stxvl instructions.
25477 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
25478 and XL_LEN_R definitions to PURE.
25480 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
25482 * config/i386/i386.c (indirect_thunk_name): Declare regno
25483 as unsigned int. Compare regno with INVALID_REGNUM.
25484 (output_indirect_thunk): Ditto.
25485 (output_indirect_thunk_function): Ditto.
25486 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
25487 in the call to output_indirect_thunk_function.
25489 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
25491 PR middle-end/83884
25492 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
25493 rather than the size of inner_type to determine the stack slot size
25494 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
25496 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
25499 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
25502 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25504 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
25505 endian Linux systems to optionally enable multilibs for selecting
25506 the long double type if the user configured an explicit type.
25507 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
25508 have no long double multilibs if not defined.
25509 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
25510 warn if the user used -mabi={ieee,ibm}longdouble and we built
25511 multilibs for long double.
25512 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
25513 appropriate multilib option.
25514 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
25516 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
25517 for building long double multilibs.
25518 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
25520 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
25522 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
25525 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
25527 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
25530 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
25533 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
25536 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25538 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
25539 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
25541 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
25543 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
25544 different rtl trees depending on TARGET_64BIT.
25545 (rs6000_gen_lvx): Likewise.
25547 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25549 * config/visium/visium.md (nop): Tweak comment.
25550 (hazard_nop): Likewise.
25552 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25554 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
25555 -mspeculate-indirect-jumps.
25556 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
25557 for -mno-speculate-indirect-jumps.
25558 (*call_indirect_elfv2<mode>_nospec): New define_insn.
25559 (*call_value_indirect_elfv2<mode>): Disable for
25560 -mno-speculate-indirect-jumps.
25561 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
25562 (indirect_jump): Emit different RTL for
25563 -mno-speculate-indirect-jumps.
25564 (*indirect_jump<mode>): Disable for
25565 -mno-speculate-indirect-jumps.
25566 (*indirect_jump<mode>_nospec): New define_insn.
25567 (tablejump): Emit different RTL for
25568 -mno-speculate-indirect-jumps.
25569 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
25570 (tablejumpsi_nospec): New define_expand.
25571 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
25572 (tablejumpdi_nospec): New define_expand.
25573 (*tablejump<mode>_internal1): Disable for
25574 -mno-speculate-indirect-jumps.
25575 (*tablejump<mode>_internal1_nospec): New define_insn.
25576 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
25579 2018-01-16 Artyom Skrobov tyomitch@gmail.com
25581 * caller-save.c (insert_save): Drop unnecessary parameter. All
25584 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25585 Richard Biener <rguenth@suse.de>
25588 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
25589 return early, inline manually is_gimple_sizepos. Make sure if we
25590 call gimplify_expr we don't end up with a gimple constant.
25591 * tree.c (variably_modified_type_p): Don't return true for
25592 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
25593 * gimplify.h (is_gimple_sizepos): Remove.
25595 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25597 PR tree-optimization/83857
25598 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
25599 vectorizable_live_operation for pure SLP statements.
25600 (vectorizable_live_operation): Handle PHIs.
25602 2018-01-16 Richard Biener <rguenther@suse.de>
25604 PR tree-optimization/83867
25605 * tree-vect-stmts.c (vect_transform_stmt): Precompute
25606 nested_in_vect_loop_p since the scalar stmt may get invalidated.
25608 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25611 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
25612 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
25613 If off is not INTEGER_CST, issue a may not be aligned warning
25614 rather than isn't aligned. Use isn%'t rather than isn't.
25615 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
25617 <case MULT_EXPR>: Improve the case when bottom and one of the
25618 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
25619 operand, in that case check if the other operand is multiple of
25620 bottom divided by the INTEGER_CST operand.
25622 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25625 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
25626 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
25627 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
25628 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
25629 * config/pa/pa.c (pa_function_arg_advance): Likewise.
25630 (pa_function_arg, pa_arg_partial_bytes): Likewise.
25631 (pa_function_arg_size): New function.
25633 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25635 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
25636 in a separate statement.
25638 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25640 PR tree-optimization/83847
25641 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
25642 group gathers and scatters.
25644 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25646 PR rtl-optimization/86620
25647 * params.def (max-sched-ready-insns): Bump minimum value to 1.
25649 PR rtl-optimization/83213
25650 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
25651 to last if both are JUMP_INSNs.
25653 PR tree-optimization/83843
25654 * gimple-ssa-store-merging.c
25655 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
25656 store_immediate_info for bswap/nop orig_stores.
25658 2018-01-15 Andrew Waterman <andrew@sifive.com>
25660 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
25662 <UDIV>: Increase cost if !TARGET_DIV.
25664 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25666 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
25667 (define_attr "cr_logical_3op"): New.
25668 (cceq_ior_compare): Adjust.
25669 (cceq_ior_compare_complement): Adjust.
25670 (*cceq_rev_compare): Adjust.
25671 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
25672 (is_cracked_insn): Adjust.
25673 (insn_must_be_first_in_group): Adjust.
25674 * config/rs6000/40x.md: Adjust.
25675 * config/rs6000/440.md: Adjust.
25676 * config/rs6000/476.md: Adjust.
25677 * config/rs6000/601.md: Adjust.
25678 * config/rs6000/603.md: Adjust.
25679 * config/rs6000/6xx.md: Adjust.
25680 * config/rs6000/7450.md: Adjust.
25681 * config/rs6000/7xx.md: Adjust.
25682 * config/rs6000/8540.md: Adjust.
25683 * config/rs6000/cell.md: Adjust.
25684 * config/rs6000/e300c2c3.md: Adjust.
25685 * config/rs6000/e500mc.md: Adjust.
25686 * config/rs6000/e500mc64.md: Adjust.
25687 * config/rs6000/e5500.md: Adjust.
25688 * config/rs6000/e6500.md: Adjust.
25689 * config/rs6000/mpc.md: Adjust.
25690 * config/rs6000/power4.md: Adjust.
25691 * config/rs6000/power5.md: Adjust.
25692 * config/rs6000/power6.md: Adjust.
25693 * config/rs6000/power7.md: Adjust.
25694 * config/rs6000/power8.md: Adjust.
25695 * config/rs6000/power9.md: Adjust.
25696 * config/rs6000/rs64.md: Adjust.
25697 * config/rs6000/titan.md: Adjust.
25699 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25701 * config/i386/predicates.md (indirect_branch_operand): Rewrite
25702 ix86_indirect_branch_register logic.
25704 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25706 * config/i386/constraints.md (Bs): Update
25707 ix86_indirect_branch_register check. Don't check
25708 ix86_indirect_branch_register with GOT_memory_operand.
25710 * config/i386/predicates.md (GOT_memory_operand): Don't check
25711 ix86_indirect_branch_register here.
25712 (GOT32_symbol_operand): Likewise.
25714 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25716 * config/i386/predicates.md (constant_call_address_operand):
25717 Rewrite ix86_indirect_branch_register logic.
25718 (sibcall_insn_operand): Likewise.
25720 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25722 * config/i386/constraints.md (Bs): Replace
25723 ix86_indirect_branch_thunk_register with
25724 ix86_indirect_branch_register.
25726 * config/i386/i386.md (indirect_jump): Likewise.
25727 (tablejump): Likewise.
25728 (*sibcall_memory): Likewise.
25729 (*sibcall_value_memory): Likewise.
25730 Peepholes of indirect call and jump via memory: Likewise.
25731 * config/i386/i386.opt: Likewise.
25732 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25733 (GOT_memory_operand): Likewise.
25734 (call_insn_operand): Likewise.
25735 (sibcall_insn_operand): Likewise.
25736 (GOT32_symbol_operand): Likewise.
25738 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25740 PR middle-end/83837
25741 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
25742 type rather than type addr's type points to.
25743 (expand_omp_atomic_mutex): Likewise.
25744 (expand_omp_atomic): Likewise.
25746 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25749 * config/i386/i386.c (output_indirect_thunk_function): Use
25750 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
25751 for __x86_return_thunk.
25753 2018-01-15 Richard Biener <rguenther@suse.de>
25755 PR middle-end/83850
25756 * expmed.c (extract_bit_field_1): Fix typo.
25758 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25761 * config/arm/iterators.md (VF): New mode iterator.
25762 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
25763 Remove integer-related logic from pattern.
25764 (neon_vabd<mode>_3): Likewise.
25766 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25768 PR middle-end/82694
25769 * common.opt (fstrict-overflow): No longer an alias.
25770 (fwrapv-pointer): New option.
25771 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
25772 also for pointer types based on flag_wrapv_pointer.
25773 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
25774 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
25775 opts->x_flag_wrapv got set.
25776 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
25777 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
25778 POINTER_TYPE_OVERFLOW_UNDEFINED.
25779 * match.pd: Likewise in address comparison pattern.
25780 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
25782 2018-01-15 Richard Biener <rguenther@suse.de>
25785 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
25786 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
25787 Reset type names to their identifier if their TYPE_DECL doesn't
25788 have linkage (and thus is used for ODR and devirt).
25789 (save_debug_info_for_decl): Remove.
25790 (save_debug_info_for_type): Likewise.
25791 (add_tree_to_fld_list): Adjust.
25792 * tree-pretty-print.c (dump_generic_node): Make dumping of
25793 type names more robust.
25795 2018-01-15 Richard Biener <rguenther@suse.de>
25797 * BASE-VER: Bump to 8.0.1.
25799 2018-01-14 Martin Sebor <msebor@redhat.com>
25802 * builtins.c (check_access): Avoid warning when the no-warning bit
25805 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
25807 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
25808 * ira-color (allocno_hard_regs_compare): Likewise.
25810 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
25813 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
25814 Use .pushsection/.popsection.
25816 2018-01-14 Martin Sebor <msebor@redhat.com>
25819 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
25821 2018-01-14 Jakub Jelinek <jakub@redhat.com>
25823 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
25824 entry from extra_headers.
25825 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
25826 extra_headers, make the list bitwise identical to the i?86-*-* one.
25828 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25830 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
25831 -mcmodel=large with -mindirect-branch=thunk,
25832 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
25833 -mfunction-return=thunk-extern.
25834 * doc/invoke.texi: Document -mcmodel=large is incompatible with
25835 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
25836 -mfunction-return=thunk and -mfunction-return=thunk-extern.
25838 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25840 * config/i386/i386.c (print_reg): Print the name of the full
25841 integer register without '%'.
25842 (ix86_print_operand): Handle 'V'.
25843 * doc/extend.texi: Document 'V' modifier.
25845 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25847 * config/i386/constraints.md (Bs): Disallow memory operand for
25848 -mindirect-branch-register.
25850 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25851 (GOT_memory_operand): Likewise.
25852 (call_insn_operand): Likewise.
25853 (sibcall_insn_operand): Likewise.
25854 (GOT32_symbol_operand): Likewise.
25855 * config/i386/i386.md (indirect_jump): Call convert_memory_address
25856 for -mindirect-branch-register.
25857 (tablejump): Likewise.
25858 (*sibcall_memory): Likewise.
25859 (*sibcall_value_memory): Likewise.
25860 Disallow peepholes of indirect call and jump via memory for
25861 -mindirect-branch-register.
25862 (*call_pop): Replace m with Bw.
25863 (*call_value_pop): Likewise.
25864 (*sibcall_pop_memory): Replace m with Bs.
25865 * config/i386/i386.opt (mindirect-branch-register): New option.
25866 * doc/invoke.texi: Document -mindirect-branch-register option.
25868 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25870 * config/i386/i386-protos.h (ix86_output_function_return): New.
25871 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
25872 set function_return_type.
25873 (indirect_thunk_name): Add ret_p to indicate thunk for function
25875 (output_indirect_thunk_function): Pass false to
25876 indirect_thunk_name.
25877 (ix86_output_indirect_branch_via_reg): Likewise.
25878 (ix86_output_indirect_branch_via_push): Likewise.
25879 (output_indirect_thunk_function): Create alias for function
25880 return thunk if regno < 0.
25881 (ix86_output_function_return): New function.
25882 (ix86_handle_fndecl_attribute): Handle function_return.
25883 (ix86_attribute_table): Add function_return.
25884 * config/i386/i386.h (machine_function): Add
25885 function_return_type.
25886 * config/i386/i386.md (simple_return_internal): Use
25887 ix86_output_function_return.
25888 (simple_return_internal_long): Likewise.
25889 * config/i386/i386.opt (mfunction-return=): New option.
25890 (indirect_branch): Mention -mfunction-return=.
25891 * doc/extend.texi: Document function_return function attribute.
25892 * doc/invoke.texi: Document -mfunction-return= option.
25894 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25896 * config/i386/i386-opts.h (indirect_branch): New.
25897 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
25898 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
25899 with local indirect jump when converting indirect call and jump.
25900 (ix86_set_indirect_branch_type): New.
25901 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
25902 (indirectlabelno): New.
25903 (indirect_thunk_needed): Likewise.
25904 (indirect_thunk_bnd_needed): Likewise.
25905 (indirect_thunks_used): Likewise.
25906 (indirect_thunks_bnd_used): Likewise.
25907 (INDIRECT_LABEL): Likewise.
25908 (indirect_thunk_name): Likewise.
25909 (output_indirect_thunk): Likewise.
25910 (output_indirect_thunk_function): Likewise.
25911 (ix86_output_indirect_branch_via_reg): Likewise.
25912 (ix86_output_indirect_branch_via_push): Likewise.
25913 (ix86_output_indirect_branch): Likewise.
25914 (ix86_output_indirect_jmp): Likewise.
25915 (ix86_code_end): Call output_indirect_thunk_function if needed.
25916 (ix86_output_call_insn): Call ix86_output_indirect_branch if
25918 (ix86_handle_fndecl_attribute): Handle indirect_branch.
25919 (ix86_attribute_table): Add indirect_branch.
25920 * config/i386/i386.h (machine_function): Add indirect_branch_type
25921 and has_local_indirect_jump.
25922 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
25924 (tablejump): Likewise.
25925 (*indirect_jump): Use ix86_output_indirect_jmp.
25926 (*tablejump_1): Likewise.
25927 (simple_return_indirect_internal): Likewise.
25928 * config/i386/i386.opt (mindirect-branch=): New option.
25929 (indirect_branch): New.
25932 (thunk-inline): Likewise.
25933 (thunk-extern): Likewise.
25934 * doc/extend.texi: Document indirect_branch function attribute.
25935 * doc/invoke.texi: Document -mindirect-branch= option.
25937 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
25940 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
25942 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
25944 * ipa-inline.c (want_inline_small_function_p): Return false if
25945 inlining has already failed with CIF_FINAL_ERROR.
25946 (update_caller_keys): Call want_inline_small_function_p before
25948 (update_callee_keys): Likewise.
25950 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25952 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
25954 (rs6000_quadword_masked_address_p): Likewise.
25955 (quad_aligned_load_p): Likewise.
25956 (quad_aligned_store_p): Likewise.
25957 (const_load_sequence_p): Add comment to describe the outer-most loop.
25958 (mimic_memory_attributes_and_flags): New function.
25959 (rs6000_gen_stvx): Likewise.
25960 (replace_swapped_aligned_store): Likewise.
25961 (rs6000_gen_lvx): Likewise.
25962 (replace_swapped_aligned_load): Likewise.
25963 (replace_swapped_load_constant): Capitalize argument name in
25964 comment describing this function.
25965 (rs6000_analyze_swaps): Add a third pass to search for vector loads
25966 and stores that access quad-word aligned addresses and replace
25967 with stvx or lvx instructions when appropriate.
25968 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
25969 New function prototype.
25970 (rs6000_quadword_masked_address_p): Likewise.
25971 (rs6000_gen_lvx): Likewise.
25972 (rs6000_gen_stvx): Likewise.
25973 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
25974 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
25975 when memory address is aligned.
25976 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
25977 this split to select lvx instruction when memory address is aligned.
25978 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
25979 instruction when memory address is aligned.
25980 (*vsx_le_perm_load_v16qi): Likewise.
25981 (four unnamed splitters): Modify to select the stvx instruction
25982 when memory is aligned.
25984 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
25986 * predict.c (determine_unlikely_bbs): Handle correctly BBs
25987 which appears in the queue multiple times.
25989 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25990 Alan Hayward <alan.hayward@arm.com>
25991 David Sherwood <david.sherwood@arm.com>
25993 * tree-vectorizer.h (vec_lower_bound): New structure.
25994 (_loop_vec_info): Add check_nonzero and lower_bounds.
25995 (LOOP_VINFO_CHECK_NONZERO): New macro.
25996 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
25997 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
25998 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
25999 fields. Make seg_len the distance travelled, not including the
26001 (dr_direction_indicator): Declare.
26002 (dr_zero_step_indicator): Likewise.
26003 (dr_known_forward_stride_p): Likewise.
26004 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
26006 (runtime_alias_check_p): Allow runtime alias checks with
26008 (operator ==): Compare access_size and align.
26009 (prune_runtime_alias_test_list): Rework for new distinction between
26010 the access_size and seg_len.
26011 (create_intersect_range_checks_index): Likewise. Cope with polynomial
26013 (get_segment_min_max): New function.
26014 (create_intersect_range_checks): Use it.
26015 (dr_step_indicator): New function.
26016 (dr_direction_indicator): Likewise.
26017 (dr_zero_step_indicator): Likewise.
26018 (dr_known_forward_stride_p): Likewise.
26019 * tree-loop-distribution.c (data_ref_segment_size): Return
26020 DR_STEP * (niters - 1).
26021 (compute_alias_check_pairs): Update call to the dr_with_seg_len
26023 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
26024 (vect_preserves_scalar_order_p): New function, split out from...
26025 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
26026 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
26027 (vect_vfa_access_size): New function.
26028 (vect_vfa_align): Likewise.
26029 (vect_compile_time_alias): Take access_size_a and access_b arguments.
26030 (dump_lower_bound): New function.
26031 (vect_check_lower_bound): Likewise.
26032 (vect_small_gap_p): Likewise.
26033 (vectorizable_with_step_bound_p): Likewise.
26034 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
26035 depencies if the vectorization factor is 1. Convert the checks
26036 for nonzero steps into checks on the bounds of DR_STEP. Try using
26037 a bunds check for variable steps if the minimum required step is
26038 relatively small. Update calls to the dr_with_seg_len
26039 constructor and to vect_compile_time_alias.
26040 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
26042 (vect_loop_versioning): Call it.
26043 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
26045 (vect_estimate_min_profitable_iters): Account for any bounds checks.
26047 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26048 Alan Hayward <alan.hayward@arm.com>
26049 David Sherwood <david.sherwood@arm.com>
26051 * doc/sourcebuild.texi (vect_scatter_store): Document.
26052 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
26054 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
26056 * genopinit.c (main): Add supports_vec_scatter_store and
26057 supports_vec_scatter_store_cached to target_optabs.
26058 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
26059 IFN_MASK_SCATTER_STORE.
26060 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
26062 * internal-fn.h (internal_store_fn_p): Declare.
26063 (internal_fn_stored_value_index): Likewise.
26064 * internal-fn.c (scatter_store_direct): New macro.
26065 (expand_scatter_store_optab_fn): New function.
26066 (direct_scatter_store_optab_supported_p): New macro.
26067 (internal_store_fn_p): New function.
26068 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
26069 IFN_MASK_SCATTER_STORE.
26070 (internal_fn_mask_index): Likewise.
26071 (internal_fn_stored_value_index): New function.
26072 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
26073 for scatter stores.
26074 * optabs-query.h (supports_vec_scatter_store_p): Declare.
26075 * optabs-query.c (supports_vec_scatter_store_p): New function.
26076 * tree-vectorizer.h (vect_get_store_rhs): Declare.
26077 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
26078 true for scatter stores.
26079 (vect_gather_scatter_fn_p): Handle scatter stores too.
26080 (vect_check_gather_scatter): Consider using scatter stores if
26081 supports_vec_scatter_store_p.
26082 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
26083 scatter stores too.
26084 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
26085 internal_fn_stored_value_index.
26086 (check_load_store_masking): Handle scatter stores too.
26087 (vect_get_store_rhs): Make public.
26088 (vectorizable_call): Use internal_store_fn_p.
26089 (vectorizable_store): Handle scatter store internal functions.
26090 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
26091 when deciding whether the end of the group has been reached.
26092 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
26093 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
26094 (mask_scatter_store<mode>): New insns.
26096 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26097 Alan Hayward <alan.hayward@arm.com>
26098 David Sherwood <david.sherwood@arm.com>
26100 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
26101 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
26102 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
26104 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
26105 Use vect_truncate_gather_scatter_offset if we can't treat the
26106 operation as a normal gather load or scatter store.
26107 (get_group_load_store_type): Take the gather_scatter_info
26108 as argument. Try using a gather load or scatter store for
26109 single-element groups.
26110 (get_load_store_type): Update calls to get_group_load_store_type
26111 and vect_use_strided_gather_scatters_p.
26113 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26114 Alan Hayward <alan.hayward@arm.com>
26115 David Sherwood <david.sherwood@arm.com>
26117 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
26118 optional tree argument.
26119 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
26121 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
26122 but continue to use the current value as a fallback.
26123 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
26124 to compare the updates.
26125 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
26126 (get_load_store_type): Use it when handling a strided access.
26127 (vect_get_strided_load_store_ops): New function.
26128 (vect_get_data_ptr_increment): Likewise.
26129 (vectorizable_load): Handle strided gather loads. Always pass
26130 a step to vect_create_data_ref_ptr and bump_vector_ptr.
26132 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26133 Alan Hayward <alan.hayward@arm.com>
26134 David Sherwood <david.sherwood@arm.com>
26136 * doc/md.texi (gather_load@var{m}): Document.
26137 (mask_gather_load@var{m}): Likewise.
26138 * genopinit.c (main): Add supports_vec_gather_load and
26139 supports_vec_gather_load_cached to target_optabs.
26140 * optabs-tree.c (init_tree_optimization_optabs): Use
26141 ggc_cleared_alloc to allocate target_optabs.
26142 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
26143 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
26145 * internal-fn.h (internal_load_fn_p): Declare.
26146 (internal_gather_scatter_fn_p): Likewise.
26147 (internal_fn_mask_index): Likewise.
26148 (internal_gather_scatter_fn_supported_p): Likewise.
26149 * internal-fn.c (gather_load_direct): New macro.
26150 (expand_gather_load_optab_fn): New function.
26151 (direct_gather_load_optab_supported_p): New macro.
26152 (direct_internal_fn_optab): New function.
26153 (internal_load_fn_p): Likewise.
26154 (internal_gather_scatter_fn_p): Likewise.
26155 (internal_fn_mask_index): Likewise.
26156 (internal_gather_scatter_fn_supported_p): Likewise.
26157 * optabs-query.c (supports_at_least_one_mode_p): New function.
26158 (supports_vec_gather_load_p): Likewise.
26159 * optabs-query.h (supports_vec_gather_load_p): Declare.
26160 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
26161 and memory_type field.
26162 (NUM_PATTERNS): Bump to 15.
26163 * tree-vect-data-refs.c: Include internal-fn.h.
26164 (vect_gather_scatter_fn_p): New function.
26165 (vect_describe_gather_scatter_call): Likewise.
26166 (vect_check_gather_scatter): Try using internal functions for
26167 gather loads. Recognize existing calls to a gather load function.
26168 (vect_analyze_data_refs): Consider using gather loads if
26169 supports_vec_gather_load_p.
26170 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
26171 (vect_get_gather_scatter_offset_type): Likewise.
26172 (vect_convert_mask_for_vectype): Likewise.
26173 (vect_add_conversion_to_patterm): Likewise.
26174 (vect_try_gather_scatter_pattern): Likewise.
26175 (vect_recog_gather_scatter_pattern): New pattern recognizer.
26176 (vect_vect_recog_func_ptrs): Add it.
26177 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
26178 internal_fn_mask_index and internal_gather_scatter_fn_p.
26179 (check_load_store_masking): Take the gather_scatter_info as an
26180 argument and handle gather loads.
26181 (vect_get_gather_scatter_ops): New function.
26182 (vectorizable_call): Check internal_load_fn_p.
26183 (vectorizable_load): Likewise. Handle gather load internal
26185 (vectorizable_store): Update call to check_load_store_masking.
26186 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
26187 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
26188 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
26189 (aarch64_gather_scale_operand_d): New predicates.
26190 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
26191 (mask_gather_load<mode>): New insns.
26193 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26194 Alan Hayward <alan.hayward@arm.com>
26195 David Sherwood <david.sherwood@arm.com>
26197 * optabs.def (fold_left_plus_optab): New optab.
26198 * doc/md.texi (fold_left_plus_@var{m}): Document.
26199 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
26200 * internal-fn.c (fold_left_direct): Define.
26201 (expand_fold_left_optab_fn): Likewise.
26202 (direct_fold_left_optab_supported_p): Likewise.
26203 * fold-const-call.c (fold_const_fold_left): New function.
26204 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
26205 * tree-parloops.c (valid_reduction_p): New function.
26206 (gather_scalar_reductions): Use it.
26207 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
26208 (vect_finish_replace_stmt): Declare.
26209 * tree-vect-loop.c (fold_left_reduction_fn): New function.
26210 (needs_fold_left_reduction_p): New function, split out from...
26211 (vect_is_simple_reduction): ...here. Accept reductions that
26212 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
26213 (vect_force_simple_reduction): Also store the reduction type in
26214 the assignment's STMT_VINFO_REDUC_TYPE.
26215 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
26216 (merge_with_identity): New function.
26217 (vect_expand_fold_left): Likewise.
26218 (vectorize_fold_left_reduction): Likewise.
26219 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
26220 scalar phi in place for it. Check for target support and reject
26221 cases that would reassociate the operation. Defer the transform
26222 phase to vectorize_fold_left_reduction.
26223 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
26224 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
26225 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
26227 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26229 * tree-if-conv.c (predicate_mem_writes): Remove redundant
26230 call to ifc_temp_var.
26232 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26233 Alan Hayward <alan.hayward@arm.com>
26234 David Sherwood <david.sherwood@arm.com>
26236 * target.def (legitimize_address_displacement): Take the original
26237 offset as a poly_int.
26238 * targhooks.h (default_legitimize_address_displacement): Update
26240 * targhooks.c (default_legitimize_address_displacement): Likewise.
26241 * doc/tm.texi: Regenerate.
26242 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
26243 as an argument, moving assert of ad->disp == ad->disp_term to...
26244 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
26245 Try calling targetm.legitimize_address_displacement before expanding
26246 the address rather than afterwards, and adjust for the new interface.
26247 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
26248 Match the new hook interface. Handle SVE addresses.
26249 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
26250 new hook interface.
26252 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26254 * Makefile.in (OBJS): Add early-remat.o.
26255 * target.def (select_early_remat_modes): New hook.
26256 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
26257 * doc/tm.texi: Regenerate.
26258 * targhooks.h (default_select_early_remat_modes): Declare.
26259 * targhooks.c (default_select_early_remat_modes): New function.
26260 * timevar.def (TV_EARLY_REMAT): New timevar.
26261 * passes.def (pass_early_remat): New pass.
26262 * tree-pass.h (make_pass_early_remat): Declare.
26263 * early-remat.c: New file.
26264 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
26266 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
26268 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26269 Alan Hayward <alan.hayward@arm.com>
26270 David Sherwood <david.sherwood@arm.com>
26272 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
26273 vfm1 with a bound_epilog parameter.
26274 (vect_do_peeling): Update calls accordingly, and move the prologue
26275 call earlier in the function. Treat the base bound_epilog as 0 for
26276 fully-masked loops and retain vf - 1 for other loops. Add 1 to
26277 this base when peeling for gaps.
26278 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
26279 with fully-masked loops.
26280 (vect_estimate_min_profitable_iters): Handle the single peeled
26281 iteration in that case.
26283 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26284 Alan Hayward <alan.hayward@arm.com>
26285 David Sherwood <david.sherwood@arm.com>
26287 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
26288 single-element interleaving even if the size is not a power of 2.
26289 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
26290 accesses for single-element interleaving if the group size is
26293 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26294 Alan Hayward <alan.hayward@arm.com>
26295 David Sherwood <david.sherwood@arm.com>
26297 * doc/md.texi (fold_extract_last_@var{m}): Document.
26298 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
26299 * optabs.def (fold_extract_last_optab): New optab.
26300 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
26301 * internal-fn.c (fold_extract_direct): New macro.
26302 (expand_fold_extract_optab_fn): Likewise.
26303 (direct_fold_extract_optab_supported_p): Likewise.
26304 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
26305 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26306 EXTRACT_LAST_REDUCTION.
26307 (get_initial_def_for_reduction): Do not create an initial vector
26308 for EXTRACT_LAST_REDUCTION reductions.
26309 (vectorizable_reduction): Leave the scalar phi in place for
26310 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
26311 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
26312 epilogue code for EXTRACT_LAST_REDUCTION and defer the
26313 transform phase to vectorizable_condition.
26314 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
26316 (vect_finish_stmt_generation): ...here.
26317 (vect_finish_replace_stmt): New function.
26318 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
26319 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
26321 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
26323 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26324 Alan Hayward <alan.hayward@arm.com>
26325 David Sherwood <david.sherwood@arm.com>
26327 * doc/md.texi (extract_last_@var{m}): Document.
26328 * optabs.def (extract_last_optab): New optab.
26329 * internal-fn.def (EXTRACT_LAST): New internal function.
26330 * internal-fn.c (cond_unary_direct): New macro.
26331 (expand_cond_unary_optab_fn): Likewise.
26332 (direct_cond_unary_optab_supported_p): Likewise.
26333 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
26334 loops using EXTRACT_LAST.
26335 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
26336 (extract_last_<mode>): ...this optab.
26337 (vec_extract<mode><Vel>): Update accordingly.
26339 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26340 Alan Hayward <alan.hayward@arm.com>
26341 David Sherwood <david.sherwood@arm.com>
26343 * target.def (empty_mask_is_expensive): New hook.
26344 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
26345 * doc/tm.texi: Regenerate.
26346 * targhooks.h (default_empty_mask_is_expensive): Declare.
26347 * targhooks.c (default_empty_mask_is_expensive): New function.
26348 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
26349 if the target says that empty masks are expensive.
26350 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
26352 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
26354 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26355 Alan Hayward <alan.hayward@arm.com>
26356 David Sherwood <david.sherwood@arm.com>
26358 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
26359 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
26360 (vect_use_loop_mask_for_alignment_p): New function.
26361 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
26362 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
26363 niters_skip argument. Make sure that the first niters_skip elements
26364 of the first iteration are inactive.
26365 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
26366 Update call to vect_set_loop_masks_directly.
26367 (get_misalign_in_elems): New function, split out from...
26368 (vect_gen_prolog_loop_niters): ...here.
26369 (vect_update_init_of_dr): Take a code argument that specifies whether
26370 the adjustment should be added or subtracted.
26371 (vect_update_init_of_drs): Likewise.
26372 (vect_prepare_for_masked_peels): New function.
26373 (vect_do_peeling): Skip prologue peeling if we're using a mask
26374 instead. Update call to vect_update_inits_of_drs.
26375 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26377 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
26378 alignment. Do not include the number of peeled iterations in
26379 the minimum threshold in that case.
26380 (vectorizable_induction): Adjust the start value down by
26381 LOOP_VINFO_MASK_SKIP_NITERS iterations.
26382 (vect_transform_loop): Call vect_prepare_for_masked_peels.
26383 Take the number of skipped iterations into account when calculating
26385 * tree-vect-stmts.c (vect_gen_while_not): New function.
26387 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26388 Alan Hayward <alan.hayward@arm.com>
26389 David Sherwood <david.sherwood@arm.com>
26391 * doc/sourcebuild.texi (vect_fully_masked): Document.
26392 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
26393 default value to 0.
26394 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
26396 (vect_analyze_loop_2): ...here. Don't check the vectorization
26397 factor against the number of loop iterations if the loop is
26400 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26401 Alan Hayward <alan.hayward@arm.com>
26402 David Sherwood <david.sherwood@arm.com>
26404 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
26405 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
26406 (dump_groups): Update accordingly.
26407 (iv_use::mem_type): New member variable.
26408 (address_p): New function.
26409 (record_use): Add a mem_type argument and initialize the new
26411 (record_group_use): Add a mem_type argument. Use address_p.
26412 Remove obsolete null checks of base_object. Update call to record_use.
26413 (find_interesting_uses_op): Update call to record_group_use.
26414 (find_interesting_uses_cond): Likewise.
26415 (find_interesting_uses_address): Likewise.
26416 (get_mem_type_for_internal_fn): New function.
26417 (find_address_like_use): Likewise.
26418 (find_interesting_uses_stmt): Try find_address_like_use before
26419 calling find_interesting_uses_op.
26420 (addr_offset_valid_p): Use the iv mem_type field as the type
26421 of the addressed memory.
26422 (add_autoinc_candidates): Likewise.
26423 (get_address_cost): Likewise.
26424 (split_small_address_groups_p): Use address_p.
26425 (split_address_groups): Likewise.
26426 (add_iv_candidate_for_use): Likewise.
26427 (autoinc_possible_for_pair): Likewise.
26428 (rewrite_groups): Likewise.
26429 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
26430 (determine_group_iv_cost): Update after split of USE_ADDRESS.
26431 (get_alias_ptr_type_for_ptr_address): New function.
26432 (rewrite_use_address): Rewrite address uses in calls that were
26433 identified by find_address_like_use.
26435 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26436 Alan Hayward <alan.hayward@arm.com>
26437 David Sherwood <david.sherwood@arm.com>
26439 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
26441 * gimple-expr.h (is_gimple_addressable: Likewise.
26442 * gimple-expr.c (is_gimple_address): Likewise.
26443 * internal-fn.c (expand_call_mem_ref): New function.
26444 (expand_mask_load_optab_fn): Use it.
26445 (expand_mask_store_optab_fn): Likewise.
26447 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26448 Alan Hayward <alan.hayward@arm.com>
26449 David Sherwood <david.sherwood@arm.com>
26451 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
26452 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
26453 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
26454 (cond_umax@var{mode}): Document.
26455 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
26456 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
26457 (cond_umin_optab, cond_umax_optab): New optabs.
26458 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
26459 (COND_IOR, COND_XOR): New internal functions.
26460 * internal-fn.h (get_conditional_internal_fn): Declare.
26461 * internal-fn.c (cond_binary_direct): New macro.
26462 (expand_cond_binary_optab_fn): Likewise.
26463 (direct_cond_binary_optab_supported_p): Likewise.
26464 (get_conditional_internal_fn): New function.
26465 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
26466 Cope with reduction statements that are vectorized as calls rather
26468 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
26469 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
26470 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
26471 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
26472 (UNSPEC_COND_EOR): New unspecs.
26473 (optab): Add mappings for them.
26474 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
26475 (sve_int_op, sve_fp_op): New int attributes.
26477 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26478 Alan Hayward <alan.hayward@arm.com>
26479 David Sherwood <david.sherwood@arm.com>
26481 * optabs.def (while_ult_optab): New optab.
26482 * doc/md.texi (while_ult@var{m}@var{n}): Document.
26483 * internal-fn.def (WHILE_ULT): New internal function.
26484 * internal-fn.h (direct_internal_fn_supported_p): New override
26485 that takes two types as argument.
26486 * internal-fn.c (while_direct): New macro.
26487 (expand_while_optab_fn): New function.
26488 (convert_optab_supported_p): Likewise.
26489 (direct_while_optab_supported_p): New macro.
26490 * wide-int.h (wi::udiv_ceil): New function.
26491 * tree-vectorizer.h (rgroup_masks): New structure.
26492 (vec_loop_masks): New typedef.
26493 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
26494 and fully_masked_p.
26495 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
26496 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
26497 (vect_max_vf): New function.
26498 (slpeel_make_loop_iterate_ntimes): Delete.
26499 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
26500 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
26501 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
26502 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
26503 internal-fn.h, stor-layout.h and optabs-query.h.
26504 (vect_set_loop_mask): New function.
26505 (add_preheader_seq): Likewise.
26506 (add_header_seq): Likewise.
26507 (interleave_supported_p): Likewise.
26508 (vect_maybe_permute_loop_masks): Likewise.
26509 (vect_set_loop_masks_directly): Likewise.
26510 (vect_set_loop_condition_masked): Likewise.
26511 (vect_set_loop_condition_unmasked): New function, split out from
26512 slpeel_make_loop_iterate_ntimes.
26513 (slpeel_make_loop_iterate_ntimes): Rename to..
26514 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
26515 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
26516 (vect_do_peeling): Update call accordingly.
26517 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
26519 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26520 mask_compare_type, can_fully_mask_p and fully_masked_p.
26521 (release_vec_loop_masks): New function.
26522 (_loop_vec_info): Use it to free the loop masks.
26523 (can_produce_all_loop_masks_p): New function.
26524 (vect_get_max_nscalars_per_iter): Likewise.
26525 (vect_verify_full_masking): Likewise.
26526 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
26527 retries, and free the mask rgroups before retrying. Check loop-wide
26528 reasons for disallowing fully-masked loops. Make the final decision
26529 about whether use a fully-masked loop or not.
26530 (vect_estimate_min_profitable_iters): Do not assume that peeling
26531 for the number of iterations will be needed for fully-masked loops.
26532 (vectorizable_reduction): Disable fully-masked loops.
26533 (vectorizable_live_operation): Likewise.
26534 (vect_halve_mask_nunits): New function.
26535 (vect_double_mask_nunits): Likewise.
26536 (vect_record_loop_mask): Likewise.
26537 (vect_get_loop_mask): Likewise.
26538 (vect_transform_loop): Handle the case in which the final loop
26539 iteration might handle a partial vector. Call vect_set_loop_condition
26540 instead of slpeel_make_loop_iterate_ntimes.
26541 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
26542 (check_load_store_masking): New function.
26543 (prepare_load_store_mask): Likewise.
26544 (vectorizable_store): Handle fully-masked loops.
26545 (vectorizable_load): Likewise.
26546 (supportable_widening_operation): Use vect_halve_mask_nunits for
26548 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
26549 (vect_gen_while): New function.
26550 * config/aarch64/aarch64.md (umax<mode>3): New expander.
26551 (aarch64_uqdec<mode>): New insn.
26553 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26554 Alan Hayward <alan.hayward@arm.com>
26555 David Sherwood <david.sherwood@arm.com>
26557 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
26558 (reduc_xor_scal_optab): New optabs.
26559 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
26560 (reduc_xor_scal_@var{m}): Document.
26561 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
26562 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
26563 internal functions.
26564 * fold-const-call.c (fold_const_call): Handle them.
26565 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
26566 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
26567 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
26568 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
26569 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
26570 (UNSPEC_XORV): New unspecs.
26571 (optab): Add entries for them.
26572 (BITWISEV): New int iterator.
26573 (bit_reduc_op): New int attributes.
26575 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26576 Alan Hayward <alan.hayward@arm.com>
26577 David Sherwood <david.sherwood@arm.com>
26579 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
26580 * internal-fn.def (VEC_SHL_INSERT): New internal function.
26581 * optabs.def (vec_shl_insert_optab): New optab.
26582 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
26583 (duplicate_and_interleave): Likewise.
26584 * tree-vect-loop.c: Include internal-fn.h.
26585 (neutral_op_for_slp_reduction): New function, split out from
26586 get_initial_defs_for_reduction.
26587 (get_initial_def_for_reduction): Handle option 2 for variable-length
26588 vectors by loading the neutral value into a vector and then shifting
26589 the initial value into element 0.
26590 (get_initial_defs_for_reduction): Replace the code argument with
26591 the neutral value calculated by neutral_op_for_slp_reduction.
26592 Use gimple_build_vector for constant-length vectors.
26593 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
26594 but the first group_size elements have a neutral value.
26595 Use duplicate_and_interleave otherwise.
26596 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
26597 Update call to get_initial_defs_for_reduction. Handle SLP
26598 reductions for variable-length vectors by creating one vector
26599 result for each scalar result, with the elements associated
26600 with other scalar results stubbed out with the neutral value.
26601 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
26602 Require IFN_VEC_SHL_INSERT for double reductions on
26603 variable-length vectors, or SLP reductions that have
26604 a neutral value. Require can_duplicate_and_interleave_p
26605 support for variable-length unchained SLP reductions if there
26606 is no neutral value, such as for MIN/MAX reductions. Also require
26607 the number of vector elements to be a multiple of the number of
26608 SLP statements when doing variable-length unchained SLP reductions.
26609 Update call to vect_create_epilog_for_reduction.
26610 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
26611 and remove initial values.
26612 (duplicate_and_interleave): Make public.
26613 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
26614 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
26616 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26617 Alan Hayward <alan.hayward@arm.com>
26618 David Sherwood <david.sherwood@arm.com>
26620 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
26621 (can_duplicate_and_interleave_p): New function.
26622 (vect_get_and_check_slp_defs): Take the vector of statements
26623 rather than just the current one. Remove excess parentheses.
26624 Restriction rejectinon of vect_constant_def and vect_external_def
26625 for variable-length vectors to boolean types, or types for which
26626 can_duplicate_and_interleave_p is false.
26627 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
26628 (duplicate_and_interleave): New function.
26629 (vect_get_constant_vectors): Use gimple_build_vector for
26630 constant-length vectors and suitable variable-length constant
26631 vectors. Use duplicate_and_interleave for other variable-length
26632 vectors. Don't defer the update when inserting new statements.
26634 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26635 Alan Hayward <alan.hayward@arm.com>
26636 David Sherwood <david.sherwood@arm.com>
26638 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
26639 min_profitable_iters doesn't go negative.
26641 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26642 Alan Hayward <alan.hayward@arm.com>
26643 David Sherwood <david.sherwood@arm.com>
26645 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
26646 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
26647 * optabs.def (vec_mask_load_lanes_optab): New optab.
26648 (vec_mask_store_lanes_optab): Likewise.
26649 * internal-fn.def (MASK_LOAD_LANES): New internal function.
26650 (MASK_STORE_LANES): Likewise.
26651 * internal-fn.c (mask_load_lanes_direct): New macro.
26652 (mask_store_lanes_direct): Likewise.
26653 (expand_mask_load_optab_fn): Handle masked operations.
26654 (expand_mask_load_lanes_optab_fn): New macro.
26655 (expand_mask_store_optab_fn): Handle masked operations.
26656 (expand_mask_store_lanes_optab_fn): New macro.
26657 (direct_mask_load_lanes_optab_supported_p): Likewise.
26658 (direct_mask_store_lanes_optab_supported_p): Likewise.
26659 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
26661 (vect_load_lanes_supported): Likewise.
26662 * tree-vect-data-refs.c (strip_conversion): New function.
26663 (can_group_stmts_p): Likewise.
26664 (vect_analyze_data_ref_accesses): Use it instead of checking
26665 for a pair of assignments.
26666 (vect_store_lanes_supported): Take a masked_p parameter.
26667 (vect_load_lanes_supported): Likewise.
26668 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
26669 vect_store_lanes_supported and vect_load_lanes_supported.
26670 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
26671 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
26672 parameter. Don't allow gaps for masked accesses.
26673 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
26674 and vect_load_lanes_supported.
26675 (get_load_store_type): Take a masked_p parameter and update
26676 call to get_group_load_store_type.
26677 (vectorizable_store): Update call to get_load_store_type.
26678 Handle IFN_MASK_STORE_LANES.
26679 (vectorizable_load): Update call to get_load_store_type.
26680 Handle IFN_MASK_LOAD_LANES.
26682 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26683 Alan Hayward <alan.hayward@arm.com>
26684 David Sherwood <david.sherwood@arm.com>
26686 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
26688 * config/aarch64/aarch64-protos.h
26689 (aarch64_sve_struct_memory_operand_p): Declare.
26690 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
26691 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
26692 (VPRED, vpred): Handle SVE structure modes.
26693 * config/aarch64/constraints.md (Utx): New constraint.
26694 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
26695 (aarch64_sve_struct_nonimmediate_operand): New predicates.
26696 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
26697 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
26698 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
26699 structure modes. Split into pieces after RA.
26700 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
26701 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
26703 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
26704 SVE structure modes.
26705 (aarch64_classify_address): Likewise.
26706 (sizetochar): Move earlier in file.
26707 (aarch64_print_operand): Handle SVE register lists.
26708 (aarch64_array_mode): New function.
26709 (aarch64_sve_struct_memory_operand_p): Likewise.
26710 (TARGET_ARRAY_MODE): Redefine.
26712 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26713 Alan Hayward <alan.hayward@arm.com>
26714 David Sherwood <david.sherwood@arm.com>
26716 * target.def (array_mode): New target hook.
26717 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
26718 * doc/tm.texi: Regenerate.
26719 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
26720 * hooks.c (hook_optmode_mode_uhwi_none): New function.
26721 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
26722 targetm.array_mode.
26723 * stor-layout.c (mode_for_array): Likewise. Support polynomial
26726 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26727 Alan Hayward <alan.hayward@arm.com>
26728 David Sherwood <david.sherwood@arm.com>
26730 * fold-const.c (fold_binary_loc): Check the argument types
26731 rather than the result type when testing for a vector operation.
26733 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26735 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
26736 * doc/tm.texi: Regenerate.
26738 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26739 Alan Hayward <alan.hayward@arm.com>
26740 David Sherwood <david.sherwood@arm.com>
26742 * doc/invoke.texi (-msve-vector-bits=): Document new option.
26743 (sve): Document new AArch64 extension.
26744 * doc/md.texi (w): Extend the description of the AArch64
26745 constraint to include SVE vectors.
26746 (Upl, Upa): Document new AArch64 predicate constraints.
26747 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
26749 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
26750 (msve-vector-bits=): New option.
26751 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
26752 SVE when these are disabled.
26753 (sve): New extension.
26754 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
26755 modes. Adjust their number of units based on aarch64_sve_vg.
26756 (MAX_BITSIZE_MODE_ANY_MODE): Define.
26757 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
26758 aarch64_addr_query_type.
26759 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
26760 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
26761 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
26762 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
26763 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
26764 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
26765 (aarch64_simd_imm_zero_p): Delete.
26766 (aarch64_check_zero_based_sve_index_immediate): Declare.
26767 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26768 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26769 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26770 (aarch64_sve_float_mul_immediate_p): Likewise.
26771 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26772 rather than an rtx.
26773 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
26774 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
26775 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
26776 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
26777 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
26778 (aarch64_regmode_natural_size): Likewise.
26779 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
26780 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
26782 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
26783 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
26784 for VG and the SVE predicate registers.
26785 (V_ALIASES): Add a "z"-prefixed alias.
26786 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
26787 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
26788 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
26789 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
26790 (REG_CLASS_NAMES): Add entries for them.
26791 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
26792 and the predicate registers.
26793 (aarch64_sve_vg): Declare.
26794 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
26795 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
26796 (REGMODE_NATURAL_SIZE): Define.
26797 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
26799 * config/aarch64/aarch64.c: Include cfgrtl.h.
26800 (simd_immediate_info): Add a constructor for series vectors,
26801 and an associated step field.
26802 (aarch64_sve_vg): New variable.
26803 (aarch64_dbx_register_number): Handle VG and the predicate registers.
26804 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
26805 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
26806 (VEC_ANY_DATA, VEC_STRUCT): New constants.
26807 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
26808 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
26809 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
26810 (aarch64_get_mask_mode): New functions.
26811 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
26812 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
26813 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
26814 predicate modes and predicate registers. Explicitly restrict
26815 GPRs to modes of 16 bytes or smaller. Only allow FP registers
26816 to store a vector mode if it is recognized by
26817 aarch64_classify_vector_mode.
26818 (aarch64_regmode_natural_size): New function.
26819 (aarch64_hard_regno_caller_save_mode): Return the original mode
26821 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
26822 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
26823 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
26824 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
26826 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
26827 does not overlap dest if the function is frame-related. Handle
26829 (aarch64_split_add_offset): New function.
26830 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
26831 them aarch64_add_offset.
26832 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
26833 and update call to aarch64_sub_sp.
26834 (aarch64_add_cfa_expression): New function.
26835 (aarch64_expand_prologue): Pass extra temporary registers to the
26836 functions above. Handle the case in which we need to emit new
26837 DW_CFA_expressions for registers that were originally saved
26838 relative to the stack pointer, but now have to be expressed
26839 relative to the frame pointer.
26840 (aarch64_output_mi_thunk): Pass extra temporary registers to the
26842 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
26843 IP0 and IP1 values for SVE frames.
26844 (aarch64_expand_vec_series): New function.
26845 (aarch64_expand_sve_widened_duplicate): Likewise.
26846 (aarch64_expand_sve_const_vector): Likewise.
26847 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
26848 Handle SVE constants. Use emit_move_insn to move a force_const_mem
26849 into the register, rather than emitting a SET directly.
26850 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
26851 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
26852 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
26853 (offset_9bit_signed_scaled_p): New functions.
26854 (aarch64_replicate_bitmask_imm): New function.
26855 (aarch64_bitmask_imm): Use it.
26856 (aarch64_cannot_force_const_mem): Reject expressions involving
26857 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
26858 (aarch64_classify_index): Handle SVE indices, by requiring
26859 a plain register index with a scale that matches the element size.
26860 (aarch64_classify_address): Handle SVE addresses. Assert that
26861 the mode of the address is VOIDmode or an integer mode.
26862 Update call to aarch64_classify_symbol.
26863 (aarch64_classify_symbolic_expression): Update call to
26864 aarch64_classify_symbol.
26865 (aarch64_const_vec_all_in_range_p): New function.
26866 (aarch64_print_vector_float_operand): Likewise.
26867 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
26868 "vN" for FP registers with SVE modes. Handle (const ...) vectors
26869 and the FP immediates 1.0 and 0.5.
26870 (aarch64_print_address_internal): Handle SVE addresses.
26871 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
26872 (aarch64_regno_regclass): Handle predicate registers.
26873 (aarch64_secondary_reload): Handle big-endian reloads of SVE
26875 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
26876 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
26877 (aarch64_convert_sve_vector_bits): New function.
26878 (aarch64_override_options): Use it to handle -msve-vector-bits=.
26879 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26880 rather than an rtx.
26881 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
26882 Handle SVE vector and predicate modes. Accept VL-based constants
26883 that need only one temporary register, and VL offsets that require
26884 no temporary registers.
26885 (aarch64_conditional_register_usage): Mark the predicate registers
26886 as fixed if SVE isn't available.
26887 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
26888 Return true for SVE vector and predicate modes.
26889 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
26890 rather than an unsigned int. Handle SVE modes.
26891 (aarch64_preferred_simd_mode): Update call accordingly. Handle
26893 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
26895 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26896 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26897 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26898 (aarch64_sve_float_mul_immediate_p): New functions.
26899 (aarch64_sve_valid_immediate): New function.
26900 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
26901 Explicitly reject structure modes. Check for INDEX constants.
26902 Handle PTRUE and PFALSE constants.
26903 (aarch64_check_zero_based_sve_index_immediate): New function.
26904 (aarch64_simd_imm_zero_p): Delete.
26905 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
26906 vector modes. Accept constants in the range of CNT[BHWD].
26907 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
26908 ask for an Advanced SIMD mode.
26909 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
26910 (aarch64_simd_vector_alignment): Handle SVE predicates.
26911 (aarch64_vectorize_preferred_vector_alignment): New function.
26912 (aarch64_simd_vector_alignment_reachable): Use it instead of
26914 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
26915 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
26917 (MAX_VECT_LEN): Delete.
26918 (expand_vec_perm_d): Add a vec_flags field.
26919 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
26920 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
26921 (aarch64_evpc_ext): Don't apply a big-endian lane correction
26923 (aarch64_evpc_rev): Rename to...
26924 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
26925 (aarch64_evpc_rev_global): New function.
26926 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
26927 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
26929 (aarch64_evpc_sve_tbl): New function.
26930 (aarch64_expand_vec_perm_const_1): Update after rename of
26931 aarch64_evpc_rev. Handle SVE permutes too, trying
26932 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
26933 than aarch64_evpc_tbl.
26934 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
26935 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
26936 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
26937 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
26938 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
26939 (aarch64_expand_sve_vcond): New functions.
26940 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
26941 of aarch64_vector_mode_p.
26942 (aarch64_dwarf_poly_indeterminate_value): New function.
26943 (aarch64_compute_pressure_classes): Likewise.
26944 (aarch64_can_change_mode_class): Likewise.
26945 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
26946 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
26947 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
26948 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
26949 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
26950 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
26951 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
26952 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
26954 (Dn, Dl, Dr): Accept const as well as const_vector.
26955 (Dz): Likewise. Compare against CONST0_RTX.
26956 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
26957 of "vector" where appropriate.
26958 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
26959 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
26960 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
26961 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
26962 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
26963 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
26964 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
26965 (v_int_equiv): Extend to SVE modes.
26966 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
26968 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
26969 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
26970 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
26971 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
26972 (SVE_COND_FP_CMP): New int iterators.
26973 (perm_hilo): Handle the new unpack unspecs.
26974 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
26976 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
26977 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
26978 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
26979 (aarch64_equality_operator, aarch64_constant_vector_operand)
26980 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
26981 (aarch64_sve_nonimmediate_operand): Likewise.
26982 (aarch64_sve_general_operand): Likewise.
26983 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
26984 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
26985 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
26986 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
26987 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
26988 (aarch64_sve_float_arith_immediate): Likewise.
26989 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
26990 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
26991 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
26992 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
26993 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
26994 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
26995 (aarch64_sve_float_arith_operand): Likewise.
26996 (aarch64_sve_float_arith_with_sub_operand): Likewise.
26997 (aarch64_sve_float_mul_operand): Likewise.
26998 (aarch64_sve_vec_perm_operand): Likewise.
26999 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
27000 (aarch64_mov_operand): Accept const_poly_int and const_vector.
27001 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
27002 as well as const_vector.
27003 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
27004 in file. Use CONST0_RTX and CONSTM1_RTX.
27005 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
27006 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
27007 Use aarch64_simd_imm_zero.
27008 * config/aarch64/aarch64-sve.md: New file.
27009 * config/aarch64/aarch64.md: Include it.
27010 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
27011 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
27012 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
27013 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
27014 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
27015 (sve): New attribute.
27016 (enabled): Disable instructions with the sve attribute unless
27018 (movqi, movhi): Pass CONST_POLY_INT operaneds through
27019 aarch64_expand_mov_immediate.
27020 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
27021 CNT[BHSD] immediates.
27022 (movti): Split CONST_POLY_INT moves into two halves.
27023 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
27024 Split additions that need a temporary here if the destination
27025 is the stack pointer.
27026 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
27027 (*add<mode>3_poly_1): New instruction.
27028 (set_clobber_cc): New expander.
27030 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27032 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
27033 parameter and use it instead of GET_MODE_SIZE (innermode). Use
27034 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
27035 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
27036 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
27037 Change innermode from fixed_mode_size to machine_mode.
27038 (simplify_subreg): Update call accordingly. Handle a constant-sized
27039 subreg of a variable-length CONST_VECTOR.
27041 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27042 Alan Hayward <alan.hayward@arm.com>
27043 David Sherwood <david.sherwood@arm.com>
27045 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
27046 (add_offset_to_base): New function, split out from...
27047 (create_mem_ref): ...here. When handling a scale other than 1,
27048 check first whether the address is valid without the offset.
27049 Add it into the base if so, leaving the index and scale as-is.
27051 2018-01-12 Jakub Jelinek <jakub@redhat.com>
27054 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
27055 fold_for_warn before checking if arg2 is INTEGER_CST.
27057 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
27059 * config/rs6000/predicates.md (load_multiple_operation): Delete.
27060 (store_multiple_operation): Delete.
27061 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
27062 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
27063 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
27064 guarded by TARGET_STRING.
27065 (rs6000_output_load_multiple): Delete.
27066 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27067 OPTION_MASK_STRING / TARGET_STRING handling.
27068 (print_operand) <'N', 'O'>: Add comment that these are unused now.
27069 (const rs6000_opt_masks) <"string">: Change mask to 0.
27070 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
27071 (MASK_STRING): Delete.
27072 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
27074 (load_multiple): Delete.
27081 (store_multiple): Delete.
27088 (movmemsi_8reg): Delete.
27089 (corresponding unnamed define_insn): Delete.
27090 (movmemsi_6reg): Delete.
27091 (corresponding unnamed define_insn): Delete.
27092 (movmemsi_4reg): Delete.
27093 (corresponding unnamed define_insn): Delete.
27094 (movmemsi_2reg): Delete.
27095 (corresponding unnamed define_insn): Delete.
27096 (movmemsi_1reg): Delete.
27097 (corresponding unnamed define_insn): Delete.
27098 * config/rs6000/rs6000.opt (mno-string): New.
27099 (mstring): Replace by deprecation warning stub.
27100 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
27102 2018-01-12 Jakub Jelinek <jakub@redhat.com>
27104 * regrename.c (regrename_do_replace): If replacing the same
27105 reg multiple times, try to reuse last created gen_raw_REG.
27108 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
27109 main to workaround a bug in GDB.
27111 2018-01-12 Tom de Vries <tom@codesourcery.com>
27114 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
27116 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
27118 PR rtl-optimization/80481
27119 * ira-color.c (get_cap_member): New function.
27120 (allocnos_conflict_by_live_ranges_p): Use it.
27121 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
27122 (setup_slot_coalesced_allocno_live_ranges): Ditto.
27124 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
27127 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
27128 (*saddl_se_1): Ditto.
27129 (*ssubsi_1): Ditto.
27130 (*ssubl_se_1): Ditto.
27132 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27134 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
27135 rather than wi::to_widest for DR_INITs.
27136 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
27137 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
27138 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
27140 (vect_analyze_group_access_1): Note that here.
27142 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27144 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
27145 polynomial type sizes.
27147 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27149 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
27150 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
27151 (gimple_add_tmp_var): Likewise.
27153 2018-01-12 Martin Liska <mliska@suse.cz>
27155 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
27156 (gimple_alloc_sizes): Likewise.
27157 (dump_gimple_statistics): Use PRIu64 in printf format.
27158 * gimple.h: Change uint64_t to int.
27160 2018-01-12 Martin Liska <mliska@suse.cz>
27162 * tree-core.h: Use uint64_t instead of int.
27163 * tree.c (tree_node_counts): Likewise.
27164 (tree_node_sizes): Likewise.
27165 (dump_tree_statistics): Use PRIu64 in printf format.
27167 2018-01-12 Martin Liska <mliska@suse.cz>
27169 * Makefile.in: As qsort_chk is implemented in vec.c, add
27170 vec.o to linkage of gencfn-macros.
27171 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
27172 passing the info to record_node_allocation_statistics.
27173 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
27175 * ggc-common.c (struct ggc_usage): Add operator== and use
27176 it in operator< and compare function.
27177 * mem-stats.h (struct mem_usage): Likewise.
27178 * vec.c (struct vec_usage): Remove operator< and compare
27179 function. Can be simply inherited.
27181 2018-01-12 Martin Jambor <mjambor@suse.cz>
27184 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
27185 * tree-ssa-math-opts.c: Include domwalk.h.
27186 (convert_mult_to_fma_1): New function.
27187 (fma_transformation_info): New type.
27188 (fma_deferring_state): Likewise.
27189 (cancel_fma_deferring): New function.
27190 (result_of_phi): Likewise.
27191 (last_fma_candidate_feeds_initial_phi): Likewise.
27192 (convert_mult_to_fma): Added deferring logic, split actual
27193 transformation to convert_mult_to_fma_1.
27194 (math_opts_dom_walker): New type.
27195 (math_opts_dom_walker::after_dom_children): New method, body moved
27196 here from pass_optimize_widening_mul::execute, added deferring logic
27198 (pass_optimize_widening_mul::execute): Moved most of code to
27199 math_opts_dom_walker::after_dom_children.
27200 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
27201 * config/i386/i386.c (ix86_option_override_internal): Added
27202 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
27204 2018-01-12 Richard Biener <rguenther@suse.de>
27207 * dwarf2out.c (gen_variable_die): Do not reset old_die for
27208 inline instance vars.
27210 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
27213 * config/rx/rx.c (rx_is_restricted_memory_address):
27214 Handle SUBREG case.
27216 2018-01-12 Richard Biener <rguenther@suse.de>
27218 PR tree-optimization/80846
27219 * target.def (split_reduction): New target hook.
27220 * targhooks.c (default_split_reduction): New function.
27221 * targhooks.h (default_split_reduction): Declare.
27222 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
27223 target requests first reduce vectors by combining low and high
27225 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
27226 (get_vectype_for_scalar_type_and_size): Export.
27227 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
27228 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
27229 * doc/tm.texi: Regenerate.
27230 * config/i386/i386.c (ix86_split_reduction): Implement
27231 TARGET_VECTORIZE_SPLIT_REDUCTION.
27233 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
27236 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
27237 in PIC mode except for TARGET_VXWORKS_RTP.
27238 * config/sparc/sparc.c: Include cfgrtl.h.
27239 (TARGET_INIT_PIC_REG): Define.
27240 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
27241 (sparc_pic_register_p): New predicate.
27242 (sparc_legitimate_address_p): Use it.
27243 (sparc_legitimize_pic_address): Likewise.
27244 (sparc_delegitimize_address): Likewise.
27245 (sparc_mode_dependent_address_p): Likewise.
27246 (gen_load_pcrel_sym): Remove 4th parameter.
27247 (load_got_register): Adjust call to above. Remove obsolete stuff.
27248 (sparc_expand_prologue): Do not call load_got_register here.
27249 (sparc_flat_expand_prologue): Likewise.
27250 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
27251 (sparc_use_pseudo_pic_reg): New function.
27252 (sparc_init_pic_reg): Likewise.
27253 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
27254 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
27256 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
27258 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27259 Add item for branch_cost.
27261 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
27263 PR rtl-optimization/83565
27264 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
27265 not extend the result to a larger mode for rotate operations.
27266 (num_sign_bit_copies1): Likewise.
27268 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27271 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
27273 Use values-Xc.o for -pedantic.
27274 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
27276 2018-01-12 Martin Liska <mliska@suse.cz>
27279 * ipa-devirt.c (final_warning_record::grow_type_warnings):
27281 (possible_polymorphic_call_targets): Use it.
27282 (ipa_devirt): Likewise.
27284 2018-01-12 Martin Liska <mliska@suse.cz>
27286 * profile-count.h (enum profile_quality): Use 0 as invalid
27287 enum value of profile_quality.
27289 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
27291 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
27292 -mext-string options.
27294 2018-01-12 Richard Biener <rguenther@suse.de>
27296 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
27297 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
27298 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
27300 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
27302 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27304 * configure.ac (--with-long-double-format): Add support for the
27305 configuration option to change the default long double format on
27307 * config.gcc (powerpc*-linux*-*): Likewise.
27308 * configure: Regenerate.
27309 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
27310 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
27311 used without modification.
27313 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27315 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
27316 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
27317 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
27318 MISC_BUILTIN_SPEC_BARRIER.
27319 (rs6000_init_builtins): Likewise.
27320 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
27322 (speculation_barrier): New define_insn.
27323 * doc/extend.texi: Document __builtin_speculation_barrier.
27325 2018-01-11 Jakub Jelinek <jakub@redhat.com>
27328 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
27329 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
27330 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
27332 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
27333 integral modes instead of "ss" and "sd".
27334 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
27335 vectors with 32-bit and 64-bit elements.
27336 (vecdupssescalarmodesuffix): New mode attribute.
27337 (vec_dup<mode>): Use it.
27339 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
27342 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
27343 frame if argument is passed on stack.
27345 2018-01-11 Jakub Jelinek <jakub@redhat.com>
27348 * ree.c (combine_reaching_defs): Optimize also
27349 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
27350 reg2=any_extend(exp); reg1=reg2;, formatting fix.
27352 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
27354 PR middle-end/83189
27355 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
27357 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
27359 PR middle-end/83718
27360 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
27361 after they are computed.
27363 2018-01-11 Bin Cheng <bin.cheng@arm.com>
27365 PR tree-optimization/83695
27366 * gimple-loop-linterchange.cc
27367 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
27368 reset cached scev information after interchange.
27369 (pass_linterchange::execute): Remove call to scev_reset_htab.
27371 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27373 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
27374 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
27375 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
27376 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
27377 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
27378 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
27379 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
27380 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
27381 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
27382 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
27383 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
27384 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
27385 (V_lane_reg): Likewise.
27386 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
27388 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
27389 (vfmal_lane_low<mode>_intrinsic,
27390 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
27391 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
27392 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
27393 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
27394 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
27395 vfmsl_lane_high<mode>_intrinsic): New define_insns.
27397 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27399 * config/arm/arm-cpus.in (fp16fml): New feature.
27400 (ALL_SIMD): Add fp16fml.
27401 (armv8.2-a): Add fp16fml as an option.
27402 (armv8.3-a): Likewise.
27403 (armv8.4-a): Add fp16fml as part of fp16.
27404 * config/arm/arm.h (TARGET_FP16FML): Define.
27405 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
27407 * config/arm/arm-modes.def (V2HF): Define.
27408 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
27409 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
27410 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
27411 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
27412 vfmsl_low, vfmsl_high): New set of builtins.
27413 * config/arm/iterators.md (PLUSMINUS): New code iterator.
27414 (vfml_op): New code attribute.
27415 (VFMLHALVES): New int iterator.
27416 (VFML, VFMLSEL): New mode attributes.
27417 (V_reg): Define mapping for V2HF.
27418 (V_hi, V_lo): New mode attributes.
27419 (VF_constraint): Likewise.
27420 (vfml_half, vfml_half_selector): New int attributes.
27421 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
27423 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
27424 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
27426 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
27427 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
27428 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
27429 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
27431 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
27432 Document new effective target and option set.
27434 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27436 * config/arm/arm-cpus.in (armv8_4): New feature.
27437 (ARMv8_4a): New fgroup.
27438 (armv8.4-a): New arch.
27439 * config/arm/arm-tables.opt: Regenerate.
27440 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
27441 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
27442 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
27443 Add matching rules for -march=armv8.4-a and extensions.
27444 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
27446 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
27449 * config/rx/rx.md (BW): New mode attribute.
27450 (sync_lock_test_and_setsi): Add mode suffix to insn output.
27452 2018-01-11 Richard Biener <rguenther@suse.de>
27454 PR tree-optimization/83435
27455 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
27456 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
27457 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
27459 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27460 Alan Hayward <alan.hayward@arm.com>
27461 David Sherwood <david.sherwood@arm.com>
27463 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
27465 (aarch64_classify_address): Initialize it. Track polynomial offsets.
27466 (aarch64_print_address_internal): Use it to check for a zero offset.
27468 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27469 Alan Hayward <alan.hayward@arm.com>
27470 David Sherwood <david.sherwood@arm.com>
27472 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
27473 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
27474 Return a poly_int64 rather than a HOST_WIDE_INT.
27475 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
27476 rather than a HOST_WIDE_INT.
27477 * config/aarch64/aarch64.h (aarch64_frame): Protect with
27478 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
27479 hard_fp_offset, frame_size, initial_adjust, callee_offset and
27480 final_offset from HOST_WIDE_INT to poly_int64.
27481 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
27482 to_constant when getting the number of units in an Advanced SIMD
27484 (aarch64_builtin_vectorized_function): Check for a constant number
27486 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
27488 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
27489 attribute instead of GET_MODE_NUNITS.
27490 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
27491 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
27492 GET_MODE_SIZE for fixed-size registers.
27493 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
27494 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
27495 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
27496 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
27497 (aarch64_print_operand, aarch64_print_address_internal)
27498 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
27499 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
27500 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
27501 Handle polynomial GET_MODE_SIZE.
27502 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
27503 wider than SImode without modification.
27504 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
27505 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
27506 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
27507 passing and returning SVE modes.
27508 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
27509 rather than GEN_INT.
27510 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
27511 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
27512 (aarch64_allocate_and_probe_stack_space): Likewise.
27513 (aarch64_layout_frame): Cope with polynomial offsets.
27514 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
27515 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
27516 polynomial offsets.
27517 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
27518 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
27519 poly_int64 rather than a HOST_WIDE_INT.
27520 (aarch64_get_separate_components, aarch64_process_components)
27521 (aarch64_expand_prologue, aarch64_expand_epilogue)
27522 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
27523 (aarch64_anchor_offset): New function, split out from...
27524 (aarch64_legitimize_address): ...here.
27525 (aarch64_builtin_vectorization_cost): Handle polynomial
27526 TYPE_VECTOR_SUBPARTS.
27527 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
27529 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
27530 number of elements from the PARALLEL rather than the mode.
27531 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
27532 rather than GET_MODE_BITSIZE.
27533 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
27534 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
27535 (aarch64_expand_vec_perm_const_1): Handle polynomial
27536 d->perm.length () and d->perm elements.
27537 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
27538 Apply to_constant to d->perm elements.
27539 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
27540 polynomial CONST_VECTOR_NUNITS.
27541 (aarch64_move_pointer): Take amount as a poly_int64 rather
27543 (aarch64_progress_pointer): Avoid temporary variable.
27544 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
27545 the mode attribute instead of GET_MODE.
27547 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27548 Alan Hayward <alan.hayward@arm.com>
27549 David Sherwood <david.sherwood@arm.com>
27551 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
27552 x exists before using it.
27553 (aarch64_add_constant_internal): Rename to...
27554 (aarch64_add_offset_1): ...this. Replace regnum with separate
27555 src and dest rtxes. Handle the case in which they're different,
27556 including when the offset is zero. Replace scratchreg with an rtx.
27557 Use 2 additions if there is no spare register into which we can
27558 move a 16-bit constant.
27559 (aarch64_add_constant): Delete.
27560 (aarch64_add_offset): Replace reg with separate src and dest
27561 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
27562 Use aarch64_add_offset_1.
27563 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
27564 an rtx rather than an int. Take the delta as a poly_int64
27565 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
27566 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
27567 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
27568 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
27569 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
27570 and aarch64_add_sp.
27571 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
27572 aarch64_add_constant.
27574 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27576 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
27577 Use scalar_float_mode.
27579 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27581 * config/aarch64/aarch64-simd.md
27582 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
27583 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
27584 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
27585 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
27586 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
27587 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
27588 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
27589 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
27590 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
27591 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
27593 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27596 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
27597 targ_options->x_arm_arch_string is non NULL.
27599 2018-01-11 Tamar Christina <tamar.christina@arm.com>
27601 * config/aarch64/aarch64.h
27602 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
27604 2018-01-11 Sudakshina Das <sudi.das@arm.com>
27607 * expmed.c (emit_store_flag_force): Swap if const op0
27608 and change VOIDmode to mode of op0.
27610 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27612 PR rtl-optimization/83761
27613 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
27614 than bytes to mode_for_size.
27616 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27618 PR middle-end/83189
27619 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
27620 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
27623 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27625 PR middle-end/83575
27626 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
27627 when in layout mode.
27628 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
27629 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
27632 2018-01-10 Michael Collison <michael.collison@arm.com>
27634 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
27635 * config/aarch64/aarch64-option-extension.def: Add
27636 AARCH64_OPT_EXTENSION of 'fp16fml'.
27637 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27638 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
27639 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
27640 * config/aarch64/constraints.md (Ui7): New constraint.
27641 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
27642 (VFMLA_SEL_W): Ditto.
27645 (VFMLA16_LOW): New int iterator.
27646 (VFMLA16_HIGH): Ditto.
27647 (UNSPEC_FMLAL): New unspec.
27648 (UNSPEC_FMLSL): Ditto.
27649 (UNSPEC_FMLAL2): Ditto.
27650 (UNSPEC_FMLSL2): Ditto.
27651 (f16mac): New code attribute.
27652 * config/aarch64/aarch64-simd-builtins.def
27653 (aarch64_fmlal_lowv2sf): Ditto.
27654 (aarch64_fmlsl_lowv2sf): Ditto.
27655 (aarch64_fmlalq_lowv4sf): Ditto.
27656 (aarch64_fmlslq_lowv4sf): Ditto.
27657 (aarch64_fmlal_highv2sf): Ditto.
27658 (aarch64_fmlsl_highv2sf): Ditto.
27659 (aarch64_fmlalq_highv4sf): Ditto.
27660 (aarch64_fmlslq_highv4sf): Ditto.
27661 (aarch64_fmlal_lane_lowv2sf): Ditto.
27662 (aarch64_fmlsl_lane_lowv2sf): Ditto.
27663 (aarch64_fmlal_laneq_lowv2sf): Ditto.
27664 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
27665 (aarch64_fmlalq_lane_lowv4sf): Ditto.
27666 (aarch64_fmlsl_lane_lowv4sf): Ditto.
27667 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
27668 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
27669 (aarch64_fmlal_lane_highv2sf): Ditto.
27670 (aarch64_fmlsl_lane_highv2sf): Ditto.
27671 (aarch64_fmlal_laneq_highv2sf): Ditto.
27672 (aarch64_fmlsl_laneq_highv2sf): Ditto.
27673 (aarch64_fmlalq_lane_highv4sf): Ditto.
27674 (aarch64_fmlsl_lane_highv4sf): Ditto.
27675 (aarch64_fmlalq_laneq_highv4sf): Ditto.
27676 (aarch64_fmlsl_laneq_highv4sf): Ditto.
27677 * config/aarch64/aarch64-simd.md:
27678 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
27679 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27680 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
27681 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27682 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
27683 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
27684 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
27685 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
27686 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
27687 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
27688 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
27689 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
27690 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
27691 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
27692 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
27693 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
27694 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
27695 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
27696 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
27697 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
27698 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
27699 (vfmlsl_low_u32): Ditto.
27700 (vfmlalq_low_u32): Ditto.
27701 (vfmlslq_low_u32): Ditto.
27702 (vfmlal_high_u32): Ditto.
27703 (vfmlsl_high_u32): Ditto.
27704 (vfmlalq_high_u32): Ditto.
27705 (vfmlslq_high_u32): Ditto.
27706 (vfmlal_lane_low_u32): Ditto.
27707 (vfmlsl_lane_low_u32): Ditto.
27708 (vfmlal_laneq_low_u32): Ditto.
27709 (vfmlsl_laneq_low_u32): Ditto.
27710 (vfmlalq_lane_low_u32): Ditto.
27711 (vfmlslq_lane_low_u32): Ditto.
27712 (vfmlalq_laneq_low_u32): Ditto.
27713 (vfmlslq_laneq_low_u32): Ditto.
27714 (vfmlal_lane_high_u32): Ditto.
27715 (vfmlsl_lane_high_u32): Ditto.
27716 (vfmlal_laneq_high_u32): Ditto.
27717 (vfmlsl_laneq_high_u32): Ditto.
27718 (vfmlalq_lane_high_u32): Ditto.
27719 (vfmlslq_lane_high_u32): Ditto.
27720 (vfmlalq_laneq_high_u32): Ditto.
27721 (vfmlslq_laneq_high_u32): Ditto.
27722 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
27723 (AARCH64_FL_FOR_ARCH8_4): New.
27724 (AARCH64_ISA_F16FML): New ISA flag.
27725 (TARGET_F16FML): New feature flag for fp16fml.
27726 (doc/invoke.texi): Document new fp16fml option.
27728 2018-01-10 Michael Collison <michael.collison@arm.com>
27730 * config/aarch64/aarch64-builtins.c:
27731 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
27732 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27733 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
27734 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
27735 (AARCH64_ISA_SHA3): New ISA flag.
27736 (TARGET_SHA3): New feature flag for sha3.
27737 * config/aarch64/iterators.md (sha512_op): New int attribute.
27738 (CRYPTO_SHA512): New int iterator.
27739 (UNSPEC_SHA512H): New unspec.
27740 (UNSPEC_SHA512H2): Ditto.
27741 (UNSPEC_SHA512SU0): Ditto.
27742 (UNSPEC_SHA512SU1): Ditto.
27743 * config/aarch64/aarch64-simd-builtins.def
27744 (aarch64_crypto_sha512hqv2di): New builtin.
27745 (aarch64_crypto_sha512h2qv2di): Ditto.
27746 (aarch64_crypto_sha512su0qv2di): Ditto.
27747 (aarch64_crypto_sha512su1qv2di): Ditto.
27748 (aarch64_eor3qv8hi): Ditto.
27749 (aarch64_rax1qv2di): Ditto.
27750 (aarch64_xarqv2di): Ditto.
27751 (aarch64_bcaxqv8hi): Ditto.
27752 * config/aarch64/aarch64-simd.md:
27753 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
27754 (aarch64_crypto_sha512su0qv2di): Ditto.
27755 (aarch64_crypto_sha512su1qv2di): Ditto.
27756 (aarch64_eor3qv8hi): Ditto.
27757 (aarch64_rax1qv2di): Ditto.
27758 (aarch64_xarqv2di): Ditto.
27759 (aarch64_bcaxqv8hi): Ditto.
27760 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
27761 (vsha512h2q_u64): Ditto.
27762 (vsha512su0q_u64): Ditto.
27763 (vsha512su1q_u64): Ditto.
27764 (veor3q_u16): Ditto.
27765 (vrax1q_u64): Ditto.
27766 (vxarq_u64): Ditto.
27767 (vbcaxq_u16): Ditto.
27768 * config/arm/types.md (crypto_sha512): New type attribute.
27769 (crypto_sha3): Ditto.
27770 (doc/invoke.texi): Document new sha3 option.
27772 2018-01-10 Michael Collison <michael.collison@arm.com>
27774 * config/aarch64/aarch64-builtins.c:
27775 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
27776 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27777 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
27778 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
27779 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
27780 (AARCH64_ISA_SM4): New ISA flag.
27781 (TARGET_SM4): New feature flag for sm4.
27782 * config/aarch64/aarch64-simd-builtins.def
27783 (aarch64_sm3ss1qv4si): Ditto.
27784 (aarch64_sm3tt1aq4si): Ditto.
27785 (aarch64_sm3tt1bq4si): Ditto.
27786 (aarch64_sm3tt2aq4si): Ditto.
27787 (aarch64_sm3tt2bq4si): Ditto.
27788 (aarch64_sm3partw1qv4si): Ditto.
27789 (aarch64_sm3partw2qv4si): Ditto.
27790 (aarch64_sm4eqv4si): Ditto.
27791 (aarch64_sm4ekeyqv4si): Ditto.
27792 * config/aarch64/aarch64-simd.md:
27793 (aarch64_sm3ss1qv4si): Ditto.
27794 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
27795 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
27796 (aarch64_sm4eqv4si): Ditto.
27797 (aarch64_sm4ekeyqv4si): Ditto.
27798 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
27799 (sm3part_op): Ditto.
27800 (CRYPTO_SM3TT): Ditto.
27801 (CRYPTO_SM3PART): Ditto.
27802 (UNSPEC_SM3SS1): New unspec.
27803 (UNSPEC_SM3TT1A): Ditto.
27804 (UNSPEC_SM3TT1B): Ditto.
27805 (UNSPEC_SM3TT2A): Ditto.
27806 (UNSPEC_SM3TT2B): Ditto.
27807 (UNSPEC_SM3PARTW1): Ditto.
27808 (UNSPEC_SM3PARTW2): Ditto.
27809 (UNSPEC_SM4E): Ditto.
27810 (UNSPEC_SM4EKEY): Ditto.
27811 * config/aarch64/constraints.md (Ui2): New constraint.
27812 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
27813 * config/arm/types.md (crypto_sm3): New type attribute.
27814 (crypto_sm4): Ditto.
27815 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
27816 (vsm3tt1aq_u32): Ditto.
27817 (vsm3tt1bq_u32): Ditto.
27818 (vsm3tt2aq_u32): Ditto.
27819 (vsm3tt2bq_u32): Ditto.
27820 (vsm3partw1q_u32): Ditto.
27821 (vsm3partw2q_u32): Ditto.
27822 (vsm4eq_u32): Ditto.
27823 (vsm4ekeyq_u32): Ditto.
27824 (doc/invoke.texi): Document new sm4 option.
27826 2018-01-10 Michael Collison <michael.collison@arm.com>
27828 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
27829 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
27830 (AARCH64_FL_FOR_ARCH8_4): New.
27831 (AARCH64_FL_V8_4): New flag.
27832 (doc/invoke.texi): Document new armv8.4-a option.
27834 2018-01-10 Michael Collison <michael.collison@arm.com>
27836 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27837 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
27838 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
27839 * config/aarch64/aarch64-option-extension.def: Add
27840 AARCH64_OPT_EXTENSION of 'sha2'.
27841 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
27842 (crypto): Disable sha2 and aes if crypto disabled.
27843 (crypto): Enable aes and sha2 if enabled.
27844 (simd): Disable sha2 and aes if simd disabled.
27845 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
27847 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
27848 (TARGET_SHA2): New feature flag for sha2.
27849 (TARGET_AES): New feature flag for aes.
27850 * config/aarch64/aarch64-simd.md:
27851 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
27852 conditional on TARGET_AES.
27853 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
27854 (aarch64_crypto_sha1hsi): Make pattern conditional
27856 (aarch64_crypto_sha1hv4si): Ditto.
27857 (aarch64_be_crypto_sha1hv4si): Ditto.
27858 (aarch64_crypto_sha1su1v4si): Ditto.
27859 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
27860 (aarch64_crypto_sha1su0v4si): Ditto.
27861 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
27862 (aarch64_crypto_sha256su0v4si): Ditto.
27863 (aarch64_crypto_sha256su1v4si): Ditto.
27864 (doc/invoke.texi): Document new aes and sha2 options.
27866 2018-01-10 Martin Sebor <msebor@redhat.com>
27868 PR tree-optimization/83781
27869 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
27872 2018-01-11 Martin Sebor <msebor@gmail.com>
27873 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27875 PR tree-optimization/83501
27876 PR tree-optimization/81703
27878 * tree-ssa-strlen.c (get_string_cst): Rename...
27879 (get_string_len): ...to this. Handle global constants.
27880 (handle_char_store): Adjust.
27882 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
27883 Jim Wilson <jimw@sifive.com>
27885 * config/riscv/riscv-protos.h (riscv_output_return): New.
27886 * config/riscv/riscv.c (struct machine_function): New naked_p field.
27887 (riscv_attribute_table, riscv_output_return),
27888 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
27889 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
27890 (riscv_compute_frame_info): Only compute frame->mask if not a naked
27892 (riscv_expand_prologue): Add early return for naked function.
27893 (riscv_expand_epilogue): Likewise.
27894 (riscv_function_ok_for_sibcall): Return false for naked function.
27895 (riscv_set_current_function): New.
27896 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
27897 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
27898 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
27899 * doc/extend.texi (RISC-V Function Attributes): New.
27901 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
27903 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
27904 check for 128-bit long double before checking TCmode.
27905 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
27906 128-bit long doubles before checking TFmode or TCmode.
27907 (FLOAT128_IBM_P): Likewise.
27909 2018-01-10 Martin Sebor <msebor@redhat.com>
27911 PR tree-optimization/83671
27912 * builtins.c (c_strlen): Unconditionally return zero for the empty
27914 Use -Warray-bounds for warnings.
27915 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
27916 for non-constant array indices with COMPONENT_REF, arrays of
27917 arrays, and pointers to arrays.
27918 (gimple_fold_builtin_strlen): Determine and set length range for
27919 non-constant character arrays.
27921 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
27923 PR middle-end/81897
27924 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
27927 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
27929 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
27931 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27934 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
27935 VECTOR_MEM_ALTIVEC_OR_VSX_P.
27936 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
27937 indexed_or_indirect_operand predicate.
27938 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
27939 (*vsx_le_perm_load_v8hi): Likewise.
27940 (*vsx_le_perm_load_v16qi): Likewise.
27941 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
27942 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
27943 (*vsx_le_perm_store_v8hi): Likewise.
27944 (*vsx_le_perm_store_v16qi): Likewise.
27945 (eight unnamed splitters): Likewise.
27947 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27949 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
27950 * config/rs6000/emmintrin.h: Likewise.
27951 * config/rs6000/mmintrin.h: Likewise.
27952 * config/rs6000/xmmintrin.h: Likewise.
27954 2018-01-10 David Malcolm <dmalcolm@redhat.com>
27957 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
27959 * tree.c (tree_nop_conversion): Return true for location wrapper
27961 (maybe_wrap_with_location): New function.
27962 (selftest::check_strip_nops): New function.
27963 (selftest::test_location_wrappers): New function.
27964 (selftest::tree_c_tests): Call it.
27965 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
27966 (maybe_wrap_with_location): New decl.
27967 (EXPR_LOCATION_WRAPPER_P): New macro.
27968 (location_wrapper_p): New inline function.
27969 (tree_strip_any_location_wrapper): New inline function.
27971 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
27974 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
27975 stack_realign_offset for the largest alignment of stack slot
27977 (ix86_find_max_used_stack_alignment): New function.
27978 (ix86_finalize_stack_frame_flags): Use it. Set
27979 max_used_stack_alignment if we don't realign stack.
27980 * config/i386/i386.h (machine_function): Add
27981 max_used_stack_alignment.
27983 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
27985 * config/arm/arm.opt (-mbranch-cost): New option.
27986 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
27989 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
27992 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
27993 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
27995 2018-01-10 Richard Biener <rguenther@suse.de>
27998 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
27999 early out so it also covers the case where we have a non-NULL
28002 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
28004 PR tree-optimization/83753
28005 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
28006 for non-strided grouped accesses if the number of elements is 1.
28008 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
28011 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
28012 * i386.h (TARGET_USE_GATHER): Define.
28013 * x86-tune.def (X86_TUNE_USE_GATHER): New.
28015 2018-01-10 Martin Liska <mliska@suse.cz>
28018 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
28019 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
28021 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
28022 CLEANUP_NO_PARTITIONING is not set.
28024 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
28026 * doc/rtl.texi: Remove documentation of (const ...) wrappers
28027 for vectors, as a partial revert of r254296.
28028 * rtl.h (const_vec_p): Delete.
28029 (const_vec_duplicate_p): Don't test for vector CONSTs.
28030 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
28031 * expmed.c (make_tree): Likewise.
28034 * common.md (E, F): Use CONSTANT_P instead of checking for
28036 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
28037 checking for CONST_VECTOR.
28039 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28041 PR middle-end/83575
28042 * predict.c (force_edge_cold): Handle in more sane way edges
28043 with no prediction.
28045 2018-01-09 Carl Love <cel@us.ibm.com>
28047 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
28049 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
28050 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
28051 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
28052 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
28053 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
28054 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
28055 * config/rs6000/rs6000-protos.h: Add extern defition for
28056 rs6000_generate_float2_double_code.
28057 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
28059 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
28060 (float2_v2df): Add define_expand.
28062 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
28065 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
28066 op_mode in the force_to_mode call.
28068 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
28070 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
28071 instead of checking each element individually.
28072 (aarch64_evpc_uzp): Likewise.
28073 (aarch64_evpc_zip): Likewise.
28074 (aarch64_evpc_ext): Likewise.
28075 (aarch64_evpc_rev): Likewise.
28076 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
28077 instead of checking each element individually. Return true without
28079 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
28080 whether all selected elements come from the same input, instead of
28081 checking each element individually. Remove calls to gen_rtx_REG,
28082 start_sequence and end_sequence and instead assert that no rtl is
28085 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
28087 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
28088 order of HIGH and CONST checks.
28090 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
28092 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
28093 if the destination isn't an SSA_NAME.
28095 2018-01-09 Richard Biener <rguenther@suse.de>
28097 PR tree-optimization/83668
28098 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
28100 (canonicalize_loop_form): ... here, renamed from ...
28101 (canonicalize_loop_closed_ssa_form): ... this and amended to
28102 swap successor edges for loop exit blocks to make us use
28103 the RPO order we need for initial schedule generation.
28105 2018-01-09 Joseph Myers <joseph@codesourcery.com>
28107 PR tree-optimization/64811
28108 * match.pd: When optimizing comparisons with Inf, avoid
28109 introducing or losing exceptions from comparisons with NaN.
28111 2018-01-09 Martin Liska <mliska@suse.cz>
28114 * asan.c (shadow_mem_size): Add gcc_assert.
28116 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
28118 Don't save registers in main().
28121 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
28122 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
28123 * config/avr/avr.c (avr_set_current_function): Don't error if
28124 naked, OS_task or OS_main are specified at the same time.
28125 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
28127 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
28129 * common/config/avr/avr-common.c (avr_option_optimization_table):
28130 Switch on -mmain-is-OS_task for optimizing compilations.
28132 2018-01-09 Richard Biener <rguenther@suse.de>
28134 PR tree-optimization/83572
28135 * graphite.c: Include cfganal.h.
28136 (graphite_transform_loops): Connect infinite loops to exit
28137 and remove fake edges at the end.
28139 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28141 * ipa-inline.c (edge_badness): Revert accidental checkin.
28143 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28146 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
28147 symbols; not inline clones.
28149 2018-01-09 Jakub Jelinek <jakub@redhat.com>
28152 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
28153 hard registers. Formatting fixes.
28155 PR preprocessor/83722
28156 * gcc.c (try_generate_repro): Pass
28157 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
28158 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
28161 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
28162 Kito Cheng <kito.cheng@gmail.com>
28164 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
28165 (riscv_leaf_function_p): Delete.
28166 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
28168 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28170 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
28172 (do_ifelse): New function.
28173 (do_isel): New function.
28174 (do_sub3): New function.
28175 (do_add3): New function.
28176 (do_load_mask_compare): New function.
28177 (do_overlap_load_compare): New function.
28178 (expand_compare_loop): New function.
28179 (expand_block_compare): Call expand_compare_loop() when appropriate.
28180 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
28181 option description.
28182 (-mblock-compare-inline-loop-limit): New option.
28184 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28187 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
28188 Reverse order of second and third operands in first alternative.
28189 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
28190 of first and second elements in UNSPEC_VPERMR vector.
28191 (altivec_expand_vec_perm_le): Likewise.
28193 2018-01-08 Jeff Law <law@redhat.com>
28195 PR rtl-optimizatin/81308
28196 * tree-switch-conversion.c (cfg_altered): New file scoped static.
28197 (process_switch): If group_case_labels makes a change, then set
28199 (pass_convert_switch::execute): If a switch is converted, then
28200 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
28202 PR rtl-optimization/81308
28203 * recog.c (split_all_insns): Conditionally cleanup the CFG after
28206 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
28208 PR target/83663 - Revert r255946
28209 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
28210 generation for cases where splatting a value is not useful.
28211 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
28212 across a vec_duplicate and a paradoxical subreg forming a vector
28213 mode to a vec_concat.
28215 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28217 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
28218 -march=armv8.3-a variants.
28219 * config/arm/t-multilib: Likewise.
28220 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
28222 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28224 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
28226 (cceq_ior_compare_complement): Give it a name so I can use it, and
28227 change boolean_or_operator predicate to boolean_operator so it can
28228 be used to generate a crand.
28229 (eqne): New code iterator.
28230 (bd/bd_neg): New code_attrs.
28231 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
28232 a single define_insn.
28233 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
28234 decrement (bdnzt/bdnzf/bdzt/bdzf).
28235 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
28236 with the new names of the branch decrement patterns, and added the
28237 names of the branch decrement conditional patterns.
28239 2018-01-08 Richard Biener <rguenther@suse.de>
28241 PR tree-optimization/83563
28242 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
28245 2018-01-08 Richard Biener <rguenther@suse.de>
28247 PR middle-end/83713
28248 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
28250 2018-01-08 Richard Biener <rguenther@suse.de>
28252 PR tree-optimization/83685
28253 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
28254 references to abnormals.
28256 2018-01-08 Richard Biener <rguenther@suse.de>
28259 * dwarf2out.c (output_indirect_strings): Handle empty
28260 skeleton_debug_str_hash.
28261 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
28263 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
28265 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
28266 (emit_store_direct): Likewise.
28267 (arc_trampoline_adjust_address): Likewise.
28268 (arc_asm_trampoline_template): New function.
28269 (arc_initialize_trampoline): Use asm_trampoline_template.
28270 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
28271 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
28272 * config/arc/arc.md (flush_icache): Delete pattern.
28274 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
28276 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
28277 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
28280 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
28283 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
28284 by not USED_FOR_TARGET.
28285 (make_pass_resolve_sw_modes): Likewise.
28287 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
28289 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
28292 2018-01-08 Richard Biener <rguenther@suse.de>
28294 PR middle-end/83580
28295 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
28297 2018-01-08 Richard Biener <rguenther@suse.de>
28299 PR middle-end/83517
28300 * match.pd ((t * 2) / 2) -> t): Add missing :c.
28302 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
28304 PR middle-end/81897
28305 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
28306 basic blocks with a small number of successors.
28307 (convert_control_dep_chain_into_preds): Improve handling of
28309 (dump_predicates): Split apart into...
28310 (dump_pred_chain): ...here...
28311 (dump_pred_info): ...and here.
28312 (can_one_predicate_be_invalidated_p): Add debugging printfs.
28313 (can_chain_union_be_invalidated_p): Improve check for invalidation
28315 (uninit_uses_cannot_happen): Avoid unnecessary if
28316 convert_control_dep_chain_into_preds yielded nothing.
28318 2018-01-06 Martin Sebor <msebor@redhat.com>
28320 PR tree-optimization/83640
28321 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
28322 subtracting negative offset from size.
28323 (builtin_access::overlap): Adjust offset bounds of the access to fall
28324 within the size of the object if possible.
28326 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
28328 PR rtl-optimization/83699
28329 * expmed.c (extract_bit_field_1): Restrict the vector usage of
28330 extract_bit_field_as_subreg to cases in which the extracted
28331 value is also a vector.
28333 * lra-constraints.c (process_alt_operands): Test for the equivalence
28334 substitutions when detecting a possible reload cycle.
28336 2018-01-06 Jakub Jelinek <jakub@redhat.com>
28339 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
28340 by default if flag_selective_schedling{,2}. Formatting fixes.
28342 PR rtl-optimization/83682
28343 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
28344 if it has non-VECTOR_MODE element mode.
28345 (vec_duplicate_p): Likewise.
28347 PR middle-end/83694
28348 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
28349 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
28351 2018-01-05 Jakub Jelinek <jakub@redhat.com>
28354 * config/i386/i386-builtin.def
28355 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
28356 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
28357 Require also OPTION_MASK_ISA_AVX512F in addition to
28358 OPTION_MASK_ISA_GFNI.
28359 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
28360 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
28361 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
28362 to OPTION_MASK_ISA_GFNI.
28363 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
28364 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
28365 OPTION_MASK_ISA_AVX512BW.
28366 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
28367 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
28368 addition to OPTION_MASK_ISA_GFNI.
28369 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
28370 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
28371 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
28372 to OPTION_MASK_ISA_GFNI.
28373 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
28374 a requirement for all ISAs rather than any of them with a few
28376 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
28378 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
28379 bitmasks to be enabled with 3 exceptions, instead of requiring any
28380 enabled ISA with lots of exceptions.
28381 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
28382 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
28383 Change avx512bw in isa attribute to avx512f.
28384 * config/i386/sgxintrin.h: Add license boilerplate.
28385 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
28386 to __AVX512F__ and __AVX512VL to __AVX512VL__.
28387 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
28388 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
28390 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
28391 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
28392 temporarily sse2 rather than sse if not enabled already.
28395 * config/i386/sse.md (VI248_VLBW): Rename to ...
28396 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
28397 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
28398 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
28399 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
28400 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
28401 mode iterator instead of VI248_VLBW.
28403 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
28405 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
28406 (record_modified): Skip clobbers; add debug output.
28407 (param_change_prob): Use sreal frequencies.
28409 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
28411 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
28412 punt for user-aligned variables.
28414 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
28416 * tree-chrec.c (chrec_contains_symbols): Return true for
28419 2018-01-05 Sudakshina Das <sudi.das@arm.com>
28422 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
28423 of (x|y) == x for BICS pattern.
28425 2018-01-05 Jakub Jelinek <jakub@redhat.com>
28427 PR tree-optimization/83605
28428 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
28429 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
28432 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
28434 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
28435 * config/epiphany/rtems.h: New file.
28437 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28438 Uros Bizjak <ubizjak@gmail.com>
28441 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
28442 QIreg_operand instead of register_operand predicate.
28443 * config/i386/i386.c (ix86_rop_should_change_byte_p,
28444 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
28445 comments instead of -fmitigate[-_]rop.
28447 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28450 * cgraphunit.c (symbol_table::compile): Switch to text_section
28451 before calling assembly_start debug hook.
28452 * run-rtl-passes.c (run_rtl_passes): Likewise.
28455 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28457 * tree-vrp.c (extract_range_from_binary_expr_1): Check
28458 range_int_cst_p rather than !symbolic_range_p before calling
28459 extract_range_from_multiplicative_op_1.
28461 2018-01-04 Jeff Law <law@redhat.com>
28463 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
28464 redundant test in assertion.
28466 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28468 * doc/rtl.texi: Document machine_mode wrapper classes.
28470 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28472 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
28473 using tree_to_uhwi.
28475 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28477 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
28478 the VEC_PERM_EXPR fold to fail.
28480 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28483 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
28484 to switched_sections.
28486 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28489 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
28490 test for d.testing.
28492 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
28495 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
28496 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
28498 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28501 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
28502 is BLKmode and bitpos not zero or mode change is needed.
28504 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28507 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
28510 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
28513 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
28514 instead of MULT rtx. Update all corresponding splitters.
28515 (*saddl_se): Ditto.
28516 (*ssub<modesuffix>): Ditto.
28517 (*ssubl_se): Ditto.
28518 (*cmp_sadd_di): Update split patterns.
28519 (*cmp_sadd_si): Ditto.
28520 (*cmp_sadd_sidi): Ditto.
28521 (*cmp_ssub_di): Ditto.
28522 (*cmp_ssub_si): Ditto.
28523 (*cmp_ssub_sidi): Ditto.
28524 * config/alpha/predicates.md (const23_operand): New predicate.
28525 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
28526 Look for ASHIFT, not MULT inner operand.
28527 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
28529 2018-01-04 Martin Liska <mliska@suse.cz>
28531 PR gcov-profile/83669
28532 * gcov.c (output_intermediate_file): Add version to intermediate
28534 * doc/gcov.texi: Document new field 'version' in intermediate
28535 file format. Fix location of '-k' option of gcov command.
28537 2018-01-04 Martin Liska <mliska@suse.cz>
28540 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
28542 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28544 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
28546 2018-01-03 Martin Sebor <msebor@redhat.com>
28548 PR tree-optimization/83655
28549 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
28550 checking calls with invalid arguments.
28552 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28554 * tree-vect-stmts.c (vect_get_store_rhs): New function.
28555 (vectorizable_mask_load_store): Delete.
28556 (vectorizable_call): Return false for masked loads and stores.
28557 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
28558 instead of gimple_assign_rhs1.
28559 (vectorizable_load): Handle IFN_MASK_LOAD.
28560 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
28562 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28564 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
28566 (vectorizable_mask_load_store): ...here.
28567 (vectorizable_load): ...and here.
28569 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28571 * tree-vect-stmts.c (vect_build_all_ones_mask)
28572 (vect_build_zero_merge_argument): New functions, split out from...
28573 (vectorizable_load): ...here.
28575 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28577 * tree-vect-stmts.c (vect_check_store_rhs): New function,
28579 (vectorizable_mask_load_store): ...here.
28580 (vectorizable_store): ...and here.
28582 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28584 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
28586 (vectorizable_mask_load_store): ...here.
28588 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28590 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
28591 (vect_model_store_cost): Take a vec_load_store_type instead of a
28593 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
28594 (vect_model_store_cost): Take a vec_load_store_type instead of a
28596 (vectorizable_mask_load_store): Update accordingly.
28597 (vectorizable_store): Likewise.
28598 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
28600 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28602 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
28603 IFN_MASK_LOAD calls here rather than...
28604 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
28606 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28607 Alan Hayward <alan.hayward@arm.com>
28608 David Sherwood <david.sherwood@arm.com>
28610 * expmed.c (extract_bit_field_1): For vector extracts,
28611 fall back to extract_bit_field_as_subreg if vec_extract
28614 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28615 Alan Hayward <alan.hayward@arm.com>
28616 David Sherwood <david.sherwood@arm.com>
28618 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
28619 they are variable or constant sized.
28620 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
28621 slots for constant-sized data.
28623 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28624 Alan Hayward <alan.hayward@arm.com>
28625 David Sherwood <david.sherwood@arm.com>
28627 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
28628 handling COND_EXPRs with boolean comparisons, try to find a better
28629 basis for the mask type than the boolean itself.
28631 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28633 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
28634 is calculated and how it can be overridden.
28635 * genmodes.c (max_bitsize_mode_any_mode): New variable.
28636 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
28638 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
28641 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28642 Alan Hayward <alan.hayward@arm.com>
28643 David Sherwood <david.sherwood@arm.com>
28645 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
28646 Remove the mode argument.
28647 (aarch64_simd_valid_immediate): Remove the mode and inverse
28649 * config/aarch64/iterators.md (bitsize): New iterator.
28650 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
28651 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
28652 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
28653 aarch64_simd_valid_immediate.
28654 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
28655 (aarch64_reg_or_bic_imm): Likewise.
28656 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
28657 with an insn_type enum and msl with a modifier_type enum.
28658 Replace element_width with a scalar_mode. Change the shift
28659 to unsigned int. Add constructors for scalar_float_mode and
28660 scalar_int_mode elements.
28661 (aarch64_vect_float_const_representable_p): Delete.
28662 (aarch64_can_const_movi_rtx_p)
28663 (aarch64_simd_scalar_immediate_valid_for_move)
28664 (aarch64_simd_make_constant): Update call to
28665 aarch64_simd_valid_immediate.
28666 (aarch64_advsimd_valid_immediate_hs): New function.
28667 (aarch64_advsimd_valid_immediate): Likewise.
28668 (aarch64_simd_valid_immediate): Remove mode and inverse
28669 arguments. Rewrite to use the above. Use const_vec_duplicate_p
28670 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
28671 and aarch64_float_const_representable_p on the result.
28672 (aarch64_output_simd_mov_immediate): Remove mode argument.
28673 Update call to aarch64_simd_valid_immediate and use of
28674 simd_immediate_info.
28675 (aarch64_output_scalar_simd_mov_immediate): Update call
28678 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28679 Alan Hayward <alan.hayward@arm.com>
28680 David Sherwood <david.sherwood@arm.com>
28682 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
28683 (mode_nunits): Likewise CONST_MODE_NUNITS.
28684 * machmode.def (ADJUST_NUNITS): Document.
28685 * genmodes.c (mode_data::need_nunits_adj): New field.
28686 (blank_mode): Update accordingly.
28687 (adj_nunits): New variable.
28688 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
28690 (emit_mode_size_inline): Set need_bytesize_adj for all modes
28691 listed in adj_nunits.
28692 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
28693 listed in adj_nunits. Don't emit case statements for such modes.
28694 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
28695 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
28696 nothing if adj_nunits is nonnull.
28697 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
28698 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
28699 (emit_mode_fbit): Update use of print_maybe_const_decl.
28700 (emit_move_size): Likewise. Treat the array as non-const
28702 (emit_mode_adjustments): Handle adj_nunits.
28704 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28706 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
28707 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
28708 (VECTOR_MODES): Use it.
28709 (make_vector_modes): Take the prefix as an argument.
28711 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28712 Alan Hayward <alan.hayward@arm.com>
28713 David Sherwood <david.sherwood@arm.com>
28715 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
28716 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
28717 for MODE_VECTOR_BOOL.
28718 * machmode.def (VECTOR_BOOL_MODE): Document.
28719 * genmodes.c (VECTOR_BOOL_MODE): New macro.
28720 (make_vector_bool_mode): New function.
28721 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
28723 * lto-streamer-in.c (lto_input_mode_table): Likewise.
28724 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
28726 * stor-layout.c (int_mode_for_mode): Likewise.
28727 * tree.c (build_vector_type_for_mode): Likewise.
28728 * varasm.c (output_constant_pool_2): Likewise.
28729 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
28730 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
28731 for MODE_VECTOR_BOOL.
28732 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
28733 of mode class checks.
28734 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
28735 instead of a list of mode class checks.
28736 (expand_vector_scalar_condition): Likewise.
28737 (type_for_widest_vector_mode): Handle BImode as an inner mode.
28739 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28740 Alan Hayward <alan.hayward@arm.com>
28741 David Sherwood <david.sherwood@arm.com>
28743 * machmode.h (mode_size): Change from unsigned short to
28745 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
28746 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28747 or if measurement_type is not polynomial.
28748 (fixed_size_mode::includes_p): Check for constant-sized modes.
28749 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
28750 return a poly_uint16 rather than an unsigned short.
28751 (emit_mode_size): Change the type of mode_size from unsigned short
28752 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
28753 (emit_mode_adjustments): Cope with polynomial vector sizes.
28754 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28756 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28758 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
28759 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
28760 * caller-save.c (setup_save_areas): Likewise.
28761 (replace_reg_with_saved_mem): Likewise.
28762 * calls.c (emit_library_call_value_1): Likewise.
28763 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
28764 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
28765 (gen_lowpart_for_combine): Likewise.
28766 * convert.c (convert_to_integer_1): Likewise.
28767 * cse.c (equiv_constant, cse_insn): Likewise.
28768 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
28769 (cselib_subst_to_values): Likewise.
28770 * dce.c (word_dce_process_block): Likewise.
28771 * df-problems.c (df_word_lr_mark_ref): Likewise.
28772 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
28773 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
28774 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
28775 (rtl_for_decl_location): Likewise.
28776 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
28777 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
28778 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
28779 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
28780 (expand_expr_real_1): Likewise.
28781 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
28782 (pad_below): Likewise.
28783 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28784 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
28785 * ira.c (get_subreg_tracking_sizes): Likewise.
28786 * ira-build.c (ira_create_allocno_objects): Likewise.
28787 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
28788 (ira_sort_regnos_for_alter_reg): Likewise.
28789 * ira-costs.c (record_operand_costs): Likewise.
28790 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
28791 (resolve_simple_move): Likewise.
28792 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
28793 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
28794 (lra_constraints): Likewise.
28795 (CONST_POOL_OK_P): Reject variable-sized modes.
28796 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
28797 (add_pseudo_to_slot, lra_spill): Likewise.
28798 * omp-low.c (omp_clause_aligned_alignment): Likewise.
28799 * optabs-query.c (get_best_extraction_insn): Likewise.
28800 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28801 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
28802 (expand_mult_highpart, valid_multiword_target_p): Likewise.
28803 * recog.c (offsettable_address_addr_space_p): Likewise.
28804 * regcprop.c (maybe_mode_change): Likewise.
28805 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
28806 * regrename.c (build_def_use): Likewise.
28807 * regstat.c (dump_reg_info): Likewise.
28808 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
28809 (find_reloads, find_reloads_subreg_address): Likewise.
28810 * reload1.c (eliminate_regs_1): Likewise.
28811 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
28812 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
28813 (simplify_binary_operation_1, simplify_subreg): Likewise.
28814 * targhooks.c (default_function_arg_padding): Likewise.
28815 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
28816 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
28817 (verify_gimple_assign_ternary): Likewise.
28818 * tree-inline.c (estimate_move_cost): Likewise.
28819 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28820 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
28821 (get_address_cost_ainc): Likewise.
28822 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28823 (vect_supportable_dr_alignment): Likewise.
28824 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28825 (vectorizable_reduction): Likewise.
28826 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
28827 (vectorizable_operation, vectorizable_load): Likewise.
28828 * tree.c (build_same_sized_truth_vector_type): Likewise.
28829 * valtrack.c (cleanup_auto_inc_dec): Likewise.
28830 * var-tracking.c (emit_note_insn_var_location): Likewise.
28831 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
28832 (ADDR_VEC_ALIGN): Likewise.
28834 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28835 Alan Hayward <alan.hayward@arm.com>
28836 David Sherwood <david.sherwood@arm.com>
28838 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
28840 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28841 or if measurement_type is polynomial.
28842 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
28843 * combine.c (make_extraction): Likewise.
28844 * dse.c (find_shift_sequence): Likewise.
28845 * dwarf2out.c (mem_loc_descriptor): Likewise.
28846 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
28847 (extract_bit_field, extract_low_bits): Likewise.
28848 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
28849 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
28850 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
28851 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
28852 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28853 * reload.c (find_reloads): Likewise.
28854 * reload1.c (alter_reg): Likewise.
28855 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
28856 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
28857 * tree-if-conv.c (predicate_mem_writes): Likewise.
28858 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
28859 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
28860 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
28861 * valtrack.c (dead_debug_insert_temp): Likewise.
28862 * varasm.c (mergeable_constant_section): Likewise.
28863 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
28865 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28866 Alan Hayward <alan.hayward@arm.com>
28867 David Sherwood <david.sherwood@arm.com>
28869 * expr.c (expand_assignment): Cope with polynomial mode sizes
28870 when assigning to a CONCAT.
28872 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28873 Alan Hayward <alan.hayward@arm.com>
28874 David Sherwood <david.sherwood@arm.com>
28876 * machmode.h (mode_precision): Change from unsigned short to
28878 (mode_to_precision): Return a poly_uint16 rather than an unsigned
28880 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
28881 or if measurement_type is not polynomial.
28882 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
28883 in which the mode is already known to be a scalar_int_mode.
28884 * genmodes.c (emit_mode_precision): Change the type of mode_precision
28885 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
28887 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28888 for GET_MODE_PRECISION.
28889 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28890 for GET_MODE_PRECISION.
28891 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
28893 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
28894 (expand_field_assignment, make_extraction): Likewise.
28895 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
28896 (get_last_value): Likewise.
28897 * convert.c (convert_to_integer_1): Likewise.
28898 * cse.c (cse_insn): Likewise.
28899 * expr.c (expand_expr_real_1): Likewise.
28900 * lra-constraints.c (simplify_operand_subreg): Likewise.
28901 * optabs-query.c (can_atomic_load_p): Likewise.
28902 * optabs.c (expand_atomic_load): Likewise.
28903 (expand_atomic_store): Likewise.
28904 * ree.c (combine_reaching_defs): Likewise.
28905 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
28906 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
28907 * tree.h (type_has_mode_precision_p): Likewise.
28908 * ubsan.c (instrument_si_overflow): Likewise.
28910 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28911 Alan Hayward <alan.hayward@arm.com>
28912 David Sherwood <david.sherwood@arm.com>
28914 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
28915 polynomial numbers of units.
28916 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
28917 (valid_vector_subparts_p): New function.
28918 (build_vector_type): Remove temporary shim and take the number
28919 of units as a poly_uint64 rather than an int.
28920 (build_opaque_vector_type): Take the number of units as a
28921 poly_uint64 rather than an int.
28922 * tree.c (build_vector_from_ctor): Handle polynomial
28923 TYPE_VECTOR_SUBPARTS.
28924 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
28925 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
28926 (build_vector_from_val): If the number of units is variable,
28927 use build_vec_duplicate_cst for constant operands and
28928 VEC_DUPLICATE_EXPR otherwise.
28929 (make_vector_type): Remove temporary is_constant ().
28930 (build_vector_type, build_opaque_vector_type): Take the number of
28931 units as a poly_uint64 rather than an int.
28932 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
28934 * cfgexpand.c (expand_debug_expr): Likewise.
28935 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
28936 (store_constructor, expand_expr_real_1): Likewise.
28937 (const_scalar_mask_from_tree): Likewise.
28938 * fold-const-call.c (fold_const_reduction): Likewise.
28939 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
28940 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
28941 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
28942 (fold_relational_const): Likewise.
28943 (native_interpret_vector): Likewise. Change the size from an
28944 int to an unsigned int.
28945 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
28946 TYPE_VECTOR_SUBPARTS.
28947 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
28948 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
28949 duplicating a non-constant operand into a variable-length vector.
28950 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
28951 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
28952 * ipa-icf.c (sem_variable::equals): Likewise.
28953 * match.pd: Likewise.
28954 * omp-simd-clone.c (simd_clone_subparts): Likewise.
28955 * print-tree.c (print_node): Likewise.
28956 * stor-layout.c (layout_type): Likewise.
28957 * targhooks.c (default_builtin_vectorization_cost): Likewise.
28958 * tree-cfg.c (verify_gimple_comparison): Likewise.
28959 (verify_gimple_assign_binary): Likewise.
28960 (verify_gimple_assign_ternary): Likewise.
28961 (verify_gimple_assign_single): Likewise.
28962 * tree-pretty-print.c (dump_generic_node): Likewise.
28963 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28964 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
28965 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
28966 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
28967 (vect_shift_permute_load_chain): Likewise.
28968 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
28969 (expand_vector_condition, optimize_vector_constructor): Likewise.
28970 (lower_vec_perm, get_compute_type): Likewise.
28971 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28972 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
28973 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
28974 (vect_recog_mask_conversion_pattern): Likewise.
28975 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
28976 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
28977 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28978 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
28979 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
28980 (vectorizable_shift, vectorizable_operation, vectorizable_store)
28981 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
28982 (supportable_widening_operation): Likewise.
28983 (supportable_narrowing_operation): Likewise.
28984 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
28986 * varasm.c (output_constant): Likewise.
28988 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28989 Alan Hayward <alan.hayward@arm.com>
28990 David Sherwood <david.sherwood@arm.com>
28992 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
28993 so that both the length == 3 and length != 3 cases set up their
28994 own permute vectors. Add comments explaining why we know the
28995 number of elements is constant.
28996 (vect_permute_load_chain): Likewise.
28998 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28999 Alan Hayward <alan.hayward@arm.com>
29000 David Sherwood <david.sherwood@arm.com>
29002 * machmode.h (mode_nunits): Change from unsigned char to
29004 (ONLY_FIXED_SIZE_MODES): New macro.
29005 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
29006 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
29007 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
29009 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
29010 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
29011 or if measurement_type is not polynomial.
29012 * genmodes.c (ZERO_COEFFS): New macro.
29013 (emit_mode_nunits_inline): Make mode_nunits_inline return a
29015 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
29016 Use ZERO_COEFFS when emitting initializers.
29017 * data-streamer.h (bp_pack_poly_value): New function.
29018 (bp_unpack_poly_value): Likewise.
29019 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
29020 for GET_MODE_NUNITS.
29021 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
29022 for GET_MODE_NUNITS.
29023 * tree.c (make_vector_type): Remove temporary shim and make
29024 the real function take the number of units as a poly_uint64
29025 rather than an int.
29026 (build_vector_type_for_mode): Handle polynomial nunits.
29027 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
29028 * emit-rtl.c (const_vec_series_p_1): Likewise.
29029 (gen_rtx_CONST_VECTOR): Likewise.
29030 * fold-const.c (test_vec_duplicate_folding): Likewise.
29031 * genrecog.c (validate_pattern): Likewise.
29032 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
29033 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
29034 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
29035 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
29036 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
29037 * rtlanal.c (subreg_get_info): Likewise.
29038 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29039 (vect_grouped_load_supported): Likewise.
29040 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
29041 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
29042 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
29043 (simplify_const_unary_operation, simplify_binary_operation_1)
29044 (simplify_const_binary_operation, simplify_ternary_operation)
29045 (test_vector_ops_duplicate, test_vector_ops): Likewise.
29046 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
29047 instead of CONST_VECTOR_NUNITS.
29048 * varasm.c (output_constant_pool_2): Likewise.
29049 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
29050 explicit-encoded elements in the XVEC for variable-length vectors.
29052 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29054 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
29056 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29057 Alan Hayward <alan.hayward@arm.com>
29058 David Sherwood <david.sherwood@arm.com>
29060 * coretypes.h (fixed_size_mode): Declare.
29061 (fixed_size_mode_pod): New typedef.
29062 * builtins.h (target_builtins::x_apply_args_mode)
29063 (target_builtins::x_apply_result_mode): Change type to
29064 fixed_size_mode_pod.
29065 * builtins.c (apply_args_size, apply_result_size, result_vector)
29066 (expand_builtin_apply_args_1, expand_builtin_apply)
29067 (expand_builtin_return): Update accordingly.
29069 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29071 * cse.c (hash_rtx_cb): Hash only the encoded elements.
29072 * cselib.c (cselib_hash_rtx): Likewise.
29073 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
29074 CONST_VECTOR encoding.
29076 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29077 Jeff Law <law@redhat.com>
29080 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
29081 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
29082 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
29083 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
29086 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
29087 explicitly probe *sp in a noreturn function if there were any callee
29088 register saves or frame pointer is needed.
29090 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29093 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
29094 BLKmode for ternary, binary or unary expressions.
29097 * var-tracking.c (delete_vta_debug_insn): New inline function.
29098 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
29099 insns from get_insns () to NULL instead of each bb separately.
29100 Use delete_vta_debug_insn. No longer static.
29101 (vt_debug_insns_local, variable_tracking_main_1): Adjust
29102 delete_vta_debug_insns callers.
29103 * rtl.h (delete_vta_debug_insns): Declare.
29104 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
29105 instead of variable_tracking_main.
29107 2018-01-03 Martin Sebor <msebor@redhat.com>
29109 PR tree-optimization/83603
29110 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
29111 arguments past the endof the argument list in functions declared
29112 without a prototype.
29113 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
29114 Avoid checking when arguments are null.
29116 2018-01-03 Martin Sebor <msebor@redhat.com>
29119 * doc/extend.texi (attribute const): Fix a typo.
29120 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
29121 issuing -Wsuggest-attribute for void functions.
29123 2018-01-03 Martin Sebor <msebor@redhat.com>
29125 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
29126 offset_int::from instead of wide_int::to_shwi.
29127 (maybe_diag_overlap): Remove assertion.
29128 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
29129 * gimple-ssa-sprintf.c (format_directive): Same.
29130 (parse_directive): Same.
29131 (sprintf_dom_walker::compute_format_length): Same.
29132 (try_substitute_return_value): Same.
29134 2018-01-03 Jeff Law <law@redhat.com>
29136 PR middle-end/83654
29137 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
29138 non-constant residual for zero at runtime and avoid probing in
29139 that case. Reorganize code for trailing problem to mirror handling
29142 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29144 PR tree-optimization/83501
29145 * tree-ssa-strlen.c (get_string_cst): New.
29146 (handle_char_store): Call get_string_cst.
29148 2018-01-03 Martin Liska <mliska@suse.cz>
29150 PR tree-optimization/83593
29151 * tree-ssa-strlen.c: Include tree-cfg.h.
29152 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
29153 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
29154 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
29156 (strlen_dom_walker::before_dom_children): Call
29157 gimple_purge_dead_eh_edges. Dump tranformation with details
29159 (strlen_dom_walker::before_dom_children): Update call by adding
29160 new argument cleanup_eh.
29161 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
29163 2018-01-03 Martin Liska <mliska@suse.cz>
29166 * cif-code.def (VARIADIC_THUNK): New enum value.
29167 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
29170 2018-01-03 Jan Beulich <jbeulich@suse.com>
29172 * sse.md (mov<mode>_internal): Tighten condition for when to use
29173 vmovdqu<ssescalarsize> for TI and OI modes.
29175 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29177 Update copyright years.
29179 2018-01-03 Martin Liska <mliska@suse.cz>
29182 * ipa-visibility.c (function_and_variable_visibility): Skip
29183 functions with noipa attribure.
29185 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29187 * gcc.c (process_command): Update copyright notice dates.
29188 * gcov-dump.c (print_version): Ditto.
29189 * gcov.c (print_version): Ditto.
29190 * gcov-tool.c (print_version): Ditto.
29191 * gengtype.c (create_file): Ditto.
29192 * doc/cpp.texi: Bump @copying's copyright year.
29193 * doc/cppinternals.texi: Ditto.
29194 * doc/gcc.texi: Ditto.
29195 * doc/gccint.texi: Ditto.
29196 * doc/gcov.texi: Ditto.
29197 * doc/install.texi: Ditto.
29198 * doc/invoke.texi: Ditto.
29200 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29202 * vector-builder.h (vector_builder::m_full_nelts): Change from
29203 unsigned int to poly_uint64.
29204 (vector_builder::full_nelts): Update prototype accordingly.
29205 (vector_builder::new_vector): Likewise.
29206 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
29207 (vector_builder::operator ==): Likewise.
29208 (vector_builder::finalize): Likewise.
29209 * int-vector-builder.h (int_vector_builder::int_vector_builder):
29210 Take the number of elements as a poly_uint64 rather than an
29212 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
29213 from unsigned int to poly_uint64.
29214 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
29215 (vec_perm_indices::new_vector): Likewise.
29216 (vec_perm_indices::length): Likewise.
29217 (vec_perm_indices::nelts_per_input): Likewise.
29218 (vec_perm_indices::input_nelts): Likewise.
29219 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
29220 number of elements per input as a poly_uint64 rather than an
29221 unsigned int. Use the original encoding for variable-length
29222 vectors, rather than clamping each individual element.
29223 For the second and subsequent elements in each pattern,
29224 clamp the step and base before clamping their sum.
29225 (vec_perm_indices::series_p): Handle polynomial element counts.
29226 (vec_perm_indices::all_in_range_p): Likewise.
29227 (vec_perm_indices_to_tree): Likewise.
29228 (vec_perm_indices_to_rtx): Likewise.
29229 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
29230 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
29231 (tree_vector_builder::new_binary_operation): Handle polynomial
29232 element counts. Return false if we need to know the number
29233 of elements at compile time.
29234 * fold-const.c (fold_vec_perm): Punt if the number of elements
29235 isn't known at compile time.
29237 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29239 * vec-perm-indices.h (vec_perm_builder): Change element type
29240 from HOST_WIDE_INT to poly_int64.
29241 (vec_perm_indices::element_type): Update accordingly.
29242 (vec_perm_indices::clamp): Handle polynomial element_types.
29243 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29244 (vec_perm_indices::all_in_range_p): Likewise.
29245 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
29247 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
29248 polynomial vec_perm_indices element types.
29249 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
29250 * fold-const.c (fold_vec_perm): Likewise.
29251 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
29252 * tree-vect-generic.c (lower_vec_perm): Likewise.
29253 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29254 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
29255 element type to HOST_WIDE_INT.
29257 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29258 Alan Hayward <alan.hayward@arm.com>
29259 David Sherwood <david.sherwood@arm.com>
29261 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
29262 rather than an int. Use plus_constant.
29263 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
29264 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
29266 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29267 Alan Hayward <alan.hayward@arm.com>
29268 David Sherwood <david.sherwood@arm.com>
29270 * calls.c (emit_call_1, expand_call): Change struct_value_size from
29271 a HOST_WIDE_INT to a poly_int64.
29273 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29274 Alan Hayward <alan.hayward@arm.com>
29275 David Sherwood <david.sherwood@arm.com>
29277 * calls.c (load_register_parameters): Cope with polynomial
29278 mode sizes. Require a constant size for BLKmode parameters
29279 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
29280 forces a parameter to be padded at the lsb end in order to
29281 fill a complete number of words, require the parameter size
29282 to be ordered wrt UNITS_PER_WORD.
29284 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29285 Alan Hayward <alan.hayward@arm.com>
29286 David Sherwood <david.sherwood@arm.com>
29288 * reload1.c (spill_stack_slot_width): Change element type
29289 from unsigned int to poly_uint64_pod.
29290 (alter_reg): Treat mode sizes as polynomial.
29292 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29293 Alan Hayward <alan.hayward@arm.com>
29294 David Sherwood <david.sherwood@arm.com>
29296 * reload.c (complex_word_subreg_p): New function.
29297 (reload_inner_reg_of_subreg, push_reload): Use it.
29299 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29300 Alan Hayward <alan.hayward@arm.com>
29301 David Sherwood <david.sherwood@arm.com>
29303 * lra-constraints.c (process_alt_operands): Reject matched
29304 operands whose sizes aren't ordered.
29305 (match_reload): Refer to this check here.
29307 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29308 Alan Hayward <alan.hayward@arm.com>
29309 David Sherwood <david.sherwood@arm.com>
29311 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
29312 that the mode size is in the set {1, 2, 4, 8, 16}.
29314 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29315 Alan Hayward <alan.hayward@arm.com>
29316 David Sherwood <david.sherwood@arm.com>
29318 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
29319 Use plus_constant instead of gen_rtx_PLUS.
29321 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29322 Alan Hayward <alan.hayward@arm.com>
29323 David Sherwood <david.sherwood@arm.com>
29325 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
29326 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
29327 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
29328 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
29329 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
29330 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
29331 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
29332 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
29333 * config/i386/i386.c (ix86_push_rounding): ...this new function.
29334 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
29336 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
29337 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
29338 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
29339 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
29340 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
29341 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
29342 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
29343 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
29344 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
29345 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
29347 * expr.c (emit_move_resolve_push): Treat the input and result
29348 of PUSH_ROUNDING as a poly_int64.
29349 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
29350 (emit_push_insn): Likewise.
29351 * lra-eliminations.c (mark_not_eliminable): Likewise.
29352 * recog.c (push_operand): Likewise.
29353 * reload1.c (elimination_effects): Likewise.
29354 * rtlanal.c (nonzero_bits1): Likewise.
29355 * calls.c (store_one_arg): Likewise. Require the padding to be
29356 known at compile time.
29358 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29359 Alan Hayward <alan.hayward@arm.com>
29360 David Sherwood <david.sherwood@arm.com>
29362 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
29363 Use plus_constant instead of gen_rtx_PLUS.
29365 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29366 Alan Hayward <alan.hayward@arm.com>
29367 David Sherwood <david.sherwood@arm.com>
29369 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
29370 rather than an int.
29372 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29373 Alan Hayward <alan.hayward@arm.com>
29374 David Sherwood <david.sherwood@arm.com>
29376 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
29377 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
29378 via stack temporaries. Treat the mode size as polynomial too.
29380 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29381 Alan Hayward <alan.hayward@arm.com>
29382 David Sherwood <david.sherwood@arm.com>
29384 * expr.c (expand_expr_real_2): When handling conversions involving
29385 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
29386 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
29387 as a poly_uint64 too.
29389 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29390 Alan Hayward <alan.hayward@arm.com>
29391 David Sherwood <david.sherwood@arm.com>
29393 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
29395 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29396 Alan Hayward <alan.hayward@arm.com>
29397 David Sherwood <david.sherwood@arm.com>
29399 * combine.c (can_change_dest_mode): Handle polynomial
29400 REGMODE_NATURAL_SIZE.
29401 * expmed.c (store_bit_field_1): Likewise.
29402 * expr.c (store_constructor): Likewise.
29403 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
29404 and polynomial REGMODE_NATURAL_SIZE.
29405 (gen_lowpart_common): Likewise.
29406 * reginfo.c (record_subregs_of_mode): Likewise.
29407 * rtlanal.c (read_modify_subreg_p): Likewise.
29409 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29410 Alan Hayward <alan.hayward@arm.com>
29411 David Sherwood <david.sherwood@arm.com>
29413 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
29414 numbers of elements.
29416 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29417 Alan Hayward <alan.hayward@arm.com>
29418 David Sherwood <david.sherwood@arm.com>
29420 * match.pd: Cope with polynomial numbers of vector elements.
29422 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29423 Alan Hayward <alan.hayward@arm.com>
29424 David Sherwood <david.sherwood@arm.com>
29426 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
29427 in a POINTER_PLUS_EXPR.
29429 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29430 Alan Hayward <alan.hayward@arm.com>
29431 David Sherwood <david.sherwood@arm.com>
29433 * omp-simd-clone.c (simd_clone_subparts): New function.
29434 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
29435 (ipa_simd_modify_function_body): Likewise.
29437 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29438 Alan Hayward <alan.hayward@arm.com>
29439 David Sherwood <david.sherwood@arm.com>
29441 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
29442 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
29443 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
29444 (expand_vector_condition, vector_element): Likewise.
29445 (subparts_gt): New function.
29446 (get_compute_type): Use subparts_gt.
29447 (count_type_subparts): Delete.
29448 (expand_vector_operations_1): Use subparts_gt instead of
29449 count_type_subparts.
29451 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29452 Alan Hayward <alan.hayward@arm.com>
29453 David Sherwood <david.sherwood@arm.com>
29455 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
29456 (vect_compile_time_alias): ...this new function. Do the calculation
29457 on poly_ints rather than trees.
29458 (vect_prune_runtime_alias_test_list): Update call accordingly.
29460 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29461 Alan Hayward <alan.hayward@arm.com>
29462 David Sherwood <david.sherwood@arm.com>
29464 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
29466 (vect_schedule_slp_instance): Likewise.
29468 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29469 Alan Hayward <alan.hayward@arm.com>
29470 David Sherwood <david.sherwood@arm.com>
29472 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
29473 constant and extern definitions for variable-length vectors.
29474 (vect_get_constant_vectors): Note that the number of units
29475 is known to be constant.
29477 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29478 Alan Hayward <alan.hayward@arm.com>
29479 David Sherwood <david.sherwood@arm.com>
29481 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
29482 of units as polynomial. Choose between WIDE and NARROW based
29485 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29486 Alan Hayward <alan.hayward@arm.com>
29487 David Sherwood <david.sherwood@arm.com>
29489 * tree-vect-stmts.c (simd_clone_subparts): New function.
29490 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
29492 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29493 Alan Hayward <alan.hayward@arm.com>
29494 David Sherwood <david.sherwood@arm.com>
29496 * tree-vect-stmts.c (vectorizable_call): Treat the number of
29497 vectors as polynomial. Use build_index_vector for
29498 IFN_GOMP_SIMD_LANE.
29500 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29501 Alan Hayward <alan.hayward@arm.com>
29502 David Sherwood <david.sherwood@arm.com>
29504 * tree-vect-stmts.c (get_load_store_type): Treat the number of
29505 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
29506 for variable-length vectors.
29507 (vectorizable_mask_load_store): Treat the number of units as
29508 polynomial, asserting that it is constant if the condition has
29509 already been enforced.
29510 (vectorizable_store, vectorizable_load): Likewise.
29512 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29513 Alan Hayward <alan.hayward@arm.com>
29514 David Sherwood <david.sherwood@arm.com>
29516 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
29517 of units as polynomial. Punt if we can't tell at compile time
29518 which vector contains the final result.
29520 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29521 Alan Hayward <alan.hayward@arm.com>
29522 David Sherwood <david.sherwood@arm.com>
29524 * tree-vect-loop.c (vectorizable_induction): Treat the number
29525 of units as polynomial. Punt on SLP inductions. Use an integer
29526 VEC_SERIES_EXPR for variable-length integer reductions. Use a
29527 cast of such a series for variable-length floating-point
29530 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29531 Alan Hayward <alan.hayward@arm.com>
29532 David Sherwood <david.sherwood@arm.com>
29534 * tree.h (build_index_vector): Declare.
29535 * tree.c (build_index_vector): New function.
29536 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
29537 of units as polynomial, forcibly converting it to a constant if
29538 vectorizable_reduction has already enforced the condition.
29539 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
29540 to create a {1,2,3,...} vector.
29541 (vectorizable_reduction): Treat the number of units as polynomial.
29542 Choose vectype_in based on the largest scalar element size rather
29543 than the smallest number of units. Enforce the restrictions
29546 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29547 Alan Hayward <alan.hayward@arm.com>
29548 David Sherwood <david.sherwood@arm.com>
29550 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
29551 number of units as polynomial.
29553 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29554 Alan Hayward <alan.hayward@arm.com>
29555 David Sherwood <david.sherwood@arm.com>
29557 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
29558 * target.def (autovectorize_vector_sizes): Return the vector sizes
29559 by pointer, using vector_sizes rather than a bitmask.
29560 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
29561 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
29562 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
29564 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
29565 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
29566 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
29567 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
29568 * omp-general.c (omp_max_vf): Likewise.
29569 * omp-low.c (omp_clause_aligned_alignment): Likewise.
29570 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
29571 * tree-vect-loop.c (vect_analyze_loop): Likewise.
29572 * tree-vect-slp.c (vect_slp_bb): Likewise.
29573 * doc/tm.texi: Regenerate.
29574 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
29576 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
29577 the vector size as a poly_uint64 rather than an unsigned int.
29578 (current_vector_size): Change from an unsigned int to a poly_uint64.
29579 (get_vectype_for_scalar_type): Update accordingly.
29580 * tree.h (build_truth_vector_type): Take the size and number of
29581 units as a poly_uint64 rather than an unsigned int.
29582 (build_vector_type): Add a temporary overload that takes
29583 the number of units as a poly_uint64 rather than an unsigned int.
29584 * tree.c (make_vector_type): Likewise.
29585 (build_truth_vector_type): Take the number of units as a poly_uint64
29586 rather than an unsigned int.
29588 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29589 Alan Hayward <alan.hayward@arm.com>
29590 David Sherwood <david.sherwood@arm.com>
29592 * target.def (get_mask_mode): Take the number of units and length
29593 as poly_uint64s rather than unsigned ints.
29594 * targhooks.h (default_get_mask_mode): Update accordingly.
29595 * targhooks.c (default_get_mask_mode): Likewise.
29596 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
29597 * doc/tm.texi: Regenerate.
29599 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29600 Alan Hayward <alan.hayward@arm.com>
29601 David Sherwood <david.sherwood@arm.com>
29603 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
29604 * omp-general.c (omp_max_vf): Likewise.
29605 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
29606 (expand_omp_simd): Handle polynomial safelen.
29607 * omp-low.c (omplow_simd_context): Add a default constructor.
29608 (omplow_simd_context::max_vf): Change from int to poly_uint64.
29609 (lower_rec_simd_input_clauses): Update accordingly.
29610 (lower_rec_input_clauses): Likewise.
29612 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29613 Alan Hayward <alan.hayward@arm.com>
29614 David Sherwood <david.sherwood@arm.com>
29616 * tree-vectorizer.h (vect_nunits_for_cost): New function.
29617 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
29618 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
29619 (vect_analyze_slp_cost): Likewise.
29620 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
29621 (vect_model_load_cost): Likewise.
29623 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29624 Alan Hayward <alan.hayward@arm.com>
29625 David Sherwood <david.sherwood@arm.com>
29627 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
29628 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
29629 from an unsigned int * to a poly_uint64_pod *.
29630 (calculate_unrolling_factor): New function.
29631 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
29633 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29634 Alan Hayward <alan.hayward@arm.com>
29635 David Sherwood <david.sherwood@arm.com>
29637 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
29638 from an unsigned int to a poly_uint64.
29639 (_loop_vec_info::slp_unrolling_factor): Likewise.
29640 (_loop_vec_info::vectorization_factor): Change from an int
29642 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
29643 (vect_get_num_vectors): New function.
29644 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
29645 (vect_get_num_copies): Use vect_get_num_vectors.
29646 (vect_analyze_data_ref_dependences): Change max_vf from an int *
29647 to an unsigned int *.
29648 (vect_analyze_data_refs): Change min_vf from an int * to a
29650 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29651 than an unsigned HOST_WIDE_INT.
29652 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
29653 (vect_analyze_data_ref_dependence): Change max_vf from an int *
29654 to an unsigned int *.
29655 (vect_analyze_data_ref_dependences): Likewise.
29656 (vect_compute_data_ref_alignment): Handle polynomial vf.
29657 (vect_enhance_data_refs_alignment): Likewise.
29658 (vect_prune_runtime_alias_test_list): Likewise.
29659 (vect_shift_permute_load_chain): Likewise.
29660 (vect_supportable_dr_alignment): Likewise.
29661 (dependence_distance_ge_vf): Take the vectorization factor as a
29662 poly_uint64 rather than an unsigned HOST_WIDE_INT.
29663 (vect_analyze_data_refs): Change min_vf from an int * to a
29665 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
29666 vfm1 as a poly_uint64 rather than an int. Make the same change
29667 for the returned bound_scalar.
29668 (vect_gen_vector_loop_niters): Handle polynomial vf.
29669 (vect_do_peeling): Likewise. Update call to
29670 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
29671 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
29673 * tree-vect-loop.c (vect_determine_vectorization_factor)
29674 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
29675 (vect_get_known_peeling_cost): Likewise.
29676 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
29677 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
29678 (vect_transform_loop): Likewise. Use the lowest possible VF when
29679 updating the upper bounds of the loop.
29680 (vect_min_worthwhile_factor): Make static. Return an unsigned int
29681 rather than an int.
29682 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
29683 polynomial unroll factors.
29684 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
29685 (vect_make_slp_decision): Likewise.
29686 (vect_supported_load_permutation_p): Likewise, and polynomial
29688 (vect_analyze_slp_cost): Handle polynomial vf.
29689 (vect_slp_analyze_node_operations): Likewise.
29690 (vect_slp_analyze_bb_1): Likewise.
29691 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29692 than an unsigned HOST_WIDE_INT.
29693 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
29694 (vectorizable_load): Handle polynomial vf.
29695 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
29697 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
29699 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29700 Alan Hayward <alan.hayward@arm.com>
29701 David Sherwood <david.sherwood@arm.com>
29703 * match.pd: Handle bit operations involving three constants
29704 and try to fold one pair.
29706 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29708 * tree-vect-loop-manip.c: Include gimple-fold.h.
29709 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
29710 niters_maybe_zero parameters. Handle other cases besides a step of 1.
29711 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
29712 Add a path that uses a step of VF instead of 1, but disable it
29714 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
29715 and niters_no_overflow parameters. Update calls to
29716 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
29717 Create a new SSA name if the latter choses to use a ste other
29718 than zero, and return it via niters_vector_mult_vf_var.
29719 * tree-vect-loop.c (vect_transform_loop): Update calls to
29720 vect_do_peeling, vect_gen_vector_loop_niters and
29721 slpeel_make_loop_iterate_ntimes.
29722 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
29723 (vect_gen_vector_loop_niters): Update declarations after above changes.
29725 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
29727 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
29728 128-bit round to integer instructions.
29729 (ceil<mode>2): Likewise.
29730 (btrunc<mode>2): Likewise.
29731 (round<mode>2): Likewise.
29733 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29735 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
29736 unaligned VSX load/store on P8/P9.
29737 (expand_block_clear): Allow the use of unaligned VSX
29738 load/store on P8/P9.
29740 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29742 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
29744 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
29745 swap associated with both a load and a store.
29747 2018-01-02 Andrew Waterman <andrew@sifive.com>
29749 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
29750 * config/riscv/riscv.md (clear_cache): Use it.
29752 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
29754 * web.c: Remove out-of-date comment.
29756 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29758 * expr.c (fixup_args_size_notes): Check that any existing
29759 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
29760 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
29761 (emit_single_push_insn): ...here.
29763 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29765 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
29766 (const_vector_encoded_nelts): New function.
29767 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
29768 (const_vector_int_elt, const_vector_elt): Declare.
29769 * emit-rtl.c (const_vector_int_elt_1): New function.
29770 (const_vector_elt): Likewise.
29771 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
29772 of CONST_VECTOR_ELT.
29774 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29776 * expr.c: Include rtx-vector-builder.h.
29777 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
29778 directly on the tree encoding.
29779 (const_vector_from_tree): Likewise.
29780 * optabs.c: Include rtx-vector-builder.h.
29781 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
29782 sequence of "u" values.
29783 * vec-perm-indices.c: Include rtx-vector-builder.h.
29784 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
29785 directly on the vec_perm_indices encoding.
29787 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29789 * doc/rtl.texi (const_vector): Describe new encoding scheme.
29790 * Makefile.in (OBJS): Add rtx-vector-builder.o.
29791 * rtx-vector-builder.h: New file.
29792 * rtx-vector-builder.c: Likewise.
29793 * rtl.h (rtx_def::u2): Add a const_vector field.
29794 (CONST_VECTOR_NPATTERNS): New macro.
29795 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
29796 (CONST_VECTOR_DUPLICATE_P): Likewise.
29797 (CONST_VECTOR_STEPPED_P): Likewise.
29798 (CONST_VECTOR_ENCODED_ELT): Likewise.
29799 (const_vec_duplicate_p): Check for a duplicated vector encoding.
29800 (unwrap_const_vec_duplicate): Likewise.
29801 (const_vec_series_p): Check for a non-duplicated vector encoding.
29802 Say that the function only returns true for integer vectors.
29803 * emit-rtl.c: Include rtx-vector-builder.h.
29804 (gen_const_vec_duplicate_1): Delete.
29805 (gen_const_vector): Call gen_const_vec_duplicate instead of
29806 gen_const_vec_duplicate_1.
29807 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
29808 (gen_const_vec_duplicate): Use rtx_vector_builder.
29809 (gen_const_vec_series): Likewise.
29810 (gen_rtx_CONST_VECTOR): Likewise.
29811 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
29812 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29813 Build a new vector rather than modifying a CONST_VECTOR in-place.
29814 (handle_special_swappables): Update call accordingly.
29815 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
29816 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29817 Build a new vector rather than modifying a CONST_VECTOR in-place.
29818 (handle_special_swappables): Update call accordingly.
29820 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29822 * simplify-rtx.c (simplify_const_binary_operation): Use
29823 CONST_VECTOR_ELT instead of XVECEXP.
29825 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29827 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
29828 the selector elements to be different from the data elements
29829 if the selector is a VECTOR_CST.
29830 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
29831 ssizetype for the selector.
29833 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29835 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
29836 before testing each element individually.
29837 * tree-vect-generic.c (lower_vec_perm): Likewise.
29839 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29841 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
29842 * selftest-run-tests.c (selftest::run_tests): Call it.
29843 * vector-builder.h (vector_builder::operator ==): New function.
29844 (vector_builder::operator !=): Likewise.
29845 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
29846 (vec_perm_indices::all_from_input_p): New function.
29847 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29848 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
29849 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
29850 instead of reading the VECTOR_CST directly. Detect whether both
29851 vector inputs are the same before constructing the vec_perm_indices,
29852 and update the number of inputs argument accordingly. Use the
29853 utility functions added above. Only construct sel2 if we need to.
29855 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29857 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
29858 the broadcast of the low byte.
29859 (expand_mult_highpart): Use an explicit encoding for the permutes.
29860 * optabs-query.c (can_mult_highpart_p): Likewise.
29861 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
29862 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29863 (vectorizable_bswap): Likewise.
29864 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
29865 explicit encoding for the power-of-2 permutes.
29866 (vect_permute_store_chain): Likewise.
29867 (vect_grouped_load_supported): Likewise.
29868 (vect_permute_load_chain): Likewise.
29870 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29872 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
29873 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
29874 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
29875 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29876 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
29877 (vect_gen_perm_mask_any): Likewise.
29879 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29881 * int-vector-builder.h: New file.
29882 * vec-perm-indices.h: Include int-vector-builder.h.
29883 (vec_perm_indices): Redefine as an int_vector_builder.
29884 (auto_vec_perm_indices): Delete.
29885 (vec_perm_builder): Redefine as a stand-alone class.
29886 (vec_perm_indices::vec_perm_indices): New function.
29887 (vec_perm_indices::clamp): Likewise.
29888 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
29889 (vec_perm_indices::new_vector): New function.
29890 (vec_perm_indices::new_expanded_vector): Update for new
29891 vec_perm_indices class.
29892 (vec_perm_indices::rotate_inputs): New function.
29893 (vec_perm_indices::all_in_range_p): Operate directly on the
29894 encoded form, without computing elided elements.
29895 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
29896 encoding. Update for new vec_perm_indices class.
29897 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
29898 the given vec_perm_builder.
29899 (expand_vec_perm_var): Update vec_perm_builder constructor.
29900 (expand_mult_highpart): Use vec_perm_builder instead of
29901 auto_vec_perm_indices.
29902 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
29903 vec_perm_indices instead of auto_vec_perm_indices. Use a single
29904 or double series encoding as appropriate.
29905 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
29906 vec_perm_indices instead of auto_vec_perm_indices.
29907 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29908 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29909 (vect_permute_store_chain): Likewise.
29910 (vect_grouped_load_supported): Likewise.
29911 (vect_permute_load_chain): Likewise.
29912 (vect_shift_permute_load_chain): Likewise.
29913 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29914 (vect_transform_slp_perm_load): Likewise.
29915 (vect_schedule_slp_instance): Likewise.
29916 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29917 (vectorizable_mask_load_store): Likewise.
29918 (vectorizable_bswap): Likewise.
29919 (vectorizable_store): Likewise.
29920 (vectorizable_load): Likewise.
29921 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
29922 vec_perm_indices instead of auto_vec_perm_indices. Use
29923 tree_to_vec_perm_builder to read the vector from a tree.
29924 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
29925 vec_perm_builder instead of a vec_perm_indices.
29926 (have_whole_vector_shift): Use vec_perm_builder and
29927 vec_perm_indices instead of auto_vec_perm_indices. Leave the
29928 truncation to calc_vec_perm_mask_for_shift.
29929 (vect_create_epilog_for_reduction): Likewise.
29930 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
29931 from auto_vec_perm_indices to vec_perm_indices.
29932 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29933 instead of changing individual elements.
29934 (aarch64_vectorize_vec_perm_const): Use new_vector to install
29935 the vector in d.perm.
29936 * config/arm/arm.c (expand_vec_perm_d::perm): Change
29937 from auto_vec_perm_indices to vec_perm_indices.
29938 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29939 instead of changing individual elements.
29940 (arm_vectorize_vec_perm_const): Use new_vector to install
29941 the vector in d.perm.
29942 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
29943 Update vec_perm_builder constructor.
29944 (rs6000_expand_interleave): Likewise.
29945 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
29946 (rs6000_expand_interleave): Likewise.
29948 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29950 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
29951 to qimode could truncate the indices.
29952 * optabs.c (expand_vec_perm_var): Likewise.
29954 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29956 * Makefile.in (OBJS): Add vec-perm-indices.o.
29957 * vec-perm-indices.h: New file.
29958 * vec-perm-indices.c: Likewise.
29959 * target.h (vec_perm_indices): Replace with a forward class
29961 (auto_vec_perm_indices): Move to vec-perm-indices.h.
29962 * optabs.h: Include vec-perm-indices.h.
29963 (expand_vec_perm): Delete.
29964 (selector_fits_mode_p, expand_vec_perm_var): Declare.
29965 (expand_vec_perm_const): Declare.
29966 * target.def (vec_perm_const_ok): Replace with...
29967 (vec_perm_const): ...this new hook.
29968 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
29969 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
29970 * doc/tm.texi: Regenerate.
29971 * optabs.def (vec_perm_const): Delete.
29972 * doc/md.texi (vec_perm_const): Likewise.
29973 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
29974 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
29975 expand_vec_perm for constant permutation vectors. Assert that
29976 the mode of variable permutation vectors is the integer equivalent
29977 of the mode that is being permuted.
29978 * optabs-query.h (selector_fits_mode_p): Declare.
29979 * optabs-query.c: Include vec-perm-indices.h.
29980 (selector_fits_mode_p): New function.
29981 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
29982 is defined, instead of checking whether the vec_perm_const_optab
29983 exists. Use targetm.vectorize.vec_perm_const instead of
29984 targetm.vectorize.vec_perm_const_ok. Check whether the indices
29985 fit in the vector mode before using a variable permute.
29986 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
29987 vec_perm_indices instead of an rtx.
29988 (expand_vec_perm): Replace with...
29989 (expand_vec_perm_const): ...this new function. Take the selector
29990 as a vec_perm_indices rather than an rtx. Also take the mode of
29991 the selector. Update call to shift_amt_for_vec_perm_mask.
29992 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
29993 Use vec_perm_indices::new_expanded_vector to expand the original
29994 selector into bytes. Check whether the indices fit in the vector
29995 mode before using a variable permute.
29996 (expand_vec_perm_var): Make global.
29997 (expand_mult_highpart): Use expand_vec_perm_const.
29998 * fold-const.c: Includes vec-perm-indices.h.
29999 * tree-ssa-forwprop.c: Likewise.
30000 * tree-vect-data-refs.c: Likewise.
30001 * tree-vect-generic.c: Likewise.
30002 * tree-vect-loop.c: Likewise.
30003 * tree-vect-slp.c: Likewise.
30004 * tree-vect-stmts.c: Likewise.
30005 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
30007 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
30008 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
30009 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
30010 (aarch64_vectorize_vec_perm_const): ...this new function.
30011 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
30012 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30013 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
30014 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
30015 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
30016 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30017 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
30019 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
30020 check for NEON modes.
30021 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
30022 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
30023 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
30024 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
30026 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
30027 the old VEC_PERM_CONST conditions.
30028 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
30029 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
30030 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
30031 (ia64_vectorize_vec_perm_const_ok): Merge into...
30032 (ia64_vectorize_vec_perm_const): ...this new function.
30033 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
30034 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
30035 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
30036 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
30037 * config/mips/mips.c (mips_expand_vec_perm_const)
30038 (mips_vectorize_vec_perm_const_ok): Merge into...
30039 (mips_vectorize_vec_perm_const): ...this new function.
30040 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
30041 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
30042 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
30043 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
30044 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
30045 (rs6000_expand_vec_perm_const): Delete.
30046 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
30048 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30049 (altivec_expand_vec_perm_const_le): Take each operand individually.
30050 Operate on constant selectors rather than rtxes.
30051 (altivec_expand_vec_perm_const): Likewise. Update call to
30052 altivec_expand_vec_perm_const_le.
30053 (rs6000_expand_vec_perm_const): Delete.
30054 (rs6000_vectorize_vec_perm_const_ok): Delete.
30055 (rs6000_vectorize_vec_perm_const): New function.
30056 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
30057 an element count and rtx array.
30058 (rs6000_expand_extract_even): Update call accordingly.
30059 (rs6000_expand_interleave): Likewise.
30060 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
30061 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
30062 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
30063 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
30064 (rs6000_expand_vec_perm_const): Delete.
30065 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
30066 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30067 (altivec_expand_vec_perm_const_le): Take each operand individually.
30068 Operate on constant selectors rather than rtxes.
30069 (altivec_expand_vec_perm_const): Likewise. Update call to
30070 altivec_expand_vec_perm_const_le.
30071 (rs6000_expand_vec_perm_const): Delete.
30072 (rs6000_vectorize_vec_perm_const_ok): Delete.
30073 (rs6000_vectorize_vec_perm_const): New function. Remove stray
30074 reference to the SPE evmerge intructions.
30075 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
30076 an element count and rtx array.
30077 (rs6000_expand_extract_even): Update call accordingly.
30078 (rs6000_expand_interleave): Likewise.
30079 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
30080 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
30082 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30084 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30086 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
30087 vector mode and that that mode matches the mode of the data
30089 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
30090 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
30091 directly using expand_vec_perm_1 when forcing selectors into
30093 (expand_vec_perm_var): New function, split out from expand_vec_perm.
30095 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30097 * optabs-query.h (can_vec_perm_p): Delete.
30098 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
30099 * optabs-query.c (can_vec_perm_p): Split into...
30100 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
30101 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
30102 particular selector is valid.
30103 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
30104 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
30105 (vect_grouped_load_supported): Likewise.
30106 (vect_shift_permute_load_chain): Likewise.
30107 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
30108 (vect_transform_slp_perm_load): Likewise.
30109 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
30110 (vectorizable_bswap): Likewise.
30111 (vect_gen_perm_mask_checked): Likewise.
30112 * fold-const.c (fold_ternary_loc): Likewise. Don't take
30113 implementations of variable permutation vectors into account
30114 when deciding which selector to use.
30115 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
30116 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
30117 with a false third argument.
30118 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
30119 to test whether the constant selector is valid and can_vec_perm_var_p
30120 to test whether a variable selector is valid.
30122 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30124 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
30125 * optabs-query.c (can_vec_perm_p): Likewise.
30126 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
30127 instead of vec_perm_indices.
30128 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
30129 (vect_gen_perm_mask_checked): Likewise,
30130 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
30131 (vect_gen_perm_mask_checked): Likewise,
30133 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30135 * optabs-query.h (qimode_for_vec_perm): Declare.
30136 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
30137 (qimode_for_vec_perm): ...this new function.
30138 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
30140 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30142 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
30143 does not have a conditional at the top.
30145 2018-01-02 Richard Biener <rguenther@suse.de>
30147 * ipa-inline.c (big_speedup_p): Fix expression.
30149 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
30152 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
30155 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
30159 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
30160 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
30161 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
30162 cond_taken_branch_cost 3->4.
30164 2018-01-01 Jakub Jelinek <jakub@redhat.com>
30166 PR tree-optimization/83581
30167 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
30168 TODO_cleanup_cfg if any changes have been made.
30170 PR middle-end/83608
30171 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
30172 convert_modes if target mode has the right side, but different mode
30175 PR middle-end/83609
30176 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
30177 last argument when extracting from CONCAT. If either from_real or
30178 from_imag is NULL, use expansion through memory. If result is not
30179 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
30180 the parts directly to inner mode, if even that fails, use expansion
30183 PR middle-end/83623
30184 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
30185 check for bswap in mode rather than HImode and use that in expand_unop
30188 Copyright (C) 2018 Free Software Foundation, Inc.
30190 Copying and distribution of this file, with or without modification,
30191 are permitted in any medium without royalty provided the copyright
30192 notice and this notice are preserved.